U.S. Department of Commerce Volume 99 Number 1 January 2001 Fishery Bulletin li.S. Department of Commerce Norman Y. Mineta Secretary National Oceanic and Atmospheric Administration D. James Baker Under Secretary for Oceans and Atmosphere National Marine Fisheries Service Penelope D. Dalton Assistant Administrator for Fisheries Scientific Editor Dr. John V. Merriner Editorial Assistant Sarah Shoffler Southeast Fisheries Science Center National Marine Fisheries Service, NOAA 101 Pivers Island Road Beaufort, NC 28516 ^t°fCo SrATES 0* Managing Editor Sharyn Matriotti National Marine Fisheries Sen/ice Scientific Publications Office 7600 Sand Point Way NE, BIN C15700 Seattle, Washington 98115-0070 The Fishery Bulletin (ISSN 0090-0656) is published quarterly by the Scientific Publications Office, National Marine Fish- eries Service, NOAA, 7600 Sand Point Way NE, BIN C 15700, Seattle, WA 98115-0070. Periodicals postage is paid at Seattle, WA, and at additional mailing offices. POST- MASTER: Send address changes for sub- scriptions to Fishery Bulletin, Superin- tendent of Documents, Attn.: Chief, Mail List Branch, Mail Stop SSOM, Washing- ton, DC 20402-9373. Although the contents of this publica- tion have not been copyrighted and may be reprinted entirely, reference to source is appreciated. The Secretary of Commerce has deter- mined that the publication of this peri- odical is necessary according to law for the transaction of public business of this Department. Use of funds for printing of this periodical has been approved by the Director of the Office of Management and Budget. For sale by the Superintendent of Documents, U.S. Government Printing Office, Washington, DC 20402. Subscrip- tion price per year: $55.00 domestic and $68.75 foreign. Cost per single issue: $34.00 domestic and $42.50 foreign. See back for order form. Editorial Committee Dr. Andrew E. Dizon Dr. Harlyn O. Halvorson Dr. Ronald W. Hardy Dr. Richard D. Methot Dr. Theodore W. Pietsch Dr. Joseph E. Powers Dr. Harald Rosenthal Dr. Fredric M. Serchuk National Marine Fisheries Service University of Massachusetts, Boston University of Idaho, Hagerman National Marine Fisheries Service University of Washington, Seattle National Marine Fisheries Service Universitat Kiel, Germany National Marine Fisheries Service Fishery Bulletin web site: fishbull.noaa.gov The Fishery Bulletin carries original research reports and technical notes on investigations in fishery science, engineering, and economics. It began as the Bulletin of the United States Fish Commission in 1881; it became the Bulletin of the Bureau of Fisheries in 1904 and the Fishery Bulletin of the Fish and Wildlife Service in 1941. Separates were issued as documents through volume 46; the last document was No. 1103. Beginning with volume 47 in 1931 and continuing through volume 62 in 1963, each separate appeared as a numbered bulletin. A new system began in 1963 with volume 63 in which papers are bound together in a single issue of the bulletin. Beginning with volume 70, number 1, January 1972, the Fishery Bulletin became a periodical, issued quarterly. In this form, it is available bv subscription from the Superintendent of Documents, U.S. Government Printing Office, Washington, DC 20402. It is also available free in limited numbers to libraries, research institutions. State and Federal agencies, and in exchange for other scientific publications. U.S. Department of Commerce Seattle, Washington Volume 99 Number 1 January 2001 Fishery Bulletin Contents The conclusions and opinions expressed in Fishery Bulletin are solely those of the authors and do not represent the official position of the National Marine Fisher- ies Service (NOAA) or any other agency or institution. The National Marine Fisheries Service (NMFS) does not approve, recommend, or endorse any proprietary product or pro- prietary material mentioned in this pub- lication. No reference shall be made to NMFS, or to this publication furnished by NMFS, in any advertising or sales pro- motion which would indicate or imply that NMFS approves, recommends, or endorses any proprietary product or pro- prietary material mentioned herein, or which has as its purpose an intent to cause directly or indirectly the advertised product to be used or purchased because of this NMFS publication. Articles 1-14 Begg, Gavin A., Wiiliam J. Overholtz, and Nancy J. Munroe The use of internal otolith morphometries for identification of haddock (Metanogrammus aeglefinus ) stocks on Georges Bank 15-28 Brown-Peterson, Nancy J., Robin M. Overstreet, Jeffrey M. Lotz, James S. Franks, and Karen M. Burns Reproductive biology of cobia, Rachycentron ccinadum, from coastal waters of the southern United States 29-39 Caretta, James V., Barbara L. Taylor, and Susan J. Chivers Abundance and depth distribution of harbor porpoise ( Phocoena phocoena ) in northern California determined from a 1995 ship survey 40-48 Dagorn, Laurent, Erwan Josse, and Pascal Bach Association of yellowfin tuna ( Thunnus albacares) with tracking vessels during ultrasonic telemetry experiments 49-62 Gharrett, Anthony J., Andrew K. Gray, and Jonathan Heifetz Identification of rockfish ( Sebostes spp.) by restriction site analysis of the mitochondrial ND-3/ND-4 and 12S/16S rRNA gene regions Companion articles 63-71 Harvey, Euan, David Fletcher, and Mark Shortis A comparison of the precision and accuracy of estimates of reef-fish lengths determined visually by divers with estimates produced by a stereo-video system 72-80 Harvey, Euan, David Fletcher, and Mark Shortis Improving the statistical power of length estimates of reef fish: a comparison of estimates determined visually by divers with estimates produced by a stereo-video system ii Fishery Bulletin 99(1 ) 81-93 Lazzari, Mark A. Dynamics of larval fish abundance in Penobscot Bay, Maine 94-107 Nelson, R. John, Maureen P. Small, Terry D. Beacham, and K. Janine Supernault Population structure of Fraser River Chinook salmon (Oncorhynchus tshawytscha ): an analysis using microsatellite DNA markers 108-122 Nichol, Daniel G. , and Erika 1. Acuna Annual and batch fecundities of yellowfin sole, Limanda aspera, in the eastern Bering Sea 123-138 Noll, Claire, Natalia V. Varnavskaya, Evgeny A. Matzak, Sharon L. Hawkins, Victoria V. Midanaya, Oleg N. Katugin, Charles Russell, Natalya M. Kinas, Charles M. Guthrie III, Hiroshi Mayama, Fumio Yamazaki, Bruce P. Finney, and Anthony J. Gharrett Analysis of contemporary genetic structure of even-broodyear populations of Asian and western Alaskan pink salmon, Oncorhynchus gorbuscha 139-150 Panfili, Jacques, and Javier Tomas Validation of age estimation and back-calculation of fish length based on otolith microstructures in tilapias (Pisces, Cichlidae) 151-167 Pella, Jerome, and Michele Masuda Bayesian methods for analysis of stock mixtures from genetic characters 168-179 Soh, SungKwon, Donald R. Gunderson, and Daniel H. Ito The potential role of marine reserves in the management of shortraker rockfish (Sebastes borealis ) and rougheye rockfish (5. aleutianus) in the Gulf of Alaska 180-192 St John, Jill, Garry R. Russ, Ian W. Brown, and Lyle C. Squire The diet of the large coral reef serranid Plectropomus leopardus in two fishing zones on the Great Barrier Reef, Australia 193-196 Notes Adam, M. Shiham, and Geoffrey P. Kirkwood Estimating tag-shedding rates for skipjack tuna, Katsuwonus pelamis, off the Maldives 197-201 DeMaster, Douglas P., Lloyd F. Lowry, Kathryn J. Frost, and Rebecca A. Bengston The effect of sea state on estimates of abundance for beluga whales ( Delphinapterus leucas) in Norton Sound, Alaska 202-209 Pierce, Daryl J., Behzad Mahmoudi, and Raymond R. Wilson Jr. Age and growth of the scaled herring, Harengula jaguana, from Florida waters, as indicated by microstructure of the sagittae 210-216 Stone, Heath H., and Langille K. Dixon A comparison of catches of swordfish, Xiphias gladius, and other pelagic species from Canadian longline gear configured with alternating monofilament and multifilament nylon gangions 217 Subscription form Abstract— Internal otolith morpho- metries, coupled with image analysis procedures and multivariate statistical analyses, were examined to investigate stock structure of haddock ( Melano - grammus aeglefinus) on Georges Bank in the northwest Atlantic. Samples were collected during spring 1995-97 from the Northeast Peak (eastern Georges Bank) and the Great South Channel (western Georges Bank) spawning com- ponents. The structure of transverse sagittal otolith sections were described for individual haddock samples from each spawning component by using a combination of linear morphometries, shape characteristics, and growth incre- ments. Analyses were structured to account for the effects of size, sex, age, and year class. Significant differences in internal otolith structure were found between eastern and western Georges Bank haddock, providing phenotypic evidence of stock separation between the two spawning components. East- ern Georges Bank haddock tended to have smaller internal otolith dimen- sions than western Georges Bank had- dock; these differences appeared to be related to growth rates. Total classifi- cation success for each spawning com- ponent varied from 61% to 83% for the different age and year-class combi- nations. Results from this study may be helpful in forming consistent stock definitions that can be used by both U.S. and Canadian fishery management agencies for rebuilding stocks of had- dock on Georges Bank. Manuscript accepted 29 June 2000. Fish. Bull. 99:1-14 (2001). The use of internal otolith morphometries for identification of haddock l Melanogrammus aeglefinus) stocks on Georges Bank Gavin A. Begg William J. Overholtz Nancy J. Munroe Northeast Fisheries Science Center National Marine Fisheries Service, NOAA 166 Water Street Woods Hole, Massachusetts 02543 Present address (for G. A. Begg): Marine Research Institute PO Box 1390 121 Reykiavik, Iceland E-mail address (for G A Begg) gavm@hafro.is Otoliths are crystalline structures com- posed of calcium carbonate and are ideal structures for use in fish stock iden- tification, containing a range of mea- surable characteristics including linear and shape morphometries, optical den- sity, and microstructural zonation and growth patterns, and elemental constit- uents (Ihssen et al., 1981; Campana and Neilson, 1985; Pawson and Jen- nings, 1996). Otoliths grow throughout the life of fish, are metabolically inert, and are typically available as a his- torical time series because of routine age and growth assessments (Campana and Neilson, 1985; Campana and Cas- selman, 1993). Linear and shape morphometries of otoliths have been widely used for fish stock identification (e.g. Messieh et ah, 1989; Dawson, 1991; Smith, 1992; Cam- pana and Casselman, 1993; Friedland and Reddin, 1994), although their use has been questioned because of within- stock differences in sex, age, and year- class variation ( Castonguay et ah , 199 1 ; Begg and Brown, 2000). Moreover, oto- lith morphometries have been found to be strongly correlated with growth, which influences development of oto- lith crystalline microstructure (Smith, 1992; Campana and Casselman, 1993). Stock definitions based on differences in otolith structure, therefore, depend not only on differential growth rates, but on the consistency of the environ- ment integrated over the life history of fish in each stock (Campana and Cas- selman, 1993). Although otolith mor- phometries cannot be used to differen- tiate stocks on a genetic basis, they can provide a phenotypic basis for stock separation that is useful for fisheries management (Casselman et al., 1981; Begg and Waldinan, 1999). Fisheries management is moving to- wards a precautionary approach to en- sure sustainable utilization of our ma- rine resources (FAO, 1995; ICES1). One requirement of the precautionary ap- proach is to consider the full impact of management actions, including explicit consideration of stock complexity (Gar- cia and Grainger, 1997). For this rea- son, there is a growing interest in the importance and recognition of individu- al spawning components within histor- ically established management units (FAO, 1995; Stephenson, 1999). The importance of individual spawn- ing components has been acknowledged in the management of haddock, Mela- nogrammus aeglefinus , a commercially important groundfish of the northwest Atlantic. The interest in haddock stock complexity is exemplified on Georges Bank (Fig. 1 ), where this species forms an important transboundary resource among U.S. and Canadian fishermen (Halhday and Pinhorn, 1990; Begg, 1 ICES (International Council for the Explo- ration of the Sea). 1997. Report of the study group on the precautionary approach to fisheries management. ICES council meeting (CM) 1997/assess 7, 41 p. 2 Fishery Bulletin 99(1 ) Northeast Fisheries Science Center stratified random survey stations from where eastern and western Georges Bank haddock samples were collected in 1995 (open squares), 1996 (open circles), and 1997 (closed circles) for stock identification based on internal otolith morphometries. 1998). During 1935 to 1960, annual landings of haddock on Georges Bank averaged 46,000 metric tons (t), before increasing to 150,000 t in 1965 owing to exceptional re- cruitment from the 1962 and 1963 year classes and in- tense fishing resulting from the entrance of distant water fleets (Clark et al., 1982). However, following these record landings the resource collapsed, and annual landings de- clined to less than 5,000 t during the mid-1970s. The re- source temporarily increased during the late 1970s and early 1980s when annual landings increased to 27,000 t owing to the large 1975 and 1978 year classes, before de- clining to a record low level of 2,300 t in 1995 (Brown, 1998). Currently, the haddock resource on Georges Bank is in a depleted condition and is the focus of rebuilding plans by both the U.S. and Canadian fishery management agen- cies (Brown, 1998; Gavaris and Van Eeckhaute, 1998). The fact that each nation conducts separate analytical assess- ments, using different stock definitions for the resource, is problematic. The United States assesses Georges Bank haddock as a single stock, encompassing historical spawn- ing components on the Northeast Peak (eastern Georges Bank) and the Great South Channel (western Georges Bank); whereas Canada assesses only the spawning com- ponent on the Northeast Peak (Brown, 1998; Gavaris and Van Eeckhaute, 1998) (Fig. 1). Since the mid-1980s, the majority of the haddock resource has been concentrated over eastern Georges Bank, where the majority of land- ings have been taken by the Canadian fishery (Brown, 1998). The changing resource status of haddock on Georg- es Bank has probably been due to greater depletion of the western Georges Bank spawning component, which may now be contributing at a much lower level to the overall productivity of the resource (Van Eeckhaute et al., 1999). Stock rebuilding plans of both nations need to examine the identity versus the separateness, of spawning components on Georges Bank and develop uniform stock definitions for the resource. Haddock stocks on Georges Bank have been examined by using a variety of techniques, although the results in terms of identification have been far from conclusive. Differences in distribution, life history, and otolith shape characteristics have indicated the existence of separate eastern and western Georges Bank spawning components (Smith and Morse, 1985; Gavaris and Van Eeckhaute, 1998; Begg et al., 1999; Begg and Brown, 2000). In con- trast, tag-recapture, ichthyoplankton surveys, and genet- ic analyses have indicated some interchange of haddock across the Bank that may reflect a single spawning com- ponent (Needier, 1930; Schroeder, 1942; Morse et al., 1987; Purcell et ah, 1996). Hence, there remains considerable uncertainty regarding the stock structure of haddock on Georges Bank. The limitations associated with traditional stock identi- fication techniques have been a major factor responsible for the remaining uncertainty regarding haddock stock structure on Georges Bank (Begg, 1998). Consequently, we used a new stock identification technique based on in- ternal otolith morphometries. We considered internal oto- lith morphometries as a generic term describing the oto- lith microstructure of individual growth zones as well as the linear and shape dimensions of transverse otolith sec- Begg et al.: Use of otolith morphometries for identification of stocks of Melanogrommus aeglefinus 3 Table 1 Details of haddock samples used in analysis of internal otolith morphometries for stock identification. Region Date captured Age group (years) Length range (cm) n Eastern Georges Bank (EGB) 10-11 Apr 1995 1 21-28 7 2 31-37 10 21-23 Apr 1996 1 21-32 6 3 41-53 9 4 45-54 9 6-8 Apr 1997 2 38-45 18 3 40-47 10 4 41-58 16 5 49-62 10 Western Georges Bank (WGB) 13 Apr 1995 1 24-31 11 16-20 Apr 1996 2 31-47 9 3 40-55 9 25 Mar-23 Apr 1997 2 36-44 17 3 39-60 44 4 47-67 37 5 50-73 10 tions. Although, linear morphometries and shape analysis of “external” or whole otoliths have been used for fish stock identification, there has been little use of “internal” otolith morphometries, probably because of the time, ex- pense, and potential difficulties in obtaining consistent otolith sections. However, most fisheries agencies now use a standardized procedure for sectioning otoliths. These standardized procedures enable consistent and compara- ble otolith sections to be obtained, providing a consistent, rapid, and readily accessible structure that can be exam- ined for stock identification. We investigated the feasibility of using internal otolith morphometries for fish stock identification, by considering Georges Bank haddock as a representative case study. For the purposes of our study, we considered a stock, or spawn- ing component, as a semidiscrete, self-reproducing group of fish with definable morphometric characteristics that are assumed to be homogeneous for management purpos- es (Begg and Waldman, 1999). Variation in these char- acteristics is assumed to be evidence that distinct geo- graphic regions are partially occupied throughout the life history of the fish, thereby providing a phenotypic basis for stock identification (Ihssen et al., 1981). Materials and methods Sample collection Haddock samples were collected in 1995, 1996, and 1997 during spring Northeast Fisheries Science Center (NEFSC) stratified random bottom trawl surveys, when the fish were assumed to be on or near their spawning grounds. Most of the adults (ages 2+) sampled were in spawning condition. Samples were collected from survey stations throughout eastern (EGB) and western Georges Bank (WGB) (Fig. 1; Table 1). At sea, haddock samples were measured (fork length [FL], to the nearest cm), sex and maturity were determined by macroscopic examina- tion of the gonads, and sagittal otolith pairs were removed from each sample. In the laboratory, one otolith from each pair was sectioned and assigned an age by following stan- dard methods for northwest Atlantic species (Pentilla and Dery, 1988). Standardized sectioning procedures ensured consistent and comparable otolith sections for morphomet- ric measurements and subsequent statistical analyses. Internal otolith morphometries Internal otolith morphometries were obtained from each otolith section by using the OPTIMAS™ (version 6.2) image analysis system (Media Cybernetics, 1996). All mea- surements were taken at a magnification of 15x. The perimeter of each otolith section was traced in a counter- clockwise direction to allow morphometries to be calcu- lated for each structure (shape). Six linear morphometries (length, width, Al, HI, A2, and H 2) were measured for each sample, where A 1 is the growth increment to the first annulus, HI is the width of the hyaline band of the first annulus, A2 is the growth increment to the second annu- lus, and H2 is the width of the hyaline band of the second annulus (Fig. 2). In addition, four shape variables (area, perimeter, circularity, and rectangularity) were measured for each sample. Circularity was defined as the perimeter of the otolith section squared divided by its area, and rect- angularity was defined as the otolith section area divided 4 Fishery Bulletin 99(1 ) Figure 2 Internal otolith morphometric variables that were measured for stock identification of haddock on Georges Bank. by the area of its enclosing rectangle oriented along the length of the section (Media Cybernetics, 1996). Data analysis Internal otolith morphometries were compared between eastern and western Georges Bank haddock for stock identification. All variables were first examined for assumptions of normality and homoscedasticity and were subsequently loge-transformed prior to statistical analysis if these assumptions were not satisfied. Following trans- formation, all variables conformed to statistical assump- tions. Analysis of covariance (ANCOVA) was then used to determine the effect of fish length on the magnitude of each variable. Those variables found to have significant interactions (P<0.05) between sampling region and fish length (i.e. samples with unequal slopes) were excluded from further statistical analyses. In contrast, those vari- ables found to have samples with equal slopes, but that significantly correlated with fish length, were corrected for fish length by using their respective common within-group slope ( 6 ) to standardize the samples. Potential confounding sources of variation among sam- ples, such as sex, age, and year-class differences, were examined and accounted for in the analyses before in- terpreting stock differences. Such confounding variables need to be examined to ensure interpretations of stock dif- ferences are real and not simply the result of sample vari- ation (Castonguay et al., 1991; Begg and Brown, 2000). Consequently, multivariate analysis of variance (MANO- VA) was used to examine differences between the sexes in their internal otolith structure, for samples from the same region and year class, and of the same age, by using the appropriate length-corrected internal otolith morpho- metric variables. One-way, fixed effects, unbalanced anal- ysis of variance (ANOVA) was then used to examine in- dividual variables to interpret any significant differences detected by the MANOVAs. Significance levels were cor- rected for multiple testing by using the Bonferroni adjust- ment factor (Sokal and Rohlf, 1995). The same tests were then used to examine the internal otolith structure of dif- ferent-age fish (sampled from the same region and year), and fish originating from different year classes (same region and age). Tukey’s honestly significant difference (HSD) tests were used for a posteriori comparisons for each significant variable. The same multi- and univari- ate tests were then used to examine the internal otolith structure of haddock sampled from eastern and western Georges Bank. These analyses were used to determine if there was any evidence for stock separation of haddock across the Bank. Principal component (PC) analysis of the length-correct- ed internal otolith morphometric variables was conduct- ed for samples of the same age and year class to provide an unbiased indication of separation between eastern and western Georges Bank haddock (i.e. there were no a priori assumptions of group membership). ANOVA was used on the significant principal components to examine differenc- es in the PC scores between the proposed groups. Forward stepwise canonical discriminant analysis was then used to detect differences in the internal otolith structure of east- ern and western Georges Bank haddock samples. The sig- nificant (P<0.05) canonical variate (CV) derived by each analysis represented the optimal combination of morpho- metric variables that provided the best overall discrimina- tion between the samples. ANOVA was used to examine differences in the CV scores. Jack-knifed cross-validation procedures were used to give unbiased estimates of clas- sification success (SPSS, Inc., 1997). Results All the internal otolith morphometric variables, except for Al and rectangularity, were log^-transformed prior to sta- tistical analysis to correct for nonnormality and inequality of variances. ANCOVA indicated a significant correlation between fish length and region for otolith width (homoge- neity of slopes test, P<0. 0222) (Table 2). Significant region- specific differences in the correlation between fish length and otolith width (i.e. regional differences in growth rate) made it necessary to remove otolith width as an analysis Begg et al.: Use of otolith morphometries for identification of stocks of Melanogrammus oeglefinus 5 Table 2 Internal otolith morphometric variables significantly correlated with fish length, and the corresponding regression coefficients ( b ) required to standardize the variables for fish length. Otolith morphometric variable Length x region Length b F df P F df P Al 0.85 15,200 0.6195 1.70 1,215 0.1936 HI 1.27 13,183 0.2333 1.22 1,196 0.2713 — A2 0.92 12,181 0.5274 2.36 1,193 0.1261 — m 1.19 12,155 0.2918 11.62 1,167 0.0008 0.01810 Length 1.16 15,200 0.3082 103.65 1, 215 0.0001 0.00922 Width 1.93 15,200 0.0222 — — — — Area 1.53 15,200 0.0971 91.91 1,215 0.0001 0.01514 Perimeter 1.48 15,200 0.1165 88.05 1, 215 0.0001 0.00837 Circularity 1.06 15,200 0.4007 2.92 1, 215 0.0889 — Rectangularity 1.63 15,200 0.0695 8.50 1, 215 0.0039 -0.00092 Table 3 Results of MANOVA and ANOVA showing significant differences in the length-corrected internal otolith moi between age groups for eastern (EGB) and western Georges Bank (WGB) haddock. •phometric variables Region Year Age group comparison (years) MANOVA Significant variable ANOVA F df P F df P EGB 1995 1 vs. 2 15.51 6,10 0.0002 length 19.16 1,15 0.0005 area 28.70 1,15 0.0001 perimeter 12.39 1,15 0.0031 1996 1,3 vs. 4 3.98 12,34 0.0007 length 32.79 2,21 0.0001 area 127.78 2,21 0.0001 perimeter 23.13 2,21 0.0001 circularity 6.93 2,21 0.0049 1997 2,3,4 vs. 5 4.05 24,135 0.0001 Al 6.87 3,50 0.0006 H\ 4.75 3,50 0.0054 length 4.90 3,50 0.0046 area 7.71 3,50 0.0002 WGB 1996 2 vs. 3 4.71 8,9 0.0163 length 11.99 1,16 0.0032 perimeter 9.19 1,16 0.0079 1997 2,3,4 vs. 5 5.46 24,297 0.0001 area 10.81 3,104 0.0001 circularity 7.46 3,104 0.0001 rectangularity 4.96 3,104 0.0029 variable. The same ANCOVAs also indicated that H 2 and otolith length, area, perimeter, and rectangularity were correlated with fish length (P<0.004), resulting in those variables being corrected for fish length by using their respective within-group regression coefficient (Table 2). Internal otolith morphometries were not significantly different between the sexes (MANOVA, P>0.05), result- ing in the sexes being pooled within each region, year and age strata to increase the statistical power used in subsequent analyses. In contrast, significant differences were found between the different age groups of haddock (Table 3). Not unexpectedly, haddock that were 1 and 2 years of age tended to have significantly smaller otoliths than haddock that were 3, 4, and 5 years of age (HSD, P<0.05). Likewise, there were significant differences in- dicative of annual growth differences found in the in- ternal otolith morphometries of haddock sampled from different year classes (Table 4). Hence, the remaining 6 Fishery Bulletin 99(1 ) Table 4 Results of MANOVA and AN OVA showing significant differences in the length-corrected internal otolith morphometric variables between year classes for eastern (EGB) and western Georges Bank (WGB) haddock. Region Age (yr) Year-class comparison (yr) MANOVA Significant variable ANOVA F df P F df P EGB 1 1994 vs. 1995 1.23 6,6 0.4034 2 1993 vs. 1995 7.82 8, 18 0.0002 — — — — 3 1993 vs. 1994 2.22 9,9 0.1255 H2 11.83 1, 17 0.0031 4 1992 vs. 1993 1.90 9, 15 0.1309 perimeter 11.57 1,23 0.0025 circularity 9.14 1,23 0.0060 WGB 2 1994 vs. 1995 1.76 8, 17 0.1549 — — — — 3 1993 vs. 1994 3.55 9,43 0.0023 HI 15.59 1,51 0.0002 Table 5 Results of MANOVA and AN OVA showing significant differences in the length -corrected internal otolith morphometric variables between haddock from eastern and western Georges Bank. Year Age (yr) MANOVA Significant variable ANOVA F df P F df P 1995 1 1.60 6, 11 0.2357 rectangularity 5.76 1, 16 0.0289 1996 3 1.18 9,8 0.4109 — — — — 1997 2 2.10 8,26 0.0726 circularity 10.37 1,33 0.0029 3 4.30 9, 44 0.0005 Al 22.76 1, 52 0.0001 A2 7.84 1, 52 0.0072 H2 7.41 1, 52 0.0088 4 3.22 9,43 0.0045 A2 3.97 1,51 0.0518 5 5.53 9, 10 0.0067 Al 9.70 1, 18 0.0060 H2 9.90 1, 18 0.0056 analyses were conducted for samples of individual ages and year classes in order to minimize the effects of these confounding variables. Significant differences in the internal otolith morpho- metries between haddock sampled on eastern and west- ern Georges Bank were found in three out of six compar- isons (Table 5). Eastern Georges Bank haddock tended to have smaller otoliths (i.e. length, area, perimeter, cir- cularity and rectangularity) than western Georges Bank haddock; this finding was derived mainly from the first growth increment (Al) which also tended to be smaller (Fig. 3). For the three significant comparisons, scatter plots of the most significant individual internal otolith morphometries for samples from each age group typically showed separation patterns between eastern and western Georges Bank haddock, albeit with some overlap (Figs. 4-6). Principal component analysis provided further support that haddock from eastern and western Georges Bank separated into two groups (Fig. 7). Eastern and western Georges Bank haddock samples, 1 and 3 (1994 year class) years of age were mainly separated on the first principal component (PC I) (AN OVA, P<0.06), whereas samples that were 2, 3 ( 1993 year class), and 5 years of age were mainly separated on the second and third principal components (PC II and PC III) (ANOVA, P<0.05). Principal compo- nents I, II, and III accounted on average for 38.7 ±5.3% SD, 22.6 ±5.9%, and 15.6 ±0.7% of the total variation in the data. Differences in length, area, and perimeter were mainly responsible for the observed separations along the first principal component, whereas Al, HI and A2 were the main variables responsible for separation along the second and third principal components. Discriminant analysis also indicated that haddock com- prise two groups on Georges Bank (Fig. 8). Significant dif- ferences in the discriminant (CV I) scores between eastern and western Georges Bank haddock were found for all age groups (ANOVA, P<0.05), except age group 4. Total clas- sification success varied from 61% to 83% for the different age and year-class combinations (Table 6). Begg et al.: Use of otolith morphometries for identification of stocks of Melanogrcimmus oeglefinus 7 -1.5i -1 -3.0- -3.5 J — i 1 1 1 1 1 — 1(94) 2(95) 3(93) 3(94) 4(93) 5(92) -2.4-1 -4.0 — i 1 1 1 1 1 — 1(94) 2(95) 3(93) 3(94) 4(93) 5(92) 2.5- 2.4- 2.3- 2.2-^ 2.1 — i 1 1 r 1 1 — 1(94) 2(95) 3(93) 3(94) 4(93) 5(92) 2.9 1 . 1 1 r- 1(94) 2(95) 3(93) 3(94) 4(93) 5(92) Age group (year class) Figure 3 Mean ± standard deviation of age-specific internal otolith growth morphometries (Al, HI, A2, H 2, and otolith length, area, perim- eter, circularity, and rectangularity) for eastern Georges Bank (squares) and western Georges Bank (circles) haddock. Discussion We found significant differences in the internal otolith structure between eastern and western Georges Bank haddock in three out of six comparisons; providing a phe- notypic basis for stock separation across the Bank. Of the three nonsignificant comparisons, two were influenced by low sample sizes (n = 18), whereas the third was margin- ally nonsignificant (P>0.07) (Table 6). Eastern Georges Bank haddock tended to have smaller internal otolith mor- phometries than western Georges Bank haddock, particu- larly during the first year of life when growth differences between progeny from the two spawning components may be most apparent. Differences in the internal otolith structure of eastern and western Georges Bank haddock corresponded with ap- parent differences in their growth rates. Commercial land- ings data indicated smaller mean lengths and weights at age for eastern than for western Georges Bank haddock, indicative of slower growth rates (and resultant smaller otoliths) for eastern Georges Bank haddock (Brown2). Likewise, other studies have found significant relation- ships between linear and shape otolith morphometries and fish growth (e.g., Mosegaard et al., 1988; Reznick et al., 1989; Secor and Dean, 1989; Smith and Kostlan, 1991; Fowler and Short, 1996). Consequently, regional differ- ences in growth rate may be a principal determinant in 2 Brown, R. W. 2000. Annual assessment data (unpubl. data). Food Web Dynamics Program, Population Dynamics Branch, Northeast Fisheries Science Center. 166 Water St., Woods Hole, MA 02543. 8 Fishery Bulletin 99(1 ) 0.4- -0.4- -1.2- -2.0- 0.6 ■ ^"oo o I °o 1 gpO a o° °0 °°io °c oo 0?) o 1.4 41 2.2 3.0 1 .6 1 1.5- 1.4- 1.3 1.2 0.6 ■ oft) O o o o o o o° 0o o ° °o ° °°o 1.4 41 2.2 3.0 -2.4-1 -3.0- -3.6- -4.2- ■o f " o ° o ° o o jpo ® ° O _ 0.6 1.4 2.2 41 3.0 0.74 1 ■=0.71 - (13 CD C ro q) 0.68 0.65 m 9° "" ■ ■c 0.73- CO £ 0.69- 0.65 o o o ° -2.2 -1.4 -0.6 Loge A2 0.2 Figure 5 Scatter plots of internal otolith morphometric variables for 4-year-old haddock ( 1993 year class) show- ing grouping patterns of eastern Georges Bank (squares) and western Georges Bank (circles). studies (Campana and Neilson, 1985; Mosegaard et al., 1988; Campana and Casselman, 1993), although there may also be a genetic contribution (Gauldie and Nelson, 1990; Friedland and Reddin, 1994). The eastern and western Georges Bank spawning com- ponents, therefore, probably comprise phenotypically sepa- rate individuals that reflect differences in otolith structure due to environmental variation. These types of morpholog- ical differences indicate growth rate differences linked to the environment, rather than any genetic differences. Our results concur with previous studies that indicate sepa- rate spawning components on Georges Bank (Smith and Morse, 1985; Begg et al., 1999; Begg and Brown, 2000), although the degree of connectivity between the two com- ponents is not known. Larvae spawned on the Northeast Peak recruit to the central part of the Bank as they de- velop and are advected from there along its southern flank (Lough and Bolz 1989), whereas some larvae spawned throughout the Great South Channel are advected along the northern flank, with the result that there is some mix- ing between progeny. Further studies need to examine mixing rates and spawning-site fidelity of individual fish originating from the eastern and western Georges Bank spawning components in order to determine if there is an underlying genetic basis for stock separation. Analysis of internal otolith morphometries may provide a more detailed description of individual fish stocks than morphometric analysis of whole otoliths because the use of internal otolith morphometries specifically incorporates individual growth zones, as well as characteristic shape qualifiers. Measurement of the first growth zone in whole otoliths has commonly been used in stock identification studies, although the results have been far from conclu- sive (e.g. Dawson, 1991; Hopkins4; Marecos5). Likewise, mixed results have been found for microstructure anal- ysis of otolith nuclear dimensions and growth incremen- tal widths (e.g. Rybock et al., 1975; Neilson et al., 1985; Mosegaard and Madsen6). Certainly, the use of more than one growth or shape dimension improved our ability to identify groups, but the utility or cost-effectiveness of in- ternal otolith morphometries may be questioned when compared to shape analysis of whole otoliths. 4 Hopkins, P. J. 1986. Mackerel stock discrimination using oto- lith morphometries. ICES CM 1986/H 7, 16 p. 5 Marecos, M. L. 1986. Preliminary analysis of horse mackerel ( Trachurus trachurus L. ) otolith (LI ) measurements. ICES CM 1986/H 72, 8 p. 6 Mosegaard, H., and K. P. Madsen. 1996. Discrimination of mixed herring stocks in the North Sea using vertebral counts and otolith microstructure. ICES CM 1996/H 17, 8 p. 10 Fishery Bulletin 99(1) 2.0 -| 1.5- 1.0- o o o o o o ■ 0.5 -I — -4.2 — i 1 — -3.7 -3.2 Log, H 2 -2.7 o -2.0-1— -4.2 -3.7 -3.2 Log, H 2 -2.7 -2.0 1 Q) O) o -J-3.0- o OHO ■ -3.6 -I — -4.2 o -3.7 -3.2 -2.7 2.2 -4.2 -3.7 -3.2 -2.7 Log, H2 Log, H 2 Figure 6 Scatter plots of internal otolith morphometric variables for 5-year-old haddock ( 1992 year class) show- ing grouping patterns of eastern Georges Bank (squares) and western Georges Bank (circles). Classification success similar to our present study was found between eastern and western Georges Bank had- dock in a comparable study where whole-otolith shape analysis was used (Begg and Brown, 2000). However, the use of internal otolith morphometries entailed a few more caveats than that of whole-otolith shape analysis; there- fore internal otolith morphometries was questioned as a preferred tool for stock identification. Although our ap- proach was semi-automated, with an aim to increasing objectivity and decreasing processing time, the use of in- ternal otolith morphometries still required some inter- pretation by the person taking the measurements. This was particularly true for growth zones, where interpreta- tion partly compromised our goals of increasing objectiv- ity and speed. Difficulties in interpreting measurements when growth zones are poorly defined can also be a source of uncertainty (Hopkins4). These sources of uncertainty should be minimized, provided the person taking the mea- surements is also an experienced reader of otolith growth increments. Internal otolith morphometries are useful for stock iden- tification, but significant overlap among variables may preclude their use for stock discrimination, that is to say for classifying unknown fish according to origin of spawning component. Although mean values may differ between stocks, measurements from individual fish may not allow them to be classified to a particular stock be- cause of individual within-stock growth differences (Paw- son and Jennings, 1996). For example, slow-growing fish from a fast-growing stock may be incorrectly classified with a slow-growing stock (Campana and Casselman, 1993). However, provided growth differences exist and analyses are conducted to account for samples of mixed ages and year classes to minimize the effects of these con- founding variables, internal otolith morphometries can be a useful phenotype-based stock identification tool. Results from this analysis on internal otolith morpho- metric differences have added to the evidence indicating separation between the eastern and western Georges Bank haddock spawning components. Although these differences do not provide a genetic basis for separation between the two spawning components, they do reflect the phenotypic characteristics of each spawning component, indicative of stock separation during life history. Studies such as this one, are needed to provide evidence of stock structure if his- torically established management units are to be changed in response to changing exploitation and resource patterns. Evidence of stock separation within the Georges Bank had- dock resource may be useful in forming stock definitions that can be used by both the U.S. and Canada in defining Begg et al.: Use of otolith morphometries for identification of stocks of Melanogrammus aeglefinus 11 2.0-1 0.5 - O a. _-| 0 -2.5 Age group 1 , year class 1 994 (PC I, P< 0.0572) ■ ■ ■ 3.0 1 2 1.0- o -1.0 -3.0 Age group 3, year class 1994 (PC I, P< 0.0060) o o ° ® ° °° oo ° o° %° °&a5 o 0„un O' -4.0 -2.0 0.0 2.0 -2.5 -0.5 1.5 3.5 3.0 1.0 O ^ -1.0 -3.0 Age group 2, year class 1995 (PC II, P<0.0096) Qd _ o O o o £ ■ -2.5 -0.5 1.5 3.5 3.0 1 2) 1.0 o o O -1.0H -3.0 Age group 4, year class 1993 (PC II, P<0.1 572) o K* o o o -3.5 -1.5 0.5 2.5 Age group 3, year class 1993 3.0 i (PC III, P<0. 0467) 2 1.0 o O Q- -1.0 -3.0 -2.5 -0.5 1.5 3.5 3.0-1 9) 1.0- O °- -1.0 -3.0 Age group 5, year class 1992 (PC II, P<0 0011) i - -2.0 -0.5 1.0 2.5 PC I score Figure 7 Principal component (PC) analysis of internal otolith morphometric variables and ANOVA results showing grouping patterns of eastern Georges Bank (squares) and western Georges Bank (circles) haddock. management units that are consistent and account for the underlying stock structure of the resource. An incorrect decision could lead to significant shifts in resource dis- tribution, changes in stock productivity, or declines in re- cruitment across stock unit boundaries. Although the pre- cautionary approach would imply that we accept the two stock hypothesis as the default scenario until proven oth- erwise, this has not been the case. It would be desirable to ensure the conservation and stock rebuilding potential of both spawning components, particularly because we do not know at present the relative contribution of each to the overall status of haddock on Georges Bank. 12 Fishery Bulletin 99(1 ) Age group 1 , year class 1994 Age group 3, year class 1994 Age group 2, year class 1 995 (CV I, P<0 0002) 5 4 Age group 4, year class 1993 5, 4 3 2 1 0 Age group 3, year class 1993 (CV I, P<0.0009) Age group 5, year class 1992 (CV I, P<0.0001) 1 1 [ 3 nil 1 _dh ; J 1 ill n n -2.3 -1.3 -0.3 0.7 1.7 -2.0 -1.0 0 1.0 2.0 CV I score Figure 8 Discriminant (CV I) scores of internal otolith morphometric variables and corresponding ANOVA results showing grouping patterns of eastern Georges Bank ( squares) and western Georges Bank (circles) haddock. Acknowledgments We would like to thank Russell Brown, Steve Murawski, Kevin Friedland, and two anonymous reviewers for their comments; Ruth Haas-Castro for assistance with OPTI- MAS; and Frank Almeida, George Bolz, Jay Burnett, and Christine Esteves for their suggestions and assistance with ageing and collection of samples. This work was per- formed while G.A.B. held a National Research Council (NOAA/NMFS/NEFSC) Research Associateship. Literature cited Begg, G. A. 1998. A review of stock identification of haddock, Melano- grammus aeglefinus, in the northwest Atlantic Ocean. Mar. Fish. Rev. 60(4): 1-15. Begg, G. A., and R. W. Brown. 2000. Stock identification of haddock Melanogrammus aegle- finus on Georges Bank based on otolith shape analysis. Trans. Am. Fish. Soc. 129:935-945. Begg et al.: Use of otolith morphometries for identification of stocks of Metanogrammus aeg/efinus 13 Begg, G. A., J. A. Hare, and D. D. Sheehan. 1999. The role of life history parameters as indicators of stock structure. Fish. Res. 43:141-163. Begg, G. A., and J. R. Waldman. 1999. An holistic approach to fish stock identification. Fish. Res. 43:35-44. Brown, R. W. 1998. Haddock. In Status of the fishery resources off the northeastern United States for 1998, p. 53-56. U.S. Dep. Commer., NOAATech. Memo. NMFS-NE-115. Campana, S. E., and J. M. Casselman. 1993. Stock discrimination using otolith shape analysis. Can. J. Fish. Aquat. Sci. 50: 1062-1083. Campana, S. E., and J. D. Neilson. 1985. Microstructure of fish otoliths. Can. J. Fish. Aquat. Sci. 42:1014-1032. Casselman, J. M., J. J. Collins, E. J. Crossman, R E. Ihssen, and G. R. Spangler. 1981. Lake whitefish (Coregonus clupeaformis ) in the Great Lakes region. Can. J. Fish. Aquat. Sci. 38:1772-1789. Castonguay, M., P. Simard, and P. Gagnon. 1991. Usefulness of fourier analysis of otolith shape for Atlantic mackerel (Scomber scombrus) stock discrimina- tion. Can. J. Fish. Aquat. Sci. 48:296-302. Clark, S. H., W. J. Overholtz, and R. C. Hennemuth. 1982. Review and assessment of the Georges Bank and Gulf of Maine haddock fishery. J. Northwest Atl. Fish. Sci. 3:1-27. Dawson, W. A. 1991. Otolith measurement as a method of identifying fac- tors affecting first-year growth and stock separation of mackerel (Scomber scombrus L.). J. Cons. Int. Explor. Mer 47:303-317. Drinkwater, K. F., and D. G. Mountain. 1997. Climate and oceanography. In Northwest Atlantic groundfish: perspectives on a fishery collapse ( J. Boreman, B. S. Nakashima, J. A. Wilson and R. L. Kendall, eds.), p. 3-25. Am. Fish. Soc., Bethesda, MD. FAO (Food and Agriculture Organization of the United Nations). 1995. Code of conduct for responsible fisheries. FAO, Rome, 41 p. Fowler, A. J., and D. A. Short. 1996. Temporal variation in the early life-history character- istics of the king george whiting ( Sillaginod.es punctata) from analysis of otolith microstructure. Mar. Freshwater Res. 47:809-18. Friedland, K. D., and D. G. Reddin. 1994. Use of otolith morphology in stock discriminations of Atlantic salmon (Salmo sala?'). Can. J. Fish. Aquat. Sci. 51:91-98. Garcia, S. M., and R. Grainger. 1997. Fisheries management and sustainability: a new per- spective of an old problem? In Developing and sustaining world fisheries resources: the state of science and manage- ment: second world fisheries congress proceedings (D. A. Hancock, D. C. Smith, A. Grant, and J. P. Beumer, eds.), p. 631-654. CSIRO Publishing, Collingwood, Australia. Gauldie, R. W. 1990. A measure of metabolism in fish otoliths. Comp. Bio- chem. Physiol. 97A:475-480. Gauldie, R. W., and D. G. A. Nelson. 1990. Otolith growth in fishes. Comp. Biochem. Physiol. 97A:119-135. Gavaris, S., and L. Van Eeckhaute. 1998. Assessment of haddock on eastern Georges Bank. Can. Dep. Fish. Oceans, DFO Atl. Fish. Res. Doc. 98/66, 75 p. Halliday, R. G., and A. T. Pinhorn. 1990. The delimitation of fishing areas in the northwest Atlantic. J. Northwest Atl. Fish. Sci. 10:1-51. Ihssen, P. E., H. E. Booke, J. M. Casselman, J. M., McGlade, N. R. Payne, and F. M. Utter. 1981. Stock identification: materials and methods. Can. J. Fish. Aquat. Sci. 38:1838-1855. Lough, R. G., and G. R. Bolz. 1989. The movement of cod and haddock larvae onto the shoals of Georges Bank. J. Fish Biol. 35 (suppl. A): 71- 79. Media Cybernetics 1996. OPTIMAS, version 6.2, eighth ed. Media Cybernet- ics, Silver Spring, Maryland, vol. I: xiii + 474 p., vol. II: viii + 630 p. Messieh, S. N., C. MacDougall, and R. Claytor. 1989. Separation of Atlantic herring (Clupea harengus ) stocks in the Gulf of St. Lawrence using digitized otolith morphometries and discriminant function analysis. Can. Tech. Rep. Fish. Aquat. Sci. 1647, 22 p. Morse, W. W., M. P. Fahay, and W. G. Smith. 1987. MARMAP surveys of the continental shelf from Cape Hatteras, North Carolina, to Cape Sable, Nova Scotia ( 1977-1984). Atlas 2: annual distribution patterns of fish larvae. U.S. Dep. Commer., NOAA Tech. Memo. NMFS-S/ NEC 47, 215 p. Mosegaard, H., H. Svedang, and K. Taberman. 1988. Uncoupling of somatic and otolith growth rates in Arctic char (Salvelinus alpinus) as an effect of differ- ences in temperature response. Can. J. Fish. Aquat. Sci. 45:1514-1524. Needier, A. W. H. 1930. The migrations of haddock and the interrelationships of haddock populations in North American waters. Con- trib. Can. Biol. 6:241-313. Neilson, J. D., G. H. Green, and B. Chan. 1985. Variability in dimensions of salmonid otolith nuclei: implications for stock identification and microstructure interpretation. Fish. Bull. 83:81-89. Pawson, M. G., and S. Jennings. 1996. A critique of methods for stock identification in marine capture fisheries. Fish. Res. 25:203-217. Pentilla, J., and L. M. Dery. 1988. Age determination methods for Northwest Atlantic species. U.S. Dep. Commer., NOAA Tech. Rep. NMFS 72:37-40. Purcell, M. K., I. Kornfield, M. Fogarty, and A. Parker. 1996. Interdecadal heterogeneity in mitochondrial DNA of Atlantic haddock (Melanogrammus aeglefinus ) from Georges Bank. Molecular Mar. Biol. Biotech. 5:185-192. Reznick, D., E. Lindbeck, and H. Bryga. 1989. Slower growth results in larger otoliths: an exper- imental test with guppies (Poecilia reticulata ). Can. J. Fish. Aquat. Sci. 46:108-12. Rybock, J. T, H. F. Horton, and J. L. Fessler. 1975. Use of otoliths to separate juvenile steelhead trout from juvenile rainbow trout. Fish. Bull. 73:654-659. Schroeder, W. C. 1942. Results of haddock tagging in the Gulf of Maine from 1923 to 1932. J. Mar. Res. 5:1-19. Secor, D. H., and J. M. Dean. 1989. Somatic growth effects on the otolith-fish size rela- tionship in young pond-reared striped bass, Morone saxati- lis. Can. J. Fish. Aquat. Sci. 46:113-121. Smith, M. K. 1992. Regional differences in otolith morphology of the deep 14 Fishery Bulletin 99(1) slope red snapper Etelis carbunculus . Can. J. Fish. Aquat. Sci. 49:795-804. Smith, M. K., and E. Kostlan. 1991. Estimates of age and growth of ehu Etelis carbuncu- lus in four regions of the Pacific from density of daily incre- ments in otoliths. Fish. Bull. 89:461-472. Smith, W. G„ and W. W. Morse. 1985. Retention of larval haddock Melanogrammus aeglefi- nus in the Georges Bank region, a gyre-influenced spawn- ing area. Mar. Ecol. Prog. Ser. 24:1-13. Sokal, R. R., and F. J. Rohlf. 1995. Biometry, third ed. Freeman and Company, New York, NY, 887 p. SPSS, Inc. 1997. SYSTAT 7.0 statistics. SPSS Incorporated, Chicago, IL, 751 p. Stephenson, R. L. 1999. Stock complexity in fisheries management: a perspec- tive of emerging issues related to population sub-units. Fish. Res. 43:247-249. Van Eeckhaute, L. A. M., S. Gavaris, and E. A. Trippel. 1999. Movements of haddock, Melanogrammus aeglefinus, on eastern Georges Bank determined from a population model incorporating temporal and spatial detail. Fish. Bull. 97:661-679. 15 Abstract— Reproductive biology of the cobia, Rachycentron Canadian , is de- scribed from four coastal areas in the southern United States. Samples were obtained from recreational fishermen between December 1995 and Novem- ber 1997 from the southeastern United States (Morehead City, NC, to Cape Canaveral, FL), the eastern Gulf of Mexico (Ft. Myers to Crystal River, FL), the north-central Gulf of Mexico (Destin, FL, to Chandeleur Islands, LA) and the western Gulf of Mexico (Port Aransas, TX). Histological evidence of spawning occurred from April through September in all areas. Some female cobia (17-32%) throughout the Gulf of Mexico had spent or regressed ovaries by July. Gonadosomatic index peaked between May and July throughout the region. Ovaries offemales from all areas contained both postovulatory follicles (POF) and oocytes in final oocyte matu- ration (FOM) during all months of the reproductive season. Batch fecundity was calculated by using three different methods: oocytes >700 pm were fixed in 1) Gilson’s fixative or 2) 10%- neutral buffered formalin (NBF), and 3) oocytes undergoing FOM were sectioned for histological examination. Mean batch fecundity ranged from 377,000 ±64,500 to 1,980,500 ±1,598,500 eggs; there was no significant difference among meth- ods. Batch fecundity calculated with the NBF method showed a positive relationship with fork length (P=0.021, r2=0.132) and ovary-free body weight (OFBW; P=0.016, r2=0.143). Relative batch fecundity was not significantly different among months during the spawning season and averaged 53.1 ±9.4 eggs/g OFBW for the NBF method and 29.1 ±4.8 eggs/g OFBW for the FOM method. Although spawning fre- quencies were not significantly differ- ent among areas (P=0.07), cobia from the southeastern United States and north-central Gulf of Mexico were esti- mated to spawn once every 5 days, whereas cobia from the western Gulf of Mexico were estimated to spawn once every 9 to 12 days. Manuscript accepted 22 August 2000. Fish. Bull. 99:15-28 (2001). Reproductive biology of cobia, Rachycentron canadum, from coastal waters of the southern United States Nancy J. Brown-Peterson Robin M. Overstreet Jeffrey M. Lotz Department of Coastal Sciences Institute of Marine Sciences The University of Southern Mississippi 703 East Beach Drive Ocean Springs, Mississippi 39564 E-mail address (for N. J. Brown-Peterson): Nancy.Brown-peterson@usm.edu James S. Franks Center for Fisheries Research and Development Institute of Marine Sciences The University of Southern Mississippi 703 East Beach Drive Ocean Springs, Mississippi 39564 Karen M. Bums Center for Fisheries Enhancement Mote Marine Laboratory 1600 Thompson Parkway Sarasota, Flordia 34236 The cobia, Rachycentron canadum (Goode, 1884), is a large coastal pelagic fish found worldwide in tropical and subtropical waters with the exception of the eastern Pacific (Shaffer and Nakamura, 1989). The cobia is a highly prized recrea- tional species and record size fish have been caught in coastal waters off the southern United States, as well as off Western Australia, Nigeria, and Kenya (International Game Fish Association, 1998). Most specimens captured in the southern United States are landed by recreational anglers along the south- eastern U.S. coast and in the Gulf of Mexico; however, some are caught inci- dentally by U.S. commercial fisheries, particularly in Florida waters (Shaffer and Nakamura, 1989). Cobia are specifically targeted by a growing number of anglers in the southern United States, as evidenced by the increase in fishing tournaments for cobia. Information on age, growth, and seasonal movement is being collect- ed through tagging programs in Virgin- ia, Florida, Mississippi, and Louisiana (International Game Fish Association, 1998), and the age and growth of cobia from the north-central Gulf of Mexico was recently described by Franks et al. (1999). Limited information on the reproductive biology of cobia from the southern United States includes de- scriptions of the eggs, larvae, and ju- veniles from Chesapeake Bay (Joseph et al., 1964), North Carolina (Hassler and Rainville, 1975), and the northern Gulf of Mexico (Ditty and Shaw, 1992). Summer spawning of cobia has been re- ported from Chesapeake Bay (Richards, 1967), North Carolina (Smith, 1995), the northern Gulf of Mexico (Dawson, 1971; Burns et al.1), Louisiana (Thomp- 1 Burns, K. M., C. Neidig, J. Lotz, and R. Overstreet. 1998. Cobia ( Rachycen- tron canadum ) stock assessment study in the Gulf of Mexico and in the South Atlan- tic. Final Rep., MARFIN Coop. Agree- ment NA57FF0294 to NMFS (NOAA), 108 p. Mote Marine Laboratory, 1600 Thomp- son Parkway, Sarasota, FL 34236. 16 Fishery Bulletin 99(1) Figure 1 Areas sampled for cobia within the southern United States. SEUS = south- eastern United States; EGOM = eastern Gulf of Mexico; NCGOM = northcen- tral Gulf of Mexico; WGOM = western Gulf of Mexico. son et al.2) and Texas (Finucane et al.3). Biesiot et al. (1994) described the biochemical changes in developing ovaries of cobia from the north- ern Gulf of Mexico and reported that spawning occurred during spring and summer. The most comprehensive information on cobia reproduc- tion to date provides a detailed description of the gonadal maturation and spawning season for cobia from the north-central Gulf of Mexico (Lotz et ah, 1996) and furnishes evidence that cobia are multiple, or batch spawners. Lotz et al. (1996) estimated batch fecundity on the ba- sis of the largest mode of oocytes, but they did not estimate spawning frequency. Our study was undertaken to document more thoroughly the reproductive biology of cobia from the southern region of the United States. Specifically, we describe and compare the spawning seasons and gonadal develop- ment from four coastal areas: the southeast- ern United States, the eastern Gulf of Mexico, the north-central Gulf of Mexico, and the west- ern Gulf of Mexico. Additionally, batch fecun- dity and spawning frequency are estimated for cobia from the region. The results are dis- cussed in light of the migratory nature of cobia throughout coastal waters of the southern United States. Materials and methods Sample collection Cobia were sampled from the coastal waters of four gen- eral regions in the southern United States (Fig. 1). The regions were defined as the southeastern United States (SEUS; Morehead City, North Carolina, to Cape Canaveral, Florida), the eastern Gulf of Mexico (EGOM; Ft. Myers to Crystal River, Florida), the north-central Gulf of Mexico (NCGOM; Destin, Florida to the Chandeleur Islands, Loui- siana), and the western Gulf of Mexico (WGOM; Port Aran- sas area, Texas). In our study, coastal waters are defined as those extending over the continental shelf for 20 km in the Atlantic Ocean and for 80 km in the Gulf of Mexico. We sampled cobia opportunistically from the recreation- al and charter boat fisheries from December 1995 to De- cember 1997. Additional samples were taken in the north - 2 Thompson, B. A., C. A. Wilson, J. H. Render, M. Beasley, and C. Cauthron. 1992. Age, growth and reproductive biology of greater amberjack and cobia from Louisiana waters. Final Rep., MARFIN Coop. Agreement NA90AA-H-MF722 to NMFS (NOAA), 77 p. Coastal Fisheries Institute, LSU Center for Coastal, Energy and Environmental Resources, Baton Rouge, LA 70803. 3 Finucane, J. H., L. A. Collins and L. E. Barger. 1978. Ichthyoplankton/mackerel eggs and larvae. Environmental stud- ies of the south Texas outer continental shelf, 1977. Final rep. to Bur. Land Manage. Natl. Mar. Fish. Serv., NOAA, Galveston, TX 77550. central Gulf of Mexico during February and March 1999. Sampling teams were present at major cobia fishing tour- naments throughout the study area; most samples from the SEUS and the NCGOM came from fishing tourna- ments. The majority of the cobia sampled from the EGOM were captured by nontournament recreational anglers. All fish from Texas were obtained from one charter boat cap- tain during regular fishing trips. Anglers were interviewed to determine the location of capture of each fish. Fork length (FL, cm) and total weight (TW, g) were recorded at the dock and gonads were excised and placed on ice for transport to the laboratory. In the laboratory, gonads were weighed to the nearest 0.1 g (gonadal weight [GW]) and the gonadosomatic index (GSI) was calculated by using the formula GSI = [ GWKTW-GW ) x 100], Fish weights were unavailable from the WGOM; hence GSIs were not calculated for this region. Sections were removed from both left and right gonads and preserved in 10% neutral buffered formalin (NBF) for histological analysis. Cobia have been shown previously to have homo- geneous oocyte development within the ovary (Lotz et al., 1996); thus, multiple sections of the same ovary were not removed for analysis. For fecundity analysis, two por- tions (approximately 5 g each) of the ovary were removed, weighed to the nearest 0.1 g, and preserved in Gilson’s fix- ative (GF) and 10% NBF, respectively. Histological analysis Tissues were placed into individually labeled cassettes and rinsed with running tap water overnight prior to de- Brown-Peterson et al.: Reproductive biology of Rachycentron cancidum 17 hydration and embedment in paraffin, following standard histological techniques. The paraffin blocks were sectioned at 4 pm by using a rotary microtome. Duplicate slides were prepared for each tissue, resulting in a total of four slides for each cobia specimen (two from each gonad). The slides were stained with Gill’s I hematoxylin and eosin phloxine (Polyscientffic Corporation) following standard histological procedures. Three separate views from each side of the gonad of each fish were examined to determine maturity stages. Ovarian maturity classes were based on those previously described for cobia by Lotz et al. (1996). The entire ovar- ian section was examined for the presence of postovulato- ry follicles (POF) and oocytes undergoing final oocyte mat- uration (FOM). POF stages were classified following the methods of Hunter et al. (1986), although age estimates for POF stages in cobia are unverified. FOM stages were classified following Brown-Peterson et al. (1988). Follow- ing inspection of the entire ovarian section, three areas were arbitrarily selected from each slide for quantification of oocytes. Oocytes in all stages of development (including atretic oocytes) and POFs were counted at lOOx and the percentage of each oocyte stage in the field of view was es- timated. Oocyte atresia stages were classified by following the methods of Hunter and Macewicz (1985a). The entire testicular section from each cobia was exam- ined to determine the maturity classification for male fish. Three arbitrarily selected portions of each section were ex- amined at lOOx and 400x to classify all stages of spermato- genesis observed. Testicular maturity stages were based on those described for cobia by Lotz et al. (1996). Partic- ular attention was given to the presence and amount of spermatogenesis in the testis. Estimates of batch fecundity Batch fecundity was estimated from the counts of oocytes in samples of ovarian tissue. Oocyte counts were obtained after teasing oocytes from tissues fixed in either GF or NBF for three to four months or from histological evalua- tion of tissue sections. The volumetric method was used to estimate fecundity for tissues fixed in GF or NBF (Bage- nal and Brauin, 1971). All oocytes freed from each GF or NBF sample were placed in 50 mL of water, stirred to homogeneity, and ten 1-mL samples were removed, combined, and the total volume brought to 50-70 mL with water. The diluted sample was stirred to homoge- neity and 1-mL subsamples were removed, counted and replaced three to six times for each ovarian sample. All oocytes >700 pm from each subsample were counted and measured by using a stereo dissecting microscope and a computerized image analysis system. Oocytes of this size were used because Lotz et al. (1996) previously showed that cobia have a distinct mode of large oocytes prior to spawning. Typically, 25-100 oocytes were measured and counted in each subsample. Estimates of batch fecundity based on histological eval- uation were obtained by counting the number of oocytes undergoing FOM in six fields under a compound micro- scope at lOOx magnification. The area of a single field of view was determined to be 0.0249 cm2 by using a stage micrometer. The number of oocytes in final maturation ob- served in a field of view was converted to the number per mL by the formula N x 0.02493/2 = Nx 0.003939. FOMs were counted as 1 if >50% of the oocyte was in the field of view and were uncounted if <50% of the oocyte was in the field of view. The total number of FOMs in a fish was then determined by multiplying the estimated number per mL by the total volume of the ovaries. In all cases, fecundity was expressed as both batch fecundity (mean number of eggs/batch) and relative fecundity (number of eggs/gram of ovary-free body weight). Ovarian volume was determined by volumetric displace- ment. The observed relationship between ovarian weight and ovarian volume was determined and that relationship was used to estimate ovarian volumes of fish for which di- rect volume measurements were unavailable. The analy- sis was restricted to fish with ovarian weights >500g. Estimates of spawning frequency Two methods based on histological observations were used to estimate spawning frequency of cobia: 1) the percentage of females in the late developing ovarian class with 0- to 24-h POF in the ovary and 2) the percentage of females in the late developing ovarian class undergoing FOM. Only fish in the late developing ovarian class were included in these analyses, because this is the only class in which cobia have the potential to spawn. For both methods, esti- mates of spawning frequency were determined according to the procedure of Hunter and Macewicz ( 1985b). The per- centage of fish in the late developing maturity stage with ovaries containing either FOMs or POFs was calculated for each month in each region. This value represents the percentage of the fish in the population that are about to spawn (FOMs) or have just spawned (POFs). Spawning frequency (the number of days between spawnings) was determined by dividing 100 (representing the total popu- lation of fish) by the percentage of fish with FOMs or POFs in the ovaries. Statistical analysis Student’s t-test was used to test for differences in GSI values between years. Batch fecundity data were tested for normality and homogeneity of variance. Simple linear regression was used to test the relationship between batch fecundity as the dependent variable and FL or ovary-free body weight as the independent variable. One way analy- sis of variance was determined for relative batch fecundity. A Mann-Whitney U test was used to compare fecundity estimates for the various methods used to determine fecundity. A chi-square test was used to test for differences in spawning frequency among areas. All statistics were computed by using SPSS-PC version 7.5 (SPSS, Inc., 1997) or Systat 8.0 (SPSS, Inc., 1998). Results were considered significant if P < 0.05. 18 Fishery Bulletin 99(1 ) Table 1 Numbers of cobia examined histologically from each sampling area. Area Males n Months of capture Females n Months of capture Southeastern United States (SEUS) 33 February-May 60 February-June Eastern Gulf of Mexico (EGOM) 43 February-December 60 March-December North-central Gulf of Mexico (NCGOM) 48 March-October 204 February-September Western Gulf of Mexico (WGOM) 23 May-August 59 May-August Totals 147 383 Results Fish collections A total of 530 cobia (147 males, 383 females) from the southern United States were collected for histo- logical analysis. The months and numbers of sam- ples collected varied by region, but in all regions fish were collected primarily during the reproduc- tive season (Table 1). Specimens ranged from 35.5 to 138.5 cm FL for females and 36.5 to 127.0 cm FL for males. Weights ranged from 0.64 to 34.93 kg for females and 0.91 to 40.82 kg for males. Accurate estimates of the size and age at sexual ma- turity of cobia from the southern United States could not be determined in this study. Small, immature spec- imens were rare in recreational catches owing to a minimum retention size of 84 cm FL for cobia in state territorial waters and the EEZ. We collected only six sexually immature specimens (all females from the EGOM ) during this study. The smallest reproductively active female encountered was 70 cm FL. Spawning season and gonadal development Cobia have a protracted spawning season (April through September) throughout the southern United States as determined from GSI values and histo- logical assessments. There was no significant differ- ence (P>0.05) in GSI values between corresponding months in 1996 and 1997 for either males or females in any region, with the exception of males in Septem- ber from the NCGOM (P=0.049). Therefore, monthly data for 1996 and 1997 by region were combined (Fig. 2). GSI values for both sexes in SEUS increased sharply from April to May (Fig. 2A), indicating the onset of the reproductive season. GSI values for both sexes of cobia from EGOM began to increase in March, peaked in July, and declined and leveled off thereafter (Fig. 2B). GSI values for females from NCGOM increased in March, peaked in May, and then declined through September (Fig. 2C). In contrast, GSI values of males from NCGOM steadily increased through July, then fell precipitously in August (Fig. 20. GSI values for males reached similar mean maxima in 8 7 6 5 4 3 2 1 A A Female ■ Male f A ' // ■ ■: . t ; 7 6 B 5 '•3' ISS 3 ▲ 2 A t 1 ■ A * ■ ■ i - A A . ' • ■ 6 5 c 4 ▲ * i 3 A A • A 2 1 * A ‘ , • ■ J FMAMJ JASOND Month Figure 2 Monthly (1996 and 1997 combined) gonadosomatic index (GSI) values for cobia from the southern United States. Values rep- resent mean ±1SE. (solid triangles=female, solid squares=male) (A) Southeastern United States. (B) Eastern Gulf of Mexico. (C) North-central Gulf of Mexico. the SEUS and the NCGOM regions but were lower in the EGOM. However, mean GSI values of females were higher in both May and June for cobia from the SEUS than during any month from the Gulf of Mexico (Fig. 2). Brown-Peterson et al.: Reproductive biology of Rachycentron canadum 19 Table 2 Percentage of cobia in each ovarian maturity class for SEUS (Morehead City, NC, to Cape Canaveral, FL). Monthly data from 1996 and 1997 were combined. Percentage atresia was calculated for each development stage for ovaries with alpha- or beta-stage atresia only. Month of capture Class February (n= 3) March (n=31) April (22=10) May (>2 = 10) June (n= 6) Early developing 66 41.9 20 0 0 % atresia 100 92 0 — — Mid-developing 34 16 10 0 0 % atresia 100 80 100 — — Late developing 0 41.9 70 100 100 % atresia — 85 66 60 0 Spent 0 0 0 0 0 % atresia — — — — — Regressed 0 3.2 0 0 0 % atresia — 0 — — — Table 3 Percentage of cobia in each ovarian maturity class for EGOM (Crystal River to Ft. Meyers, FL). Monthly data from 1996 and 1997 combined. Percentage atresia was calculated for each development stage for ovaries with alpha- or beta-stage atresia only. Class Month of capture March (n= 2) April (n= 7) May (n= 3) June (22= 2) July (22=6) August (22=6) September (22=3) October (21=11) November (22 = 13) December (22=7) Immature 100 0 0 50 0 16 0 18 7 0 Early developing 0 29 0 0 0 0 0 0 0 0 % atresia — 50 — — — — — — — — Mid-developing 0 14 66 0 0 0 0 0 0 0 % atresia — 0 50 — — — — — — — Late developing 0 57 33 50 83 67 33 0 0 0 % atresia — 50 0 0 0 50 0 — — — Spent 0 0 0 0 17 0 0 18 8 14 % atresia — — — — 0 — — 100 100 100 Regressed 0 0 0 0 0 17 66 64 85 86 % atresia — — — — — 100 100 86 45 83 Histological analysis showed that all males from all ar- eas were ripe during all months. Spermatogenic activity varied over the reproductive season, but males captured during February-May exhibited active spermatogenesis throughout the testis. No spermatogenesis occurred dur- ing August and September, but the testis contained sper- matozoa. Males from EGOM during October through De- cember had spermatozoa in the testis, although 50% or more of the males in November and December had the tes- tis classified as spent. Histological examination of ovaries revealed all classes of maturity, from early developing through regressed. Dif- ferences in monthly ovarian maturity between correspond- ing months in 1996 and 1997 were minimal; thus, month- ly data were combined (Tables 2-5). A majority of ovaries from female cobia were in the late developing class by March in the NCGOM (Table 4) and by April in SEUS (Table 2) and EGOM (Table 3). Data from the WGOM (Ta- ble 5) were limited to only a portion of the reproductive season. Ovarian recrudescence began in February in the 20 Fishery Bulletin 99(1) Table 4 Percentage of cobia in each ovarian maturity class for NCGOM (Destin, FL, to Chandelier Islands, LA). Monthly data from 1996 and 1997 were combined. Percentage atresia was calculated for each developmental stage for ovaries with alpha- or beta-stage atresia only. Month of capture Class February 0=10) March 0 = 6) April (rc=20) May 0=112) June 0=1) July 0=25) August 0=8) September 0=22) Early developing 60 0 0 4.5 0 0 0 0 % atresia 60 — — 20 — — — — Mid developing 0 33 10 7 0 4 0 0 % atresia — 100 100 62 — 0 — — Late developing 0 67 90 88.5 100 64 37.5 18 % atresia — 75 44 38 0 31 33 25 Spent 10 0 0 0 0 28 50 23 % atresia too — — — — 86 100 80 Regressed 30 0 0 0 0 4 12.5 59 % atresia 0 — — — — 0 0 100 Table 5 Percentage of cobia in each ovarian maturity class for WGOM (Port Aransas, TX) in 1996. Percentage atresia was calculated for each developmental stage for ovaries with alpha- or beta-stage atresia only. Class Month of capture May 0=1) June 0=8) July 0=48) August 0=2) Mid developing 0 0 4 50 % atresia — — 50 0 Late developing 100 100 73 50 % atresia 0 0 5.7 0 Spent 0 0 21 0 % atresia — — 100 — Regressed 0 0 2 0 % atresia — — 100 — SEUS and the NCGOM, and females in the late develop- ing class occurred in both areas by March (Tables 2 and 4). Spent females were initially observed in July in the Gulf of Mexico (Tables 3-5), although some females remained in the late developing class through September (Tables 3 and 4). Some females in the regressed class occurred in July throughout the Gulf of Mexico (Tables 3-5). In July, lengths of spent and regressed fish ranged from 88.0 to 93.0 cm FL in the EGOM, from 85.5 to 102.1 cm FL in the NCGOM, and from 86.4 to 128.3 cm FL in the WGOM. Ovarian tissue in all classes of maturity showed atre- sia throughout the reproductive season. Alpha- and beta- stage atresia of yolked oocytes (Fig. 3A) was most preva- lent in spent fish, but also occurred in females in the early, mid and late developing ovarian classes (Tables 2-5). Atre- sia of hydrated oocytes (Fig. 3B) occurred in ovaries in the late developing and spent classes only (5-10%). Atresia of nonyolked oocytes was difficult to recognize but was com- mon (50-90%) in early developing ovaries. The later stag- es of atresia (gamma and delta, Fig. 30 occurred in ova- ries in all maturity classes. Many late developing females (60-85%) from the SEUS exhibited alpha- or beta-stage atresia during March through May (Table 2). Similarly, fe- males from NCGOM exhibited high levels of atresia (75%) in ovaries in the late developing class during March (Table 4). Less atresia (38-50%) occurred during April and May in the ovaries of cobia from the Gulf of Mexico (Tables 3 and 4) as compared with cobia from the SEUS (Table 2). Female cobia underwent final oocyte maturation (FOM) in all four areas sampled during April through September. The early stages of FOM were characterized by early lip- id coalescence (Fig. 4A), followed by complete lipid coales- cence and migration of the nucleus to the periphery of the oocyte (Fig. 4B). The final stages of FOM, characterized by breakdown of the nuclear membrane, yolk coalescence and hydration, were not observed in any sample. Final oo- cyte maturation was a synchronous process and most oo- cytes within an ovary were in the same stage of FOM (Fig. 4C). The percentage of mature oocytes (>600 pm) undergoing FOM varied from 5% to 84% of the oocytes within a lOOx microscopic field of view for all ovaries ex- amined. Females from the EGOM had the highest mean percentage of mature oocytes undergoing FOM (49%). The mean percentage of mature oocytes undergoing FOM was lower, but similar, among the other regions (NCGOM, 19%; SEUS, 16%; WGOM, 15%). The mean percentage of females in the late-developing ovarian class undergoing Brown-Peterson et at: Reproductive biology of Rcichycentron canadum 21 FOM ranged from 11% in the WGOM to 59% in the EGOM. On average, 19% of the females from SEUS and NCGOM were un- dergoing FOM. The presence of POFs in cobia ovaries in- dicated spawning had commenced, although POFs were uncommonly observed; the great- est density of POF recorded was five per lOOx microscopic field of view. Postovulatory folli- cles occurred in cobia ovaries in the late-de- veloping class from all regions during April through September, suggesting that although cobia were in the late-developing class by March, spawning commenced in April. The 0- tol2-h POF stage (Fig. 5A) was infrequently observed (16%) and was absent in females from the WGOM. The 24-h POF stage (Fig. 5B) was the most frequently observed (51%) and was most common in cobia from EGOM and NCGOM. The 48-h POF stage (33%, Fig. 5C) was difficult to distinguish from gamma- and delta-stage atresia, and most commonly occurred in cobia from theWGOM. Spawning frequency Estimates of monthly spawning frequency for the POF and FOM methods (Table 6) were consistent throughout the spawning season in all regions. Both methods showed good agreement for fish from SEUS and NCGOM. The POF method indicated a more frequent estimate of spawning rate in the NCGOM. On the other hand, the FOM method resulted in a more frequent spawn- ing rate for cobia from the WGOM. Cobia from SEUS and NCGOM were estimated to spawn every 4 to 5 days, whereas those from WGOM spawned every 9 to 12 days. Chi- square analysis showed no significant differ- ence in spawning frequency estimates for the three regions for either the POF (P=0.08) or the FOM (P=0.409) method. Months where fewer than five specimens had late develop- ing ovaries were eliminated from the anal- ysis because spawning frequency estimates would probably be inaccurate. No estimate was made for cobia from the EGOM, and spawning frequencies for the WGOM were based on fish captured in July. Spawning frequency estimates for SEUS were based on data from April, May, and June, whereas estimates for NCGOM were based on data from April, May, and July. Relationship of ovary weight and volume Direct volume measurements were per- formed for 86 females with ovarian weights Figure 3 Stages of atresia in cobia ovaries. (A) Alpha (a), beta (j3), and gamma (y) stage atresia in a late-developing or spent ovary. Scale bar = 0.1 mm. (B) Alpha-stage atresia of a hydrated oocyte (0). Scale bar = 0.1 mm. (C) Beta ( /3), gamma (y) and delta (<5) stages of atresia in a spent ovary. Scale bar = 0.1 mm. 22 Fishery Bulletin 99(1 ) Figure 4 Final oocyte maturation (FOM) in late-developing cobia ovaries. (A) Early stage of FOM showing initial lipid coalescence (L). Scale bar = 0.1 mm. (B) More advanced stage of FOM. Lipids have coalesced to form a single large droplet (L) and the nucleus (N) is beginning to migrate to the periphery of the oocyte. Scale bar = 0.1 mm. (C) Oocytes undergoing syn- chronous FOM. Scale bar = 0.2 mm. >500 g. The observed relationship was linear, and the best fit equation as judged by the least squares criterion was mL = -8.54 + 0.96g (r2=0.978). The relationship indicated that cobia ovaries from 500 to 1600 g were less dense than seawater and that ovary den- sity remained constant over that range. Batch fecundity Batch fecundity estimates were compared for samples from 11 females with ovaries fixed in GF, for 40 females with ovaries fixed in NBF, and for 26 ovarian samples examined by the histological method and where oocytes were undergoing FOM from April through Septem- ber. Samples of this type were limited; there- fore we combined observations from SEUS, EGOM, and NCGOM. Fecundity estimates for individual fish varied widely; however, there was no significant difference in mean estimates among the three methods (Mann- Whitney [/-test, P>0. 05; Table 7). Mean batch fecundity ranged from 377,000 ±64,500 eggs (CV=2.677) with the histological method to 1,980,500 ±1,598,500 eggs (CV=0.875) with the GF method. Batch fecundity estimates for all three methods showed substantial variation. A Kol- mogorov-Smirnov test of normality showed that batch fecundity was normally distribut- ed for NBF samples (40 df, P>0.05) and FOM samples (26 df, P>0.05) and could be ana- lyzed by using parametric statistics. Batch fecundity determined with the GF method was not used for further analyses owing to the small sample size. Regression analysis showed a significant, positive relationship be- tween batch fecundity (BF) and FL (P=0.021, r2=0.132) and BF and ovary-free body weight (OFBW; P=0.016, r2=0.143) for NBF samples. The relationship between BF and OFBW for NBF samples (Fig. 6A) was described by BF = OFBW1 717 -36.813. There was no significant relationship be- tween BF and FL (P=0.105) or OFBW (P=0.097) for FOM samples. The relationship between BF and OFBW for FOM samples (Fig. 6B) was described by BF = 19.29 x OFBW + 1,113,713 [r2=0.11, P=0.097], Mean batch fecundity, as determined from the NBF method, averaged 854,100 ±166,200 eggs and ranged from 247,100 ±204,400 eggs in August (n= 2) to 923,000 ±237,700 eggs in May (n= 26). Mean batch fecundity, as Brown-Peterson et al.: Reproductive biology of Rachycentron canadum 23 determined with the FOM method, aver- aged 377,000 ±64,500 eggs and ranged from 212,500 ±122,700 eggs in August (n= 5) to 637,000 ±376,600 eggs in September (n-3) . Relative batch fecundity (Table 8) did not vary significantly from April through Sep- tember as determined by the NBF method (F=0.636, df=37, P=0.639) and the FOM method (F=0.468, df=24, P=0.759). Relative batch fecundity for the NBF method aver- aged 53.1 ±9.4 eggs/g ovary-free body weight throughout the reproductive season (77=39, Table 8) and it was lowest in August and highest in June. Relative fecundity values determined with the FOM method were low- er than those with the NBF method, averag- ing 29.1 ±4.8 eggs/g ovary-free weight (77 =25, Table 8), and were lowest in June and high- est in July. Potential annual fecundity for cobia was estimated from batch fecundity and spawn- ing frequency estimates. A female cobia weighing 20 kg from SEUS or NCGOM may potentially spawn 20,952,000 (FOM meth- od) to 38,232,000 (NBF method) eggs be- tween April and September. In contrast, the same size female from WGOM would po- tentially spawn 8,730,000 (FOM method) to 21,240,000 eggs (NBF method) between April and September, with the data provided here. Discussion Our data show that the reproductive biol- ogy of cobia is similar throughout the coastal waters of the southern United States. Although sample sizes from some regions and months were small due to reliance on recreational catches for samples, we feel the data adequately represent the reproductive population from the four regions. Spawning commences in April throughout the region, as evidenced by the presence of oocytes undergoing FOM as well as 24-h POFs. These findings are in agreement with pre- vious studies of cobia reproduction in the southeastern U.S. Atlantic Ocean (Smith, 1995) and the north-central Gulf of Mexico (Biesiot et al., 1994; Lotz et al., 1996; Thomp- son et al.2). Collections of larval cobia from the Gulf of Mexico during May through Sep- tember (Ditty and Shaw, 1992) also confirm the spawning season. Reproductive activity of female cobia probably ceases during Sep- tember in the Gulf of Mexico and extends at least through June in the SEUS. Smith (1995) reported that cobia from North Caro- lina spawned through July. Eggs and larvae Figure 5 Postovulatory follicles (POF) in late-developing cobia ovaries. (A) 0- to 12-h POF (arrows). Scale bar = 0.1 mm. (B) 24-h POF (arrows). Scale bar = 0.2 mm. (C) 48-h POF (arrow). Scale bar = 0.1 mm. 24 Fishery Bulletin 99(1 ) Table 6 Mean estimated spawning frequencies of cobia from three regions in the southern United States. Spawning frequencies are esti- mated from the percentage of ovaries in the late developing ovarian class containing either postovulatory follicles (POF) or under- going final oocyte maturation (FOM). Spawning frequency estimates were based on data from April to June in SEUS, from April, May, and July in NCGOM, and during July in WGOM. Spawning frequency Region Southeastern United States (SEUS) (rc=23) Northcentral Gulf of Mexico (NCGOM) (ti=135) Western Gulf of Mexico (WGOM) (rz =35 ) % POFs 19.4 24.8 8.1 Frequency (POFs) 5.2 days 4.0 days 12.3 days % FOM 19.4 19.8 10.8 Frequency (FOM) 5.2 days 5.0 days 9.2 days Ovary-free body weight (kg) Figure 6 Relationship between batch fecundity (BF) and ovary- free body weight ( OFB W ) for cobia from the southern United States. Cobia were captured from April through September of 1996 and 1997 from the southeastern United States, the eastern Gulf of Mexico, and the north-central Gulf of Mexico. (A) Batch fecundity determined from formalin-fixed oocytes >700 pm.BF = OFBW 1 717 - 36.813. (B) Batch fecundity determined from histological sections of oocytes undergoing final oocyte maturation. BF = 19.290 x OFBW + 1,113,713. Table 7 Batch fecundity estimates (no. of eggs) of cobia from the southern United States determined with three different methods. All means were not statistically different (Mann- Whitney (/-test, P>0.05). NBF = neutral buffered formalin. FOM = final oocyte maturation. Fecundity method Measurement Gilson’s fixative (77 = 11) 10% NBF (72=40) Histology (FOM) (77=26) Percentage of ovary counted 0.02 0.02 0.005 Mean number of eggs 1,980,500 854,000 377,000 Standard error 1,598,500 166,200 64,500 Coefficient of variation 2.677 1.246 0.873 Minimum number of eggs 2,700 8,000 22,900 Maximum number of eggs 17,848,800 5,132,000 1,390,000 of cobia were collected from the Chesapeake Bay from mid- June through mid-August (Joseph et ah, 1964), whereas cobia eggs from North and South Carolina were collected from mid-May through the end of August (Hassler and Rainville, 1975; Shaffer and Nakamura, 1989). Gonadosomatic index values are indicators of the dura- tion of the reproductive season for cobia, and they correlat- ed well with our histological findings. However, Jons and Miranda (1997) advised caution in their use because of re- gional and temporal variations in GSI values. Therefore, GSI values should not be used for comparing or indexing Brown-Peterson et al.: Reproductive biology of Rcichycentron canodum 25 Table 8 Monthly mean relative batch fecundity expressed as number of eggs/g ovary-free body weight for cobia in the southern United States. Batch fecundity was determined from oocytes >700 pm in neutral buffered formalin (NBF) and from histological sections of oocytes undergoing final oocyte maturation (FOM). All means were not statistically different (AN OVA, P>0.05). Month NBF FOM n Relative fecundity ±1SE n Relative fecundity ±1SE April i 46.6 0 — May 25 51.2 ±12.2 8 24.6 ±4.9 June 2 115.7 ±102.2 3 21.9 ±12.9 July 5 44.0 ±9.2 6 40.2 ±12.9 August 2 29.8 ±25.2 5 24.6 ±14.9 September 4 57.7 ±27.5 3 33.2 ±13.0 Overall 39 53.1 ±9.4 25 29.1 ±4.8 maturity stages, particularly in multiple spawning fish. The GSI profile for both male and female cobia from the SEUS is similar to that described by Smith (1995) for co- bia from North Carolina. On the other hand, GSI peaks for cobia from the northern Gulf of Mexico vary among stud- ies. Biesiot et al. (1994) reported that female GSI values peaked in April, Lotz et al. (1996) found peak female GSI values in May, and Thompson et al.2 reported peak female GSI values in June. Our data for females mirror those pre- sented by Lotz et al. (1996); however, we suspect annual differences. Our study was not designed to determine the size or age at first maturity for cobia. However, our limited data sug- gest that both sexes of cobia may achieve sexual maturity at a smaller size than that reported by Lotz et al. (1996) for the north-central Gulf of Mexico. This apparent differ- ence in the size at sexual maturity could be partially ex- plained by regional differences; most small female cobia in the present study ( <85 cm FL) were captured in EGOM, whereas Lotz et al. (1996) sampled in NCGOM. Male cobia are probably capable of spawning through- out the year because of the presence of sperm in the tes- tis. A more detailed analysis of the histological pattern and spermatogenesis of male cobia will be discussed sepa- rately (Brown-Peterson et al.4). A similar longer reproduc- tive season for male fish has been reported for other spe- cies with protracted spawning seasons including common snook ( Centropomus undecimalis, Grier and Taylor, 1998; 4 Brown-Peterson, N. J., H. J. Grier and R. M. Overstreet. 2000. Manuscript in preparation. Reproductive classes in male cobia (Rcichycentron canadum ) defined by changes in the germinal epithelium. Abstract and presentation at the 80th annual meet- ing of the American Society of Ichthyologists and Herpetolo- gists, June 2000, La Paz, B.C.S., Mexico. Taylor et al., 1998), spotted seatrout ( Cynoscion nebulosus, Brown-Peterson et al., 1988), red drum ( Sciaenops ocella- tus, Grier et al., 1987), and blue tilapia ( Oreochromis au- reus, Grier and Abraham, 1983). Cobia have a protracted spawning season, yet a portion of the females in the population may spawn during April- June only. Other females remain in spawning condition throughout September in the Gulf of Mexico. Lotz et al. (1996) and Biesiot et al. (1994) reported a similar occur- rence in cobia from the NCGOM, and a high percentage of female cobia off Louisiana are spent and regressed by Ju- ly (Thompson5). A comparable phenomenon, i.e. asynchro- nous cessation of spawning, was reported in the weakfish (C. regalis ) in Chesapeake Bay (Lowerre-Barbieri et al., 1996). It is difficult to explain early cessation of spawning by some female cobia. Spent and regressed fish in July and August had a broad length distribution (85 tol28 cm FL), suggesting that multiple age classes in the fishery have an abbreviated reproductive season. Perhaps some females delay ovarian maturation and spawn between July and September; this theory would account for the small per- centage of females in the early- and mid-developing classes in May and June. Differences in the amount of oocyte atresia during the spawning season between cobia from the SEUS and the Gulf of Mexico suggest differential spawning success dur- ing the early portion of the reproductive season. High per- centages of alpha and beta atresia in cobia in the late-de- veloping class from SEUS during March, April, and May suggest that many oocytes do not reach final maturation and that the atresia may be related to variable or unfa- vorable environmental conditions during spring (March- May) in the region. Hay and Brett ( 1988) showed a similar occurrence for Pacific herring ( Clupea harengus pallasi) of increased atresia at the beginning of the reproductive season — condition they attributed to environmental fac- tors rather than the female’s physiological ability. Low- erre-Barbieri et al. (1996) used similar reasoning to ex- plain the increased percentage of alpha and beta atresia present in the ovaries of weakfish captured during the be- ginning of the spawning season in Chesapeake Bay. Co- bia from NCGOM also showed high percentages of atresia during March, the beginning of the reproductive season but prior to the initiation of spawning in that area. The lower percentage of females in the late-developing class with atretic oocytes in the Gulf of Mexico during April and May suggests a relatively high spawning success during the early portion of the spawning season in the Gulf of Mexico, and it also may suggest more stable environmen- tal conditions in this region during late spring. In the absence of hydrated oocytes in any cobia that we or Lotz et al. (1996) examined, we used three different methods to estimate fecundity of cobia containing large (>700 pm) oocytes. The wide range of results found among methods highlights the variations to be expected when es- timating batch fecundity of a multiple spawning species. 5 Thompson, B. A. 1999. Personal commun. Coastal Fisheries Institute, LSU Center for Coastal, Energy and Environmental Resources, Baton Rouge, LA 70803. 26 Fishery Bulletin 99(1 ) It is reasonable to assume that a female will not always spawn the same number of oocytes during each spawning event and that this variation in batch size may not be related to body size, as indicated for fecundity estimates with the FOM method. Furthermore, although only large oocytes were counted with the GF and NBF methods, not all females with oocytes >700 pm underwent FOM, as determined by histological inspection. Because histologi- cal inspection is not always practical when fecundity esti- mates are taken, we feel our estimates should include all fish with oocytes >700 pm. Both the GF and NBF methods resulted in higher, but not significantly different, fecundi- ty estimates than those yielded by the histological meth- od, suggesting that the wide variation among individual fish obscures any meaningful difference among methods. We believe our most accurate fecundity estimates are based on the actual histological counts of oocytes undergo- ing FOM. Our approach is supported by Hunter and Mace- wicz’s (1985b) finding that oocytes undergoing FOM can be used for fecundity estimates in fish with rapid FOM when hydrated oocytes are unavailable. Although the ex- act time frame of FOM is unknown for cobia, we presume it is relatively rapid. For example, fish in the early stages of FOM were captured in the morning. Cobia are pre- sumed to spawn during the day, probably the late after- noon, on the basis of collections of fertilized eggs (Ditty and Shaw, 1992). Other multiple spawning fish from simi- lar latitudes (e.g. C. nebulosus [see Brown-Peterson et al., 1988], black drum, Pogonias cromis [see Fitzhugh et al., 1993], and C. undecimalis [see Taylor et al., 1998]) under- go FOM within 12 h. Several large scombrids also have rapid FOM (McPherson, 1993; Schaefer, 1996; Farley and Davis, 1998). From this evidence, we conclude that accu- rate batch fecundity estimates can be made for cobia by using oocytes from fish undergoing FOM. The estimated mean batch fecundity values from the present study (1.9 x 106 eggs with the GF method, 8.5 x 105 with the NBF method, and 3.8 x 105 with the histo- logic method) are lower than previous mean estimates by Lotz et al. (1996) of 4.8 x 107 and Richards (1967) of 2-5 x 106 eggs. Differences in methods no doubt explain the wide range in estimates. We used only oocytes >700 pm for fecundity estimates rather than all oocytes >550 pm used by Lotz et al. (1996) and Richards (1967). Our meth- ods ensured that only oocytes likely to undergo hydration within the following 24 h were included in fecundity es- timates. Lotz et al. (1996) suggested that their batch fe- cundity values may have been an overestimate because all the oocytes counted may not have been released dur- ing spawning. Richards (1967) probably also overestimat- ed the batch fecundity of cobia, although his estimates are close to ours obtained by using the GF method. Batch fecundity was not estimated for any species by us- ing direct histological counts of oocytes undergoing FOM; thus, it is difficult to compare our results with other pub- lished results. Even though the estimates appear low when compared with more traditional methods of estimat- ing fecundity, there is less variation in the counts. The relatively small sample size {n= 26) used in our study for the FOM method may have resulted in an underestima- tion of batch fecundity. Increasing the sample size from <30 to 298 fish resulted in an increase as great as 33% in batch fecundity estimates for Atlantic mackerel (Scomber scombrus, see Watson et al., 1992). Although our batch fe- cundity estimates for cobia are realistic first approxima- tions, additional samples are necessary to produce a more accurate mean estimate of batch fecundity, a crucial value for accurate spawning stock biomass assessments. In ad- dition, the large variations in batch fecundity among indi- vidual fish are probably a biologically accurate represen- tation of variations in batch size in this multiple spawning species. Thus, assigning a single value to the batch fecun- dity of cobia does not give a biologically accurate portrayal of spawning stock biomass. The mean relative fecundity of 29.1 ±4.8 to 53.1 ±9.4 eggs/g ovary-free body weight calculated for cobia is low when compared with co-occurring inshore and estuarine fish in the region (Brown-Peterson et al., 1988; Fitzhugh et al., 1993). Cobia, like the co-occurring tripletail (Lo- botes surinamensis) , wahoo ( Acanthocybium solandri), common dolphinfish (Coryphaena hippurus), king mack- erel ( Scomberomorus caualla), and greater amberjack {Se- riola dumerili), is a large, subtropical pelagic fish and ex- hibits a very different life history than smaller nearshore and estuarine species. Fecundity data are available only for two of these co-occurring species: wahoo, with an esti- mated relative fecundity of 57.7 eggs/g (Brown-Peterson et al., 2000) and tripletail, with an estimated relative fe- cundity of 47.6 eggs/g (Brown-Peterson and Franks, in press). Values from both species compare favorably with our estimates for cobia. Other pelagic species for which relative batch fecundity values are available include At- lantic mackerel (55.5 eggs/g; Watson et al., 1992), south- ern bluefin tuna ( Thumnus maccoyii, 57 eggs/g; Farley and Davis, 1998) and yellowfin tuna (Thunnus albacares, 68 eggs/g; Schaefer, 1996). When the relative fecundity of the cobia is compared with that of other species with similar habitats and life histories, our estimate appears within the range of reported values for other pelagic species. Our study represents the first report of spawning fre- quency for cobia. The FOM and the POF methods pro- duced estimates of spawning at five-day intervals for co- bia in the SEUS and NCGOM. However, these estimates are based on three months of data for a potential six- month spawning season and thus may not represent the spawning frequency throughout the entire reproductive period for each region. Regardless, this spawning frequen- cy is lower than that reported for other large pelagic spe- cies, such as narrow barred Spanish mackerel (S. com- merson, 2-3 d, McPherson, 1993), southern bluefin tuna (daily spawners, Farley and Davis, 1998), yellowfin tuna (1-2 d, Schaefer, 1996) and wahoo (2-6 d, Brown-Peterson et al., 2000), as well as for the smaller pelagic carangids (3 d, Clarke and Privitera, 1995), spotted seatrout (2-7 d, Brown-Peterson et al., 1988), common snook (1. 1-2.5 d, Taylor et al., 1998), and red drum (2-4 d, Wilson and Neiland, 1994). Perhaps, the longer intervals between spawnings for cobia may be due to the longer distances that cobia need to travel between feeding and spawning grounds in comparison with the distances traveled by the Brown-Peterson et al.: Reproductive biology of Rachycentron canadum 27 species just mentioned. The exact location of cobia spawn- ing is unknown; early surveys have suggested spawning immediately outside the mouth of Chesapeake Bay (Jo- seph et ah, 1964), whereas later data from the southeast- ern United States have indicated that cobia spawn off- shore of North Carolina (Hassler and Rainville, 1975) and South Carolina (Shaffer and Nakamura, 1989). Egg col- lections in Crystal Bay, FL (Ditty and Shaw, 1992), imply that spawning occurs nearshore in the Gulf of Mexico, al- though other egg and larval evidence (Shaffer and Naka- mura, 1989; Ditty and Shaw, 1992) suggest spawning oc- curs in the Gulf of Mexico on the shelf 50-90 km from shore. Further evidence for offshore spawning was the col- lection of small larvae (3. 8-6. 8 mm) 50-90 km off the coast of Texas (Finucane et al.3). Because most of our samples were captured no more than 40 km from shore, cobia in immediate pre- or postspawning condition may not occur in those locations. In addition, our hook-and-line method of capture may be biased against cobia in immediate pre- or postspawning condition owing to changes in feeding be- havior at these stages. Although spawning frequencies of cobia from the three study areas were not significantly different, the apparent lower spawning frequency of cobia in the WGOM may be biologically relevant. Cobia from SEUS and NCGOM were estimated to be capable of spawning up to 36 times during the six-month spawning season, whereas fish from WGOM were estimated as capable of spawning 15 to 20 times during the spawning season. Hydrologic features of the ar- eas may explain the differences. The southeastern United States and the north-central Gulf of Mexico have substan- tial inputs of freshwater from major river systems which may result in high productivity in those areas (Livingston et al., 1997) and hence abundant food sources. In contrast, there is little freshwater input along the western Gulf of Mexico. Another possible explanation for the differences in spawning frequencies may be that many cobia in the Gulf of Mexico spawn in a single location that is closer to the north-central region than to the western region. The lack of 12-h POFs and the predominance of 48-h POFs seen in the ovaries of cobia from the western Gulf of Mexico may be due to the longer distance that western cobia must travel from the spawning grounds. This hypothesis as- sumes that cobia in the Gulf of Mexico do not have distinct breeding areas or subpopulations — a hypothesis support- ed by Hrincevich’s (1993) work on the molecular genetics of cobia. Although Hrincevich ( 1993) did find a high degree of heterogeneity in cobia mtDNA, this heterogeneity did not support the hypothesis that discrete stocks of cobia ex- ist in the northern Gulf of Mexico. The genetic data, in combination with data from tagging studies in the north- ern Gulf of Mexico (Franks et al.6), suggest that cobia in- 6 Franks, J. S., J. T. McBee, and M. T. Allen. 1992. Studies on the seasonal movements and migratory patterns of the cobia, Rachycentron canadum, in Mississippi marine waters and adja- cent Gulf waters. Interim Contract Rep. to Miss. Dep. Wildl., Fish, and Parks/Bur. Mar. Res, and U.S. Fish and Wildl. Serv, Atlanta, GA 30303. [Available from Gulf Coast Res. Lab., Ocean Springs, MS 36566-7000.] termix not only within the Gulf of Mexico but also along the southeastern Atlantic coast of the United States. Thus, the overall similarities in the reproductive biology of cobia throughout the southern United States are not surprising. The information provided in our study on batch fecundity and spawning frequency of cobia should aid effective man- agement of cobia stocks, as well as underscore areas where additional research is needed. Acknowledgments We thank the Mote Marine Laboratory (MML) sampling team (Teresa DeBruler, Carole Neidig, Marion Hersey, Diana Skapura, Roger DeBruler, and Sasha Koulish [MML staff] and Jonnie Walker, Rob Roberts and Don and Toma Marshall [MML volunteers] ), the Gulf Coast Research Lab- oratory (GCRL) sampling team (Don Barnes, Casey Nich- olson, Nate Jordan, Jody Peterson, Jason Steckler, Nicola Garber, and Melanie Griggs), Scott Holt (The University of Texas Marine Science Institute), and Hal Osborne (Texas Department of Parks and Wildlife) for assistance with col- lecting cobia samples. We also thank the many anglers, charter boat captains, tournament directors, and student interns who assisted our sampling efforts. Histological support for this project was provided by Marie Wright, Kim Lamey, Mary Tussey, and Tershara Matthews (GCRL); Leslie Christmas (GCRL) furnished the fecundity counts. We thank Susan Carranza (GCRL) for photographic devel- opment and drawing of Figure 1. Mark Peterson (GCRL) assisted with statistical analysis and reviewed the manu- script. Joseph W. Smith (National Marine Fisheries Ser- vice, Beaufort, NC) was the technical monitor for this project. This project was funded by the National Oceanic and Atmospheric Administration, National Marine Fish- eries Service (grant numbers NA57FF0294, NA86FL476, and NA96FL0358). Literature cited Bagenal, T. B., and E. Braum. 1971. Eggs and early life history. In Methods for assess- ment of fish production in fresh waters, (W. E. Ricker, ed.), p. 159-181. IBP (Int. Biol. Prgramme), Handbook 3, 2nd ed., Blackwell Sci. Publ., Oxford, England. Biesiot, P. M., R. M. Caylor, and J. S. Franks. 1994. Biochemical and histological changes during ovarian development of cobia, Rachycentron canadum, from the northern Gulf of Mexico. Fish. Bull. 92:686-696. Brown-Peterson, N. J.. and J. S. Franks. In press. Aspects of the reproductive biology of tripletail, Lobotes surinamensis, in the northern Gulf of Mexico. Proc. Gulf Caribb. Fish. Inst. 52. Brown-Peterson, N. J., J. S. Franks, and A. M. Burke. 2000. Preliminary observations on the reproductive biology of wahoo , Acanthocybium solandri, from the northern Gulf of Mexico and Bimini, Bahamas. Proc. Gulf Caribb. Fish. Inst. 51:414-427. Brown-Peterson, N., P. Thomas, and C. R. Arnold. 1988. Reproductive biology of the spotted seatrout, Cynoscion nebulosus, in south Texas. Fish. Bull. 86:373-388. 28 Fishery Bulletin 99(1 ) Clarke, T. A., and L. A. Privitera. 1995. Reproductive biology of two Hawaiian pelagic caran- gid fishes, the bigeye scad, Selar crumenopht hal mus , and the round scad, Decapturus macarellus. Bull. Mar. Sci. 56:33-47. Dawson, C. E. 1971. Occurrence and description of prejuvenile and early juvenile Gulf of Mexico cobia, Rachycentron canadum. Copeia 1971:65-71. Ditty, J. G., and R. F. Shaw. 1992. Larval development, distribution, and ecology of cobia, Rachycentron canadum, (Family: Rachycentridae) in the northern Gulf of Mexico. Fish. Bull. 90:668-677. Farley, J. H. and T. L. O. Davis. 1998. Reproductive dynamics of southern bluefin tuna, Thunnus maccoyii. Fish. Bull. 96:223-236. Fitzhugh, G. R., B. A. Thompson, and T. G. Snider III. 1993. Ovarian development, fecundity and spawning fre- quency of black drum Pogonias cromis in Louisiana. Fish. Bull. 91:244-253. Franks, J. S., J. R. Warren, and M. V. Buchanan. 1999. Age and growth of cobia, Rachycentron canadum, from the northeastern Gulf of Mexico. Fish. Bull. 97:459-471. Grier, H. J., and M. Abraham. 1983. A model for testicular recrudescence in Oreochromis aureus. In International symposium on tilapia in aqua- culture (L. Fishelson and Z. Yaron, eds.), p. 200-209. Tel Aviv Univ. Press, Tel Aviv. Grier, H. J.. and R. G. Taylor. 1998. Testicular maturation and regression in the common snook. J. Fish Biol. 53:521-542. Grier, H. J., R. G. Taylor, and R. O. Reese. 1987. The mechanism of tubule elongation during testicu- lar recrudescence in the redfish, Sciaenops ocellatus (Per- ciformes). In Proceedings of the V congress of European ichthyologists (S. O. Kullander and B. Fernholm, eds.), p. 285-291. Swedish Museum of Natural History, Stock- holm. Hassler, W. W., and R. P. Rainville. 1975. Techniques for hatching and rearing cobia, Rachycen- tron canadum, through larval and juvenile stages. Publ. UNC-SC-75-30, Univ. N.C. Sea Grant Coll. Prog., Raleigh, NC, 26 p. Hay, D. E., and J. R. Brett. 1988. Maturation and fecundity of Pacific herring ( Clupea harengus pallasi ): an experimental study with compar- isons to natural populations. Can. J. Fish. Aquat. Sci. 45:399-406. Hrincevich, A. W. 1993. Mitochondrial DNA analysis of cobia Rachycentron canadum population structure using restriction fragment length polymorphisms and cytochrome B sequence vari- ation. M.S. thesis, Univ. Southern Mississippi, Hatties- burg, MS, 91 p. Hunter, J. R., and B. J. Macewicz. 1985a. Rates of atresia in the ovary of captive and wild northern anchovy, Engraulis mordax. Fish. Bull. 83:119- 136. 1985b. Measurement of spawning frequency in multiple spawning fishes. In An egg production method for esti- mating spawning biomass of pelagic fish: application to the northern anchovy, Engraulis mordax (R. Lasker, ed.), p. 79-94. U.S. Dep. Commer., NOAA Tech. Rep. NMFS 36. Hunter, J. R., B. J. Macewicz, and H. R. Sibert. 1986. The spawning frequency of skipjack tuna, Katsuwonus pelamis, from the south Pacific. Fish. Bull. 84:895-903. International Game Fish Association. 1998. World record game fishes. International Game Fish Association, Pompano Beach, FL, 352 p. Jons, G. D., and L. E. Miranda. 1997. Ovarian weight as an index of fecundity, maturity and spawning periodicity. J. Fish Biol. 50:150-156. Joseph, E. B., J. J. Norcross, and W. H. Massman. 1964. Spawning of the cobia, Rachycentron canadum, in the Chesapeake Bay area, with observations of juvenile speci- mens. Chesapeake Sci. 5:67-71. Livingston, R.J., X. Niu, F.G. Lewis III, and G.C. Woodsum. 1997. Freshwater input to gulf estuaries: long-term control of trophic organization. Ecol. Appl. 7:277-299. Lotz, J. M., R. M. Overstreet, and J. S. Franks. 1996. Gonadal maturation in the cobia, Rachycentron can- adum, from the northcentral Gulf of Mexico. Gulf Res. Reports 9:147-159. Lowerre-Barbieri, S. K., M. E. Chittenden Jr., and L. R. Barbieri. 1996. The multiple spawning pattern of weakfish in the Chesapeake Bay and Middle Atlantic Bight. J. Fish. Biol 48:1139-1163. McPherson, G. R. 1993. Reproductive biology of the narrow barred Spanish mackerel ( Scomberomorus commerson Lacepede, 1800) in Queensland waters. Asian Fish. Sci. 6:169-182. Richards, C. E. 1967. Age, growth and fecundity of the cobia, Rachycentron canadum, from the Chesapeake Bay and adjacent mid- Atlantic waters. Trans. Am. Fish. Soc. 96:343-350. Schaefer, K. M. 1996. Spawning time, frequency, and batch fecundity of yel- lowfin tuna, Thunnus albacares, near Clipperton Atoll in the eastern Pacific Ocean. Fish. Bull. 94:66—76. Shaffer, R. V., and E. L. Nakamura. 1989. Synopsis of biological data on the cobia, Rachycentron canadum, (Pisces: Rachycentridae). U.S. Dep. Commer., NOAA Tech. Rep. NMFS 82 [FAO Fish Synop. 153], 21 p. Smith, J. W. 1996. Life history of cobia , Rachycentron canadum (Osteich- thyes: Rachycentridae), in North Carolina waters. Brim- leyana 23:1-23. SPSS, Inc. 1997. SPSS PC version 7.5. SPSS, Inc., Chicago, IL, 628 p. SPSS, Inc. 1998. SYSTAT 8.0: statistics. SPSS, Inc, Chicago, IL, 1086 p. Taylor, R. G., H. J. Grier, and J. A. Whittington. 1998. Spawning rhythms of common snook in Florida. J. Fish Biol. 53:502-520. Watson, J. J, I. G. Priede, P. R. Witthames, and A. Owari-Wadunde. 1992. Batch fecundity of Atlantic mackerel. Scomber scom- brus L. J. Fish Biol. 40:591-598. Wilson, C. A., and D. L. Nieland. 1994. Reproductive biology of red drum, Sciaenops ocella- tus, from the neritic waters of the northern Gulf of Mexico. Fish. Bull. 92:841-850. 29 Abstract— We estimate abundance and describe the depth distribution of harbor porpoise (Phocoena phocoena ) in north- ern California from a November 1995 ship survey. This was the first ship sur- vey designed to systematically survey different depth strata out to 91 m (50 fathoms) in this region. Harbor por- poise abundance in northern California is estimated at 5686 (CV=0.29, log-nor- mal 95% CI=3498-9242), based on 377 km of transect effort and 153 porpoise sightings. Using a confidence interval of differences approach, we determined that our abundance estimate was sig- nificantly different (P=0.G8; a=0.10) from an aerial survey estimate obtained 1 to 2 months earlier in the same region, where abundance was estimated at 13,145 (CV=0.39, log-normal 95% CI=6318-27,357). Possible explanations for differences in estimates include sea- sonal movement of porpoise to other areas or depths, insufficient transect effort during the ship survey, or under- estimates of the fraction of porpoise groups missed on the trackline due to large swells. Porpoise distribution in northern California was not random with respect to water depth; signif- icantly more porpoise than expected occurred at depths of 20 to 60 m and fewer porpoise than expected occurred at depths >60 m. Manuscript accepted 21 July 2000. Fish. Bull. 99:29-39 (2001). Abundance and depth distribution of harbor porpoise ( Phocoena phocoena) in northern California determined from a 1995 ship survey James V. Carretla Barbara L. Taylor Susan J. Olivers Southwest Fisheries Science Center National Marine Fisheries Serv/ice, NOAA P.O. Box 271 La Jolla, California 92038 E-mail address (for J. V. Caretta) Jim.Caretta@noaa.gov The abundance and distribution of har- bor porpoise (Phocoena phocoena ) in California waters has recently been described from a series of aerial surveys conducted by the U.S. National Marine Fisheries Service (NMFS) (Forney et al., 1991; Barlow and Forney, 1994; Forney 1995; Forney, 1999). The most recent abundance estimate was 5732 (CV=0.28) for central California and 11,066 (CV=G.39) for northern Califor- nia, based on aerial surveys conducted from 1993 to 1997. 1 Declining trends in porpoise abundance have recently been described in central California based on aerial surveys conducted from 1986 to 1995 (Forney, 1995, 1999). This per- ceived decline is unexpected because the central California population was expected to recover following a sub- stantial reduction in fishery-related mortality beginning in the late 1980s (Barlow and Forney, 1994; Barlow and Hanan, 1995; Julian and Beeson, 1998). Forney (1999) noted that harbor porpoise abundance was negatively correlated with positive sea surface temperature anomalies off central Cal- ifornia and suggested that perceived declines in porpoise abundance may be due to interannual movement of animals in and out of the study area in response to changing oceanographic conditions, rather than an actual popu- lation decline. Because the aerial survey study area was limited to nearshore waters <91 m (50 fathoms) in depth, one possible explanation was that por- poise might move into deeper waters in response to anomalous periods of warm water (Forney, 1999). Another explanation is that harbor porpoise moved north into northern California in response to warm-water episodes (For- ney’s trend data showed an increase in porpoise abundance in northern Cali- fornia during the same period of decline in central California, although sample sizes from northern California were too small to make unequivocal state- ments about trends in population size). Forney ( 1995, 1999) recommended that directed vessel surveys examine the influence of oceanographic and envi- ronmental variables on the distribution and abundance of harbor porpoise and focus on possible changes in the depth distribution of porpoise and the distri- bution of prey species (Forney 1995, 1999). Current knowledge of harbor porpoise stock structure off California suggests that porpoise do not make long-distance movements; thus it is pos- sible that a perceived population decline in central California is the result of small-scale changes in porpoise distri- bution, given that aerial survey tran- sects have remained unchanged since 1986 (Forney 1995, 1999). Today, NMFS recognizes (and man- ages separately) four stocks of harbor porpoise along the U.S. west coast; 1) 1 Forney, K. A. 1999. The abundance of California harbor porpoise estimated from 1993-97 aerial line-transect surveys. Ad- ministrative report LJ-99-02, National Marine Fisheries Service, Southwest Fish- eries Science Center, 16 p. [Available from Southwest Fisheries Science Center, P.O. Box 271, La Jolla, CA 92038.] 30 Fishery Bulletin 99(1) central California; 2) northern California; 3) Oregon-Wash- ington outer coast; and 4) inland Washington waters (Bar- low et al., 1997, 1998, Forney et al., 1999). Evidence for geographic subdivision of harbor porpoise populations along the U.S. west coast is apparent from pollutant stud- ies (Calambokidis and Barlow, 1991), which reveal latitu- dinal gradients in chlorinated hydrocarbon levels in wa- ters off Washington, Oregon, and California. Molecular genetic studies have also demonstrated larger-scale geo- graphic population subdivisions within the northeast Pa- cific, where four putative populations (Alaska, British Co- lumbia, Washington, and California) are recognized (Rosel et al., 1995). More recent genetic analyses, based on more sampling locations and different genetic markers, have provided evidence to support recognition of these four pop- ulations. Smaller-scale geographic subdivision within Cal- ifornia remains unresolved.2 We present the results of a 13-30 November 1995 line- transect ship survey in waters off California, which was designed to estimate abundance and describe the depth distribution of harbor porpoise. This was the first ship sur- vey to cover this region in late autumn and to survey sys- tematical lly different depth strata out to 91 m (50 fath- oms). The 1995 ship survey contrasts with previous NMFS ship surveys that were conducted earlier in the year along the 18-m isobath and that relied on models of porpoise density at different depths to calculate abundance (Bar- low, 1988). We estimate porpoise abundance for two re- gions within California (Fig. 1): Bodega Bay to the Cali- fornia-Oregon border, hereafter referred to as “northern California,” and Point Sur to San Francisco Bay in cen- tral California, hereafter referred to as “region 2,” as delin- eated by Barlow ( 1988). We emphasize results from north- ern California because relatively little survey effort was conducted in central California. We compare statistically aerial and ship survey abundance estimates in 1995 with those from recent aerial surveys in northern California and region 2 in central California, using a confidence in- terval of differences (CItl) method proposed by Lo (1994). We also describe the depth distribution of harbor porpoise in northern California from the 1995 ship survey sighting data. Methods Observers searched for marine mammals during daylight hours from the 53-m NOAA research vessel McArthur. Nine observers rotated through five duty stations on the flying bridge 10 m above sea level, spending 30 minutes at each station. Two observers at the periphery of the flying bridge scanned from the trackline to the ship’s beam (0 to 90 degrees) with 7x binoculars and two inner observers scanned from the trackline to 45 degrees on each side. The middle observer recorded data into a laptop computer and searched for porpoise groups by naked eye. A sixth 2 Chivers, S. J. 1999. Personal commun. NMFS, Southwest Fisheries Science Center, P.O. Box 271, La Jolla, CA. 92038. independent observer opportunistically searched from the ship’s bridge level (~7.5 m above sea level) to estimate the fraction of porpoise groups missed by the primary team of five observers. A series of predetermined saw-tooth transect lines be- gan at Point Conception, California (34°27'N) and ended at the California-Oregon border (42°00/N). Transect lines ex- tended from approximately the 20-m isobath to the 91-m (50 fathom) isobath. Some depths greater than 91m were surveyed, especially where the axes of submarine canyons intersected our transects. The ship could not routinely op- erate in water depths shallower than 20 m. The ship pro- ceeded along the transect lines in passing mode, i.e. it did not deviate from course even when porpoise groups were sighted. The study area was divided into several a prioj-i analysis regions (Fig. 1), the same as those used by Barlow (1988) and Barlow and Forney (1994). The region referred to as “northern California” incorporates region 4 and that part of region 5 that is south of the California-Oregon bor- der (Fig. 1). Areas of each region were previously calculat- ed by Forney.3 An event-driven data acquisition program (PPCRUISE) was used to record all sighting and effort data. The pro- gram was run on a laptop computer linked to the ship’s GPS system to obtain geographic position data. For each harbor porpoise group sighted, bearing and distance from the ship were recorded. Observers obtained the bearing to harbor porpoise groups using mounted protractors on the flying bridge. Distance to harbor porpoise groups was read from calibrated reticle marks imprinted on the eyepiece of 7x binoculars (Barlow and Lee, 1994) or estimated by eye for some close groups. Depth soundings were recorded sys- tematically every two minutes while the ship was under- way. The position of each porpoise sighting was calculated to correct for differences between the position of the ship and porpoise groups. Depths at which porpoise were sight- ed were determined from National Ocean Service (NOS) hydrographic survey data4 by using a triangulated irregu- lar surface model in Arc Info GIS. To estimate abundance, we used transect data collected only during calm sea states, defined as Beaufort 0 through 2 (wind speeds up to 6 knots, no whitecaps present). Por- poise abundance was estimated out to the 91-m isobath. Effort segments in deeper water were excluded from line- transect analysis to allow direct spatial comparison with recent aerial survey results (Barlow and Forney, 1994; Forney1). Porpoise abundance in region i was estimated 3 Forney, K. 1988. Contour mapping and the calculation of areas between 10m depth contours along the coasts of Califor- nia, Oregon and Washington. Administrative report LJ-88-23, National Marine Fisheries Service, Southwest Fisheries Center, 18 p. [Available from Southwest Fisheries Science Center, P.O. Box 271, La Jolla, CA 92038.] 4 National Oceanic and Atmospheric Administration (NOAA), National Geophysical Data Center (NGDC). 1999. National Ocean Service (NOS) hydrographic survey data, U.S. coastal waters, version 4.0, available on CD-ROM. Website: http:/ / www. ngdc. n oaa.gov / ngdc. h tm 1 Carretta et al Abundance and depth distribution of Phocoena phocoena off northern California 31 Figure 1 Study regions referred to in text. “Northern California” is defined as region 4 as well as part of region 5 south of the California-Oregon border. Central California includes regions 1-3. The thin line just offshore from Point Conception to Cape Blanco, Oregon, delineates the 91-m (50-fathom) depth contour. by using standard line-transect methods (Buckland et al., 1993): ^ A,n,S/( 0) 2Lig(0) where A( = size of the study area (km2) in region i\ ni - number of sightings in region i; S' = mean group size in region i; fl 0) = sighting probability density (/km) at zero per- pendicular distance L( = length of transect line (in km) surveyed in region i, and g(0) - probability of seeing a porpoise group directly on the trackline. The histogram of radial sighting distances revealed spikes at distances corresponding to whole and half-ret- icle binocular readings, which suggests that observers tended to round-off radial sighting distances. In the ab- sence of rounding error, the distance interval between any two adjacent radial sighting distances should have been relatively equal for both near and distant sight- ings, but these intervals usually increase with distance because of reduced precision in reading reticles at great- er distances. Theoretically, a linear regression of dis- 32 Fishery Bulletin 99(1) tance from the observer against the interval between ad- jacent radial sighting distances should have a slope of zero in the absence of rounding error. Following Barlow,5 we modeled the rounding error in radial sighting distanc- es with this regression method, where the slope of the re- gression represents the percentage of rounding error. As a result, we smeared radial sighting distances by a ran- dom factor between ±4%. Observers also showed a ten- dency to round bearing angles to the nearest 5 degrees; therefore we smeared angles by a random number be- tween ±5 degrees to further reduce the effects of round- ing (Buckland et al., 1993). We noted a problem in the recording of radial sighting distances when the ship was closer than 5.6 km (3 nmi) to shore. In these cases, the shoreline was closer than the true horizon and radial sighting distances were some- times recorded by using the shoreline as the horizon (doc- umented from database comments and inferred in other cases). The use of the shoreline as a false horizon intro- duces errors because resulting radial (and perpendicular) sighting distances are positively biased. Thirty sightings were identified for which shoreline bias probably resulted in positively biased distance readings. Frequency distribu- tions of “shoreline-biased” and “unbiased” perpendicular sighting distances were significantly different (Kolmogo- rov-Smirnov test, PcO.OOl) and the “shoreline-biased” dis- tribution revealed a higher proportion of sightings beyond 400 m. To reduce this bias, we excluded the “shoreline- biased” sightings prior to data truncation and to fitting the detection function to the perpendicular distance data. Buckland et al. (1993) recommend truncation of 5% to 10% of the largest perpendicular distances prior to model fit- ting. We truncated all sightings beyond 1 km, which elim- inated 10% of all sightings. Three models (hazard rate, half-normal, and uniform) were fitted to the perpendicular distance data by using the program DISTANCE (Laake et al.6) and the most parsimonious model was selected by DISTANCE based on minimizing Akaike’s Information Criterion (AIC: Akaike, 1973; Buckland et al., 1993). We also fitted the above models by using other truncation distances and bin intervals to examine the sensitivity of abundance estimates to these values. Because large porpoise groups are more likely to be de- tected at greater distances than single animals or pairs, it is possible to introduce bias into abundance estimates by over- estimating mean group size. Truncation of distance data will reduce potential overestimation of group size because the largest groups are eliminated, thus minimizing this bias. An additional step to reduce overestimation is to test for dependence between group size and detection distance 5 Barlow, J. 1987. Abundance estimation for harbor porpoise ( Phocoena phocoena ) based on ship surveys along the coasts of California, Oregon, and Washington. Administrative report LJ-87-05, National Marine Fisheries Service, Southwest Fisher- ies Center, 36 p. [Available from Southwest Fisheries Science Center, P.O. Box 271, La Jolla, CA 92038.] 6 Laake, J. L., S. T. Buckland, D. R. Anderson, and K. R Burn- ham. 1996. DISTANCE user’s guide, 82 p. Colorado Coopera- tive Fish and Wildlife Research Unit, Colorado State University, Fort Collins, CO 80523. by regressing the log of the observed group size against the detection probability at distance x [log(s;) versus g(x)], as recommended by Buckland et al. (1993). If the regres- sion is significant at a = 0.15, the mean group size is re- placed with the regression-based estimate of mean group size at zero distance, where theoretically, group size bias should not occur. We used this regression method in the program DISTANCE to determine which group sizes to use for estimating abundance. We tested the null hypothesis that harbor porpoise are randomly distributed with respect to depth by comparing the proportion of porpoise sightings to the proportion of survey effort within 20-m depth intervals. We used depth soundings as a measure of survey effort because soundings were taken at regular 2-minute intervals while the ship was underway. If harbor porpoise are randomly distributed by depth, then the proportion of porpoise sightings to depth soundings should be relatively equal for a given depth in- terval. Distributions were compared by using a nonpara- metric Kolmogorov-Smirnov goodness-of-fit test. Precision of the abundance estimates was estimated with two methods. Log-normal confidence intervals were calculated analytically with formulae presented in Buck- land et al. (1993). Bootstrap confidence intervals and coef- ficients of variation (CV) were calculated as follows. Effort and sighting data from region i were divided into 5-km ef- fort segments (for Beaufort 0-2 sea states only). A TRUE- BASIC computer program (BOOTPORP) was written to randomly draw (with replacement) effort segments within each region until the number of kilometers drawn equaled the number of kilometers actually surveyed. A pseudo- abundance estimate was then calculated from this boot- strap sample and the process was repeated 2000 times. The CV of the point estimates were calculated as the stan- dard error of the 2000 bootstrap estimates divided by the original point estimate. Bootstrap 95% confidence inter- vals were determined by identifying the 2.5th and 97.5th percentiles of the 2000 bootstrap estimates. For each boot- strap sample, the effective half-strip width [ESW or 1//10)], was treated as a random variable drawn from a normal distribution with a mean and standard error equal to that obtained from the detection model fitted to the trun- cated perpendicular distances. The probability of detect- ing a trackline group of porpoise, g( 0), was estimated for each bootstrap as a random variable drawn from a bino- mial distribution with mean = 0.769 (SE=0.117). Thisg(O) value was calculated by Barlow5 with independent observ- er methods, using a nearly identical vessel and the same observer configuration that we used. Owing to an insuf- ficient number of sightings by the independent observers during our survey, we could not independently estimate a value forg(0). We statistically compared abundance estimates obtained from the 1995 ship survey with estimates from aerial sur- veys conducted 1 to 2 months earlier (Forney, 1999), using the “confidence interval of differences” approach proposed by Lo (1994) and adopted by Forney and Barlow (1998) for bootstrap confidence intervals. Commonly used compara- tive methods, such as those based on whether confidence intervals overlap or whether one population mean is in- Carretta et al. Abundance and depth distribution of Phocoena phocoena off northern California 33 eluded within the confidence interval of a second mean, have been shown to be biased, because a levels do not ap- proach the intended value of 0.05 (Lo, 1994). Therefore, we used a third method proposed by Lo (1994), based on the confidence interval of the difference (CIrf), between two population means. Through computer simulation, we gen- erated 5000 log-normal pseudo-abundance estimates for the aerial and ship surveys (IV*), using the mean estimate and CV from each respective survey. The difference be- tween ship and aerial pseudo-estimates was calculated as d* - N* , -N* u ly ship ly air and a 95% confidence interval of the differences (CId) was calculated from the 5000 d * values with the percentile method. Aerial and ship survey estimates were considered significantly different if the resulting CIrf did not include zero. Under the alternative hypothesis that abundance estimates were significantly different, we estimated the statistical power of this test by constructing one thousand 95% CIrf intervals through simulation, using the observed effect size and variance from the ship and aerial surveys in 1995. The probability of committing a type-II error /3 was calculated as the fraction of 1000 intervals that included zero (indicating no significant difference at a=0.05). An initial power analysis at a = 0.05 revealed that the power to detect a difference as large as the one observed between aerial and ship estimates was low (=0.13). We therefore generated a power curve in order to objectively reselect an a level for the CIrf test that would provide an approximate power of 0.80 (Cohen, 1988). This resulted in an a level = 0.10; therefore, all statistical comparisons between aerial and ship estimates were considered statistically signif- icantly different if the 90% CIrf did not include zero. Following Forney and Barlow ( 1998), we estimated the sig- nificance level for this comparison by iteratively construct- ing a range of confidence intervals from the simulated data (i.e. 80%, 90%, 95%-, 96%, 97%...) and we identified the threshold a level (two-tailed) where the CI(/ just included zero. Results We surveyed a total of 594 km of transect in California during calm sea states (Beaufort 0-2) and detected 170 harbor porpoise groups within the truncation distance of 1 km (Figs. 2 and 3, Table 1). Most survey effort (377 km) occurred in northern California, where 153 groups of harbor porpoise were seen, mostly in the vicinity of Cape Mendocino. No harbor porpoise were seen within regions 1 and 3 in central California, but the amount of survey effort in these regions (60 and 91 km, respectively) was low. Owing to persistent coastal fog, transect coverage within region 3 was limited to the offshore area near the Farallon Islands. In region 2, only 17 porpoise groups were detected within the 1-km truncation distance, but only 91 km of trackline was surveyed. The perpendicular distance data were best fitted with the half-normal model without adjustment terms and had 125 124 123 122 121 120W Figure 2 On-effort transect effort (594 km) shown as thick gray lines in central and northern California during calm (Beaufort 0-2) sea states. the lowest AIC value of all competing models ix2 goodness- of-fit test, P=0.84, Fig. 4). Several truncation distances and interval groupings were explored when fitting a detection function to the distance data, and all fits resulted in esti- mates of abundance within 9% of each other. In general, the lowest abundances were obtained with hazard-rate models. Here, we report only the results obtained with the half-normal model with a truncation distance of 1 km. We used the observed mean group size of 2.45 for region 2 and 2.65 for northern California to estimate porpoise abun- dance because the regression of the log of observed group size versus g(x) was not significant (r2=0.11, P<0.89), sug- gesting no school size bias within the truncation distance of 1 km. Abundance of harbor porpoise in northern California was estimated at 5686 (log-normal 95% CI=3498-9242; bootstrap CV=0.29, bootstrap 95% CI=2760-8394) out to the 91 m isobath, based on 377 km of transect effort and 153 porpoise sightings in calm sea states (Beaufort 0-2). A statistical comparison of our estimate with the aerial 34 Fishery Bulletin 99(1 ) 125 124 123 122 121 120W Figure 3 On-effort sightings of harbor porpoise recorded in cen- tral and northern California during calm (Beaufort 0-2) sea states. survey estimate obtained 1-2 months earlier (A=13,145 CV=0.39)1 revealed that the 90% confidence interval of dif- ferences (CIrf) did not include zero, indicating that aerial and ship abundance estimates were significantly different at a = 0. 10 (90% Cld =-17,275 to -257, P=0.08, Fig. 5, A and B). Porpoise abundance for region 2 was estimated at 1041 (log-normal 95% CI=559-1941; bootstrap CV=0.44, boot- strap 95% CI=587-4138), based on 91 km of survey effort and 17 porpoise sightings. Our estimate for region 2 was significantly different (90% CIf/=-4084 to -162, P=0.07, a=0.10) from an estimate obtained with aerial surveys 1 to 2 months earlier in the year (A=2861 CV=0.39).l In northern California, harbor porpoise were not dis- tributed randomly with respect to water depth (Kolmogo- rov-Smirnov test, PcO.OOl; Fig. 6A). High proportions of porpoise sightings with respect to survey effort were found between 20 to 60 m, and fewer porpoise than expected were found in waters deeper than 60 m. Because calm oajyipcor-N'tiotDWCvi't® dooo T-T-d-^ c\i cvj cxi Perpendicular distance in km Figure 4 Half-normal model fit to the perpendicular distance data. Data were truncated at 1 km, which eliminated 10% of all observations. The model fitted the data well, as determined by a chi-square goodness-of-fit test (P=0.84). sighting conditions often occur in nearshore shallow wa- ters where porpoise densities are relatively high (Barlow, 1988), we also examined relative abundance at depth for rough sea states (Beaufort 3-4) to see if the overall depth patterns held. The depth distribution of porpoise was sim- ilar in rough sea states: more porpoise than expected occurred between 20 to 40 m and fewer porpoise than expected were seen at depths of 40 to 120 m ( Kolmogorov - Smirnov test, P<0.002, Fig. 6B). During our survey, the area from shore to the 20-m isobath went largely unsur- veyed because of the ship’s draft and the presence of local navigational hazards. Owing to an insufficient number of sightings within region 2 (n= 25, depth range=20 to 93 m), we did not attempt to statistically describe the depth dis- tribution of harbor porpoise for that region. Discussion Abundance of harbor porpoise off northern California Our estimate of porpoise abundance for northern Cali- fornia (5686, log-normal 95% CI=3498-9242) is consider- ably lower than estimates for the same area from 1988 to 1993 aerial surveys (9250, 95% CI=5943-14,397; Barlow and Forney, 1994). Barlow and Forney (1994) estimated 2.05 porpoise/km2 in northern California from aerial sur- veys (corrected for missed groups), whereas we estimated 1.26 porpoise/km2 (also corrected for missed groups). From 1984-85 ship surveys conducted in September, Barlow ( 1988) estimated 12,700 porpoise (2.09 porpoise/km2) from Bodega Head, California to Cape Blanco, Oregon, an area approximately 23% larger than our our study area in northern California. From aerial survey data collected 1 to 2 months prior to our ship survey, Forney1 reported a pre- liminary estimate of 13,145 porpoise (CV=0.39, log-nor- mal 95% CI=6316-27,357; 2.92 porpoise/km2) for northern California. This aerial estimate is significantly different from our ship survey estimate, as determined by the con- fidence interval of differences test (90% CIrf=— 17,275 to -257, P=0.08, a=0.10). Had we used the traditional a level Carretta et al. Abundance and depth distribution of Phocoena phocoena off northern California 35 of 0.05, these two estimates would not have been significantly different, but the power to detect such a difference was low (0.13). This result highlights recent criticisms of statistical significance testing (Johnson, 1999) and under- scores the importance of estimating statistical power to aid in decision-making. Regardless, real differences exist between aerial and ship estimates obtained in 1995, as evidenced in Figure 5, A and B, which may be due to a number of factors. Animals may have moved out of the study area between the time of the two studies. Sea- sonal movements of harbor porpoise are known to occur on the Atlantic coast (Polacheck et al., 1995), and in Glacier Bay, Alaska (Taylor and Dawson, 1984). In California, the situa- tion is less clear because polluntant evidence suggests limited latitudinal movement along the coast (Calambokidis and Barlow, 1991) and genetic stock structure remains unresolved (Chivers2). Several researchers have reported highest porpoise densities during late summer and autumn throughout California (Barlow, 1988, Monterey Bay; Calambokidis et al.7, cen- tral California; Dohl et al.8, California-wide; Goetz, 1983, northern California; Sekiguchi, 1995, Monterey Bay). Barlow (1988) and Seki- guchi (1995) observed the lowest harbor por- poise densities in Monterey Bay during winter. Forney (1995, 1999) reported harbor porpoise abundance was negatively correlated with pos- itive sea-surface temperature anomalies in the Monterey Bay region in autumn, a finding that suggests that animals move in response to changing oceanographic conditions. Pyle and Gilbert (1996) reported harbor porpoise sight- ings near Southeast Farallon Island in central California only from March to November, even though observers searched year-round over a ten-year period (1982-92). Collectively, the data suggest small-scale seasonal movement of harbor porpoise along the California coast. Most cetacean species in California show sea- sonal trends in distribution and abundance (Forney and Barlow, 1998) and despite the re- striction of harbor porpoise to neritic habitat 7 Calambokidis, J., C. Ewald, G. H. Steiger, S. M. Cooper, I. D. Szczepaniak, and M. A. Webber. 1990. Harbor porpoise studies in the Gulf of the Farallones. Final contract report CX 8000-8-0001 to the Gulf of the Farallones National Marine Sanc- tuary, 57 p. Fort Mason Center Bldg. 201, San Francisco, CA 94123. 8 Dohl, T. P., R. C. Guess, M. L. Duman, and R. C. Helm. 1983. Cetaceans of central and northern California, 1980-1983: status, abundance, and distribution. Contract rep. 14-12-0001-29090, p. 135-152. Pacific OCS Region, Minerals Man- agement Service, U.S. Department of the Interior, Washington D.C. 36 Fishery Bulletin 99(1 ) along the California coast, they probably respond to seasonal or interannual oceanographic changes. Large swells (3 m) during our ship survey may have caused observers to miss more porpoise groups on the trackline than is accounted for by our g{0) correction factor. The independent observer recorded an insuffi- cient number of sightings during our survey; therefore we could not independently estimate g(0). Instead, we adopted theg(O) value of 0.769 (SE=0.117) estimated by Barlow5 who used the same searching method and vessel type as we did. During our survey, the largest swells (3 m) occurred in northern California, where porpoise densities are highest. Further, swell height increased with latitude in California and the slope of the regression was significant (Zar, 1984; AN OVA, P«0.005). Even in calm sea states, porpoise groups may be easily missed because animals surface behind large swells and thus are not observed. Our transect lines frequently were oriented directly into the pre- vailing northwest swell or directly down-swell as a consequence of our sampling design, possibly contrib- uting to more groups being missed on the transect line than expected. If swells were considerably larger during 1995 survey than during the 1980s surveys, then Barlow’s g(0) value may not reflect the fraction of trackline groups missed and our estimate of abun- dance would be negatively biased by an unknown, but nontrivial, amount. Larger winter swells may also help explain why porpoise densities from winter ves- sel surveys are generally lower than those from sum- mer and autumn surveys (Goetz, 1983; Barlow, 1988; Sekiguchi, 1995; Calambokidis et al.7). Abundance in northern California may be under- estimated because the area inshore of the 20-m isobath, known to have relatively high porpoise densities, was not routinely surveyed. Only 2% of all survey effort and 4% of all sightings occurred at depths shallower than 20 m. Shallow water effort is included in our abundance esti- mate, thus we implicitly assume that porpoise density from shore to 20 m is equal to that in the remainder of our study area. Depth data from aerial surveys conduct- ed from 1988 to 1997 support this assumption. Porpoise encounter rates (weighted by effort for each depth stra- tum) are nearly equal for the area from shore to 90 m (0.33 porpoise/km) and from 20 to 90 m (0.32 porpoise/ km) (NMFS9). However, these data were collected primar- ily during September and October and may not reflect the depth distribution of harbor porpoise in November. During our survey, if porpoise densities were significantly higher from shore to 20 m than from 20 to 90 m, our es- timate of porpoise abundance would be negatively biased by an unknown amount. The same bias might result if significant numbers of harbor porpoise moved into waters deeper than 90 m between the time of the aerial and ship surveys. B 1 1 B a m _ . fd o o o o Simulated abundance o o o o ■ SHIP N = 5686 CV = 0.29 0 AERIAL A/= 13,145 CV = 0.39 a; -O E Difference in abundance (ship minus aerial) Figure 5 Distribution of simulated log-normal abundance estimates for 1995 ship and aerial surveys in northern California (A). Dif- ference between 5000 simulated abundance estimates for 1995 ship and aerial survey in northern California (B). The distribution of harbor porpoise in northern Califor- nia is highly clustered, especially near Cape Mendocino, where densities are highest, and our level of survey effort may not have fully captured the spatial heterogeneity in porpoise densities in northern California. The bootstrap CV of ship abundance estimate was not extremely precise (0.29) and reflects the spatial variability in porpoise en- counter rates in northern California on small (5-km boot- strap unit) scales. The estimated number of transect effort in kilometers needed to attain higher levels of precision for the northern California estimate, following the method described in Buckland et al. (1993, p. 303-304) were ap- proximately 800 and 3200 km, respectively, to obtain CVs of 0.20 and 0.10 (we surveyed 377 km). These effort esti- mates assume a Poisson distribution of porpoise, and are probably conservative. Repeated sampling of our transect lines over a longer period may have yielded an estimate of abundance more similar to recent aerial surveys. An adaptive sampling method, as used by Palka and Pollard (1999), would be useful on future surveys to increase sam- pling effort in known high-density regions. Abundance of harbor porpoise off central California Region 1 has the lowest densities of harbor porpoise in California (Barlow, 1988; Forney et al., 1991; Dohl et al.3) and we did not detect any porpoise during 60 km of survey 9 NMFS (National Marine Fisheries Service). 1999. Unpubl. data. Protected Resources Div., Southwest Fisheries Science Center, P.O. Box 271, La Jolla, CA. 92038. Carretta et al. Abundance and depth distribution of Phocoeno phocoena off northern California 37 Effort at depth ■ Porpoise at depth 0.4 B 0.4 0 3 0.2 0.1 0 sr 100 m and fewer than expected occur at depths <75 m (Raum-Suryan, 1995; Raum-Suryan and Harvey, 38 Fishery Bulletin 99(1 ) 1998). We attribute these differences in depth distribu- tion between California and the San Juan Islands to habi- tat: porpoises near the San Juan Islands occupy an inland waterway characterized by numerous islands with steep bathymetry, wheras the open coastline of California has relatively gentler bathymetry. In both areas harbor por- poise do not occur far from land. Along the outer coasts of Oregon and Washington, Green et al.12 reported approx- imately 25% of all harbor porpoise at depths of 100 to 200 m; further, the depth distribution of harbor porpoise changed seasonally, in summer, 56% of porpoise were inshore of the 100-m isobath and during winter, this pro- portion increased to 86%. Much of what we know about harbor porpoise abun- dance and depth distribution in California comes from aerial and ship data collected in late summer and early autumn, largely because weather conditions are more fa- vorable at this time of year. For northern California, in particular, there is a bias towards collection of autumn data: the depth distribution model of Barlow (1988) con- tained only September data and past abundance surveys have relied on summer and autumn data (Barlow, 1988; Forney, 1995; 1999). The exception to this seasonal col- lection of data are year-round aerial surveys conducted by Dohl et al.8 off central and northern California. How- ever, Dohl’s surveys were not ideally suited to estimate porpoise abundance because their transect lines were ori- ented perpendicular to the shoreline and most survey ef- fort occurred far offshore of known porpoise habitat. Ad- ditional winter and spring aerial surveys are needed in California to investigate seasonal differences in harbor porpoise distribution and abundance. To permit direct sea- sonal comparisons, we suggest that winter-spring surveys use existing transect lines from past summer-autumn aer- ial surveys. We recommend aerial surveys over ship sur- veys because they permit coverage of shallow water areas where larger research vessels cannot operate; are more cost-effective for surveying large areas, thus maximizing data collection during brief fair-weather periods; and re- duce the influence of large ocean swells that compromise the effectiveness of a ship hne-transect survey. Acknowledgments We thank the marine mammal observers who spent many hours collecting the sightings and effort data: W. Arm- strong, M. Donahue, B. Hanson, B. Odom, D. Outram, K. Raum-Suryan, R. Rowlett, (and S.J.C. and B.L.T. ). We also thank the officers and crew of the NOAA ship McArthur for all their support. Charles Stinchcomb assisted in iden- tifying shoreline-biased sightings. Karin Forney provided 12 Green, G. A., J. J. Brueggeman, R. A. Grotefendt, C. E. Bowlby, M. L. Bonnell, and K. C. Balcomb III. 1992. Ceta- cean distribution and abundance off Oregon and Washington, 1989-1990. Chapter 1 in Oregon and Washington marine mammal and seabird surveys (J, J. Brueggeman, ed.), p. 1-140. Final report to OCS (Offshore Continental Shelf) Study MMS 91-0093. (Available from Ebasco Environmental, 10900 NE 8th St., Bellevue, WA 98004.] unpublished data on harbor porpoise encounter rates at depth from previous aerial surveys and patiently reviewed the CIrf method. Tim Gerrodette provided a FORTRAN subroutine used to correct sighting positions in relation to vessel positions. Rich Cosgrove provided NOS bathy- metric data used to determine the depth distribution of porpoise. This survey was funded by the NMFS Office of Protected Resources. Lastly we thank Jay Barlow, Debbie Palka, Kim Raum-Suryan, and two anonymous reviewers for their critiques of the manuscript. An earlier draft of this manuscript served as working paper 98-PSRG-5 at the November 1998 Pacific Scientific Review Group meet- ing in Seattle, WA. Literature cited Akaike, H. 1973. Information theory and an extension of the maximum likelihood principle. In International symposium on infor- mation theory, 2nd ed., (B. N. Petran and F. Csaaki, eds.), p. 267-281. Akadeemiai Kiadi, Budapest, Hungary. Barlow, J. 1988. Harbor porpoise ( Phocoena phocoena ) abundance esti- mation in California, Oregon, and Washington: I. Ship Sur- veys. Fish. Bull. 86:417-432. Barlow, J., and K. A. Forney. 1994. An assessment of the 1994 status ofharbor porpoise in Cal- ifornia. U.S.Dep.Commer., NOAA Technical Memo., NOAA - TM-NMFS-SWFSC-205, 17 p. [Available from South- west Fisheries Science Center, P.O. Box 271, La Jolla, CA 92038.] Barlow, J., K. A. Forney, P. S. Hill, R. L. Brownell Jr., J. V. Carretta, D. P. DeMaster, F. Julian, M. S. Lowry, T. Ragen, and R. R. Reeves. 1997. U.S. Pacific marine mammal stock assessments: 1996. U.S. Dep. Commer., NOAA Technical Memo., NOAA-TM- NMFS-SWFSC-248, 223 p. [Available from Southwest Fisheries Science Center, P.O. Box 271, La Jolla, CA 92038.] Barlow, J., and D. Hanan. 1995. An assessment of the status of harbour porpoise in central California. Rep. Int. Whal. Comm, (special issue) 16:123-140). Barlow, J., and T. Lee. 1994. The estimation of perpendicular sighting distance on SWFSC research vessel surveys for cetaceans: 1974 to 1991. U.S. Dep. Commer., NOAA Tech. Memo., NOAA- TM-NMFS-SWFSC-207, 46 p. (Available from Southwest Fisheries Science Center, P.O. Box 271, La Jolla, CA 92038.] Barlow, J., P. S. Hill, K. A. Forney, and D. P. DeMaster. 1998. U.S. Pacific marine mammal stock assessments: 1998. U.S. Dep. Commer, NOAA Technical Memo., NOAA-TM- NMFS-SWFSC-258, 40 p. (Available from Southwest Fish- eries Science Center, P.O. Box 271, La Jolla, CA 92038.] Buckland, S. T., D. R. Anderson, K. P. Burnham, and J. L. Laake. 1993. Distance sampling: estimating abundance of biologi- cal populations. Chapman and Hall, London, 446 p. Calambokidis, J., and J. Barlow. 1991. Chlorinated hydrocarbon concentrations and their use in describing population discreteness in harbor porpoises from Washington, Oregon, and California. In Marine Mammal Strandings in the United States (J.E. Reynolds Carretta et al. Abundance and depth distribution of Phocoena phocoena off northern California 39 and D.K. Odell, eds.), p. 101-110. U.S. Dep. Commer., NOAA Technical Report NMFS 98. Cohen, J. 1988. Statistical power analysis for the behavioral sciences. Lawrence Erlbaum Assoc. Pubh, Hillsdale, NJ, 567 p. Forney, K. A. 1995. A decline in the abundance of harbor porpoise, Phocoena phocoena, in nearshore waters off California, 1986-93. Fish. Bull. 93(4):741-748. Forney, K. A. 1999. Trends in harbor porpoise abundance off central Cali- fornia, 1986-95: evidence for interannual changes in distri- bution? J. Cetacean Res. and Manage. 1:73-80. Forney, K. A., and J. Barlow. 1998. Seasonal patterns in the abundance and distribution of California cetaceans, 1991-1992. Mar. Marnm. Sci. 14(3): 460-489. Forney, K. A., D. A. Hanan, and J. Barlow. 1991. Detecting trends in harbor porpoise abundance from aerial surveys using analysis of covariance. Fish. Bull. 89:367-377. Forney, K. A., M. M. Muto, and J. Baker. 1999. U.S. Pacific marine mammal stock assessments: 1999. U.S. Dep. Commer., NOAA Technical Memo., NOAA-TM- NMFS-SWFSC-282, 62 p. [Available from Southwest Fish- eries Science Center, PO. Box 271, La Jolla, CA 92038.] Goetz, B. J. 1983. Harbor porpoise (Phocoena phocoena) movements in Humboldt Bay, California and adjacent waters. M.A. the- sis, Humboldt State Univ., Areata, CA, 118 p. Johnson, D. H. 1999. The insignificance of statistical significance testing. J. Wildl. Manage. 63<3):763-772. Julian, F., and M. Beeson. 1998. Estimates for marine mammal, turtle, and seabird mortality for two California gillnet fisheries: 1990-95. Fish. Bull. 96:271-284. Lo, N. C. H. 1994. Level of significance and power of two commonly used procedures for comparing mean values based on con- fidence intervals. Calif. Coop. Oceanic Fish. Invest. Rep. 35:246-253. Palka, D., and J. Pollard. 1999. Adaptive line transect survey for harbor porpoises. In Proceedings of the symposium on surveys, status and trends of marine mammal populations, Seattle, Wash- ington, USA, February 25-27 1998, (G. W. Garner, S. C. Amstup, J. L. Laake, B. F .J. Manly, L. L. McDonald, and D. G. Robertson, eds.), p 3-11. A.A. Balkema Publishers, Rotterdam, Netherlands. Polacheck, T., F.W. Wenzel, and G. Early. 1995. What do stranding data say about harbor porpoises? Int. Whal. Comm, (special issue) 16:169-179. Pyle, P, and L. Gilbert. 1996. Occurrence patterns and trends of cetaceans recorded from Southeast Farallon Island, California, 1973 to 1994. Northwestern Naturalist 77:1-8. Raum-Suryan, K. L. 1995. Distribution, abundance, habitat use, and respira- tion patterns of harbor porpoise (Phocoena phocoena) off the Northern San Juan Islands, Washington. M.S. thesis. Moss Landing Marine Laboratory, San Jose State LTniver- sity, San Jose, CA, 77 p. Raum-Suryan, K .L. and J. T. Harvey. 1998. Distribution and abundance of and habitat use by harbor porpoise, Phocoena phocoena, off the northern San Juan Islands, Washington. Fish. Bull. 96:808-822. Rosel, P. E., A. E. Dizon, and M. G. Haygood. 1995. Variability of the mitochondrial control region in pop- ulations of the harbour porpoise, Phocoena phocoena , on interoceanic and regional scales. Can. J. Fish. Aquat. Sci. 52:1210-1219. Sekiguchi, K. 1995. Occurrence, behavior and feeding habits of harbor porpoises (Phocoena phocoena ) at Pajaro Dunes, Monterey Bay, California. Aquat. Mamm. 21( 2):91— 103. Taylor, B.L. and PK. Dawson. 1984. Seasonal changes in density and behavior of harbor porpoise (Phocoena phocoena) affecting census methodol- ogy in Glacier Bay National Park, Alaska. Rep. Int. Whal Comm. 34:479-483. Zar, J.H. 1984. Biostatistical analysis, 2nd ed. Prentice Hall. Engle- wood Cliffs, NJ, 718 p. 40 Abstract — Three yellowfin tuna (Thun- nus albaca res) carrying ultrasonic depth- sensitive transmitters developed a strong association with the tracking vessel, following it at speeds up to 5 knots (2.6 m/s). Two fish associated with the tracking vessel during day- time, and the other fish during day and night periods. Swimming behavior appeared to depend on the speed of the vessel. The tuna remained within a few meters of the surface when the vessel was traveling at high speeds but moved deeper when the vessel drifted. The behavior of these fish is compared to those of other yellowfin tuna tracked in other situations (associated with fish- aggregating devices or unassociated with devices). The reasons for these associations are not known but some hypotheses are advanced. Manuscript accepted 10 July 2000. Fish. Bull. 99:40-48 (2001). Association of yellowfin tuna ( Thunnus albacares ) with tracking vessels during ultrasonic telemetry experiments Laurent Dagorn Institut de Recherche pour ie Developpement (IRD) BP 5045 34032 Montpellier Cedex 1, France E-mail address: dagorn@ird.fr Erwan Josse Institut de Recherche pour le Developpement (IRD) BP 70 29280 Plouzane, France Pascal Bach Institut de Recherche pour le Developpement (IRD) BP 5045 34032 Montpellier Cedex 1, France Tunas associate with floating objects, such as logs, anchored man-made fish- aggregating-devices (FADs) (see Freon and Misund, 1999, for a review), and fishing boats (Fonteneau and Diouf, 1994). Numerous ultrasonic telemetry experiments have been conducted (Cayre and Chabanne, 1986; Holland et al., 1990; Cayre, 1991; Cayre and Marsac, 1993; Marsac et ah, 1996; Bach et ah, 1998; Josse et al., 1998; Marsac and Cayre, 1998; Brill et ah, 1999) to determine the behaviors of tunas associated with anchored FADs, but no published studies have examined fish associated with drifting objects. More- over, during ultrasonic telemetry exper- iments, the assumption is that neither the transmitter nor the tracking oper- ation alters the behavior of the fish. Some yellowfin tuna, however, have developed associations with the track- ing vessel — a rare behavior previously observed on two occasions (Cayre et ah, 1996; Brill et ah, 1999). In other words, in these situations, the vessel is not fol- lowing the fish but the fish is following the vessel. In our study, we examined the move- ments of three yellowfin tuna, which clearly followed the tracking vessel dur- ing ultrasonic telemetry experiments. Our objective was to characterize these associations and to compare them with other types of association behavior. We discuss these observations in relation to some hypotheses on the nature of tuna associations with floating objects and propose ideas for future studies. Materials and methods Fish movements were monitored with acoustic telemetry techniques from the research vessel RV Alis. Tracking oper- ations were conducted between Octo- ber 1995 and April 1996 in French Polynesia. The depth-sensitive acous- tic transmitters carried by the fish and the ultrasonic receiving equipment were built by VEMCO (Shad Bay, Nova Scotia, Canada), and are described in detail in Dagorn et al. (2000). Fish were caught on vertical longline gear and transmitters were attached externally with either nylon tie-wraps (as described by Holland et al., 1990) or a stainless steel dart (as described by Brill et al., 1993). During tracking operations, simulta- neous acoustic data were collected be- Dagorn et ai.: Association of Thunnus albacares with tracking vessels 41 1 52 22 W Time (hours) 12:00 16:00 20:00 0:00 4:00 8:00 12 00 Figure 1 Horizontal (upper graph) and vertical (lower graph) movements of yellowfin tuna 1, 60 cm FL. Tracking lasted 22 h (October 1995). The period of associa- tion between the fish and the tracking vessel is shown by the bold line. Arrows indicate the direction of horizontal movement in the upper graph. The light gray patch represents a prey patch observed by the echo sounder. tween 10 and 500 m depth with a SIMRAD (SIMRAD, Horten, Norway) EK500 scientific sounder connected to a hull-mounted SIMRAD ES38B split-beam transducer (fre- quency 38 kHz, beam angle 6.9°). Acoustic data, along with vessel position, were simultaneously logged on a per- sonal computer running SIMRAD EP 500 software (Sim- rad, 1994). Vessel speeds were estimated from straight-line calculations by using positions of the tracking vessel based on data from the Global Positioning System for the first two fish. Speeds were taken directly from the knot meter of the vessel for fish 3, which provided a greater volume of data on real-time movements of the tracking vessel. When the crew suspected the fish had become associ- ated with the tracking vessel, experiments were developed to test the association (complete turns, changes in vessel speed and direction, etc.). Fish that clearly followed the vessel during such tests were considered to be associated. Results Of the fourteen yellowfin tuna that were tagged and tracked in French Polynesia from 1985 to 1997, three indi- viduals clearly exhibited strong and lengthy associations with the research vessel. Tuna 1 (60 cm FL) was caught at a depth of 120 m at midday close to a FAD anchored near Maupiti Island, lo- cated within the Leeward Islands of the Society Archipel- ago. The fish was tracked for 22 h as indicated in Figure 1. This fish associated with the FAD immediately after release but shifted to a free-swimming (unassociated) phase directed offshore (eastward movement) until 17:14 h, crossing for the first time a patch of mid-water prey ob- served on the echo-sounder. After crossing the patch for the second time (beginning of the night), the tuna returned to the FAD but did not re-associate; rather it began a cir- 42 Fishery Bulletin 99(1 ) 146 27 W Time (hours) 8:00 10:00 12 00 14:00 16:00 18:00 20:00 Figure 2 Horizontal and vertical movements of yellowfin tuna 2, 100 cm FL. Tracking lasted 11 h (December 1995). The period of association between the fish and the tracking vessel is shown by the bold line. Arrows indicate the direction of hori- zontal movement in the upper graph. cular movement around the island. The fish became as- sociated with the tracking vessel after traveling half-way around the island. The association occurred simultaneous- ly with a change in its vertical movement pattern, shifting from a movement pattern between the surface and 150 m to a surface oriented behavior that kept the fish within a few meters of the surface (Fig. 1). As the boat continued to move around the island, the fish maintained the asso- ciation and swam within 10 m of the surface, except for an excursion to 50 m around 07:30. However, this dive oc- curred simultaneously with a XBT (expendable bathythe- mograph) launch and the fish may have followed the in- strument as it went down, a behavior also observed by Block et al. (1997). When the tracking vessel reached the FAD where the fish had been caught the day before, the fish broke its association with the ship and dove, likely to join a small tuna school observed on the echo-sounder un- der the FAD at around 150 m depth. Yellowfin tuna 2 ( 100 cm FL) was captured at 9:00 while it was associated with a FAD off Ahe Island (Tuamotu Ar- chipelago) and after being followed for 11 h. Immediately after its release, the fish returned to the depth at which it was caught (between 200 and 250 m) and left the FAD heading northwest (Fig. 2). Before 11:00, the fish rose to the surface and became associated with the boat; the main engine was then shut down and the vessel drifted. Around 11:15, the boat began to move and the fish followed. The fish remained strongly associated with the moving vessel, swimming within the first 10 m below the surface until 16:25, when it began to break off the association, making some rapid dives to 70 m. Contact was lost at 18:38, after a sudden departure of the fish during a heavy rain squall. The fish was briefly relocated at 19:14. Yellowfin tuna 3 (108 cm FL) was caught at 07:38, close to a FAD located off the island of Tahiti (Fig. 3). After release, the yellowfin tuna returned to the depth where Dagorn et al.: Association of Thunnus albacores with tracking vessels 43 15CT54W 1 8° 1 0 S 06 00 Day 20 April 1000 14:00 B 1800 18:00 06:00 0 Day 21 April 10:00 14:00 50 100 150 t2 200 Depth (m) Night 20/21 April 22.00 02:00 06 00 0 50 100 150 200 Depth (m) D Night 21/22 April 18:00 18:00 22:00 02:00 06:00 1 >vf J 50 100 -f- 150 200 Depth (m) 06:00 0 Day 22 April 1000 1400 18:00 50 100 4 150 -4 f 200 -L 4 5 Depth (m) G 06 00 10:00 14:00 18 00 18.00 Depth (m) H 18:00 Night 22/23 April 22:00 02:00 06 00 22 00 02 00 06 00 Figure 3 Horizontal and vertical movements of yellowfin tuna 3, 108 cm FL. Tracking lasted 91 h (April 1998). This fish was always associated with the tracking vessel. Arrows indicate the direction of horizontal movement in the upper graph. Changes in the movements of the tracking vessel are rep- resented by vertical arrows on the graphs representing the vertical movements of the fish. 1 = the vessel left FAD1 to move to FAD2 during the night; 2 = the vessel arrived at FAJD2; 3 = the vessel left FAD2 to go to FAD3; 4 = the vessel arrived at FAD3. 5 = the vessel left the FAD to go close to the Maiao island; 6 = the vessel came back to FAD3 and start drifting; 7 = the vessel moved away from FAD to conduct trawl and acoustic survey operations. 44 Fishery Bulletin 99(1 ) A o 20 40 I 0 12 3 4 Vessel speed (knots) 5 Time (hours) 0:00 4:00 8:00 12:00 1600 20:00 10 6 !L w 1 - 4 2 C 0 20 T I <1 " " 1 40 1 * :|i i i h 1 100 4 r t T , 0 1 2 3 4 5 Vessel speed (knots) Figure 4 Relationships between swimming depths of the fish and the speed of the tracking vessel during associations. (A) Mean values (±SD) corresponding to each association: Tl, T2, T3 for tunas 1, 2, 3; dl, d2, representing day times (white dots) and nl, n2 representing night times (black dots) of the first two 24-h cycles of the track of tuna 3. The two other graphs repre- sent details of the association between tuna 3 and the track- ing vessel during the third day (22 April 1996) where many changes in the vessel speeds occurred: (B) vertical movements and vessel speeds in relation to time; and (C) depth versus vessel speed (mean value ±SD). it was captured (approximately 150 m) and immedi- ately became associated with the tracking vessel. It re- mained at a depth between 80 m and 150 m while the vessel drifted close to the FAD. When the ship left the FAD (FAD1) at 16:20 to go to an other FAD (FAD2 South of Moorea Island), the fish moved closer to the surface and followed in a similar manner to that ex- hibited by yellowfin tuna 1 and 2. It remained within a school of other yellowfin tuna (individual size rang- ing from 20 to 50 kg) that could be seen from the ship, swimming just below the surface. During the second day (21 April), the fish remained associated with the ship as it drifted close to FAD2. Mean swimming depth was 75 m. During the second night (21-22 April), the vessel moved from FAD2 to FAD3, southwest of Maiao Island, and the fish followed, swimming close to the surface as during the first night. When the vessel and the fish arrived at FAD3 in the morning of 22 April, the fish left the surface and came under the ship. Around 07:45, the vessel left the FAD to shelter close to Maiao Island to find better sea conditions, with the tagged yel- lowfin tuna and the school following closely. The fish remained associated when the vessel returned to the FAD at 12:30. In the afternoon, the vessel drifted close to the FAD until 19:00, after which it made rapid ac- celerations away from the FAD to break off the asso- ciation. The strategy was successful, and the fish re- mained associated with the FAD. During the fourth day (23 April), trawl and acoustic survey operations were conducted away from the FAD. The fish did not associ- ate with the vessel during this period, but its presence at the FAD was regularly observed. Attempts to re- associate the fish with the tracking vessel were not suc- cessful because the fish returned to the FAD when the ship moved 0.5 nmi from the FAD. The fish remained associated with the FAD until operations terminated. Figure 4 shows the relationships between swimming depths and the speed of the vessel during all observed associations. Shallower swimming depths were observed when the vessel was moving at higher speeds, both dur- ing daytime (tuna 1 and 2) and nighttime (tuna 3) as- sociations. Figure 4 also shows in detail the response of tuna 3 to rapid changes in vessel speed during the third day (22 April), when the vessel was moving be- tween FAD3 and the Maiao Island, or drifting close to the FAD or to the island. This day represents an exam- ple of frequent changes in vessel speed and correspond- ing changes in swimming depths of the fish. Discussion Individual yellowfin tuna have been documented asso- ciating with tracking vessels in the Indian Ocean (a 108-cm yellowfin tuna, Cayre et ah, 1996) and near the main Hawaiian Islands (a 167-cm yellowfin tuna, Brill et al., 1999). However, these authors merely noted the occurrence of the associations without providing fur- ther analyses or comments on this striking behavior. It is noteworthy that all these fish were yellowfin tuna. It Dagorn et al.: Association of Thunnus albacares with tracking vessels 45 seems, however, that this behavior is not size dependent (sizes ranged from 60 to 167 cm FL) nor is it related to the size of the tracking vessel: 12-m vessel for Cayre et al. (1996), 20-m and 53-m vessels for Brill et al. (1999), and 28-m vessel in our study. Horizontal movements The horizontal movements of fish associated with tracking vessels duplicated the horizontal movements of the track- ing vessel. Therefore, the observed paths are not compa- rable with horizontal movements of tagged fish that were not associated with tracking vessels. Our results, however, give information on the duration of associations and pos- sible “competition” between FADs and tracking vessel to attract the tagged fish. Considering the different patterns of movements of tu- nas observed at anchored FADs, Holland (1996) proposed three horizontal patterns: 1 ) fish that leave the FAD and show no tendency to return to it over the duration of the track; 2) fish that spend the entire duration of the track (day and night) within a few hundred meters of the FAD, and 3) fish that spend daylight hours at the FAD site, leave at night and return to the same or an adjacent FAD the next day. Fish 1 and 2, as well as the 167-cm yellow- fin tuna tracked by Brill et al. (1999) near the main Ha- waiian Islands, associated with the tracking vessel during daytime, which corresponds to the third class defined by Holland (1996). Conversely, Cayre et al. (1996) reported a nighttime association between a 108-cm yellowfin tuna and the tracking vessel. Moreover, fish 3 remained associ- ated with the tracking vessel for more than two 24 h (con- tinuous day and night cycles) which corresponds to the second pattern defined by Holland (1996). Although our sample size was small, the three fish of our study, and the two other yellowfin tuna that exhibited such associa- tion (Cayre et al., 1996; Brill et al., 1999), exhibited dif- ferent lengths of associations, at different periods of the diurnal cycle, all of which also correspond to the variety of patterns observed for fish associated with anchored FADs. These features, however, cannot be used to determine if tuna treat drifting and anchored floating objects different- ly, as first proposed by Holland et al. ( 1990). The possible competition between FADs and the track- ing vessel to aggregate tuna is an interesting feature of our results. The three tuna were caught close to and were considered associated with a FAD. Yellowfin tuna 1 and 2 left their FADs after release and did not associate im- mediately with the tracking vessel, whereas yellowfin tu- na 3 associated with the tracking vessel after release. In the last part of their track, yellowfin tuna 1 and 3 clearly abandoned the vessel to associate with FADs. The pres- ence of yellowfin tuna 2 was also noticed close to the FAD a few hours after the end of the tracking. Attempts to re- aggregate yellowfin tuna 3 while associated with the third visited FAD were not successful. However, it is noteworthy that this tuna chose to associate with the tracking vessel rather than to FADs 1 or 2 on the previous days, showing a different motivation than those exhibited toward FAD3. Cayre et al. (1996) attempted to abandon the associated 108-cm yellowfin tuna by rapid vessel accelerations, trying to make the fish associate with a FAD, but without success. We cannot determine the reason for a possible preference of fish toward vessels or FADs. We can only propose that tuna regard the vessel and FADs in a similar manner, or that the choice in aggregating between these two struc- tures depend on factors (external or internal stimuli) that we could not record during our experiments. Vertical movements It is known that the swimming depth of yellowfin tuna is controlled by the diurnal cycle: surface swimming at night and deep swimming at daytime (Carey and Olson, 1982; Cayre and Chabanne, 1986; Holland et al., 1990; Cayre, 1991). However, besides this diurnal behavior, it seems that fish travel closer to the surface when associated with a moving vessel: yellowfin tuna 1 and 2 swam very close to the surface (mean swimming depth=5.3 m ±2.9 for tuna 1 and 8.4 m ±4.5 for tuna 2) when they were associated with the moving vessel during daytime. Yellowfin tuna 3 also exhibited very shallow swimming depths when associ- ated with the tracking vessel, but during nighttime (mean swimming depth=11.3 ±6.6 and 9.3 ±4.8 for the first two nights). Cayre et al. (1996) and Brill et al. (1999) did not report any relationship between the swimming depth of the fish and the speed of the tracking vessel. Figure 4 indi- cates definite relationships between fish swimming depths and the speed of the vessel during the associations. How- ever, while this vertical reaction of the associated fish to the different vessel speeds have been observed during day- time, we should mention that no observations were made to examine the response of the fish to low vessel speeds during nighttime, which should have been deeper than the depths exhibited by yellowfin tuna 3 during the first two nights. This association behavior is similar to that observed by Holland et al. (1990), in that fish tend to be closer to the surface when associated with FADs. We propose that floating objects generally induce the fish to swim closer to the surface and that this tendency increases when floating objects are moving fast. In addition to the shallower swimming of yellowfin tuna when associated with moving vessels, the amplitude of ver- tical oscillations are drastically reduced. We suggest that when fish are associated with a vessel, they reduce the am- plitudes of their vertical oscillations, and that the mean swimming depth is partly controlled by the speed of the vessel (i.e. the distance from the fish to the tracking vessel decreases when the vessel speed increases). The reasons for this change are not known. More data are clearly needed to examine the exact effects of a floating object (including its speed) on the vertical pattern of associated fish and to distinguish these effects from those due to the diurnal cycle, thermoregulation, or foraging behavior. Hypotheses to explain why tunas associate or disassociate with tracking vessels The reasons why pelagic fish associate with floating objects are still not known conclusively (Freon and Misund, 1999). 46 Fishery Bulletin 99(1 ) Tagging operations certainly represent a stress for the fish, especially when performed after a traumatic capture and removal from the water to attach the tag. If we assume that a fish considers this particularly large floating object (the tracking vessel) to be a shelter against the stress or possible injury caused by tagging procedures, the asso- ciation with the vessel could then be interpreted as an antipredator behavior (shelter from predator hypothesis, Suyehiro, 1952, cited in Freon and Misund, 1999). In fact, Block et al. (1992) and Brill et al. (1993) did observe badly injured fish swimming within a few meters of the surface (one Indo-Pacific blue marlin and one striped marlin, respectively), which corresponds to the swimming pattern exhibited by our fish when following the moving tracking vessel. If an injury or significant stress occurred during the capture or tagging operations, one could expect to see post- tagging antipredator behavior. The time delay between release and the onset of association behavior differs from one fish to another and ranges up to 16 hours after release (tuna 1), which argues against a stress-related associa- tion caused by the tagging operation. Moreover, observa- tions of a school of yellowfin tuna exhibiting the same association as yellowfin tuna 3 prove that nontagged and apparently noninjured and nonstressed tuna develop the same association. Our observations thus do not support the “shelter from predator” hypothesis as an explanation for the attraction of tuna to the tracking vessel. The role of social behavior to explain the association of fish with floating objects has been expressed in the “meeting point” hypothesis (Dagorn, 1994; Dagorn and Freon, 1999; Freon and Misund, 1999). This hypothesis proposes the enhancement of fish aggregation by floating objects through improving the encounter rate between small schools or between isolated individuals, or both. Ac- cording to this hypothesis, tuna associate with various floating objects (drifting logs, anchored FADs, boats) to in- crease their chances of encountering conspecifics. Yellow- fin tuna 1 and 2 seemed to be isolated during the tracking, whereas yellowfin tuna 3 was a member of a school. Yel- lowfin tuna 1 broke its association with the tracking ves- sel, joining individuals (observed by the echo-sounder) lo- cated under a FAD. It is not possible to know if the fish left the tracking vessel because of the FAD or because of the conspecifics. This observation, however, appears to support the “meeting point” hypothesis: this tuna and those of the aggregation benefited from their respective associations to find more conspecifics. Yellowfin tuna 3 was visually ob- served to be with a school during nights when the school swam close to the boat, and acoustically observed when it was associated with FAD3. The school was estimated to be composed of 80 individuals while associated with FAD3. Our observations were not precise enough to determine if new individuals joined the school during the 4-day experi- ment, nor if the school broke its association with the boat to join a group already aggregated to FAD3, in a manner similar to that shown by yellowfin tuna 1. However, we be- lieve that the present observations do not reject the meet- ing point hypothesis. Yellowfin tuna 1 and 3 left the tracking vessel to stay close to anchored FADs. It is difficult to know, however, if they broke the vessel association to associate with FADs or to join conspecifics located close to the FADs, or both. Contact with yellowfin tuna 2 was lost owing to a heavy rain. We do not know if the fish voluntarily broke off the association with the vessel or if it simply lost contact with the vessel. For instance, if the fish used the sound of the vessel to stay close, it is possible that the sound of the ves- sel was masked by the rain. Because it is very important to know why tuna associate with floating objects (or vessels in the present case), it is al- so essential to understand why tuna leave floating objects. Although the sample size of our study was very small, it seems that the presence of other floating objects, conspecif- ics, or bad sea conditions can be responsible. Understand- ing the reasons why tuna form and break off aggregations is of major importance when studying the consequences of aggregation on tuna movements and distribution (Dagorn and Freon, 1999). Future studies The objective of a sonic tagging experiment is to observe movements of a fish in its natural environment. The vari- ety of experiments conducted throughout all the tropical oceans (Cayre and Chabanne, 1986; Holland et al., 1990; Cayre, 1991; Cayre and Marsac, 1993; Marsac et al. 1996; Bach et al., 1998; Josse et al., 1998; Marsac and Cayre, 1998; Brill et al., 1999; Dagorn et al., 2000) have contrib- uted to a considerable increase in knowledge on the behav- ioral ecology of tropical tunas. Nevertheless, when a fish associates with a tracking vessel, although a very rare event, this objective has been violated. Among the track- ing experiments on 14 yellowfin tuna in French Polynesia, the distinction between vessel-associated and unassoci- ated individuals was very obvious. Moreover, this striking behavior has never been observed on other tuna species during tracking experiments (i.e. bigeye tuna, Thunnus obesus, and skipjack tuna, Katsuwonus pelamis). We con- sider that there is no possible doubt on the nature of the movements exhibited by a tagged individual (free movements or patterns associated with the vessel), which insures the validity of the interpretations of sonic tagging results. However, these rare events can be used to study the associations of fish with floating objects in a general sense. Rather than interrupting the tracking operation, we propose to develop particular experiments to improve our knowledge on tuna behavior. During the associations with vessels described in this paper, fish sometimes fol- lowed the vessel at speeds of up to 5 knots (2.6 m/s). It could be useful to use this behavior to study in situ the relationship between endurance time and velocity. More- over, it is important to collect data on the duration of asso- ciations. Observations of the biological environment of the associated fish would also be very useful to test the valid- ity of certain concepts, such as the “meeting point” hypoth- esis. During our experiments, we observed the biological environment (i.e. both prey and conspecifics) using an echo-sounder. The sounder assisted us to observe patches of prey (for yellowfin tuna 1) and the tuna aggregation Dagorn et al.: Association of Thunnus olbacares with tracking vessels 47 joined by yellowfin tuna 1, and also helped to determine the size of the school that yellowfin tuna 3 was drawn away from. However, the use of a multibeam sonar that can observe the few meters below the surface could pro- vide complementary information, especially for such close observations. Similar sonar units have been successfully used to observe the structure and the behavior of small pelagic fish schools (Gerlotto et al., 1999) and would be particularly appropriate for the observation of both tuna schools and prey in the vicinity of a tracking vessel. A long-range multibeam omnidirectional sonar could also provide useful information on the horizontal distribution and spatial dynamics of tuna schools around the vessel. Rather than disregarding tuna aggregations around tracking vessels, we propose to continue collecting infor- mation on such events. The tracking vessel represents a useful and fully instrumented, mobile floating object adapted to conduct detailed ethological observations to improve our knowledge of the behavior of tuna aggregated around floating objects. Acknowledgments This work was done within the frame of the ECOTAP Program (studies of tuna behavior using acoustic and fishing experiments), which is a joint program between SRM (Service des Ressources Marines de Polynesie Frangaise), IFREMER (Institut frangais de recherche pour l’exploitation de la mer), IRD (Institut de Recherche pour le Developpement). The authors wish to thank all the sci- entists who participated in the cruises, as well as the offi- cers and crew of the IRD RV Alis for providing valuable help during all the cruises. We are also very grateful to David Itano and Dave McConaghay, who kindly improved the English text, and the three referees who made many constructive suggestions. Literature cited Bach, P., L. Dagorn, E. Josse, F.-X. Bard, R. Abbes, A. Bertrand, and C. Misselis. 1998. Experimental research and fish aggregating devices (FADs) in French Polynesia. SPC FAD (Fish Aggregating Device) Inf. Bull. 3:3-19 Block, B. A., D. T., Boothand, and F. G. Carey. 1992. Depth and temperature of the blue marlin, Makairci nigricans, observed by acoustic telemetry. Mar. Biol. 114:175-183 Block, B. A., J. E. Keen, B. Castillo, H. Dewar, E. V. Freund, D. J. Marcinek, R. W. Brill, and C. Farwell. 1997. Environmental preferences of yellowfin tuna ( Thun- nus albacares ) at the northern extent of its range. Mar. Biol. 130:119-32. Brill, R. W., B. A. Block, C. H. Boggs, K. A. Bigelow, E. V. Fruend, and D. J. Marcinek. 1999. Horizontal movements and depth distribution of large adult yellowfin tuna (Thunnus albacares) near the Hawai- ian Islands, recorded using ultrasonic telemetry: implica- tions for the physiological ecology of pelagic fishes. Mar. Biol. 133:395-408. Brill, R. W„ D. B. Holts, R. K. C. Chang, S. Sullivan, H. Dewar, and F. G. Carey. 1993. Vertical and horizontal movements of striped marlin (Tetrapturus audax ) near the Hawaiian Islands, deter- mined by ultrasonic telemetry, with simultaneous mea- surement of oceanic currents. Mar. Biol. 117:567-574. Carey, F. G., and R. J. Olson. 1982. Sonic tracking experiments with tunas. Collect. Vol. Sci. Pap. ICC AT 17:458-466. Cayre, P. 1991. Behaviour of yellowfin tuna (Thunnus albacares ) and skipjack tuna (Katsuwonus pelamis) around fish aggregat- ing devices (FADs) in the Comoros Islands as determined by ultrasonic tagging. Aquat. Living Resour. 4:1-12 Cayre, P, and J. Chabanne. 1986. Marquage acoustique et comportement de thons trop- icaux (albacore: Thunnus albacares, et listao: Katsuwonus pelamis ) au voisinage d’un dispositif concentrateur de pois- sons. Oceanogr. Trop. 21(21:167-83 Cayre, P, and F. Marsac. 1993. Modeling the yellowfin tuna (Thunnus albacares ) ver- tical distribution using sonic tagging results and local envi- ronmental parameters. Aquat. Living Resour. 6:1-14 Cayre, C., D. Norungee, F. Marsac, and F. Conand. 1996. Rapport de la campagne ACMAR06 (Marquage acoustique de thonides) a file Maurice, du 11 au 16 fevrier 1996. Doc. Sci. AT/COI/PTR2, 29 (4), 13 p. [Available from the Indian Ocean Commission (IOC), Avenue Sir Guy Forget, Quatre Bornes, lie Maurice.] Dagorn, L. 1994. Le comportement des thons tropicaux modelise selon les principes de la vie artificielle. Ph.D. diss., ENSA de Rennes, France, 250 p. Dagorn, L., P. Bach, and E. Josse. 2000. Movement patterns of large bigeye tuna (Thunnus obesus ) in the open ocean, determined using ultrasonic telemetry. Mar. Biol. 136: 361-371. Dagorn, L., and P. Freon. 1999. Tropical tuna associated with floating objects: a simu- lation study of the meeting point hypothesis. Can. J. Fish. Aq. Sci. 56(6):984-993 Fonteneau, A., and T. Diouf 1994. An efficient way of bait-fishing for tunas recently developed in Senegal. Aquat. Living Resour. 7:139-51 Freon, P, and O.A. Misund. 1999. Dynamics of pelagic fish distribution and behaviour: effects on fisheries and stock assessment. Fishing News Books, Blackwell’s, Oxford, 348 p. Gerlotto, F., M. Soria, and P. Freon. 1999. From 2D to 3D: the use of multibeam sonar for a new approach in fisheries acoustics. Can. J. Fish. Aquat. Sci. 56( 1 ):6— 12 Holland, K. N. 1996. Biological aspects of the association of tunas with FADs. SPC FAD (Fish Aggregating Device) Inf. Bull. 2: 2-7 Holland, K. N., R. W. Brill, and R. K. C. Chang. 1990. Horizontal and vertical movements of yellowfin tuna and bigeye tuna associated with fish aggregating devices. Fish. Bull. 88:493-507 Josse, E., P. Bach, and L. Dagorn. 1998. Simultaneous observations of tuna movements and their prey by sonic tracking and acoustic surveys. In Advances in invertebrates and fish telemetry (J. P. Lagardere M.-L. Begout-Anras, and G. Claireaux, eds.), p.61-69. Hydrobiologia 371/372. 48 Fishery Bulletin 99(1 ) Marsac, F., and P. Cayre. 1998. Telemetry applied to behaviour analysis of yellowfin tuna (Thunnus albacares, Bonnaterre, 1788) movements in a network of fish aggregating devices. In Advances in invertebrates and fish telemetry (J. P. Lagardere, M.-L. Begout-Anras, and G. Claireaux, eds.), p. 155-171. Hydro- biologia 371/372. Marsac, F., P. Cayre, and F. Conand. 1996. Analysis of small scale movements of yellowfin tuna around fish aggregating devices (FADs) using sonic tag- ging. In Proceedings of the expert consultation on Indian Ocean tunas, 6th session, Colombo, Sri Lanka, 25-29 Sep- tember 1995 (A. A. Anganuzzi, K. A. Stobberup, and N. J. Webb, eds.), p. 151-159. IPTP (Indo-Pacific Tuna Project) Coll. 9. Simard. 1994. SimradEP500: echo processing system. Simrad Sub- sea A/S, Horten, Norway. 49 Abstract— Species-specific restriction site variation in the 12S/16S rRNA and ND-3/ND-4 mtDNA regions was used to distinguish among 15 rockfish species of the genus Sebastes common to the waters of Alaska. Intraspecific variation exhibited by eight of the spe- cies (based on five individuals of each species) did not obscure the interspe- cific variation, except possibly between S. zacentrus and S. var-iegatus. Intra- specific nucleotide diversity averaged 0.0024 substitutions per nucleotide, whereas interspecific nucleotide diver- gence averaged 0.0249. In contrast, the average nucleotide divergences between Sebastes and two other scorpaenid spe- cies, Helicolenus hilgendorfi and Sebas- tolobus alascanus, were 0.0805 and 0.1073, respectively. Cladistic and phe- netic analyses supported some, but not all, of the subgenera assignments of Sebastes. A scheme for distinguishing among the species studied was pre- sented. Restriction sites of 10 restric- tion endonucleases were mapped in the two PCR-amplified mtDNA regions by using double digests. In all, we detected 153 sites corresponding to 640 (13.5%) of the 4815 nucleotides in the two regions combined. The ND-3/ND-4 region exhibited substantially more intraspecific, interspecific, and interge- neric variation than the 12S/16S rRNA region. Manuscript accepted 22 August 2000. Fish. Bull. 99:49-62 (2001). Identification of rockfish ( Sebastes sppJ by restriction site analysis of the mitochondrial ND-3/ND-4 and 12S/16S rRNA gene regions Anthony J. Gharrett Andrew K. Gray Fisheries Division, School of Fisheries and Ocean Sciences University of Alaska Fairbanks 11120 Glacier Highway Juneau, Alaska 99801 E-mail address (for A. J. Gharrett): ffaig@uaf.edu Jonathan Heifetz Auke Bay Laboratory Alaska Fisheries Science Center National Marine Fisheries Service, NOAA 11305 Glacier Highway Juneau, Alaska 99801-8626 The species-rich genus of Sebastes rock- fish has challenged both fisheries scien- tists and ichthyologists since they were first described from Alaskan waters by Tilesius (S. ciliatus\ 1813, cited in Esch- meyer, 1998) and Richardson (S. cauri- nus ; 1845). Both the large number of species, about 100 worldwide (Ishida, 1984; Kendall, 1991), and the metamor- phoses that occur in larval and juve- nile fish produce a confusing number of forms. The diversity of species and forms combine to limit our knowledge of the biology, including life histories, of rockfishes. To date, identification to spe- cies is not possible for many larvae and juveniles (e.g. Kendall, 1991; Moser, 1996), and distinguishing between some adult species may be difficult. For exam- ple, adult S. variegatus is similar to S. zacentrus and adult S. mystinus, S. melanops, and S. ciliatus are often mis- identified (Love1). The inability to iden- tify species constrains surveys of larval abundance and, consequently, ecologi- cal studies that are important for con- servation and management of rockfish and other species. In addition, the ques- tions facing biologists and fishery man- agers require tools that can resolve intraspecific population (stock) struc- ture, as well as methods for identifying species. The size of the genus and the pau- city of information about some of the species have also contributed to a cha- otic history of their systematics and many aspects of the phylogeny have not been resolved (Kendall2 and re- viewed in Cramer 11895] and Phillips 11957]. Cuvier (1829, cited in Eschmey- er,1998) first described the genus Se- bastes for northern Atlantic specimens. The number of genera recognized for the species presently placed in Sebastes has expanded and contracted repeated- ly, reaching a maximum of 15 (Jordan et al., 1930) and now these genera are generally considered subgenera. When combined with five northwestern Pa- cific Ocean (Matsubara, 1943) and one northern Atlantic Ocean subgenus, Se- bastes comprises about 22 subgenera (Kendall, 1991). Identification and systematics of fish depend largely on morphological char- acters; morphology alone, however, does not always provide sufficient criteria, especially for identification of larval and juvenile forms. Genetic informa- tion, obtained by using biochemical or molecular methods, has been used to address systematic problems. In some 1 Love, M. 2000. Personal commun. Marine Sciences Institute, University of California, Santa Barbara, CA. 93106. 2 Kendall, A.W. 2000. Personal commun. NMFS Alaska Fisheries Center, 7600 Sand- point Way NE, Seattle, WA 98115. 50 Fishery Bulletin 99(1 ) instances, genetic differences can be used to differentiate between species that have overlapping morphologies. For example, cryptic species of southern Atlantic Ocean Se- bastes species were recognized from mtDNA analysis (Ro- cha-Olivares et al., 1999a). Genera and many species of rockfish can be distinguished from protein electrophoresis differences (e.g. Tsuyuki et al., 1968; Johnson et al., 1972). More recently, allozyme data (Seeb, 1986) and mtDNA variation (Johns and Avise, 1998; Rocha-Olivares, 1998a; Seeb, 1998; Rocha-Olivares et al. 1999b, 1999c) have been used to address questions about the evolution and sys- tematics of Sebastes. Genetic differences may provide the means for identifying rockfish larvae and juveniles that cannot be identified from their morphology (Seeb and Ken- dall, 1991). Recently, Rocha-Olivares (1998b) devised a PCR-based approach for identification of Sebastes species. The advantage of his approach is that it is fast. The disad- vantage is that failed PCR reactions are part of the iden- tification scheme. However, failed reactions can also result from poor quality DNA or intraspecific variation and lead to misidentification of the specimens. Intraspecific genetic variation can also provide information about population structure (e.g. Wishard et al., 1980; Seeb et al., 1988; Ro- cha-Olivares and Vetter, 1999). Vertebrate mitochondrial DNA (mtDNA) is compact (about 16,500 base pairs) and has been completely se- quenced in a variety of organisms including carp (Cyprinus carpio; Chang et al., 1994) and rainbow trout (Oneorhyn- ehus mykiss\ Zardoya et al., 1995). Because mitochondria are transmitted primarily through maternal genes (Gyl- lensten et al., 1991), mtDNA is haploid and clonally in- herited (Meyer, 1993). Restriction fragment analyses of PCR-amplified regions of mtDNA provide a rapid and practical method for detecting nucleotide sequence varia- tion in mtDNA between individuals or species. Sequence variation detected by restriction enzymes produces bina- ry character-state data that can be used in phylogenetic analyses (e.g. Dowling et al., 1992). An advantage of re- striction site surveys over sequencing is that they are practical for detecting variation in large sequence spans. The number of nucleotides screened in restriction site sur- veys depends on the number of restriction enzymes used and their match with the DNA sequence. We have developed primers that can be used to PCR amplify regions of Sebastes mtDNA. The amplified regions provide material for addressing species identification and stock identification questions about rockfish. In addition, the haplotypes observed provide information for address- ing systematic relationships among Sebastes. Our objective in this study was to examine the potential that restriction fragment analyses of PCR-amplified mtDNA regions have for the study of rockfish biology. We asked the following specific questions: 1 ) Is there interspecific hap- lotype variation? 2) Is there intraspecific haplotype varia- tion? 3) Does intraspecific variability compromise the use of mtDNA restriction fragments in species identification? 4) Can a simple strategy for identifying species be de- vised? 5) If there is interspecies variation, how do similari- ties between species correlate with (presumed) systematic relationships? To answer these questions, we conducted restriction site analyses on five individuals from each of 15 different Sebastes species common in Alaskan waters and mapped the sites using double digests to determine indi- vidual-based haplotypes. From these data, we examined intra- and inter-specific divergences and used both phe- netic and cladistic procedures to examine relationships among the haplotypes. We also mapped the sites for short- spine thornyhead ( Sebastolobus alascanus) and Helicole- nus hilfendorfi to facilitate analysis. Finally, we developed a mtDNA restriction fragment-based strategy for identify- ing Sebastes species. Materials and methods Adult specimens of 15 different species of Sebastes rockfish and Sebastolobus alascanus were collected from the east- ern Gulf of Alaska (Table 1). These species are the most abundant of the approximately 25 species reported in the region. In the field, species identification was confirmed by using the pictoral guide of Kramer and O’Connell (1988) and the key and descriptions in Hart (1973). H. Ida (Kita- sato University, Sanriku, Japan) provided samples of Heli- colenus hilgendorfi from Japanese coastal waters. Samples of heart tissue from each specimen were preserved in 95% ethanol or a solution of 20%’ dimethyl sulfoxide (DMSO), 0.25M ethylenediaminetetraacetic acid (EDTA) at pH 8 and saturated with NaCl (Seutin et al., 1991). Total cellular DNA was isolated by phenol-chloroform extraction (Wallace, 1987) or with Puregene DNA™ iso- lation kits (Gentra Systems, Inc., Minneapolis, MN). Two target regions were PCR-amplified from total cellular DNA with primers that we developed for coho salmon (Of i- corhynchus kisutch ) mtDNA studies. The ND3/ND4 region begins in the glycyl tRNA gene and spans the NADH- dehydrogenase subunit-3, arginyl tRNA, NADH-dehydro- genase subunit-4L, and NADH-dehydrogenase subunit-4 genes, ending in the histidyl tRNA gene. The 12S/16S re- gion extends from near the phenylalanyl tRNA end of the 12S rRNA gene through the valyl tRNA gene to near the leucyl tRNA end of the 16S rRNA gene (Table 2). From restriction digests, we estimated that the ND3/ND4 and 12S/16S regions comprised 2385 and 2430 base pairs (bp), respectively, as compared with 2331 and 2402, respective- ly, for O. mykiss. Target sequences were amplified by heat- ing to 94°C for 5 min, followed by 30 cycles for 1 min at 94°C, 1 min at 55°C, and 3 min at 72°C using Taq polymerase from Perkin Elmer (Norwalk, CT) according to manufacturer’s directions. ND3/ND4 amplification re- quired 3mM MgCl2, whereas amplification of 12S/16S re- quired 2mM MgCl2. Single digests of subsamples of the PCR-amplified mtDNA regions were made by using 10 restriction endonucleases. BsHJ I, Cfo I, Dde I, Hinf I, Mbo I, Msp I, and Rsa I have 4-nucleotide recognition sites; Bst N I recognizes an ambiguous 5-nucleotide site; and Hind II and Sty I rec- ognize ambiguous 6-nucleotide sites. Digestions were car- ried out under conditions recommended by the manu- facturers. Fragments were separated by electrophoresis through 1.5% agarose (a mixture composed of one part Gharrett et al.: Identification of Sebastes spp. by restriction site analysis 51 Table 1 Rockfish and related species and subgenera of Sebastes spp. used in mitochondrial DNA haplotype comparisons. The number des- ignates the species and the letter indicates the particular composite haplotype observed. Designation Common name Species Subgenus 1, a and b Pacific ocean perch Sebastes alutus Acutomentum 2, a and b rosethorn rockfish Sebastes helvomaculatus Sebastomus 3 quillback rockfish Sebastes maliger Pteropodus 4, a and b redbanded rockfish Sebastes babcocki Rosicola 5, a and b black rockfish Sebastes melanops Sebastosomus 6 yellowtail rockfish Sebastes flavidus Sebastosomus 7, a-d sharpchin rockfish Sebastes zacentrus Allosebastes 8 harlequin rockfish Sebastes variegatus Allosebastes 9 redstripe rockfish Sebastes pronger Allosebastes 10, a and b rougheye rockfish Sebastes aleutianus Zalopyr 11, a and b yelloweye rockfish Sebastes ruberrimus Sebastopyr 12 shortraker rockfish Sebastes borealis Zalopyr 13 light dusky rockfish Sebastes ciliatus Sebastosomus 14 silvergray rockfish Sebastes brevispinis Acutomentum 15, a and b copper rockfish Sebastes caurinus Pteropodus 16 helicolenus Helicolenus hilgendorfi 17, a-d shortspine thornyhead Sebastolobus alascanus Table 2 Primers used for polymerase chain reaction amplification of rockfish ( Sebastes , Helicolenus, and Sebastolobus spp.) mtDNA regions, a = Thomas and Beckenbach (1989); b = Cronin et al. (1993); c = Gharrett7;d = Anderson et al. (1981); e = Anderson et al. (1982); f = Roe et al. (1985); g = Chang et al., 1994; h = Zardoya et al. (1995). Region amplified Sequence Location in O. mykissh Source ND3/ND4 5' TAACGCGTATAAGT G ACTT CCAA 3' 5' TTTT GGTT CCTAAGACCAAT GG AT 3' bp 10574-10596 bp 12881-12904 from a (similar to b) from a and c (similar to b) 12S/16S 5' AATT CAGCAGT G ATAAACATT 3' 5' AG ATAG AAACT G ACCT GG ATT 3' bp 1234-1254 bp 3615-3635 consensus: d, e, f, g consensus: d, e, f, g 1 Gharrett, A. J. 2000. Unpubl. Oncorhynchus kisutch sequences. Fisheries Division, Univ. Alaska, Fairbanks, 11120 Glacier Hwy., Juneau, AK 99801. Ultra Pure™ agarose [BRL Gibco, Grand Island, NY! and two parts Synergel™ [Diversified Biotech Inc., Bos- ton, MA]) in 0.5xTBE buffer (TBE is 90 mM tris-boric acid, and 2 mM EDTA, pH 7.5). DNA in the gel was stained with ethidium bromide and photographed on an ultraviolet light transilluminator. Digests that produced small unresolvable fragments on agarose gels were sub- jected to electrophoresis on 8% polyacrylamide gels (29:1 acrylamide:bisacrylamide) in 2xTAE (TAE is 40 mM tris- acetic acid and 1 mM EDTA, pH 8.0). DNA in poly- acrylamide was stained with SYBR Green 1 Nucleic Acid Stain™ (Molecular Probes, Eugene, OR). Molecular weight markers used to estimate restriction fragment sizes were 100 base pair (bp) or 25-bp ladders (BRL Gibco, Grand Island, NY). Restriction sites were mapped by using dou- ble digests. Double digests were examined both in agarose and polyacrylamide by using 100- and 25-bp ladders. Com- posite haplotypes for all 10 restriction enzymes and both mtDNA regions were determined for each individual. Generalized (relaxed Dollo) parsimony trees (Swofford et al., 1996) were computed from shared restriction sites by a heuristic search with PAUP* 4.0 (Swofford, 1998), which assumed unordered states. Because the likelihood of the loss of a site is higher than the restoration of a lost site, we conducted analyses that assumed 1 ) no added cost, 2) twice the cost, and 3) four- times the cost for restor- ing a site. Multiple maximum parsimony trees from each analysis were combined to produce a majority consensus tree using PAUP* 4.0 (Swofford, 1998). A maximum-like- lihood tree was estimated with the program RESTML in 52 Fishery Bulletin 99(1 ) PHYLIP 3.57c (Felsenstein3), assuming that all restriction sites were 4 bp long (PHYLIP, Felsenstein3). Nucleotide divergences (proportion of nucleotide substitutions) and their standard errors were estimated according to Nei and Tajima (1983), Nei (1987), and Nei and Miller (1990) by using REAP (McElroy et ah, 1990). Results Restriction fragment patterns from double digests were used to construct restriction site maps for comparisons of species and detection of intraspecies variation (Appen- dix 1). The map includes 153 restriction sites, 36 of which were common in all haplotypes and 28 of which were cla- distically uninformative because the presence or absence occurs only in a single haplotype. Many of the cladistically uninformative sites, however, were useful in species delin- eation. These data represent 153 restriction sites (79.3 on average) corresponding to 640 nucleotides (332.05 on aver- age) per haplotype. Among the 85 fish examined were 30 different compos- ite haplotypes (Table 3); each species had haplotypes that were distinct from those of other species, although S. var- iegatus composite haplotype 8 differed at a single site from S. zacentrus composite haplotype 7c (Table 4). All other pairs of species differed by 5 or more sites. Intraspecific variation was observed in nine of the seventeen species even when only five specimens of each species were ana- lyzed. The most variable species were S. zacentrus and Se- bastolobus alascanus, each of which had four haplotypes. In the study, differences between haplotypes ranged from a single site difference or 0.0014 nucleotide substitutions per site to 65 restriction site differences and 0.120 nucleo- tide substitutions per site (Table 4). Nucleotide divergence within variable species averaged 0.0024 subsitutions ( 1.56 site changes), whereas divergences between Sebastes spe- cies averaged ten-fold higher, 0.0249 (15.4 site changes), ranging from 0.0015 (1 site change) to 0.0384 (25 site changes). Nucleotide divergences between Sebastes spe- cies and Sebastolobus alascanus averaged 0.1073 (59.2 site changes) and divergences between Sebastes species and H. hilgendorfi averaged 0.0805 (43.5 site changes). Distribution of the variation between the two different mtDNA regions (ND3/ND4 and 12S/16S) reflects their rates of evolution. In the 12S/16S region, which is more conservative, 27 of 58 restriction sites were shared by all haplotypes. Nucleotide diversities between Sebastes spe- cies averaged 0.0094 nucleotide changes per nucleotide (a total of 3.29 sites differences in the region), divergences be- tween Sebastes and H. hilgendorfi averaged 0.0641 (12.67 site differences), and divergences between Sebastes and Sebastolobus alascanus averaged 0.0561 (18.03 site differ- ences). In contrast, in the ND3/ND4 region only 9 of 95 sites were common to all haplotypes; and nucleotide diver- 3 Felsenstein, J. 1993. PHYLIP (Phylogeny Inference Pack- age) version 3.57c. Distributed by the author. Department of Genetics, Box 357360, Univ. Washington, Seattle, WA 98195-7360. gences between Sebastes species averaged 0.0471 (12.11 site differences) and divergences between Sebastes and H. hilgendorfi and between Sebastes and Sebastolobus alas- canus averaged 0.1373 (31.75 site differences) and 0.1929 (40.93 site differences), respectively. The maximum like- lihood and majority consensus tree for the 60 maximum parsimony trees that imposed a cost of two for regained restriction sites had identical topologies (Fig. 1). The to- pologies of parsimony trees, which had either no addition- al cost or a cost of four, were somewhat different. Several groups of species were present in all three parsimony to- pologies. The S. zacentrus-S. variegatus pair, mentioned above, and each of four species pairs — S. melanops-S. flavi- dus, S. babcocki-S. heluomaculatus, S. proriger-S. brevispi- nis, and S. maliger-S. caurinus — clustered tightly at sub- terminal nodes. A more interior cluster of species included S. melanops, S. flavidus, S. babcocki, and S. helvomacula- tus. In addition, S. maliger and S. caurinus clustered sep- arately from all other Sebastes species and the Sebastes species were distinct from H. hilgendorfi and Sebastolobus alascanus. The mtDNA variation we observed among Sebastes spe- cies provides a tool for identifying species. From our da- ta, numerous schemes could be devised that distinguish among the Sebastes species examined. We propose a sim- ple scheme that minimizes the number of digests required and involves separation of restriction fragments from the ND3/ND4 PCR product on an agarose-Synergel™ gel us- ing only four restriction enzymes. Mho I digests produce 11 different haplotypes (haplotypes A-K; Figure 2 A; Ta- ble 3); S. alutus (B), S. melanops (E), S. babcocki (G and H), S. ruberrimus (I), and S. caurinus (J) are species spe- cific. If Mbo I haplotypes A (S. heluomaculatus or S. flavi- dus) or C (S. maliger or S. caurinus ) are observed, digest the ND3/ND4 PCR product with Hind II; Hind II haplo- type B is specific for S. heluomaculatus and Hind II hap- lotype C is specific for S. maliger (Fig. 2B; Table 3). If Mbo I haplotypes F (S. ciliatus or S. borealis) or K (S. aleutia- nus, S. proriger, or S. brevispinis) are observed, digest the ND3/ND4 PCR product with BstN I; BstN I haplotype A is specific for S. ciliatus and BstN I haplotype G is specific for S. brevispinis (Fig. 2C; Table 3). Mbo 1 and BstN I hap- lotypes do not distinguish between S. aleutianus and S. proriger , but Cfo I haplotype B is specific for S. aleutianus (Fig. 2D, Table 3). The combined haplotype of Mbo I, Hind II, BstN I, and Cfo I can be used to identify S. borealis (KAFD) and S. proriger (FAFD) (Fig. 2). The single differ- ence between S. zacentrus and S. variegatus is the pres- ence of a 123-bp fragment in Rsa I digests of S. zacentrus (Table 2; Appendix 1). This simple scheme takes advantage of unique single- site differences for several of the species. Although a neigh- bor-joining tree (Saitoh and Nei, 1987) appeared stable to intraspecific variation for increased sample sizes of three species (data not shown), a single site change that produc- es apparent convergence between taxa in our scheme is conceivable. Increased certainty can be achieved by con- ducting digests with all four enzymes. With this strategy there will be at least two site differences between every pair of species, except S. proriger and S. brevispinis, which Gharrett et al.: Identification of Sebastes spp. by restriction site analysis 53 Table 3 Composite haplotypes for Sebastes spp., Helicolenus hilgendorfi, and Sebastolobus alascamts in the 12S/16S and ND3/ND4 mtDNA regions. The species codes are listed in Table 2. The haplotype codes refer to haplotypes in Table 4. Five individuals were analyzed for each species. Where intraspecific variation was observed, alternative haplotypes are presented. Species 12S/16S haplotypes BstN I BstXJ I Cfo I Dc/c I Hind II Hint I Mbo I Msp I Rsa I Sty I la A A A D A B C B A A lb A A A D A A C B A A 2a A A A C A B B A D A 2b A A A C A B C A D A 3 A A A D A B A A B A 4a A A A C A B C B B A 4b A A A C A B C B B A 5a A A A D A B c B D A 5B A A A D A B c B D A 6 A A A A A B c B D A 7a A A A B A B c A C A 7b A A A B A B c A C A 7c A A A B A B c A C A 7d A A A B A B c A C A 8 A A A B A B c A C A 9 A A A D A B c A C A 10a A A A D A B c B A A 10b A A A D A B c B A A 11a A A A D A C c A A A lib A A A D A C c A A A 12 B A A A A B c A C A 13 A A A D A B c A A A 14 A A A D A A c A C A 15a A A A D A B A A A A 15b A A A D A B A A A A 16 D B B F B B F B A A 17a C B B E A D E B E B 17b C B B E A D E B E B 17c C B B E A D E B E B 17d C B B E A D E B E B Species ND3/ND4 haplotypes fi.s/N I BstU I Cfo I Ode I Hind II Hint I Mbo I Msp I Rsa I St} I la B C A J A A B B C c lb B c A J A A B B C c 2a F c D K B A A B C c 2b F c D K B A A B E c 3 F B D L C D C C B A 4a F C D G A A G B B C 4b F C D F A A H B B C 5a D C D F A B E A A C 5B D C D F A B E B A C 6 E c D H A C A B A c 7a F A D E A A D B C c 7b F A D A A A D B C c 7c F C D E A A D B C c continued 54 Fishery Bulletin 99(1 ) Table 3 (continued) Species ND3/ND4 haplotypes ftsvN I Bstl J I Cfo I Dde I Hind II Hint I Mho I Msp I Rsa I Sty I 7d F A D E A A D B C B 8 F C D E A A D B B C 9 F C D E A A K E B C 10a F C B N A E K D D C 10b F C B N A F K D D C 11a C C A B A D 1 B D C lib C C D B A D 1 B D C 12 F C D D A A F B B C 13 A D A M A A F B F C 14 G C D C A A K D B D 15a F B C L A G C C B A 15b F B C L A G J C B A 16 J G F Q A A N G H A 17a H E E 0 A H M F G F 17b 1 E E 0 A H M F G F 17c H F E 0 A H M F G E 17d H E E P A H M F G F can be resolved by using Msp I, and S. zacentrus and S. variegatus (see above). We do not recommend using Dde I because it has many sites, often produces small fragments requiring both agarose and polyacrylamide gels for reso- lution, and is, therefore, time consuming to analyze. How- ever, the restriction patterns of Dde I are nearly species specific. Discussion Sufficient interspecies restriction site variation occurred in the ND3/ND4 and 12S/16S mtDNA regions in Sebastol- obus alascanus , Helicolenus hilgendorfi , and 15 Sebastes species to distinguish among them. Intraspecific varia- tion was observed in nine of the seventeen species, but it did not interfere with our ability to distinguish between species. We used the interspecific variation to devise a strategy to identify the species we studied. Intraspecific variability can serve as a basis for stock identification. A broader survey, particularly for S. zacentrus and S. var- iegatus, might reveal overlaps in haplotype compositions that compromise the ability to distinguish between some species pairs. This would be most likely if there were gene flow between the species or if the species had recently di- verged. Otherwise, extending the analysis to other mtDNA regions and additional restriction endonucleases should in- crease resolution. Of course, additional intraspecific varia- tion has the potential to obscure the topology of trees. To test this possibility, we examined trees that included the additional haplotypes observed in samples of 40 to 126 indi- viduals each from S. caurinus (n=79),S. cileutianus (u = 126), and S. borealis 0?=40) (data not shown). The additional haplotypes (5, 13, and 5, respectively) increased the num- ber of branches at the tip of the species limbs but did not in- fluence or obscure relationships with other species. We are currently investigating the population structure of S. aleu- tianus, S. borealis, S. alutus, S. caurinus, and Sebastolobus alascanus by using mtDNA restriction site variation. Because of the similarity of many Sebastes species, there is a chance that very similar species can be misidentified. In fact, a young dusky rockfish (S. ciliatus) and a young yellowtail rockfish {S. flavidus ) were misidentified in the field as black rockfish (S. melanops) prior to our mtDNA analysis. Also, it is possible that closely related species may hybridize (e.g. Seeb, 1998). Because hybrids carry on- ly the maternal lineage and because only the maternal contributor can be identified, mtDNA analysis is a poor tool for identifying hybrids. In addition to providing a tool that can distinguish among a variety of rockfish species, the data appear to provide criteria that may prove useful in unraveling some questions about rockfish systematics. Both outgroups are distinct from Sebastes ; H. hilgendorfi is more closely re- lated than Sebastolobus alascanus. The 15 Sebastes spe- cies studied include eight subgenera, five of which were represented by two or more species. Despite the uncertain- ty in some of the subgenus assignments,2 our analyses of mtDNA restriction sites show some concordance with sub- generic assignments. Unfortunately, the only recently re- viewed subgenus is Sebast077ius (Chen, 1971), for which we have only a single representative ( S . lielvo7naculatus). A phylogeny of subgenera is unavailable. Several species pairs were persistent in the analyses. Within Sebastes, S. tnaliger and S. caurinus (subgenus Pteropodus) were distinct from the other Sebastes species. Gharreft et al.: Identification of Sebastes spp. by restriction site analysis 55 Table 4 Differences between haplotypes (see Table 3 and Appendix 1) of rockfish (Sebastes, Helicolenus, and Sebastolobus spp.). Above the diagonal are the number of restriction site differences. Below the diagonal are estimates of evolutionary differences (nucleotide substitutions per site; Nei and Tajima 1981; Nei and Miller 1990). An average of 79.3 sites and 332.05 bases were examined for each haplotype (McElroy et al.1990). Species la lb 2a 2b 3 4a 4b 5a 5b 6 la 1 14 14 21 13 13 16 15 16 lb 0.0015 15 15 22 14 14 17 16 17 2a 0.0224 0.0243 2 19 9 9 14 13 14 2b 0.0222 0.0240 0.0031 19 9 9 14 13 14 3 0.0335 0.0354 0.0308 0.0302 16 18 21 20 21 4a 0.0208 0.0226 0.0143 0.0141 0.0253 2 11 10 13 4b 0.0210 0.0229 0.0145 0.0143 0.0290 0.0031 11 10 13 5a 0.0280 0.0279 0.0225 0.0222 0.0334 0.0175 0.0177 1 6 5b 0.0242 0.0260 0.0207 0.0204 0.0315 0.0158 0.0160 0.0015 5 6 0.0267 0.0286 0.0230 0.0227 0.0340 0.0213 0.0216 0.0094 0.0078 7a 0.0192 0.0210 0.0194 0.0191 0.0306 0.0246 0.0249 0.0260 0.0242 0.0265 7b 0.0210 0.0229 0.0212 0.0210 0.0325 0.0265 0.0269 0.0280 0.0260 0.0285 7c 0.0174 0.0192 0.0176 0.0174 0.0287 0.0227 0.0230 0.0242 0.0223 0.0246 7d 0.0210 0.0228 0.0213 0.0210 0.0327 0.0264 0.0268 0.0279 0.0260 0.0284 8 0.0192 0.0210 0.0194 0.0191 0.0272 0.0211 0.0214 0.0226 0.0208 0.0231 9 0.0205 0.0223 0.0207 0.0204 0.0283 0.0190 0.0193 0.0206 0.0188 0.0243 10a 0.0223 0.0242 0.0295 0.0256 0.0336 0.0208 0.0211 0.0223 0.0239 0.0296 10b 0.0238 0.0257 0.0310 0.0272 0.0351 0.0224 0.0227 0.0206 0.0221 0.0312 11a 0.0261 0.0281 0.0333 0.0293 0.0339 0.0280 0.0283 0.0333 0.0313 0.0376 11b 0.0277 0.0296 0.0313 0.0274 0.0320 0.0260 0.0264 0.0313 0.0294 0.0356 12 0.0208 0.0226 0.0209 0.0206 0.0285 0.0192 0.0194 0.0210 0.0192 0.0216 13 0.0190 0.0208 0.0329 0.0325 0.0302 0.0277 0.0281 0.0336 0.0317 0.0381 14 0.0339 0.0324 0.0342 0.0337 0.0383 0.0324 0.0328 0.0270 0.0285 0.0308 15a 0.0321 0.0340 0.0325 0.0321 0.0075 0.0272 0.0311 0.0355 0.0336 0.0361 15b 0.0351 0.0371 0.0356 0.0351 0.0104 0.0303 0.0342 0.0385 0.0366 0.0391 18 0.0740 0.0767 0.0835 0.0779 0.0745 0.0724 0.0735 0.0827 0.0843 0.0885 17a 0.1043 0.1028 0.1148 0.1084 0.1043 0.0981 0.1041 0.0996 0.1011 0.1055 17b 0.1055 0.1041 0.1161 0.1097 0.1057 0.0995 0.1055 0.1010 0.1025 0.1070 17c 0.1031 0.1016 0.1136 0.1071 0.1029 0.0968 0.1029 0.0983 0.0998 0.1043 17d 0.1028 0.1013 0.1133 0.1069 0.1029 0.0966 0.1028 0.0981 0.0996 0.1040 Species 7a 7b 7c 7d 8 9 10a 10b 11a 11b la 12 13 11 13 12 13 14 15 16 17 lb 13 14 12 14 13 14 15 16 17 18 2a 12 13 11 13 12 13 18 19 20 19 2b 12 13 11 13 12 13 16 17 18 17 3 19 20 18 20 17 18 21 22 21 20 4a 15 16 14 16 13 12 13 14 17 16 4b 15 16 14 16 13 12 13 14 17 16 5a 16 17 15 17 14 13 14 13 20 19 5b 15 16 14 16 13 12 15 14 19 18 8 18 17 15 17 14 15 18 19 22 21 7a 1 1 1 2 7 14 15 14 13 7b 0.0016 2 2 3 8 15 16 15 14 7c 0.0015 0.0031 2 1 6 13 14 13 12 7d 0.0016 0.0032 0.0032 3 8 15 16 15 14 8 0.0031 0.0047 0.0015 0.0048 5 12 13 12 11 9 0.0109 0.0126 0.0093 0.0126 0.0078 9 10 11 10 continued 56 Fishery Bulletin 99(1 ) None of the other subgenera were as coherent. The hap- lotypes of S. zacentrus and S. variegatus (subgenus Allo- sebastes) were very similar and the haplotype of a third member, S. proriger , generally clustered nearby. Similarly, the haplotypes of S. maliger and S. flavidus (subgenus Se- bastosomus) were tightly clustered, but the branch for the haplotype of the third member, S. ciliatus, was distal; and different tree construction methods inconsistently placed Table 4 (continued) Species 7a 7b 7c 7d 8 9 10a 10b 11a lib 10a 0.0227 0.0246 0.0208 0.0245 0.0193 0.0140 1 14 13 10b 0.0243 0.0262 0.0224 0.0260 0.0208 0.0155 0.0015 15 14 11a 0.0229 0.0248 0.0210 0.0247 0.0195 0.0174 0.0226 0.0242 1 lib 0.0210 0.0229 0.0192 0.0228 0.0177 0.0157 0.0208 0.0223 0.0015 12 0.0143 0.0161 0.0126 0.0160 0.0111 0.0093 0.0174 0.0189 0.0211 0.0193 13 0.0227 0.0246 0.0243 0.0245 0.0227 0.0207 0.0292 0.0308 0.0266 0.0282 14 0.0239 0.0258 0.0221 0.0258 0.0206 0.0122 0.0236 0.0252 0.0274 0.0255 15a 0.0292 0.0311 0.0272 0.0311 0.0257 0.0269 0.0322 0.0338 0.0361 0.0340 15b 0.0322 0.0342 0.0303 0.0342 0.0288 0.0299 0.0353 0.0368 0.0391 0.0371 16 0.0795 0.0823 0.0811 0.0821 0.0839 0.0843 0.0740 0.0756 0.0801 0.0817 17a 0.1151 0.1136 0.1166 0.1186 0.1151 0.1101 0.0996 0.1011 0.1014 0.1029 17b 0.1164 0.1149 0.1179 0.1199 0.1164 0.1115 0.1010 0.1025 0.1031 0.1046 17c 0.1140 0.1124 0.1155 0.1173 0.1140 0.1090 0.0983 0.0998 0.1002 0.1017 17d 0.1136 0.1120 0.1151 0.1170 0.1136 0.1086 0.0981 0.0996 0.0999 0.1014 Species 12 13 14 15a 15b 16 17a 17b 17c 17d la 13 12 21 20 22 41 58 59 57 57 ib 14 13 20 21 23 42 57 58 56 56 2a 13 20 21 20 22 45 62 63 61 61 2b 13 20 21 20 22 43 60 61 59 59 3 18 19 24 5 7 42 59 60 58 58 4a 12 17 20 17 19 40 55 56 54 54 4b 12 17 20 19 21 40 57 58 56 56 5a 13 20 17 22 24 45 56 57 55 55 5b 12 19 18 21 23 46 57 58 56 56 6 13 22 19 22 24 47 58 59 57 57 7a 9 14 15 18 20 43 62 63 61 61 7b 10 15 16 19 21 44 61 62 60 60 7c 8 15 14 17 19 44 63 64 62 62 7d 10 15 16 19 21 44 63 64 62 62 8 7 14 13 16 18 45 62 63 61 61 9 6 13 8 17 19 46 61 62 60 60 10a 11 18 15 20 22 41 56 57 55 55 10b 12 19 16 21 23 42 57 58 56 56 11a 13 16 17 22 24 43 56 57 55 55 lib 12 17 16 21 23 44 57 58 56 56 12 13 14 17 19 44 61 62 60 60 13 0.0212 19 18 20 43 62 63 61 61 14 0.0221 0.0307 23 25 50 59 60 58 58 15a 0.0271 0.0288 0.0369 2 41 62 63 61 61 15b 0.0301 0.0317 0.0399 0.0030 43 64 65 63 63 16 0.0811 0.0786 0.0928 0.0734 0.0765 61 62 60 60 17a 0.1117 0.1135 0.1055 0.1118 0.1147 0.1145 1 3 1 17b 0.1131 0.1150 0.1070 0.1131 0.1160 0.1158 0.0014 4 2 17c 0.1105 0.1124 0.1045 0.1105 0.1135 0.1129 0.0043 0.0057 4 17d 0.1102 0.1120 0.1040 0.1104 0.1133 0.1130 0.0014 0.0029 0.0058 Gharrett et al.: Identification of Sebastes spp. by restriction site analysis 57 S. ciliatus on the tree (not shown). Haplotypes of S. aleu- tianus and S. borealis (subgenus Zalopyr) were found in the same general region of the tree, but are not sister taxa. Likewise, the two representatives of Acutomentum, S. alutus and S. brevispinis, were not monophyletic sister taxa. Disparities, such as we observed between relation- ships of haplotype and assignments of subgenera, have also been reported for allozyme comparisons (Seeb, 1986) and mtDNA cytochrome b sequences (Johns and Avise, 1998; Rocha-Olivares, 1998a; Rocha-Olivares et al., 1999a, 1999b). The members of subgenera Acutomentum and Al- losebastes, in particular, seem discordant with trees. It is important to recall that the systematics is not unequivo- cal and controversies date back more than a century (e.g. Cramer, 1895). Therefore, discrepancies between the mo- lecular-based comparisons and current systematic place- Maximum likelihood Maximum parsimony LS. alutus A Acutomentum S. alutus B Acutomentum S ciliatus Sebastosomus S aleutianus A Zalopyr , S. aleutianus B Zalopyr p* S ruberrimus A Sebastopyr S ruberrimus B Sebastopyr S borealis Zalopyr S brevispinis Acutomentum S . proriger Allosebastes S. zacentrus A Allosebastes S zacentrus D Allosebastes S zacentrus B Allosebastes —1 so S. zacentrus C Allosebastes 1 1 S. variegatus Allosebastes , S. helvomaculatus A Sebastomus , S helvomaculatus B Sebastomus Jj°l 90 . S babcocki A Rosicola S. babcocki B Rosicola S. melanops A Sebastosomus S. melanops B Sebastosomus S flavidus Sebastosomus S. maliger Pteropodus S. caurinus A Pteropodus S caurinus B Pteropodus H. hilgendorfi Sebastolobus alascanus A Sebastolobus alascanus B Sebastolobus alascanus D Sebastolobus alascanus C _L 0.10 0.05 0 Nucleotide substitutions Figure 1 Maximum likelihood (PHYLIP; Felsenstein3) and maximum parsimony consensus trees (PAUP ; Swofford, 1998) trees estimated from restriction site data for haplotypes of 15 Sebastes spp., Helicolenus hilgendorfi , and Sebastolobus alascanus. Haplotype and restriction site data are provided in Appendix 1 and Table 3, respec- tively. On the maximum likelihood tree, and indicate distances that are significantly positive (P<0.05 and P<0.01, respectively). On the maximum parsimony consensus tree, the numbers at nodes indicate the percent- age of trees that included those branches. 58 Fishery Bulletin 99(1) ments do not necessarily discredit the validity of the mo- lecular comparisons. Use of restriction site data in mtDNA holds promise for the identification and systematics of Sebastes and sug- gests the possibility of applications for stock identifica- tion. Larval and juvenile rockfish carry mtDNA that is ad- equate for PCR amplification (e.g. see Seeb and Kendall, 1991; Rocha-Olivares 1998b). Combining molecular iden- tification with morphometry may solve many of the prob- lems of identification that accompany rockfish studies. The apparent coherence of closely related rockfish species that we observed in both cladistic and phenetic analyses sug- gests that we should focus our applications on groups of species that are presumed to be close relatives. The con- sensus tree depicting relationships among interior clades within the Sebastes parsimony tree did not unequivocally position those clades either in this study or analyses of the cytochrome b region (Johns and Avise, 1998; Rocha-Oliva- res, 2000). Consequently, determination of higher level re- lationships among Sebastes requires analysis of additional mtDNA regions. Moreover, because the divergence of mtD- NA sequences provides only one perspective of the evolu- tion of Sebastes divergence, the relationships inferred by mtDNA analyses must be corroborated by analysis of the interspecific divergence of nuclear genes. Acknowledgments We gratefully acknowledge the many crew members and scientists aboard the research vessels John N. Cobb and Miller Freeman who participated in collecting specimens for our study. L. Densmore and T. Dowling provided con- structive comments on early drafts of this manuscript. Three anonymous reviewers provided constructive com- ments. A.W. Kendall Jr. and M. S. Love contributed advice and insight that helped us develop this paper. J. A. Gharrett and D. Churikov reviewed drafts of this paper. This work was supported by the National Marine Fisheries Service Auke Bay Laboratory and the U.S. Geological Services (Bio- logical Resources Division) Western Regional Office in Seattle, WA (R.W.O. 32). Literature cited Anderson, S., A. T. Bankier, G. T. Barrell, M. H. L. DeBruijn, and A. R. Coulson. 1981. Sequence and organization of the human mitochondrial genome. Nature 290:457-465. Anderson, S., M. H. L. DeBruijn, A. R. Coulson, I. C. Eperon, and F. Sanger. 1982. Complete sequence of bovine mitochon- drial DNA. J. Mol. Biol. 56:683-717. Chang, Y-S., F-L. Huang, and T-B. Lo. 1994. The complete nucleotide sequence and gene organization of carp (Cyprinus carpio ) mitochon- drial genome. J. Mol. Evol. 38:138-155. Chen, L.-C. 1971. Systematics, variation, distribution, and biology of rockfishes of the subgenus Sebastomus (Pices, Scorpaenidae, Sebastes). Bull. Scripps Inst. Ocean. Univ. Cal., Univ. Cal. Press, Berke- ley and Los Angeles, CA, 115 p. Cramer, F. 1895. On the cranial characters of the genus Sebas- todes (rock-fish). Contributions to biology from the Hopkins Seaside Laboratory, Leland Stanford Junior University Publications. Reprinted from Proe. Cal. Acad. Sci., series 2, vol. V, 54 p. Cronin. M. A., W. J. Spearman, R. L. Wilmot, J. C. Patton, and J. W. Bickham. 1993. Mitochondrial DNA variation in Chinook ( Oncorhynchus tshawytscha ) and chum (O. beta) detected by restriction enzyme analysis of poly- merase chain reaction (PCR) products. Can. J. Fish. Aquat. Sci. 30:708-715. Dowling, T. E., W. R. Hoeh, G. R. Smith, and W. M. Brown. 1992. Evolutionary relationship of shiners in the Gharrett et al: Identification of Sebostes spp. by restriction site analysis 59 genus Luxilus (Cyprinidae) as determined by analysis of mitochondrial DNA. Copeia 1992:306-322. Eschmeyer, W. N. (ed.). 1998. Catalog of fishes. California Academy of Sciences, San Francisco, CA, 2905 p. Gyllensten, U., D. Wharton, A. Josefsson, and A. C. Wilson. 1991. Paternal inheritance of mitochondrial DNA in mice. Nature 352:255-257. Hart, J. L. 1973. Pacific fishes of Canada. Bulletin 189, Fisheries Research Board of Canada, Ottawa, 740 p. Ishida, M. 1984. Taxonomic study of the sebastine fishes in Japan and its adjacent waters. Master’s thesis, Hokkaido Univ., Hakodate, Hokkaido, Japan, 267 p. Johns, G. C., and J. C. Avise. 1998. Tests for ancient species flocks based on molecular phylogenetic appraisals of Sebastes rockfishes and other marine fishes. Evolution 52(4 ):1 135-1 146. Johnson, A. G., F. M. Utter, and H. O. Hodgins. 1972. Electrophoretic investigation of the family Seorpaeni- dae. Fish. Bull. 70:403-413. Jordan, D. S., B. W. Evermann, and H. W. Clark. 1930. Check-list of the fishes and fish-like vertebrates of North and Middle America north of the northern boundary ofVenezuela and Columbia. U. S. Com. Fish., Rep. 1928, part 2, 670 p. Kendall, A. W., Jr. 199 1 . Systematics and identification of larvae and juveniles of the genus Sebastes. Environ. Biol. Fish. 30:173-190. Kramer, D. E., and V. M. O’Connell. 1988. Guide to Northeast Pacific rockfishes genera Sebastes and Sebastolobus. Alaska Sea Grant Advisory Bulletin 25, Alaska Sea Grant College Program, Univ. Alaska, Fair- banks, 78 p. Matsubara, K. 1943. Studies on the scorpaenoid fishes of Japan: anatomy, phylogeny, and taxonomy. Trans. Sigenkagaku Kenky- usyo 1, 490 p. McElroy, D. M., P. Moran, E. Bermingham, and I. Kornfield. 1990. REAP: an integrated environment for the manipu- lation and phylogenetic analysis of restriction data. J. He red. 83:157-158. Meyer, A. 1993. Evolution of mitochondrial DNA in fishes. In Bio- chemistry and molecular biology in fishes, vol. 2: Molecular biology frontier (P. W. Hochachka and T. P. Mommsen, eds.), p. 1-38. Elsevier, Amsterdam. Moser, H. G. 1996. Scorpaenidae: scorpionfishes and rockfishes. In Cal- ifornia Cooperative Oceanic Investigations, Atlas 33: the early stages of the fishes in the California Current region (H. G. Moser, ed.), p. 733-795. Allen Press, Lawrence, KS. Nei, M. 1987. Molecular Evolutionary Genetics. Columbia Univ. Press, New York, NY, 512 p. Nei, M., and J. C. Miller. 1990. A simple method for estimating average number of nucleotide substitutions within and between populations from restriction data. Genetics 97:145-163. Nei, M. and F. Tajima. 1983. Maximum likelihood estimation of the number of nucleotide substitutions from restriction sites data. Genet- ics 105: 207-217. Phillips, J. B. 1957. A review of the rockfishes of California (Family Scor- paenidae). State of California, Department of Fish and Game, Marine Fisheries Branch, Fish Bull. 104, 158 p. Richardson, J. 1845. Ichthyology, part 2. In The zoology of the voyage of H.M.S. Sulphur, under the command of Captain Sir Edward Belcher, during the years 1836-1842, no. 9 (R. B. Hinds, ed.), p. 71-98, pis. 55-64. Smith, Elder & Co., London. [Cited in Eschmeyer (1998).] Rocha-Olivares, A. 1998a. Molecular evolution, systematics, zoogeography, and levels of intraspecific genetic differentiation in the species of the antitropical subgenus Sebastomus, Sebastes (Scor- paeniformes,Teleostei) using mitochondrial DNA sequence data. Ph.D. diss., Univ. Cal., San Diego, CA, 240 p. 1998b. Multiplex haplotype-specific PCR: a new approach for species identification of the early life stages of rock- fishes of the species-rich genus Sebastes Cuvier. J. Exper. Mar. Biol. Ecol. 231:279-290. Rocha-Olivares, A., H. G. Moser, and J. Stannard. 2000. Molecular identification and description of pelagic young of rockfishes Sebastes constellatus and Sebastes ensi- fer. Fish. Bull. 98:353-363. Rocha-Olivares, A., R. H. Rosenblatt, and R. D. Vetter. 1999a. Cryptic species of rockfishes (Se6astes:Scorpaenidae) in the southern hemisphere inferred from mitochondrial lineages. J. Hered. 90: 404-411. 1999b. Evolution of a mitochondrial cytochrome b gene sequence in the species-rich genus Sebastes (Teleostei, Scoraenidae) and its utility in testing monophyly of the sub- genus Sebastomus. Mol. Phylogenet. Evol. 11: 426-440. 1999c. Molecular evolution, systematics, and zoogeography of the rockfish subgenus Sebastomus ( Sebastes , Scorpae- nidae) based on mitochondrial cytochrome b and control region sequences. Mol. Phylogenet. Evol. 11: 441-458. Rocha-Olivares, A., and R. D. Vetter. 1999. Effects of oceanic circulation on the gene flow, genetic structure, and phylogeography of the rosethorn rock- fish ( Sebastes helvomaculatus). Can. J. Fish. Aquat. Sci. 56:803-813. Roe, B. A., D-P. Ma, R. K. Wilson, and J. F-H. Wong. 1985. The complete nucleotide sequence of the Xenopus lae- vis mitochondrial genome. J. Biol. Chem. 260:9759-9774. Saitoh, N., and M. Nei. 1987. The neighbor-joining method: a new method for recon- structing phylogenetic trees. Mol. Biol. Evol. 4:406-425. Seeb, L. W. 1986. Biochemical systematics and evolution of the scorpae- nid genus Sebastes. Ph.D. diss., Univ. Washington, Seat- tle, WA, 176 p. 1998. Gene flow and introgression within and among three species of rockfishes, Sebastes auriculatus, S. caurinus , and S. maliger J. Hered. 89:393-404. Seeb, L. W., and A. W. Kendall Jr. 1991. Allozyme polymorphisms permit the identification of larval and juvenile rockfishes of the genus Sebastes. Envi- ron. Biol. Fish. 30:191-201. Seeb, L. W., F. M. Utter, and D. R. Gunderson. 1988. Genetic variation and population structure of Pacific ocean perch ( Sebastes alutus). Can. J. Fish. Aquat. Sci. 45:78-88. Seutin, G., B. N. White, and P. T. Boag. 1991. Preservation of avian blood and tissue samples for DNA analysis. Can. J. Zool. 69:82-90. Swofford, D. L. 1998. PAUP . Phylogenetic analysis using parsimony ( and other methods), version 4. Sinauer Associates, Sunderland, MA. 60 Fishery Bulletin 99(1) Swofford, D. L., G. J. Olsen, P. J. Waddell, and D. M. Hillis. 1996. Phylogenetic inference In Molecular systematics, 2nd ed. (D. M. Hillis, C. Moritz, and B. K. Mable, eds.), p. 407-514. Sinauer Assoc., Sunderland, MA. Thomas, W. K., and A. T. Beckenbach. 1989. Variation in salmonid mitochondrial DNA: evolution- ary constraints and mechanisms of substitution. J. Mol. Evol. 29:233-245. Tsuyuki, H., E. Roberts, R. H. Lowes, W. Hadaway, and S. J. Westrheim. 1968. Contribution of protein electrophoresis to rockfish (Scor- paenidae) systematics. J. Fish. Res. Board Can. 25:2477- 2501. Wallace, D. W. 1987. Large- and small-scale phenol extractions. In Meth- ods in enzymology, vol. 12: Guide to molecular cloning tech- niques (S. L. Berger and A. R. Kimmel, eds.), p. 33-41. Academic Press, San Diego, CA. Wishard, L. N., F. M. Utter, and D. R. Gunderson. 1980. Stock separation of five rockfish species using natu- rally occurring biochemical genetic markers. Mar. Fish. Rev. 42:64-73. Zardoya, R., A. Garrido-Pertierra, and J. M. Bautista. 1995. The complete nucleotide sequence of the mitochon- drial DNA genome of the rainbow trout, Oncorhynchus mykiss. Mol. Evol. 41:942-951. Gharrett et al.: Identification of Sebastes spp by restriction site analysis 61 Appendix 1 Restriction site locations for Sebastes spp., Helicolenus hilgendorfi, and Sebastolobus alascanus in the 12S/16S and ND3/ND4 mtDNA regions. The Sebastes species are listed in Table 2; five individuals of each species are represented. Sites were mapped by double digests. Haplotypes for each restriction endonuclease are presented for each mtDNA region. X’s denote presence and O’s denote absence of a site. (X)’s are sites that occur in the primers and were present in all PCR products. They were not used in the analysis. 12S/16S Msp I Cfo I Rsa I hapotypes Mbo I haplotypes haplotypes Dde I haplotypes haplotypes sites A B c D E sites A B C D E F sites A B sites A B C D E F sites A B 293 O O o 0 X 201 O X X O o 0 65 X X 44 X X X X X X 537 X X 507 O O X X o 849 X X X X o X 766 X X 55 X X X X X X 602' X O 588 O O 0 0 X 1015 X X X X X 0 1259 O X 976 X X X X X 0 665' X X 761 X X X X X 1403 X X X X X X 1390 X X 1043 X X O X X o 1898 X X 950 o O o o X 1507 X O X X X X 1535 X X 1056 X X 0 X 0 0 2268 X X 1000 X X X X 0 1746 O O O X 0 0 2226 X X 1304 X X X X X 0 1071 X X X X 0 1984 X X X X X X 2403 X X 1735 o X X X X 0 1263 o 0 0 0 X 2059 X X X X X X 2181 X O X X X 0 1308 0 X 0 X X 2228 X X X X X X 2393 X X X X X X 1358 X X X X X 2318 X X X X X X 2164 0 o 0 o X 2388 X X X X X X Sty I BstC I Hin fl haplotypes Hin d II haplotypes haplotypes Bst N I haplotypes haplotypes sites A B c D sites A B sites A B sites A B c D sites A B 982 O O X 0 1717 X O 295 X O 326 X X X X 87 X X 1291 X X X o 2053 O X 988 O o 0 X 600' X O 2094 0 X X o 2254 X X 1687 O o X 0 664' X X 1741 O X 0 0 729 X X 2416 (X) (X) (X) X) 1872 X X 2128 X X ND3/ND4 Cfo I haplotypes Bst U I haplotypes Hin fl haplotypes sites A B c D E F sites A B c D E F G sites A B C D E F G H 709 X O X X X X 4 (X) (X) (X) (X) (X) (X) (X) 130 O X O O o X O O 1221 o O o o X X 344 O 0 0 o 0 X o 389 O X X O 0 0 o X 1436 o O 0 o X O 1499 O 0 0 0 X o 0 494 X X X X X X X X 1741 0 X X X 0 O 1854 O 0 o X o 0 0 853 O O O O 0 0 o X 1813 o O X 0 0 O 2025 O X X 0 o 0 0 1448 O O O O 0 0 0 X 2306 O X 0 0 X X X 1537 O O O 0 o 0 X O 1755 O O O 0 X X 0 O 1888 O O O X 0 0 0 X 2232 X X X X X X X X Sty I haplotypes Msp I haplotypes Rsa I haplotypes sites A B c D E F sites A B c D E F G sites A B C D E F G H 49 X X X X X X 30 X X X X X X X 346 O O O X X O X X 194 O O o X X X 933 O O 0 0 o X O 742 O O O O 0 0 O X 365 X X X X 0 O 1199 O O 0 0 o X O 1077 O O O O 0 X 0 O 391 O O 0 0 X X 1261 O O 0 o 0 O X 1231 O O O o 0 0 X O 534 X O X X X X 1738 X X X X X X X 1492 O O O 0 0 0 0 X 1258 0 0 o o 0 X 1826 X X 0 X X 0 0 1863 O O X 0 X 0 0 X 1570 o X X X X X 1844 O O o X X o o 1985 O X X X X X X X 2311 X X X X X X 2073 0 o 0 0 o X o 2110 O X X 0 X o o continued 62 Fishery Bulletin 99(1 ) 63 Abstract— Visual estimates of reef- fish length are a nondestructive and useful way of determining the biomass, mean length, or length frequency of reef fish. Consequently, visual estimates of reef-fish length are often an important component of reef-fish monitoring pro- grams, many of which increasingly use volunteers. We compared estimates of the length of plastic fish silhouettes determined visually by experienced sci- entific and novice SCUBA divers. Novice divers showed a similar level of accu- racy ( mean error: 2.3 cm ) to that of expe- rienced scientific divers (mean error: 2.1 cm). Significant improvements in accuracy and precision were provided by a stereo-video system (mean error: 0.6 cm). After minimal training in the use of hardware and software, vol- unteers can obtain a high degree of measurement accuracy and precision with a stereo-video system, allowing them to assist with monitoring reef-fish lengths. Manuscript accepted 14 July 2000. Fish. Bull. 99:63-71 (2001). A comparison of the precision and accuracy of estimates of reef-fish lengths determined visually by divers with estimates produced by a stereo-video system Euan Harvey Department of Marine Science University of Otago 304 Castle Street Dunedin, New Zealand Present address: Department of Botany University of Western Australia Western Australia 6907, Australia E-mail address: euanh@cyllene.uwa.edu.au David Fletcher Department of Mathematics and Statistics University of Otago 539 Castle Street Dunedin, New Zealand Mark Shortis Department of Geomatics University of Melbourne Parkville, Melbourne Victoria 3052, Australia The marine environment is affected by many anthropogenic activities (e.g. the exploitation of marine resources for food, medicine, and curios) with the result that many environments are compromised by siltation and pollution. Reef fish, an important commercial, recreational, and cultural resource, are likewise being exploited at an increas- ing rate. Consequently, it is important that environmental managers and sci- entists have objective evidence of the magnitude and effect of these impacts. Data on the community structure of reef-fish assemblages, including precise and repeatable estimates of the length frequency and abundance of individual species, provide this information. Infor- mation on the length frequency of a fish population when linked with even a rudimentary knowledge of the biol- ogy of the species may allow estimates of recruitment to the adult population, fishing intensity, and rates of recovery from fishing (McCormick and Choat, 1987) or other impacts. Accurate and precise length and abundance data are difficult to obtain because fish occupy different habitats and display varying behavior over a range of spatial and temporal scales. Environmental surveys are commonly used to determine the abundance and length frequency of reef-fish assemblag- es with SCUBA divers who count and visually estimate the length of individ- ual reef fish (Jones and Chase, 1975; Harmelin-Vivien and Bouchon-Navaro, 1981; Russ, 1985; Bellwood, 1988; Bell- wood and Alcala, 1988; Kulibicki, 1989; Samoilys, 1989; Francour, 1991, 1994). Visual census techniques have many advantages, compared with other sam- pling techniques, in that they are quan- titative, quick, nondestructive, and re- peatable (English et al., 1994). The disadvantages of visual census tech- niques is that the observers undertak- ing the sampling need to be trained and must have experience to identify, count, and estimate the length of reef fish accurately (English et al., 1994). In addition, visual census techniques in- volving SCUBA divers are restricted to shallow depths owing to the constraints of decompression diving. 64 Fishery Bulletin 99(1 ) Two important sources of error affect the accuracy and precision of visual length estimates: 1 The estimation of the length of reef fish underwater is complicated by the air-water interface in the diver’s mask which causes objects to be magnified in size by a factor of 1.3 and to appear to be closer to the observer than they actually are; 2 Researchers using SCUBA have been shown to be inef- ficient when performance underwater is compared with similar activities in air (Hollien and Rothman, 1975). The accuracy and precision of diver’s estimates of reef- fish length is probably affected by the detrimental phys- iological effects related to SCUBA diving (Baddeley, 1965; Baddeley et ah, 1968; Baddeley, 1971). The level of precision and accuracy associated with vi- sual length estimates influences comparisons of data over different temporal or spatial scales in two distinct ways. First, bias in the estimates can make the results of the analysis less reliable. Second, any lack of precision in the estimates arising from both sampling error and measure- ment error tends to reduce the power of the statistical analysis. The Great Barrier Reef Marine Park Authority (GBRMPA, 1979), Bell et al. (1985), and Polunin and Rob- erts (1993) have performed calibration studies in order to quantify measurement error and have concluded that such error is relatively small and can be ignored. However, they did not directly assess the extent to which such error reduc- es the precision of a statistical analysis. Harvey and Shortis (1996) demonstrated that subjectivity in visual length es- timates can be overcome, and the accuracy and precision of length estimates enhanced, by using a simple and rela- tively inexpensive underwater stereo-video system. The objectives of our study were 1) to compare the ac- curacy and precision of stereo-video measurements of reef- fish length with visual estimates made by novice and ex- perienced scientific SCUBA divers; 2) to evaluate the ef- fect of operator training and experience on the precision and accuracy of length measurements made with a stereo- video system; and 3) to assess the effect of water clarity on the accuracy and precision of stereo-video measurements. Methods of optimizing the accuracy and precision of ste- reo-video length estimates are discussed and the accura- cy and precision of experienced divers’ estimates are com- pared to one other study. Materials and methods Comparison of length estimates made by novice and experienced scientific divers with estimates generated by a stereo-video system The accuracy and precision of length estimates were tested by a simple procedure that is used routinely for calibrat- ing diver estimates of the lengths of reef fish (GBRMPA, 1979; Bell et al., 1985; English et al., 1994). Typically, poly- vinyl chloride (PVC) sticks or silhouettes of fish are placed in the water and their lengths estimated. The accuracy of the diver estimate is then assessed from the difference between the real size and the estimate. In this study, 16 plastic silhouettes of fish, ranging from 10 to 49 cm in length, were placed at distances of between 3.0 and 6.6 meters from a transect rope. Each diver moved along the transect rope estimating the length of each of the sixteen silhouettes when they were opposite them. Each diver repeated this process five times, i.e. five transects each with estimates of the lengths of the sixteen silhou- ettes. Five transects were completed with the stereo-video system in the same manner. The distances from the tran- sect rope and the order of the individual silhouettes were kept constant throughout the trial. All silhouettes were weighted in a way that their length orientation remained perpendicular to the transect. Therefore, the error mea- sures presented here do not take into consideration that in the field the divers may have to make length estimates where orientation of the live fish in relation to the diver changes. Novice divers Novice divers were defined as experienced SCUBA divers who had made few, if any, estimates of the lengths of reef fish underwater. Eight novice divers made length estimates in a saltwater pool at the Portobello Marine Laboratory (PML) between May 1994 and Janu- ary 1995. No more than two transects were completed on any one day, except one diver who completed four consec- utive transects in one day. Between transects and dives, data were not available to divers in order to avoid memo- rization of previous estimates or of silhouette lengths. The novice divers made only 594 of a possible 640 length esti- mates because 46 silhouettes (7%) were not recorded. The majority of the missing estimates were from the smaller silhouettes (approximately 10 cm long) or from a silhou- ette placed farthest from the transect (6.6 meters). Experienced scientific divers Experienced scientific divers were considered active marine scientists who had been, or who were currently involved in research that required them to make estimates of reef fish length. Six experienced scientific divers estimated the length of the plastic silhou- ettes between October 1994 and June 1996. Three of the divers made their estimates in a saltwater pool at PML, whereas the other three made their estimates in fresh- water swimming pools elsewhere. A total of 480 length estimates were made, and all silhouettes were recorded. Owing to time constraints, all of the length estimates were made on consecutive transects during one dive. Stereo-video measurements For interested readers a comprehensive description of the design and calibration of the stereo-video can be found in Harvey and Shortis ( 1996, 1998) and will not be described here. Stereo-video length estimates were made in the same way as those made by the divers; the stereo-video system recorded the silhouette as it was moved along the tran- sect line by a diver. Measurements were made in the Uni- versity of Melbourne swimming pool in July 1994. 80 sil- houettes were recorded (16 silhouettesx5 transects). Four sets of images were rejected because the orientation of the Harvey et a! : Comparison of estimates of reef-fish lengths made by divers and a stereo-video system 65 silhouette to the stereo-video rig was greater than 50 de- grees, resulting in a deterioration of the accuracy and pre- cision of measurements (Harvey and Shortis, 1996). Thus, 76 pairs of images were retained for analysis, from each of which ten measurements were made. To test whether measurements made by an inexperi- enced volunteer using the stereo-video software would dif- fer from those made by an experienced operator, two vol- unteers were given brief instructions on how to operate the stereo-video software and were asked to make mea- surements of the silhouettes. Measurements made by the inexperienced operators were then compared with those made from the same images by an experienced operator To determine whether changes in the water clarity af- fect the accuracy and precision of length estimates made by the stereo-video system, a further three transects were sampled in a saltwater swimming pool at the PML during August 1994. Water clarity was measured with a tape mea- sure to record the distance to the farthest, clearly visible silhouette. A silhouette was clearly visible at up to 25 me- ters in the swimming pool at the University of Melbourne and up to 5.5 meters in the saltwater pool at PML. We expected that the accuracy of measurements ob- tained with the stereo-video system would improve with operator experience. To evaluate the effect of experience, the same operator remeasured the images recorded in the swimming pool in Melbourne, approximately one year af- ter the initial analysis. In the intervening period the op- erator routinely used the stereo-video system for analysis of fish lengths and was by far the most experienced person using the system. Analysis of data Four measures of error were used to summarize the accu- racy of each length estimate. If the observed fish length is O and the true length is T, these four measures are Error: E = O - T\ Relative error: RE = (O - T)/T = E!T\ Absolute error: AE = j O - T | = | E | ; Relative absolute error: RAE = \ 0 - T\/T = \E \ IT = \ RE \ . The four measures provide different types of information. The error E will be positive or negative according to whether the observed length is an overestimate or an underestimate. If the mean of E is close to zero, the reason might be that all the estimates are accurate or that some are overestimates and others are underestimates to approximately the same degree. The absolute error AE ignores the direction of the error, and thus would provide different mean values for these two scenarios. The rela- tive errors RE and RAE are of interest because it might be expected that these would be consistent across a range of fish lengths. The measure of accuracy used by St John et al. (1990) was simply RE + 1. The results for the novice and experienced scientific div- ers were summarized prior to further analysis by calcu- lating, for each silhouette and each observer, the mean length estimate over all transects. The repeat transects were therefore used solely to provide a reliable estimate of the error made by each observer on each silhouette. If the transects were truly independent, use of the mean would tend to provide a conservative estimate of the error a diver would make on one transect. Because the transects were swum in quick succession, they were not independent and therefore the degree to which their error was underesti- mated should be small. Each mean length estimate was then used to derive the four summaries of error, for each silhouette and each observer. For each measure of error, a two-factor analysis of variance was performed, the factors being type of observer and silhouette length. The individual observers provided the replication needed for this analysis. To summarize the estimates made by using the stereo- video, the mean length estimate over all ten measure- ments was calculated (for each silhouette, transect, and operator). The mean of these over all five transects was then calculated for each silhouette and operator and con- verted to the four summary measures of error. For each measure of error, a one-factor analysis of variance was per- formed, with silhouette length being the factor. The indi- vidual operators (one experienced and two inexperienced) provided the replication needed for this analysis. The two types of diver data and all video data were ana- lyzed separately, rather than in a single analysis of vari- ance because for each silhouette the variation between video operators was found to be much lower than that between experienced scientific divers and novice divers, making the usual assumption of equal variance invalid (Underwood, 1981). Comparison of the video technique with that of the experienced scientific divers and novice divers was therefore made by comparing 95% confidence intervals. Tbe purpose of the analysis of variance for the video data was to provide an estimate of the standard er- ror associated with the mean (for each measure of error) over all silhouettes. To compare the data of experienced and inexperienced stereo-video operators, the mean estimated length over the ten repeat measurements obtained per image was calcu- lated. These means were then converted to the four mea- sures of error. Each measure of error was then analyzed by using a two-factor analysis of variance, with operator and silhouette as the factors. In this analysis, therefore, unlike those for comparing estimates from divers with those pro- duced with the stereo-video system, the transects provided replication against which to compare operators. This anal- ysis is reasonable, because in comparing the operators, the transects can be regarded as independent. To compare the use of a stereo-video system under the two water-clarity conditions, a three-factor analysis of variance was performed on the individual errors. In this analysis, the factors were water clarity, silhouette, and transect (nested within each combination of water clarity and silhouette). Prior to this analysis, the estimates for one of the silhouettes in 5.5-m water clarity were removed because this silhouette was 6.6 meters from the transect and could not be seen clearly. For the 1994 and 1995 stereo-video data used to assess operator experience, a three-factor analysis of variance 66 Fishery Bulletin 99(1 ) was performed on the individual errors, the factors be- ing year, silhouette, and transect (nested within each combination of year and silhouette). With the stereo-video system, the mean of ten mea- surements should result in greater accuracy and pre- cision than use of a single measurement. In order to assess the extent of this improvement, the above anal- ysis was performed a further five times by using the first 1, 2, 4, 6, and 8 measurements. One of the most commonly cited papers on the train- ing of divers to estimate fish lengths with accuracy and precision is that of Bell et al. (1985). The data for this paper originated from a report published by the Great Barrier Reef Marine Park Authority (GBRMPA, 1979). Tables 7, 10, 12, and 15 in that report contain underwater length estimates of pieces of orange PVC conduit (cut into 50 lengths ranging from 6 to 94 cm) determined by experienced scientific divers. Three ex- perienced divers undertook four transects estimating the lengths of the conduit seen. The estimates of a fourth diver are also included in Tables 10, 12, and 15 of the report. We have chosen to disregard this div- er’s data because he was considered inexperienced at the time. We have summarized these results by first calculating the mean length estimate over all four transects for each silhouette and diver. Each of these means was then expressed by using the four summa- ry measures of error described: error, absolute error, relative error, and relative absolute error. For each measure of error, a one-factor analysis of variance was performed, with silhouette being the factor. The indi- vidual divers provided the replication needed for this analysis. Results Comparison of length estimates made by novice and experienced scientific divers with estimates generated by a stereo-video system Figure 1 shows the range of length estimate means over five transects for the novice and experienced scientific divers and for the stereo-video system. The variability of the estimates was greatest for the novice divers and slightly less for the experienced scientific divers. By com- parison, the length estimates made by the stereo-video system showed little variability around the true lengths. The coefficient of variation (CV=standard deviation/ mean) was significantly lower for the stereo-video than it was for either the experienced scientific or the novice div- ers (Fig. 2). For all four measures of error (E, RE, AE, and RAE), there was no significant interaction between type of diver and silhouette size, suggesting that any differences be- tween experienced scientific and novice divers were con- sistent across the silhouettes. For this reason, the results are presented as means across all silhouettes (Fig. 3). For both E and RE, the difference between the experienced sci- entific and novice divers was highly significant (P<0.0001); for AE and RAE, the difference was close to significant at the 5% level (P=0.08 and P= 0.05 respectively; F x gi=5.96 [Ej, =1.75 [AE], =6.28 [RE], =1.94 [RAE]). For both the diver and stereo-video data, there were sig- nificant differences (at the 5% level) between silhouettes, for all measures of error except RAE on the diver data. In- spection of the silhouette means showed no clear pattern for these differences. Any pattern would be difficult to interpret because the silhouettes were placed at different distances from the transect line in order to provide a range of sizes at a range of distances from the transect line and thus make the comparison of means in Figure 3 widely applicable. The GBRMPA divers had a mean measurement error of -2.4 cm (SE=0.2 cm) which is similar to the -2.1 cm (SE=0.6 cm) mean error recorded by the experienced scientific divers used in our study ( Fig. 3 ). Because 26 of the silhouettes used in the GBRMPA study were larger than any of those used in our study, it might be argued that the relative errors are more directly comparable. The mean RE for the GBRMPA divers was -4.6% (SE=0.5%), compared with a mean of -8.6% (SE=1.9%) for the experienced scientific divers used in our study. These results suggest that the experienced divers in the two studies had comparable skills. Harvey et al.: Comparison of estimates of reef-fish lengths made by divers and a stereo-video system 67 40% T 35% 30% 25% 20% 15% 1 0% ~f j 5% + 0% - Xx .$■ | 6 if x Experienced o Novice o Video 0 10 20 30 40 True length (cm) —4 1 50 80 Figure 2 Coefficients of variation between divers or operators of length estimates for experienced scientific divers, novice divers, and a stereo-video system. For all four measures of error, there was no significant interaction between software operators and silhouette, but there was a significant operator effect (Fig. 4). This finding suggests that the differences between operators were consistent across silhouettes. Although the differ- ences between the operators were statistically significant, they were small compared with the differences between divers and the stereo-video system (cf. Figs. 3 and 4). (F9 1fin=6.0, P=0.G03 [E j ; F9 ian =14.6, P is 5%' a> cc 0% GBRMPA £ s V Figure 3 Means and 95% confidence intervals for each measure of error, for each method of estimating fish length. E = experienced divers; N = novice divers; V = stereo-video system, GBPRMA= Great Barrier Reef Marine Park Authority data. of repeat measurements is increased from 1 to 10 (Fig. 7). There is improvement in the error for both inexperienced operators and the experienced operator, even after ten re- peat measurements (Fig. 7). Discussion Our results highlight the differences in the accuracy and precision of length estimates of silhouettes of reef fish made by novice and experienced scientific divers in com- parison with those produced by a stereo-video system. Our length estimates were made under ideal conditions where the plastic models were fixed in position. Under real field conditions fish move, occur at different distances from the divers and have behavioral and morphological differences that can influence length estimates. Consequently, the measures of error presented in our study can most likely be considered a best case scenario. As with Darwall and Dulvy (1996), the results obtained by novice divers were similar to, but slightly less accurate than, those obtained by experienced scientific divers (mean errors of 2.3 cm and -2.1 cm respectively). Our results also demonstrate that significant improvements in accuracy can be obtained by using a stereo-video system (mean error -0.6 cm). Similar improvements in precision were also recorded: mean CVs 68 Fishery Bulletin 99(1 ) 0.0 -0.2 ' -0.4 E -°-6 1 (_> o lu "0.8 -1.0 | -1.2 1 0% T o > o > -1% 2 -2% a5 -3% cc -4% 1 -5% a .. 1.40 1.20 t ! _ 1.00 E 0 0.80 $ cd CD 1 0.60 - o c/) n < 0.40 | 0.20 i 0.00 5% y 4% | o CD § 3% o > 1 2% 1% I- 0% Figure 4 Means and 95% confidence intervals for each measure of error, for each stereo-video operator. 9% | 8% 7% -- 6% 5% + 4% 3% 2% + 1% 0% O o + Good O Poor + <*> % o + + o<^ 0 2 3 True length (cm) Figure 5 Relative absolute error for stereo-video estimates of silhouette length under good and poor water clarity. were 18.2%, 18.6%, and 1.6% for novice scientific divers, experienced scientific divers, and the ste- reo-video system, respectively. The results of the experienced divers’ estimates are similar to those obtained in other published research studies (Bell et ah, 1985). In general, water clarity does not appear to affect the accuracy or precision of the measure- ments made from the stereo-video system. The ac- curacy of the of the stereo-video system is limited to the ability of the operator to accurately point out image locations of interest that are then re- corded to subpixel resolution. Discrete sampling of the CCD (charged coupled device) sensors, com- bined with noise artifacts from the video tape re- cording and frame grabbing, tends to smear the edges of the images and blur details (Shortis et al., 1993). This result is particularly noticeable under recording conditions with good water clar- ity and high contrast where there is a problem with the detection of the edge of the silhouette. The edges or outline of the points of interest (in this instance, the snout and fork of the tail of each silhouette) become significantly blurred within the computer image owing to sampling and noise effects on each pixel. There is a tendency for the observer to select a location inside the true edge of the point of interest because the location most nearly matches the local appearance of the body of the object of interest. This results in the under- estimation of silhouette length, as demonstrated in Figure 1. Under poor water clarity, the edges of the silhouettes have less contrast with the dark background. This lack of contrast permits more accurate pointing to the edges of the object of in- terest because the sampling and noise effects gen- erate a smaller disparity between the real and apparent edges (Fig. 8). Less accurate measure- ments will be made when the contrast becomes so low that the operator cannot discriminate be- tween the object of interest and the background. Hence, where there is sufficient contrast to discriminate the ob- ject from the background, lower contrast will realize more accurate values that ameliorate the underestimation of length. Advantages of stereo-video census techniques The use of a stereo-video system for the measurement of reef-fish length has many advantages. It significantly decreases measurement error and is relatively insensitive to operator experience. The data suggest that the stereo- video system provides far greater accuracy and precision than even experienced scientific divers, and potentially allows inexperienced volunteers to participate in monitor- ing programs without compromising the accuracy or pre- cision of the data collected. This degree of accuracy and precision may be important in research where the objec- tive is to detect small (5-30%) changes in the mean length of a population or assemblage of reef fish, with a high level Harvey et al.: Comparison of estimates of reef-fish lengths made by divers and a stereo-video system 69 of statistical power (Harvey et al., 2000). With minimal training, volunteers can assist with the analysis of images. In addition, a remote stereo- video system can be used to record length-fre- quency data without harm to the fish from far greater depths and over longer periods of time than is possible by employing SCUBA divers. These advantages will likely have applications in fisheries management and deep sea biological surveys. Unlike a diver who often has to make an im- mediate decision on the identity and length of a fish, a stereo-video system observer can review the images later and repeatedly. Multiple imag- es of the same fish are recorded, enabling the se- lection of paired images with the best angle of orientation to the cameras. Moreover, where im- ages of fish are at an acute angle to the cam- eras, the system is still able to make accurate measurements provided that the head and tail of the fish can be seen in both images (Harvey and Shortis, 1996). In addition, once an image is on screen, multiple measurements of one fish can be made over a short period of time (example 10 measurements in 30 seconds), further reduc- ing measurement error. Our results suggest that at least five such measurements yield the most accurate results. Disadvantages of an stereo-video system One of the disadvantages of a stereo-video system is the financial cost of the equipment: two video cameras with underwater housings; a PC com- puter; and a suitable frame grabber required to convert video sequences to the readable format of digital images. Time constraints also need to be considered. Although approximately 3 min- utes are required to complete a calibration in the water, approximately one hour is required in the laboratory to capture the 32 images and process them into a calibration file (Harvey and Shortis, 1998). The major time constraint occurs in the selection and synchroniza- tion of the paired video images from the left and right videotapes. Once the calibration and synchronization pro- cesses are complete, recording of image locations with the measurement observation system is reasonably efficient. The observer visually locates object features of interest, positions a cursor on the feature and clicks the mouse. Some physical constraints also need to be considered with a design like ours. A bar 1.5 m wide is used to separate the video cameras and may become entangled if used in large algal beds such as those of Macrocystis pyrifera. In addi- tion, the underwater video housings create water resis- tance and the system can be difficult to maneuver into a strong current. This system is unsuitable for censusing cryptic species, because its base separation is too large to maneuver into small crevices. The system is best deployed off a boat, i.e. a boat large enough to safely carry the stereo- rig, a calibration cube (see Harvey and Shortis, 1996), 0.0 -0.5 -1.0 -15 0% -1% ■■95 ■■94 £ 1.0 3 0.5 0.0 ■■94 •■95 -2% -3% -4% -5% ■■95 ■■94 6% 4% iS 2% 0> cr ■■94 ■■95 0% Figure 6 Means and 95% confidence intervals, for each measure of error, for 1994 and 1995. Number of repeat measurements Figure 7 Improvements in the accuracy of stereo-video mea- surements due to increasing the number of repeat measurements. 70 Fishery Bulletin 99(1 ) Ideal edge Image profile Image profile Actual edge a> Typical location manually selected Image profile and sampled pixels High contrast image High contrast image Figure 8 The effects of different contrast in images for location of the edges of objects. divers, and diving equipment. Shore-based diving would be extremely difficult. Before a stereo-video system is used to collect and mea- sure field data, operators should practice taking measure- ments in the laboratory with a set of test objects of known size, so that they may learn to distinguish the edge of the object accurately within a range of water clarity. Conclusions Many of the problems outlined in our study can be over- come in the near future. The development of digital video cameras and frame grabbing boards have improved image quality and the ease of image acquisition and synchroni- zation. These developments will also facilitate the analy- sis of streams of video images rather than single images, thereby greatly increasing the speed of image processing. In the future, the combination of stereo-videography with neural networks and fuzzy logic could facilitate automated pattern recognition, classification, and measurement of reef fish from video images. Worldwide, temperate and tropical reefs are being threat- ened by anthropogenic disturbances. There is a need to describe the ecological structure and function of reef-fish assemblages and to monitor the effects of disturbance on these populations. This type of research has traditionally been carried out by government agencies and academic in- stitutions. However, limitations in funding and resources are forcing these agencies to use supplementary sources of data. Consequently, the use of volunteers in monitoring programs is increasing (Halusky et ah, 1994; Mumby et ah, 1995; Darwall and Dulvy, 1996). Hunter and Maragos (1992) suggested that new technology in computing and underwater video systems may allow recreational and vol- unteer SCUBA divers to assist with surveys of coral reefs without compromising the data quality. We have demon- strated that volunteers can in fact carry out surveys of the length of reef fish using a stereo-video without compromis- ing data quality. Although the differences in the accuracy of stereo-video measurements made by an experienced op- erator and volunteers is statistically significant, these are negligible in comparison to errors in visual estimates by divers. Our study specifically addresses length estimates and does not address estimation of abundance. However, if someone is estimating all the lengths of a fish observed within a sample unit they are also effectively recording rel- ative abundance. The accuracy and precision of abundance estimates is an important issue. Factors that need address- ing include the accuracy of estimates made by a diver at a distance from the fish and the behavioral differences of fish at various spatial and temporal scales. These are complex issues that are beyond the scope of the present study. Quantitative sampling of reef-fish length frequency or biomass for monitoring programs requires reliable identi- fication skills and the ability to make precise and accurate estimates of reef-fish length. It is suggested that volunteer SCUBA divers could be trained to use a stereo-video sys- tem both in the water and in the laboratory for making measurements. Volunteers, under the guidance of profes- sional scientists, could assist with monitoring programs of reef-fish length frequency and abundance without having any effect on the quality of data recorded. The data col- lected and analyzed from stereo-video images are proven to be considerably more precise and accurate than visual estimates undertaken by experienced scientific divers. Harvey et al.: Comparison of estimates of reef-fish lengths made by divers and a stereo-video system 71 Acknowledgments We acknowledge financial aid from the New Zealand De- partment of Conservation (research grant 1822), the Uni- versity of Otago Research Committee, the Southland Regional Council (New Zealand) and the Division of Sci- ences at the University of Otago, Dunedin, New Zealand, and support from Sony New Zealand Ltd, THC Milford Sound, Fiordland Lobster Company, and Mobil and Gore Services Ltd. The authors are also grateful to James Seager, of the Department of Geomatics at the University of Mel- bourne, for assistance with software development, and to June Hill, Callum Duncan, Lisa Kelleher, and Andrea Brown for assisting with field trials. Additionally we thank all the divers who willingly gave their time to assist with our research. We would also like to acknowledge the Great Barrier Reef Marine Park Authority for allowing us access to data. This manuscript was greatly improved by com- ments from Howard Choat, June Hill, Philip Mladenov, Keith Probert, and four anonymous reviewers. Literature cited Baddeley, A. D. 1965. The influence of depth on manual dexterity of free divers : a comparison between open sea and pressure cham- ber testing. J. Appl. Psych. 50:81-85. 1971. Diver performance. In Underwater science (J. D. Woods and J. N. Lythgoe (eds.), p. 33-67. Oxford Univ. Press, London. Baddeley, A. D., J. W. De Figueredo, J. W. Hawkswell Curtis, and A. N. Williams. 1968. Nitrogen narcosis and underwater performance. Ergo- nomics 11(2): 15V — 164. Bell, J. D., G. J. S. Craik, D. A. Pollard, and B. C. Russell. 1985. Estimating length frequency distributions of large reef fish underwater. Corals Reefs 4:41-4. Bellwood, D. R. 1988. On the use of visual survey methods for estimating reef fish standing stocks. Fishbyte (April): 14-16. Bellwood, D. R., and A. C. Alcala. 1988. The effect of minimum length specification on visual census estimates of density and biomass of coral reef fishes. Coral Reefs 7:23-27. Darwall, W. R. T., and N. K. Dulvy. 1996. An evaluation of the suitability of non-specialist vol- unteer researchers for coral reef fish surveys. Mafia Island, Tanzania — a case study. Biol. Conserv. 78:223-231. English, S., C. Wilkinson, and V. Baker (eds.). 1994. Survey manual for tropical marine resources, 368 p. [Available from the Australian Institute of Marine Science. P.M.B. No. 3, Townsville Mail Centre, Australia 4810.] Francour, P. 1991. The effect of protection level on a coastal fish commu- nity at Scandola, Corsica. Rev. Ecol. Terre Vie 46:65-81. 1994. Pluriannual analysis of the reserve effect on ichthyo- fauna in the Scandola natural reserve (Corsica, Northwest- ern Mediterranean). Oceanol. Acta 17(3):309— 317. GBRMPA (Great Barrier Reef Marine Park Authority). 1979. Workshop on coral trout assessment techniques: work- shop series 3. G.B.R.M.P.A., Queensland, Australia, 64 p. Halusky, J. G., W. J. Seaman, and E. W. Strawbridge. 1994. Effectiveness of trained volunteer divers in scientific documentation of artificial aquatic habitats. Bull. Mar. Sci. 55(2-3):939-959. Harmelin-Vivien, M. L., and Y. Bouchon-Navaro. 1981. Trophic relationships among chaetodontid fishes in the Gulf of Aquaba (Red Sea). In Proc. 4th int. coral reef symp. 2:537-544. Harvey, E., and M. Shortis. 1996. A system for stereo-video measurement of subtidal organisms. Mar. Tech. Soc. J. 29(4)10-22. 1998. Calibration stability of an underwater stereo-video system: implications for measurement accuracy and preci- sion. Mar. Tech. Soc. J. 32(2):3— 17. Harvey, E., D. Fletcher, and M. R. Shortis. 2000. Improving the statistical power of length estimates of reef fish: a comparison of estimates determined visually by divers with estimates produced by a stereo-video system. Fish. Bull. 99(1)72-80. Hollien, H., and H. B. Rothman. 1975. The effectiveness of diver’s work with and without the aid of communication systems. Mar. Tech. Soc. J. 9( 8 ):3— 10. Hunter, C., and J. Maragos. 1992. Methodology for involving recreational divers in long- term monitoring of coral reefs. Pac. Sci. 46(3)381-382. Jones, R. S., and J. A. Chase. 1975. Community structure and distribution of fishes in an enclosed high island lagoon in Guam. Micronesia 1 1 : 127- 148. Kulbicki, M. 1989. Correlation between catch data from bottom longlines and fish censuses in the SW lagoon of New Caledonia. In Proc. 6th int. coral reef symp. 2, p. 305-312. Townsville, Australia. McCormick, M. I., and J. H. Choat. 1987. Estimating total abundance of a large temperate- reef fish using visual strip-transects. Mar. Biol., 96(4): 469-478. Mumby, P. J., A. R. Harborne, P. S. Raines, and J. M. Ridley. 1995. A critical assessment of data derived from Coral Cay Conservation volunteers. Bull. Mar. Sci. 56(3):737-751. Polunin, N. V. C., and C. M. Roberts. 1993. Greater biomass and value of target coral reef fishes in two small Caribbean marine reserves. Mar. Ecol. Prog. Ser. 100:167-176. Russ, G. R. 1985. Effects of protective management on coral reef fishes in the central Phillipines. In Proc. 5th int. coral reef symp. 4, p. 219-224. Tahiti. Samoilys, M. A. 1989. Abundance and species richness of the coral reef fish on the Kenyan Coast: the effects of protective management and fishing. In Proc. 6th int. coral reef symp. 2:261-266. Townsville, Australia. Shortis, M. R., W. L. Snow, B. A. Childers, and W. K. Goad,. 1993. The influence of storage media on the accuracy and repeability of photogrammetric measurements using CCD cameras. In Videometrics II, p. 80-92. SPIE (Interna- tional Society for Optical Engineering) 2067. St. John, J., Russ, G. R., and W. Gladstone. 1990. Accuracy and bias of visual estimates of numbers, size structure and biomass of coral reef fish. Mar. Ecol. Prog. Ser. 64:253-262. Underwood, A. J. 1981. Techniques of analysis of variance in experimental marine biology and ecology. Oceanogr. Mar. Biol. Ann. Rev. 19(5 13-605 ):5 13-605. 72 Abstract— We calculated the power of visual length estimates by novice and experienced scientific SCUBA divers and estimates generated by a stereo- video system to detect changes in the mean length of three common species of reef fish from New Zealand. Length estimates from a stereo-video system had much greater power for blue cod (mean length=33.1 cm., range 19.5-50.1 cm.) and snapper (mean length=31.7 cm., range 23-71 cm.). For a third spe- cies, red cod (mean length=42.5 cm., range 13-74 cm.), the statistical power of diver and stereo-video estimates was much less for an equivalent number of samples owing to the greater vari- ation in the true mean length of red cod recorded at different sites. At 90% power, a stereo-video system detected a 15% (~5-cm) change in the mean length of blue cod with 63% less samples (10) than those required by the experienced scientific divers (27). Novice scientific divers required 28 samples. Manuscript accepted 14 July 2000. Fish. Bull. 99:72-80 (2001). Improving the statistical power of length estimates of reef fish: a comparison of estimates determined visually by divers with estimates produced by a stereo-video system Euan Harvey Department of Marine Science University of Otago 304 Castle Street Dunedin, New Zealand Present address: Department of Botany University of Western Australia Western Australia 6907, Australia E-mail address: euanh@cyllene.uwa.edu.au David Fletcher Department of Mathematics and Statistics University of Otago 539 Castle Street Dunedin, New Zealand Mark Shortis Department of Geomatics University of Melbourne Parkville, Melbourne Victoria 3052, Australia Visual censuses of reef fish have been used to monitor fish communities as indicators of environmental degrada- tion (Hourigan et al., 1988; Fausch et ah, 1990) and as a fisheries manage- ment tool for assessing the condition of reef fish stocks (Ault et al., 1998). Ault et al. ( 1998) used data on the aver- age length of a fish stock as an index of fishing effects. Information on the length frequency or mean length of a fish population when linked with even a rudimentary knowledge of the biol- ogy of the species may allow estimates of recruitment to the adult population, fishing intensity, and rates of recovery from fishing (McCormick and Choat, 1987). Environmental surveys commonly use SCUBA divers to count and visual- ly estimate the length of individual reef fish (Jones and Chase, 1975; Harmel- in-Vivien and Bouchon-Navaro, 1981; Bellwood and Alcala, 1988; Samoilys, 1989; English et al., 1994). These visu- al censuses have many advantages in comparison with other sampling tech- niques: they are quantitative, quick, nondestructive and repeatable (Eng- lish et al., 1994). Visual census tech- niques have been widely adopted and are used to monitor changes in the relative abundance or mean length of reef fish within marine protected ar- eas (Bell, 1983; McCormick and Choat, 1987; Alcala, 1988; Cole et al., 1990; Francour, 1991, 1994; Russ and Acala, 1996) and as a tool for assessing the standing stock or biomass of individual species of reef fish (Craik, 1981; Russ, 1985; Medley et al., 1993; Polunin and Roberts, 1993; Hart et al., 1996). Bio- mass is estimated from the relationship between length and the weight of an in- dividual fish of a certain species (Kul- bicki, 1989; Kulbicki et al.,1993). How- ever, the question not yet addressed is how useful are data from visual length estimates for detecting changes in the mean length or length frequency of a population of reef fish? The advantages of assessing the sta- tistical power of environmental moni- toring programs has been discussed by Harvey et a!.: Improving the statistical power of length estimates of reef fish 73 a number of authors (Green, 1979; Andrew and Mapstone, 1987; Gerrodette, 1987; Hayes, 1987; Peterman, 1990a, 1990b; Fairweather, 1991). Statistical power is defined as the probability of correctly rejecting a null hypothesis and is 1-/3, where [i is the probability of a type-II error (An- drew and Mapstone, 1987; Gerrodette, 1987; Fairweather, 1991). An example of a type-II error in environmental monitoring would be to conclude that no impact has oc- curred when one has. Therefore, low statistical power can be disastrous for environmental monitoring because ad- verse environmental impacts go undetected (Fairweather, 1991). Despite this problem, few marine ecologists and bi- ologists make use of power analysis (Fairweather, 1991). Power analysis has been used to determine the optimum size of sample units and levels of replication needed to detect an effect of a particular size with a desired level of probability (Andrew and Mapstone, 1987; Gerrodette, 1987; Fairweather, 1991). Power is a function of sample size, the probability of a type-I error (a) and the effect size (Gerrodette, 1987). Fairweather (1991) discussed the is- sues associated with deciding upon an appropriate level of power. Low power can be attributed not only to the sample design, but also to biases and errors inherent in the sam- pling method (Andrew and Mapstone, 1987) and power analysis must account for the uncertainty of measurement error (Gerrodette, 1987). Historically, reef-fish ecologists have failed to calculate and publish the power of their sampling programs. Fur- thermore, it is frequently assumed by many researchers that their visual estimates of reef-fish length are both accurate and precise. In the published literature on reef fish studies containing data on visual length estimates, we found only three examples out of forty-three papers in which the authors stated the accuracy of their in situ vi- sual length estimates (Sweatman, 1985; Polunin and Rob- erts, 1993; Green, 1996). The aims of our study were 1) to examine the accuracy and precision of length estimates made by a number of ex- perienced and novice scientific SCUBA divers, and so de- termine their power to detect changes in the mean length of populations of three common species of reef fish from around New Zealand coastal waters and 2) to demonstrate that the power to detect changes in mean length can be greatly improved for two of the three species by using an underwater stereo-video system instead of divers’ visual estimates. The three fish species that we consider are blue cod ( Pa - rapercis colias), red cod ( Pseudophycis backus), and snap- per ( Pagrus auratus ). All three species support commer- cial trawl (red cod and snapper), long line (snapper), and trap (blue cod) fisheries. Blue cod and snapper are also the focus of popular recreational fisheries and thus are impor- tant species in New Zealand. Methods and materials To assess the extent to which measurement error will affect the power of visual estimates to detect changes in mean length of a population of fish, we considered the fol- lowing simple scenario. Suppose we are interested in com- paring the mean lengths of two fish populations and we collect length estimates by randomly selecting dive loca- tions within each site. At each location, the dive involves the visual collection of data from a strip-transect or point- count method to measure the length of each of a number of fish of the species concerned. Later, we will assume that the same numbers of fish are encountered on each dive. This is clearly unrealistic because the numbers encoun- tered will obviously differ: it merely helps to simplify the discussion of power analysis. The analysis we consider here involves first transforming the estimated lengths by using natural logarithms, calculating the mean log-length at each location in each site, and then comparing sites by a standard Ltest, with the locations acting as replicates. The reason for considering log-length rather than length is twofold. First, it may be more prudent to perform such an analysis on the log-scale, for the usual reason of want- ing to satisfy the assumptions of the t-test. Second, the power analysis can then be framed in terms of our ability to detect a percentage change in mean length. As a conse- quence, the standard allometric relationship between log- length and log-weight (Kulbicki, 1989) implies that the results given here for the power to detect a percentage change in mean length will also apply to an equivalent proportional change in mean weight. The estimated length of fish j at dive location i can be written as where xif = the true length of the fish; and e = the relative accuracy of the estimate (see St John et ah, 1990). This equation shows that variation in estimated length will arise from two sources: first, from the natural varia- tion, both between and within dive locations, of the true lengths of the fish; second, from the variation, between and within dives, in the relative accuracy of the estimate. It is this second component of variation that will be influ- enced by using stereo-video system as opposed to experi- enced or novice scientific divers. The relative magnitudes of the two sources of variation will determine the benefits to be expected from improving the measurement of length. Thus, if the natural variation in true length is large in relation to the measurement error, there will be little sta- tistical benefit in reducing the latter. On a log-scale this equation can be written as logy,, = logxy + logCy The variation in logjt(/ between and within dive locations can be expressed in a one-way random effects model as logv(/ = a, + bjj, with Var(a() = o2a and Var(6i;) = o\ (Sokal and Rolf, 1995). Thus o2 and a2, are the between-dive and within-dive a o variance components, respectively. 74 Fishery Bulletin 99(1 ) Now consider the variation in loge;;. We can write where dl ■ = an effect applying to all estimates made during dive i; and = an effect applying solely to the estimate for fish j during that dive. The value of dl will be influenced by the conditions at loca- tion i, as well as by the diver used at that location. The value of £jj will be influenced by the activity of the fish and its orientation to the diver. We can now write Logey = logrf, + log £jj, with Var(logd-) = a2d and, VarOoge^) = ct2, analogous to the equation for log* . Again, a2d and ct2 are the between-dive and within-dive variance components. The power analysis that follows involves predicting the variation we would expect in logy, for a given number of dive locations in) and a fixed number of fish at each loca- tion (m). The equations above can be combined to show that this variation has four components. Because logyy = a, + btJ + log d, + log Ejj, we have Vflo gyu) = ct2 + a\ + a2d + ct2. Going one step further, the predicted variance of the mean of logy, over all fish (j) and all dives (i) at that site can be written as Ua T Ud T y = — + EL. n n Assuming the number of dives and fish recorded per dive is the same at the second site, the predicted power to detect a difference D in the mean log-lengths at the two sites is given by D SED 0,2 where Fr [.] = cumulative distribution function; tuh = the upper a/2% point for the ^-distribution with 2(n-l) degrees of freedom, and SED = V2V, which is the standard error of the dif- ference in the two mean log-lengths (see Sokal and Rohlf, 1995, p. 263). Note that if the analysis involved the comparison of s>2 sites, the degrees of freedom for the f-distri- bution would be s(/?-l). Because the anal- ysis is on a log-scale, the difference D is calculated as D=log(R+l), where R is the percentage change of interest. In order to evaluate the power, we needed estimates of the four variance components. The first two, ct2 and o\, were estimated by using catch data on true length for populations of red cod, blue cod, and snapper from around New Zealand from trawl and trap surveys by the Fisheries Division of the National Institute of Water and Atmospheric Research. The length data for red cod and snapper came from a number of locations around New Zealand (Fig. 1). The depths at which these fish were collected ranged between 20 and 40 m. The locations at which they were collected were grouped into sites, such that two locations that were within approximate- ly 30 km of each other were considered to be at the same site. For red cod, there were 12 sites, each with between two and four locations: at each location, lengths were record- ed for between 12 and 20 fish. For snapper, there were six sites, two of which contained 14 locations, and the other four had just two locations each. At each location, lengths were recorded for between 10 and 91 fish. The log-lengths were then analyzed by using nested analysis of variance. Site and location were specified to be random factors, and location was nested within the site. The location and residual vari- ance components were used as estimates of ct2 and a re- spectively. The data for blue cod came from one site, at five locations off Stewart Island. There were between 47 and 51 fish lengths recorded per location. The log-lengths were ana- lyzed by using analysis of variance, with location being spec- ified as a random factor. The location and residual variance components were used as estimates of a2 and o2h, respec- tively. The remaining two variance components, o2d and ct2 were estimated by using data on the measurement error of novice scientific divers, experienced scientific divers, and a stereo-video system. These errors were determined by using a simple testing procedure for calibrating diver estimates of the lengths of reef fish. Silhouettes of fish were placed in the water and their lengths estimated by following the meth- ods of the GBRMPA (1979), Bell et al. (1985), and English et al. (1994). There were eight novice divers and six expe- rienced divers, each of who swam five repeat transects. On each transect they estimated the length of sixteen silhou- ettes. The same procedure was used for the stereo-video es- timates: for each image the estimate used was the mean of ten measurements. The novice divers and three of the experienced divers made their length estimates in a salt- water pool. For the stereo-video system and the other three experienced divers, measurements were made in a swim- ming pool. Definitions of novice and experienced scientific divers are the same as those given in Harvey et al. (2000), which contains a full presentation of the data and detailed description of the method used. A full description of the de- sign and calibration of the system can be found in Harvey and Shortis (1996, 1998). The novice and experienced diver estimates were both analyzed by using nested analysis of variance of the loga- rithm of the relative accuracy (estimated length divided by true length). Diver and silhouette were specified to be random factors, and silhouette was nested within diver. The diver and silhouette variance components were used as estimates of o2d and a2 , respectively. The stereo-video estimates were analyzed by using analysis of variance of the logarithm of the relative accuracy, and silhouette Harvey et al.: Improving the statistical power of length estimates of reef fish 75 was specified as a random factor. The silhouette vari- ance component was used as an estimate of . Therefore, it is essential that fish eggs and larvae develop in favorable habitats that maximize the probability of survival. Bigelow (1926) recognized the significance of the coastal shelf for the production of fish larvae within the Gulf of Maine, noting that most larvae were found within the 200-m contour. He also observed that larval drift was generally to the southwest and that abundance increased progressively to the west with the result that few larvae were observed off eastern Maine and in the Bay of Fundy. Other surveys (Fish and Johnson, 1937; Marak, 1960; Marak and Colton, 1961; 1962) further defined the composition of the ichthy- oplankton of the Gulf of Maine. Ich- thyoplankton of inshore waters of the Gulf of Maine has been documented for the Damariscotta, Sheepscot, and Sullivan Harbor estuarine systems and nearby waters in the central area of the Maine coast (Graham and Boyar, 1965; Graham, 1972; Chenoweth, 1973; Hauser, 1973; Lee, 1975; Laroche, 1980; 1982; Shaw, 1981; Townsend, 1981; 1983; 1984). However, the ichthyoplank- ton of Penobscot Bay has not been studied despite the fact that it is the largest embayment in the region and that coastal environments, such as bays and estuaries, may constitute favorable habitats for the early life stages of a large number of marine fishes (Frank and Leggett, 1983). This study describes the results of a two-year, spring survey of larval fishes in Penobscot Bay, Maine. The objectives of the study were 1) to describe the structure of the larval fish community, 2) to determine the temporal and spa- tial variation in species diversity and abundance, and 3) to relate these vari- ations to differences in location and en- vironmental variables. Materials and methods Field methods Penobscot Bay is a large (80-km) drowned river valley typical of the Maine coast. It has a drainage area of over 21,000 km2 (Haefner, 1967). The study area is about 40 km long and varies in depth from 15 to 110 m. Sixteen ichthyoplankton stations (Fig. 1) were selected to encompass Penob- scot Bay for the larval surveys. Six upper bay stations ( R1-R6) were located in the northern estuarine portion of the bay between Isleboro Island and the mainland. Seven midbay stations (B1-B7) were located in the central portion of the bay. Three lower bay stations (02-04) were located in the sourthern estuarine portion of the bay adjacent to the islands of North Haven and Vinalhaven. In 1997, seven, two- or three-day cruises (97I-97VII) were conducted bi- weekly from 4 April through 25 June 1997 to coincide with spring and sum- mer spawning times for many fishes. Data collection involved towing a 1.0-m, 333-micron mesh plankton net equipped with a General Oceanics flowmeter dur- ing daylight hours (Fig. 1). The net was hauled for 20 minutes in stepped oblique fashion at the surface, at 10 m, and at 20 m, or to within 5 m of the bottom. At each station, a vertical pro- file of salinity and temperature was col- lected with a Seabird 19 CTD (conduc- tivity, temperature, and depth) probe. 82 Fishery Bulletin 99(1) Figure 1 Map of stations sampled biweekly with a 1.0-m plankton net in Penobscot Bay, Maine, from 4 April to 25 June 1997 and from 18 March to 30 April 1998. II = Iselboro Island; NH = North Haven Island; and VH = Vinalhaven Island. In 1998, four one- or two-day cruises (98I-98IV) were conducted biweekly from 18 March through 30 April 1998 and data collection was the same as in 1997 except that ten ichthyoplankton stations were sampled in the lower bay only. Eight of these stations were sampled in 1997 including five midbay stations (B1-B5) and three lower bay stations (02-04). Two additional lower bay stations, 01 and 05, were added in 1998. Larvae from both years were preserved in 5% formalin for later identification to the lowest taxon possible by the Atlantic Reference Center of the Huntsman Marine Biological Laboratory in St. An- drews, New Brunswick, and for quantitative determina- tion of larval fish densities (number of larvae per filtered 100 m3). Fish larvae were measured for standard length, or in some cases notochord length, to the nearest mm. Plankton volume standardized by volume filtered was de- termined for each tow by displacement of the unidentified plankton. Data analysis Larval fish abundance and environmental data were analyzed by using three multivariate techniques: princi- pal components analysis (PCA), multivariate analysis of variance (MAN OVA), and canonical correlation analysis (CCA). PCAs of the variance-covariance matrices derived from both environmental and larval abundance data were performed to reduce intercorrelated variables to a smaller number of uncorrelated variables. This procedure provided a concise description and comparison of complex spatial Lazzari: Dynamics of larval fish abundance in Penobscot Bay, Maine 83 and temporal patterns of the larval fish assemblage and environmental data (Gauch, 1982). Varimax rotation was performed on factors from the PCAs in both cases because rotated solutions tend to extract components that cor- relate highly with a smaller number of variables than unrotated PCAs (Stevens, 1986). This step aided in the interpretation of the factors. Rotated factor scores from the PCA were used as the dependent variables in the MAN OVA because they were uncorrelated, thus satisfying the assumption of independence for parametric statistical tests. Environmental data were log-transformed to satisfy assumptions of univariate normality and then summa- rized by PCA. The resulting factor scores were grouped by month, and means were compared by MANOVA to as- sess the null hypothesis that environmental variables did not differ among months. This analysis was followed by a Tukey-Rramer multiple range test to detect which month- ly means differed. This test controls for increases in the type-I error rate associated with unequal sample sizes (Day and Quinn, 1989). Temporal changes in environmen- tal variables were interpreted by evaluating product mo- ment correlations of the environmental variables and fac- tor scores from PCA. I considered only variables that were significantly correlated with an individual factor in inter- preting that factor (significance levels were adjusted for multiple comparisons, where p'= 1 - ( 1 - p)1,k and k equals the number of comparisons [Sokal and Rohlf, 1981]). The results of these analyses were used to discriminate among months. The analysis of larval fish assemblage composition was similar to that performed for environmental data. However, fish taxa not present in 5% of the samples were eliminat- ed to reduce the influence of rare taxa (taxa that were ex- cluded were present in five of 102 samples in 1997 and two of 40 samples in 1998) as potential outliers (Gauch, 1982). A matrix of angularly transformed [arcsine of the square root of a proportion (Sokal and Rohlf, 1981)] relative abun- dances for the m = 12 taxa in n = 102 samples in 1997, and the m = 13 taxa in n = 40 samples in 1998 were con- structed and summarized by PCA. Varimax rotated factor scores generated from PCA of the relative abundance ma- trix were compared among months by using MANOVA (unbalanced design). The null hypothesis was that mean factor scores (and taxonomic composition) did not change among months. This was followed by a Tukey-Rramer test of significantly affected factors to identify the extent to which monthly means differed. Factor scores were then grouped by location and served as the dependent variables in MANOVA to test for dif- ferences in sample location (upper, mid, or lower bay). The null hypothesis was that mean factor scores (and taxonomic composition) did not change among locations. A Tukey-Rramer test was performed to identify location mean factor scores that differed significantly. Temporal and spatial changes in the taxonomic composition of lar- val fish assemblages were then interpreted by identify- ing which factors were affected by month and location treatments and by evaluating product-moment correla- tions of angularly transformed relative abundance data Table t Mean (± standard error) environmental variables for the entire study period, reported by month for 1997 and 1998. Temperature (°C) and salinity recorded as mean for top 20 m of the water column and volume displacement (Pvol) in mL of unidentified plankton per 100 m3. Month Temperature Salinity Depth (m) Pvol 1997 April 3.37 (0.13) 30.56(0.13) 74.0 (9.4) 17.3 (1.7) May 5.49 (0.12) 30.24(0.08) 58.0 (11.8) 4.2 (0.5) June 8.52 (0.22) 30.44 (0.05) 46.8(8.2) 13.7 (2.1) 1998 March 2.41 (0.01) 29.90(0.05) 38.8(18.8) 34.4 (0.5) April 4.23 (0.14) 30.14(0.05) 56.5 (10.7) 25.5 (0.9) and significant PCA factor scores (i.e. “loadings” on fac- tors significantly affected by time and location; Pielou, 1984). The relationships of the factor scores from PCA of the relative larval fish abundance matrix and the log- transformed environmental variables were examined by CCA. Standard lengths, or notochord lengths, from 1997 and 1998 were compared among cruises (sample dates) for each species by using Kolmogorov-Smirnov two-sample tests. The null hypothesis was that size distribution did not differ significantly between sample dates. Results A total of 102 plankton net samples and CTD measure- ments were collected during biweekly sampling between 4 April and 25 June 1997 in Penobscot Bay. Sixteen stations were sampled on five of the seven sampling trips; how- ever, inclement weather (wind speeds >25 knots) during the first week of April (4-7, 971) and from late April to early May (29 April-1 May 97III) limited the number of stations that could be sampled during these trips to 10 and 12, respectively. The number of cubic meters of water filtered per 20-minute tow varied from 137.5 to 755.7 m3 depending on station location and sample date. The total volume of filtered in 1997 was 44,622 m3. In 1998, 40 plankton net samples and CTD measure- ments were collected during approximately biweekly sam- pling between 18 March and 30 April 1998. Ten stations were sampled on each of the four sampling trips. The num- ber of cubic meters of water filtered during each 20-minute tow varied from 54.2 to 582.5 m3 depending on station lo- cation and sample date. The total volume of water filtered in 1998 was 12,459 m3. Environmental data Station depth ranged from 20 m at mean low water at the upper bay station R4 to 91 m at station 02 (Table 1). 84 Fishery Bulletin 99(1) Table 2 Correlations among environmental variables and factor scores from varimax-rotated PCA of the log-transformed environmental data. Underlined values significant at P = 0.05, adjusted for multiple comparisons such that P < 0.01 for a significant correlation. Depth Pvol Temp Salinity Factor 1 Factor 2 1997 Depth 1.00 0.08 0.98 Pvol 0.07 1.00 0.84 -0.06 Temperature -0.13 -0.35 1.00 -0.68 -0.05 Salinity 0.20 0,44 -0.24 1.00 0.71 0.25 1998 Depth 1.00 0.08 0.83 Pvol -0.08 1.00 -0.90 0.08 Temperature 0.18 -0.71 1.00 0.90 -0.09 Salinity -0.16 -0.29 0.21 1.00 0.42 -0.68 Station mean water temperatures for the top 20 m of the water column in 1997 were variable and ranged from 2.3°C on 4 April to 10.8°C on 25 June. Station mean salin- ity ranged between 27.8 ppt on 30 April and 31.2 ppt on 16 April. Plankton volume ranged from 0.8 to 44.9 mL per 100 m3 and generally was greatest during early April (971) with a secondary peak in volume observed in late June (97VII). Greatest plankton volumes occurred at the lower bay stations. In 1998, mean water temperatures ranged from 2.4°C on 18 March to 5.3°C on 30 April. Mean salin- ity ranged between 29.6 ppt on 18 March and 30.6 ppt on 30 April. Station plankton volume in 1998 ranged from 6.1 to 280 mL per 100 m3 and was greatest during late April (98IV) at the lower bay stations. We compared environmental variables among months to aid in the characterization of Penobscot Bay. PCA of log-transformed variables identified two factors that ex- plained 69% of the total variance in 1997 and 76% in 1998 (Table 2). Factor 1 was related to temperature, plank- ton volume, and salinity in 1997 and in 1998. Factor 2 was related to depth in both years. MAN OVA of mean factor scores indicated significant differences by month in 1997 (Wilk’s lambda=0.792, P<0.001) and 1998 (Wilk’s lambda=0.778, PcO.001). Temperature was negatively cor- related and plankton volume and salinity were positively correlated with PCA factor 1 in 1997 (Table 2). Post hoc comparisons of monthly scores for factor 1 indicated that the mean scores for April were different from those for May and June. In 1998, temperature was positively cor- related and plankton volume negatively correlated with PCA factor 1 (Table 2) and post hoc comparisons indicated that the mean scores for March were different from those for April. Salinity was correlated with both PCA factor 1 (positively) and factor 2 (negatively) in 1998. Depth was positively correlated with factor 2 in both 1997 and 1998. Post hoc comparisons of monthly scores for PCA factor 2 indicated that the mean scores for April were different from those for May and June in 1997; the mean score for March was different from that for April in 1998. Fish larvae Differences existed between the two years in the kinds of fish larvae found in Penobscot Bay. At least, 26 species (23 individuals [1.5%] not identified to species) belonging to 15 families were identified (Table 3). A total of 23 species (15 families) were found in 1997, and 16 species (9 fami- lies) in 1998; 13 species were common to both years. Dif- ferences in taxonomic composition were mainly due to the collection of rare specimens during the year, such as inquiline snailfish (Liparis inquilinus) in 1997 and pol- lock ( Pollachius virens ) in 1998. In addition, several spe- cies that were present in 1997, including American eel (Anguilla rostrata), capelin (Mallotus villosus), fourbeard rockling ( Enchelyopus cimbrius), Atlantic tomcod (Micro- gadus tomcod ), cunner ( Tautogolabrus adspersus), Atlantic mackerel (Scomber scombrus), alligatorfish ( Aspidophoroi - des monopterygius), windowpane flounder (Scophthalmus aquosus ), and American plaice (Hippoglossoides plattes- soides), did not occur in the 1998 collections. In contrast, pollock, snake blenny (Lumpenus lampraetiformis) and Arctic shanny (Stichaeus punctatus) were collected only in 1998. The assemblage of species was slightly more diversi- fied in 1997 than in 1998 with Shannon-Wiener indices of H'=3.1 and 2.8, respectively. The three most abundant spe- cies in 1997 (U. subbifurcata, Ammodytes sp., P america- nus) constituted about 59% of all captures compared with 68% in 1998 (M. octodecimspinosus , P gunnellus, Ammo- dytes sp.). A total of 779 larvae of 23 species of fishes were collect- ed during the seven cruises in 1997 (Table 3). The most commonly occurring larvae (>35% of the samples) were At- lantic seasnail (Liparis atlanticus), winter flounder (Pleu- ronectes americanus), radiated shanny (Ulvaria subbifur- cata), sand lance (Ammodytes sp.), and H. plattessoides. Rare larvae (<1% of the samples) included Atlantic cod ( Gadus morhua), Gulf seasnail ( Liparis coheni ), L. inqui- linus, shorthorn sculpin (Myoxocephalus scorpius), and S. aquosus. The highest number of species (14) occurred dur- Lazzari: Dynamics of larval fish abundance in Penobscot Bay, Maine 85 Table 3 Egg type, total density (number/100 m3), and percent frequency of occurrence (FO) of larval fishes collected by 1.0 m plankton net from April to June 1997 (102 samples) and March to April 1998 (40 samples) in Penobscot Bay. D = demersal; P = pelagic. Family 1997 1998 Species Egg type Density FO Density FO Anguillidae Anguilla rostrata Clupeidae — 0.004 2 Clupea harengus D 0.009 4 0.072 15 Osmeridae Mallotus villosus D 0.022 5 Gadidae Enchelyopus cimbrius P 0.034 11 Gadus morhua P 0.002 1 0.022 5 Microgadus tomcod D 0.007 2 Pollachius virens P 0.007 3 Labridae Tautogolabrus adspersus P 0.007 3 Stichaeidae Lumpenus lampraetiformis D 0.086 23 Ulvaria subbifurcata D 0.462 41 0.014 5 Stichaeus punctatus D 0.014 3 Pholidae Pholis gunnellus Cryptacanthodeidae D 0.099 27 1.428 75 Cryptacanthodes maculatus Ammodytidae D 0.054 12 0.294 33 Ammodytes sp. D 0.350 35 0.596 35 Scombridae Scomber scombrus P 0.016 4 Cottidae Hemitripterus americanus D 0.064 21 0.215 43 Myoxocephalus aenaeus D 0.022 9 0.237 35 M. octodecimspinosus D 0.034 9 1.550 40 M. scorpius D 0.004 1 0.366 33 Agonidae Aspidophoroides monopterygius Cyclopteridae D 0.007 3 Liparis atlanticus D 0.182 51 0.043 13 L. coheni D 0.002 1 0.187 23 L. inquilinus D 0.002 1 Bothidae Scopthalmus aquosus D 0.002 1 Pleuronectidae Hippoglossoides platessoides P 0.094 30 Pleuronectes americanus D 0.229 47 0.029 8 ing mid-April (9711). The most abundant larvae were U. subbifurcatci (26% of total number), Ammodytes sp. (20%), P. americanus ( 13%), and L. atlanticus ( 10%). Total density of all larvae combined ranged from 0.6 larvae per 100 m3 in mid-May (97IV) to 3.9 larvae per 100 m3 in late June (97VII). The high larval densities in late June were a re- sult of the large number of U. subbifurcata collected at the three lower bay stations. In 1998, 726 larvae of 16 species of fishes (Table 3) were collected over the four sampling cruises. The most common 86 Fishery Bulletin 99(1) 1997 1998 Figure 2 Total density (number/100 m3) of larval fishes by egg type and sample date in Penobscot Bay, Maine: 4-7, 15-16, and 29-30 April; 12-13, 28-29 May; 9-10 and 24-25 June 1997; and 18 March, 7, 15, 30 April 1998. larvae (>35% of the samples) were rock gunnel ( Pholis gunnellus), sea raven ( Hemitripterus americanus), long- horn sculpin (Myoxocephalus octodecimspinosus), grubby (M. aenaeus), and Ammodytes sp., whereas P virens and S. punctatus were the least common (<3%). Myoxocepha- lus octodecimspinosus (30% of total number), P. gunnellus (27%), and Ammodytes sp. (11%) were the most abundant larvae collected. The greatest number of species (12) were found during mid-March (981). Densities of all larvae com- bined declined from 11.4 larvae per 100 m3 in mid-March (981) to about one larva per 100 m3 in mid-April (98III). The high larval densities in mid-March resulted from the high numbers of M. octodecimspinosus and P. gunnellus larvae collected at that time. Larvae originating from demersal eggs dominated the catches in both years (Table 3). In terms of abundance, most larvae collected in 1997 came from demersal eggs (70%) that dominated the catch at all stations. Larvae from demersal eggs composed most of the catch (>90%) until mid (79%) and late June (60%) when the number of larvae originating from pelagic eggs increased (Fig. 2). Overall, five of 23 species (22%) of larvae came from pe- lagic eggs in 1997, but by late June (97VII), four of the eight species (50%) collected hatched from pelagic eggs. Larvae from demersal eggs dominated the catches in 1998 to an even greater extent than in 1997 (Fig. 2). Two of the 16 species (13%) and four larvae (0.04%) collected in 1998 originated from pelagic eggs (three specimens of G. morhua and one of P. virens). Ulvaria suhbifurcata were the most abundant larvae in 1997 with densities up to 9.1 per 100 m3 at the lower bay stations, whereas in 1998, M. octodecimspinosus and P gunnellus were the most abun- dant larvae, with station densities up to 12.9 and 6.1 lar- vae per 100 m3, respectively. Densities of sand lance (Am- modytes sp.) were highest (4 per 100 m3) at the upper bay and midbay stations in 1997 and 1998, respectively. Pleu- ronectes americanus were more abundant in 1997 with densities up to 1.7 larvae per 100 m3 at the midbay sta- tions (B2, B3, B4). Larvae from pelagic eggs were rare in both years. Several species appeared in plankton samples from most sampling dates in 1997 (Table 4). One species, H. platte- soides was taken during all seven cruises but was absent from the 1998 collections. Ammodytes sp. occurred from early April (971) through late May (97V) in 1997 and from mid-March (981), early April (9811), and late April (98IV) in 1998. Pleuronectes americanus and L. atlanticus were present from mid-April (9711) to late June (97VII) and H. americanus and P. gunnellus were found from early April (971) to mid June (97VI). Four of the five species of fish larvae originating from pelagic eggs were taken during only one cruise. In 1998, three species, P. gunnel- lus, H. americanus and M. aenaeus, were collected on all four cruises. Atlantic herring ( Clupea hare/igus) and wry- mouth ( Cryptacanthodes maculatus ) occurred from mid- March (981) through mid April (98III). Myoxocephalus oc- todecimspinosus, M. scorpius, and L. coheni were taken only during the first two cruises. Size of the most abundant species changed little over the survey time period (Table 4). For example, winter floun- der larvae were 2-4 mm during mid April (9711) and 3-7 mm during late June (97VII) indicating successive spawn- ing events or no observed growth. The length-frequency distributions of most larvae were not significantly differ- ent among sample dates (Kolmogorov-Smirnov two sam- ple test, P>0.05). However, the proportion of larger larvae increased over time for three taxa collected over a two month period in 1997 (i.e. Ammodytes sp., P. gunnellus, H. plattessoides. Table 4). Few species showed a change in length frequency distribution during the shorter 1998 sur- vey (Kolmogorov-Smirnov two sample test, P>0.05). Temporal changes in the abundance of fish larvae were observed in both years. Two peaks in mean density of lar- val fish were observed among all locations in 1997, where- as only a single peak was seen during the shorter 1998 survey. In 1997, the mean density was initially about 1.9 larvae per 100 m3 in early (971) and mid-April (9711), then declined to 0.6 larvae per 100 m3 by mid May (97IV). Lar- val density increased to 1.2 larvae per 100 m3 by late May (97V) and peaked at 3.9 larvae per 100 m3 in late June (97VII). Most of the initial reduction in density was due to fewer Ammodytes. sp., C. maculatus, and H. americanus in the collections, whereas the June increase was due to U. subbiffurcata, and to a lesser extent to E. cimbrius and P. americanus. Densities of fish larvae in 1998 declined from a peak of 11.4 larvae per 100 m3 in mid-March (981) to about 1 larva per 100 m3 in mid-April (98III), before increasing to 1.4 larvae per 100 m3 on late April. Fewer Ammodytes. sp., C. maculatus . L. coheni, the three Myoxo- cephalus sp., and P. gunnellus accounted for the initial re- duction. The increase during the latter part of the 1998 survey was due to greater numbers of L. atlanticus, P. americanus, and U. subbiffurcata larvae in collections. Relative abundance patterns of 12 of the most abundant taxa (by the 5% criterion) in 1997 and 13 taxa in 1998 Lazzari: Dynamics of larval fish abundance in Penobscot Bay, Maine 87 Table 4 Length-frequency distributions of the more abundant larval fishes collected in Penobscot Bay in 1997 and 1998; and results of the Kolmogorov-Smirnov two-sample test (P<0.05). Notochord or standard length (mm) Year and species Cruise n 1 2 3 4 5 6 7 8 9 10 >10 Mean KS test 1997 U. subbifurcata I 0 II 0 III 0 IV 8 2 6 6.8 VII>VI V 21 1 9 11 6.5 VI 54 4 31 17 2 6.3 VII 119 25 44 28 10 8 3 1 6.5 P. gunnellus1 I 20 2 8 7 2 1 14.6 III>I,H II 14 2 1 3 1 2 1 2 2 16.5 III 6 1 1 4 20.3 IV 1 1 25.0 V 1 1 14.0 VI 1 1 13.0 VII 0 C. maculatus1 I 12 2 1 4 5 20.4 ns II 10 3 7 22.1 III 0 IV 1 1 38.0 V 0 VI 0 VII 0 Ammodytes sp. I 7 1 2 2 2 9.0 V>II,III II 94 1 4 12 15 11 18 16 17 8.6 III 44 1 1 4 8 10 7 13 9.7 IV 4 1 1 2 12.0 V 2 2 36.5 VI 0 VII 0 H. americanus1 I 10 2 6 1 1 14.1 rv>i,n II 11 1 2 3 2 1 2 15.8 III 3 1 1 1 18.3 IV 3 3 21.7 V 1 1 23.0 VI 1 1 14.0 VII 0 L. atlanticus I 0 II 3 1 2 4.7 II>VI III 11 7 3 1 4.5 IV 10 3 7 3.7 V 18 8 8 2 3.7 VI 17 10 5 2 3.5 VII 21 1 9 8 2 1 3.7 P. americanus I 0 II 1 1 2.0 VII>IV III 18 1 11 6 3.3 IV 24 1 9 14 3.5 continued 88 Fishery Bulletin 99(1) Table 4 (continued) Year and species Cruise n 1 2 Notochord or 3 4 5 standard 6 7 ength (mm 8 9 ) 10 >10 Mean KS test P. amerieanus (cont.) V 16 1 7 5 3 3.6 VI 17 6 8 1 2 3.9 VII 24 7 7 3 5 2 4.5 H. platessoides I 1 1 5.0 VII>VI II 4 2 1 1 4.8 III 2 1 1 5.0 IV 3 1 1 1 4.7 V 4 2 1 1 5.0 VI 17 1 3 3 9 1 4.5 VII 10 2 1 2 4 1 6.2 1998 P. gunnellus1 I 134 2 16 46 37 18 9 5 1 13.8 ns II 47 1 12 14 8 7 1 1 2 1 13.7 III 12 3 2 3 1 1 2 14.1 IV 3 1 1 1 14.7 C. maculatus1 I 18 3 6 5 4 19.6 ns II 12 4 6 2 20.0 III 10 1 3 6 19.5 IV 0 Ammodytes sp. I 72 5 16 22 15 4 4 6 7.7 ns II 9 3 1 1 2 2 8.9 III 0 IV 1 1 10.0 H. amerieanus1 I 8 1 4 1 1 1 14.5 m>i II 15 1 3 3 4 1 2 1 16.7 III 5 3 1 1 17.8 IV 2 2 23.0 L. coheni I 19 4 3 4 1 1 6 7.5 n>i II 7 1 6 11.0 IV 0 V 0 M. aenaeus I 22 2 15 5 6.1 ns II 9 4 4 1 5.9 III 2 1 1 6.5 V 2 1 1 5.5 M. octodecimspinosus I 200 1 12 46 45 40 56 9.6 n>i II 16 2 1 2 11 10.4 III 0 IV 0 M. scorpius I 38 1 1 6 30 11.5 n>i II 13 13 14.0 VI 0 VII 0 1 scale + 10 mm. were examined in detail (Table 5). Relative abundances of fishes changed over the course of the study in both 1997 and 1998 (Fig. 3). Three factors (species assemblages) were retained from the PCA of the relative abundance matrices in each year. In 1997, MANOVA and post hoc comparisons of mean factor scores indicated a significant temporal ef- fect on factors 1 and 2 (Wilk’s lambda=0.198, PcO.OOOl). Values for taxa that were most abundant in June (E. cirri- Lazzari: Dynamics of larval fish abundance in Penobscot Bay, Maine 89 Table 5 Product-moment correlations between principal component scores and the angularly transformed relative abundances of larval fishes. Underlined values significant at P < 0.05, adjusted for multiple comparisons such that P < 0.004 for a significant correlation. Rotated factor 1 explained 36% of the total variance, factor 2 explained 20%, and factor 3 explained 15%’ in 1997, and rotated factor 1 explained 35% of the total variance, factor 2 explained 28%, and factor 3 explained 11% in 1998. Taxon 1997 1998 Factor 1 Factor 2 Factor 3 Factor 1 Factor 2 Factor 3 Clupea harengus 0.05 0.51 -0.04 Mallotus villosus 0.02 -0.16 0.08 Enchelyopus cimbrius 0.44 -0.14 0.11 Lumpenus lampraetiformis -0.07 0.34 0.02 Ulvaria subbifurcata 0.97 -0.22 0.10 -0.11 -0.11 0.05 Pholis gunnellus -0.22 0.46 -0.41 0.67 0.40 0.58 Cryptacanthodes maculatus -0.14 0.23 -0.39 0.63 -0.01 0.11 Ammodytes sp. -0.17 0.98 0.05 0.06 0.92 0.31 Hemitripterus americanus -0.15 0.37 -0.27 0.13 0.02 0.73 Myoxocephalus aenaeus -0.12 0.31 -0.13 0.41 0.58 0.35 M. octodecimspinosus -0.14 0.07 -0.41 0.80 0.58 -0.07 M. seor-pius 0.56 0.66 0.17 Liparis atlanticus 0.39 -0.13 0.55 -0.42 0.16 -0.05 L. coheni 0.67 0.10 0.40 Pleuronectes americanus 0.06 -0.23 0.89 -0.20 -0.09 -0.13 Hippoglossoides platessoides 0.48 -0.07 0.27 □ Usu □ Pgu □ Pel a Pam □ Msp □ Lsp □ Cma ■ Asp — i Mar Apr 1998 Figure 3 Percent density (number/100 m3) by month of the dominant larval fishes collected with 1.0-m plankton net in Penobscot Bay, Maine, from April to June 1997 and from March to April, 1998. Asp -Ammo- dytes sp., Cma = Cryptaeanthod.es maculatus. Lsp = Liparis spp., Msp = Myoxocephalus spp., Pam = Pleuronectes americanus. Pel = pelagic spp., Pgu = Pholis gunnellus, Usu = Ulvaria subbiffurcata. brius, H. plattesoides, U. subbifurcata ) were uni- formly positively correlated with factor 1. Species that were most abundant earlier in the year (Am- modytes sp., H. americanus, M. aenaeus, P. gunnel- lus) were positively correlated with factor 2. Values for several species (C. maculatus, M. octodecimspi- nosus, and P. americanus) were not significantly correlated to either factor 1 or 2, but were cor- related to factor 3. In 1998, MANOVA and post hoc comparisons of mean factor scores indicated a significant temporal effect on factors 1 and 2 (Wilk’s lambda=0.341, P<0.0001). Values for taxa that were more abundant in March ( Ammodytes sp., C. maculatus, C. harengus, L. coheni, M. aenae- us, M. octodecimspinosus, M. scorpius ) were uni- formly positively correlated with factor 1 or factor 2, or with both factors. Values for H. americanus were not significantly correlated to either factor 1 or 2, but were highly correlated to factor 3 as their relative abundances increased in April. Assemblage composition was similar among sampling locations as indicated by MANOVA of mean factor scores in 1997 and 1998. In 1997, MANOVA (Wilk’s lambda=0.985, P>0.10) indicated that means for lower bay, midbay and upper bay sites were not significantly different from each oth- er for any factor. Similarly, MANOVA (Wilk’s lamb- da=0.831, P>0.08) indicated that means for lower estuary and bay sites were not significantly different for any factor in 1998. CCA revealed significant overall association between fac- tor scores from the PCA of the relative fish abundances and environmental variables in 1997 (Wilk’s lambda=0.17, 90 Fishery Bulletin 99(1 ) P<0.0001) and 1998 (Wilk’s lambda=0.21, PcO.OOOl). In 1997, the first canonical correlation (Table 6) indicated that the relative abundance of larval fishes with high posi- tive loading on factor 1 (E. cimbrius, H. plattesoides, and U. subbifurcata ) exhibited a positive association with tem- perature. These taxa also exhibited a weaker negative as- sociation with plankton volume and salinity. Fish taxa that loaded positively on factor 2 were negatively associated with temperature and positively associated with plankton volume and salinity. These species included Ammodytes sp., H. americanus, M. aenaeus, and P gunnellus. Fish taxa that loaded positively on factor 3 (L. atlanticus and P. amer- icanus) were positively associated with temperature and were present in every month sampled. In 1998, the first canonical correlation (Table 6) indicated that the relative abundance of larval fishes with high positive loading on factor 1 (C. maculatus, L. coheni, and P. gunnellus) and factor 2 (Ammodytes sp., C. harengus, and M. aenaeus) all exhibited a negative association with temperature and a positive association with plankton volume. There was no correlation between factor 3 scores and environmental data for the first canonical correlation (Table 6). Discussion Species composition Ichthyoplankton typical of the Gulf of Maine and the Northwest Atlantic was observed in Penobscot Bay: lipa- rids, sculpins ( Myoxocephalus spp. ), Ammodytes sp., P. gun- nellus, U. subbifurcata , and P. americanus. The dominant species are similar to those previously reported for areas of the Maine coast, including the Sheepscot estuary (Graham and Boyar, 1965; Hauser, 1973; Shaw, 1981), the central Maine coast and estuaries (Chenoweth, 1973), the Dam- ariscotta River estuary (Lee, 1975; Laroche, 1980; 1982; Townsend, 1981; 1984), and Sullivan Harbor (Townsend, 1984). Of the twenty-two species collected in the Damar- iscotta estuary and 21 species collected in Sullivan Harbor (Townsend, 1984), 17 and 18 of these species, respectively, occurred in Penobscot Bay. The dominant species in these two systems was P. gunnellus along with the three species of Myoxocephalus and a stichaeid, Lumpenus lampraetifor- mis (Townsend, 1984). Twenty-two kinds of boreal larvae with centers of abundance north of the mid-Atlantic coast were found in the Damariscotta and Sheepscot estuaries and along the central coast of Maine (Chenoweth, 1973), and 18 of these taxa were also found in Penobscot Bay. Pholis gunnellus, Liparis sp., C. maculatus, L. lumpraeti- formis, and Cottidae accounted for 91% of the total catch; the number of larvae declined sharply in spring with the end of the larval stage of these dominant species before reaching a low point in July and August (Chenoweth, 1973). The ichthyoplankton of the St. Lawrence estuary, principally osmerids, gadids, cottids, cyclopterids and pleu- ronectids, consisted of 25 species (Able, 1978) and 15 of these also occurred in Penobscot Bay. Significance of demersal eggs The ichthyoplankton community of Penobscot Bay is dom- inated by larvae that hatch from demersal eggs within the estuary, whereas larvae that hatched from pelagic eggs are rare. This finding was first noted for the Mystic River estuary (Pearcy and Richards, 1962) and subsequently for the Sheepscot and Damariscotta estuaries (Chenoweth, 1973). Able (1978) observed that the St. Lawrence estuary was almost exclusively inhabited by larvae that hatched from demersal eggs, primarily cottids, stichaeids, P. ameri- canus, and C. harengus and that the larvae from pelagic eggs were merely strays from more offshore waters. In all of these areas, larvae that hatched from demersal eggs dominated the catch within the estuary, whereas larvae that hatched from pelagic eggs were more common at the mouth of the estuary or adjacent ocean. Table 6 Canonical correlation coefficients for relative larval fish abundance factor scores and environmental variables in 1997 and 1998. Underlined values significant at P = 0.05, adjusted for multiple comparisons such that P < 0.004 for a significant correlation. 1997 1998 Variables Standardized coefficients Correlation with canonical variables Standardized coefficients Correlation with canonical variables Larval fish relative abundance Factor 1 0.63 0.66 -0.66 -0.65 Factor 2 -0.58 -0.61 -0.74 -0.74 Factor 3 0.46 0.49 -0.16 -0.09 Environment Depth -0.06 -0.19 0.24 0.36 Plankton -0.02 -0.37 -0.28 -0.82 Temperature 1.00 1.00 0.66 0.94 Salinity 0.07 -0.21 0.19 0.37 Lazzari: Dynamics of larval fish abundance in Penobscot Bay, Maine 91 The dominant fishes in Penobscot Bay lay demersal and adhesive eggs (Bigelow and Schroeder, 1953). Fourteen of 16 species (87%) of larvae collected in Penobscot Bay in 1998 and and 17 of 22 (78%) of the larvae collected in 1997 hatched from demersal eggs. Twenty-two species of larvae were taken in the central coastal area of Maine and a com- paratively large number of pelagic-egg species (41%) com- pared with demersal-egg species, as might be expected in an area strongly influenced by coastal water (Chenoweth, 1973). In upper Sheepscot Bay, 42 species of fish larvae were taken (Shaw, 1981), but this relatively high number reflects the long time series (nine years) of this survey and the inclusion of several coastal species that were rare in the total catch. Even so, these catches were dominated by demersal egg-laying species (76%). In the Damariscotta estuary and Sullivan Harbor, only one of 22 species (<5%) taken as larvae came from pelagic eggs (Townsend, 1981; 1984). The ichthyoplankton of the St. Lawrence estuary were almost exclusively forms from demersal eggs (Able, 1978) and only seven of 25 (28%) larval forms collected came from pelagic eggs. However, pelagic eggs did occur in the St. Lawrence estuary, but these were usually nonresi- dent species, such as S. scombrus and G. morhua, whose larvae were virtually absent from the upper bay. Gadus morhua were very rare in Penobscot Bay and the Sheep- scot estuary (Chenoweth, 1973; Shaw, 1981) and were ab- sent from the Damariscotta estuary and Sullivan Harbor (Townsend, 1984). Typically, the abundance of larvae from demersal eggs peaked in the winter-spring period whereas larvae from pelagic eggs peaked in the summer (Che- noweth, 1973). Egg type has major implications on larval fish ecology. Coastal shelf spawners, such as G. morhua , E. cimbrius, H. plattesoides , and T. adspersus lay pelagic eggs that may be dispersed over a wide area by the counterclock- wise current flow in the Gulf of Maine (Sherman et ah, 1984). Nearshore spawners, such as P americanus, P. gun- nellus , Myoxocephalus spp., U. subbifurcata , and C. haren- gus lay demersal eggs and depend on protected areas of the coastline for nursery areas. Larvae that hatch from de- mersal eggs are less likely to be transported out of their nearshore nursery grounds, rather they are entrained in the landward moving bottom layer (Pearcy and Richards, 1962; Norcross and Shaw, 1984). Variation in distribution and abundance Larval fish assemblage composition in Penobscot Bay varied considerably over time but not among sampling locations. Interannual and monthly differences in both spe- cies composition and abundance of ichthyoplankton in the bay appeared to be associated with differences in environ- mental conditions, particularly with temperature. Tempo- ral patterns of larval fish abundance corresponded with seasonality of reproduction (Bigelow and Schroeder, 1953). Larvae that hatched from demersal eggs from late winter through early spring, such as Myoxocephalus spp.,Ammo- dytes sp., and Pholis guunellus, were abundant in Penob- scot Bay in March and April and gradually their numbers in plankton samples declined. Such a spawning pattern suggests that adults use the bay and nearshore areas for an extended period of time. Larvae of taxa that hatch later in spring such as P. americanus, L. atlanticus and U. subbifur- cata, were abundant in Penobscot Bay in May and June. These results suggest that temperature is the principal factor controlling the temporal occurrence of fish larvae in Penobscot Bay. Temperature is a primary factor asso- ciated with the occurrence and distribution of young stag- es of fish inhabiting temperate and cold oceans (Oben- chain, 1981; Frank and Leggett, 1982; Laprise and Pepin, 1995). This association may be critical in ecosystems char- acterized by a short growing season (Conover, 1992). How- ever, the influence of temperature on larval occurrence no doubt resulted from species-specific responses of spawn- ing adults. The late winter-early spring species, such as Myoxocephalus spp., Ammodytes sp., and Pholis gunnel- lus, had a negative association with temperature, whereas late-spring spawners, such as P. americanus , L. atlanti- cus, and U. subbifurcata, had a positive association. The synchronous emergence of the larvae of several species of fishes in the planktonic environment may be a strategy to further reduce predation (Frank and Leggett, 1983) re- gardless of temperature. The association observed in this study between salinity and ichthyoplankton may be arti- factual. Although the influences of temperature and sa- linity could not be separated, it seems unlikely that the minute differences in salinity observed between samples would greatly affect marine animals. Penobscot Bay plays an important role in the early life history of fish inhabiting the central coast of Maine by offering favorable habitat for ichthyoplankton. A hydro- graphic front off of Penobscot Bay separates the coastline into two different hydrographic regimes (Bigelow, 1927). West of Penobscot Bay, a combination of increased runoff and reduced tidal mixing, favors more rapid development of vertical stratification in spring and summer. East of Pe- nobscot Bay, tidal mixing is enhanced and the develop- ment of vertical stratification is much reduced t hroughout the summer months. Vertical stability of the water column is controlled more by salinity stratification than by tem- perature, and peaks of stability are the result of influxes of low-salinity surface waters (Townsend, 1984). Therefore, the peak in abundance of both phytoplankton and zoo- plankton occurs earlier in the western Gulf of Maine and gradually spreads to the east with the onset of vertical stratification during spring and summer (Bigelow, 1927). Marine fishes may be categorized as those that use the estuaries, such as Penobscot Bay, as primary spawning and nursery areas and those that do not. Winter-early- spring spawning species ( P. gunnellus, Ammodytes sp., H. americanus, Myoxocephalus spp. ) belong to the first group of resident demersal fishes whose larvae hatch from de- mersal eggs and use the bays and estuaries as nursery ar- eas. Typically, these larvae disappear from collections in April and May. Two species that spawn in late spring, P. americanus and U. subbifurcata, also use Penobscot Bay as a spawning and nursery habitat and have areas of greatest larval abundance in the midbay and lower bay, re- spectively. The larvae of E. cimbrius, G. morhua, T. adsper- sus, S. scombrus and H. plattessoides, although present, 92 Fishery Bulletin 99(1 ) were not abundant in spring and early summer; therefore Penobscot Bay does not appear to be a primary nursery area for these species. Possibly the numbers of spawning adults of these species are low in the bay or the pelagic eggs that they produce are dispersed before hatching, or both the number of spawning adults are low and the pe- lagic eggs they produce are dispersed before hatching. Pe- nobscot Bay appears to act as a nursery for many fishes; therefore degradation of water quality during the vernal period would have wide reaching effects on the nearshore fish community. Acknowledgments I would like to thank Captain Kevin Lapham for his excel- lent command of the RV Nucella and Lou VanGuelpen of the Atlantic Reference Center at Huntsman Marine Laboratory, New Brunswick. Stan Chenoweth and three anonymous reviewers provided helpful comments to an early draft. This study was supported by a grant from the National Environmental, Data, and Satellite Information Service (NESDIS) through the Island Institute, Rockland, Maine. Literature cited Able, K. W. 1978. Ichthyoplankton of the St. Lawrence estuary: compo- sition, distribution and abundance. J. Fish. Res. Board Can. 35:1518-1531. Bigelow, H. B. 1926. Plankton of the offshore waters of the Gulf of Maine. Bull. U.S.Bur. Fish., vol. XL, 1924, part II, 509 p. 1927. Physical oceanography of the Gulf of Maine. Bull. U.S. Bur. Fish. 40:511-1027. Bigelow, H. B., and W. C. Schroeder. 1953. Fishes of the Gulf of Maine. U.S. Fish and Wildl. Serv. Fish. Bull. 53(74), 477 p. Chenoweth, S. B. 1973. Fish larvae of the estuaries and Coast of central Maine. Fish. Bull. 71( 1):105-113. Conover, D. O. 1992. Seasonality and the scheduling of life history at dif- ferent latitudes. J. Fish Biol. 41 (suppl. B):161-178. Cushing, D. H. 1975. Marine ecology and fisheries. Cambridge Univ. Press, Cambridge, England, 278 p. Day, R. W., and G. P. Quinn. 1989. Comparisons of treatments after an analysis of vari- ance in ecology. Ecological Monographs 59:433-463. Fish, C. J., and M. W. Johnson. 1937. The biology of the zooplankton in the Bay of Fundy and Gulf of Maine with special reference to production and distribution. J. Biol. Board Can. 3:189-322. Frank, K. T., and W. C. Leggett. 1982. Coastal water mass replacement: its effect on zoo- plankton dynamics and the predator-prey complex associ- ated with larval capelin (Mallotus villosus ). Can. J. Fish. Aquat. Sci. 39:991-1003. 1983. Multispecies larval fish associations: accident or adap- tation? Can. J. Fish. Aquat. Sci. 40:754-762. Gauch, H. G. 1982. Multivariate analysis in community ecology. Cam- bridge Univ. Press, Cambridge, England, 298 p. Graham, J. J. 1972. Retention of larval herring within the Sheepscot estu- ary of Maine. Fish. Bull. 70 (21:299-305. Graham, J. J., and H. C. Boyar. 1965. Ecology of herring larvae in the coastal waters of Maine. Internationa Commission for the Northwest Atlan- tic Fisheries (ICNAF) Spec. Publ. 6:625-634. Haefner, P. A. 1967. Hydrography of the Penobscot Bay (Maine) estuary. J. Fish. Res. Board Can. 24:1553-1571. Hauser, W. J. 1973. Larval fish ecology of the Sheepscot River-Montsweag Bay estuary, Maine. PhD. diss., Univ. Maine, Orono, ME, 79 p. Hjort, T. J. 1914. Fluctuations in the great fisheries of northern Europe viewed in light of biological research. Rapp. P.-V Cons. Perm. Int. Explor. Mer 20, 228 p. Laprise, R., and P. Pepin. 1995. Factors influencing the spatio-temporal occurrence of fish eggs and larvae in a northern, physically dynamic coastal environment. Mar. Ecol. Prog. Ser. 122:73-92. Laroche, J. L. 1980. Larval and juvenile abundance, distribution, and larval food habits of the larvae of five species of sculpins (Family: Cottidae) in the Damariscotta River estuary, Maine. PhD. diss., Univ. Maine, Orono, ME, 169 p. 1982. Trophic patterns among larvae of five species of scul- pins (Family: Cottidae) in a Maine estuary. Fish. Bull. 80:827-840. Lee, W. Y. 1975. Succession and some aspects of population dynamics of copepods in the Damariscotta River estuary, Maine. PhD. diss., Univ., Maine, Orono, ME, 181 p. Leggett, W. C., and E. Deblois. 1994. Recruitment in marine fishes: Is it regulated by star- vation and predation in the egg larval stages? Neth. J. Sea Res. 32:119-134. Marak, R. R. 1960. Food habits of larval cod, haddock and coalfish in the Gulf of Maine and Georges Bank area. J. Cons. Int. Explor. Mer 25:147-157. Marak, R. R., and J. B. Colton Jr. 1961. Distribution of fish eggs and larvae, temperature and salinity in the Georges Bank, Gulf of Maine area, 1953. U.S. Fish Wildl. Serv. Spec. Sci. Rep. Fish. 398, 61 p. 1962. Distribution of fish eggs and larvae, temperature and salinity in the Georges Bank, Gulf of Maine area, 1956. U.S. Fish Wild. Serv. Spec. Sci. Rep. Fish. 411, 66 p. Norcross, B. L., and R. F. Shaw 1984. Oceanic and estuarine transport of fish eggs and larvae: a review. Trans. Am. Fish. Soc. 113(2):153-165. Obenchain, C. L. 1981. A study of the larval fish community in the New York Bight, July 1974 to June 1976. Rapp. P.-V Reun. Cons. Int. Explor. Mer 178:217-219. Pearcy, W. G., and S. W. Richards. 1962. Distribution and ecology of fishes of the Mystic River, Connecticut. Ecology 43(2):248-259. Pielou, E. C. 1984. The interpretation of ecological data. John Wiley and Sons, New York, NY, 286 p. Lazzari: Dynamics of larval fish abundance in Penobscot Bay, Maine 93 Shaw, R. 1981. Seasonal species composition, diversity, spatial distri- butions and tidal retention and transport of ichthyoplank- ton in the Sheepscot River-Back River-Montsweag Bay estuary system, Maine. PhD. diss., Univ. Maine, Orono ME, 285 p. Sherman, K.W., W. Smith, W. Morse, and M. Berman. 1984. Spawning strategies of fishes in relation to circu- lation, phytoplankton production and pulses in zooplank- ton off the northeast United States. Mar. Ecol. Prog. Ser. 18:1-19. Sokal, R.R., and F. J. Rohlf. 1981. Biometry. W. H. Freeman and Company, New York, NY, 859 p. Stevens, J. 1986. Applied multivariate statistics for the social sciences. Lawrence Erlbaum and Associates, Hillsdale, NJ, 341 p. Townsend, D.W. 1981. Comparative ecology and population dynamics of larval fishes and zooplankton in two hydrographically dif- ferent areas on the Maine coast. PhD. diss., Univ. Maine, Orono, ME, 282 p. 1983. The relations between larval fishes and zooplankton in two inshore areas of the Gulf of Maine. J. Plankton Res. 5(2): 145—173. 1984. Comparison of inshore zooplankton and ichythyo- plankton in the Gulf of Maine. Mar. Ecol. Prog. Ser. 15: 79-90. 94 Abstract— Microsatellite DNA analy- sis was applied in a genetic study of 20 chinook salmon populations from four regions within the Fraser River drainage of British Columbia, Canada. Twelve populations were represented by samples collected in different years. A total of 2612 fish were examined at three microsatellite loci. Each locus was highly polymorphic, with 30 alleles at OfslOl, 28 alleles at OtslOO, and 35 alleles at Ofsl02. Average observed het- erozygosities were 86%, 88%, and 71%, respectively. In a dendrogram analysis of pairwise genetic distances, four geo- graphically based groups were observed consisting of the lower Fraser River, the middle Fraser River, the upper Fraser River, and the Thompson River. An analysis of molecular variance showed that 97.57% of the genetic variance was within populations and 1.80% of the genetic variance was partitioned among populations. We detected significantly different allele frequencies among pop- ulations within regional groupings and temporal stability in allele frequencies in populations for which multiple years of samples were analyzed. Regional divergence may reflect colonization pat- terns following the last ice age, and divergence among populations within regions may reflect local adaptation. The elucidation of population structure of chinook salmon of the Fraser River watershed will be useful information for management designed to conserve genetic biodiversity. Manuscript accepted 14 July 2000. Fish. Bull. 99:94-107 (2001). Population structure of Fraser River chinook salmon ( Oncorhynchus tshawytscha): an analysis using microsatellite DNA markers R. John Nelson Maureen P. Small Terry D. Beacham K. Janine Supernault Pacific Biological Station Nanaimo, British Columbia V9R 5K6, Canada Present address (for R J Nelson): SeaStar Biotech Inc. 32056-3749 Shelbourne St. Victoria, British Columbia V8P 5S2 Canada E-mail address (for R. J. Nelson) |nelson@seastarbio.com The Fraser River watershed produces greater numbers of Pacific salmon than any other river system in British Colum- bia (B.C.). Approximately 65 tributar- ies of the Fraser River are used as spawning and rearing habitat for chi- nook salmon (Oncorhynchus tshawyts- cha), and these streams produce up to one third of the commercial catch of chinook salmon from Brirish Columbia (Fraser et ah, 1982). Although chinook salmon account for only 1% to 5% of the total escapements of salmon within the watershed (Northcote and Atagi, 1997), these fish are an important cultural, sporting, and food resource. Chinook salmon populations in the Fraser water- shed have been negatively impacted by a variety of forces, in some cases reducing ( Bradford, 1994) or completely eliminating (Slaney et ah, 1996) local populations. Historical efforts to maintain and en- hance salmon runs through trans- plantation have had mixed results, il- lustrating that the characteristics of a population influence its ability to thrive in a given environment (Wood, 1995). Also, transplantation of fish and hatchery production practices may al- ter genetic composition of wild stocks (Waples, 1994). In fisheries, it is impor- tant not to over harvest small popula- tions that may contain unique adap- tive traits. For the above reasons it is advantageous to understand how pop- ulation structures evolve in order to protect individual salmon runs and to preserve biodiversity. Most of the chinook salmon popula- tions of B.C. were founded after the ice of the Wisconsin glaciation retreat- ed approximately 10,000 to 15,000 year ago (McPhail and Lindsey, 1986). If chi- nook salmon recolonization is similar to that of sockeye and coho salmon (Wood et ah, 1994; Small et ah, 1998a), re- establishment of the present day B.C. chinook salmon populations may have occurred from at least two different sources, Beringia to the north and Cas- cadia to the south (Gharrett et ah, 1987), and possibly from a refuge in the Queen Charlotte Islands (Warner et ah, 1982). The genetic character of the founding fish may be reflected in pres- ent day genetic structure, but because Pacific salmon return to their natal streams to spawn (Scheer, 1939; Quinn, 1984), reproductive isolation can lead to divergence of phenotypic and geno- typic characters. Neutral genetic mark- ers can be used to measure the degree of reproductive isolation and potential for local adaptation. Over the years, a variety of methods have been used to examine population structure. Allozyme analysis has long been a mainstay in fish genetics re- search and has been used to determine population structure in chinook salm- on of Alaska (Gharrett et ah, 1987), from California to Alaska (Utter et ah, 1989), of the Yukon River (Beacham et Nelson et al.: Population structure of Oncorhynchus tshawytscha of the Fraser River 95 al., 1989; Wilmot et al., 1992), of California and Oregon (Bartley et al. 1992), and in British Columbia (Teel et al. 2000). These studies suggest that chinook salmon popula- tions were genetically heterogeneous and that populations could be placed into genetically defined groups correspond- ing to geographic regions. In some of the earlier studies there was little genetic distinction between geographical- ly separate groups because the allozyme markers showed low polymorphism. However, using 25 polymorphic allo- zymes, Teel et al. (2000) detected strong population diver- gence within the Fraser River and among major rivers in British Columbia. DNA markers can be more polymorphic than allozyme markers and thus may be more sensitive to population structure; with higher levels of polymorphism, there is an increased likelihood for populations to contain unique al- leles or to have frequency differences in alleles that are shared among populations. Among the DNA-based mark- ers, mitochondrial DNA has been used to examine genetic structure in chinook salmon populations of the West Coast of North America (Wilson et al., 1987; Cronin et al., 1993). These studies suggest that there is structuring among West Coast chinook salmon populations. However, the low resolution of this method limits its utility. Minisatellite DNA has been used to study Canadian chinook salmon populations. Beacham et al. (1996) found that chinook salmon formed two major regional groups in British Co- lumbia: a southern group consisting of populations of the Fraser River, Vancouver Island, and the southern main- land; and a northern group consisting of populations of the Skeena River, the Yukon River, and the northern main- land. However, owing to technical complexity, the tech- nique is unsuitable for studies involving large numbers of individuals. Microsatellite DNA loci are highly polymorphic and technically easy to use (Nelson et al., 1998) and provide powerful markers for elucidating population structure. Microsatellite loci have provided information regarding population divergence in chinook salmon (Banks et al., 1996) and other salmonids (Angers et al., 1995; Scribner et al., 1996; Nelson et al., 1998; Small et al., 1998a, 1998b). In our study we exploited the ease of analysis and the highly polymorphic nature of microsatellite DNA loci to study population structure of chinook salmon. We surveyed variation at three microsatellite loci within 20 Fraser River chinook salmon populations and examined temporal stability of microsatellite allele frequencies. We used this information to hypothesize the genetic structure of chinook salmon populations within the Fraser River watershed. Materials and methods DNA extraction Liver or scale samples were analyzed from 2612 individual chinook salmon from 20 populations of the Fraser and Thompson River watersheds (Fig. 1.). Sample sizes ranged from 30 to 347 fish (Table 1). Liver samples were obtained from spawning wild adults. Hatchery adults were sources for the Chehalis-red and Chilliwack-red samples. The nomenclature “-red” refers to the red flesh color of the fish in the population. DNA was extracted from liver and scales archived on scale cards according to the methods of Nelson et al. ( 1998). Liver samples taken prior to 1994 were subjected to DNA extractions as described in Beacham et al. (1996). Each 25 pL of polymerase chain reaction (PCR) required either 100 ng of genomic DNA, 0.1 to 1 pL of liver extracts, or 5 to 10 pL of scale extract. PCR amplification The loci amplified in this work were OfslOO (Nelson et al., 1998), OislOl (Small et al., 1998a) and Ofsl02 (Nelson and Beacham, 1999). PCR amplification was carried out in 96-well microtiter plates with a MJ PTC- 100 thermal cycler (MJ research, Watertown, MA). 25-pL PCR reactions contained 10 pmol (0.4 pM) of each primer, 80 pM of each nucleotide, 20 mM tris-pH 8.8, 2 mM MgS04, 10 mM KC1, 0.1% triton x-100, 10 mM (NH4)S04, and 0.1 mg/mL of bovine serum albumin. Primer set OtslOO required a 10% final volume of glycerol in the PCR. PCR temperature cycles were preceded by a denaturation incubation of 3 min at 94°C; samples then were held at 80°C while 1 unit of DNA polymerase was added. PCR cycle parameters and primer sequences for each locus are presented in Table 2. Three pLs of lOx loading dye (50 mM EDTA pH 8.0, 30% glycerol, 0.25%' bromphenol blue) were added to each reac- tion and ten pLs of this solution was loaded on each gel lane for electrophoresis. Gel electrophoresis Microsatellite alleles were size-fractionated on nondena- turing polyacrylamide gels 17 cm wide by 14.5 cm long. Gels consisted of a 19:1 ratio of acrylamide to bis-acryl- amide. Gel contained 2x TAE buffer (Maniatis et al., 1982) as did the gel box reservoirs. Electrophoretic conditions are described in Table 2. Each gel included three 20 base- pair (bp) marker lanes (GenSura Labs Inc., Del Mar, CA) to create a molecular size grid for sizing amplified micro- satellites, 24 population samples, and one “standard fish” to estimate the precision of allele sizes (Table 3). Standard deviations were calculated for alleles from two different standard fish for each primer set. Gels were stained with 0.5 pg/mL of ethidium bromide in water and visualized with ultraviolet light (Fig. 2). Digital images of gels were obtained as described in Nelson et al. (1998). Individual alleles were identified by using the procedure outlined in Small et al. (1998a). A four-bp bin was used for all Ots 101 alleles. A four-bp bin was used to define smaller alleles of OtslOO and five- to eight-bp bins were used for larger alleles. A four-bp bin was used for the smaller alleles of Ots 102 and five- to six- bp bins were used for larger alleles. These bin sizes (see Table 1 for bin designations) were four or more standard deviations wide according to estimates derived from the standard fish. Bins are referred to as “alleles” throughout the text. 96 Fishery Bulletin 99(1 ) Table 1 Allele frequencies, observed heterozygosity (H0), and expected heterozygosity (H) at loci OtslOl, OtslOO, and Ofsl02 for 20 chinook salmon zygosity. The allele number (in basepairs) is the lower limit of the bin. The weighted mean of allele frequencies for regions (L Fr=lower Fraser; 1 OfslOl Chilliwack Tete Chilliwack Harrison White LFr Quesnel Stuart Nechako Chilko Bridge Cottonwood Mid Fr Jaune Red Alleles 326 181 507 186 294 187 120 56 51 893 249 29 142 0 0 0 0 0.002 0 0 0 0 0.001 0 0 146 0 0 0 0.008 0.003 0 0.004 0 0 0.003 0 0 150 0 0.003 0.001 0.054 0.02 0.045 0.008 0 0 0.029 0.002 0.017 154 0.003 0.003 0.003 0.056 0.066 0.048 0.042 0.027 0.029 0.053 0.008 0 158 0 0.008 0.003 0.062 0.029 0.059 0.025 0.045 0.127 0.048 0.028 0 162 0.005 0.003 0.004 0.016 0.02 0.064 0.017 0.089 0.078 0.036 0.06 0.052 166 0.035 0.017 0.029 0.046 0.02 0.027 0.017 0.089 0 0.029 0.014 0.034 170 0.014 0.025 0.018 0.054 0.01 0.003 0.017 0.027 0 0.019 0.004 0 174 0.015 0.039 0.024 0.024 0.01 0.019 0.05 0.009 0 0.02 0.036 0 178 0.046 0.028 0.039 0.019 0.015 0.019 0.071 0.071 0.059 0.03 0.042 0.017 182 0.067 0.014 0.048 0.03 0.029 0.037 0.025 0.027 0.049 0.031 0.032 0.034 186 0.069 0.05 0.062 0.054 0.066 0.08 0.038 0.045 0.118 0.064 0.034 0.069 190 0.075 0.11 0.088 0.032 0.097 0.061 0.117 0.089 0.108 0.079 0.112 0.017 194 0.112 0.066 0.096 0.129 0.075 0.04 0.146 0.098 0.118 0.092 0.078 0.121 198 0.126 0.155 0.136 0.097 0.138 0.123 0.117 0.063 0.108 0.117 0.104 0.121 202 0.118 0.08 0.105 0.081 0.109 0.107 0.083 0 0.088 0.091 0.08 0.138 | 206 0.09 0.072 0.084 0.056 0.068 0.061 0.083 0.045 0.059 0.064 0.058 0.069 210 0.063 0.077 0.068 0.048 0.071 0.04 0.025 0.134 0.049 0.057 0.098 0.034 214 0.044 0.075 0.055 0.038 0.048 0.029 0.004 0.018 0 0.031 0.026 0 218 0.015 0.003 0.011 0.022 0.024 0.053 0.021 0.036 0 0.028 0.042 0.086 222 0.026 0.022 0.025 0.016 0.01 0.021 0.008 0.036 0 0.015 0.06 0.138 226 0.026 0.041 0.032 0.022 0.017 0.011 0.004 0 0 0.013 0.04 0.017 230 0.017 0.03 0.022 0.011 0.015 0.013 0.021 0 0 0.013 0.014 0.017 234 0.014 0.014 0.014 0 0.01 0.011 0.013 0.009 0 0.008 0.012 0 238 0.005 0.008 0.006 0 0.005 0.011 0.021 0.027 0 0.008 0.012 0 242 0.006 0.017 0.01 0.005 0.007 0.008 0.017 0.018 0 0.008 0 0 246 0.003 0.019 0.009 0.008 0.007 0.005 0 0 0.01 0.006 0 0.017 250 0.003 0.011 0.006 0.008 0 0.003 0 0 0 0.002 0 0 254 0.002 0.008 0.004 0.005 0.005 0 0 0 0 0.003 0 0 259 0 0.003 0.001 0 0.002 0 0.008 0 0 0.002 0 0 Hobs 0.844* 0.889 0.86 0.855 0.881 0.877 0.933 0.839 0.686* 0.868 0.863* 0.897 Hexp 0.926 0.934 0.927 0.94 0.932 0.941 0.925 0.929 0.922 0.939 0.936 0.931 Ots 1 00 Chilliwack Tete Chilliwack Harrison White LFr Quesnel Stuart Nechako Chilko Bridge Cottonwood Mid Fr Jaune Red Alleles 347 168 515 183 226 174 122 55 53 813 254 30 150 0.001 0 0.001 0 0 0 0 0.018 0 0.001 0 0 207 0.01 0 0.007 0 0.011 0.011 0 0.018 0 0.007 0 0 211 0.016 0.009 0.014 0.008 0.018 0.009 0.004 0 0 0.009 0 0 215 0 0 0 0.003 0.004 0.003 0 0.018 0 0.004 0 0 219 0 0 0 0.003 0 0 0 0 0 0.001 0 0 227 0.001 0.009 0.004 0 0 0 0 0 0 0 0 0 231 0.001 0 0.001 0 0 0 0 0 0 0 0 0.017 235 0.004 0 0.003 0.003 0 0.006 0.078 0.009 0 0.014 0 0 Nelson et al.: Population structure of Oncorhynchus tshawytscha of the Fraser River 97 Tabie 1 populations from the Fraser River drainage. Populations out of Hardy- Weinberg equilibrium are Mid Fr=middle Fraser; U Fr=upper Fraser) are given. Sample sizes are provided below population ndicated by * next to the observed hetero- names. Chehalis Red Bowron Holmes Indianpoint Slim UFr L Shuswap M Shuswap Eagle Coldwater Nicola Thompson 30 55 49 40 70 522 192 195 33 37 233 690 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0.003 0 0 0 0 0.001 0 0 0 0.013 0 0.003 0.023 0.008 0 0 0.019 0.015 0.033 0.018 0 0.013 0 0.009 0.023 0.023 0 0 0.028 0.022 0.017 0.036 0 0 0 0.018 0.013 0.015 0 0.027 0.004 0.011 0.017 0.036 0.02 0.063 0 0.043 0.029 0.005 0.03 0.027 0.026 0.021 0 0.009 0.02 0.025 0 0.013 0.013 0.021 0.121 0.054 0.028 0.028 0 0 0 0.013 0 0.003 0.003 0.023 0 0.014 0.039 0.021 0 0.027 0.02 0.013 0 0.023 0.005 0.008 0.015 0.014 0.002 0.006 0.05 0.064 0.092 0.013 0.057 0.048 0.008 0.003 0.106 0.054 0.019 0.017 0.017 0.036 0 0.025 0.057 0.032 0.063 0.085 0.106 0.068 0.105 0.086 0.15 0.091 0.051 0.05 0.007 0.048 0.044 0.074 0.091 0.122 0.06 0.064 0.15 0.082 0.153 0.087 0 0.093 0.065 0.095 0.242 0.054 0.026 0.068 0.05 0.091 0.143 0.138 0.007 0.081 0.182 0.215 0.121 0.041 0.099 0.153 0.133 0.136 0.061 0.175 0.164 0.12 0.195 0.208 0.015 0.068 0.09 0.148 0.1 0.036 0.143 0.1 0.029 0.08 0.068 0.044 0.03 0.068 0.058 0.056 0.067 0.091 0.092 0.05 0.1 0.071 0.109 0.054 0.03 0.054 0.071 0.074 0.067 0.091 0.102 0.063 0.121 0.093 0.094 0.046 0.045 0.108 0.073 0.072 0.017 0.018 0 0.025 0.093 0.03 0.042 0.026 0.015 0.095 0.097 0.057 0 0.018 0.01 0.025 0.007 0.031 0.005 0.018 0 0.068 0.06 0.03 0.033 0.027 0.01 0.063 0.136 0.065 0 0.008 0 0.027 0.026 0.012 0.067 0.045 0.031 0.025 0.114 0.049 0 0.021 0 0 0.021 0.013 0.017 0.036 0.041 0.025 0.05 0.025 0 0 0 0.014 0.009 0.004 0.017 0 0 0 0.036 0.011 0 0.003 0.015 0.014 0.013 0.007 0 0.009 0 0 0.014 0.009 0.008 0 0 0.014 0 0.003 0 0 0 0 0.007 0.001 0.003 0 0 0 0.015 0.006 0 0 0.01 0 0 0.002 0 0 0 0 0.011 0.004 0 0 0 0 0 0 0.003 0 0 0 0 0.001 0 0 0 0 0 0 0 0 0.015 0 0.002 0.001 0 0 0 0 0 0 0 0 0 0 0 0 0.867 0.964 0.776 0.975 0.814 0.86 0.865 0.841 0.836* 0.784 0.888 0.851 0.933 0.927 0.898 0.925 0.9 0.933 0.891 0.882 0.879 0.946 0.94 0.92 Chehalis Red Bowron Holmes Indianpoint Slim UFr L Shuswap M Shuswap Eagle Coldwater Nicola Thompson 30 50 46 39 70 519 210 192 31 62 268 713 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0.01 0 0 0 0.001 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0.024 0 0.002 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0.001 0 0 0 0 0 0 0 0.04 0 0.013 0 0.005 0 0 0 0 0.002 0.001 con tin ued 98 Fishery Bulletin 99(1 ) Table 1 (continued) O/slOO (continued) Chilliwack Tete Chilliwack Harrison White LFr Quesnel Stuart Nechako Chilko Bridge Cottonwood Mid Fr Jaune Red Alleles 347 168 515 183 226 174 122 55 53 813 254 30 239 0.001 0.003 0.002 0.098 0.027 0.032 0.115 0.027 0.009 0.056 0.002 0.1 243 0 0.006 0.002 0.049 0.051 0.029 0.078 0.027 0.085 0.05 0.016 0.067 247 0.006 0.006 0.006 0.022 0.02 0.02 0.012 0 0 0.017 0.033 0.033 251 0.006 0.009 0.007 0.005 0.029 0.009 0 0 0 0.011 0.081 0.033 255 0.004 0 0.003 0.044 0.027 0.055 0.012 0.045 0.009 0.034 0.039 0.017 260 0.019 0.03 0.022 0.101 0.093 0.055 0.012 0.018 0.019 0.065 0.063 0.117 266 0.039 0.024 0.034 0.063 0.1 0.069 0.004 0.027 0.028 0.061 0.039 0.05 271 0.058 0.083 0.066 0.022 0.044 0.023 0.045 0.091 0.057 0.039 0.059 0.1 276 0.032 0.068 0.044 0.06 0.024 0.034 0.004 0.027 0.047 0.033 0.065 0.05 281 0.13 0.083 0.115 0.137 0.08 0.083 0.07 0.073 0.198 0.099 0.091 0.067 288 0.101 0.134 0.112 0.145 0.053 0.075 0.078 0.055 0.057 0.082 0.106 0.017 293 0.111 0.089 0.104 0.046 0.058 0.124 0.053 0.118 0.142 0.078 0.128 0.133 299 0.118 0.077 0.105 0.057 0.097 0.055 0.102 0.1 0.094 0.08 0.089 0.067 305 0.086 0.092 0.088 0.06 0.066 0.057 0.07 0.082 0 0.06 0.057 0.067 311 0.13 0.119 0.126 0.038 0.084 0.069 0.131 0.027 0.075 0.073 0.053 0 318 0.084 0.107 0.091 0.025 0.075 0.095 0.119 0.1 0.17 0.082 0.02 0.033 326 0.026 0.045 0.032 0.003 0.035 0.08 0.012 0.091 0.009 0.036 0.033 0.033 334 0.007 0.003 0.006 0.005 0.004 0.003 0 0.027 0 0.005 0.026 0 342 0.006 0 0.004 0.003 0 0.006 0 0 0 0.002 0 0 358 0.003 0.003 0.003 0 0 0 0 0 0 0 0 0 Hobs 0.876 0.869 0.868 0.907 0.947 0.908 0.893 0.8* 0.849 0.905 0.882 0.867 Hexp 0.911 0.917 0.913 0.923 0.938 0.937 0.918 0.927 0.887 0.939 0.929 0.933 Ots 102 Chilliwack Tete Chilliwack Harrison White LFr Quesnel Stuart Nechako Chilko Bridge Cottonwood Mid Fr Jaune Red Alleles 302 180 482 174 255 129 114 47 48 767 262 29 . 134 0 0 0 0 0 0 0 0 0 0 0 0 163 0 0 0 0 0 0 0 0 0 0 0 0 167 0 0 0 0 0 0.008 0.004 0.011 0 0.003 0 0 171 0.003 0 0.002 0 0.002 0 0 0 0.021 0.002 0 0 175 0 0.008 0.003 0.02 0.045 0.004 0.009 0 0.01 0.022 0 0.086 179 0.015 0.008 0.012 0.052 0.049 0.07 0.031 0.106 0 0.051 0.008 0.138 183 0.065 0.106 0.08 0.178 0.267 0.163 0.364 0.138 0.229 0.233 0.302 0.224 187 0.038 0.017 0.03 0.066 0.055 0.066 0.039 0.053 0 0.053 0.006 0.017 192 0.023 0 0.015 0 0.004 0.027 0.004 0 0.01 0.007 0 0 197 0.003 0.014 0.007 0.006 0.025 0.008 0.035 0.011 0 0.017 0 0 201 0.007 0.003 0.005 0.02 0.004 0.008 0.013 0 0.01 0.01 0 O b -4 205 0.008 0.014 0.01 0.011 0.002 0.008 0 0 0 0.005 0 0 209 0.018 0.047 0.029 0.026 0.012 0.023 0.009 0 0.042 0.018 0.013 0.034 213 0.041 0.047 0.044 0.086 0.065 0.07 0.066 0.245 0.115 0.085 0.073 0.034 217 0.142 0.106 0.129 0.075 0.057 0.101 0.083 0.043 0.313 0.087 0.042 0.069 221 0.151 0.122 0.14 0.057 0.043 0.054 0.083 0.043 0.01 0.052 0.055 0.086 226 0.098 0.106 0.101 0.023 0.029 0.07 0.018 0.064 0 0.033 0.029 0.017 231 0.046 0.081 0.059 0.057 0.122 0.089 0.026 0.053 0.021 0.077 0.015 0.017 Nelson et at: Population structure of Oncorhynchus tshawytscha of the Fraser River 99 Table t (continued) Chehalis Red 30 Bowron 50 Holmes 46 Indianpoint 39 Slim 70 UFr 519 L Shuswap 210 M Shuswap 192 Eagle 31 Coldwater 62 Nicola 268 Thompson 713 0.033 0.08 0.011 0.09 0.057 0.032 0.002 0 0 0 0.007 0.003 0.033 0.08 0.033 0.103 0.057 0.039 0.002 0.01 0 0 0 0.004 0 0 0.011 0 0.007 0.02 0 0.003 0 0 0 0.001 0 0 0.033 0.051 0 0.048 0.007 0.003 0.016 0 0.007 0.006 0.017 0.02 0.022 0.013 0.007 0.027 0.055 0.044 0.016 0.008 0 0.029 0.167 0.12 0.087 0.141 0.164 0.099 0.048 0.036 0.016 0 0 0.025 0.017 0.03 0.065 0.026 0.029 0.038 0.083 0.065 0.065 0.073 0.083 0.076 0.15 0.04 0.054 0.051 0.1 0.069 0.138 0.094 0.065 0.016 0.018 0.076 0.017 0.11 0.098 0.051 0.121 0.075 0.136 0.096 0.258 0.016 0.014 0.083 0.133 0.15 0.185 0.064 0.093 0.104 0.155 0.201 0.081 0.056 0.023 0.115 0 0.03 0.054 0.077 0.007 0.067 0.067 0.07 0.032 0.073 0.087 0.073 0.183 0.09 0.087 0.09 0.136 0.122 0.112 0.188 0.065 0.073 0.14 0.135 0.1 0.1 0.12 0.064 0.114 0.093 0.076 0.063 0.097 0.065 0.08 0.074 0.017 0.06 0.054 0.077 0.05 0.056 0.069 0.065 0.065 0.145 0.158 0.102 0.017 0.02 0.022 0.038 0.036 0.039 0.026 0.052 0.065 0.032 0.08 0.052 0.05 0 0.043 0.026 0.014 0.022 0.01 0.005 0.032 0.242 0.131 0.067 0.05 0.01 0.011 0.026 0.007 0.026 0.005 0.005 0.097 0.153 0.106 0.053 0.017 0.01 0 0 0 0.014 0.007 0 0.032 0.024 0.062 0.025 0 0 0.011 0 0 0.001 0.002 0 0 0 0.002 0.001 0 0 0 0 0 0 0 0 0 0 0 0 0.933 0.86 0.826 0.923 0.814 0.871 0.914 0.88 0.806 0.855 0.907 0.886 0.9 0.92 0.913 0.923 0.9 0.929 0.9 0.885 0.903 0.871 0.918 0.919 Chehalis Red 30 Bowron 50 Holmes 44 Indianpoint 40 Slim 66 UFr 521 L Shuswap 185 M Shuswap 167 Eagle 37 Coldwater 54 Nicola 231 Thompson 674 0 0 0 0 0 0 0.027 0 0.054 0 0 0.01 0 0 0 0 0 0 0.03 0 0.027 0 0 0.01 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0.008 0.012 0 0.009 0.026 0.015 0 0.02 0.023 0.025 0.045 0.016 0 0.012 0.041 0.167 0.134 0.065 0 0.07 0.08 0.063 0.091 0.041 0.024 0.033 0.149 0.102 0.18 0.093 0.05 0.19 0.193 0.138 0.167 0.233 0.116 0.207 0.135 0.046 0.058 0.114 0.017 0.04 0.011 0 0.015 0.012 0.084 0.174 0.108 0.065 0.017 0.083 0 0.01 0 0.013 0.008 0.003 0 0.009 0 0 0.011 0.006 0 0 0 0.025 0.015 0.004 0.003 0 0 0 0 0.001 0.033 0.02 0.045 0.013 0 0.01 0.003 0 0 0 0 0.001 0 0 0.011 0.025 0.015 0.005 0.003 0.006 0 0 0.002 0.003 0.017 0.03 0.057 0.087 0.061 0.032 0.027 0.051 0 0.019 0.019 0.028 0.167 0.24 0.205 0.162 0.152 0.12 0.035 0.057 0.041 0.148 0.115 0.077 0.183 0.14 0.091 0.087 0.121 0.079 0.049 0.036 0.041 0.157 0.195 0.104 0.1 0.06 0.023 0.038 0.015 0.051 0.024 0.009 0.027 0.046 0.009 0.017 0.05 0.04 0.011 0.038 0.03 0.03 0.014 0.021 0 0.009 0.015 0.015 0.017 0.04 0 0.038 0.03 0.02 0.022 0.006 0 0.046 0.022 0.019 continued Fishery Bulletin 99(1) 100 Table 1 (continued) Ots 102 (continued) Alleles Harrison 302 Chilliwack White 180 LFr 482 Quesnel 174 Stuart 255 Nechako 129 Chilko 114 Bridge 47 Cottonwood 48 Mid Fr 767 Tete Jaune 262 Chilliwack Red 29 236 0.043 0.067 0.052 0.083 0.045 0.085 0.053 0.032 0.021 0.059 0.053 0.034 241 0.065 0.05 0.059 0.055 0.045 0.043 0.022 0.085 0.031 0.045 0.132 0.069 246 0.05 0.047 0.049 0.009 0.024 0.016 0.031 0.053 0 0.02 0.074 0.017 251 0.03 0.042 0.034 0.026 0.022 0.019 0.009 0 0.01 0.018 0.048 0.034 256 0.038 0.022 0.032 0.032 0.012 0.008 0.044 0.011 0.073 0.024 0.122 0.034 261 0.028 0.039 0.032 0.052 0.022 0.023 0.022 0.032 0.063 0.032 0.021 0 267 0.022 0.017 0.02 0 0.029 0.016 0.009 0 0 0.014 0 0.034 273 0.015 0.003 0.01 0.009 0.016 0.008 0.009 0 0 0.01 0 0 279 0.005 0.003 0.004 0.011 0.002 0 0 0 0.021 0.005 0 0 285 0.01 0.014 0.011 0.006 0 0.004 0 0 0 0.002 0 0 291 0.013 0 0.008 0 0 0.004 0.013 0 0 0.003 0 0.034 297 0.01 0.014 0.011 0.009 0 0.004 0 0 0 0.003 0.004 0 303 0.003 0 0.002 0.029 0.004 0.004 0.004 0.021 0 0.01 0.004 0 309 0.002 0.006 0.003 0 0 0 0 0 0 0 0 0 325 0.005 0 0.003 0.003 0 0 0 0 0 0.001 0 0 321 0.002 0 0.001 0 0 0 0 0 0 0 0 0 333 0.002 0 0.001 0 0 0 0 0 0 0 0 0 Hobs 0.714* 0.628* 0.699 0.632* 0.761* 0.713* 0.684* 0.723 0.833 0.714 0.775* 0.862 Hexp 0.927 0.928 0.927 0.931 0.894 0.93 0.842 0.894 0.833 0.906 0.859 0.897 Table 2 Primer sequences, PCR cycle, and gel-running conditions for microsatellite loci OtslOO, OfslOl, and Otsl02. Gel conditions Locus Primer sequence PCR cycle denature Anneal Extend No. of cycles % acrylamide Voltage (V) Time (h) Ots 100 F 5TGAACATGAGCTGTGTGAG-3' R 5'-ACGGACGTGCCAGTGAG-3'. 94°C/20s 57°C/20s 72°C/20s 30 7 60 18 OfslOl F 5'ACGT CT G ACTT CAAT GAT GTTT-3' R 5'TATTAATTAT CCT CCAACCCAG-3' 94°C/30s 53°C/30s 72°C/30s 30 8 70 17 Ots 102 F 5'AGG AT CCAATAAGG AGT G ATA-3' R 5ACTAGGTATCCCCTTAACCA-3' 94°C/20s 50°C/10s 72°C/20s 30 6 60 17 Data analysis A pedigree analysis was performed on chinook salmon families to document the inheritance of microsatellite alleles at each locus. Chinook salmon families were obtained from domesticated strains originating in Robertson Creek and Big Qualicum River, B.C. For each locus, we examined inheritance in 6 families (12 offspring per family). Popu- lation genetic data were analyzed by using GENEPOP version 3.1 (Raymond and Rousset, 1995a) and ARLEQUIN version 1.1 (Schneider et al., 1997). Allele frequencies and heterozygosities were estimated for each population at each locus and conformation to Hardy- Weinberg equilibrium (HWE) was tested with a simulated Fisher’s exact test (Guo and Thompson, 1992). A gametic disequilibrium test was also performed. Differences in allele frequencies among dif- ferent sample years, populations, and regions (populations grouped into regions) were examined by using pairwise tests in ARLEQUIN which is analogous to Fisher’s exact test (described in Raymond and Rousset, 1995b). An analy- sis of molecular variation (AM OVA) was also performed by using ARLEQUIN to measure the distribution of molecu- lar variance at several levels: among individuals, among samples taken in different years for the same population, among all populations (with year classes combined), and among regions. F-statistics (Wright, 1951) and their stan- Nelson et al.: Population structure of Oncorhynchus tshawytscha of the Fraser River 101 ' Table 1 (continued) j r Chehalis Red 30 Bowron 50 Holmes 44 Indianpoint 40 Slim 66 UFr 521 L Shuswap 185 M Shuswap 167 Eagle 37 Coldwater 54 Nicola 231 Thompson 674 0.033 0.02 0 0.125 0.038 0.047 0.016 0.006 0 0.019 0.006 0.01 0.2 0.01 0.136 0.013 0.053 0.102 0.005 0 0 0 0.009 0.004 0.017 0.01 0.034 0.025 0.008 0.046 0.014 0.003 0 0 0.006 0.007 0.083 0 0.034 0.025 0.03 0.039 0.008 0.009 0 0 0.004 0.006 0.017 0.01 0.023 0.025 0.03 0.073 0.016 0.009 0 0 0 0.007 0 0.03 0.023 0 0.008 0.016 0.016 0.006 0.014 0 0.002 0.007 0 0 0 0 0 0.002 0.019 0.003 0 0 0 0.006 0.017 0 0 0 0 0.001 0.003 0.006 0.014 0 0.004 0.004 0 0 0 0 0 0 0.016 0.015 0 0 0 0.008 0 0 0 0 0 0 0.011 0.021 0.081 0.009 0 0.013 0 0 0 0.025 0.023 0.007 0.103 0.066 0.068 0 0 0.048 0 0.01 0 0.013 0.038 0.009 0.116 0.096 0.095 0 0.011 0.065 0 0 0 0 0.008 0.003 0.057 0.093 0.041 0.111 0.089 0.08 0 0.01 0 0 0 0.001 0.041 0.027 0 0.046 0.065 0.044 0 0 0 0 0 0 0.016 0.003 0.027 0 0 0.007 0 0 0 0 0 0 0.038 0.003 0.041 0 0 0.013 0 0 0 0 0 0 0.038 0.003 0 0 0 0.011 0.9 0.8 0.75 0.875 0.829 0.808 0.535* 0.575* 0.297* 0.796 0.68* 0.602 0.9 0.88 0.886 0.925 0.857 0.896 0.941 0.898 0.92 0.889 0.883 0.933 dard deviations were calculated for each locus and for all loci combined according to Weir and Cockerham ( 1984) by using FSTAT (Goudet, 1995). FSTAT also provided pair- wise values for populations and regions. We used the notations F„F ,andF for Weir and Cockerham’s ( 1984) 6, f, and F, respectively. In all analyses, probability values were adjusted to correct for the number of simultaneous tests as discussed by Lessios (1992). Thus, the significance level is taken to be 0.05/number of simultaneous comparisons. Graphical analysis of genetic relationship between pop- ulations was performed. A neighbor-joining (NJ) (Saitou and Nei, 1987) dendrogram was constructed by using PHY- LIP 3.5c (Felsenstein1). The allele frequency matrix was resampled 1000 times in a bootstrap resampling proce- dure and Cavalli-Sforza and Edwards (1967) chord dis- tances among populations were estimated for each matrix. A consensus NJ dendrogram of chord distances was gener- ated to determine the stability of nodes within the dendro- gram. For presentation, a NJ dendrogram with the origi- nal branch lengths was constructed and bootstrap values over 50% were plotted onto the nodes of the dendrogram. In several populations we suspected a null allele at Ots 102. These populations had moderate heterozygote de- ficiencies and a large number of individuals whose other two loci amplified but that failed to amplify at Ots 102, even after PGR stringency was relaxed. We generated an estimate of the frequency of the null allele by recoding, as null allele homozygotes, two out of every three of the indi- viduals failing to amplify at Ots 102. Corrected allele fre- Table 3 Observed precision of allele size determination (bp) for repeated analysis of heterozygous fish at loci Ots 101, OtslOO, and Otsl02. “n” is the number of times each fish was analyzed. Standard deviation (SD) was calculated and the range of observed measurements is shown. Locus n Mean SD Range OtslOl 44 221.77 1.08 220-224 103 199.47 0.96 196-201 44 167.93 0.79 166-169 103 153.49 0.94 151-155 Ots 100 18 367.17 1.86 360-367 103 322.07 1.73 318-325 18 281.28 1.23 279-283 103 251.81 1.13 249-254 Ots 102 74 270.35 1 34 268-273 70 225.54 1.10 223-228 70 188.30 1.08 186-190 74 181.15 0.86 180-183 1 Felsenstein, J. 1993. PHYLIP (Phylogeny Inference Package), version 3.4. Univ. Washington, Seattle, WA. (Available from author, Department of Genetics, Box 357360, Univ. Washington, Seattle, WA 98195-7360.] 102 Fishery Bulletin 99(1 ) Figure 1 Map of the Fraser River watershed showing locations of chinook salmon populations with inset of the map of British Columbia. Numbers are placed at the collection sites where the populations were obtained: 1 = Harrison, 2 = Chilliwack-white, 3 = Chilliwack -red, 4 = Chehalis-red, 5 = Bridge, 6 = Coldwater, 7 = Nicola, 8 = Lower Shuswap, 9 = Middle Shus- wap, 10 = Eagle, 11 = Chilko, 12 = Quesnel, 13 = Stuart, 14 = Nechako, 15 = Cottonwood, 16 = Bowron, 17 = Holmes, 18 = Tete Jaune, 19 = Indianpoint, 20 = Slim. quencies in the presence of a null allele were generated by using the maximum likelihood method of the utilities op- tion in GENEPOP. The data set with corrected allele fre- quencies was tested in a NJ dendrogram analysis. Results Pedigree analysis and allele assignment All loci displayed normal Mendelian inheritance; each hetero- zygous parent (two allele bands) passed each of its two bands to approximately 50% of its offspring, and each homozygous parent (one allele band) passed its single band to 100% of its offspring. Assignment of alleles was based on the empirically determined standard error of band size estimation as reported in Table 3. The size range for each allele was set to allow for 95% confidence of allele assignment. Heterozygosity and allele frequencies Heterozygosity was consistently high at Ots 10 1 and Ots 100 (Table 1), ranging from 0.636 to 0.975 at Ots 101(0. 86 aver- age), and from 0.80 to 0.947 (0.88 average) at OtslOO. Het- Nelson et al.: Population structure of Oncorhynchus tshawytscha of the Fraser River 103 KL L Ots 101 L -300 -200 KL L Ots 100 L -400 -300 K L L Ots 102 L -300 -280 -260 -240 -220 -200 -180 Figure 2 Photographs of nondenaturing polyacrylamide gels stained with ethidium bro- mide showing typical alleles for OfslOl, OfslOO, and Ofsl02. “K” indicates the 1-kilobase-pair ladder and “L" indicates the 20 base-pair ladder. erozygosity values had a wider range at Ots 102, ranging from a low of 0.297 in Eagle River to a high of 0.9 in Che- halis-red (average of 0.706). The apparently low hetero- zygosity found in some populations at Ots 102 may have been partially due to the presence of a null allele. With the exception of Bridge River, all allele frequencies showed significant variation among the different populations. In pairwise tests, Bridge River was not significantly differ- ent ( P nondifferentiation=0. 00004) from Harrison River, Chilliwack-white, Chilliwack-red, Chehalis-red, Coldwater River, Middle Shuswap River, Eagle River, Stuart River, and Cottonwood River. All other populations were signifi- cantly different from each other. Hardy-Weinberg equilibrium and treatment of the null allele We tested each of the 20 populations for significant devi- ations from HWE proportions at each locus. Conforma- tion to HWE was rejected at the 5% level (P<0. 05/20) in four populations (Harrison River, Tete Jaune River, Cot- tonwood River, and Eagle River) at OfslOl, in one pop- ulation (Bridge River) at Ots 100, and in 12 populations at Ots 102 (see Table 1). All rejections of conformation to HWE were due to a deficiency of heterozygotes. Single locus Fls values for OfslOl, OfslOO, and Ofsl02 were 0.079, 0.050, and 0.261, respectively, and 0.127 for 104 Fishery Bulletin 99(1) 59% 6 1 °< J87% \ 54% 70% Quesnel 1 2 MF Stuart 13 MF Nechako 14 MF _ Chilko 11 MF Lower Shuswap 8 TR 70% L___' Middle Shuswap 9 TR Eagle 10 TR 59% Coldwater 6 TR Nicola 7 TR Bridge 5 MF ■fr 67% Tete Jaime 18 LTF Red Chilliwack 3 UF Slim 20 UF Indianpoint 19 UF Red Chehalis 4 UF — ■— — Cottonwood 16 MF Holmes 17 UF ■ Bowron 15 UF Hanison 1 LF Wlrite Chilhwack 2 LF Genetic distance 0.03 i 0.02 —I — 0.01 —I — Figure 3 Neighbor-joining dendrogram of 20 chinook salmon populations from the Fraser River watershed. The dendrogram shows branch lengths, and the bootstrap values at the nodes were the percent- age of dendrograms in which the populations beyond the node were grouped together out of 1000 dendrograms. The numbers follow- ing the place names correspond to map numbers in Figure 1. “MF”, “TR, “UF” and “LF” indicate middle Fraser, Thompson River, upper Fraser and lower Fraser River respectively. all loci combined. All values were significantly greater than zero (P<0.005), indicating that pop- ulations tended towards disequilibrium at each locus. Gametic disequilibrium tests indicated that loci were unlinked. Because of a suspected null allele at Ots 102, most populations showed a deficiency of heterozy- gotes at this locus; this locus also had a high and significant Fis value. When the dendrogram analy- sis was redone with the corrected data set, the re- gional structure was improved, in that all the mid- dle Fraser River populations grouped together and bootstrap values were higher for regional nodes. Thus the major conclusion of this study was un- affected by the suspected null allele. However be- cause the null allele was not observed in the pedi- gree analysis, its presence remains speculative and we have kept the original data set for the analysis presented here. Observed homozygote excesses are likely not due to our inability to resolve alleles be- cause we observed a large number of individuals that amplified at the other two loci of the study but failed to amplify at Ots 102, suggesting the pres- ence of a null homozygote. Observed heterozygote deficiency could also be due to a partial disequi- librium; however no single population showed dis- equilibrium at all three loci. Temporal stability of allele frequencies, and population and regional variability Comparisons of samples from different years for the same population suggested that there is temporal stability in allele frequencies within populations. With the exception of a single comparison within the Harrison River samples (the 1988 samples dif- fered from the 1992 samples [P nondifferentia- tion=0.0003]), sample sets within populations were indistinguishable. Analysis of molecular variance showed that 97.57% (P<0.01, 5719 df) was within populations and that 1.80% (P<0.01, 16 df) of the total genetic variance was among populations. A small but significant 0.63%, (P<0.01, 3 df) of the genetic variance was apportioned between regions. Because population year classes were not signif- icantly different from each other and variability among populations exceeded variability among year classes, year classes were combined into single populations for the rest of the analysis. All regions were significantly different from each other in pairwise tests (P<0.008). Dendrogram To determine if there was a pattern to allele frequency differences, we constructed a NJ dendrogram (Fig. 3) of pairwise genetic distances. This analysis suggested that geography at least partially underlies genetic relation- ships among chinook salmon populations in the Fraser River drainage. The NJ dendrogram consisted of four ma- jor branches, and are congruent with regional groupings (Fig. 3). The lower Fraser River branch includes Harrison River and Chilliwack and is supported in 96% of the den- drograms. The Thompson River branch is supported at 70% and includes Bridge River (from the middle Fraser River) in 59% of the dendrograms. The middle Fraser River branch is supported in 59% of the dendrograms. The larg- est branch, supported in 32% of the dendrograms, includes all the upper Fraser River populations, as well as the Chill- iwack-red and Chehalis-red hatchery populations, and Cot- tonwood River from the middle Fraser River region. This graphic analysis of genetic relationships shows re- gional groupings consisting of lower, middle, and upper Fraser River population groups, and a well-defined Thomp- son River group. These groupings are modestly supported because most of the genetic variance is within populations and only a small amount of the variance is among regions. Population and regional differences Wright’s F statistics (Wright, 1951) were calculated to determine the degree of structuring between and within the regional population groupings. When all the popula- tions were kept separate, the Fst values indicated diver- Nelson et al.: Population structure of Oncorhynchus tshawytscha of the Fraser River 105 gence among populations with single locus values of 0.011 (SD=0.002) for Ots 101, 0.021(SD=0.QQ4) for OtslOO and 0.038 (SD=0.007) for Otsl02. The multiple locus value was 0.023 (SD=0.008). To determine the pat- terns of genetic relationship between sample groups within the individual regional groupings, pairwise Fst values were calculated for each pair of samples (Table 4). The highest pairwise Fst values were observed between populations of different regional groups. In order to esti- mate the degree of genetic isolation between the regional groupings sug- gested by dendrogram analysis, popu- lations were combined by geographic region and pairwise Fsl values were calculated (Table 5). The greatest value observed was between the Thompson River and upper Fraser River regions (0.0161) and the lowest value was between the upper and middle Fraser regions (0.006). Fst values were significant at each locus: Ofs 101 0.006 (SD=0.002), OfslOO 0.008 (SD=0.00)1, and Ofsl02 0.022 (SD=0.008), for combined loci 0.012 (SD=0.005) (all P<0.005). Discussion Our results show significant genetic diversity within and between chi- nook salmon populations spawning in the tributaries of the Fraser River. The genetic relationship between the populations from the different trib- utaries suggest that there are four regional assemblages: upper Fraser River, middle Fraser River, lower Fraser River, and the Thompson River. These regional assemblages are concordant with the interpre- tation of the population structure seen by Teel et al. (2000) based on allozyme analysis. This concordance was observed with two very differ- ent marker sets. We employed three markers with average observed het- erozygosities of 86%, 88%, and 71%, whereas out of the 25 polymorphic enzymes used by Teel et al., the high- est heterozygosity was 0.441. The concordance of these two marker types with geography suggests that the genetic differences observed are not due simply to genetic drift but 106 Fishery Bulletin 99(1 ) Table 5 Table of pairwise Fst values among major regions in the Fraser River drainage. Lower Fraser Mid Fraser Upper Fraser Lower Fraser 0.0000 Mid Fraser 0.0115 0.0000 Upper Fraser 0.0147 0.0062 0.0000 Thompson 0.0146 0.0125 0.0161 rather reflect patterns of historical colonization and pres- ent day gene flow (or lack thereof). Concordance of both allozyme and microsatellite data indicates that the simple method used in our study to designate alleles does not greatly bias or skew the results. During the retreat of the Wisconsin glacial ice sheet, the headwaters of the Fraser and Thompson Rivers were ice free before the lower Fraser River channel was ice free, and therefore drained through the Columbia River. McPhail and Lindsey (1986) suggested that freshwater fish colonized the upper Fraser River and the Thompson River by means of the Columbia River during this time. Al- lozyme analysis of chinook salmon show that the Thomp- son River populations are distinct from the population of the other Fraser River tributaries (Utter et al., 1989; Teel et al., 2000). Similarly, coho salmon from the Thompson River are genetically distinct from coho salmon in the low- er Fraser River (Small et al., 1998a). In our analysis, the Thompson River populations formed a distinct group, con- sistent with the hypothesis that the Thompson River wa- tershed may have been colonized by a different founder group than other regions of the Fraser River. In addition, strong genetic substructuring within the Thompson River watershed was observed. This structuring, also observed by Teel et al. (2000), suggests that there may be sufficient genetic isolation within the Thompson River watershed to allow for the persistence of locally adapted populations. If the upper Fraser and Thompson Rivers were both col- onized by means of the Columbia River, then tests of the genetic relationship between these groups might show the upper Fraser and Thompson Rivers more closely related to each other than to other regions. However, based upon our study, they are the most distantly related, suggesting that either the upper Fraser was not colonized by the same population that founded the Thompson River populations, or that migration may have obscured the origins of fish inhabiting this region. This hypothesis is currently being tested by analyzing chinook salmon populations from the Skeena and Nass Rivers (possible source populations) and by increasing the resolution of the genetic structure of coho salmon in the Fraser River watershed by analyzing more microsatellite loci. The high bootstrap support for the lower Fraser group suggests that this region was colonized by a single founder population. This hypothesis seems likely because the riv- er mouths are separated by approximately 15 kilometers. The close genetic association between the Harrison and Chilliwack Rivers (Fs/=0.003) indicates either that stray- ing occurs routinely between them or that colonization was so recent that the populations have not diverged. The red-flesh Chehalis and Chilliwack populations were introduced in the 1980s from broodstocks originating in the upper and middle Fraser River. Sources of the brood stocks included the Bowron, Chilko, and Quesnel Rivers and Slim Creek2 These populations are artificially main- tained by selecting red-flesh fish for broodstock. Our anal- ysis places these populations with the upper Fraser River populations, reflecting their origins. The regional groupings and patterns of genetic relation- ships within each region provide a starting point for dis- cussion of the events that led to the repopulating of these regions by chinook salmon and the degree of isolation of dif- ferent populations. Although only in the early stages, this information forms a base upon which to begin assigning management and fishery enhancement priorities such that genetic diversity present in wild populations is preserved. This information, along with life history and ecological da- ta, will also be useful for the determination of whether a regional grouping of populations can be considered an evo- lutionarily significant unit (Waples, 1991). Acknowledgments Samples were collected with the assistance of J. Candy and the staff of the Habitat and Assessment Branch of the Department of Fisheries and Oceans. Funding was provided by the Department of Fisheries and Oceans. C. Wood and R. E. Withler provided helpful discussion. We thank five anonymous reviewers for improvements to the manuscript. Literature cited Angers, B., L. Bernatchez, A. Angers, and L. Desgroseillers. 1995. Specific microsatellite loci for brook charr reveal strong population subdivision on a microgeographic scale. J. Fish. Biol, (suppl A) 47:177-185. Banks, M. A., B. A. Baldwin, and D. Hedgecock. 1996. Research on chinook salmon ( Oncorhynchus tshaw- ytscha) stock structure using microsatellite DNA. Bull. Natl. Inst. Aquacult. suppl. 2:5-9. Bartley, D., B. Bentley, J. Brodziak, R. Gomulkiewicz, M. Mangel, and G. A. E. Gall. 1992. Geographic variation in population genetic structure of chinook salmon from California and Oregon. Fish. Bull. 90:77-100. Beacham, T. D., C. B. Murray, and R. E. Withler. 1989. Age, morphology and biochemical genetic variation of Yukon River chinook salmon. Trans. Am. Fish. Soc. 118: 46-63. Beacham, T. D., R. E. Withler, and T. A. Stevens. 1996. Stock identification of chinook salmon ( Oncorhynchus tshawytscha) using minisatellite DNA variation. Can. J. Fish. Aquat. Sci. 53:380-394. 2 Cross, C. 1998. Personal commun. Department of Fisheries and Oceans, 555 West Hastings St. Vancouver, B.C., V6B 5G3, Canada. Nelson et al.: Population structure of Oncorhynchus tshawytscha of the Fraser River 107 Bradford, M. J. 1994. Trends in the abundance of chinook salmon ( Oncorhyn - chus tshawytscha ) of the Nechako River, British Columbia. Can. J. Fish. Aquat. Sci. 51:965-973. Cavalli-Sforza, L. L., and A. W. F. Edwards. 1967. Phylogenetic analysis: models and estimation proce- dures. Am. J. Hum. Gen. 19:233-257. Cronin, M. A., W. J. Spearman, R. L. Wilmot, J. C. Patton, and J. W. Bickham. 1993. Mitochondrial DNA variation in chinook ( Oncorhyn- chus tshawytscha ) and chum salmon (O. Keta) detected by restriction enzyme analysis of polymerase chain reaction (PCR) products. Can. J. Fish. Aquat. Sci. 50:708-715. Fraser, F. J., P. J. Starr, and A. Y. Fedorenko. 1982. A review of the chinook and coho salmon of the Fraser River. Can. Tech. Rep. Fish. Aquat. Sci. 1126, 132 p. Gharrett, A. J., S. M. Shirley, and G. R. Tromble. 1987. Genetic relationships among populations of Alaskan chinook Salmon ( Oncorhynchus tshawytscha). Can. J. Fish. Aquat. Sci. 44:765-774. Goudet, J. 1995. FSTAT, version 1.2: a computer program to calculate F-statistics. J. Hered. 86:485-486 Guo, S. W., and E. A. Thompson. 1992. Performing the exact test of Hardy- Weinberg propor- tions for multiple alleles. Biometrics 48:361-372 Lessios, H. A. 1992. Testing electrophoretic data for agreement with Hardy- Weinberg expectations. Mar. Biol. 112:517-523. Maniatis, T., E. F. Fritsch, and J. Sambrook. 1982. Molecular cloning: a laboratory manual. Cold Spring Harbor Laboratory, Cold Spring Harbor Laboratory, New York, NY. McPhail, J. D., and C. C. Lindsey. 1986. Zoogeography of freshwater fishes of Cascadia (the Columbia system and rivers north to the Stikine). In Zoogeography of North American freshwater fishes (C. H. Hocutt and E. O. Wiley, eds.), p. 615-637. J. Wiley and Sons, Inc., New York, NY. Nelson, R. J., and T. D. Beacham. 1999. Isolation and cross species amplification of micro- satellite loci useful for study of Pacific salmon. Animal Genetics 30:228-229. Nelson, R. J., T. D. Beacham, and M. P. Small. 1998. Microsatellite analysis of the population structure of a Vancouver Island sockeye salmon (Oncorhynchus nerka ) stock complex using nondenaturing gel electrophoresis. Mol. Mar. Biol. Biotech. 7:312-319. Northcote, T. G., and D. Y. Atagi. 1997. Pacific salmon abundance trends in the Fraser River watershed compared with other British Columbia Systems. In Pacific salmon and their ecosystems: status and future options (D. J. Stouder, P. A. Bisson, and R. J. Naimons, eds.), p. 199-219. Chapman and Hall, New York, NY. Quinn, T. P. 1984. Homing and straying in Pacific Salmon. In Mecha- nisms of migration in fishes ( J. D. McCleave, G. P. Arnold, J. J. Dodson, and W. H. Neill, eds.), p. 357-362. Plenum Press, New York, NY. Raymond, M. and F. Rousset. 1995a. GENEPOP (versions 1.2 and 3.1): population genetics software for exact tests and ecumenism. Heredity 86:248- 249. Raymond, M., and F. Rousset. 1995b. An exact test for population differentiation. Evolu- tion 49:1280-1283. Saitou, N. and M. Nei. 1987. The neighbor-joining method: a new method for reconstructing phylogenetic trees. Mol. Biol. Evolution 4: 406-425. Scheer, B. T. 1939. Homing instinct in salmon. Q. Rev. Biol. 14:408-430. Schneider, S., J. M. Kueffer, D. Roessli, and L. Excoffier. 1997. Arlequin, verison 1.1: a software for population genetic data analysis. Genetics and Biometry Laboratory, Univ. Geneva, Switzerland. Scribner, K.T., J. R. Gust, and R. L. Fields. 1996. Isolation and characterization of novel salmon mic- rosatellite loci: cross-species amplification and population genetic application. Can. J. Fish. Aquat. Sci. 53:833-841. Slaney, T. L., K. D. Hyatt, T. G. Northcote, and R. J. Fielden. 1996. Status of anadromous salmon and trout in British Columbia and Yukon. Fisheries 21:20-35. Small, M.P, T. D. Beacham, R. E. Withler, and R. J. Nelson 1998a. Discriminating coho salmon ( Oncorhynchus kisutch ) populations within the Fraser River, British Columbia. Mol. Ecol. 7:141-155. Small, M. P, R. E. Withler, and T. D. Beacham. 1988b. Population structure and stock identification of Brit- ish Columbia coho salmon, Oncorhynchus kisutch , based on microsatellite DNA variation. Fish. Bull. 96:843-858. Teel, D. J., G. B. Milner, G. A. Winans, and W. S. Grant. 2000. Genetic population structure and origin of life-his- tory types in chinook salmon in British Columbia, Canada. Trans. Am. Fish. Soc. 129:194-209. Utter, F., G. B. Milner, G. Stahl, and D. Teel. 1989. Genetic population structure of chinook salmon, Oncorhynchus tshawytscha, in the Pacific Northwest. Fish. Bull. 87:239-264. Waples, R. S. 1991. Definition of “species” under the Endangered Species Act: application to Pacific salmon. U.S. Dep. Commer., NOAA Tech. Memo., NMFS, F/NWC-194, p. 29. Waples, R. S. 1994. Genetic considerations in recovery efforts for Pacific Salmon. Cons. Biol. 8:884-886. Warner, B. G., R. W. Mathewes, and J. J. Clague. 1982. Ice-Free conditions on the Queen Charlotte Islands, British Columbia, at the height of the late Wisconsin gla- ciation. Science (Washington D.C.) 218:675-677. Weir, B. S. and C. C. Cockerham. 1984. Estimating F-statistics for the analysis of population structure. Evolution 38:1358-1370. Wilmot, R. L., R. Everett, W. J. Spearman, and R. Baccus. 1992. Genetic stock identification ofYukon River chum and chinook salmon 1987 to 1990. Alaska Fish Wild. Res. Cent. U.S. Fish Wild. Serv. Prog. Rep. 132 p. Wilson, G. M., W. K. Thomas, and A. T. Beckenbach. 1987. Mitocondrial DNA analysis of Pacific Northwest pop- ulations of ( Oncorhynchus tshawytscha). Can. J. Fish. Aquat. Sci. 44:1301-1305. Wood, C. C. 1995. Life history variation and population structure in sock- eye salmon. Am. Fish. Soc. Symp. 17:195-216. Wood, C. C„ B. E. Riddell, D. T. Rutherford, and R. E. Withler. 1994. Biochemical genetic survey of sockeye salmon ( Oncorhynchus nerka) in Canada. Can. J. Fish. Aquat. Sci. 51(suppl. 11:113-131. Wright, S. 1951. The genetical structure of populations. Ann. Eugen. 15:323-354. 108 Abstract— Annual and batch fecundi- ties of yellowfin sole, Limanda aspera, in the eastern Bering Sea were deter- mined. Most individuals had a determi- nate, group-synchronous mode of oocyte development as evidenced by a dis- tinctly separate distribution of fully yolked oocytes in more advanced ova- ries. The spawning of batches by an indi- vidual female occurs in uninterrupted succession, as indicated by the presence of oocytes with migratory-stage nuclei in nearly all females undergoing oocyte hydration. Total fecundity ranged from 295,615 to 3,635,108 oocytes per female and is described by 3.3225 x TL 3-6312, where T = total fish length (cm). The length-fecundity relationship was found to be the same in both southeast and northwest areas of the eastern Bering Sea, despite known growth differences between the two areas. Individual females spawn from 8 to 11 batches. The first batch spawned is generally smaller in number than succeeding batches. After spawning begins, fish remain in the nearshore spawning area (<30 m bottom depth) until spent. The presence of residual chorion tissue from unspawned ova in the ovary lumen of some maturing females indicates that at least some females are capable of spawning more than one series of batches within one reproductive season. The annual fecundity, therefore, for such individuals is consequently con- sidered indeterminate Manuscript accepted 10 August 2000. Fish. Bull. 99:108-122 (2001). Annual and batch fecundities of yellowfin sole, Limanda aspera, in the eastern Bering Sea Daniel G. Nichol Erika I. Acuna Resource Assessment and Conservation Engineering Division Alaska Fisheries Science Center National Marine Fisheries Service, NOAA 7600 Sand Point Way NE, BIN C15700 Seattle, Washington 98115-0070 E-mail address (for D. G. Nichol): dan nichol@noaa gov Yellowfin sole, Limanda aspera,1 have historically been among the more abun- dant fishes in the eastern Bering Sea, where biomass estimates have exceeded 2 million metric tons (t) annually since 19802 * 4 (Wilderbuer et ah, 1992; Nichol, 1998). Yellowfin sole has been an impor- tant commercial trawl species with annual catches averaging 135,630 t froml991 to 1998. 2 Yellowfin sole spawn from May through August in nearshore waters of Bristol Bay northward to at least Nunivak Island (Fadeev, 1970; Nichol, 1995) at depths less than 30 m (Nichol, 1995). Yellowfin sole, like many other flatfishes, are batch spawners (Nichol, 1995). They spawn pelagic eggs that have been observed in the plank- ton with diameters of approximately 0.76-0.85 mm (Waldron, 1981; Mata- rese et ah, 1989). Adult yellowfin sole generally undergo long migrations from wintering grounds near the shelf-slope break to spring-summer grounds at bottom depths less than 50 m (Wak- abayashi, 1989; Nichol, 1998). Annual fecundity has been used as a measure of reproductive output for many species; its use as a parameter in fishery population models makes it relevant to stock assessment. Annual fecundity is defined as the total num- ber of eggs spawned by a female in a single year. Batch fecundity, defined as the number of eggs released at one time, can be estimated for any female from counts of hydrated oocytes as long as the female is not ovulating or spawn- ing. Directly estimating annual fecundi- ty from counts of advance-stage oocytes is feasible only if the number of eggs to be spawned is determinate or fixed prior to spawning (Hunter et al., 1985, Hunt- er et ah, 1992). Females with a determi- nate fecundity, just prior to spawning, will have a fixed number of advanced- yolked oocytes that are separated from the other less developed oocytes by a distinct gap in size; these can either be spawned in batches or all at once. In contrast, species whose ovaries are characterized by continuous oocyte size distributions may be able to develop unyolked oocytes continually and add them to the stock of advanced-yolked oo- cytes even after spawning begins. These species will typically develop multiple groups of oocytes with size distribu- tions that overlap. The method to esti- mate annual fecundity in such species with the least error uses batch fecun- dity, spawning frequency, and spawn- ing season duration (Yamamoto, 1956; Hunter and Macewicz, 1985a). Prior to estimating annual fecundity, we exam- ined oocyte size distributions to deter- mine which mode of oocyte development yellowfin sole undergoes. Assuming that yellowfin sole females have a determinate fecundity, two po- tential problems associated with esti- mating annual fecundity must be ad- dressed. First, because yellowfin sole spawn eggs in batches, annual fecundi- ty may be underestimated if partially spawned fish are included in estimates. 1 Scientific name follows Cooper and Chap- leau ( 1998). 2 Wilderbuer, T. K., and D. Nichol. 1998. Yellowfin sole. Section 3 in Stock assess- ment and fishery evaluation report for the groundfish resources of the Bering Sea/ Aleutian Islands regions. North Pacific Fishery Management Council, 605 West 4th Ave, Suite 306, Anchorage, AK 99501. Nichol and Acuna: Annua! and batch fecundity of Limanda aspera in the eastern Bering Sea 109 Figure 1 Location of yellowfin sole (Limanda aspera ) ovary collections within the eastern Bering Sea trawl survey area, 1993. Stations where yellowfin sole ovaries were collected are circled. The diagonal line separates the northwest from the southeast sampling areas. Identifying partially spawned ovaries can be difficult with unaided macroscopic obser- vation. With microscopic observation, how- ever, identification of postovulatory follicles (POFs) within an ovary verifies that at least one batch has been spawned (Hunter and Macewitz, 1985a; Hunter et al., 1992). Histo- logical analysis of collected ovaries was em- ployed in our study in part to identify POFs. Second, annual fecundity may be overesti- mated if the eventual loss of oocytes due to atresia (oocyte resorption) is not taken into account. Oocyte atresia can occur in vary- ing degrees throughout the spawning sea- son, depending on the species and environ- mental conditions (Hunter and Macewicz, 1985b; Macewicz and Hunter, 1994; Walker et al., 1994; McFarlane and Saunders, 1997). This bias can be minimized if ovaries chosen for fecundity estimates are well developed and near spawning condition. Understanding the spawning character- istics along with other life-history param- eters may help determine how fish abun- dance varies with changing environmental or fishing conditions. Moreover, knowledge of the fecundity and the spawning charac- teristics can help define how species, such as yellowfin sole, relate to other species from a phylogenetic perspective. In our study, we evaluate the annual and batch fecundities of yellowfin sole with ref- erence to its spawning habits. Materials and methods Summary of collections A total of 767 ovary pairs were collected in June and July of 1993 during a groundfish trawl survey in the eastern Bering Sea, conducted by the Alaska Fisheries Science Center (Nichol, 1995). Collections were made at 83 sta- tions at depths shallower than 50 in (Fig. 1). Whole ova- ries were extracted from approximately 10 females per station and then preserved in 10% formalin (3.7% form- aldehyde) buffered with 19 g/liter sodium acetate-trihy- drate. Females were selectively chosen by size: 2 females 25-30 cm total length (TL), 4 females 31-35 cm TL, and 4 females >35 cm TL. All females less than 25 cm TL were immature and therefore were not collected. Each pre- served ovary lobe was weighed to the nearest milligram. Ovaries were assigned a maturity code based on a 5-point maturity classification scale for macroscopic examination (Table 1). These assignments were verified with histologi- cal evaluation. Histological evaluation of ovaries Histological cross sections were prepared from the middle portion of one ovary from each fish. Most sections (93%) were taken from the blind side; however, sections were occasionally taken from the eyed-side if fixation was not complete. Ovary tissues were embedded in paraffin, sec- tioned at 6 pm, then stained with hematoxylin and eosin. Each ovary section was examined for the following oo- cyte stages: early perinuclear, late-perinuclear, partially yolked (PY), advanced-yolk (AY), migratory-nucleus (MN), unovulated hydrated (HY), and ova (ovulated, hydrated) (Howell, 1983). We also noted the presence of postovula- tory follicles (POF), atretic oocytes, and residual chorion tissue. The maximum diameter of AY through MN oocytes were measured with an ocular micrometer. Oocyte diameters were measured from five of the largest spherical nonatretic oocytes with a centrally located nucleus. The average maxi- mum oocyte diameters were then computed for each fish. Examination of oocyte atresia Occurrence of alpha (a) stage atresia (Hunter and Mace- wicz, 1985b) among advanced vitellogenic oocytes (AY, MN, and HY) and PY oocytes was recorded for 75 females by using histological ovary cross sections. Samples included 28 maturing (maturity-code 2) females with AY-stage oocytes and no evidence of POFs, 13 fish with AY and HY oocytes and no POFs, 23 fish that had spawned at least one batch (POFs present), and 11 fish with evidence of residual chorion tissue in the ovary lumen and no other evidence of batch spawning (no HY, ova, POFs). A 2 x 2 mm grid was used to count atretic and nonatretic PY oocytes, and AY through HY oocytes. One ovary section 110 Fishery Bulletin 99(1 ) was examined for each female. Grid counts were repeated 4 to 12 times at different locations within each ovary sec- tion, until counts totaled approximately 180 oocytes. Testing for homogeneity within the ovary Prior to subsampling for total fecundity, we tested (two- way AN OVA; t-tests) for differences in oocyte density (number of oocytes per gram of ovary tissue) and mean oocyte diameter between eyed-side and blind-side ovarian lobes and among three ovary positions (anterior, middle, posterior). Twenty ovaries histologically identified with AY oocytes (maturity-code 2) and no evidence of prior batch spawnings (i.e. no POFs) were selected from fish over a broad length range. AY oocytes were defined as those with yolk filling more than half the volume of the oocyte. Tissue samples averaging 10.5 mg (SD=2.9559) and 267 oocytes (SD=90.0) were taken from anterior, middle, and poste- rior positions along the long axis of each ovarian lobe (six subsamples per fish). All AY through MN stage oocytes from each tissue sample were counted manually with the aid of a dissecting microscope. Fifty of these oocytes per tissue sample were randomly selected for oocyte area mea- surements. Oocyte areas were measured with a micro- scopic image analysis system with Optimas 5.0 software (BioScan, Inc., 1992). Black and white images were gener- ated with a video camera attached to a dissecting micro- scope (transmitted light) and were viewed on a 13-inch (diagonal) color monitor. The resolution was set to 640 x 480 pixels, corresponding to 4.149 pin/pixel at 25x magnifica- tion. Oocyte diameter was calculated from oocyte area by number of oocytes per gram of ovary tissue and unyolked oocytes in terms of percent frequency of the 50 measured. Total fecundity and hatch fecundity estimation Because of findings from the testing of homogeneity with- in ovaries (see “Results” section), two tissue subsamples were taken from the ovaries of each fish: one from the posterior third of either ovary, and one from either the middle or anterior third of either ovary. To estimate total fecundity, defined as the standing stock of AY through HY oocytes, only ovaries with no evidence of prior batch spawnings (no ova, no POFs, nor residual chorion mate- rial) were used. To eliminate less developed ovaries that contained many PY oocytes, we also limited fecundity samples to those with maximum oocyte diameters >0.35 mm as measured from histological slides. Area measure- ments of AY through MN stage oocytes with image analy- sis and counts were conducted in the same manner as the above test for homogeneity. To determine the proportional mass of each ovary sec- tion ( WFp , WFm, and WFa), one of the paired ovaries from each of the 20 fish tested for oocyte homogeneity was cut into thirds, mid-way between anterior-middle and mid-way between middle-posterior tissue positions. Each section was weighed to the nearest 0.001 g. The propor- tional mass (section wt/sum of section wts) of anterior, middle, and posterior sections averaged respectively 0.536 (SE=0.0095), 0.311 (SE=0.0086) and 0.153 (SE=0.0059) of the total ovary mass. Total fecundity was computed as Diameter = 2 ■ (1) Total fecundity ^(WF,) + ^ wt,, ' wt . ( WF POW , Oocyte size distributions Oocyte areas were measured to determine whether there was a hiatus between distributions of AY oocytes and less advanced oocytes. Areas of oocytes in partially yolked (PY ), unyolked, and AY stages were measured by using preserved tissue (as above), and calculated diameters were plotted for 75 of the 324 females examined for fecundity. These were the same females used to examine ooctye atresia, thus representing prespawning through partially spawned females, as well as those with residual chorion material present. PY-stage oocytes were considered as those with yolk that filled less than half the volume of the oocyte. Sub- samples were taken from the anterior or middle portion of one ovary and weighed to the nearest 0.001 g. All PY and AY oocytes were counted separately. For the purpose of oocyte counts and measurements, MN oocytes were not dis- tinguished from AY oocytes. No attempt was made to count unyolked oocytes or measure unyolked oocytes less than approximately 0.05 mm. Random oocyte area measure- ments by image analysis (at 25x) included 50 AY oocytes, 50 PY oocytes (if they existed), and 50 unyolked oocytes. The three oocyte diameter distributions were then plotted for each fish as PY and AY oocyte distributions in terms of where no „ = wt„ = wt = WF. = WF.. POW = number of AY-HY oocytes in tissue sample from posterior position of either ovary; number of AY-HY oocytes in tissue sample from either anterior or middle positions of either ovary; weight of tissue sample from posterior position of either ovary; weight of tissue sample (g) from either middle or anterior position of either ovary; weighting factor computed as the average proportional mass of the posterior third of either ovary; 0.153 g O?=20); weighting factor computed as the average proportional mass of anterior and middle ovary sections combined; 0.536 g + 0.311 g = 0.837 g (n=20)] and paired ovary weight (g). Batch fecundity was estimated from those ovaries con- taining HY oocytes (maturity-code 3). Subsample weights averaged 52 mg. Both HY oocytes (batch oocytes) and AY oo- cytes (remaining oocytes ) were counted. Batch fecundity was computed as above (Eq. 2), substituting nop and noam with Nichol and Acuna: Annual and batch fecundity of Limanda aspero in the eastern Bering Sea 111 Table 1 Maturity code criteria for yellowfin sole (Limanda aspera) ovaries based on macroscopic examination. Corresponding histological descriptions are included. LP = late perinucleus stage oocyte; PY = partially yolked oocyte; AY = advanced yolked oocyte; MN = oocyte with migratory nucleus; HY = unovulated hydrated oocyte. POFs = postovulatory follicles. PY is defined as an oocyte with yolk globules that filled less than half the volume of the oocyte. Maturity code Condition Macroscopic examination Histological examination 1 immature Ovary clear to slightly pink or grey-pink. No dis- tinct oocytes. Ovarian wall thin and taut around ovary interior. LP is the most advanced oocyte stage. Ovarian wall diameter thin (generally <2 pm). 2 maturing Ovary usually opaque with distinct vitellogenic oocytes. A network of veins covers the ovary. PY or AY oocytes present (yolk globules present). MN oocytes may be present. POF may be present. 3 hydrated As in code 2, but some portion of oocytes are trans- lucent (hydrated-unovulated). Hydrated oocytes are larger than the opaque oocytes and are ran- domly scattered about the ovary. HY oocytes (yolk coalesced) present, each surrounded by a follicle. AY oocytes present if prior to last batch. MN oocytes usually present if prior to last batch. POF may be present. 4 spawning Hydrated (translucent) oocytes in lumen of ovary (ovulated). A continuous band of hydrated oocytes may also be visible from the ovary sides. Eggs may run with slight pressure. If all oocytes are translucent, they represent the last batch of eggs to be spawned in the season. Ova present. AY oocytes present if prior to last batch. POFs present. MN oocytes usually present if prior to last batch. 5 spent Deflated ovary, often with blood. Ovary wall thick and often flaccid around ovary interior. LP to PY oocytes present. Ovarian wall diameter thick (generally >3 pm). numbers of HY oocytes in the respective subsample. Ovaries were categorized by stages in the batch spawning succes- sion: a first batch was recognized by the presence of HY and AY oocytes, and absence of POFs. A middle batch was recog- nized by the presence of HY oocytes, AY oocytes, and POFs. A final batch was recognized by the presence of HY oocytes and the absence of remaining AY or MN oocytes. Fecundity-total length relationships were computed for total and batch fecundity by using Gauss-Newton nonlinear least squares regression (SAS Institute, 1989). The num- ber of batches spawned from a female for a given fish length was then estimated as the estimated total fecun- dity divided by the estimated batch fecundity. Results Histological evaluation Histological evaluation of 767 ovary pairs, in most cases (93%), verified the general ovary codes that were assigned macroscopically (Table 1). Resulting groupings are shown in Table 2. We note here that maturing (maturity-code 2) ovaries included females that had spawned one or more batches (POF present) but contained no hydrated oocytes (HY) or ova; these ovaries were indistinguishable macroscopically from advancing ovaries that had not yet spawned a batch. Early perinucleus stage oocytes were present in all ovaries examined. Late perinucleus stage oocytes were present in all but one of the mature fish ova- ries (??=665; maturity-codes 2-5). Migratory nucleus stage (MN) oocytes, vitellogenic oo- cytes that are precursors to hydrated oocytes, were pres- ent in 96% of the females undergoing oocyte hydration and 91% of spawning females. The high occurrence of MN oo- cytes among fish about to spawn a batch suggests that fish do not wait extended periods between batch spawnings. Postovulatory follicles were observed in 121 of the ova- ries examined. POFs were present in 50 of 75 females un- dergoing hydration, indicating for these fish that at least one batch was previously spawned. For the other 25 fe- males with hydrated ovaries containing no POFs, HY oo- cytes represented the first batch of a series of batches. We could not distinguish more than one apparent age of POF among the 50 fish that had begun spawning. This indicates that yellowfin sole either resorb POFs before a succeeding batch is ovulated or that POFs of different ages (from dif- ferent batches) are present but are indistinguishable. Evidence of more than one series of batdi spawnings A total of 16 females with maturing ovaries (maturity- code 2) contained residual chorion tissue within the ovary lumen (Table 2; Fig. 2). The chorion tissue in these cases was sufficiently large to conclude that they were unspawned ova from a previous spawning. The ovaries ranged from those in early stages of yolk accumulation 112 Fishery Bulletin 99(1) Advanced Yolked Residual Chorion Oocytes . Tissue 0.3 mm Figure 2 Histological section of a yellowfin sole (Limanda aspera ) ovary with advanc- ing yolked oocytes and remnants of chorion tissue leftover from previous spawning. The 38-cm-TL female was captured 17 June 1993 at a bottom depth of 27 m. The yolked oocytes averaged 0.31 mm in diameter, well below the mean diameter of yolked oocytes from spawning individuals (>0.44 mm). Table 2 Occurrence of oocyte stages, postovulatory follicles, and other postspawning remnants within immature, maturing, hydrated, spawning, and spent yellowfin sole (Limanda aspera ) ovaries. Histological examination of ovary cross sections. Oocyte stage or intra-ovarian structure Ovary maturity code Total fish Immature (1) Maturing (2) Hydrated (3) Spawning (4) Spent (5) Early perinucleus 102 415 75 23 152 Late perinucleus 58 415 75 23 151 Partially yolked 8 349 17 7 82 Advanced-yolked 0 409 72 23 0 Migratory nucleus 0 63 72 21 0 Hydrated-unovulated (HY) 0 0 75 3 0 Ova 0 0 0 23 0 Postovulatory follicles 0 13 50 19 40 Atretic-HY 0 0 0 0 21 Atretic ova 0 7 9 1 52 Residual chorion tissue 0 15 1 0 29 Number of fish 102 415 75 23 152 767 (mean AY diameters=0.27 mm), to those that contained AY oocytes near maximum size (mean AY diameter=0.43 mm). These ovaries did not contain POFs or MN stage oocytes; therefore ovaries were not advanced enough for there to be recent spawning from the existing AY standing stock. It is possible that these fish had already spawned at least one series of batches, were spent, and were readying Nichol and Acuna: Annual and batch fecundity of Limanda aspero in the eastern Bering Sea 113 Table 3 Two-way analysis of variance results comparing differences of oocyte density (top), and mean oocyte diameter (bottom) between blind and eyed-side ovaries and among anterior, middle, and posterior tissue subsampling positions in yellowfm sole, Limanda aspera. Paired r-tests ( H0=dpost-dant', H0=dpnsl-dmid) among ovary positions indicate significantly (P<0.05) greater oocyte densities (d) in posterior tissue samples compared with either anterior or middle positions. SS = sum of squares; MSE = mean square error. Source df SS MSE F P-value Oocyte density (no. oocytes/g ovary tissue) Fish 19 3,391,989,272 178,525,751 62.32 0.0001 Ovary 1 3,466,212 3,466,212 1.21 0.2708 Position 2 25,951,981 12,951,981 4.53 0.0125 Ovary x position 2 9,505,128 4,752,564 1.68 0.1914 Error 95 268,371,656 2,824,965 Corrected total 119 3,699,284,249 Mean oocyte diameter Fish 19 5.7136 0.3007 183.25 0.0001 Ovary 1 0.0003 0.0003 0.19 0.6655 Position 2 0.0034 0.0017 1.02 0.3600 Ovary x position 2 0.0022 0.0011 0.68 0.5044 Error 5975 9.8053 0.0016 Corrected total 59997 15.5248 1 n = 6000; 2-ovary x 3-ovary positions x 20-fish x 50-oocytes/fish. another standing stock of oocytes for spawning within the same reproductive season. Homegeneity of oocytes between and within ovaries Two-way ANOVA on mean oocyte diameters indicated that the means were not significantly different between ova- ries or among the anterior, middle, and posterior ovary positions (Table 3). Additionally, our analysis revealed that oocytes were slightly more concentrated (number of oocytes/g ovary tissue) in posterior tissue samples com- pared with middle and anterior samples (paired £-tests, Ho- dpoSt-^ant=Q and Ho- dpost-^nud=0’ n= 20> ^-values <0.05; Table 3 ). Comparison of oocyte density between middle and anterior tissue samples and between eyed-side and blind- side ovary lobes revealed no significant differences ( paired £-test; P-value >0.05; Table 3). These results prompted us to take fecundity subsamples from two ovary locations (posterior and middle-anterior position from either ovary lobe) and to incorporate weighting factors (WF\ Eq. 2) for all calculations of total fecundity. The evidence for greater posterior-end oocyte densities that was found in 20 test fish prior to fecundity subsam- pling, was also present after all ovaries were subsam- pled. Oocyte densities were significantly greater in the posterior ovary end versus the anterior-middle positions among all ovaries sampled for fecundity (paired t-test, H0 : dpost-dant/mid= 0; £=2.59; n= 323; P=0.0100). The densi- ty of hydrated (HY) oocytes within maturity-code-3 ova- ries, used for batch fecundity estimates, was also greater in subsamples taken from the posterior ovary end (paired '-test, H0. dpost-dant/mid= 0; £=3.40; n= 75; P=0.0011). Com- putation of batch fecundity, therefore, incorporated the same weighting factors described for computation of total fecundity. Oocyte size distributions Oocyte diameter plots indicated a hiatus in distribution between AY oocytes and PY oocytes for ovaries with mean AY diameters >0.38 mm (Fig. 3). This separation of AY oocytes from other oocyte stages in more advanced ovaries (closer to spawning) is indicative of a determinate mode of oocyte development. In prespawning fish (maturity-code 2, no POFs), oocyte size distributions overlapped for fish with mean AY diameters less than 0.38 mm. PY oocytes, however, were a relatively minor portion in terms of oocyte density of the overall distribution for ovaries with mean AY diameters greater than 0.38 mm (Fig. 3). Oocyte distri- butions in maturing (maturity-code 2) fish with residual chorion tissue present in the ovary lumen exhibited the same hiatus between PY and AY oocytes where mean AY diameters were >0.38 mm (Fig. 3). The size distribution of AY oocytes for those fish that were either spawning their first batch (maturity-code 4, no POFs; n = 13) or had begun spawning (maturity-code 2 or 3, POFs present; n= 23) was separated from PY and non- yolked oocyte distributions in 33 of the 36 fish examined. PY oocytes were either absent or were a very small frac- tion of the number of AY oocytes present. From the histo- logical analysis, we found that 68 of the 86 fish that had begun spawning (maturity-codes 2 and 3 with POFs, and maturity-code 4) did not contain any PY oocytes, further indicating that only one stock of oocytes is advanced prior to the spent ovary condition. 114 Fishery Bulletin 99(1 ) Mean oocyte diameters of yolked oocytes (AY through MN) were generally larger among fish that had begun spawning compared with those that had not released a batch (Fig. 4). Ovaries that contained POFs had mean yolked oocyte diameters that on average were greater than 0.44 mm. Yolked oocytes in advancing ovaries with no evi- dence of recent spawning (no POFs) averaged less than 0.44 mm in diameter. Again, most of the ovaries with re- sidual chorion material were significantly less developed than those with evidence of recent spawning (Fig. 4). Occurrence of oocyte atresia Of the 75 females examined histologically for the occur- rence of atresia, only 15 (20%) had ovaries containing a-atretic AY, MN, or HY oocytes (Table 4). Within each of these 15 females, the average proportion of a-atretic AY-HY oocytes among all AY-HY oocytes was 0.034. Thus, among all females examined including those with no atresia (n= 75), the relative frequency of a-atretic AY-HY oocytes among normal AY-HY oocytes was less than 1%. The occurrence of atretic AY-HY oocytes as well as atretic PY oocytes was higher among females that were either close to spawning or partially spawned compared with those in a prespawning condition (Table 4). As shown with the oocyte distribution plots, normal PY oocytes were more common among females that were in a prespawning condition. Lower frequencies of normal PY oocytes, but higher proportions of atretic PY oocytes among females that were partially spawned or close to spawning, indi- cated that many PY oocytes do not develop after spawning is initiated. We observed no cases of major atresia (>50% atretic) among advancing oocytes for all ovaries examined histologically. Given the low occurrence of oocyte atresia, we assumed a negligible effect of atresia on estimates of annual fecundity. Nichol and Acuna: Annual and batch fecundity of Limanda aspera in the eastern Bering Sea 115 n= 11 Code 2, Residual - I • — chorion material present Code 2-3, POF - (jj (partially spawned) a. & jk Code 3, no POF - ro (HYs = first batch) O Code 2, MNs, no POF - n = 46 n = 24 I • 1 n = 45 I — • — I Code 2, no MNs, no POF - n = 99 ' 1 > 1 — ' 1 ■ 1 > 1 ■ 1 ■ 1 0 32 0 34 0 36 0 38 0 40 0 42 0 44 0 46 Mean oocyte diameter (mm) Figure 4 Comparison of mean yolked oocyte diameters among ovaries at different levels of devel- opment, from maturing (maturity-code 2) ovaries with no migratory nucleus stage (MN) oocytes (>0.38 mm) to maturity-code 3 ovaries with postovulatory follicles (POFs) pres- ent. Error bars indicate 95% confidence intervals. See Table 1 for description of ovary maturity codes. Table 4 Occurrence of alpha (a) atretic oocytes among advanced vitellogenic (AY through HY) and partially-yolked (PY) oocytes within yel- lowfin sole ( Limanda aspera ) ovaries (n=75 females). AY = advanced-yolked; HY = unovulated hydrated. Maturity code No. of females sampled No. of females with a-atretic AY- HY oocytes Proportion of a-atretic among all AY-HY (min. -max.) No. of females with PY oocytes No. of females with a-atretic PY oocytes Code 2, no POFs (prespawning) 28 2 0.01-0.025 24 3 Code 3, no POFs (first batch) 13 5 0.004-0.058 4 3 Codes 2 and 3, POFs present (partially spawned) 23 6 0.005-0.037 6 3 Code 2, residual chorion tissue present 11 2 0.061-0.134 9 2 Total fecundity In choosing ovaries to be used to estimate total fecundity, we selected ovaries that were advanced enough that the advancing stock (AY to MN) could be discriminated from less advanced oocytes (PY), yet were not advanced enough for there to have been potential spawning. First, we elimi- nated ovaries with mean oocyte diameters <0.38 mm due to the overlap in AY and PY distributions. Secondly, we elimi- nated ovaries with mean oocyte diameters greater than 0.44 mm (n= 20) in an attempt to eliminate partially spawned ovaries that could not be identified with histological eval- uation; POFs were not found in these ovaries; however, because the level of oocyte advancement (mean AY diam- eter) was similar to ovaries that did contain POFs (Fig. 4), the occurrence of recent batch spawning was possible. Inclusion of ovaries with mean AY-oocyte diameters <0.38 mm as well as those >0.44 nun can potentially bias estimates of total fecundity. Estimates of total fecundity, measured from fish where distributions of PY and AY oocytes overlapped (Fig. 3), were higher than estimates from ovaries with more advanced oocytes (Fig. 5; Table 5). We compared yellowfin sole length-fecundity relation- ships of fish with ovaries that had mean AY-oocyte diam- eters >0.38 mm and <0.44 (n = 148) to those with mean diameters <0.38 mm (n=80). Linear comparisons of the \og(Iength)-\og(fecundity ) relationships between the two data sets indicated similar slopes (F=0.08; P= 0.7725; df=l, 116 Fishery Bulletin 99(1) Table 5 Yellowfin sole ( Limanda aspera ) length-fecundity coefficients for nonlinear least-squares fit using equation F = aLb, where F = fecundity and L = fish length (cm). MOD = mean oocyte diameter of advanced yolked oocytes. SE = standard error of estimate. Confidence intervals (Cl) indicate approximate 95% bounds for predicted values of the mean (SAS Institute, 1989). Fecundity type a Constants b n r2 Fecundity estimate for 35-cm female (Value ±95% Cl) Estimate SE Estimate SE Total MOD > 0.38-0.44 mm7 3.322 2.909 3.631 0.243 148 0.60 1,343,807 ±64,406 MOD < 0.38 mm 4.947 5.564 3.550 0.311 80 0.64 1,496,401 ±105,108 All data (MOD <0.44 mm) 2.988 2.074 3.672 0.192 248 0.62 1,397,492 ±56,054 Batch 1.648 3.138 3.188 0.528 75 0.32 137,862 ±17,111 1 Ovaries with mean AY through MN oocyte diameters >0.38 mm and <0.44 mm were considered the most appropriate for use in estimating total fecundity. 4 5 4 0 - -L— MOD 0 38 - 0 44 mm -X-- MOD < 0 38 mm 3 5- O MOD > 0 44 mm 0.0 -I 1 1 ' r— — ' p » 1 • 1 ' 1 1 > 1 > 1 < r ' 1 24 26 28 30 32 34 36 38 40 42 44 46 Total length (cm) Figure 5 Total fecundity as a function total length (cm). Data for yellowfin sole (Limanda aspera) ovaries with mean AY diameters (MOD) 0.38 mm to 0.44 mm (n=148) were compared with ovaries with MOD <0.38 mm (n=80), ovaries with MOD >0.44 mm (n=20), and ovaries with residual chorion material present (n=7 ). Yellowfin sole ovaries with MOD 0.38 mm to 0.44 mm were considered most appropriate for use in total fecundity. Curves indicate predicated values from nonlinear regression. MOD = mean oocyte diameter of AY through MN oocytes. 221) but different intercepts (F=6.77; P=0.01; df =1, 222). We also tested whether the inclu- sion of ovaries with mean oocyte diameters >0.44 mm would have a significant effect on the fish-length-fecundity relationship (Table 6). When ovaries with mean oocyte diameters >0.44 mm were included, estimates of total fe- cundity at length were significantly reduced. No such effect was apparent when the data were limited to ovaries with mean oocyte di- ameters <0.44 mm (Table 6). Total fecundity as a function of fish length for yellowfin sole was estimated as F = 3.3225 x TL3 6312, where F = standing stock of AYs; and TL - total fish length in centimeters (Table 5). Total fecundity ranged from 295,615 oocytes in a 29-cm-TL fish to 3,635,108 oocytes in a 39-cm-TL fish (Fig. 6). Batch fecundity Batch fecundity was estimated for 75 females and ranged from 2,400 to 408,000 oocytes and larger fish generally had larger batches (Figs. 6 and 7). The number of oocytes in the first batch was in most cases lower than in succeed- ing batches, although the variability of the first batch for a given fish length was quite high (Fig. 7). Twenty-five of the fish examined contained ovaries with unovulated hydrated (HY) oocytes and no evidence of prior batch spawnings, thus representing the first batch. Three fish contained ovaries with only one hydrated batch remaining (no AYs left), thus representing the last batch. The remain- ing “middle batch” fish (n= 47) possessed ovaries contain- ing unovulated hydrated oocytes and POFs, indicating that at least one batch had been previously spawned. Batch fecundity, irrespective of batch order, is described in terms of fish length by B = 1.6481 x TL3 1879, Nichol and Acuna: Annual and batch fecundity of Limanda aspera in the eastern Bering Sea 117 Table 6 Linear regression results from the model logtF) = log(a) + b\og(L) + log(c); the log-transformation of model F = aLhc, where F = total fecundity, L = fish length (cm), and a, b, c are coefficients. Coefficient c represents the interaction of mean oocyte diameter on F. A series of regressions were run, first by using all data where the mean yolked oocyte diameter was >0.38, then with data sets that excluded ovaries with high mean oocyte diameters (i.e. ovaries with mean oocyte diameters >0.46 mm, >0.45 mm, >0.44 mm). Mean oocyte diameter range of ovaries used in analysis Estimate P-value of H0: log(c) = 0 Log(a) b Log(c) n > 0.38 mm 6.40 3.43 -1.22 0.003 168 0.38-0.46 mm 6.35 3.44 -1.22 0.006 165 0.38-0.45 mm 5.84 3.42 -1.06 0.026 160 0.38-0.44 mm7 3.07 3.44 -0.33 0.525 148 1 This data set, which excluded ovaries with mean oocyte diameters > 0.44 mm, indicated no significant effect of mean oocyte diameter on the estimate of total fecundity. where B = batch fecundity; and TL = total fish length in centimeters (Table 5). Number of batches The number of batches that a female of a given length spawns was estimated as F/B (above). Yellowfin sole are estimated to spawn an aver- age of 8 to 11 batches prior to the spent ovary condition. The frequency of batches appears to be slightly higher for larger fish. Discussion Determinate or indeterminate fecundity? Fish with determinate fecundity are defined as those with ovaries whose advancing stock of oocytes represent the entire number of eggs to be spawned that year (Hunter et ah, 1989; Hor- wood and Walker, 1990; Hunter et al., 1992). Fish with indeterminate fecundity, in contrast, are defined as those whose ovaries continu- ously mature yolked oocytes from unyolked oocytes; thus the “standing stock” of advanced oocytes does not represent the total number of eggs to be spawned that year (Hunter et ah, 1989). Oocyte size frequencies of various stage oocytes (i.e. advanced-yolked, partially yolked, unyolked) have been used as criteria to determine if a fish has determinate fecundity (Hunter et ah, 1989; Hunter et ah, 1992; Horwood and Walker, 1990). If a break or “hiatus” has occurred between the advanced-yolked distribution and the less advanced ones, then the advanced-yolked distribution could be considered the determinate stock. This break in oocyte distributions was observed for yellowfin sole, and it is clear that batches are spawned from the advanced-yolked mode. A deter- minate mode of oocyte development in yellowfin sole is further indicated by the fact that total fecundity was sub- stantially lower in fish that had spawned at least one batch (Fig. 8) — an indication that there is no oocyte recruit- ment to the advanced oocyte stock once spawning begins. Hence, although yellowfin sole spawns eggs intermittently in batches, they undergo a group synchronous mode of oocyte development. In contrast to other fish with determinate fecundity, yel- lowfin sole fecundity estimates were higher among indi- viduals with less-developed ovaries where PY and AY oo- cyte distributions overlapped (Fig. 5). Hunter et al. ( 1992) indicated the opposite for Dover sole; they found that fe- cundity among individuals where the advanced-stock size 118 Fishery Bulletin 99(1 ) Figure 8 Comparison of mean total fecundity (no POF and mean oocyte diam- eter >0.38 mm) at length and the mean fecundity observed in yellowfin sole ( Limanda aspera) that had already spawned at least one batch (POFs present). distribution was not well separated from less developed oocytes was lower because oocyte recruitment to the advanced stock was not complete. We suggest two reasons for this dif- ference between the two species. First, the discrimination between PY and AY oocytes and the criteria used to separate them were likely different for the two studies. Exact dis- crimination between oocytes that will not de- velop and those that are viable cannot occur until the distributions are completely sepa- rate. Second, rates of oocyte atresia between the two species may be different. Higher rates of PY and AY atresia may account for a de- crease in fecundity as the AY distribution ad- vances. As Hunter et al. (1992) noted, the fate of PY oocytes is uncertain; perhaps some in- dividuals retain them for an additional series of batch spawnings. Although multiple groups of oocytes are not continuously developed as are those observed for indeterminate fishes, such as the northern anchovy, Engraulis mordax (Hunter and Le- ong, 1981), yellowfin sole appear to have the potential to recover spent ovaries and spawn another series of batches. Remnants of chori- ons from hydrated oocytes in the lumen of ova- ries containing a unimodal stock of advancing yolked oocytes with diameters significantly less than those of spawning fish (Fig. 4) suggest that at least some individuals can produce more than one series of batches in a single year. Annual fe- cundity for each of these individuals would be significantly greater than the estimated total fe- cundity. Although yellowfin sole exhibit a deter- minate group-synchronous mode of oocyte devel- opment, some individuals may spawn more than one series of batches; therefore, annual fecundity must be considered indeterminate. Comparison of total fecundity between southeast and northwest areas Total fecundity for yellowfin sole was com- pared for the northwest and southeast areas of the eastern Bering Sea. Although Nichol (1997) found that yellowfin sole growth and size at maturity were greater in the northeast area of the eastern Bering Sea compared with the southeast area (Fig. 1), we found no dif- ferences in fecundity between areas. Neither slopes (ANCOVA: F=0.09; P=0.77; df =1, 144) nor intercepts (ANCOVA: F=0.56; P>0.10; df = 1, 145) were different for the log-transformed (linear) length-fecundity relationships. Comparison of results with those from other authors With the exception of fecundity data presented by Fadeev (1970), the relationship between yellowfin sole fecundity and fish length appears similar between east and west sides of the Bering Sea and among years (Fig. 9). Ivankov and Ivankova (1974), who reported that yellowfin sole in the northwestern Sea of Japan spawn up to five batches, presented slightly lower values of total fecundity com- pared with our results. Tikhonov (1977) presented sim- ilar length-fecundity relationships for yellowfin sole off Nichol and Acuna: Annual and batch fecundity of Limonda aspera in the eastern Bering Sea 119 Figure 9 Comparison of predicted values of total fecundity as a function of fish length for yellowfin sole (Limanda aspera ), as presented by various authors. Error bars presented for the present study indicate 95% confidence intervals for the mean. the west coast of Kamchatka for five dif- ferent years, 1963—69. Because methods (i.e. selection criteria for ovaries) likely varied among studies, critical interpre- tation of these comparisons is difficult. Comparison of yellowfin sole with other species Yellowfin sole is more fecund and spawns smaller eggs than other flatfish species in the eastern Bering Sea (Table 7). Most flatfishes in the eastern Bering Sea, with the exception of northern rock sole ( Lep - idopsetta polyxystra, Orr and Matarese, 2000), spawn pelagic eggs intermittently (in batches). Longhead dabs, also within the genus Limanda ( proboscidea ), have similar reproductive characteristics to yel- lowfin sole in respect to their high fecun- dity at length, small diameter of eggs, shallow spawning location, and spring- summer spawning season (Table 7). Yellowfin sole are most closely related to yellowtail flounder, Limanda ferrun- ginea (Cooper and Chapleau, 1998), a western Atlantic species ranging from the Gulf of St. Lawrence to Chesapeake Bay (Howell, 1983; Zamarro, 1992). Al- though only morphological characters were used to determine the phylogenetic relationships of these two species among pleuronectids (Cooper and Chapleau, 1998), yellowfin sole and yellowtail flounder also share very similar life history characteristics. Fecundity ranges, egg diameter, egg type, spawning interval, and spawning seasons are very simi- lar between the two species (Table 7). Like yellowfin sole females, yellowtail flounder females develop oocytes in a group-synchronous manner (Howell, 1983). In addition, they have MN oocytes present in their ovaries that con- tain hydrated oocytes, indicating that spawning of batches is similarly continuous (Zamarro, 1992). Yellowtail floun- der spawn approximately 7 batches compared with 8-11 for yellowfin sole in the eastern Bering Sea. In short, close- ly related species share more than just morphological sim- ilarities; reproductive characteristics also reflect a shared evolutionary history. Spawning habits Yellowfin sole in the eastern Bering Sea have been ob- served from prespawning to spawning conditions from mid-May through August (Fadeev, 1970; Nichol, 1995). Observations of eggs and early-stage larvae in ichthyo- plankton surveys conducted in the eastern Bering Sea (Musienko, 1963; Waldron, 1981) indicate that spawning may not completely end until September. Given that the spawning season is protracted and that spawning of a series of batches is fairly rapid, individuals may have the potential to recover spent ovaries and spawn more than one series of batches within a single year. The possibility that residual chorion tissue is left over from the previous year’s spawning seems unlikely given that it would have been retained within the ovary for more than 8 months (Sep-May). Yellowfin sole remain in the spawning area ( <30 m bot- tom depth) until a series of batches have been spawned. The absence of partially spawned fish, those with ovaries containing POFs and AY oocytes, found outside the spawn- ing area ( >30 m) indicate that after spawning begins (first batch) fish remain in the spawning area until spent. Again, the presence of MN oocytes in most ovaries under- going oocyte hydration indicates that there is very little lag period between batch spawnings. This evidence refutes an earlier assertion by Nichol (1995) that yellowfin sole may migrate in and out of the spawning area between batch spawnings. However, some females do migrate out of the spawning area after a series of batches has been completely spawned; both spent and maturing (maturity- code 2) females with evidence of completed batch spawn- ings (i.e. with residual chorion tissue) were observed in spawning ( <30 m) and nonspawning ( >30 m) waters. Acknowledgments Claire Armistead, Jan Haaga, Mark Conrad, and Gary Walters aided with tissue collections. Frank Morado, Lisa Table 7 Fecundity and spawning habits of nine species of flatfishes found in the eastern Bering Sea and yellowtail flounder (Limanda ferruginea) from the Atlantic coast of North America. Spawning interval refers to whether eggs are spawned intermittently (batch) or all at once (synchronous). Numbers in parentheses are the range of fish lengths used for the respective fecundity estimates. All estimates are for the eastern Bering Sea except where noted. 120 Fishery Bulletin 99(1) T3 £ too _ P £ > o o3 Si a — tic ^ P 03 g i > a) p c a .5 m bD bo W P ■ s - p "3 r b£ be W n d p p a co V PH 10 "a CJ 03 o3 03 0) P CO 1 P •“C § § Q 1 < I tA 03 § 1 A Ph C A > ^0 o3 Xi O LO co CD co" LO 03 CM I LO 03 CM I O CD CO I oo o '6b co o o ^ £ 6 CM cj p ? i o o o LO I o o I o 03 X3 03 -O P >4 '5b 03 C0 I lO o I CM co o 00 § s co" LO co o o © CM CM cm" So l o o o cd" CM O o ^ s s O CJ 2 oa \ 6 I o o CO rH LO I CM CM 4 CO I o 03 o o o" s © ° o o o CM 00 © p p I >4 03 I 00 oo I o o o o o CM o' S © u ■"f IM i 7 .A n, CO a C3 C Q C3 o3 o a c o3 P C§ J 3 ^ xl P * m i >4 03 03 Xi °6 ^6 '03 °b O Ph CJ *4 p C/3 % O 5jd 5b Ph O ’5b ’5b 5b P 6b 6b iS iS C/3 <13 r2 rP iS a J2 'p Ph 03 Ph 6 03 Ph o3 P3 0 Ph p Ph p X3 0 Ph 'p Ph O I LO CD LO I .. co o CO o w o o ^ °. a LO O 7 ok O CM O w lO co" CO CO I £ CQ e c < S -2 ai <33 AC 03 00 X, § 1 1 A § -a <2 « l c 2 2 ta 2; j= « s 2 £ 09 AT JA rH C 00 to 2 CQ — -a .b s 1% u £ P a; .o Is O £ 6 § i ~ cox? to o3 H C CO . o __ AC co ^3 u X C o £ 0J cj J 2 co c a> c 2 C > 0 S3 2 s o g .03 03 N £ ^ CS) 0.05) was observed for any Sakhalin drainage, and only a few loci suggested heterogeneity (Table 2), which vanishes when corrections Noll et al : Analysis of genetic structure of Asian and western Oncorhynchus gobuscha 12 7 Table 1 (continued) Enzyme Enzyme number Locus Tissue7 Buffer2 Level of variability0 Lactate dehydrogenase 1.1.1.27 LDH-A1*45 M 1 3 LDH-A2*4-5 M 1 1 LDH-B1*4 H 1 2 LDH-B2* L 1 2 LDH-C*45 E 7 1 Malate dehydrogenase 1.1.1.37 sMDH-Al,2* L 6 3 sMDH-Bl,2*4’5 M 2,3 3 Malic enzyme 1.1.1.40 mMEP-1*4’5 M 3,4 3 Mannose-6-phosphate isomerase 5.3.1.8 MPI*4-5 H 2 2 Peptidase: Cytosol non-specific dipeptidase (glycyl-leucine) 3.4.-.- PEP A*4 M 2 1 Tripeptide aminopeptidase (zinc enzyme) (leucyl-glycyl-glycine) 3.4.-.- PEPB *4 M 1 3 Leucyl-tyrosine peptidase 3.4.-.- PEP-LT* M 2,4 3 Phosphoglucomutase 5. 4.2. 2 PGM-2*4-5 M 3 2 Phogluconate dehydrogenase 1.1.1.44 PGDH*45 E 3 3 Phosphoglycerate kinase 2. 7. 2. 3 PGK-1* L 8 1 PGK-2* L 8 1 X-proline dipeptidase 3.4.13.9 PEPD-1*45 M 2 2 PEPD-2*4’5 M 2 3 Superoxide dismutase 1.15.1.1 sSOD- 1 *4-5 L 7 1 mSOD* H 1,2 1 Triose-phosphate isomerase 5.3.1. 1 TPI-P" E 7 1 TPI-2*4 E 7 2 TPI-3*4 E 7 1 TPl-4 *4 E 7 2 1 E = eye; H = heart; L = liver; M = muscle; preferred tissue listed first. 2 1 = lithium hydroxide (Ridgway et al., 1970); 2 = Tris-EDTA-borate (Boyer et al., 1963); 3 = amine citrate, pH 6.1 (Clayton and Tretiak, 1972); 4 = Tris-citrate, pH 7.0 (Shaw and Prasad, 1970); 5 = Tris-citrate discontinuous (Schaal and Anderson, 1974); 6 = amine-citrate-EDTA, pH 7.2 (modified from Clayton and Tretiak, 1972); 7 = Tris-glycine (Holmes and Masters, 1970); 8 = amine-citrate, pH 6.8 (modified from Clayton and Tretiak, 1972). 3 1 = monomorphic; 2 = low (frequency of most prevalent allele >0.95); 3 = high (frequency of most prevalent allele <0.95). 4 Loci included in 36-locus version of neighbor-joining tree and gene diversity analysis. 5 Loci included in 21-locus version of neighbor-joining tree. are made for multiple testing. This is especially notable for the series of collections from the Znamenka (9 collec- tions, P=0.43) and Ochepukha (6 collections, P=0.52) riv- ers. No heterogeneity was detected among Sakhalin drain- ages (G= 131.01, 127 df; P=0.39). The three Japanese (Hokkaido) collection sites were the Tokushibetsu River on the Sea of Okhotsk coast, the Kushiro River facing the northwestern Pacific Ocean, and the Yurappu River southwest of the Kushiro River on Hok- kaido’s eastern coast. The synthetic Yurappu stock is de- rived in large part from Okhotsk coast stocks, including the Tokushibetsu stock. The genetic profile of the Yurap- pu stock resembles that of the Tokushibetsu stock except at sAAT-3* (P=0.018, analysis not shown), which suggests a founder effect or subsequent divergence. The Yurappu sample was not used in subsequent analyses. Overall het- erogeneity was observed between the Kushiro and Tokush- ibetsu Rivers (G=48.87, 33 df;P=0.037) (Table 2), although tests at the four loci suggest heterogeneity was not signifi- cant after correction for multiple testing. 128 Fishery Bulletin 99(1 ) Table 2 Hierarchical homogeneity tests. Levels are among collections within streams, among streams within drainages, among drainages within a region, among Asian regions, and between Asia and Prince William Sound, Alaska. Degrees of freedom of tests at lower levels of hierarchy may differ from (be lower than) those necessary for the overall hierarchical analysis. — indicates that no test was possible because there were data for only one collection. The hierarchical nature of the analysis precludes multiple testing corrections in the table because several different hypotheses are possible. Collection site sAAT-4* sAH* GDA* FDHG* sAAT-3* CK-C1 * PGDH* TPI-2* G df G df G df G df G df G df G df G df Asia Hokkaido 0.96 1 1.46 1 0.66 4 1.8 2 0.48 1 1.36 1 8.79“ 3 0 1 Sakhalin Dolinka — — — - — — 0.71 3 — Lutoga — 1.81 1 3.55 4 0.82 2 0.03 1 — 7.49 9 0 1 Monetka — 1.64 1 5.45 4 0.01 2 0.01 1 0.02 1 6.89“ 3 0 1 Ochepukha Other — — — — — — 7.01 15 — Znamenka — 0 1 3.14 4 0.09 2 0.002 1 — 25.31 24 0 1 Within Ochepukha — 0 1 3.14 4 0.09 2 0.002 1 — 32.32 39 0 1 Between Ochepukha — 4.67“ 1 1.01 4 4.74“ 2 0.001 1 0.70 1 4.42 3 0 1 Total Ochepukha — 4.67 2 4.15 8 4.82 4 0.003 2 0.70 1 36.74 42 0 2 Within Sakhalin — 8.12 4 13.15 16 5.65 8 0.03 4 0.72 2 51.83 57 0 4 Among Sakhalin — 1.21 3 10.84 12 7.33 6 0.96 3 7.53 3 6.40 9 0 3 Total Sakhalin — 9.34 7 23.99 28 12.98 14 1 7 8.25 5 58.22 66 0 7 Northern Sea of Okhotsk — — — 0.20 2 — — 3.29 3 0 1 Western Kamchatka — 0 2 10.07 8 8.03 8 10.61“ 4 1.08 2 11.67 12 0 4 Eastern Kamchatka — 0 1 6.28 4 0.02 2 1.31 1 0.14 1 3.19 3 0 1 Within Asia 0.96 1 10.80 11 41.00 44 23.04 28 13.40 13 10.84 9 85.16 87 0 14 Among Asia 6.646 1 3.04 3 39.75d 12 19.89“ 8 56.59d 3 3.00 3 65.95d 12 0 4 Total Asia 7.60“ 2 13.84 14 80.75“ 56 42.93 36 69.99d 16 13.84 12 151. lld 99 0 18 North America Prince William Sound 7.8 4 2.45 4 10.02 16 3.20 8 3.96 4 4.90 4 18.58“ 12 3.38 4 Total within 15.40“ 6 16.29 18 90.77 72 46.13 44 73.95d 20 18.74 16 169.69d 111 3.38 22 Between Alaska and Asia 0.95 1 0 1 107.53d 4 23.51d 2 269. 38d 1 8.836 1 43.12d 3 39.30d 1 Total for collections 16.34“ 7 16.29 19 198. 30d 76 69.64“ 46 343.33d 21 27.57 17 212. 81d 114 42.68h 23 TPI-4* rriAH-4* G3PDH-1 * GPI-B2* GPI-A* LDH-A1* LDH-B1* Collection site G df G df G df G df G df G df G df Asia Hokkaido 5.30“ 1 0 2 0.14 3 1.33 3 2.9 2 0 3 1.45 2 Sakhalin Dolinka — — 0.02 3 — — — — Lutoga 0.07 1 2.76 2 1.69 9 0 3 1.01 2 1.05 3 1.05 2 Monetka 1.39 1 1.73 2 0.47 3 1.45 3 0 2 0 3 0 2 Ochepukha Other — — 3.16 15 — — — — Znamenka 1.39 1 0.76 2 12.71 24 0 3 1.39 2 1.43 3 0 2 Within Ochepukha 1.39 1 0.76 2 15.88 39 0 3 1.39 2 1.43 3 0 2 Between Ochepukha 0.81 1 0.20 2 0.01 3 0 3 0.83 2 0.22 3 2.20 2 Total Ochepukha 2.20 2 0.97 4 15.89 42 0 6 2.22 4 1.65 6 2.20 4 Within Sakhalin 3.66 4 5.45 8 18.06 57 1.45 12 3.23 8 2.70 12 3.25 8 Among Sakhalin 1.41 3 5.41 6 4.54 9 3.93 9 2.22 6 6.76 9 2.14 6 Total Sakhalin 5.06 7 10.86 14 22.60 66 5.38 21 5.45 14 9.47 21 5.39 14 Northern Sea of Okhotsk 0.19 1 0.64 2 0.33 3 0.19 3 0.19 2 — 0 2 continued Noll et al.: Analysis of genetic structure of Asian and western Oncorhynchus gobuscha 129 Table 2 (continued) TPI-4* mAH-4* G3PDH-1* GP1-B2* GPI-A* LDH-A1* LDH-B1* Collection site G df G df G df G df G df G df G df Western Kamchatka 8.87 4 7.85 8 14.91 12 13.24 12 6.60 8 2.99 12 2.07 8 Eastern Kamchatka 1.67 1 2.53 2 0.003 3 1.58 3 0 2 2.47 3 0 2 Within Asia 21.10 14 21.88 28 37.98 87 21.72 42 15.14 28 14.92 39 8.91 28 Among Asia 17.06'’ 4 14.21 8 10.69 12 13.85 12 9.58 8 7.59 12 1.60 8 Total Asia 38. 166 18 36.08 36 48.66 99 35.56 54 24.72 36 22.51 51 10.5 36 North America Prince William Sound 0 4 2.98 8 7.32 12 3.17 12 6.55 8 13.316 12 6.78 8 Total within 38.16" 22 39.06 44 55.98 111 38.73 66 31.27 44 35.82 63 17.28 44 Between Alaska and Asia 7.36ft 1 28.86"' 2 42.64d 3 28.71"' 3 6.12" 2 23.20"' 3 4.09 2 Total for collections 45.52ft 23 67.93" 46 98.62 114 67.44 69 37.39 46 59.02 66 21.37 46 LDH-B2* PE PIT PGM -2* PEPD-1* PE PD -2* FH GR l- Collection site G df G df G df G df G df G df G df Asia Hokkaido Sakhalin 0 3 0.88 2 1.37 2 0 1 2.56 4 0 2 2.90 2 Dolinka — — — — — — — Lutoga 1.05 3 — 1.05 2 0 1 1.52 8 0 2 0.35 2 Monetka Ochepukha 0 3 — 1.41 2 2.04 1 6.28" 4 2.78 2 2.95 2 Other — — — — 7.46 20 — — Znamenka 1.39 3 — 1.39 2 4.25 3 11.10 28 5.25 2 0.08 2 Within Ochepukha 1.39 3 — 1.39 2 4.25 3 18.56 48 5.25 2 0.08 2 Between Ochepukha 3.01 3 — 0.81 2 1.10 1 5.81 4 2.44 2 0.03 2 Total Ochepukha 4.40 6 — 2.20 4 5.35 4 24.37 52 7.68 4 0.11 4 Within Sakhalin 5.45 12 — 4.66 8 7.39 6 32.16 64 10.46 8 3.41 8 Among Sakhalin 8.11 9 — 0.62 6 2.24 3 13.24" 12 6.49 6 3.67 6 Total Sakhalin 13.56 21 — 5.28 14 9.63 9 45.41 76 16.95 14 7.08 14 Northern Sea of Okhotsk - — 0 2 0 1 0.57 4 — — Western Kamchatka 3.00 9 7.00 6 0 8 8.34 4 16.05 16 6.34 6 10.05 8 Eastern Kamchatka 0 3 1.30 2 0 2 0.27 1 0.81 4 0.72 2 0.18 2 Within Asia 16.56 36 9.17 10 6.65 28 18.24 16 65.40 104 24.00 24 20.21 26 Among Asia 5.23 12 9.78 6 10.08 8 6.45 4 74.59"' 16 79.18"' 8 76.78"' 8 Total Asia 21.79 48 18.95 16 16.73 36 24.69 20 139.99 120 103.18"' 32 96.98"' 34 North America Prince William Sound 3.20 12 5.60 8 8.16 8 2.37 4 3.09 16 11.02 8 3.26 8 Total within 24.99 60 24.55 24 24.89 44 27.06 24 143.09 136 114.20"' 40 100.24"' 42 Between Alaska and Asia 12.25'’ 3 6.27" 2 2.94 2 0.32 1 18.69' 4 6.72" 2 33.91"' 2 Total for collections 37.24 63 30.82 26 27.82 46 27.38 25 161.78 140 120.92"' 42 134.15"' 44 sMDH- sMDH- Total for Al 2* Bl, 2* mMEP- 1 * MPI PEP 11 mAH -2* loci Collection site G df G df G df G df G df G df G df Asia Hokkaido Sakhalin 0.09 4 2.74 6 0.77 2 0.001 2 6.64" 2 4.30"' 1 48.87" 62 Dolinka — 1.95 6 5.95“ 2 — — — 8.63 14 Lutoga 0.77 4 4.47 18 3.67 6 0.21 2 0.05 2 — 34.47 89 con tin tied 130 Fishery Bulletin 99(1) Table 2 (continued) sMDH- sMDH- Total for Al,2 * Bl,2* mMEP-1* MPI PEPB* mAH-2* loci Collection site G df G df G df G df G df G df G df Monetka 1.75 4 1.20 6 0.01 2 0.34 2 1.33 2 2.07 1 41.20 59 Ochepukha Other — 25.79 30 3.28 10 — — — 46.70 90 Znamenka 0.11 4 37.29 48 4.14 16 1.35 2 8.47 6 0 1 121.04 186 Within Ochepukha 0.11 4 63.08 78 7.43 26 1.35 2 8.47 6 0 1 167.74 276 Between Ochepukha 0.03 4 6.60 6 1.25 2 1.87 2 3.16 2 — 45.91 58 Total Ochepukha 0.15 8 69.68 84 8.67 28 3.22 4 11.62 8 0 1 213.66 334 Within Sakhalin 2.67 16 77.30 114 18.30 38 3.77 8 13.01 12 2.07 2 297.94 496 Among Sakhalin 8.52 12 16.07 18 1.10 6 2.28 6 3.54 6 4.44 3 131.01 177 Total Sakhalin 11.19 28 93.37 132 19.40 44 6.05 14 16.55 18 6.51 5 428.95 673 Northern Sea of Okhotsk 0.60 4 0 6 0.66 2 0 2 9.55h 2 — 16.41 41 Western Kamchatka 0.16 12 18.47 24 4.87 8 5.21 8 6.08 8 6.55 3 190.11 220 Eastern Kamchatka 4.08 4 4.02 6 3.38 2 1.57 2 2.14 2 1.42 1 39.07 61 Within Asia 16.11 52 118.60 174 29.07 58 12.82 28 40.96 32 18.78 10 723.41 1057 Among Asia 74.76'' 16 45.82fe 24 8.14 8 15.98° 8 76.91'' 8 15.69'’ 4 768.81'' 236 Total Asia 90.87° 68 164.42 198 37.21 66 28.81 36 117.87'' 40 34.47'1 14 1492.22'' 1293 North America Prince William Sound 5.67 16 13.22 24 4.12 8 6.27 8 11.49 8 8.31° 3 176.78“ 247 Total within 96.53 84 177.64 222 41.33 74 35.08 44 129.36rf 48 42.78 17 1669.0H 1540 Between Alaska and Asia 9.40 4 12.09 6 4.06 2 1.73 2 12.886 2 2.33 1 717.87'' 62 Total for collections 105.93 88 189.73 228 45.39 76 36.81 46 142. 24d 50 45.11' 18 2386.85'' 1602 "P < 0.05, hP < 0.01, P < 0.001, dP < 0.0001 No overall heterogeneity was observed in tests among collections within each of the remaining geographic re- gions: northern Okhotsk coast, eastern Kamchatka, and western Kamchatka (Table 2). The northern Okhotsk and western Kamchatka collections each had a single locus that indicated heterogeneity, but neither test was signifi- cant after correction for multiple testing. The Alaska (Prince William Sound) collections exhibited low heterogeneity over all loci (G=176.78, 147 df; P=0.047) (Table 2); tests at three loci suggested that heterogeneity was not significant after correction for multiple testing. Although not the focus of this paper, the overall hetero- geneity suggests local genetic structure in even-year pink salmon within Prince William Sound. The collections of even-year pink salmon within the Asian geographic regions were relatively homogeneous, but we found strong heterogeneity among regions (G=1492.22, 1257 df; PclCP4) (Table 2). Of the nine loci that individu- ally suggest heterogeneity, eight ( GDA *, sAAT-3 *, PGDH *, PEPD-2 *, FH*, GR*, sMDH-Al *, and PEPB *) showed sig- nificant heterogeneity (P<0.05) after correction for multi- ple testing. The ratios of heterogeneities among regions to heterogeneity within regions (approximate P’s) for each locus (not shown) were significant (P<0.05) for 13 of 28 lo- ci, and comparisons of GDA*, sAAT-3 *, PGDH*, PEPD-2 *, FH*, GR*, sMDH-Al *, sMDH-B2* , and PEPB were high- ly significant (PcO.OOl). At the next level of hierarchy, between continents, Asian samples (in aggregate) differed overall from the Alaska samples with which they were compared (G=717.87, 62 df; P«10-6) (Table 2); many of the loci examined contribute to the difference. After correction for multiple tests, GDA*, FDHG*, sAAT-3 *, PGDH*, TPI-2 *, in AH -4*, G3PDH-1*, GPI-B2*, LDH-A1*, and GR* were strongly significant (PcO.OOl). Of these loci, GDA*, sAAT-3 *, PGDH*, mAH-4 *, G3PDH-1* , LDH-A1 *, and GR* had appreciable variation (common allele <0.95) in at least some populations. These results suggest that at least seven allozyme loci may prove useful for distinguishing among even-year pink salmon stocks from different regions of the northern Pacific Ocean (Hawkins et ah, 1998). Data from 36 loci common to all regions were used to estimate pairwise chord distances (Cavalli-Sforza and Edwards, 1967) with which we constructed an unrooted neighbor-joining tree (Fig. 2). The tree supports a geo- graphic basis for the variability observed among collec- tions and suggests a geographic relationship among re- gions. Four clusters are evident along the tree axis: in linear order, they consisted of the collections from south- ern Okhotsk (Hokkaido Island and Sakhalin Island), west- ern Kamchatka, eastern Kamchatka, and Alaska. The greatest distance was between the Alaska cluster and all the other collections. When the Magadan samples were included in a similar tree with data from 34 loci, they Noll et al.: Analysis of genetic structure of Asian and western Oncorhynchus gobuscha 131 '\ PWS(17D) s°c? • - • - PWS(17C) \ \ v PWS(17B) Hokkaido E. Kamchatka till 1 i ■ — i — — i — i i — i 1 1— — a 0 0.01 0.02 0.03 0 04 0.05 0 06 0 07 Genetic distance Figure 2 Neighbor-joining tree of even-year pink salmon broodlines from Japan, Russia, and Prince William Sound based on Cavalb-Sforza and Edwards (1967) chord distances from allele frequencies at 36 loci. Significance between nodes or for collections joined at nodes were tested within regions by using log-likelihood ratios (G-tests; Sokal and Rohlf, 1995) of the 36 loci to test homogeneity of branches joined at a node, and between basins by tests of homogeneity among the populations of the two adjacent regions. The numbers following the population names correspond to their locations in Figure 1. PWS = Prince William Sound. clustered with the eastern and western Kamchatka collec- tions (not shown). The arrangement of the other clusters remained as in Figure 2. The variation at 36 loci common to all collections was partitioned by using gene diversity analysis (GST’ s; Chakraborty and Leimar, 1987) and analysis of variance (0ST’s; Weir and Cockerham, 1984; Weir, 1996) (Table 3). Analysis of variance accounts for bias inherent in analyses that use finite samples from and small numbers of popula- tions. Differences between the Gsr’ s and dST’s are most ap- parent at upper levels of the hierarchy where fewer groups are compared. Although biased, the GST’ s have been wide- ly used, so we summarize results of partitioning variation with Gst’s but included 0ST’s parenthetically. On average, variation within streams accounted for 97.2% (95.57%) of the total, with 0.63% (0.05%) attributable to differences among streams within regions, 0.65% (1.29%) to differences among regions within continents, and 1.52% (3.09%) to differences between continents. We combined our data with available data (Altukhov et al., 1983; Salmenkova and Omelchenko, 1983; Zhivitovsky et al., 1989; Kartavtsev, 1991; Kartavtsev et al., 1992) to conduct a log-likelihood ratio analysis to test for homo- geneity within and among Asian regions. The primary (variable) loci for which data were available were PGDH* (omitting collections missing the *, 95 allele), G3PDH-1 *, and MDH-B1,2*. These data included collections from Pri- morie, western Sakhalin Island, and the Kuril Islands in addition to the regions we reported (Table 4). There was no overall heterogeneity within regions and only a single test suggested heterogeneity. However, PGDH* and MDH-B1,2* exhibited strong heterogeneity (P<10 4) among regions. Data from an earlier study of genetic diversity in Alaska pink salmon (Gharrett et al., 1988) were included to pro- duce an unrooted neighbor-joining tree based on 21 loci common (Appendix 1) to that study and the present one (Fig. 3). Data from all collections were condensed by re- gion, except for Japan, where the collections were hetero- geneous. Some regions were represented by a single collec- tion; data for the Aleutian Islands collections were pooled because they exhibited no heterogeneity, but as a group differed from other Western Alaska regions (Gharrett et al., 1988). The Magadan collections were omitted from our analysis because data from several of the 21 loci were un- available. As in the previous analysis, the tree showed a clear geographic basis for genetic variation. A cluster con- taining the Japan and Sakhalin Island collections adjoined the western Kamchatka group. A longer span joined the latter with a cluster consisting of the eastern Kamchat- 132 Fishery Bulletin 99(1 ) Table 3 Gene diversity analysis of even-year pink salmon broodlines. Relative gene diversities are estimated from the variable loci in the 36-locus set that these collections had in common. Estimates and standard deviations are jackknife estimates over loci and based on the total expected heterozygosity at each level of hierarchy (G’s; Chakraborty and Leimar, 1987) or analysis of variance (0’s; Weir, 1996) in parentheses. Source Number of collections Gst GSw (0S-0S) G LR Wg-6P) grt (Op) Asia 13 0.0150 ±0.0007 0.0051 ±0.0001 0.0099 ±0.0007 (0.0144 ±0.0011) (0.0003 ±0.0001) (0.0140 ±0.0012) Southern Sea of Okhotsk 6 0.0024 ±0.0001 0.0016 ±0.0001 0.0007 ±0.0001 (0.0000 ±0.0002) (-0.0003 ±0.0001) (0.0003 ±0.0001) Hokkaido 2 0.0022 ±0.0002 (-0.0008 ±0.0004) Southern Sakhalin 4 0.0014 ±0.0001 (-0.0002 ±0.0001) Western Kamchatka 5 0.0085 ±0.0003 (0.0019 ±0.0003) Eastern Kamchatka 2 0.0053 ±0.0005 (0.0030 ±0.0010) Alaska (Prince William Sound) 5 0.0089 ±0.0004 (0.0005 ±0.0005) Total 18 0.0280 ±0.0021 0.0063 ±0.0002 0.0065 ±0.0005 0.0152 ±0.0019 (0.0443 ±0.0055) (0.0005 ±0.0001) (0.0129 ±0.0011) (0.0309 ±0.0055) Table 4 Log-likelihood ratio analysis of allozyme data available for Asian even-year pink salmon broodlines (Altukhov et al., 1983; Salmen- kova and Omelchenko, 1983; Zhivotovsky et ah, 1989; Kartavtsev, 1991; Kartavtsev et al., 1992; our paper). Data include three alleles or allele pools (*100, *95, and *90) at PGDH* for 4587 individuals from 42 collections, two alleles or allele pools (*100 and all others) at G3PDH * for 6398 fish from 53 collections, and three alleles or allele pools (*100, fast alleles, and slow alleles at sMDH-Bl,2 * for 6336 individuals from 53 collections. Region PGDH* G (df) G3PDH* G (df) MDHB-1,2* G (df) Hokkaido 5.84 (2) 0.52(1) 0.80(2) Primoriya 6.08(6) 8.10(4) 16.66* (8) Western Sakhalin Island 0.71 (2) 1.04(3) 6.03 (6) Eastern Sakhalin Island 19.67 (16) 14.99 (12) 34.94 (24) Kuril Islands 33.36(26) 11.83(13) 16.67(26) Northern Sea of Okhotsk 4.37(4) 2.02(5) 15.88(10) Western Kamchatka 7.50(10) 8.53 (5) 9.42 (10) Eastern Kamchatka 3.04 (2) 0.03 (2) 6.59 (4) Total within region 80.58 (68) 47.07 (45) 107.00(90) Among regions 110.05"* (14) 13.92(7) 50.21*** (14) Total 190.61*** (82) 60.99(52) 157.21** (104) * = P < 0.05; ** = P < 0.001; ***= P < 0.0001. ka, Aleutian Islands, Norton Sound, and Bristol Bay col- lections. A cluster consisting of Kodiak Island and Prince William Sound collections was separated from the other groups by the longest distance in this tree. The neighbor- joining tree suggested three large geographic aggregations of pink salmon populations which correspond to marine basins: the Sea of Okhotsk, the Bering Sea, and the Gulf of Alaska. Using those basins as the basis for gene diver- sity hierarchy, we partitioned the variation. We estimated the proportion of the total variation attributable to differ- Noll et al.: Analysis of genetic structure of Asian and western Oncorhynchus gobuscha 133 Sea of Okhotsk Bering Sea Gulf of Alaska W Kamchatka Japan (Pacific) 3 ■ i • i • ' i I I f I , — T \ 0 0 01 0.02 0.03 0.04 0.05 0 06 0 07 Genetic distance Figure 3 Neighbor-joining tree based on allele frequencies at 21 loci, showing Cavalli-Sforza and Edwards ( 1967) chord distances among pink salmon from Japan, Russia, and Alaska. Sig- nificance between nodes or of collections joined at nodes were tested within basins by using log-likelihood ratios (G-tests; Sokal and Rohlf, 1995) of the 21 loci to test homogene- ity of branches joined at a node, and between basins by tests of homogeneity of the two adjacent branches. Jack-knifing populations to test the stability of the tree produced two local rearrangements that involved only the Aleutian Islands and Bristol Bay collections. The eastern Pacific Ocean and Bering Sea collections had the following sample sizes: Bris- tol Bay, n = 146; Aleutian Islands, n = 642; Norton Sound, n = 201; and Kodiak Island, n = 66 (Gharrett et ah, 1988). Table 5 Gene diversity analysis of even-year pink salmon broodlines. Relative gene diversities are estimated from the 21 -locus set common to collections of this study and that of Gharrett et al. ( 1988). Estimates and standard deviations are jackknife estimates based on the total expected heterozygosity at each level of hierarchy. Number of Source regions Gst gsg Ggt Sea of Okhotsk 3 0.0031 ±0.0006 Bering Sea 4 0.0079 ±0.0003 Gulf of Alaska 2 0.0043 ±0.0006 Total 9 0.0230 ±0.0025 0.0055 ±0.0002 0.0175 ±0.0025 ences among regions ( GST ) as well as to the average varia- tion among regions within a basin ( Gsg ) and among basins (Ggt) (Table 5). The diversity within areas was low, rang- ing from 0.31 to 0.79%, and averaging 0.55%. The diver- sity among areas was 1.75%, and the average within each stream was 98.25%. We also analyzed genetic variability by comparing aver- age expected heterozygosities (Nei, 1978). Eastern Pacific pink salmon populations appeared to have higher hetero- zygosities than western populations. With the 36-locus da- ta set, the average heterozygosity of the five Alaska collec- tions was 0.074 ±0.004 (mean ±SE) compared with 0.056 ±0.004 for 13 Asian collections. Using the 21 loci common to the three geographic regions shown by Figure 3, we es- timated heterozygosities of 0.061 ±0.002 for three regions within the Sea of Okhotsk, 0.073 ±0.004 for four regions in the Bering Sea, and 0.099 ±0.003 for the two regions in the Gulf of Alaska. Heterozygosities for these samples of allozyme loci increased from west to east. Discussion Heterogeneity among even-year pink salmon populations in Asian regions contrasted strongly with the relative homogeneity we observed within regions. For example, 134 Fishery Bulletin 99(1 ) western Kamchatka and eastern Sakhalin Island face each other across the Sea of Okhotsk but have different allele frequencies at PEPD-2*100 [0.71 ±0.02 (SE) for western Kamchatka versus 0.61 ±0.01 for eastern Sakha- lin), GR*100 (0.90 ±0.01 versus 0.97 ±0.005), PEPB*100 (0.91 ±0.01 versus 0.79 ±0.01), and PGDH*100 (0.76 0.02 versus 0.82 ±0.01). Therefore, it is unlikely that there is substantial gene flow between these regions. If large movements of spawning pink salmon occur (fluctu- ating stock hypothesis; e.g. Zhivotovsky and Glubokovsky, 1989; Shuntov et al. 1994), our results suggest that they are restricted to within-region movements for even-year pink salmon. Moreover, the relative homogeneity observed within regions does not necessarily indicate large numbers of strays. Relatively small exchanges between drainages (say 10 fish or so per generation) can arrest genetic diver- gence (e.g. Gharrett, 1994). This level of straying does not, however, provide demographic insurance against overfish- ing or environmental catastrophes in the short run (Hop and Gharrett, 1989). The contemporary genetic structure of even-year pink salmon has practical implications for fishery scientists. Genetic differences among populations can be used as markers for stock identification. It is often necessary to manage a species that has as many small populations as pink salmon as a regional assemblage. The regional basis for genetic structure observed in Asian pink salmon lends itself to stock separation analyses (Hawkins et al., 1998). Most of the genetic variability observed in even-year pink salmon is attributable to within-population variabil- ity (e.g. Gharrett et al., 1988; Beacham et al., 1988). How- ever, low levels of divergence among populations do not preclude hierarchical population structure. In our study we saw little or no temporal structure among Sakhalin Island populations. This finding does not necessarily con- flict with observations of temporal structure observed by Altukhov et al. (1983) who used much larger numbers of fish (but fewer alleles), by McGregor et al. (1998) who used multiple years of data and numerous loci, and by Brykov et al. ( 1999) who used highly variable mitochondrial DNA haplotypes, because their tests had much greater statisti- cal power. In addition, allozymes may not be appropriate for detecting some kinds of population structure because a very low level of gene flow can “homogenize” frequen- cies of neutral or nearly neutral loci. Note, for example, the genetic component observed for time of return within a spawning season (Smoker et al., 1998) and the persis- tence of a genetic marker for time of spawning (Lane et al., 1990). The strong regional structure we observed in Asian even-year pink salmon populations is a stark contrast to the nearly complete absence of structure reported for odd- year Kamchatka pink salmon (Varnavskaya and Beacham, 1992; Shaklee and Varnavskaya, 1994). However, the lat- ter surveys covered smaller geographic ranges and in- cluded neither Sakhalin Island nor Japanese pink salmon populations. These combined studies involving numerous allozyme loci confirm the work of Zhivotovsky et al. (1989) who, using four allozyme loci, also recognized the brood- year differences; and who estimated that 1.6% (we esti- mated 1.75%) and 0.6% of the total genetic variability were attributable to interregional divergence for even- and odd-year broods, respectively. The differences observed between Asian and North American even-year pink salmon are not surprising; they have been reported previously for both even-year (Zhivo- tovsky et al., 1989) and odd-year broodlines (Varnavska- ya and Beacham, 1992; Shaklee and Varnavskaya, 1994). However, the strong coherence of populations within each of the major North Pacific basins (Sea of Okhotsk, Bering Sea, and Gulf of Alaska) is striking. Each basin includes a range of habitats and climates that suggests that the genetic similarity among populations within a geographic region does not result from convergent or homogeneous selection. The differences among populations of different basins are also reflected by the different average hetero- zygosities. The apparent directional change in heterozy- gosities could be interpreted in terms of differences in effective population size, age of the populations, or the ex- tent of environmental variation. However, speculation is probably not warranted because three different variables can be arranged in a monotonic pattern in two of six dif- ferent possible orders. A more evocative explanation of the genetic structure combines geographic and oceanographic influences, as well as recent geologic history. Populations in the contemporary Sea of Okhotsk, Bering Sea, and Gulf of Alaska are sep- arated geographically by land masses and oceanographi- cally by the different currents that flow into or between the oceanic basins and that influence migration routes of the fish (Royce et al., 1968). The geographic separation was greatly exacerbated by the limits and effects of late Pleisto- cene glaciation. In their northern range, pink salmon pop- ulations experienced increased isolation between the ma- rine basins as a result of lower sea level, loss of freshwater habitat to increased ice cover, and more extensive sea ice. Just as recent fluctuations in salmon productivity have resulted from relatively minor climate changes (Mantua et al., 1997), less favorable freshwater and marine envi- ronmental conditions undoubtedly decreased the sizes and numbers of populations dramatically, further isolating the remaining populations in this region. During the past several 100,000 years, there have been periodic changes in climate, sea level, glacial extent, and oceanographic conditions. It is important to realize that our modern, interglacial conditions are an extreme (San- cetta and Silvestri, 1986) and that the oceanic record of global ice volumes (from the d180 record in marine sedi- ments) and geologic evidence (Mann and Hamilton, 1995) from the north Pacific realm during that period indicate that lower sea levels, more extensive glaciation, colder sea surface temperatures, and more extensive and southerly sea ice were typical (Bartlein et al., 1991; Rohling et ah, 1998). The relative proportion of the 180 isotope (<5180) in a stratum of a core is related to the portion of the earth’s wa- ter tied up in ice at the time corresponding to the stratum, and consequently the sea level in relation to the modern sea level. At the last glacial maximum (LGM: ca. 14,000-20,000 years before the present [BP] ), the paleogeography of the Noll et a!.: Analysis of genetic structure of Asian and western Oncorhynchus gobuscho 135 North Pacific coast line was severely altered, reducing the sizes of the enclosed basins as well as the circulation among them. 5180 records (e.g. Bartlein et al., 1991; Roh- ling, 1998), and limited data from Beringia (Hopkins, 1982) suggest that during most of the mid-Wisconsin glaciation, sea level was 50 m lower than at present and that at the LGM, the level was 120 to 130 m lower in areas not af- fected by ice loading or tectonic activity (Fairbanks, 1989). Sakhalin Island was connected to the mainland and Hok- kaido Island, blocking the northern outflow of the Sea of Japan; and the connection of a smaller Sea of Okhotsk with the Pacific Ocean through the Kurile islands was restrict- ed. In the Bering Sea, the extensive continental shelf was exposed, blocking circulation through the Bering Strait. To the south, the Aleutian Islands and the Alaska Peninsula were joined to about 170°W, and many of the islands to the west were also connected, thus limiting the connections be- tween the Gulf of Alaska and the Bering Sea. In Cook In- let, glaciers may have remained advanced throughout the middle Wisconsin glaciation (Reger and Updike, 1983). Pink salmon spawning habitat along the Gulf of Alaska was nearly completely eliminated at the LGM and much of the Beringian coastline was probably ice bound much of the year, minimizing available freshwater habitat. Coastal areas of the Gulf of Alaska, including most of the continen- tal shelf, were extensively glaciated during the Quaterna- ry Period, although a few isolated areas of the outer coast may have been ice free (Hamilton, 1986); and the ice cover of the eastern Aleutians coalesced with the Alaska Penin- sula and ice caps covered all the major islands (Thorson and Hamilton, 1986). In these areas there may have been ephemeral streams from melted snow or ice at the south- ern margin, near the present shelf break. Although the rivers draining the Yukon-Kuskokwim region flowed over the exposed shelf and probably served as a refugium, the Bering Sea appears to have had sea ice cover much of the year (Sancetta, 1983; Sancetta and Robinson, 1983); and seasonal sea ice may have persisted as far south as 54°N for 6-8 months a year during the LGM (de Vernal and Pedersen, 1997). On fhe Asian side, glaciation was much less extensive and included some alpine glaciation; but few glaciers extended to tidewater, except possibly on the southeast side of the Kamchatka Peninsula (Bespalyy, 1984; Velichko, 1984; Anderson, 1981). Freshwater habi- tat was probably not reduced to the extent of the Gulf of Alaska coast. Nevertheless, the Sea of Okhotsk, like the Bering Sea, probably had sea ice cover much of the year (Sancetta, 1983; Sancetta and Robinson, 1983). Harsh conditions greatly reduced marine surface water productivity over the entire region (Morley et al., 1991; Keigwin et al., 1992; deVernal and Pederson, 1997). Micro- fossils indicate that the subarctic Pacific Ocean was sim- ilar to the present day Sea of Okhotsk, with cold fresh surface water and a highly stratified water column. Sea surface temperatures estimated by CLIMAP were 2° to 4°C colder than present throughout the year over most of the area (Moore et al., 1980). Off Japan, temperatures were even colder (>6°C) at the LGM, indicating that the cold Oyashio Current penetrated farther south than it does at present (Moore et al., 1980). Most of the description above considered the LGM, but 5180 records indicate that periodically other major glacial advances occurred at about 135,000 BP (180 stage 6), about 225,000 BP (180 stage 8), and so forth (Bartlein et al., 1991). Another less extensive advance may have oc- curred in the North Pacific region about 75 ka BP (Hop- kins, 1982). It is likely that those events also influenced the distribution and demographics of salmon species in- cluding pink salmon. Many of the streams populated by pink salmon are short coastal streams that are transient. Consequently, in maxi- mizing productivity opportunities, pink salmon may exhib- it a higher level of gene flow (straying) than other Pacific salmon (Quinn, 1984). Geologic evidence suggests that the LGM did not affect Asian streams as broadly or severely as Alaskan streams. However, the freshwater environments were less favorable than at present, and it would be expect- ed that the harsh marine environment severely reduced productivity, probably creating a situation in which many local populations repeatedly went extinct. The ability to exploit spawning habitat rapidly would have been an ad- vantage during the LGM and many local extinctions were probably followed by recolonization. As a consequence, it is likely that a few systems provided the source for stock colo- nization following deglaciation. In the eastern range where habitat was ice-covered, re-establishment of pink salmon probably depended on colonization from the Bering or more southerly refugia, or possibly from the “off-year” broodline, if the rigid two year life cycle relaxes in marginal environ- ments, such as appears to have happened in the Lauren- tian Great Lakes (e.g., Kwain and Chappel, 1978). Overall, the geological events should have accentuated geographic boundaries and increased the importance of random drift. Divergence between Asian and North Amer- ican populations suggests colonization from different re- fugia, and significant, but lesser, regional divergence sup- ports homing in pink salmon, at least regionally. However, the low overall divergence observed among both Asian and North American populations (Gsr-0 .023) suggests that the populations studied are either recently diverged and derived from a single population or from genetically simi- lar ancestral populations, or that there is sufficient gene flow to arrest divergence. One of the advantages of study- ing pink salmon is that there are two broodlines occupy- ing much of the same range. It will be interesting to ex- amine the genetic structure of odd-broodline pink salmon in the same range, which will represent a second natural experiment with which to examine the influences of geog- raphy, oceanography, and geologic history on the genetic structure of pink salmon populations. Acknowledgments We are indebted to many individuals who assisted in pro- viding samples and grateful to Hanhvan Thi Nguyen and Bichhang Thi Nguyen for their valuable assistance in the laboratory. R. Wilmot provided helpful comments and help with federal bureaucratic processes. Two anonymous reviewers provided constructive criticism. 136 Fishery Bulletin 99(1) Literature cited Aebersold, P. B., G. A. Winans, D. J. Teel, G. B. Milner, and F. M. Utter. 1987. Manual for starch gel electrophoresis: a method for the detection of genetic variation. U.S. Dep. Commer., NOAA Tech. Rep. NMFS 61. Allendorf, F., N. Ryman, and F. Utter. 1987. Genetics and fishery management: past, present, and future. In Population genetics and fishery management (N. Ryman and F. Utter, eds.), p. 1-20. Univ. Washington Press, Seattle, WA. Allendorf, F., and G. A Thorgaard. 1984. Tetraploidy and the evolution of salmonid fishes. In Evolutionary genetics of fishes (B. J. Turner, ed.), p. 1-53. Plenum, New York, NY. Altukhov, Yu. P, E. A. Salmenkova, V. T. Omelchenko, and V. N. Efanov. 1983. Genetic differentiation and population structure of pink salmon of the Sakhalin-Kuril region. [English transl.] Biologiya Morya 2:46-51. Anderson, B. G. 1981. Late Weichselian ice sheets in Eurasia and Green- land. In The last great ice sheets (J. H. Denton and T. J. Hughes, eds.), p. 3-65. John Wiley & Sons, New York, NY. Aspinwall, N. 1974. Genetic examination of North American populations of the pink salmon, Oncorhynchus gorbuscha, possible evi- dence for the neutral mutation-random drift hypothesis. Evolution 28:295-305. Bartlein, P. J., P. M. Anderson, M. E. Edwards, and P. F. McDowell. 1991. A framework for interpreting paleoclimatic variations in eastern Beringia. Quatern. Internat. 10-12:1991. Beacham, T. D., R. E. Withler, and A. P. Gould. 1985. Biochemical genetic stock identification of pink salmon ( Oncorhynchus gorbuscha ) in Southern British Columbia and Puget Sound. Can. J. Fish. Aquat. Sci. 42:1474-1483. Beacham, T. D., R. E. Withler, C. B. Murray, and L. W. Barner. 1988. Variation in body size, morphology, egg size, and bio- chemical genetics of pink salmon in British Columbia. Trans. Am. Fish. Soc. 117:109-126. Bespalyy, V. G. 1984. Late Pleistocene mountain glaciation in Northeastern USSR. In Late Quaternary environments of the Soviet Union (A. A. Velichko, H. E. Wright Jr., and C. W. Barnosky, eds.), p. 31-33. Univ. Minnesota Press, Minneapolis, MN. Boyer, S. H., D. C. Fainer, M. A. Naughton. 1963. Myoglobin: Inherited structural variation in man. Sci- ence (Washington D.C. >140:1228-1231. Brykov, V. A., N. Polyakova, L.A. Skurikhina, A.D. Kukhlevsky, O N. Kirillova, D. Churikov, A.I. Pudovkin, and A. J. Gharrett. 1999. Analysis of mtDNA indicates weak temporal genetic heterogeneity in pink salmon spawning runs in two rivers on Sakhalin Island. J. Fish Biol. 55:617-635. Cavalli-Sforza, L. L., and A. W. F. Edwards. 1967. Phylogenetic analysis: models and estimation proce- dures. Evolution 21:550-570. Chakraborty, R., and O. Leimar. 1987. Genetic variation within a subdivided population. In Population genetics and fishery management (N. Ryman and F. Utter, eds.), p. 89-120. Univ. Washington Press, Seattle, WA. Clayton, J. W., and D. N. Tretiak. 1972. Amino citrate buffer for pH control of starch gel elec- trophoresis. J. Fish. Res. Board Can. 29:1169-1172. Cooper, D. W. 1968. The significance level in multiple tests made simulta- neously. Heredity 23:614—617. de Vernal, A., and T. F. Pedersen. 1997. Micropaleontology and palynology of core PAR87A-10: A 23,000 year record of paleoenvironmental changes in the Gulf of Alaska, northeast North Pacific. Paleoceanogra- phy 12:821-830. Fairbanks, R. G. 1989. A 17,000-year glacio-eustatic sea level record: influ- ence of glacial melting rates on the Younger Dryas event and deep ocean circulation. Nature 342:637-642. Gagalchii, N. G. 1986. Population genetic analysis of the Kamchatka pink salmon Oncorhynchus gorbuscha (Walb.). Cand. Sci. diss., Leningrad State Univ., Leningrad. Gharrett, A. J. 1994. Genetic dynamics of a small population system: a model applicable to interaction between hatchery and wild fish. Aquacult. Fish. Manage. 25(suppl. 2):79-97. Gharrett, A. J., C. Smoot, A. J. McGregor, and P. B. Holmes. 1 988. Genetic relationships of even-year northwestern Alas- kan pink salmon. Trans. Am. Fish. Soc. 117:536-545. Gharrett, A. J., and M. A. Thomason. 1987. Genetic changes in pink salmon (Oncorhynchus gor- buscha) following their introduction into the Great Lakes. Can. J. Fish. Aquat. Sci. 44:787-792. Glubokovsky, M. K., L. A. Zhivotovsky, R. M. Victorovsky, A. M. Bronevsky, K. I. Afanas’ev, V. V. Efremov, L. N. Ermolenko, B. A. Kalabushkin, V. G. Kovalev, A. N. Makoedov, T. V. Malinina, S. P. Pustovoit, and G. A. Rubtsova. 1989. Population organization of the pink salmon. [English transl.] Genetika 25:1275-1285. Hamilton, T. D. 1986. Late Cenozoic glaciation in the central Brooks Range. In Glaciation in Alaska: the geologic record (T. D. Hamilton, K. M. Reed, and R. M. Thorson, eds.), p. 9-49. Alaska Geo- logical Society, Anchorage, AK. Hawkins, S., N. Varnavskaya, J. Pohl, and R. Wilmot. 1998. Simulations of the even-year Asian pink salmon (Oncorhynchus gorbuscha) genetic baseline to determine accuracy and precision of stock composition estimates. N. Pac. Anadr. Fish Comm. Bull. 1:213-219. Heard, W. R. 1991. Life history of pink salmon (Oncorhynchus gorbus- cha). In Pacific salmon life histories (C. Groot and L. Mar- golis, eds.), p. 119-230. UBC Press, Vancouver, Canada. Holmes, R. S., and C. J. Masters. 1970. Epigenetic interconversion of the multiple forms of mouse liver catalase. FEBS Letters 11:45-48. Hop, H., and A. J. Gharrett. 1989. Genetic relatedness of Arctic grayling in the Koyu- kuk and Tanana rivers, Alaska. Trans. Am. Fish. Soc. 118:290-295. Hopkins, D. M. 1982. Aspects of the paleogeography of Beringia during the late Pleistocene. In Paleoecology of Beringia (D. M. Hop- kins, J. V. Matthews, C. E. Schweger, and S. B. Young, eds.), p. 3-28. Academic Press, New York, NY. Kartavtsev, Yu. F. 1991. Temporal and spatial variability in allelic frequencies in populations of pink salmon, Oncorhynchus gorbuscha. [English transl.] J. Ichthyol. 31(2):87-98. Kartavtsev, Yu. F., V. V. Efremov, M. V. Smirnov, E. V. Ivankova, and N. E. Polyakova. 1992. Analysis of genotype distributions at allozyme loci Noll et al.: Analysis of genetic structure of Asian and western Oncorhynchus gobuscha 137 in populations of pink salmon, Oncorhynchus gorbuscha. [English transl.l J. Ichthyol. 32(4):102— 1 15. Keigwin, L. D., G. A, Jones, and P. N. Froelich. 1992. A 15,000 year paleoenvironmental record from Meiji Seamount, far northwestern Pacific. Earth Planet. Sci. Lett. 111:425-440. Kwain, W., and J. A. Chappel. 1978. First evidence for even-year spawning pink salmon, Oncorhynchus gorbuscha, in Lake Superior. J. Fish. Res. Board Can. 35:1373-1376. Lane, S., A. J. McGregor, S. G. Taylor, and A. J. Gharrett. 1990. Genetic marking of an Alaskan pink salmon popula- tion, with an evaluation of the mark and marking process. Am. Fish. Soc. Symp. 7:395-406. Mann, D. H., and T. D. Hamilton. 1995. Late Pleistocene and Holocene paleoenvironments of the North Pacific coast. Quartern. Rev. 14:449-471. Mantua, N. J., S. R. Hare, Y. Zhang, J. W. Wallace, and R. C. Francis. 1997. A Pacific interdecadal climate oscillation with impacts on salmon production. Bull. Am. Meteorol. Soc. 78:1069- 1079. McGregor, A. J. 1983. A biochemical genetic analysis of northern Southeast Alaskan pink salmon ( Oncorhynchus gorbuscha). M.S. thesis, Univ. Alaska, Juneau, AK, 94 p. McGregor, A. J., S. Lane, M. A. Thomason, L. A. Zhivotovsky, W. W. Smoker, and A. J. Gharrett. 1998. Migration timing, a life history trait important in the genetic structure of pink salmon. N. Pac. Anadr. Fish Comm. Bull. 1:262-273. Moore, T. C., Jr., L. H. Burckle, K. Geitzenauer, B. Luz, A. Molina-Cruz, J. H. Robertson, H. Sachs, C. Sancetta., J. Thiede, P. Thompson, and C. Wenkam. 1980. The reconstruction of sea surface temperatures in the Pacific ocean of 18,000 B.P Marine Micropaleontology 5:215-247. Morley, J. J., L. E. Heusser, and N. J. Shackleton. 1991. Late Pleistocene/Holocene radiolarian and pollen records from sediments in the sea of Okhotsk. Paleocean- ography 6:121-131. Nei, M. 1978. Estimation of heterozygosity and genetic distance from a small number of individuals. Genetics 89:583-590. Park, L. K., and P. Moran. 1994. Developments in molecular genetic techniques in fish- eries. Reviews in fish biology and fisheries 4:272-299. Pella, J. J., and G. B. Milner. 1987. Use of genetic marks in stock composition analysis. In Population genetics and fishery management (N. Ryman and F. Utter, eds.), p. 247-276. Univ. Washington Press, Seattle, WA. Quinn, T. P. 1984. Homing and straying in Pacific salmon. In Mecha- nisms of migration in fishes ( J. D. McCleave, G. P. Arnold, J. J. Dodson, and W. H. Neill, eds.), p. 357-362. Plenum Press, New York, NY. Reger, R. D., and R. G. Updike. 1983. A working model for late Pleistocene glaciation of the Anchorage Lowland, upper Cook Inlet, Alaska. In Glacia- tion in Alaska — extended abstracts from a workshop (R. M. Thorson and T. D. Hamilton, eds.), p. 71-74. LTniv. Alaska Museum Occasional Paper 2, Fairbanks, AK. Ridgway, G. J., S. W. Sherburne, and R. D. Lewis. 1970. Polymorphisms in the serum esterases of Atlantic herring. Trans. Am. Fish. Soc. 99:147-151. Roffi D. A., and P. Bentzen. 1989. The statistical analysis of mitochondrial polymor- phisms: chi-square and the problem of small samples. Mol. Bio. Evol. 6:539-545. Rohling, E.J., M. Fenton, F.J. Jorissen, P. Bertrand, G. Ganssen, and J. P. Caulet. 1998. Magnitudes of sea-level lowstands of the past 500,000 years. Nature 396:162-165. Royce, W. F., L. S. Lenwood, and A. C. Hartt. 1968. Model as of oceanic migrations of Pacific salmon and comments on guidance mechanisms. U.S. Fish Wildlf. Serv. Fish. Bull. 66:441-462. Saitou, N., and M. Nei. 1987. The neighbor-joining method: a new method for con- structing phylogenetic trees. Mol. Biol. Evol. 4:406-425. Salmenkova, E. A., and V. T. Omelchenko. 1983. Polymorphism of NADP-dependent isocitrate dehy- drogenase and malate dehydrogenase in populations of chum and pink salmon. [In Russian, English abstract.] Biologiya Morya 3:24-28. Sancetta, C. 1983. Effect of Pleistocene glaciation upon oceanographic characteristics of the North Pacific Ocean and Bering Sea. Deep Sea Res. 30:851-869. Sancetta, C., and S. W. Robinson. 1983. Diatom evidence on Wisconsin and Holocene events in the Bering Sea. Quaternary Res. 20:232-245. Sancetta, C., and S.-M. Silvestri. 1986. Pliocene-Pleistocene evolution of the North Pacific Ocean-atmosphere system, interpreted from fossil diatoms. Paleoceanography 1:163-180. Schaal, B. A., and W. W. Anderson. 1974. An outline of techniques for starch gel electrophoresis of enzymes from the American oyster Crassostrea virginica Gmelin. Georgia Marine Science Center Tech. Rep. 74-3. Seeb, L. W., and P. A. Crane. 1999. High heterogeneity in chum salmon in Western Alaska, the contact zone between northern and southern lineages. Trans. Am. Fish. Soc. 128:58-87. Shaklee, J. B., F. W. Allendorfi D. C. Morizot, and G. S. Whitt. 1990. Gene nomenclature for protein-coding loci in fish. Trans. Am. Fish. Soc. 119:2-15. Shaklee, J. B , D. C. Klaybor, S. Young, and B. A. White. 1991. Genetic stock structure of odd-year pink salmon, Oncorhynchus gorbuscha , Walbaum, from Washington and British Columbia and potential mixed-stock fisheries appli- cations. J. Fish Biol. 29(suppl. A 1:21-34. Shaklee, J. B., and N. N. Varnavskaya. 1994. Electrophoretic characterization of odd-year pink salmon (Oncorhynchus gorbuscha ) populations from the Pacific coast of Russia and comparison with selected North American populations. Can. J. Fish. Aquat. Sci. 51(suppl. 1 ): 158-171. Shaw, C. R., and R. Prasad. 1970. Starch gel electrophoresis of enzymes — a compilation of recipes. Biochem. Genet. 4:297-320. Shuntov, V. P, V. V. Lapko, A. A. Balanov, and A. V. Startsev. 1994. Interannual changes in anadromous migrations of salmons in waters of the Sakhalin-Kuril Region. (NPAFC [North Pacific Andadramous Fisheries Comm.] Doc. 107) Laboratory of Applied Biocenology, Pacific Research Insti- tute of Fishery and Oceanography (TINRO) 1994. Vladi- vostok, Russia, 20 p. Smoker, W. W., A. J. Gharrett, and M. S. Stekoll. 1998. Genetic variation of return date in a population of pink salmon: A consequence of fluctuating environment 138 Fishery Bulletin 99(1) and dispersive selection? Alaska Fishery Res. Bulletin 5:46-54. Sokal, R. R., and F. J. Rohlf. 1995. Biometry, third ed. W.H. Freeman and Co., New York, NY, 887 p. Takagi, K., K. V. Aro, A. C. Hartt, and M. B. Dell. 1981 . Distribution and origin of pink salmon (Oncorhynchus gorbuscha) in offshore waters of the North Pacific Ocean. Committee on Biology and Research. Int. North Pac. Fish. Comm. Document 2178, Bulletin 40, International North Pacific Fisheries Commission, Vancouver, Canada. Thorson, R. M., and T. D. Hamilton. 1986. Glacial geology of the Aleutian Islands. In Glacia- tion in Alaska: the geologic record (T. D. Hamilton, K. M. Reed, and R. M. Thorson, eds.), p. 171-191. Alaska Geo- logical Society, Anchorage, AK. Utter, F. M., H. O. Hodgins, and F. W. Allendorf. 1974. Biochemical genetic studies in fishes: potentialities and limitations. In Biochemical and biophysical perspec- tives in marine biology, vol. 1 (D. C. Malins and J. R. Sar- gent, eds.), p. 213-238. Academic Press, San Francisco, CA. Varnavskaya, N. V., and T. D. Beacham. 1992. Biochemical genetic variation in odd-year pink salmon ( Oncorhynch us gorbuscha ) from Kamchatka. Can. J. Zool. 70:2115-2120. Varnavskaya, N. V., C. C. Wood, and R. J. Everett. 1994a. Genetic variation in sockeye salmon ( Oncorhynchus nerka) populations of Asia and North America. Can. J. Fish. Aquat. Sci. 51:132-146. Varnavskaya, N. V., C. C. Wood, R. J. Everett R. L. Wilmot, V. S. Varnavsky, V. V. Midanaya, and T. P. Quinn. 1994b. Genetic differentiation of subpopulations of sockeye salmon ( Oncorhynchus nerka) within lakes of Alaska, Brit- ish Columbia, and Kamchatka, Russia. Can. J. Fish. Aquat. Sci. 51:147-157. Velichko, A. A. 1984. Late Pleistocene spatial paleoclimatic reconstruc- tions. In Late quaternary environments of the Soviet Union (A. A. Velichko, H. E. Wright Jr., and C. W. Bar- nosky, eds.), p. 261-285. Univ. Minnesota Press, Minne- apolis, MN. Weir, B. S. 1996. Genetic data analysis II. Sinauer Assoc. Inc., Sun- derland, MA. Weir, B. S., and C. C. Cockerham. 1984. Estimates of F-statistics for the analysis of popula- tion structure. Evolution 38:1358-1370. Wilmot, R. L., R. J. Everett, W. J. Spearman, R. Baccus, V. V. Varnavskaya, and S. V. Putivkin. 1994. Genetic stock structure of western Alaska chum salmon and a comparison with Russian Far East stocks. Can. J. Fish. Aquat. Sci. 51:84-94. Winans, G. A., P. B. Aebersold, S. Urawa, and N. V. Varnavskaya. 1994. Determining continent of origin of chum salmon ( Oncorhynchus keta) using genetic stock identification tech- niques: status of allozyme baseline in Asia. Can. J. Fish. Aquat. Sci. 51:95-113. Wright, S. 1978. Evolution and the genetics of populations, vol. 4: Vari- ability within and among natural populations. Univ. Chi- cago Press, Chicago, IL. Zhivotovsky, L. A., and M. K. Glubokovsky. 1989. Role of migration and selection in genetic differenti- ation of pink salmon (Oncorhynchus gorbuscha). [English transl.l Doklady Akademii Nauk SSSR 308:1235-1240. Zhivotovsky, L. A., M. K. Glubokovsky, R. M. Victorovsky, A. M. Bronevsky, K. I. Afanasev, V. V. Efremov, L. N. Ermolenko, B. A. Kalabushkin, V. G. Kovalev, A. N. Makoedov, T. V. Malinina, S. P. Pustovoit, and G. A. Rubstova. 1989. Genetic differentiation of pink salmon. Genetika 25: 1261-1274. [English transl.]. 139 Abstract— Two tilapia species, Saro- therodon melanotheron (brackish water fish) and Oreochromis niloticus (fresh- water fish), were marked with tetracy- cline and reared in Cote d’Ivoire (West Africa) in intensive (fish fed) and exten- sive (fish unfed) conditions. Juvenile and adult otoliths of the two species were examined. They were cut into trans- verse thin sections (10 to 40 pm), and otolith microincrements were counted on the sulcus along the ventral axis. Results for both species showed that microincrements are laid down daily. The number of days of growth reflects the number of microincrements (regres- sion with slope not different from 1 and intercept not different from 0; P>0.05). This technique has a tendency to under- estimate age (P<0.05): for S. melano- theron., the mean bias error is 4.4 d for juveniles (48 to 169 d of growth) and 8.1 d for adults (34 to 185 d of growth); for O. niloticus , the mean bias error is 0.9 d for juveniles (31 to 62 d of growth) and 5.1 d for adults (36 to 65 d of growth). Back-calculation of individual length at marking is very sensitive to an uncoupling between otolith and fish- specific growth rates. With the present data, back-calculated lengths overesti- mated actual size. When otolith and fish growth were coupled, length was back- calculated accurately. Manuscript accepted 9 August 2000. Fish. Bull. 99:139-150 (2001). Validation of age estimation and back-calculation of fish length based on otolith microstructures in tilapias (Pisces, Cichlidae) Jacques Panfili Institut de Recherche pour le Developpement, LASAA B P 70 29280 Plouzane, France E-mail address: panfili@ird fr Javier Tomas Port Erin Marine Laboratory Port Erin Isle of Man IM9 6JA, United Kingdom “Tilapia” is the common name for three genera of endemic cichlids from Africa, Oreochromis, Sarotherodon, and Tila- pia. Species belonging to these three genera are widely distributed in tropi- cal areas and have colonized all kinds of continental waters as natural or intro- duced species. Their adaptative capabil- ities have also been used in developing an aquaculture for these species world- wide (Lowe-McConnell, 1982; Wootton, 1984; Pullin et al., 1988). Both the repro- duction and growth strategies in tila- pias, however, differ among populations depending on environmental conditions (Kolding, 1993). Therefore, correct esti- mates of age is of great importance in assessing different growth strategies, as well as in characterizing different populations, by means of parameters such as size at first maturity. Despite the need for accurate age estimates, age at first maturity has not been the subject of many studies in tilapias and the question remains whether the observed differences in size at matu- rity are due to growth or to age dif- ferences (Eyeson, 1983; Legendre and Ecoutin, 1989; Duponchelle and Panfili, 1998). Moreover, tilapias can be sexu- ally active a few months after hatching as shown in studies on tilapia reared in aquaculture (Eyeson, 1983). Since Pannella’s (1971) work, otolith microincrements have been widely used to estimate the age of fish in days, useful for studies on larvae and juve- nile fish (Jones, 1992). Several authors have studied microincrements in tila- pia otoliths, but their results have not been applied to field research. One pos- sible reason is that there is not a stan- dard and simple method available for choosing and preparing the otolith. Pre- vious works have used different oto- liths and preparation techniques: sagit- ta cut transversally and acid etched for Tilapia guineensis (Fagade, 1980),sagit- ta cut transversally and observed with scanning electron microscope for Oreo- chromis niloticus (Tanaka et al., 1981), lapillus observed whole in photonic mi- croscopy for Tilapia marine (Rosa and Re, 1985), sagitta cut sagittally and ob- served in scanning electron microscopy for Oreochromis aureus (Karakiri and Hammer, 1989), and sagitta cut trans- versally, etched and stained for Oreo- chromis niloticus (Zhang and Runham, 1992). Each of these authors, except Fag- ade ( 1980), has validated the daily depo- sition of microincrements. Nevertheless, all fish used for validation were juve- niles and were reared under controlled conditions in aquaria, far from their nat- ural environment. To date, no attempt has been made to estimate growth for tilapia in natural waters nor age at ma- turity by using microincrements. The first aim of our study was to as- sess whether examination of microin- crements in otoliths yields valuable esti- mates of age (in days) for both juve- niles and adults of Sarotherodon mel- anotheron (brackishwater species) and Oreochromis niloticus (freshwater spe- cies). To obtain results that could be ap- plied to field studies, fish were reared 140 Fishery Bulletin 99(1 ) in an environment as similar as possible to natural condi- tions. The second aim was to develop a standard method for otolith preparation and examination to be used for both species. Both objectives are particularly useful in the case of S. melanotheron for which no validation has previously been published. We chose to carry out experiments in two aquaculture stations in Cote d’Ivoire (Africa), where the fish and the otoliths were simultaneously marked with ex- ternal tags and tetracycline, respectively, to follow fish and otolith growth. Tetracycline labels have been used widely to mark calcified structures since the first assays (Weber and Rigway, 1967; Meunier, 1974), and we improved this universal marker for marking tilapia otoliths. During the validation process, special attention was paid to the accu- racy of the age estimations (Campana and Jones, 1992). Geffen (1992) reported in her review on validation that there is a lack of analysis of the variation in increment number at a given age and we focused on that particular point. Our data on individual fish and otolith tagging were useful to test a growth back-calculation model commonly used for fishes (Francis, 1990, for review; Campana and Jones, 1992; Smedstad and Holm, 1996) and particularly to test the problem of uncoupling between somatic and otolith growth reported in previous studies (Mosegaard et al., 1988; Reznick et ah, 1989; Secor and Dean, 1989). The final aim of our study was to obtain an accurate, precise, and simple tool for age estimation for future life history studies on tilapias. Materials and methods Rearing experiments Fish were reared in two aquaculture stations in Cote d’Ivoire (West Africa), a country that experiences a tran- sitional equatorial climate with two dry and two rainy seasons (Durand and Skubich, 1982; Durand and Guiral, 1994). Juveniles and adults of Sarotherodon melanotheron and Oreochromis niloticus were used in the experiments. Juvenile fish were obtained from synchronous layings, whereas adults, males and females, were caught in the nat- ural environment. Sarotherodon melanotheron was reared at the Layo station (Centre de Recherches Oceanologiques) located on the Ebrie lagoon. One-hundred and ninety-eight adults of S. melanotheron ranging between 90 and 130 mm FL (fork length) were marked and randomly assigned to three 4-m3 cages (Cl, C2, C3) immersed in the lagoon. Another forty four adults between 170 and 210 mm FL were marked and kept in a 25-m3 cage (C4). After hatch- ing, juveniles were transferred to a 3-m3 concrete tank supplied with a constant flow of water from the lagoon. Fish were fed daily with formulated pellets. Oreochromis niloticus was reared at the Bouake station (Institut des Savanes). One-hundred and fifty-two adults of O. niloti- cus were marked and released in two 400-m2 ponds (Al and A2) that had been previously enriched with organic matter (density of 0.2 fish/m2). Juveniles were kept in two 50-m2 ponds (J1 and J2) after hatching. No food was sup- plied; fish were sustained by natural resources. Marking and sampling Juveniles were not marked, therefore the “date of mark- ing” actually refers to the date of birth. Adults caught in the field were marked by injecting tetracycline into the peritoneal cavity (50 mg/kg of live weight) and tagged with plastic T-bar anchor tags. Fish were measured (standard length at marking ( SLm , millimeters), weighed (grams), and sex was determined at the date of marking (Table 1). All O. niloticus adults and only S. melanotheron adults kept in cage C4 were tagged. After marking, both species were sampled in cages and ponds at monthly intervals. After capture, all individuals were measured (standard length at capture ( SLC , millimeters), weighed (grams), sexed and their otoliths (sagittae) removed. Randomly selected otoliths were prepared for analysis. Table 1 shows the dates of marking and sampling, otolith subsamples, and the number of days between marking and recapture for the two species. Otolith preparation Only the right otolith was prepared according to the tech- nique described by Secor et al. (1992). After testing all pos- sible planes of the section (i.e. sagittal, transverse, frontal), we chose the transverse section plane. Each otolith was then embedded in polyester resin before being sectioned transversally (with an Isomet saw) to avoid extra polish- ing and taking care to leave material on both sides of the core’s plane. The resulting section was attached to a glass slide with thermoplastic glue (CrystalBond), ground with wet sand paper (grit ranging from 400 to 1200 per paper), and polished (polishing cloth with alumina paste ranging from 3 to V3 pm) on one side until reaching the primor- dium. The block was then turned over, affixed again to a slide with the polished face down, and ground and pol- ished to remove extra material until the core area was reached. The thickness of the resulting sections ranged between 10 and 40 pm. Microincrement readability was improved by polishing the surfaces with 1/3 pm alumina paste. Otoliths with over-ground surfaces or with damage in the reading axis were discarded. Otolith interpretation and variables measured Terminology used in our study refers to that of Kalish et al. (1995). Microincrements were interpreted and counted as number of D-zones (reading) along the sulcus, chosen as the standard axis for reading (Fig. 1). Otoliths of adults were read under an epifluorescent microscope (Leica, 50W HBO lamp, 355-420 nm D filter) because the tetracycline deposit emits a yellow-green fluorescence under UVB at 390 nm. Microincrements were counted between the tet- racycline mark and the otolith edge on a monitor coupled with a video with 1250x magnification. In juvenile oto- liths, microincrements were counted from the primordium to the outer edge of the otolith under 400x and lOOOx mag- nifications on the monitor coupled with video. Each otolith was read twice by the same reader, first from the primordium to the edge and then back from the Panfili and Tomas: Validation of age estimation and back-calculation of fish length in tilapias 141 Table 1 Number of otoliths for each sampling date and number of days between marking and sampling (in brackets) for Sarotherodon melanotheron and Oreochromis niloticus marked in Layo and Bouake. 1 = adults tagged with anchor tags. Date of marking Species or birthdate Date of sampling Sarotherodon melanotheron Adults 22 Nov 93 22 Dec 93 20 Jan 94 25 Feb 94 31 Mar 94 22 Apr 94 Cage Cl 19 Nov 93 1 (34 d) 1 (64 d) 1 (93 d) 1(129 d) 1 (163 d) 1 (185 d) Cage C2 19 Nov 93 1 (34 d) 1 (64 d) 1 (93 d) 1 (129 d) 1 (163 d) 1 (185 d) Cage C3 19 Nov 93 1 (34 d) 1 (64 d) 1 (93 d) 1 (129 d) 1 (163 d) 1(185 d) Cage C4; 19 Nov 93 4 (34 d) 4 (64 d) 4 (93 d) 4(129 d) 4 (163 d) 4 (185 d) Juveniles 22 Dec 93 20 Jan /94 26 Feb 94 31 Mar 94 22 Apr 94 Tank 4 Nov 93 5 (48 d) 5 (77 d) 5 (114 d) 5 (147 d) 5 (169 d) Oreochromis niloticus Adults 3 Mar 94 1 Apr 94 Pond A F 26 Jan 94 15 (36 d) 15 (65 d) Pond A2; 26 Jan 94 15 (36 d) 15 (65 d) Juveniles Pond J 1 2 Feb 94 7 Mar 94 5 Apr 94 15 (33 d) 16 (62 d) Pond J2 14 Feb 94 17 Mar 94 15 Apr 94 15 (31 d) 16 (60 d) edge to the primordium following the same growth axis (Campana, 1992). If no significant difference was found between these two read- ings for the whole sample after a paired f-test, the mean was used to estimate the age (Cam- pana and Jones, 1992). To maintain unbiased readings, physical data (size, date of capture) were not given to the reader. Three variables were measured on each oto- lith: 1) the otolith diameter at capture ( ODc ) which corresponds to the maximum length on the anteroposterior axis of the otolith before sectioning; 2) the otolith radius at capture (Rc) which corresponds to the distance between the primordium and the edge along the sulcus ax- is; and 3) the otolith radius at marking (Rm) which corresponds to the distance between the primordium and the tetracycline mark along the sulcus axis for the adults marked (Fig. 1). Variables were measured with TNPC image processing software (Visilog software platform, Noesis, France). Validation of the time of deposition of microincrements The term validation refers here to the accuracy of the age estimation in tilapias by counting the number of micro- increments in the otoliths (Nmc). The number of days involved in the validation experiment ( D in days) refers to the microincrements deposited between hatching and the date of capture for juveniles and between the date of tetracycline labeling and the date of capture for adults. Pannella’s hypothesis (1971) of a daily deposition rate for microincrements was tested for the two development stages of both species. The accuracy of the method was tested by establishing the relation between the number of microincrements counted and the number of days involved in the experi- ment. A Student test on the linear regression was used to establish if the microincrements were deposited daily (slope equal to 1) and if the deposition started on the 142 Fishery Bulletin 99(1 ) first day (intercept equal to 0). The difference between the number of microincrements and the number of days of growth was plotted against time. Validation of back-calculation for Oreochromis niloticus Using tagged individuals, we assessed the validity of back- calculating fish length at marking. Several back-calcula- tion models are described in the literature and each one assumes a different relationship between fish growth and otolith growth (Francis, 1990). Francis (1990) recommends Whitney and Carlander’s model where a constant propor- tionality between fish growth and otolith growth through- out the life of the fish is assumed. Moreover, Francis ( 1990) recommends the regression of fish length against otolith length. Two approaches were considered in this study. If the relationship between fish length and otolith length was linear, SLc = a + b x Rc (1) GSL(%/d) = log (SLc ) - log( SLm ) I) x 100 GOL{% Id) log(fi„)-log(fl„)xl00 (5) (6) where SLc = standard length at capture; SLm - standard length observed at marking; Rc = otolith radius at capture; Rm = otolith radius at marking; and D ~ number of days of growth. The relationships between these specific growth rates and the differences between observed and back-calculated lengths at marking were established. Results Microincrement identification and tetracycline labeling with a body proportional hypothesis (BPH), then the back- calculation formula would be SL, i h a + bx Rm a + bx R xSL. (2) If the relationship between fish length and otolith length was nonlinear and for example multiplicative, Log(SLc) = c + d x LogiRc) (3) with a body proportional hypothesis (BPH), then the back- calculation formula would be Log ( SLmh ) - d x Log + Log(SLc ), (4) where SLr = standard length at capture; SLinb = back-calculated standard length at marking; Rc = otolith radius at capture; Rm = otolith radius at marking (tetracycline mark); and a, b, c, d = constants. The validation of the back-calculation model was carried out on otoliths of adult O. niloticus labeled with tetracy- cline. For each individual, fish length measured at mark- ing iSLm, millimeters) was compared with the individual fish length back-calculated at marking from the otolith ( SLinb , millimeters). The mean of the differences between measured and back-calculated fish lengths was compared by means of a Student test with a theoretical mean equal to 0. The specific growth rates of the fish length ( GSL in %/d ) or of the otolith radius length (GOL in %/d) were calculated following Ricker s formula (1975): Preparing tilapia otoliths for examination is time-consum- ing; it takes sixty to ninety minutes to prepare one otolith. A final polishing with 1/3 pm alumina powder is an impor- tant improvement for microstructure reading when deal- ing with thin sections ranging between 10 and 40 pm of thickness. The central core of the otolith of both species seems to correspond to the fusion of several primordia (up to six) even though it remains a small structure easily recognizable during the grinding process. Two accessory growth centers are visible on any transverse section of juvenile otoliths between the 13th and the 28th microin- crements. They are located on both sides of the core on the dorsal and ventral halves of the otolith, where they appear to control the growth of the otolith on the dorso- ventral axis. Microstructures are more clearly identified along the sulcus or along the dorsoventral axis as typically alternated L-zones and D-zones. Cross-checking and nar- rower microstructures are more common on the dorsoven- tral axis than in the sulcus region. As a result, we chose to interpret the microincrements along the sulcus axis (Fig. 1). Microincrements were counted from the hatching check which was clearly identifiable in the core area. For at least the first 30 microincrements the otolith grows predominantly along the dorsoventral axis and towards the external face. To avoid underestimating the number of microstructures during this first growth stage, the first 15 to 20 microincrements were counted along the core- ventral axis. Other microstructures along the ventral axis, interpreted as subdaily increments, rapidly increased in number and made the reading difficult. The shift to the sulcus axis was completed by following any conspicuous microincrement along the core-ventral axis to the sulcus region where the reading was finished (Fig. 1), provided accessory growth centers had not been encountered in this area. The tetracycline mark was present in all otoliths ex- amined, except one otolith of S. melanotheron sampled in April 1994. The mark was more intense on O. niloticus Panfili and Tomas: Validation of age estimation and back-calculation of fish length in tilapias 143 Table 2 Results of microincrement readings for adults and juveniles Sarotherodon melanotheron . D = number of days between marking (adults) or birth (juveniles) and capture; SD = standard deviation; Cl (95%) = confidence interval for mean at 95%; CV = coefficient of variation. Number of microincrements D (d) n Mean SD CV (%) Cl (95%) Adults 34 6 28.3 2.8 10.4 25.4-31.3 64 6 56.1 8.9 16.6 46.7-65.5 93 3 88.5 10.0 12.2 63.7-113.3 129 7 119.4 7.7 6.7 112.3-126.2 163 4 150.7 23.5 16.5 113.4-188.1 185 3 176.7 11.0 6.7 149.4-203.9 Juveniles 48 4 48.6 2.3 5.0 45.0-52.3 77 4 76.9 2.2 3.0 73.3-80.4 114 5 109.9 2.1 2.0 107.3-112.5 147 4 132.9 17.1 13.7 105.6-160.1 169 5 164.5 6.1 3.9 156.9-172.1 Table 3 Linear regressions between the number of microincrements iNinc) and the number of days between marking or birth and capture (Dl, D = a + b x Ninc, and Student tests for slope (6=1) and intercept (a=0) for Sarotherodon melanotheron. F = result of model ANOVA; ns = no significant difference (P>0.05); r2 = coefficient of determination. Intercept Slope Group n F r2 (%) a t a = 0 b t b = 1 Juveniles 22 524.7 96.3 4.348 0.88 ns 0.923 -1.91 ns Adults 29 658.4 96.1 -4.844 -1.11 ns 0.968 -0.85 ns Juveniles and adults 51 1176.3 96.0 -1.899 -0.57 ns 0.957 -1.54 ns otoliths than S. melanotheron even though the quantities injected in both species were identical (50 mg per kg of live weight). In all cases the tetracycline deposit coincid- ed with a check in the otolith structure, confirming that the marking is synonymous with stress to the fish. Other checks present along the sulcal axis interrupted the mi- crostructure deposition pattern without any regularity. The growth of the otoliths in the region between the tet- racycline deposit and the edge (i.e. between marking and capture) was always smaller for S. melanotheron than for O. niloticus for comparable sizes regardless of the date of capture. Validation of microincrement deposition in otoliths of Sarotherodon melanotheron Microincrement interpretation on adult S. melanotheron otoliths is difficult and requires a minimum magnification of lOOOx for the microincrements between the tetracycline mark and the outer edge. Results of otolith reading for adults and juveniles are summarized in Table 2. The mean of the two readings was used as the value of the micro- increment count of each otolith because the difference between the two readings was not significant in the whole sample (paired f-test, t=0.02, P>0.05). Results showed a tendency to underestimate true age, even though this underestimation was less pronounced in juvenile fish than in adults (Table 2). The mean underestimation was 8.1 d for adults and 4.4 d for juveniles; both means were sig- nificantly different from 0 (respectively f=4.19 and t= 2.36, P<0.05). The number of increments were plotted against the days of growth. An ANOVA conducted with the result- ing linear models (Table 3) with adults or juvendes, or both, showed a coefficient of determination (r2) signifi- cantly different from 0 (PcO.OOl). The slopes were not dif- ferent from 1 and the intercepts were not different from 0 (Table 3). Thus the microincrements counted on the trans- verse section of S. melanotheron otoliths can be considered structures that are deposited daily. The technique appears to accurately estimate the age of S. melanotheron in days. The dispersal of residuals in Figure 2 was more im- portant in large individuals. Underestimation slightly in- 144 Fishery Bulletin 99(1 ) Table 4 Results of microincrement readings for adults and juveniles Oreochromis niloticus. D = number of days between marking (adults) or birth (juveniles) and capture; SD = standard deviation; Cl (95%) = confidence interval for mean at 95%; CV = coefficient of variation. Number of microincrements D (d) n Mean SD CV (%) Cl (95%) Adults 36 25 31.1 2.5 8.2 30.0-32.1 65 27 59.7 2.6 4.4 58.7-60.8 Juveniles 31 12 31.9 2.5 8.2 30.2-32.1 33 9 30.5 1.3 4.5 29.5-31.5 60 9 59.4 3.7 6.4 56.6-62.3 62 13 60.4 2.3 3.9 59.0-61.8 50 O 40 ■ □ 30 ■ o 20 O ;o O □ ? 10 . O 8 o 8 o □ Q © □ O o □ 0 o D 18 a . 1 . ° □ □ o □ 1 ° B o ° □ □ 30 D 60 90 120 150 180 210 -10 o -20 0(d) Figure 2 Individual differences between the number of days of growth ( D ) and the number of microincrements counted on transverse otolith section (NinC) for adults (O) and juveniles O Sarotherodon melanotheron. creased with the duration of the experiment although the coefficient of variation did not in- crease with time (Table 2). Variability in preci- sion was nevertheless higher for adults than for juveniles. Figure 2 also shows that the dif- ference between the number of increments and the number of growth days was equal to zero for only a few individuals. Validation of microincrement deposition in otoliths of Oreochromis niloticus The distinction of microincrements was much greater in O. niloticus than in S. melanotheron otoliths resulting in an easier interpretation of microincrement deposition in O. niloticus. Results of microincrement counts for adults and juveniles of O. niloticus are summarized in Table 4. As for S. melanotheron , the two readings of each otolith were not significantly different across the whole sample; thus the mean of the two counts was used as the otolith microincrement value (paired £-test, £=0.85, P>0.05). Microincrement values in otoliths of adults after 36 and 65 d underestimated the true age of the fish by 4.9 and 5.2 d, respec- tively. Both values significantly differed from 0 (respectively £=9.78 and £=10.44, P<0.05) but were not significantly different from each other (P>0.05). Therefore the deviation between the true age and the estimated age was similar between 36 and 65 d and approximately equal to five days. In otoliths of juveniles, the mean of the differences be- tween the number of days and microincrements counted did not differ significantly from 0 in ponds J1 (0.9 d) and J2 (0.6 d) (£=1.18 and £=0.45 [P>0.05], respectively). For juveniles reared in pond Jl, age was underestimated af- ter 33 and 62 d of growth (Table 4), whereas ages of juve- niles in pond J2 (31 and 60 d of growth) were accurately estimated (Table 4). A one-level nested AN OVA was car- ried out to check whether the pond or the age, or both, had had an effect in the difference between the true age and the estimated age. Results showed a significant effect of the pond (Fa 39)=7.57, P< 0.05) but not of age (P(2 39)=1.13, P>0.05). Furthermore, a multiple rank test showed that no significant differences existed between the means at 31, 60, and 62 d and between the means at 33, 60, and 62 d (P>0.05), but that a significant difference existed between 31 and 33 d (P<0.05). Thus an effect of the pond on otolith deposition could not be confirmed. The relation between the number of microincrements and the number of days before capture was established to test the accuracy of the age estimation (Table 5). The re- sulting r2 was significantly different from 0 for adults or ju- veniles, or for both (P<0.001). The slopes of the model were not different from 1 (Table 5, P>0.05), which shows that the deposition rate of microincrements is daily. The num- Panfili and Tomas: Validation of age estimation and back-calculation of fish length in tilapias 145 Table 5 Linear regressions between the number of microincrements (Ninc) and the number of days between marking or birth and capture (D),Z) = a + b x Ninc and Student tests for slope (6=1) and intercept (a=0) for Oreochromis niloticus. F = result of model ANOVA; ns = no significant difference (P>0.05); r2 = coefficient of determination; s = significant difference (P<0.05). Intercept Slope Group n F r2 (%) a t a = 0 6 t 6 = 1 Juveniles 43 1122.6 96.5 0.293 0.21 ns 0.975 -0.85 ns Adults 52 1264.9 97.0 -4.522 -3.47 s 0.989 -0.45 ns Juveniles and adults 95 1696.5 94.8 -1.347 -1.12 ns 0.963 -1.59 ns ber of days of growth explains the num- ber of microincrements counted. Never- theless, the intercept differed from 0 for adults (P<0.05). It therefore seems that the deposition of new increments did not start on the first day after marking and that this difference (5 d) remained con- stant for one to two months of growth (Tables 4 and 5), suggesting that the in- crement technique is accurate for esti- mating the age of O. niloticus in days. Residual dispersal was similar for adults and juveniles and seems constant through time (coefficient of variation in Table 4 and Fig. 3). Age was especially overestimated in juvenile fish (Fig. 3). As for S. melanotheron, a difference be- tween the number of increments and the number of growth days equal to zero was very rare with O. niloticus otoliths. The trend in the deviation of the age estima- tion was the same over time for a given pond (Fig. 3). Validation of back-calculation of length in Oreochromis niloticus 14 -- 12 -- 10 -- 8 -- 6 -- Q 0 -2 -- -4 -- -6 -- -8 -- — I- 10 20 O □ □ □ □ □ , □ — h- 40 □ o o □ o o □ o □ o □ o o □ □ o □ o □ □ □ □ o □ □ -I ¥— B 1 □ o 50 D 70 □ □ □ o D(d) Figure 3 Individual differences between the number of days of growth (D) and the number of microincrements counted on transverse otolith section (IV ) for adults (O) and juveniles (□) Oreochromis niloticus. The relation between fish length and oto- lith length was determined by establishing the regression of fish standard length at capture on the otolith radius at capture (Table 6). Both the linear and the multiplica- tive models were tested by an ANOVA ( F calculated. Table 6) and had highly significant relationships (P<0.001). A comparison of the variances suggested that the coefficient of determination in the multiplicative model was signifi- cantly higher than that in the linear model (P<0.05). As a result, the regression used for the back-calculation of fish length was the multiplicative form. The observed disper- sion of residuals reinforced this choice. The observed low value of r2 for the linear regression was due to the importance of the dispersion of points around the model for adults (Fig. 4). Nevertheless, this dispersion was also observed in the relation between fish length and otolith diameter at capture (Fig. 4), prior to any otolith preparation. In this case the linear regression had a higher coefficient of determination (Table 6). The dispersion can be explained by the natural growth varia- tion which appears with age, and which is strong for this species. The back-calculated formula used to compare the ob- served length at marking and the back-calculated length with otolith transverse sections was therefore Log(SLmh) = 0.899278 x Log ' Rn + Log(SLi ). (7) Back-calculated lengths at marking were overestimated in the whole sample and this tendency did not depend on fish size at marking (Fig. 5). The mean of the differences 146 Fishery Bulletin 99(1) (mean=8.5 mm, standard deviation=8.7 mm) was signifi- cantly different from 0 (t=6.94, P< 0.05). As a result, back- calculation of fish length overestimated the length of fish that had grown for one or two months. The relation between the specific growth rate of the fish (Gsl) and the difference between back-calculated and ob- served length at marking was positive because overesti- mation increases with the specific growth rate, regardless of the duration of the experiment (Fig. 6). The actual under- or overestimation of back-calcu- lated length seems to depend on the coupling between the specific growth rates of the otolith and the fish (Fig. 7). An uncoupling between so- matic and otolith growth rates would explain the deviation of the back-calculated length from the measured value (Fig. 7). That is to say, if the specific growth rates of the fish and the oto- lith are identical, the back-calculation gives a very good approximation of the length at mark- ing (i.e. the difference is almost equal to zero), whereas if the otolith growth rate is higher than the fish growth rate, then the back-calculation underestimates the length at marking. Inverse- ly if the somatic growth rate is higher than the otolith growth rate, the back-calculation will overestimate the length at marking. Moreover, when the uncoupling between otolith and so- matic growth rates rises, the overestimation of the back-calculation increases (Fig. 7). Discussion Otolith preparation and interpretation of microincrements in tilapias Transversal sections (in contrast to sagit- tal or frontal sections) of otoliths are the clearest and most reliable way to inter- pret microincrements in tilapia otoliths. Although some authors have worked on sagittal sections (Fagade, 1980; Karakiri and Hammer, 1989) the concavoconvex shape of the otoliths in adults makes it very difficult to obtain a plane that would include both the core area and the otolith edge. Zhang and Runham (1992), who prepared transverse sections of adult O. niloticus otoliths with their histological technique (Zhang et al., 1991), obtained sections with clear microincrements. Rosa and Re (1985) reported that sagittae in Table 6 Relations between SL, (standard length at capture) and Rc (otolith radius at capture) or ODr (otolith diameter at capture) for Oreochromis niloticus. F is calculated from the ANOVA for testing the model, r 2 = coefficient of determination. Model n Regression F r2 (%) Linear 108 SLC = 17.13 + 130.31 x Rc 612.5 85.1 Multiplicative 108 Log(SLc) = 5.02 + 0.899 x Log(Rc ) 1117.3 91.3 Linear 119 SLc = -0.609 + 20.33 x OD( 4163.8 97.2 Panfili and Tomas: Validation of age estimation and back-calculation of fish length in tilapias 147 Figure 7 Relations between the specific growth rate (otolith length [thin curve] or standard length [bold curve] ) and the individual difference of the stan- dard length back-calculated at marking ( SLmb ) and the standard length observed at marking ( SLm ) for Oreochromis niloticus. juvenile Tilapia marine were too thick to allow observation of the microstructures without previous preparation and decided to work on lapilli. Despite the long and time-consuming process (about 1 h for preparing each trans- verse section of a sagittal, it appears to be the best way to observe otolith microincrements for both juvenile and adult tilapia. To avoid parallax when observing microstructures on thick preparations (Campana, 1992; Neilson, 1992), a thickness of 10 to 40 pm and a fine polishing of the sections are necessary. Zhang and Runham (1992) observed micro- increments of juvenile O. niloticus otoliths un- der a maximum magnification 400x with light microscopy, whereas Rosa and Re (1985) used magnifications ranging from 600x to 1250x while working on Tilapia mariae lapilli. Our results show that microincrement observation and interpretation require a minimum magni- fication of lOOOx for adult tilapias. These high magnifications with compound microscopes are required to observe microincrements in the sulcus area where they are found to have such a compressed arrangement that no space is left for subdaily increments (Zhang and Runham, 1992). Therefore, to interpret microincrements in tilapia otoliths accurately, we strongly rec- ommend preparing thin transverse sections (10-40 pm), polishing them finely (V3 mm), and observing them under high magnifications (minimum of lOOOx). In interpreting microstructures in tilapia otoliths, four types of problems were encoun- tered: 1) difficulty in interpreting microstruc- tures in the otolith region that correspond to first-growth stages; 2) difficulty in having to switch the reading axis (starting in the dorso- ventral area and finishing along the sulcus ar- ea); 3) difficulty in reading some zones; and 4) difficulty in identifying microstructures near the outer edge of the otolith. Microincrements around the hatching check are very faint and narrow and require high magnifications to be identified. Narrow increments were also re- ported on Tilapia mariae lapilli by Rosa and Re (1985). The presence of accessory growth centers on both sides of the core area in the dorsoventral plane of the otolith were also ob- served by Karakiri and Hammer (1989) and Zhang and Runham ( 1992) on O. niloticus otoliths. These authors esti- mated the date of formation of these accessory growth cen- ters to be between 21 and 30 days and between 16 and 28 days after hatching, respectively. Secondary growth cen- ters in Ivorian tilapia otoliths were located between the 13th and the 28th microincrements; therefore our results agree with observations made by these authors. It is like- ly that the presence of accessory growth centers repre- sents a shift in the growth of the otolith, meaning that growth along the dorsoventral axis is favored at this stage. The use of the dorsoventral axis to read microincrements may induce reader error. Ambiguities arise because of the numerous subdaily structures deposited during the fast growing period (Zhang and Runham, 1992); thus reading along the sulcus region is recommended. However, as the growth of the otolith along the sulcus is relatively indis- tinct during the first 15 to 20 microincrements, proper in- formation can only be gathered along the core-ventral ax- is. Certain regions in otoliths, especially near the edge, have been difficult to read for numerous other species (Campana, 1992). Tilapia otoliths also exhibit unreadable 148 Fishery Bulletin 99(1 ) parts as reported and observed by Zhang and Runham (1992). Otoliths of both species are difficult to interpret and attention should be paid to the need for trained read- ers capable of properly interpreting microincrements in ti- lapia otoliths. Validation of microincrement deposition in tilapias Tetracycline remains a universal marker for otoliths (Beamish and McFarlane, 1987; Brothers, 1990; Geffen, 1992). In our study only one S. melanotheron did not reveal any tetracycline deposit on its otolith. Because the fish is necessarily handled, injecting tetracycline induces a strong calciotraumatic effect on the otolith (Meunier and Boivin, 1978; Pannella, 1980; Campana and Neilson, 1985; Panfili and Ximenes, 1992). As a result, the mark is easily recognizable in the otolith and corresponds to a check in the otolith structure. This check probably reflects a ces- sation in the growth of the otolith that could last several days. In our study, underestimation of age was constant (equal to 5 days) for adult O. niloticus which indicates that the growth of the otolith was stopped with the tetracy- cline injection and resumed five days later. When using microincrement counts, the effect of marking should be considered, but it is difficult to estimate the time elapsed between injection and resumed growth of the otolith. We estimated the age of juvenile and adult O. niloticus and S. melanotheron accurately using the same microin- crement examination and interpretation on thin otolith transverse preparations. Tanaka et al. (1981) and Zhang and Runham ( 1992) had similar results on transverse sec- tions of juvenile O. niloticus otoliths. The former observed otoliths with scanning electron microscopy and the latter observed stained otoliths. Karakiri and Hammer (1989) also reported daily increment deposition on sagittal sec- tions of Oreochromis aureus otoliths observed with scan- ning electron microscopy. Our technique of preparation is the only one that has been validated for two different spe- cies and for several developmental stages. Precision in age estimation with otolith microincre- ments was calculated for both species and showed that the error ranges between 4.4 d in juveniles and 8.1 d in adults for S. melanotheron , and between 0.9 d in juveniles and 5.1 d in adults for O. niloticus. Validation of back-calculation and influence of individual growth rates We validated the back-calculation model with a body pro- portional hypothesis (BPH) developed by Whitney and Carlander (1956, in Francis, 1990) and commonly rec- ommended and used in the literature (Francis, 1990; Smedstad and Holm, 1996; Horppila and Nyberg, 1999). Back-calculation models rely on the assumption that oto- lith size and fish size are related and that a relation between them can be established. It is assumed that 1) the frequency of formation of each structure is constant and 2) the width of each increment is proportional to the growth of the fish (Campana and Jones, 1992). Cau- tion should be taken when calculating the relationship because, as stated by Francis (1990) and Campana and Jones (1992), if the aim is to backcalculate a mean fish length from any otolith dimension, the resulting regres- sion must have fish length as a dependent variable and otolith dimensions as independent variables. It is there- fore very important to set up the most suitable relation- ship relating fish length and otolith length. Some works have shown that many factors influence this relation- ship. Wright et al. ( 1990) reported that the relation of fish length to otolith length was linear for smolts belonging to the high mode (fast growth) and curvilinear for smolts belonging to the low mode (slow growth). The relation- ship is also affected by food supply (Rice et al., 1985) or seasonal changes (Thomas, 1983). Furthermore, Reznick et al. (1989) showed that slow growing guppies ( Poecilia reticulata) have larger otoliths than fast growing gup- pies of similar lengths, even though both groups of fish shared the same genetic background, had the same feed- ing schedule, and were reared under the same conditions. In this context, the use of the model of Whitney and Car- lander is particularly suitable because it assumes that if a fish is 10% smaller than the mean length of the popula- tion for a given otolith size, this deviation will be constant throughout the life of the fish. We chose the curvilinear model for the relationship between fish length and oto- lith length because it fitted the existing data better. Brad- ford and Geen ( 1987) also found no significant difference between the curvilinear model and the linear model and therefore used the former because it adjusted total data better. Smedstad and Holm ( 1996) compared several back- calculation formulae for cod otoliths and concluded that the nonlinear one was better. These relationships seem to depend on the axis of the otolith chosen for back-calcula- tion. Back-calculation results could have been less vari- able if the diameter of the whole otolith had been used in the relationship with fish length instead of the radius on the transverse otolith section (Fig. 4; Table 6). Unfortu- nately, because it was impossible to interpret the microin- crements along the anteroposterior axis (diameter), that axis was discarded for back-calculation. A prerequisite of back-calculation is the assumption that the frequency of formation of microincrements is constant along the axis of analysis (Campana and Jones, 1992) and that assump- tion could not be made in the anteroposterior axis otoliths of our study. Our study shows that back-calculated fish lengths are greater than measured fish lengths at marking among fish that have grown between one and two months. These re- sults agree with those obtained by Rijnsdorp and Visser1 on plaice ( Pleuronectes platessa ) grown for 19 months, al- though the back-calculation model used by these authors was the Dahl-Lea model (1920, in Francis, 1990). Two main points can be related to the observed overestima- tion. First, it is related to fish growth: the larger the fish growth, the larger the overestimation, although previous authors found an inverse relationship. This overestima- 1 Rijnsdorp, A. D., and T. A. M. Visser. 1987. Tetracycline label- ling of otoliths in plaice. ICES, C.M. 1987/G:33, 12 p. Panfili and Tomas: Validation of age estimation and back-calculation of fish length in tilapias 149 tion may be result of back-calculation model being differ- ent or the marking-recapture interval being longer. Sec- ond, rather than fish growth rates alone, it appears that the coupling between fish growth and otolith growth plays a major role in explaining the observed overestimation of fish length. As shown in our study, underestimation of back-calculated fish length corresponds more to otolith growth rates compared with fish growth rates, and vice versa. These aspects must be discussed in light of the re- lationship between fish size and otolith size, the influence of fish growth on the overestimation of fish length, and the evidence of uncoupling between fish and otolith growth rates (Mosegaard, et ah, 1988; Reznick, et al., 1989; Secor and Dean, 1989). Figure 6 shows that the higher the fish growth rates, the higher the deviation of the back-calcu- lated fish length, implying that the faster a fish grows the further the fish is from the model. Secor and Dean (1989) considered that the ratio of otolith size to fish size increas- es in starved fish, as well as in fish with slow growth rates. Thomas (1983) considered this relationship to correct the underestimation resulting from using the Lee back-cal- culation model. Because fish growth rates have an influ- ence on the observed estimations, they raise the question of what was the influence of fish growth rates on otolith growth rates along the sulcus axis in our study? Geffen ( 1992) proposed establishing the relationship between fish and otolith growth rates prior to back-calculation. Unfor- tunately this method is difficult to achieve in the field be- cause it would mean marking fish from any studied popu- lations, which unfortunately rarely happens. Finally, Bradford and Geen (1987) advised caution when back-calculating fish length because otolith growth seems to be more conservative than fish growth. In our study, this assumption takes force because tilapias used in the experiments were starved before the beginning of the rearing experiments and experienced high growth rates after placement in ponds. Otolith growth rates fol- lowed fish growth rates within a certain range. When fish growth decreased below a certain limit, the otolith contin- ued to grow. When fish growth increased, otolith growth also increased to a certain extent. This finding confirmed that the rate of growth in otoliths is conservative com- pared with the rate of somatic growth. Furthermore, the otolith represents an essential part of the equilibrium and sensory system of fish and thus cannot follow only fish growth rates. As a result, high growth rates in fish will imply a bigger dispersion of the data, or heterocedas- ticity around the relation of fish length to otolith length, which is observed here for larger individuals. The uncou- pling of fish and otolith growth thus explains the differ- ence between back-calculated and measured fish lengths at marking. Therefore caution should be taken when es- tablishing the relationship of fish length to otolith length by using a representative sample of the individuals in their natural environment. In conclusion, the model de- veloped by Whitney and Carlander represents a valid model for studies in the field because it considers individ- ual variability in the relationship of fish length to otolith length but further work is needed to validate the use of other back-calculation models. Acknowledgments We specially thank Saurin Hem (IRD, Montpellier, France) and Philippe Cecchi (IRD, Bouake, Cote d’Ivoire) for their valuable help in marking and rearing experiments. We ack- nowledge the CRO (Centre de Recherches Oceanologiques, Abidjan, Cote d’Ivoire) and the IDESSAGnstitut des Savanes, Bouake, Cote d’Ivoire) for their logistic support during the rearing experiments. Thanks are also due to Jacques Baron (Universite de Bretagne Occidentale, Brest, France) for his assistance and guidance in the statistical treatments. Marcus Belchier helped with the English text. We thank him espe- cially. We also appreciate the comments of one anonymous reviewer which improved the manuscript. Literature cited Beamish, R. J., and G. A. McFarlane. 1987. Current trends in age determination methodology. In The age and growth of fish (R. C. Summerfelt and G. E. Hall, eds.), p. 15-42. Iowa State Univ. Press, Ames, IA. Bradford, M. J., and G. H. Geen. 1987. Size and growth of juvenile Chinook salmon back-cal- culated from otolith growth increments. In The age and growth of fish (R. C. Summerfelt, and G. E. Hall, eds.), p. 453-461. Iowa State Univ. Press, Ames, IA. Brothers, E. B. 1990. Otolith marking. Am. Fish. Soc. Symp. 7:183-202. Campana, S. E. 1992. Measurement and interpretation of the microstruc- ture of fish otoliths. In Otolith microstructure examina- tion and analysis (D. K. Stevenson and S. E. Campana, eds.), p. 59-71. Can. Spec. Publ. Fish. Aquat. Sci. 117. Campana, S. E., and C. M. Jones. 1992. Analysis of otolith microstructure data. In Otolith microstructure examination and analysis (D. K. Steven- son, and S. E. Campana, eds.), p. 73-100. Can. Spec. Publ. Fish. Aquat. Sci. 117. Campana, S. E., and J. D. Neilson. 1985. Microstructure of fish otoliths. Can. J. Fish. Aquat. Sci. 42:1014-1032. Duponchelle, F, and J. Panfili. 1998. Variations in age and size at maturity of female Nile tilapia, Oreochromis niloticus , populations from man-made lakes of Cote d’Ivoire. Environ. Biol. Fishes 52:453-465. Durand, J. R., and D. Guiral. 1994. Hydroclimat et hydrochimie. In Environnement et ressources aquatiques de Cote d’Ivoire: les milieux lagun- aires (J. R. Durand, P. Dufour, D. Guiral, and S. G. F. Zabi, eds.), p. 59-90. ORSTOM (Office de la Recherche Scienti- fique et Technique d’Outre-MerlEditions, Paris. Durand, J. R., and M. Skubich. 1982. Les lagunes ivoiriennes. Aquaculture 27:21 1-250. Eyeson, K. N. 1983. Stunting and reproduction in pond-reared Sarother- odon melanotheron. Aquaculture 31:257-267. Fagade, S. O. 1980. The structure of the otoliths of Tilapia guineensis (Dumeril) and their use in age determination. Hvdrobio- logia 69:169-173. Francis, R. I. C .C. 1990. Back-calculation of fish length: a critical review. J. Fish Biol. 36:883-902. 150 Fishery Bulletin 99(1 ) Geffen, A. J. 1992. Validation of otolith increment deposition rate. In Otolith microstructure examination and analysis (D. K. Stevenson and S. E. Campana, eds.), p. 101-113. Can. Spec. Publ. Fish. Aquatic Sci. 117. Horppila, J., and K. Nyberg. 1999. The validity of different methods in the backcalcula- tion of the lengths of roach — a comparison between scales and cleithra. J. Fish Biol. 54:489-498. Jones, C. M. 1992. Development and application of the otolith incre- ment technique. In Otolith microstructure examination and analysis (D. K. Stevenson, and S. E. Campana, eds.), p. 1—11. Canadian Special Publication of Fisheries and Aquatic Sciences 117. Kalish, J. M., R .J. Beamish, E. B. Brothers, J. M. Casselman, C. Francis, H. Mosegaard, J. Panfili, E. D. Prince, R. E. Thresher, C. A. Wilson, and P. J. Wright. 1995. Glossary. In Recent developments in fish otolith research (D. H. Secor, J. M. Dean, and S.E. Campana, eds.), p. 723-729. Belle W. Baruch Library in Marine Science, vol. 19, Univ. South Carolina Press, Columbia, SC. Karakiri, M., and C. Hammer. 1989. Preliminary notes on the formation of daily incre- ments in otoliths of Oreochromis aureus. J. Appl. Ichthyol. 5:53-60. Folding, J. 1993. Population dynamics and life-history styles of Nile tilapia, Oreochromis niloticus , in Ferguson’s Gulf, lake Tur- kana, Kenya. Environ. Biol. Fishes 37:25-46. Legendre, M., and J. M. Ecoutin. 1989. Suitability of brackish water tilapia species from the Ivory Coast for lagoon aquaculture. I — Reproduction. Aquat. Liv. Res. 2:71-79. Lowe-McConnell, R. H. 1982. Tilapias in fish communities. In The biology and culture of tilapias, (R. S. V. Pullin and R. H. Lowe-McCon- nell, eds.), p. 83-113. ICLARM (International Center for Living Aquatic Resources Management) Conference Pro- ceedings, vol. 7, Manila. Meunier, F. 1974. La technique de marquage vital des tissus squelet- tiques des poissons. Bull. Fra. Pise. 255:51-57. Meunier, F. J., and G. Boivin. 1978. Action de la fluoresceine, de l’alizarine, du bleu de cal- ceine et de diverses doses de tetracycline sur la croissance de la truite et de la carpe. Ann. Biol. Anim. Bioch. Bio- phvs. 18:1293-1308. Mosegaard, H . H. Svedang, and K. Taberman. 1988. Uncoupling of somatic and otolith growth rates in arctic char ( Salvelinus alpinus ) as an effect of differences in tem- perature response. Can. J. Fish. Aquat. Sci. 45:1514-1524. Neilson, J. D. 1992. Sources of error in otolith microstructure examina- tion. In Otolith microstructure examination and analysis (D. K. Stevenson and S. E. Campana, eds.), p. 115-126. Can. Spec. Publ. Fish. Aquat. Sci. 117. Panfili, J., and M. C. Ximenes. 1992. Measurements on ground or sectioned otoliths: possi- bilities of bias. J. Fish Biol. 41:201-207. Pannella, G. 1971. Fish otoliths: daily growth layers and periodical pat- terns. Science (Washington D.C.) 173:1124-1127. 1980. Growth pattern in fish sagittae. In Skeletal growth of aquatic organisms: biological records of environmental change (D. C. Rhoads and R. A. Lutz, eds.), p. 519-560. Plenum Press, New-York and London. Pullin, R .S. V., T. Bhukaswan, K. Tonguthai, and J. L. MacLean. 1988. The second international symposium on tilapia in aquaculture. ICLARM Conference Proceedings 15, Bang- kok and Manila, 623 p. Reznick, D., E. Lindbeck, and H. Bryga. 1989. Slower growth results in larger otoliths: an exper- imental test with guppies (Poecilia reticulata). Can. J. Fish. Aquat. Sci. 46:108-112. Rice, J. A., L. B. Crowder, and F. P. Binkowski. 1985. Evaluating otolith analysis for bloater Coregonus hoyi: do otoliths ring true? Trans. Am. Fish. Soc. 114:532-539. Ricker, W. E. 1975. Computation and interpretation of biological statis- tics of fish populations. Fish. Res. Board Can. Bull. 191, 409 p. Rosa, H. C., and P. Re. 1985. Influence of exogenous factors on the formation of daily microgrowth increments in otoliths of Tilapia mariae (Boulenger, 1899) juveniles. Cybium 9:341-357. Secor, D. H., and J. M. Dean. 1989. Somatic growth effects on the otolith-fish size rela- tionship in young pond-reared striped bass, Morone saxati- lis. Can. J. Fish. Aquat. Sci. 46:113-121. Secor, D. H., J. M. Dean, and E. H. Laban. 1992. Otolith removal and preparation for microstructural examination. In Otolith microstructure examination and analysis (D. K. Stevenson and S. E. Campana, eds.), p. 19-57. Can. Spec. Publ. Fish. Aquat. Sci. 117. Smedstad, O. M., and J. C. Holm. 1996. Validation of back-calculation formulae for cod oto- liths. J. Fish Biol. 49:973-985. Tanaka, K., Y. Mugiya, and J. Yamada. 1981. Effects of photoperiod and feeding on daily growth patterns in otoliths ofjuvenile Tilapia nilotica. Fish. Bull. 79:459-466. Thomas, R. M. 1983. Seasonal variation in the relationship between oto- lith radius and fish length in the pilchard off south west Africa. S.-Afr. J. Mar. Sci. 1:133-138. Weber, D., and G. J. Rigway. 1967. Marking Pacific salmon with tetracycline antibiotics. J. Fish. Res. Board Can. 24:849-865. Wootton, R. J. 1984. Introduction: strategies and tactics in fish reproduc- tion. In Fish reproduction: strategies and tactics (G. W. Potts and R. J. Wootton, eds.), p. 1-12. Academic Press, London. 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. Zhang, Z., and N. W. Runham. 1992. Otolith microstructure pattern in Oreochromis niloti- cus ( L. ). J. Fish Biol. 40:325-332. Zhang, Z., N. W. Runham, and T. J. Pitcher. 1991. A new technique for preparing fish otoliths for examination of daily growth increments. J. Fish Biol. 38:313-315. 151 Bayesian methods for analysis of stock mixtures from genetic characters Abstract— An implementation of Bay- esian methods to assess general stock mixtures is described. An informative prior for genetic characters of the sep- arate stocks in a mixture is derived from baseline samples. A neutral, low- information prior is used for the stock proportions in the mixture. A Gibbs sampler — the data augmentation algo- rithm— is used to alternately generate samples from the posterior distribution for the genetic parameters of the sepa- rate stocks and for the stock proportions in the mixture. The posterior distribu- tion incorporates the information about genetic characters in the baseline sam- ples, including relatedness of stocks, with that in the stock-mixture sample to better estimate genotypic composi- tion of the separate stocks. Advantages over usual likelihood methods include greater realism in model assumptions, better flexibility in applications, espe- cially those with missing data, and consequent improved estimation of stock-mixture proportions from the con- tributing stocks. Two challenging appli- cations illustrate the technique and its advantages. Manuscript accepted 13 September 2000. Fish. Bull. 99:151-167 (2001). Jerome Pella Auke Bay Laboratory Alaska Fisheries Science Center National Marine Fisheries Service, NOAA 1 1305 Glacier Hwy. Juneau, Alaska 99801-8626 E-mail address: Jerry.Pella@noaa.gov Michele Masuda Auke Bay Laboratory Alaska Fisheries Science Center National Marine Fisheries Service, NOAA 11305 Glacier Hwy. Juneau, Alaska 99801-8626 Fisheries that exploit mixed stocks are very common, and their management oftentimes requires assessment of com- position of the mixed catches (Begg et ah, 1999). Multilocus genotypes of fish are a natural tag by which to infer their origins. The unknown proportions from stocks comprising a stock mix- ture, or its stock composition, can be estimated from genotype counts in a random sample of the stock-mixture individuals if relative frequencies (RFs) of the genotypes vary among the con- tributing stocks. Larger differences in genotypic RFs among stocks result in more accurate and precise stock composition estimates. The conditional maximum likelihood (CML) method (Fournier et ah, 1984; Millar, 1987; Pella and Milner, 1987) has most commonly been used for stock-mixture analysis. Baseline samples drawn from the sepa- rate contributors are used in estimating the RFs of the observed stock-mixture genotypes in each stock. The CML stock composition estimate maximizes a like- lihood function of the stock-mixture genotypes as if their RFs in the base- line stocks were known without error. The baseline multilocus genotype RFs determine the outcome of a stock-mix- ture analysis. Larger errors in these estimated RFs result in larger stock composition errors. Usually the vari- ation in CML stock composition esti- mates from baseline and stock-mixture sampling is evaluated by the bootstrap method. The estimation of the baseline mul- tilocus genotype RFs depends on the mode of inheritance of the observed markers. Among molecular markers de- veloped for fish, allozymes, mitochondri- al DNA (mtDNA), minisatellite DNA, and microsatellite DNA are widely known for their utility in stock-mixture analysis. For mtDNA, the entire hap- lotype passes as a unit from female to offspring and the baseline multilocus haplotype RFs are estimated directly by their observed RFs in the baseline samples. For allozymes, minisatellite DNA, and microsatellite DNA, the mul- tilocus genotypes pass from parents to offspring under the usual rules of dip- loid inheritance. The expected multilo- cus genotype RFs for diploids equal the products of the genotypic RFs at the in- dividual loci (or subsets of them) that pass independently from parents to off- spring. In the special case when Hardy- Weinberg equilibrium holds at a locus, its expected genotypic RFs are deter- mined by its allele RFs; the homozy- gote RFs equal the squares of their allele RFs and the heterozygote RFs equal twice the product of their allele RFs. To compute the estimated base- line multilocus genotype RFs for dip- loids, observed RFs of alleles or geno- types in the baseline samples replace corresponding unknown RFs. The few- 152 Fishery Bulletin 99(1 ) er the number of unknown RFs that need to be estimated from the available baseline samples, the more reliable the estimated baseline multilocus genotype RFs become (Al- tham, 1984). Therefore, under Altham’s principle, allele RFs are to be estimated if Hardy- Weinberg equilibrium holds, and genotype RFs at the locus otherwise. If alleles among a linked subset of loci are not inherited indepen- dently, the multilocus genotype RFs for the subset have to be estimated directly from their observed RFs in the base- line samples. The search for genetic variation by which to distinguish among populations of fish and other marine organisms has provided an embarrassment of riches. The numbers of mtDNA haplotypes (Epifanio et al., 1995; Bowen et al., 1996; Rosel et al., 1999) and alleles at minisatellite and microsatellite loci (O’Connell and Wright, 1997) can often be so large that reliable estimation of their RFs in stocks is a concern. Baseline sample sizes are usually limited, and the relative precision of estimated RFs for the numer- ous haplotypes, alleles, or genotypes (HAGs) declines with the magnitude of their RFs. Resulting stock composition estimates for the stock mixture may suffer. Grouping or binning of HAGs may help to control variation in estima- tion of their baseline RFs (O’Connell and Wright, 1997). However, for stock-mixture analysis, the practical details of grouping so as to balance the loss of information about the mixture against the gain in precision of the baseline RFs are unresolved. Bayesian methods developed for es- timating allele RFs at a locus (sec. 3.7 of Lange, 1997) and for estimating cell probabilities in contingency tables (Bishop et al., 1975; Sutherland et al., 1975; Leonard, 1977) offer another attack on the problem. The Bayesian methods are applied later in estimating the RFs by using genetic similarities of stocks. Conditional maximum likelihood does not use the infor- mation in the stock-mixture sample to improve the esti- mates of baseline multilocus genotype RFs, and the omis- sion becomes ever more meaningful with the accumulation of mixture individuals from a series of analyses performed on stock mixtures of the same baseline populations. The unconditional maximum likelihood (UML) (Pella and Mil- ner, 1987) or unconditional least squares (ULS) (Xu et al., 1994) methods have been suggested to remedy this short- coming for analysis of a single mixture. For either ap- proach, estimates are provided both for the stock propor- tions and baseline genotypic RFs by optimizing a criterion of fit to counts in both the baseline and stock-mixture sam- ples. However, the fitting criteria may have local optima (Smouse et al., 1990), and effective search for the global optimum and corresponding estimates from both methods is unresolved. A practical compromise for either method is to find a particular local optimum by starting the search from the CML estimate of stock proportions and stock ge- notypic RFs, the latter evaluated from the baseline sam- ples alone. None of the past approaches — CML, UML, or ULS — makes use of the genetic similarities among stocks to es- timate the relative frequencies of haplotypes, alleles, or genotypes more accurately in the separate stocks. Com- mon HAGs are shared universally among stocks; HAGs with moderate RFs are shared at least regionally, and rare HAGs occur only sporadically. Instead, the similarities in HAG RFs are viewed solely as limiting success in distin- guishing the origins of the stock-mixture individuals. Im- proved estimates of HAG RFs, to replace simple observed values, would generally benefit accuracy and precision of stock composition estimation, especially as the number of rare or uncommon HAGs increases (e.g. Xu et al., 1994). Es- timation of RFs for rare HAGs in separate stocks from base- line samples is especially problematic. Even when present in a population, they may well be absent from the baseline sample. The Bayesian proposal for stock-mixture analysis will shrink the observed baseline HAG RFs of individual stocks toward better-established grand, regional, or group means in order to control HAG RF estimation error. All past approaches — CML, UML, or ULS — produce es- timates of stock proportions that become increasingly bi- ased as the true stock-mixture proportions become more uneven (Pella and Milner, 1987; Xu et ah, 1994). Contribu- tions from abundant stocks are underestimated and those from less common or even absent stocks are overestimat- ed. No effective general solution for this bias has been pro- posed. The Bayesian proposal results in a probability dis- tribution for the stock composition estimates, the location of which can be characterized by various measures, such as the mean, median, and mode, which differ in their bias when viewed as potential point estimators. Finally, the previous estimation methods appear limit- ed in capacity to attack practical problems that fail to fit the standard mold of a sampled stock mixture and com- plete baseline. In particular, missing information for en- tire stocks from the baseline is very difficult to accommo- date (Smouse et al., 1990). Despite their availability for a decade or more, nothing has been accomplished using these methods to incorporate genetic similarities of base- line stocks to deal more realistically with missing data. The Bayesian proposal will initially fill in missing base- line HAG RFs with appropriate grand, regional, or group means, proxies that are revised later during analysis of the stock-mixture sample. Bayes methods have the potential to correct for these shortcomings better than the likelihood or least squares methods. In our study we describe the rationale for this new approach to stock-mixture analysis, develop the sta- tistical models, and outline the numerical algorithms by which to quantify uncertainty in stock proportions of the mixture as well as in the baseline HAG RFs in the separate contributing stocks. Software developed for per- forming the computations and summarizing results is available at our anonymous ftp site, with address ftp:/ / wwwabl.afsc.noaa.gov / sida / mixture-analysis / bayes. Two applications with special difficulties are used as illustra- tions. First, a winter stock mixture thought to be composed of four Northwest Atlantic harbor porpoise ( Phocoena pho- coena) populations is assessed. These porpoise populations are characterized by mtDNA haplotypes, the number of which exceeds baseline and stock-mixture sample sizes. Second, a Southeast Alaska steelhead trout (Oncorhynchus my kiss) stock mixture is resolved to two populations, only one of which could be sampled separately. Allozymes, mi- Pella and Masuda: Bayesian methods for analysis of stock mixtures from genetic markers 153 crosatellites, and mtDNA were available for independent and confirming assessments of stock-mixture proportions from the two populations. Methods The premise of the Bayes method for estimating an un- known quantity, 6, is that some information about 6 is available before sampling begins. This information is in the form of a prior probability density, JiiO). After sam- pling, the new data obtained, Y , are used to revise the prior to the posterior probability density for the unknown, M6 \ Y). The posterior is obtained by application of Bayes’s theorem, which states that the posterior is proportional to the product of the prior and the likelihood of the sample, 7riF| 6), viz. 7r(0 1 Y ) - k ''/r(0)/r(Y | 0), (1) where k = f 7T(0)/r( Y | O)c/0. 6 Once the posterior for 6 is known, a variety of point esti- mates— mode, median, or mean — as well as the Bayesian posterior probability interval (the apparent counterpart of frequentist confidence intervals, but which actually is a direct probability statement about the unknown) for 9 can be derived from it. In stock-mixture analysis, the un- knowns separate into two evident blocks [9 = (p,Q)]: 1) the stock proportions of the mixture, p, and 2) the param- eters, Q , needed to determine the genetic composition — stock-mixture haplotype or multilocus genotype RFs — of the baseline stocks. For haploids, Q represents the base- line haplotype RFs. For diploids, Q represents the array of baseline allele and genotype RFs that are needed under Altham’s principle to compute the stock-mixture genotype RFs in the baseline stocks. The new information comes from the baseline and stock-mixture samples for which the likelihood functions are unchanged from earlier likeli- hood methods. The stock-mixture sample provides counts of haplotypes or multilocus genotypes, and the baseline samples provide counts from the separate stocks of the haplotypes or alleles and genotypes at the loci comprising the mixture multilocus genotypes. The standard stock-mixture analysis for complete base- line and stock-mixture samples by Bayes methods will be outlined here, with details given in following sections. Ex- tension to nonstandard applications will be indicated by example. First, a prior for 9 = (p,Q ) is developed, which is a product of block priors for its components,/) and Q The prior proposed for p, which will be called “uninformative,” allows any substantive stock-mixture sample information regarding p to overwhelm that from the prior. The prior for Q , used to analyze the stock-mixture sample, is infor- mative and will be derived from the baseline samples to quantify uncertainty in the genotypic composition of the contributing stocks. Second, the standard likelihood func- tion for the haplotype or multi locus genotype counts seen in the stock-mixture sample is described. Third, and last, the data augmentation algorithm, a Gibbs sampler, is used to alternately generate a sequence of samples from the posterior distributions for p and Q. The stock identities of the mixture-sample individuals are reassigned at each sampling cycle by using a chance mechanism that reflects their uncertainty. The stock identities simplify greatly the revision of the prior distributions for p and Q to account for the stock-mixture sample information; just as with the baseline samples, counts of mixture individuals and their HAGs by stock are available at each cycle. Assignment of individuals to stock origin contrasts with their fractional allocation by the CML method (Pella and Milner, 1987). These samples from the posterior distributions are used to quantify the final uncertainty in p and Q after observing the stock-mixture sample. Prior for stock-mixture proportions, nip) The prior for p can incorporate information about the stock-mixture composition other than that in the stock- mixture sample if such is available. More commonly, how- ever, such information is either unavailable or else the researcher prefers to let that of the stock-mixture sample dominate, just as it does with the earlier likelihood or least squares methods. Therefore, the prior proposed will be restricted, providing no useful information about the stock-mixture composition. Such an uninformative prior for the stock proportions of a c-stock mixture must be defined over the stock composition simplex, S(p) = Ip: 0 < p, < 1, ^ p, = 1 j, and have negligible effect on the posterior distribution. The Dirichlet probability density can accommodate these requirements, and it is natural to use it as a prior with compositional count data both for computational conve- nience and for its interpretation as additional data. Prior draws of p from the Dirichlet probability density, a , > 0,i - l,...,c, have means, variances, and covariances given by E(p;) = a , / a0, var( p, ) = «,(«„ - or, )/ («,;(«„ + D), cov(p,,p;,) = -aial. /(a'g(a0 + - 1,2,... ,c, and (3) «n = /= 1 154 Fishery Bulletin 99(1 ) If a prior draw of p ~ D(a1,a2, . . . , ac) (the notation “x ~ f" means “x is distributed as the probability density or prob- ability function f”) was obtained for the stock proportions of a stock mixture, and then a stock-mixture sample of size M was drawn such that the individuals could be correctly identified to stock origin, their counts, Z=(zvz2, . . . , zc), would have a conditional multinomial distribution, M' 7t(Z | p,M) = - Pi Pt— Pc, y I? I . . . y i *l'*2m *cm or Z\p ~ MultiM , p ). The posterior for p, given Z, would be the Dirichlet (computational convenience), p | Z ~D(z1 + av . . . , zc + ac). Notice that the prior parameters enter the posterior density in parallel with the sample counts and therefore could be viewed as counts obtained before the stock mixture was sampled (additional data) (sec. 3.5, Gelman et al., 1995). In fact, the mixture individuals are identified to stock origin (with unavoidable random error) during each cycle of the data augmentation algorithm later when samples are generated from the posterior. With the stock origins identified at a cycle, the uncertainty in p is described by the Dirichlet posterior with parameters equal to the sums of stock counts and prior parameters (z, +a-). With equal values summing to 1 assigned to its param- eters or “prior counts,” cq = a2 - . . . = ac - c-1, the Dirichlet prior meets our initial requirements. Specifically, the den- sity is defined over the stock composition simplex, and the additional data, which is neutral in the sense of favoring equal stock proportions (mean stock proportions are c_/), would be equivalent to adding just a single individual to the stock-mixture sample. Means, variances, and covari- ances (substitute zi + c] for a, in Eq. 3) of the resulting posterior distribution of p | Z, D(zl + c-1, . . . , zc + c_1), ap- proximate closely with increase of stock-mixture sample size, the observed stock proportions, their estimated vari- ances, and their estimated covariances, respectively, from standard frequentist analysis of the multinomial sample, Z. Therefore, given the stock assignments of the mixture individuals, the posterior distribution for p will be a rea- sonable description of its uncertainty for both Bayesian and frequentist statisticians. Prior for genetic parameters given baseline samples, n(Q\Y) The genetic compositions — haplotype or multilocus geno- type RFs — of the separate stocks are determined by their RFs of haplotypes, alleles, or genotypes, Q An estimate of Q from the baseline samples must be used in place of the unknown Q to estimate the stock genetic compositions. When baseline samples are large, the observed and unbi- ased value of Q, together with measures of precision, may be sufficient to anchor the stock-mixture analysis. Com- monly, baseline sample sizes are more limited and some tradeoff between bias and precision (sec 1.4.2 of Carlin and Louis, 1996; Bishop et al., 1975) in estimation of Q may well be advisable. The essential idea is to shrink the observed RFs of HAGs for individual stocks toward cen- tral values that are more reliably determined and are consistent with the genetic similarity of the stocks. An informative Bayes prior distribution for these unknown genetic parameters underlying the stock-mixture sample can be derived from the baseline samples and would pro- vide for such shrinkage. The statistical modeling begins with the allele RFs at a single locus but applies equally to haplotypes, alleles, or genotypes. Later, the modeling is extended to cover multiple loci. The Bayesian scenario begins with an imaginary experi- ment in which the RFs of the T distinct alleles for a single locus are drawn for each of the c baseline stocks (sec. 3.7 of Lange, 1997). Denote the resulting unobserved RFs for the ith stock by q=(qiv qi2,-.., qiT). The draws from the stocks are independent and from a common Dirichlet probability density, which is the Bayes prior for baseline sampling, n(ql ) =Z)(/3 j,j32, . . . , PT). The justification for the Dirichlet prior for baseline sampling parallels that for the earlier stock- mixture composition prior, n (p), that is, computational con- venience and its simple interpretation as additional data. Next, baseline samples of n1,n2,...,nc alleles of the locus are available from the c stocks. The counts of the different alleles — y.=(y. ^i2, . . . ,ylT) for the ith stock — have the mul- tinomial distribution, Multin •, q:), and therefore the base- line posterior for the unknown allele RFs in each stock is also a Dirichlet distribution, q,\yt ~ D(P1+yiVP2+yl2, ... , /fT+y(T). The posterior mean of q,\y, can be written as a weighted average of the observed and prior mean RFs (Bishop et al., 1975; Sutherland et al., 1975), E(q„ I A 37 ) = (y„ + p, )/(«, +/?.) = n, + p. Ll n, P • [A n,+p.\ p. j t = 1,2,.. ,,T, (4) where the observed RF is y(>/rq, its prior mean is P,/ p., and p. = Y.Pr If the baseline sample is missing ( n=0 ), the posterior mean equals the prior mean. Otherwise, the pos- terior mean ranges between the observed and prior mean RFs (as a function of /3. >0). Shrinkage from the usual esti- mator of qr the observed allele RFs, toward the prior mean increases with the prior “sample size,” p ., but so does bias in estimates of the allele RFs. Therefore, the magnitude of the prior parameters should be no larger than necessary to satisfactorily control estimation error. Although the choice of a Dirichlet baseline prior was partly for convenience, the resulting posterior density has good properties. The posterior mean is a reliable estima- tor for the unknown allele RFs: it is strongly consistent, becomes unbiased for large baseline sample size, and mod- erates the extremes of the usual estimates — the observed RFs — among baseline stocks. All posterior means for the allele RFs are positive, so that absence of an allele from a stock’s baseline sample implies it is only rare and was missed in sampling rather than it is nonexistent. The values for the baseline prior parameters, PVP2, . . . , PT, have been arbitrary. To complete the specification of the baseline posterior, which will serve as the stock-mixture Pella and Masuda: Bayesian methods for analysis of stock mixtures from genetic markers 155 prior for the allele RFs, their values must be assigned. Two approaches — empirical Bayes and pseudo-Bayes — are con- sidered in which the prior parameters are functions of al- lele counts in the baseline samples. The empirical Bayes method was previously developed for geneticists to esti- mate allele RFs (Lange, 1997). In this method, the values assigned to the /Is are those which maximize the Bayes prior predictive distribution (Gelman et al., 1995) for the allele counts in the baseline samples. This distribution is the marginal distribution of the allele counts, which results from averaging their multinomial distribution, Mult(nt , qt), weighted by the prior probability of qr D((3V j32, ... , PT). The prior predictive distribution is parameterized by the /Is alone, and the optimizing values can be computed from the allele counts (Appndx. 1). Limited experience during this study indicates that, with large baseline samples of a few baseline stocks, or lesser baseline samples for large numbers of baseline stocks, the empirical Bayes method can provide values for the prior parameters, which result in sensible weighting of the observed sample and prior means. Commonly, baseline sampling is more limited, and pragma- tism requires a less-demanding alternative method. The pseudo-Bayes method is based on several practical considerations to determine values for the baseline prior parameters. First, the baseline prior parameters, fiv p0,..., PT, have no intrinsic value, other than as tuning parame- ters by which to perform stock-mixture analysis. Nonethe- less, a sound rationale and simple computational formu- las for their values are desirable. Second, the prior mean should reflect the similarity of the allele RFs among the baseline stocks. Third, the weights assigned to the prior and observed allele RFs should allow a realistic evalua- tion of the uncertainty in the genetic composition of the stock, yet not cause misleading bias in the estimated stock composition. Loci with large variation among stocks have more effect on estimated stock composition than those with small variation. Therefore, shrinkage from observed allele RFs toward prior means for loci with large varia- tion should be less than for loci with small variation. If the prior parameter sum, /!., is substantially smaller than the baseline sample sizes, the bias will be limited. How- ever, with j3.=0 , all weight goes to the observed RFs. Then, when a baseline sample misses an allele that is present, sampling error will be underestimated (as it is with boot- strapping under the CML method). Fourth, and last, the weight assigned to the observed RFs for a stock should be positively related to its baseline sample size. The pseudo-Bayes method of this proposal is original to estimating allele RFs and appears in practice to satisfy the aforementioned criteria. The prior mean will be cen- tered within the observed allele RFs for the stocks of the baseline samples with P, = P--yt, t -1,2 t, where p. = is an estimate (Appndx. 2) of the value for j8. that minimizes the baseline risk, or expected squared-errors between the posterior means at Equation 4 and the unknown allele RFs of all baseline stocks, and — 1 y y = — > — = is the baseline center, or unweighted arith- c >=i n‘ metic mean, of the observed RFs for the hth allele among stocks. With this definition for the /Is, the prior mean equals the baseline center. The posterior mean for any stock is the weighted average of its observed allele RFs and the base- line center as at Equation 4. Although the central allele RFs for the entire set of baseline stocks anchors the esti- mation of Q in this description, extensions to accommodate regional or other groupings of stocks could be accomplished as simply by anchoring on regional or group centers. Complete analysis of the baseline requires repeated and separate application of the empirical Bayes or pseudo- Bayes methods to each locus. Suppose a total of H loci com- pose the stock-mixture multilocus genotypes. Let the /zth locus have Jh alleles with prior parameters j3/( = (j8/(1, Ph2, . . . , PhJ/ ) and allele RFs in the ith stock of qih = (qihl, qih 2, • • • - QihjJ- If Qi denotes the ith stock’s combined arrays, qrn,qh i=l i= 1 h = l H V \jD(Phl,ph2,...,phJi) , V h=l y that is, prior draws for allele RFs are independent among stocks and loci. The baseline samples are drawn independently from the stocks. Denote by Y( = iyivyi2, • • • ,;y(7/) the H arrays of al- lele counts in the baseline sample for the ith stock, and by Y, the entire baseline collection of Yj,Y2, . . . ,YC. Then the Bayesian posterior density for the allele RFs of the entire baseline is the product of Dirichlet densities, c c H k(q i y> = i ^)=nri;r<9 iy<-*)= /=1 i = l h = l c H (5) | i| f\D(phX+ylln,...,phJk +yihJ), i=\ /» = 1 and each density in the product has a mean vector, for the stock and locus, equal to a weighted average of the observed allele RFs and corresponding prior means (as at Eq. 4). Although the statistical modeling of the baseline samples has been described with alleles and loci, it applies equally to any combination of independent components: alleles at loci, haplotypes at mtDNA, and genotypes at loci in Hardy- Weinberg disequilibrium. Stock-mixture sample likelihood function for unknowns, g(X\d) The stock-mixture sample likelihood function is propor- tional to the probability of drawing the observed stock- mixture genotypes as a function of the unknowns, p and 156 Fishery Bulletin 99(1 ) Q. Denote the count of the yth allele of the /?th locus for the mth mixture individual by xmh •. Let the collection of such counts, Xm, denote the multilocus genotype of the mth individual, and let the array X denote the collection of such arrays for the M individuals composing the stock- mixture sample. Further, let the RF of individuals with the genotype Xm in the ith stock, which depends on that stock’s allele RFs, be denoted as flXm \ Q{). The RF of the genotype in the stock mixture is the weighted sum, Ip.KXJQ ;), and so the likelihood function for the stock- mixture sample is Dirichlet priors with multinomial counts from the stock mixture. The stock identities of the mixture individuals are deter- mined by chance in the data augmentation algorithm. Let zm=(zmi,zm2, • • • , zrnc) indicate the stock origin of the mth mixture individual by a single “1” at the coordinate of the contributing stock, and c-1 “0”s at the remaining coordi- nates. For later reference, let Z=lzv z2, . . . , zM) denote the stock origins of all the mixture individuals. If p and Q were known, the proportion of mixture individuals with genotype Xm that came from the ith stock could be calculated as in c g(X|p,Q) = ]"~[ ^p,/(Xm | Qj (6) In the Bayesian view, X is fixed and g(X |p,Q) is a random function of the unknowns, p and Q. Again, although the likelihood function for the stock-mixture genotypes has been described with alleles and loci, it applies equally to stock-mixture genotypes of any combination of indepen- dent components: alleles at loci, haplotypes at mtDNA, and genotypes at loci in Hardy- Weinberg disequilibrium. Posterior distribution of the unknowns, nlO\X, Y) The Bayesian assessment of the unknown stock propor- tions in the stock mixture and of the baseline RFs of hap- lotypes, alleles, or genotypes is provided by their joint posterior distribution. This posterior distribution is propor- tional to the product of the prior density for the unknowns and the likelihood function of the stock-mixture sample, given the unknowns. The prior density for the stock-mix- ture proportions is the uninformative Dirichlet of Equa- tion 2. The baseline posterior at Equation 5 becomes the stock-mixture prior for the HAG RFs. Prior information on stock-mixture composition and the HAG RFs is rea- sonably considered independent, so the joint prior for the unknowns is the product (Eqs. 2 and 5), Wmi = Pif(Xm \Qi^^Pkf(Xm IQ*}, i ~ 1,2,... C. (8) k=l Equivalently, the probability that a randomly drawn mix- ture individual with genotype Xm came from the ith stock is wmi of Equation 8. The data augmentation algorithm draws the missing stock identity, zm, for each mixture indi- vidual from the multinomial distribution, zm ~ Multi l,wm), where the probabilities for the stocks listed by tvm={wml, wm2, . . . ,wmc) are computed from the current samples of p and Q. Colloquially, the stock identity of each stock-mix- ture individual is randomly assigned with the probability for any stock equal to the stock-mixture fraction of the genotype contributed by the stock. In broad outline, the data augmentation algorithm used to draw posterior samples is straightforward. After the ini- tial sample is obtained (as described later), a sequence of samples is drawn with each sample dependent only on the preceding sample, that is, the algorithm is a Markov chain Monte Carlo (MCMC) method. At the £th sample, two steps are performed: 1 Draw stock identities of the mixture individuals, zik'~ Mult( l,M>(fe^), using Equation 8 for genotype Xm and the current values p=p,k> and Q-Q, m- 1,2, . . . , M. 2 Draw p(k+v and Q(k+1) from their respective posterior densities, nip \X,Y,ZM), and nlQ\X,Y,Z(k)). 7T(p,Q) = 7T(p)7T(Q | Y). 11) The posterior distribution for p and Q with the stock- mixture sample observed, nlp.Q |X,Y), is proportional to the product of their likelihood at Equation 6 and their prior at Equation 7. Analytic evaluation of the posterior distribution is impractical because of the prodigious com- putation required, caused by the combinatorial explo- sion of terms in the likelihood function with increase in stock-mixture sample size (Bernardo and Giron, 1988). Instead, a sufficient number of samples are drawn se- quentially from the posterior distribution to accurately describe it. The data augmentation algorithm (Tanner and Wong, 1987; Diebolt and Robert, 1994) can be used to draw the sequence of samples. The idea underlying the algorithm is that the estimation problem would be much simplified if the stock identities of the mixture individ- uals were known. Given the stock identities, the poste- rior distribution for the stock proportions and HAG RFs in the baseline stocks simply requires updating of the The stock identities, Z(k>, of the stock-mixture sample are sufficient statistics for p (Pella et al., 1996). With them available, the genetic data of the stock-mixture sample is of no value to estimation of p. Therefore, the posterior for p is obtained by updating the Dirichlet prior for p with the counts of stock identities for the mixture individuals, n(p\X,Y,Zik)) = n(p\Zik)) = The posterior density for HAG RFs of the genetic com- ponents, niQ\X,Y,Zlk>), updates the stock-mixture prior, or baseline posterior, MQ \ Y), at Equation 5 for the HAG counts from the identified mixture individuals as H n(Qt | X, Y,Z"°) = ]Jnlqib | X,yih,Z‘k)) = (10) Pella and Masuda: Bayesian methods for analysis of stock mixtures from genetic markers 157 n n D ivi Ph 1 "h Vih 1 + (Zmi Xmh 1 )> • ■ ■ > Pli.J,, + V ihj ,, m=l M (10) (cont.) i = X2,...,c. Notice that each of the updated Dirichlet parameters for the HAG RF in the ith stock equals the sum of the prior parameter, the HAG count in the baseline sample, and the HAG count for the mixture individuals identified to the stock. The data augmentation algorithm cycles the two steps and eventually outputs a sequence, or chain, of samples of stock proportions and baseline genetic parameters from the posterior Bayes distribution. However, the early sam- ples of a chain are influenced by the starting values of p and Q. To make valid inferences, early burn-in samples must be discarded and a sufficient number of subsequent samples must be generated to accurately describe the pos- terior. Statistics to determine the number of samples to generate per chain — the Raftery and Lewis ( 1996) conver- gence diagnostic — or to monitor convergence of multiple chains to the desired posterior density — the Gelrnan and Rubin (1992) shrink factor — are widely used in practice and are comparatively inexpensive to compute. Although these two statistics are used in the applications that fol- low, MCMC research is currently very active (e.g. Brooks and Roberts, 1999), and alternatives may prove to be su- perior for these purposes. Main interest is usually in the stock composition of a stock mixture and so in the later ap- plications the statistics have been applied only to samples of p even though they can be applied to samples of Q as well. However, convergence of samples forp without corre- sponding convergence for Q is not thought to be possible. The diagnostic outlined by Raftery and Lewis ( 1996) de- termines the number of samples required for estimating quantiles (q) of posterior distributions with a specified ac- curacy (?•) and probability (s). The Fortran implementa- tion of the diagnostic, called gibbsit,1 is applied in later examples (with r/=0.975, /-=0.02, and s=0.95) to each of several chains of stock proportions generated from differ- ent starting values. The diagnostic first requires that an initial pilot sample be generated for each chain, which is used to compute its recommended number of samples. An additional number of samples are generated to satisfy the maximum recommended. The combined samples — original pilot samples and the additional samples — are used with gibbsit as pilot samples to compute recommended sample sizes again. Further samples are generated if the maxi- mum recommended sample size for any stock exceeds the number so far generated. This iterative scheme is applied to the first chain, beginning with a pilot sample size of 235 (the initial number recommended from the chosen val- ues of q, r, and s). The other chains are run the length of 1 Fortran program gibbsit (version 2.0) can be obtained without cost from the general archive at http: / / lib.stat.cmu.edu / . the first chain and then analyzed separately by gibbsit. If gibbsit suggests that any of the chains should be longer than the first chain, then all the chains are run for the largest number of samples recommended for all stocks and all chains. Gelrnan and Rubin (1992) recommended running a small number of independent chains with dispersed start- ing points to reduce the possibility that a chain is ac- cepted as representative of the posterior distribution be- fore convergence has occurred. To monitor convergence of the chains to the posterior density, a univariate statistic, called the shrink factor (Gelrnan and Rubin, 1992), is com- puted2 for each of the stock proportions. One chain per stock is started, with most of the stock mixture initially contributed by that stock. Once chains of samples are gen- erated and length determined from the Raftery and Lew- is diagnostic, shrink factors for all stocks are computed to verify that the chains have converged. The shrink fac- tor is computed from the second halves of the chains and compares the variation within a single chain for a given parameter to the total variation among the chains. Esti- mates of shrink factors close to one indicate convergence, and acceptable values are less than 1.2 (see Kass et al., 1998). Because the shrink factor is computed from the sec- ond halves of chains, the first halves of chains are discard- ed as burn-in samples. The purpose of discarding initial burn-in samples is to remove dependence on the starting values. Samples subsequent to the burn-in samples may be thought of as coming from the desired posterior dis- tribution. Once convergence of chains has been verified, the MCMC samples (after burn-in discard) of stock compo- sition estimates are combined and summarized with var- ious statistics (equivalent to parameters because of the large samples): means, standard deviations, and empirical percentiles (2.5, 50, and 97.5). Baseline haplotype, allele, or genotype RFs can be summarized similarly. Checking the fit of the stock-mixture model Current stock-mixture modeling presumes that a stock mixture composed of random genotypes from the contrib- uting stocks occurs and that a simple random sample of the mixture individuals has been drawn. Further, it is assumed that simple random samples of HAGs from all contributing stocks are available by which to estimate, with an appropriate and known genetic model, the stock- mixture genotype RFs in each of the separate stocks. Samples are considered small in relation to the popula- tions sampled, so that the multinomial distribution can be used to describe sampling variation in counts. These assumptions may be plausible in many applications, but violations can also occur. For example, baseline samples of juvenile salmon, drawn before their families have mixed, would have extra-multinomial variation (Waples 2 Our current Fortran program for the diagnostic is a translation of Gelman’s S function itsim (free from the Statlib S archive at http: / / lib.stat.cmu.edu / ), and its modification (version 0.4) in CODA (Best et al., 1995) (free from the MRC Biostatistics Unit, University of Cambridge, at http://www.mrc-bsu.cam.ac.uk/ bugs/). 158 Fishery Bulletin 99(1 ) and Teel, 1990), as would stock-mixture samples if the populations segregated (McKinnell et al., 1997). The Bayes method opens the way for checking that the models fit. Lack of fit is indicated if the observed samples are unusual realizations of the Bayes posterior predictive distribution (Gelman et al., 1995). Test statistics should be designed to detect suspected problems in stock-mix- ture analysis: unrepresentative samples, unsatisfactory priors, presence of extra stocks, etc. In particular, the hap- lotype, allele, or genotype counts in the actual baseline samples should not be outliers of their corresponding pre- dictive distribution. When violations to the assumptions are detected, the posterior distribution of stock propor- tions and baseline genetic parameters would be mislead- ing. New samples drawn by improved design, or alternate sampling models, could be needed to make the stock-mix- ture analysis trustworthy. Samples are easily drawn from the posterior predictive distribution. The kth predictive baseline sample for the HAG counts in a stock is simply a multinomial sample with size equal to that of the actual sample and with prob- abilities equal to the HAG RFs in the /eth posterior sam- ple. The Mh predictive stock-mixture sample is obtained in two steps. First, a multinomial sample of M individuals identified to stock is drawn, with probabilities equal to the stock proportions in the /?th posterior sample. Second, the stock-mixture genotype of each individual is generated bv sequentially drawing the HAGs of the multiple characters by using the HAG RFs for its stock from the kth. posterior sample. Applications Two applications are considered next to illustrate use of the Bayesian method. In the first application, large num- bers of mtDNA haplotypes are present in the baseline and stock-mixture samples and pose special difficulty in analy- sis. The fairly common availability of mtDNA data makes this application of general interest. In the second applica- tion, only one of two populations in a stock mixture could be sampled separately. The Bayesian solution for the miss- ing baseline samples from the second population should be of special interest to biologists concerned with assessing stock mixtures of anadromous and resident populations in streams (Busby et al., 1996; Michael, 1983), and of gen- eral interest for extensions to the standard stock-mixture analysis. Example 1 : mtDNA samples from harbor porpoise (Pho- coeno phocoena) of the northwest Atlantic Ocean (Rosel et al., 1999) Rosel et al. ( 1999 ) obtained mtDNA sequence data for samples from four summer breeding popula- tions— Gulf of Maine-Bay of Fundy, Gulf of St. Lawrence, Newfoundland, and West Greenland — of harbor porpoises in the northwest Atlantic and from a wintering group along the mid-Atlantic states. The authors were reason- ably certain that the wintering group comprised one or more of the summer populations. Because of special con- servation concerns for the Gulf of Maine-Bay of Fundy population, the authors wished to determine if it alone could have been the wintering group. Contingency table analysis of the mtDNA haplotype frequencies indicated only that the Gulf of Maine-Bay of Fundy population was almost surely not alone (P<0.06), if at all present. Rosel et al. (1999) used a stock-mixture analysis by the CML method to attempt to delimit the population contributions better with the mtDNA data. Here the Bayesian method is applied to the same data for comparison. Summer sample sizes for each of the populations were between 40 and 80 individuals, and the winter sample size was 41. A total of 67 distinct haplotypes was observed in the summer sam- ples, and the winter sample of 41 individuals included an additional 8 singleton haplotypes previously unseen. Among the total of 253 individuals of all samples, the five most numerous haplotypes were represented by 45 (18%), 42 (17%), 15 (6%), 9 (4%), and 7 (3%) individuals. Most haplotypes were sporadic in samples; the most common counts in the summer and winter samples being 0 and 1. The occurrence of a few fairly common and many scarce haplotypes is characteristic of mtDNA data (Xu et al., 1994) and poses special difficulty in estimation. For exam- ple, under the CML method, stock-mixture haplotypes contributed by a particular stock have another apparent source if absent from its baseline sample. Four chains of samples were generated by data augmen- tation with both the empirical Bayes and pseudo-Bayes methods for specifying the baseline prior “count” param- eters. The total prior “sample size,” [i. computed by the methods was 22 (pseudo-Bayes) and nearly 2000 (em- pirical Bayes). An initial pilot chain of 235 samples was analyzed by using the Fortran implementation of gibbsit, which indicated that chains of 2012 samples should be run (given <7=0.975, r=0.02, and s=0.95). The four chains were begun with diverse values for population propor- tions: one chain was begun for each population, with it composing 0.95 of the stock mixture and the other three populations composed equal parts (thirds) of the remain- der (0.05). The four chains had mixed sufficiently, or con- verged, by their second halves so that the Gelman-Rubin shrink factors were less than 1.03 for any one population. The samples from the second halves were pooled to rep- resent 4024 draws from the posterior distribution. Predic- tive baseline samples were generated from the posterior samples for haplotype RFs, and indicated lack of fit only for the empirical Bayes method (Fig. 1). Therefore, only the posterior distribution from the pseudo-Bayes method will be described further. Parameters for population pro- portions computed from the posterior sample (Fig. 2) in- clude the mean, mode, median, standard deviations, and equal-tail bounds of posterior intervals (Table 1). Condi- tional maximum likelihood estimates for the winter sam- ple were computed for comparison, along with bootstrap evaluation of their precision from 1000 resamplings. Cor- responding statistics of the bootstrap sample for the CML method are the means, standard errors, and 95% confi- dence bounds (Table 2). This CML analysis differs from that of Rosel et al. (1999) by using 1) the counts of all individual haplotypes instead of pooling to form subsets with larger counts, and 2) an alternate method for con- structing confidence bounds. Rosel et al. (1999) used the Pella and Masuda: Bayesian methods for analysis of stock mixtures from genetic markers 159 0.3 0.3 0.2 0.2 0.1 0.0 .■III.. 0.1 0.0 6 I 0 10 20 30 40 50 0 10 20 30 40 50 0.3 0.2 0.1 0.0 ' 0 10 20 30 40 50 0.3 0.2 0.1 0.0 20 30 40 50 30 40 50 Type count Figure 1 Histograms of predictive baseline population (Gulf of Maine-Bay of Fundy, top; Gulf of St. Lawrence, second; New- foundland, third; and West Greenland, bottom) sample counts for the most common haplotype in the pooled summer and winter samples, by empirical Bayes (left) and pseudo-Bayes (right) methods. The actual count of the haplotype in each baseline sample is shown as a spike. Table 1 Parameters of the posterior density for harbor porpoise population proportions composing the winter stock mixture. Reported pro- portions do not necessarily sum to 1.0 because they are rounded. Population Mean Mode7 SD 2.5% Posterior quantiles Median 97.5% Gulf of Maine-Bay of Fundy 0.12 0.02 0.13 0.00 0.08 0.46 Gulf of St. Lawrence 0.48 0.69 0.19 0.14 0.48 0.84 Newfoundland 0.15 0.02 0.16 0.00 0.10 0.52 West Greenland 0.24 0.26 0.18 0.00 0.22 0.66 7 The mode is computed by 4-dimensional binning of the Markov chain Monte Carlo samples for stock proportions, each bin with sides of 0.05, and then normalizing the bin center having maximum count. percentile interval (Efron and Tibshirani, 1993) for confi- dence bounds. The alternate method, called the non- symmetric percentile bootstrap (Lunneborg, 2000), is expected to have superior coverage properties to the stan- dard percentile method for the usual skew distributions of stock-mixture composition estimates. The confidence 160 Fishery Bulletin 99(1 ) Table 2 The conditional maximum likelihood point estimate for harbor porpoise population proportions composing the winter stock mix- ture, and its bootstrap standard error and 95% confidence bounds (nonsymmetric percentile method). Reported proportions do not necessarily sum to 1.0 because they are rounded. 95% Confidence bounds Point Population estimate SE' Lower Upper Gulf of Maine-Bay of Fundy 0.19 0.14 0.00 0.37 Gulf of St. Lawrence 0.40 0.16 0.13 0.77 Newfoundland 0.18 0.15 0.00 0.35 West Greenland 0.24 0.15 0.00 0.48 1 These standard errors are reduced by 30% to 50% from those reported by Rosel et al. (1999). At our earlier recommendation, the authors pooled subsets of haplotypes without a well-grounded basis in order to avoid the small counts of individual haplotypes used here. The point estimate is unchanged, but the confidence intervals differ mainly because a new method was used in their computation. bounds were computed with an update of the program, spam 3.2 (Debevec et al., in press), available on the inter- net at http: / / www.cf.adfg.state.ak.us / geninfo / research / genetics / Software / SpamPage. htm . Both the pseudo-Bayes and conditional maximum likeli- hood methods are in agreement on the population compo- sition of the winter sample in five respects (Tables 1 and 2). First, any of the populations could be involved in the stock mixture and comprise much (upper posterior bounds range from 0.46 to 0.84, and upper confidence bounds, from 0.35 to 0.77) of it. Second, the contributions by any of the populations are very imprecisely determined from the mtDNA counts (widths of all interval estimates exceed 0.35, and standard deviations or standard errors range from 0.13 to 0.19). Third, more than one population seems to be present, given that none of the interval estimates in- cludes 1.0. Fourth, the most frequent estimates, or modes, of the Bayes posterior imply that the mix is almost entire- ly composed of the Gulf of St. Lawrence and West Green- land populations (Fig. 2). Fifth, and last, the Gulf of St. Lawrence population was almost certainly present (lower 95% posterior bound for the proportion, 0.14; correspond- ing lower 95% confidence bound, 0.13). The claim that the Gulf of St. Lawrence population was wintering along the mid-Atlantic coast is important to the conservation issue. Is its presence conspicuous to a direct examination of the genetic samples? The answer is yes if one knows where to look. The Bayes method actually identifies the stock origins of the stock-mixture individu- als during generation of each sample from the posterior distribution, and so the posterior identity distribution — the relative frequency of assignment to each population — for each stock-mixture individual is available. The identi- ty distributions of mixture individuals showed 4 of the 33 (12%) winter porpoise were more likely than not (posterior probabilities-0.60, 0.64, 0.67, 0.72) to be from the Gulf of St. Lawrence, and 4 more were more certainly (posterior probabilities=0.87, 0.88, 0.88, and 0.89) from that popula- tion. Corresponding probabilities (Eq. 5, Pella and Milner, 1987) from the CML method for the first (0.62, 0.69, 0.70, 0.81) and second groups (1, 1, 1, 1) agreed reasonably. The summer samples contained the following numbers of the same 8 haplotypes: Gulf of Maine-Bay of Fundy, 2 of 80 (3%); Gulf of St. Lawrence, 14 of 40 (35%); Newfoundland, 4 of 42 (10%); and West Greenland, 6 of 50 (12%). Except for the Gulf of St. Lawrence population, in which these haplotypes were fairly common, their observed RFs in the other populations were half or less of that (24%) in the winter sample. With the observed haplotype RFs assumed to be accurate, the probability is less than 0.05 that 8 of 33 individuals with the haplotypes came from any popu- lation other than Gulf of St. Lawrence. The conjunction of necessary sampling errors — higher frequencies of the 8 haplotypes in the other populations or lower frequency in the stock mixture — without the presence of Gulf of St. Lawrence is deemed highly improbable from the Bayes computations. Without the posterior probabilities of stock identities, a search for direct evidence of the presence of particular populations in the winter sample would have been futile. A total of 25 sets of simulated baseline and stock-mix- ture samples of harbor porpoise mtDNA haplotypes was generated for each of four experiments. Sizes of the sim- ulated samples equaled those of the actual data. The ex- perimental conditions that were controlled include the pro- portions from the four populations in the stock mixtures and their haplotype RFs. In three of the experiments, the Gulf of St. Lawrence population comprised 0.95 of the stock mixture, and the other stocks comprised equal thirds of the remaining 0.05. In the fourth experiment, the four pop- ulations contributed equal parts (0.25) to the stock mix- ture. The haplotypes of the samples were drawn with re- placement from either the original baseline samples (0% addition) or augmented baseline samples for which half (50% addition) or all (100% addition) of the missing hap- lotypes were replaced by singletons. The conditional maxi- mum likelihood method was applied to each set of simu- lated samples just as it had to the actual samples. The Bayes method was similarly applied, but with a single ex- ception— a long fixed sequence of 5000 samples (first 2500 discarded as burn-in) was generated for all sets to reduce processing labor. Average point estimates among the 25 Pella and Masuda: Bayesian methods for analysis of stock mixtures from genetic markers 161 25 20 15 10 5 0 0.0 0.2 0.4 0.6 0.8 1.0 Gulf of Maine - Bay of Fundy 0.0 0.2 0.4 0.6 0.8 1.0 Gulf of St. Lawrence iS CD cc 25 20 15 10 5 0 iii! 0.0 0.2 0.4 0.6 0.8 1.0 Newfoundland 10 8 6 4 2 0 0.0 0.2 0.4 0.6 0.8 1.0 West Greenland Population proportion Figure 2 Histograms of samples from the posterior distribution of unknown proportions of summer populations of harbor porpoise (Gulf of Maine-Bay of Fundy, Gulf of St. Lawrence, New- foundland, and West Greenland) in the wintering group. sets of samples for each experiment — Bayes mode, Bayes mean, and conditional maximum likelihood (CML) esti- mate— and their standard errors were computed (Table 3). The main lesson of these simulations is that the Bayes method, as configured, performs reasonably well in the frequency sense, that is, under repeated sampling. The Bayes posterior mode seems to be a practical point es- timator for population proportions in stock mixtures: it was less biased than the Bayes posterior mean and the conditional maximum likelihood estimate when the ex- perimental conditions caused bias. As is characteristic of stock-mixture composition bias, uneven population contri- butions combined with large variation in estimated con- tributions were aggravating. When the populations con- tributed equally (stock mixture 4), bias was negligible for any estimator, given the large variation of estimated popu- lation proportions, but with unequal contributions (stock mixtures 1-3), the bias became increasingly severe be- cause the haplotypes were added to the populations and increased the variation in estimated contributions. Lower bias of the Bayes mode was not without cost because its variation among sets of samples was generally larger than that of the Bayes mean or conditional maximum likeli- hood estimate for the more-difficult third and fourth stock mixtures. Example 2: Sashin Creek steelhead ( Oncorhynchus mykiss) stock mixture Sashin Creek on Baranof Island in South- east Alaska contains a population of anadromous rainbow trout, or steelhead, in its lower portion. In addition, a self- sustaining population above a barrier waterfall was cre- ated in 1926 by a transplant from the lower to the upper portion (which includes two lakes). Although the falls was a barrier to upstream migration, migrating juveniles from the upper portion apparently survived the plunge to the lower river. Samples of mature adults returning from the ocean, obtained from the lower portion, were compared with similar samples from the upper population for allo- zymes (21 loci with 2-6 alleles per locus), microsatellites (10 loci with 3-26 alleles per locus), and mtDNA (5 haplo- types). An excess of homozygotes at loci (Wahlund effect) provided evidence that the samples came from a mixture of both populations. In particular, the allozyme, PGK2, appeared to be fixed (100%) in the upper population, yet the fixed allele represented less than 50% of the PGK2 alleles in the stock-mixture sample from the lower por- tion. Biologists3 were able to infer that roughly 25% of the stock mixture probably originated from the upper popula- 3 Thrower, F. 2000. NMFS, Auke Bay Laboratory, Juneau, AK 99801-8626. 162 Fishery Bulletin 99(1) Table 3 Average point estimates- —Bayes mode, Bayes mean and conditional maximum likelihood (CML) estimate — and their standard errors (in parentheses) for 25 simulated samplings of four stock mixtures composed of harbor porpoises from the Gulf of Maine- Bay of Fundy, Gulf of St. Lawrence, Newfoundland, and West Greenland. The haplotypes of stock mixtures were drawn from the original baseline samples (0%) or augmented baseline samples for which half (50%) or all (100%) of the missing haplotypes were replaced by singletons. Baseline and stock-mixture sample sizes were those reported by Rosel et al. (1999) and analyzed earlier in this section. Reported proportions do not necessarily sum to 1.0 because they are rounded. Stock mixture Gulf of Maine-Bay Gulf of and estimator of Fundy St. Lawrence Newfoundland West Greenland Stock mixture 1 : 0% 0.95 0.01666 0.01666 0.01666 Bayes mode 0.91 (0.07) 0.03 (0.02) 0.02 (0.00) 0.04 (0.07) Bayes mean 0.76 (0.17) 0.07 (0.11) 0.09 (0.12) 0.08 (0.12) CML mean 0.82 (0.09) 0.04 (0.06) 0.07 (0.08) 0.06 (0.08) Stock mixture 2: 50% 0.95 0.01666 0.01666 0.01666 Bayes mode 0.85 (0.15) 0.04(0.07) 0.07(0.11) 0.04 (0.07) Bayes mean 0.71 (0.20) 0.09 (0.13) 0.11 (0.13) 0.09 (0.13) CML mean 0.68 (0.13) 0.09 (0.10) 0.14 (0.09) 0.10 (0.09) Stock mixture 3: 100% 0.95 0.01666 0.01666 0.01666 Bayes mode 0.72 (0.30) 0.08 (0.19) 0.04 (0.06) 0.16 (0.27) Bayes mean 0.59(0.24) 0.12 (0.17) 0.12 (0.16) 0.17 (0.20) CML mean 0.56 (0.13) 0.12 (0.13) 0.13 (0.11) 0.19 (0.14) Stock mixture 4: 0% 0.25 0.25 0.25 0.25 Bayes mode 0.25(0.26) 0.18 (0.29) 0.21 (0.22) 0.36 (0.34) Bayes mean 0.27 (0.22) 0.23 (0.22) 0.24 (0.21) 0.26 (0.23) CML mean 0.30 (0.14) 0.21 (0.14) 0.24(0.13) 0.25 (0.15) tion. Their method depended on the fixed condition of the locus in the upriver population: removal of about 25% of such mixture individuals resulted in the remainder meet- ing Hardy- Weinberg equilibrium. This approach was dif- ficult to generalize to the other loci, most of which were Highly variable. Further, the approach could not provide a complete description of the genetic composition of the lower-river population. To use the information better, all loci available for each type of genetic data were analyzed to provide a Bayes posterior distribution of the population proportions and their allele (allozymes and microsatellites) or haplotype RFs (mtDNA). Each type of genetic data was treated separately in order to examine the consistency of population composition estimates from independent data. The stock-mixture prior distributions for the baseline characters required some change to accommodate the sin- gle-population baseline. As is routine, loci were assumed to have been inherited independently, and their alleles were in Hardy-Weinberg equilibrium for either popula- tion. However, the baseline prior parameters (/3s) for each genetic character of the upstream population were un- informative: their sum, the baseline prior “sample size,” equaled just 1, and each equaled the inverse of the number ( Jh ) of HAGs. (empirical Bayes or pseudo-Bayes methods for computing the prior parameters were not applicable with a single baseline population.) With the baseline sam- ple counts for locus h denoted as yh = iyhl, ■ ■ ■ ,y/,jitY, the stock-mixture prior (or baseline posterior) of HAG RFs for the upstream population was n n{QuP I ) = IT ^yhx + yhJll +J~h Notice that the stock-mixture prior “sample size” was the unit-augmented actual sample size, nh+ 1, where nh=^yh/. The corresponding downstream population stock-mix- ture prior reflected the even greater uncertainty in that population’s characteristics by a downstream stock-mix- ture prior “sample size” equal to only 1, yet with average HAG RFs closely approximating those from the counts, xh = ( xhv . . . ,xhJhY, seen in the stock-mixture sample, viz. n Xli 1 + ^ h xi,jIi +tV +1 X xhj + 1 The prior for population proportions, n(p), was the stan- dard Dirichlet, D( 0.5, 0.5). Three chains were generated, beginning from diverse upriver population proportions of 0.95, 0.50, and 0.05 (Fig. 3). The chain lengths for allo- zymes and microsatellites were 10,000 samples with the first 5000 discarded as burn-in. The posterior sample com- prised the 15,000 samples from their second halves. The chain lengths for mtDNA were 100,000 samples and the posterior sample comprised the 150,000 samples from their second halves. Pella and Masuda: Bayesian methods for analysis of stock mixtures from genetic markers 163 Iteration Figure 3 The three chains of posterior samples for the unknown proportion from the upper Sashin Creek steelhead population present in the lower Sashin Creek stock mixture, based on microsatellite data. Each chain was initiated with a different value for the unknown upper proportion, 0.95, 0.50, and 0.05. All chains tracked to high proportions during the early phase and later stabilized at the equilibrium posterior distribution. Table 4 Parameters of the marginal posterior density for the upriver steelhead proportion in the lower-river stock mixture. Posterior quantiles Data Mean Mode SD 2.5% Median 97.5% Allozymes 0.25 0.22 0.07 0.12 0.24 0.40 Microsatellites 0.24 0.22 0.06 0.13 0.24 0.37 MtDNA 0.41 0.00 0.19 0.01 0.46 0.70 The posterior marginal distributions of the upriver propor- tion from allozymes and microsatellites were nearly identical and their modes (0.22) or means (0.24, 0.25) were support- ive of the earlier approximate assessment of 0.25 (Table 4). Either data set points with high probability (0.95) to an up- river population presence between 0.1 and 0.4 of the lower- river stock mixture. The mtDNA was not nearly as informa- tive in regard to population composition, judging from the 164 Fishery Bulletin 99(1) resulting posterior marginal distribution. Its standard de- viation roughly equaled threefold, and its posterior interval length, twofold, that from allozymes or microsatellites. Discussion Analysts, accustomed to using likelihood or least squares methods for stock-mixture problems, should not be deterred by the novelty of the proposed Bayes method. Instead, the Bayesian implementation should be seen as eminently prac- tical and sensible. The data augmentation algorithm rec- ognizes each mixture individual as an entity and labels it with a stock origin. Given the stock assignments, the observed stock proportions are obvious estimates of the stock composition. In concurrence, the means, variances, and covariances of the Bayes posterior distribution for stock pro- portions approximate closely the observed stock proportions, their estimated variances, and their estimated covariances, respectively, from frequentist methods. Given the current stock proportions and genetic parameters of an MCMC chain, the random labeling accurately reflects the uncer- tainty in stock origins. Each mixture individual is assigned to one of the baseline stocks, by using probabilities of stocks proportional to each stock’s contribution of its genotype to the mixture. Stock proportions and genetic parameters of the MCMC chain gravitate toward their true values because draws from the posterior, which integrates the baseline and stock-mixture information, are more probable nearby. One goal in developing a Bayesian method of stock-mix- ture analysis was to replace the conditional maximum like- lihood assumption of ignorable baseline sampling error by modeling that acknowledged the uncertainty in genetic com- position of the baseline stocks. Ignorable baseline sampling error is especially unrealistic in applications for which un- common genotypes are present. Stock-mixture individuals, particularly those with uncommon genotypes, may be con- tributed by stocks whose baseline samples imply their ab- sence. Current bootstrap resampling of the baseline samples does not accommodate reasonably this mismatch between stock-mixture presence and apparent absence of a genotype in a baseline stock. The individual is presumed to come from a stock different from that of the contributor. Such mis- matches become frequent when many rare and uncommon genotypes occur, such as in mtDNA data. The simulations for harbor porpoise showed that as greater numbers of rare haplotypes occurred in the populations, the bias of the CML estimator became severe. When none of the baseline sam- ples can explain presence of a stock-mixture genotype, the CML assumption leaves only an outside source. Data sets generated during bootstrap resampling can require an ap- parent outside source even when the original samples did not. Pooling of uncommon types to circumvent their effects on estimation should be preceded by careful study to assure information useful to stock-mixture composition is not lost. Some potential to improve stock-mixture assessment by Bayesian methods arises from the prior for 9=(p, Q), which has no counterpart in the likelihood approach. The Bayes proposal for stock-mixture analysis emulates the objectiv- ity of likelihood methods by letting stock-mixture sample information dominate that of the neutral low-information prior for stock proportions. If information about p is truly unavailable, or the researcher prefers to withhold it and let the “data do the talking,” the neutral low-information prior will be adequate. However, the resulting composition estimates may be so imprecise as to be of limited practi- cal value. If additional information is available, either cus- tomizing the prior to include it, or updating the posterior (it becomes the prior) with the additional information may improve precision. As an example of updating, the three independent data sets for Sashin Creek steelhead trout could be integrated sequentially into a single posterior for population proportions. In attempting to maintain objectivity for description of the uncertainty in genotypic composition of the separate stocks, the empirical Bayes method for specifying the baseline prior parameters was examined. The empirical Bayes method for choosing prior parameters for the haplotype, allele, or geno- type RFs provided excessive weight to the prior mean for harbor porpoise, with the prior “sample size” parameter, /3., for harbor porpoise of nearly 2000, many-fold the total of actual sample sizes. In addition, the empirical Bayes meth- od consistently weighted prior means heavily on several ap- plications examined and not reported. Information in these typical baseline samples is evidently inadequate for estima- tion of the prior parameters. A full Bayesian analysis, which views them as random variables (sec. 5.3 of Gelman et al., 1995), would require an informative prior for them. Because such an informative prior was not evident, the pseudo-Bayes approach (Bishop et ah, 1975) was adopted. Under the pseudo-Bayes approach, the posterior mean for HAG RFs of stocks interpolates between observed val- ues for individual stocks and a baseline central value for all stocks, with the shrinkage, or weighting, determined by the values of the baseline prior parameters. The best choice for values of the prior parameters remains an open question. Possibly, the prior parameter values could be chosen for their performance in experiments of simulated stock-mix- ture analyses. However, the computations involved would be extensive and without guarantee beforehand of a clear solution. This proposal included an objective criterion — minimum squared-error risk of baseline allele RFs — by which to determine weighting between the prior and ob- served HAG RFs from the baseline samples alone. Re- searchers who find choice of weighting a deterrent to ap- plication of the Bayes method can set the baseline prior parameters to zero with the qualification that variation of stock proportions may be understated as with the CML method. In return, the Bayes algorithm easily includes the information in the stock-mixture and baseline samples in assessing stock proportions. The simulations for harbor porpoise showed that the weighting from the pseudo-Bayes method resulted in good frequency performance. In many practical applications, fisheries managers re- quire a point estimate of stock composition. The well- known bias of the conditional maximum likelihood esti- mate has been troublesome for this reason. Any corrections for its bias have referred estimated stock proportions to simple one-dimensional graphical relationships between simulation averages and known stock proportions. The Pella and Masuda: Bayesian methods for analysis of stock mixtures from genetic markers 165 simulations for harbor porpoise showed that the Bayes- ian posterior mode had considerably less bias in situa- tions for which the CML estimate was severely biased. The mode has intuitive appeal as the most frequent estimate from the posterior. Its promise for situations requiring point estimates needs to be explored by simulation in fur- ther applications. In addition, computation of the multi- dimensional mode requires some smoothing of the poste- rior samples, such as the binning used here (see footnote, Table 1). Stock grouping, followed by summing of individ- ual stock proportions for group totals, may also be neces- sary because finding posterior modes becomes more prob- lematic with large numbers of stocks. The Dirichlet distribution was the basis for probability modeling because it is a natural choice. First, it is defined for random compositions (i.e. arrays with nonnegative elements that sum to one). Second, the posterior for multinomial data can be written in closed form and is also Dirichlet. Third, the prior parameters can be inter- preted as additional data. Fourth, and last, it is easy to sample by computer. Flowever, compositional data are al- so nicely modeled with the logistic normal density (Aitchi- son, 1986; Billheimer et al., 1998), whose flexibility and relation to normal theory may have advantages in stock- mixture analysis. Use of geographical structure for the stock proportions in complex stock mixtures comprising many stocks is an area for exploration with the logistic normal and Bayesian hierarchical methods. Acknowledgments We much appreciated the opportunity to examine the ge- netic samples of harbor porpoises and steelhead trout. The harbor porpoise samples were made available by Patricia E. Rosel of the National Ocean Service, NOAA, 219 Fort Johnson Road, Charleston, SC 29412, and the steelhead trout samples were made available by Frank Thrower, National Marine Fisheries Service, Auke Bay Laboratory, Juneau, AK 99801-8626. The manuscript was revised after many insightful comments received from colleagues. Our special thanks for their reviews and ideas go to Eric Ander- son of the University of Washington, Joel Reynolds of the Alaska Department of Fish and Game, Peter Smouse of Rutgers University, Jim Murphy and Richard Wilmot of the Auke Bay Laboratory, and three anonymous reviewers of the Fishery Bulleti n. Literature cited Aitchison, J. 1986. The statistical analysis of compositional data. Chap- man & Hall, New York, NY, 416 p. Altham, P. M. E. 1984. Improving the precision of estimation by fitting a model. J. Roy. Statist. Soc. B 46:118-119. Begg, G. A., K. D. Friedland, and J. B. Pearce. 1999. Stock identification-its role in stock assessment and fisheries management: a selection of papers presented at a symposium of the 128'1’ annual meeting of the American Fisheries Society in Hartford, Connecticut, USA, 23-27 August 1998. Fish. Res. 43/1-3, 249 p. Bernardo, J. M., and F. J. Giron. 1988. A Bayesian analysis of simple mixture problems. In Bayesian statistics 3 ( J. M. Bernardo, M. H. DeGroot, D. V. Lindley, and A. F. M. Smith, eds.), p. 67-78. Oxford Univ. Press, Oxford. Best, N. G., M. K. Cowles, and S. K. Vines. 1995. CODA: convergence diagnosis and output analysis software for Gibbs sampling output, version 0.3. MRC Biostatistics Unit, Cambridge, UK, 41 p. + addendum. Billheimer, D., P. Guttorp, and W. F. Fagan 1998. Statistical analysis and interpretation of discrete compositional data. NRCSE Tech. Rep. 11, Univ. Wash- ington, Seattle, WA, 29 p. + appendix. Bishop, Y. M. M., S. E. Fienberg, and P. W. Holland. 1975. Discrete multivariate analysis: theory and practice. MIT Press, Cambridge, MA, 557 p. Bowen, B. W., A. L. Bass, A. Garcia-Rodriguez, C. E. Diez, R. van Dam, A. Bolten, K. A. Bjorndal, M. M. Miyamoto, and R. J. Ferl. 1996. Origin of hawksbill turtles in a Caribbean feeding area as indicated by genetic markers. Ecol. Applications 6:566-572. Brooks, S. P, and G. Roberts. 1999. On quantile estimation and Markov chain Monte Carlo convergence. Biometrika 86:710-722. Busby, P. J., T. C. Wainwright, G. J. Bryant, L. J. Lierheimer, R. S. Waples, F. W. Waknitz, and I. V. Lagomarsino. 1996. Status review of west coast steelhead from Washing- ton, Idaho, Oregon, and California. U.S. Dep. Commer., NOAA Tech. Memo. NMFS-NWFSC-27, 261 p. Carlin, B. P, and T. A. Louis. 1996. Bayes and empirical Bayes methods for data analysis. Chapman & Hall, New York, NY, 399 p. Debevec, E. M., R. B. Gates, M. Masuda, J. Pella, J. Reynolds, and L. W. Seeb. In press. SPAM (version 3.2): statistics program for analyz- ing stock mixtures. J. Heredity. Diebolt, J., and C. P. Robert. 1994. Estimation of finite stock-mixture distributions through Bayesian sampling. J. Roy. Statist. Soc. B 56:363-375. Efron, B., and R. J. Tibshirani. 1993. An introduction to the bootstrap. Chapman & Hall, New York, NY, 436 p. Epifanio, J. M., P. E. Smouse, C. J. Kobak, and B. L. Brown. 1995. Mitochondrial DNA divergence among populations of American shad ( Alosa sapidissima ): how much variation is enough for mixed-stock analysis? Can. J. Fish. Aquat. Sci. 52:1688-1702. Fournier, D. A., T. D. Beacham, B. E. Riddell, and C. A. Busack. 1984. Estimating stock composition in mixed stock fisheries using morphometric, meristic, and electrophoretic charac- teristics. Can. J. Fish. Aquat. Sci. 41:400-408. Gelman, A., and D. B. Rubin. 1992. Inference from iterative simulation using multiple sequences. Stat. Sci. 7:457— 511. Gelman, A., J. Carlin, H. S. Stern, and D. B. Rubin. 1995. Bayesian data analysis. Chapman & Hall, New York. NY, 526 p. Kass, R.E. (Moderator), B. P. Carlin, A. Gelman, and R. M. Neal (Panelists). 1998. Markov chain Monte Carlo in practice: a roundtable discussion. Am. Statist. 52:93-100. 166 Fishery Bulletin 99(1) Lange, K. 1997. Mathematical and statistical methods for genetic analysis. Springer, New York, NY, 265 p. Leonard, T. 1977. Bayesian simultaneous estimation for several multi- nomial distributions. Commun. Statist. -Theor. Meth. A6: 619-630. Lunneborg, C. E. 2000. Data analysis by resampling: concepts and applica- tions. Duxbury Press, Pacific Grove, CA, 568 p. + xvii. McKinnell, S., J. J. Pella, and M. L. Dahlberg. 1997. Population-specific aggregations of steelhead trout (Oncorhynchus mykiss ) in the North Pacific Ocean. Can. J. Fish. Aquat. Sci. 54:2368-2376. Michael, J. H., Jr. 1983. Contribution of cutthroat trout in headwater streams to the sea-run population. Calif. Fish Game 69(2):68-76. Millar, R. B. 1987. Maximum likelihood estimation of mixed stock fish- ery composition. Can. J. Fish. Aquat. Sci. 44:583-590. O’Connell, M., and J. M. Wright. 1997. Microsatellite DNA in fishes. Rev. Fish Biol. Fish. 7:331-363. Pella, J. J., and G. B. Milner. 1987. Use of genetic marks in stock composition analysis. In Population genetics and fisheries management (N. Ryman and F. Utter, eds.), p. 247-276. Univ. Washington Press, Seattle, WA. Pella, J., M. Masuda, and S. Nelson. 1996. Search algorithms for computing stock composition of a stock-mixture from traits of individuals by maximum likelihood. U.S. Dep. Commer., NOAATech. Memo. NMFS- AFSC-61, 68 p. Raftery, A. E., and S. M. Lewis. 1996. Implementing MCMC. In Markov chain Monte Carlo in practice (W. R. Gilks, S. Richardson, and D. J. Spiegelhal- ter, eds.), p. 115-130. Chapman & Hall, London. Rosel, P. E., S. C. France, J. Y. Wang, and T. D. Kocher. 1999. Genetic structure of harbor porpoise Phocoena pho- coena populations in the Northwest Atlantic based on mito- chondrial and nuclear markers. Mol. Ecol. 8:S41-S54. Smouse, P. E., R. S. Waples, and J. A. Tworek. 1990. A genetic stock-mixture analysis for use with incom- plete source population data. Can. J. Fish. Aquat. Sci. 47:620-634. Sutherland, M., P. W. Holland, and S. E. Fienberg. 1975. Combining Bayes and frequency approaches to esti- mate a multinomial parameter. In Studies in Bayesian econometrics and statistics in honor of Leonard J. Savage (S. Fienberg and A. Zellner, eds.), p. 585-617. North-Hol- land Publishing Company, New York, NY. Tanner, M. A., and W. Wong. 1987. The calculation of posterior distributions by data aug- mentation (with discussion). J. Am. Statist. Assoc. 82: 528-550. Waples, R. S., and D. J. Teel. 1990. Conservation genetics of Pacific salmon. I. Temporal changes in allele frequency. Conserv. Biol. 4:144-156. Xu, S., C. J. Kobak, and P. E. Smouse. 1994. Constrained least squares estimation of mixed popu- lation stock composition from mtDNA haplotype frequency data. Can. J. Fish. Aquat. Sci. 51:417-425. Appendix 1 — The empirical Bayes method for maximum likelihood estimation of the Dirichlet prior parameters from the prior predictive distribution (adapted from Lange, 1997). If the allele RFs at a locus for the ith stock, q|=(qil, . . . , qiT)', are distributed as the Dirichlet density, D(qi \ /3;,/32, . . . , (iT), and the allele counts, yi=(yil,...,yiT)', in a random sample of nj alleles have the multinomial distribution, Multfn^q,), then the prior predictive distribution (Gelman et al.. 1995) for the allele counts, obtained by integrating the product of the probability distributions, Multfn^q) and D(qt \ /3;,/3.„ . . . ,/3r), over the simplex, S(qt), is Estimation of the /3’s begins with the logarithm of Equa- tion 1, which can be viewed as the ith component of the loglikelihood or support function for the unknown /3’s, LogL,(/3„...,PT) = log V Yii- ■ ■ ■ Yit- ) + log f(/3.) - log r(?;, + /3.) + (2) i X(log r(y„ + /3, ) - log R/3,)). /=1 The total support function for all baseline samples is the sum of the individual support functions at Equation 2, FT - n/3.) 1 1 Q„ = P<,;) + [D1*1 -6(*)ir]‘1S1*’, (8) y/'(w) = —5— log Hh;) <5m is the trigamma function. The observed information ma- trix, or negative Hessian of the total support function, can be written as ' d2LogL ' D-611', (5) where plk) = the approximation of the maximum likeli- hood estimate at the Mh step, D - denotes the matrix D at Equation 6 when evaluated at P'kl, folk) _ denotes the minimum of either the scalar, b, at Equation 7 when evaluated at P'k) or the ratio, (l-e)/|l'(D(k)) _11] with s being an arbi- trary constant in (0,1), and Slk) = the vector of scores at Equation 3 evaluated at Plk). where D is a diagonal matrix with main diagonal ele- An arbitrary choice for P' 11 such as the unit column, 1, can ments be used to start the search. Appendix 2 — Minimum squared-error risk estimate of J3. with the prior mean fixed (an extension of sec. 12.2.3 of Bishop et ai., 1975). Let the baseline risk criterion be the expected value of the squared distance of any matrix estimator of baseline RFs from the true values, c 1 R( Q,Q) = y ^ n,E(q„ ~q„f Q, Q ) — 2 T 1 -X5-10 years), have long life spans (50-140 years), and are particularly vulnerable to overfish- ing. Several species off the Pacific coast currently subsist at dangerously low levels (Murray et al., 1999), and it is anticipated that rebuilding these stocks will require long periods of time (sev- eral decades in some cases). Many ecol- ogists and fishery managers see marine reserve (or no-take refuge) networks as an attractive supplement to current management systems in the conser- vation and management of rockfish stocks (Murray et al., 1999). Our study explores the potential use of harvest refugia in the specific case of shortraker iS. borealis ) and rougheye (S. aleutia- nus) rockfish in the Gulf of Alaska. We review the current management system for shortraker and rougheye rockfish in the Gulf of Alaska, describe problems confronted under this system, and suggest refuge management as a supplement to the current management regime. The comparison of future bio- mass and fishing mortality between the current system and the refuge system will be discussed on the bassis of twenty- year projections from a population dy- namics model. The potential impacts of harvest refugia on other Gulf of Alaska fisheries will also be discussed. Current management system Groundfish in the Gulf of Alaska are currently managed by the North Pacific Fishery Management Council (NPFMC). Scientists on the Gulf of Alaska ground- fish plan team compile a Stock Assess- ment and Fishery Evaluation report every year. This report includes rec- ommended acceptable biological catch (ABC) levels for each stock and stock complex (including the shortraker- rougheye complex) under the Fishery Management Plan. The NPFMC then Soh et a!.: Role of marine reserves in the management of Sebastes borealis and S. aleutianus 169 Table 1 Gulf-wide discard rates (%) for the four slope rockfish management subgroups in the commercial fishery during 1991-97. Data are from weekly production and observer reports. Management subgroup 1991 1992 1993 1994 1995 1996 1997 Pacific ocean perch 15.7 21.5 79.2 60.3 19.8 17.2 14.3 Shortraker and rougheye rockfiish 42.0 10.4 26.8 44.8 30.7 22.2 22.0 Northern rockfish — — 26.5 17.7 12.7 16.5 27.8 Other slope rockfish 20.0 29.7 48.9 65.6 72.5 75.6 52.1 considers ABC recommendations, social and economic factors, and determines total allowable catches (TACs). Shortraker and rougheye rockfish are considered as a management unit in the setting of TACs. In fishery management, fishing seasons are imposed to regulate fishing effort based on three catch levels: total al- lowable catch (TAC), acceptable biological catch (ABC), and overfishing level (OFL). There is an open, directed fishery (with allowances for bycatch) until the catches reach the TAC level. The TAC level is often lower than the ABC but frequently they have been the same for the shortraker- rougheye rockfish subgroup. Once the ABC is reached, these species are prohibited to be caught and must be returned to the sea. Finally, the fishery is closed when it reaches the overfishing level. A directed fishery for shortraker- rougheye rockfish occurred only in 1991 and 1992; the fishery has since been designated as a bycatch-only fishery. A constant exploitation-rate strategy has been used for the shortraker and rougheye rockfish subgroup. Fishing mortality for determining ABC was set equal to natural mortality ( F=M rule), which, like F0 p was generally con- sidered to be conservative (Deriso, 1987). This has been used to determine ABC since 1991, when these stocks were established as an independent management subgroup of the slope rockfish assemblage. The main reason for using this simple management procedure compared with the more sophisticated one used for Pacific ocean perch and other species is a lack of historical catch data and bio- logical information. In 1997, revised definitions of ABC and OFL replaced the Plan Team’s previous procedures (NMFS, 1996): ABC is a preliminary description of the ac- ceptable harvest (or range of harvests) for a given stock or stock complex and overfishing is defined as any amount of fishing in excess of a prescribed maximum allowable rate. For rougheye rockfish, fishing mortality rates for OFL and ABC are determined by ^OFL - ^30% ar*d F.ABC - ^40% ’ where ^X7< refers to the fishing mortality rate associated with an equilibrium level of spawning per recruit equal to X% of the equilibrium level of spawning per recruit in the absence of any fishing (Clark, 1991; 1993). For shortraker rockfish, the equations are Fofl - M and ^abc - 0-75 x M because there is less information for this species. Fish discards are a serious problem under the current management system because fish are discarded when the catch of a management category attains ABC. Discard rates for the four slope rockfish subgroups in the Gulf of Alaska were estimated as shown in Table 1 (Heifetz et al., 1997). There are two situations which result in discards at sea. One case happens under "bycatch only" status, where theoretically, only a fixed percentage of the haul-by-haul catch can be retained and the rest must be discarded to protect the stocks from a directed fishery. This “retained” catch should be available only when the bycatch status is still open. The other case occurs under prohibited status. If the total harvest exceeds ABC levels, all catch must be discarded. Shortraker and rougheye rockfish are a slow-growing, long-lived species. Maximum age for rougheye rockfish, for example, is about 140 years (Chilton and Beamish, 1982) and the von Bertalanffy growth parameter, K, for this species is about 0.05. Both shortraker and rougheye rockfish are very large fish; Lx s 55 cm for rougheye rock- fish and L^= 72 cm for shortraker rockfish. McDermott (1994) estimated natural mortality of rougheye and short- raker as 0.030-0.039 and 0.027-0.042, respectively, using the gonadosomatic index method (Gunderson, 1997a). Ma- turity analysis showed that about 50 percent are mature at 43.87 cm FL (about 20 years) for rougheye rockfish, whereas about 50 percent are mature at 44.90 cm FL for shortraker rockfish (McDermott, 1994). Rougheye rockfish undertake only limited migrations once they have recruited to the fishing ground (Gunder- son, 1997b) and this is generally considered to be true for shortraker rockfish as well. Historical catch locations show specific habitat preferences and a patchy distribu- tion for shortraker and rougheye rockfish in the Gulf of Alaska. Shortraker and rougheye rockfish present in “hot spots” can be readily overexploited by the fishing industry by targeting these spots and harvesting the fish through specialized harvesting skills. Refuge management Harvest refugia are areas protected from some or all fish- ing activities. The general objectives of creating harvest refugia can be discussed with respect to the fish commu- nity, fishery practices, and ecological aspects of fish habitat 170 Fishery Bulletin 99(1 ) (Lindeboom, 1995; Bohnsack and Ault, 1996; Allison et al., 1998; Lauck et al., 1998; Yoklavich (ed.), 1998). Con- servation of ecological and demographic characteristics, protection of fish habitat, and ensuring recruitment supply under environmental uncertainty and manage- ment shortcomings are among the general benefits of harvest refugia. Harvest refugia may also serve as con- trol communities for comparison with nonrefuge areas, allowing us to determine the effects of exploitation on the ecological community and to disentangle the effects of fishing and environmental change. However, we still lack detailed and scientifically de- fensible knowledge regarding the effects of harvest refugia. Because of a lack of information on spatial processes for fish populations, there are only a few quantitative assessments of the effects of refugia on current yields and on future abundance of populations (Polacheck, 1990; Roberts and Polunin, 1991; DeMar- tini, 1993; Holland and Brazee, 1996). The primary in- formation needed for effective design of harvest refugia includes spatial structure of the population, population dynamics, larval drift trajectories, movements of ben- thic life stages, descriptions of fish habitats, and a dis- tribution of fishing effort. Limited amounts of such in- formation are available for shortraker and rougheye rockfish. Therefore, the establishment of harvest refu- gia in our study aims to improve the efficacy of the cur- rent harvest-rate strategy by providing a safer man- agement scheme by which to conserve shortraker and rougheye spawning populations from possible deple- tion. This scheme can be used in combination with the current management strategy as an additional control on harvesting and as an attempt to solve problems that characterize the current management policy. Bycatch season Shortraker rockfish Rougheye rockfish (1041 data points) (1285 data points) 40 • 40 ■ 30 30 • • 20 20 • :! Jlfc. _ . ^ 0 05 1 1.5 0 05 1 XI O 03 O Prohibited season 13 03 Shortraker rockfish Rougheye rockfish X (234 data points) (384 data points) 40 40 30 30 20 20 10 10 itfln*9 “* • • 0 0 0 5 1 0 0 5 1 Catch proportion Figure 1 Haul catch of shortraker and rougheye rockfish and the pro- portion of the total catch made up by these species during the bycatch and prohibited season during 1991-96. Methods Documentation of fishery targeting practices Current management allows vessels to “top-off” their catch during the bycatch season. The purpose of the “bycatch season” is to protect the population from a directed fishery and to allow for “natural” bycatch in other directed fish- eries. In theory, targeting of a species can be avoided by allowing retention of only a certain fraction of that spe- cies, haul by haul. Such bycatch management measures have failed in instances where fishing deliberately targets bycatch species when their natural bycatch levels are lower than the specified limit. For example, if Pacific ocean perch (POP) is the target species, then vessels can retain bycatch species such as shortraker and rougheye rockfish up to a certain percentage of the total POP catch during the vessel trip. If they prefer, however, they could first fill their bins with POP and then target shortraker and rougheye rock- fish. Through “topping-off,” the fishery, in essence, becomes a directed fishery for the bycatch species. To examine the targeting practice for these bycatch spe- cies, historical fishing seasons were divided into an allow- able bycatch season and a prohibited season, and catch information for each fishing season was compared. Dur- ing the prohibited season, neither high catches nor high proportions of shortraker and rougheye rockfish occurred (Fig. 1). During the bycatch season, however, both high catches and high proportions were observed. This compar- ison supports the hypothesis that vessels do target high- value species and that their fishing practices are focused on specific “hot spot” areas. Higher catch-per-unit-of-effort (CPUE) points in regions of short-tow duration (Fig. 2) provide further evidence of the practice of targeting these species. As a result, serial targeting in productive areas would induce early attainment of the ABC and result in discarding afterwards. Design of harvest refugia Basic design components of marine refugia are location, size, shape, and number. These components depend on the purpose of the harvest refugia. In our application, the main goal of establishing harvest refugia for shortraker and rougheye rockfish is to protect adult fish from serial overfishing in areas of high short- Soh et al: Role of marine reserves in the management of Sebastes borealis and 5. aieutianus 171 Shortraker rockfish 60 , 50 - 40 30 20 •* 03 § Rougbeye rockfish 0 2 4 6 8 10 12 Duration (hour) Figure 2 Distribution of haul CPUE (t/haul) versus tow dura- tion for shortraker and rougheye rockfish in the Gulf of Alaska (domestic observer data for trawlers during 1987-96). raker and rougheye production. Precautions are needed because these stocks are very slow growing, long-lived, and patchy in their distribution. Once “hot spot” stocks are depleted, several decades may be required for them to re- cover (Francis, 1988). Because this study aimed to develop a method for an objective design of refuge networks and to evaluate the feasibility of alternative management sce- narios, no specific area was suggested for practical usage. Size of refugia was determined by using quantiles from cumulated catch data. The selection of the size for har- vest refugia depends on the goals of refuge management in general. Here, three sizes were arbitrarily selected and the effects of refuge size on the fish community and fishery practices were then evaluated. These refuge sizes are referred to as SSR (small-size refugia) MSR (middle- size refugia), and LSR (large-size refugia) and include all 1987-96 catches that exceeded the cutoff points for the 99.8%, 99.5%, and 99.0% quantiles, respectively (Fig. 3). The SSR, MSR, and LSR refugia defined in this manner represented 2.4, 5.4, and 9.5% of the total habitat occupied by shortraker or rougheye rockfish, or by both species. Because the fishery data only provided net retrieval lo- cations, the exact catch locations of each tow were un- known. This lack of starting-tow locations led us to postu- late a circular area around the retrieval point to represent the occurrence of the catch. Although the average tow dis- tance was about 18 km, the width of the continental slope region containing high catches for both species was usual- ly less than 15 km, and trawling was generally conducted along depth contours. Therefore, the basic cell of the har- vest refugium was arbitrarily chosen as a 20-km diameter circle, a group of which comprised the refugia for each spe- cies. Two species-specific harvest refugia were constructed by using GIS. Because shortraker and rougheye rockfish are currently managed together, the two independent re- serves were united in the final design of the harvest refu- gia. Once a GIS database was established, it became pos- sible to visualize the impact of different design criteria on the spatial distribution of the refuge network almost instantaneously. Modeling population behavior under refuge management There were two main goals for the assessment and projec- tion of shortraker and rougheye rockfish stock conditions. The first was to examine the current stock status, to proj- ect future biomass, and to evaluate uncertainties in the key parameters under the current management system. The other was to examine the effectiveness of harvest refugia in the management of these species and compare it with the current management system. We employed a population dynamics model to address these goals (Soh, 1998). The model was coded by using AD Model Builder (Fournier, 1994). Determination of parameter uncertain- ties was based on the Markov Chain Monte Carlo method implemented as part of the AD Model Builder software libraries. Two separate time schedules were used in the modeling process: an assessment section (1961-1996) and a projection section (1997-2016). Inputs to the model included reconstructed catch histories (1961-1996), priors of the parameters, survey biomass estimates, and fixed fishing mortality rates. Two conditions of fishing mortality were selected for fu- ture projection because fishing mortalities in establishing ABC and actual fishing mortalities at sea were not equal. Fishing mortalities used to calculate recommended ABC level were F=0.023 for shortraker and F= 0.025 for rough- eye rockfish (which were natural mortalities determined by NMFS and referred to as “F for ABC”) during 1991-96. These ABC levels were not attained, however, and actual average fishing mortalities estimated by the model were F=0.G63 for shortraker and F=G.015 for rougheye rockfish (referred to as “actual F”) during 1991-96. If no alterna- tive actions were applied to the current management pol- icy, then projection results under the “actual F” scenario would be more realistic than those under “F for ABC”. In a refuge management system, no fishing was allowed in the refuge areas, and a fixed exploitation rate was ap- plied to the harvestable areas. A major difference between projections of the current management and those of ref- uge management systems is that refuge management in- corporates an adjustment for a discard rate of 29.5% (Ta- ble 1) into the system on the assumption that no discards will be necessary under refuge management. As a result. 172 Fishery Bulletin 99(1) Figure 3 Three sizes of marine refugia are designed and the radius of the basic circle is 10 km. SSR = small-size refugia; MSR = middle-size refugia, and LSR = large-size refugia. the current actual level of ABC (which includes discards), reduced by average discards, will be applied to future rec- ommended ABC under refuge management. That is, rec- ommended ABC in the harvestable areas will be the rec- ommended ABC under current management multiplied by (1.0 - discard rate), assuming that this average discard rate is constant throughout the future projection years. This reduced ABC does not affect the fishing industry’s actual harvest level if they can keep all the catch until ABC is attained. Considering that they are a very large and valuable spe- cies, discards of shortraker and rougheye rockfish are as- sumed to result primarily from season limits, rather than size limits. Skippers consistently develop their area-spe- cific skills to maximize total income, and high catch rates can result in a short fishing season (Hilborn, 1985). Be- cause harvest refugia in our study were established in high catch-rate zones, or preference zones for shortraker and rougheye rockfish, the harvesting period was assumed to be sufficiently prolonged where shortraker and rough- eye rockfish were less dense and regulatory-induced dis- cards would be virtually eliminated as a result. Model description A stock reduction analysis model (Kimura et al., 1984 } with one-parameter-based Beverton and Holt stock-recruit- ment relationship (Kimura, 1988) was applied for the stock assessment and future projections. Both analyses were combined to estimate the following relevant parameters: Soh et al.: Role of marine reserves in the management of Sebostes borealis and S. aleutianus 173 i j A Q d a M A ~'a.3C P (0 k s (Jobs Qpred SC or FC gobs B Bref g non ref R Rref gnonref years for the assessment period from 1961 to 1996; years for the future projection period from 1997 to 2016; weighting factor between survey and fishery information for the combined catch history { = 1.0 when using survey data alone, 0.0 when using fishery data alone); survey gear efficiency; average annual discard rate; proportion of total biomass in refugia, esti- mated by cumulated catch data; natural mortality rate; recruitment strength in the Beverton and Holt model; fixed exploitation rate for determining ABC; Brody coefficient; annual growth rate of a fish in weight, = ; age at recruitment; survival rate (s- = e~[Fl+M))\ observed catches for i and predetermined pre- dicted catches for j; predicted catches based on catch equation; reconstructed catches based on survey (S) or fishery (F) information; biomass index from surveys; Gulf-wide total predicted biomass; predicted biomass in refugia; predicted biomass in nonrefugia; Gulf-wide total recruits; predicted recruits in refugia; predicted recruits in nonrefugia. mass level was assumed to be in an equilibrium state. As a result, initial biomass and the annual recruitment bio- mass to the unexploited stock can be shown as follows: B1 - B0 = B 1961 R1 = R19S1 = B 1 s„ + p ( s„ s0 ) 1- p CO s0 where survival rate s0 = e~M. However, biomass in the second year, B2, can be defined from the SRA model as B2 — B-[9g2 - ( l+p)SjB| pSjSqBq + R2 p&JSj-ffj, where Bx = B0 and survival rate s- = eAFl+M). Rt is the recruitment biomass at the stock size B( and is modeled by using a stochastic Beverton and Holt stock- recruitment relationship: R; = R,ef' , e( - AhCfcr;) for 2 k, where k = the recruitment age (assumed k=30). Stock assessment Catch histories, based on survey and fishery information, were reconstructed for both shortraker and rougheye rod; fish (Soh, 1998). Because surrey data were available from 1961 and fishery data from 1977, survey information was used in the construction of catch history for the years 1961-76, whereas an information weighting factor (A) was applied to combine the two catch histories for the years of 1977-90. Since 1991, independent catch data for the shortraker-rough- eye roekfish subgroup have been available. As a result, the combined catch history, which was used as observed catches (Cofcs) in the model, can be described as follows: /~iobs ^1961 SC ^1991 fiobs '“'1976 sc '-'1976 /^obs '“'1977 ^SCl977 +( 1 - A)FC1977 , r^obs '“'1990 AsC1990 + ( 1-A)fC3990 obs ^1991 Fc ^ 1991 f^obs _ 1996 F c ^1996 13 1 is defined as an initial biomass at the beginning of 1961 and R1 is the recruitment in 1961. The pristine bio- Since 1963, biomass has been able to be estimated by using the SRA model based on the Deriso’s delay-differ- ence equation (1980): Bj = (1 + p)s,_iB,_i - ps ,-\S i-%B i + Rt - pwst _1Ri j for i > 3. Future projection of biomass and recruitment The following biomass and recruitment models were applied throughout the projection period under the cur- rent management system: B, = (1 +p)sl_lBi_l - ps j^s j_.2Bj_2 + Rj - pcosJ_lR]_l B_l± R = R Jk e' ,f ~ N(0,o2). l-Al-Ai l B„ ) Under the current management system, FABC was fixed during the projection period. Predetermined predicted catches (“target catches” or “observed catches”) could then be calculated from 174 Fishery Bulletin 99(1 ) Cobs _ 77* D j —^ABC^J- The model stock was divided into two subcomponents under the refuge management system: one for refugia and the other for harvestable areas. Stock within refugia was based on the proportion of Gulf-wide biomass in refugia, a, using the spatial distribution of the historical catches to represent the spatial distribution of the actual exploitable biomass. Under the refuge management system, target or observed catches were fixed to the level of ABC under the current management system minus the average annual discards: C? = a-d)FABCBJ, where d = the average annual discard rate for shortraker and rougheye rockfish. Gulf-wide recruitment was allocated into refugia and non- refugia areas by using 2(77 log Q Q 2(7 ? log M Y[ ,Mnnor)\ 2o: [ r a )\ log A prior / J +u- 3, B'fi = ( 1 + p)s„B'l '\ ~ Ps lB'fi2 + Rrf - p(osHRr;\ B"""rcl =(1+p)s,_1B;T/ - PSj-iSj-iB'iT1 + R7""vr ~ 1 R"T'. All fishing mortalities were estimated by using the New- ton-Raphson method (Taylor and Mann, 1983). According to Bayes theorem, posterior log-likelihood is the sum of prior and experimental log-likelihood, logL(0|X) = logL( 9) + logL(X | 9). Therefore, posterior values of the oth- er parameters were evaluated by minimizing the total negative log-likelihood: Fixed exploitation-rate strategy Under an UF for ABC” fishing mortality (F=0.023), the ending biomass of shortraker rockfish would increase by about 800 metric tons (t) in twenty years, but the stock would decrease by over 7000 t if current fishing intensity (actual F=0.063) were continued (Table 2). Rougheye rock- fish declined under both UF for ABC” and “actual F" fishing scenarios (either F=0.025 or F=0.015) by about 5000-12,000 t. Although the actual fishing mortality was much lower than the fishing intensity for the recom- mended ABC, rougheye stocks still declined. Annual recruits of shortraker and rougheye rockfish ranged from about 1% to 3% of annual biomass based on the recruitment scenario, with a Beverton and Holt shape parameter A=0. 889. For both scenarios of “F for ABC” and “actual F” fishing conditions, similar recruitment trends occurred in both species. Because the age at recruitment was assumed to be 30 (Nelson, 1986), future recruitment was not affected by the future fishing mortality schedule during the projection years. Recruitment strength varied Soh et al.: Role of marine reserves in the management of Sebastes borealis and S. aleutianus 175 relatively little (Fig. 4) and was greatest during the early 1990s. The estimates of posterior parameter values are shown in Table 3. The mean value of the posterior Q at maximum likelihood was about 0.78 for both species. This value of Q implies underestimation of the actual biomass by the sur- veys and is consistent with submersible observations by Krieger and Ito (1999), who concluded that the above-bot- tom distribution of shortraker and rougheye rockfish, to- gether with their preference for steep-slope boulder hab- itats, would result in a value of Q that was less than 1.0. Natural mortality was estimated to be closer to prior expectations as was the shape parameter (A) in the Be- verton and Holt recruitment curve (Beverton and Holt, 1957). Presumably the closeness of prior values and pos- terior estimates reflects the lack of information about the mortality and stock-recruitment relationship. The lower posterior value of A for shortraker rockfish represents more emphasis on fishery information and the increased value of A for rougheye rockfish represents more emphasis on survey information in the combination of the two catch histories during 1977-90. Survey catch data show that shortraker rockfish are found primarily along the continental slope, whereas rougheye rockfish are distributed more broadly on the continental shelf. Generally, survey coverage in the slope region is relative- ly sparse, and data are consequently acknowledged to be insufficient. In this regard, more emphasis on fishery in- formation for shortraker rockfish is desirable. The 1996 biomass represented decreases to about 46% and 69% of pristine level (B1961) for shortraker and rougheye rock- fish, respectively. Refugia management strategy The effects of refuge size on ending biomass, and differ- ences in fishing intensity between current management and refuge management were considered in the projection model of refuge management. Ending biomass in twenty- year future projections and average twenty-year fishing mortality rates were compared for two levels of F (“F for ABC” and “actual F”) and three refuge sizes. - - - Shortraker rockfish Rougheye rockfish 900 “ 700 W '5 | 600 400 1 ‘ ‘ 1 1940 1960 1980 2000 2020 Year Figure 4 Predicted recruitment trajectories for shortraker and rougheye rockfish using the Beverton and Holt model with shape parameter A = 0.889. Table 2 Starting and ending biomasses (t) of shortraker and rough- eye rockfish under the current management regime, based on a 20-year future simulation (1997-2016). “F for ABC” is the fishing mortality currently used to calculate recom- mended ABC, and actual average fishing mortality during 1991-96, estimated by stock reduction analysis, is denoted as “actual F'. Shortraker rockfish Rougheye rockfish F for ABC Actual F F for ABC Actual F Year (F= 0.023) (F= 0.063) (F=0.025) (F=0.015) 1997 20,061 20,061 55,896 55,896 2016 20,847 12,623 43,351 50,443 Table 3 Estimates of posterior parameter values and their standard errors. Values in parentheses are standard errors. (B1996=biomass in 1996, M=natural mortality, Q=survey catchability coefficient, A=shape parameter of Beverton and Holt recruitment curve, A=weighting parameter for survey information in the reconstruction of catch history, and depletion level \B i9g6IBl96l )). Shortraker rockfish Rougheye rockfish Parameter Prior Posterior Prior Posterior ■®1996 bA — 20,667 (10,260) — 56,040 (30,982 M 0.030 (0.2) 0.031 (0.006) 0.025 (0.2) 0.025 (0.005) Q 1 (0.6) 0.776 (0.325) 1 (0.6) 0.785 (0.407) A 0.889 (0.2) 0.889 (0.178) 0.889 (0.2) 0.889 (0.178) A 0.5 0.266(0.565) 0.5 0.554(0.707) ^1996^1961 - 0.464 (0.123) — 0.685 (0.120) 176 Fishery Bulletin 99(1) Table 4 Summary of simulations showing ending biomasses (t) and average fishing mortalities from 20-year future projections. Three refuge sizes and two fishing intensity schedules are compared under current and refuge management systems. F values under refuge management pertain only to areas outside of refugia. (SSR=small-size refugia; MSR=middle-size refugia; LSR=large- size refugia) Ending biomass (year=2016) Average F (during 1997-2016) F for ABC Actual F F for ABC Actual F (F= 0.023) (F=0.063) (F=0.023) (F=0.063) Species Current Refuge Current Refuge Current Refuge Current Refuge Shortraker rockfish SSR 23,442 18,270 0.018 0.043 MSR 20,847 23,437 12,623 18,236 0.023 0.022 0.063 0.055 LSR 23,425 18,153 0.030 0.090 Rougheye rockfish SSR 50,187 54,764 0.019 0.012 MSR 43,351 50,176 50,444 54,760 0.025 0.023 0.015 0.014 LSR 50,160 54,753 0.030 0.018 Table 5 Percentage of Gulf-wide catches in middle-size refugia ( MSR), by species and depth, from domestic observer data ( 1987-1996 ). POP = Pacific Ocean perch. Depth (m) Shortraker rockfish Rougheye rockfish POP Thornyhead Northern rockfish Dusky rockfish Sablefish Rex sole 100-713 28 29 15 12 12 17 6 12 200-713 27 28 12 12 1 1 5 6 300-713 23 25 2 9 0 0 3 1 Incorporation of harvest refugia into current manage- ment for shortraker and rougheye rockfish resulted in in- creased ending biomass for both levels of F (Table 4). Un- der the “F for ABC” scenario, ending biomass increased about 12% for shortraker rockfish and about 16% for rougheye rockfish with refuge management. Under the “actual F” scenario, ending biomass increased about 44% for shortraker and about 8% for rougheye rockfish when refugia were employed. Figure 5 shows the future biomass trajectories for the two management systems. For both fishing intensities, higher Gulf-wide biomass levels were projected in refuge management for both species. Size differences for harvest refugia had relatively little impact on the ending biomass estimates. Table 4 shows similar ending biomass in refuge management irrespec- tive of refuge sizes for both “F for ABC” and “actual F” scenarios. This is because the same catch removal was ap- plied, irrespective of refuge size. Impacts of harvest refugia on other target fisheries Table 5 summarizes the proportion of catches within refu- gia compared with those from the whole Gulf for major species in the slope region. About 30% of the Gulf-wide catches of shortraker and rougheye rockfish occurred at depths greater than 100 m in MSR. Out of this, over 80% still remained in MSR in depths greater than 300 m. Dusky rockfish had the next highest proportion in MSR but their distribution was primarily confined to the 100-200 m depth interval. Shortspine thornyhead remained at 9% of the Gulf-wide catch in refugia deeper than 300 m. The catch proportions of other major fishes in refugia deeper than 300 m declined to 3% or less. Clearly, if harvest refugia can be established with depth limits greater than 300 m, then the impacts of no-take zones for shortraker and rough- eye rockfish on other fisheries seem minor. Discussion Understanding the changes in fishing effort under refuge management is important for its practical application. In refuge management, fishing is allowed only in harvestable areas where fish density is assumed to be less than in refuge areas. However, high levels of fishing intensity out- side of refugia are to be avoided because high fishing effort Soh et al.: Role of marine reserves in the management of Sebastes borealis and S. aleutianus 17 7 Shortraker rockfish E o bo Rougheye rockfish - — — Historical biomass Current (F for ABC) Refugia (F for ABC) -* — *- Current (Actual F) Refugia (Actual F) • Survey index / Q Figure 5 Future Gulf-wide biomass projections for shortraker and roughete rockfish under current management and under refuge management. “F for ABC” is F = 0.023 for shortraker and F = 0.025 for rougheye rockfish; and "actual F" is F = 0.063 for shortraker and rougheye rockfish. Q is a survey catchability coefficient. could deplete stocks in nonrefuge areas if adults are sedentary. The three sizes of refugia resulted in different levels of fishing mortality out- side of refugia (Table 4). In the case of middle-size refugia (MSR), fishing mortalities outside of refugia differed little from current Gulf-wide levels. No increased fishing intensity outside of refugia is required under the MSR- based refuge management system, even though harvestable areas are reduced and density is lowered in these areas. It should be remembered, however, that in the projection of refuge man- agement, annual quotas were set to the current annual quotas for short- raker and rougheye rockfish minus dis- cards, and fishermen were assumed to keep all bycatch until quotas were at- tained. In larger-size refugia (LSR), in- creased fishing mortality was required in harvestable areas to attain the same amount of quota. Theoretically, this may be acceptable because harvestable areas were reduced and fish density in these areas was less than that out- side MSR. In the case of small-size re- fugia (SSR), fishing mortalities were re- duced compared with those under the current management system. Reduced fishing intensity in harvestable areas seem counter-intuitive compared with fishing efforts under the current man- agement system, but this results from applying a reduced annual quota in the refuge management system. However, reduced annual quotas do not mean re- ductions in landings for the fishing in- dustry because it will probably be pos- sible to keep all catches rather than discard some of them. In this example, lower fishing effort outside of refugia, attainment of the same level of landings for the industry, reduced discards, and increased ending biomass were achieved in refuge management by establishing medium-size or small-size harvest refugia. Several additional benefits of refuge management for shortraker and rougheye rockfish can be suggested. Harvest refugia can eliminate serial overfishing of the substocks in areas of high densities, i.e. “hot spot” areas. Larger short- raker and rougheye rockfish have been intensively removed during the past two decades. Length-frequency distribu- tions for shortraker and rougheye rockfish show a reduc- tion of larger fish between foreign (1975-1985) and domes- tic (1987-1996) observer harvesting periods (Soh, 1998). As shown in Figure 5, shortraker rockfish are being overex- ploited if the goal is to maintain current stock levels. This overexploitation occurs because shortraker and rougheye rockfish are managed together as one subgroup with one combined quota, but shortraker rockfish are generally larg- er and commercially more valuable than rougheye rockfish. Discards are an important problem and should be re- duced. However, under the current management system, it may be difficult to reduce discards at sea because skippers search for areas with higher catch rates in order to maxi- mize their fishing success. Although the shortraker-rough- eye rockfish fishery has been “bycatch only” since 1993, targeting and retaining fish under “topping-off” strategies have continued during the bycatch season (Fig. 1). This type of targeting practice by fisheries allows earlier at- tainment of ABC and a prolonged discarding period. Hot spots, or high density zones, for shortraker and rougheye rockfish can be considered as control areas for monitoring and sustaining the surrounding fish stocks. Because we had no information regarding the life stages of these two species prior to recruitment to their fishing grounds, we were at least interested in maintaining ad- 178 Fishery Bulletin 99(1 ) equate spawner biomass in the hot spots. Adult shortraker and rougheye rockfish are reported to live on habitats con- taining steep slopes (>20°) and numerous boulders (Krieg- er and Ito, 1999). If these habitats are degraded by fishing activity and if stocks in these areas become reduced, then Gulf-wide stock stability might be at risk. By establishing harvest refugia, essential fish habitat for adult shortraker and rougheye rockfish can be conserved. Harvest refugia can also be used to ensure an appro- priate proportion of shortraker and rougheye rockfish in catches. Because these two species are managed as one subgroup of the slope rockfish assemblage, the fishing sea- son is subject to the attainment of the combined level of two TACs. This management strategy has led to a discrep- ancy between the intended ABCs and the actual catches of the two species. Trawl survey estimates of exploitable biomass and recommended ABC show species proportions (shortraker to rougheye rockfish) of 3:7 and 4:6, respec- tively; whereas, commercial catches show a 6:4 ratio. Obvi- ously, this reversed proportion of exploitation between the two species will incur a faster decrease of the shortraker stock than is expected. Separate management of these two species may be ap- propriate but because of their overlapping habitat, there could still be excessive waste through discards. Out of all the trawl hauls with either shortraker or rougheye rock- fish sampled by domestic observers during 1987-97, about 50% of the total hauls contained both species. By weight, such hauls containing both species accounted for about 72% of the total catch of shortraker and 78% of the total catch of rougheye rockfish. This finding suggests consid- erable habitat overlap, even though survey data indicate that individual species have somewhat different bathy- metric and habitat preferences (Soh, 1998; Ito, 1999). Nev- ertheless, refuge networks can probably be designed to take advantage of differences in the habitat preference of shortraker and rougheye rockfish to achieve the desired proportions of these species in commercial catches. In summary, refuge management can potentially be used to solve a number of management problems. By se- lecting a moderate size, number, and spatial distribution for harvest refugia, the apparent short-term costs or risks can be reduced as much as possible. A solution to the prob- lems of bycatch waste, protection from stock depletion, preservation of essential fish habitat, preservation of the existing spatial distribution of substocks, and prevention of serial depletion of substocks can potentially be achieved through refuge management. Uncertainties in the results of our study resulted from the fact that populations may behave and adapt differ- ently under different environments with different fishing pressures. Furthermore, there is an uncertainty regarding the sustainability of recruitment owing to a lack of knowl- edge of larval and juvenile stages and the impact of an- nual or interdecadal variability in oceanic conditions. For practical application of a refuge system, additional research is still needed. Optimum shape, location, and size of refugia can best be achieved through adaptive management, by using monitoring and observations from existing refugia to redesign the network. Potential costs or risks, or both, are also expected. The fishing industry needs to be involved in gaining a full understanding of any loss incurred by the displacement of fishing effort from specified areas. Harvest refugia may prove to be un- successful if illegal fishing cannot be controlled. Investi- gation as to how to prevent such illegal behavior and to promote a practical and effective monitoring system still needs to be undertaken. Acknowledgments This paper is based on a dissertation by S. K. Soh, in partial fulfillment of the requirements of a Ph.D. at the University of Washington. The research was supported by grants from Washington Sea Grant and the Joint Institute for the Study of the Atmosphere and Ocean ( JISAO) under NOAA cooperative agreement no. NA67RJ0155, contribu- tion no. 668. Literature cited Allison, G. W., J. Lubchenco, and M. H. Carr. 1998. Marine reserves are necessary but not sufficient for marine conservation. Ecol. Appl. 8 (suppl. 1):S79-S92. Beverton, R. J. H., and S. J. Holt. 1957. On the dynamics of exploited fish populations. Fish- eries Investment Series 2, vol. 19. U.K. Ministry of Agri- culture and Fisheries, London. Bohnsack, J. A. and J. S. Ault. 1996. Management strategies to conserve marine biodiver- sity. Oceanography 9:73-82. 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. Pub. Fish. Aquat. Sci. 60, 102 p. Clark, W. G. 1991. Groundfish exploitation rates based on life history parameters. Can. J. Fish. Aquat. Sci. 48:734-750. 1993. The effects of recruitment variability on the choice of a target level of spawning biomass per recruit. In Proceed- ings of the international symposium on management strat- egies for exploited populations (G. Kruse, D. M. Eggers, R. J. Marasco, C. Pautzke, and T. J. Quinn II, eds.), p. 233-246. Alaska Sea Grant College Program Report 93-02. DeMartini, E. E. 1993. Modeling the potential of fishery reserves for manag- ing coral reef fishes. Fish. Bull. 91:414-427. Deriso, R. B. 1980. Harvesting strategies and parameter estimation for an age-structured model. Can. J. Fish. Aquat. Sci. 37:268-282. 1987. Optimal F0 Y criteria and their relationship to maxi- mum sustainable yield. Can J. Fish. Aquat. Sci. 44(suppl. 2):339-348. Fournier, D. A. 1994. AUTODIF, A C++ array language extension with auto- matic differentiation for use in nonlinear modeling and sta- tistics. Otter Research Ltd. P.O.BOX 2040, Sidney, British Columbia V8L 3S3, Canada. Francis, R. C. 1986. Two fisheries biology problems in West Coast ground- fish management. N. Am. J. Fish. Manage. 6:453-462. Soh et ai.: Role of marine reserves in the management of Sebastes borealis and S. aleutianus 179 Gunderson, D. R. 1997a. Trade-off between reproductive effort and adult sur- vival in oviparous and viviparous fishes. Can J. Fish. Aquat. Sci. 54:990-998. 1997b. Spatial patterns in the dynamics of slope rockfish stocks and their implications for management. Fish. Bull. 95:219-230. Heifetz, J., J. N. Ianelli, and D. M. Clausen. 1997. Slope rockfish. In Stock assessment and fishery evalua- tion report for the groundfish resources of the Gulf of Alaska, p. 247-288. [Available from North Pacific Fishery Management Council, 605 W 4th Ave, Suite 306, Anchorage, AK 99501.1 Hilborn, R. 1985. Fleet dynamics and individual variation: why some people catch more fish than others. Can. J. Fish. Aquat. Sci. 42:2-13. Holland, D. S., and R. J. Brazee. 1996. Marine reserves for fisheries management. Marine Resource Economics. 11:157-171. Ito, D. 1999. Assessing shortraker and rougheye rockfishes in the Gulf of Alaska: addressing a problem of habitat specificity and sampling capability. Ph.D. diss., Univ. Washington, Seattle, WA, 205 p. Kimura, D. K. 1988. Stock-recruitment curves as used in the stock-reduc- tion analysis model. J. Cons. Int. Explor. Mer 44:253-258. Kimura, D. K., J. W. Balsiger, and D. H. Ito. 1984. Generalized stock reduction analysis. Can. J. Fish. Aquat. Sci. 41:1325-1333. Krieger, K. and D. Ito 1999. Distribution and abundance of shortraker rockfish, Sebastes borealis, and rougheye rockfish, S. aleutinus, deter- mined from a manned submersible. Fish. Bull. 97:264-272. Lauck, T., C. W. Clark, M. Mangel, and G. R. Munro. 1998. Implementing the precautionary principle in fisheries management through marine reserves. Ecological Appli- cations 8(1):S72-S78. Lindeboom, H. J. 1995. Protected areas in the North Sea: an absolute need for future marine research. Helgolander Meeresunters 49:591-602. McDermott, S. F. 1994. Reproductive biology of rougheye and shortraker rockfish, Sebastes aleutianus and Sebastes borealis. M.S. thesis, Univ. Washington, Seattle, WA, 76 p. Murray S. N., R. F. Ambrose, J. A. Bohnsack, L. W. Botsford, M. H. Carr, G. E. Davis, P. K. Dayton, D. Gotshall, D. R. Gunderson, M. A. Hixon, J. Lubchenco, M. Mangel, A. MacCall, D. A. McArdle, J. C. Ogden, J. Roughgardon, R. M. Starr, M. J. Tegner, and M. M. Yoklavich. 1999. No-take reserve networks: sustaining fishery popula- tions and marine ecosystems. Fisheries 24 ( 1 1 ): 1 1—25. NMFS (National Marine Fisheries Service). 1996. Environmental assessment/regulatory impact review for Amendment 44 to the Fishery Management Plan for the groundfish fishery of the Bering Sea and Aleutian Island area and Amendment 44 to the Fishery Management Plan for the groundfish fishery of the Gulf of Alaska to redefine acceptable biological catch and overfishing, Appendix B. Alaska Fisheries Science Center, NMFS, 7600 Sand Point Way NE„ Seattle, WA 98115-0070, 24 p. Nelson, B. D. 1986. Population parameters of rougheye rockfish (Sebastes aleutianus). M.S. thesis, Univ. Alaska, Juneau, AK, 103 p. Polacheck, T. 1990. Year around closed areas as a management tool. Nat- ural Resource Modeling 4(3):327-354. Roberts, C. and N. Polunin. 1993. Marine reserves: simple solutions to managing com- plex fisheries. Ambio 22(6):363-368. Soh, S. K. 1998. The use of harvest refugia in the management of short- raker and rougheye rockfish ( Sebastes borealis! Sebastes aleutianus) in the Gulf of Alaska. Ph.D. diss., Univ. Wash- ington, Seattle, WA, 194 p. Taylor, A. E., and W. R. Mann. 1983. Advanced calculus. John Wiley and Sons, New York, NY. Yoklavich, M., editor. 1998. Marine harvest refugia for west coast rockfish: a workshop. U.S. Dep. Commer, NOAA Tech. Memo. NMFS- SWFSC-255, 159 p. 180 Abstract— The diet of Plectropomus leopardus (Serranidae, Lacepede 1802) was examined on two pairs of reefs in the Cairns Section of the Great Barrier Reef Marine Park, Australia. For both pairs, one reef was open to fishing and the other had been closed to fishing for eight years; however zoning appeared to be ineffective as there was no differ- ence in the size structure of leopard cor- algrouper populations on either open or closed reefs. Two fishing methods were used to sample reefs concurrently, and the size structure and diet of P. leopardus that were speared ran- domly (n= 587) were compared to sam- ples caught by line (n= 85). Adult P. leopardus were highly piscivorous (96% of prey was fish by number), and two families of fishes, Pomacentridae and Labridae, composed approximately half of their diet (index of relative impor- tance=48.4%). Numerical composition of fish in the diet varied significantly among reefs, but there were no patterns related to reef closures when fish prey were classified by taxa or by their habi- tat. Fishes categorized as living in the demersal reef habitat were the domi- nant prey consumed, followed by mid- water fishes. When the data from reefs were pooled, the abundance of families in the diet differed between locations (north and south) but not between fish- ing zones. Dietary overlap was high between the different fishing zones and was very high in relation to naturally occurring spatial and temporal vari- ability in the diet of P. leopardus found in other studies. With line fishing larger and hungry fish are caught, and the few data on natural prey suggest ten- tatively that line catches are biased toward P. leopardus feeding on pelagic fishes. The coral reefs and surrounding waters provide the major food source of P. leopardus , whereas sandy areas are much less important. Our data suggest that the coral trout fishery is resilient to changes in abundances of particular prey species because the diet of P. leop- ardus is broad and because the two major prey families are diverse and abundant on coral reefs. Manuscript accepted 25 April 2000. Fish. Bull. 99:180-192 (2001). The diet of the large coral reef serranid Plectropomus leopardus in two fishing zones on the Great Barrier Reef, Australia Jill St John Garry R. Russ Department of Marine Biology James Cook University Townsville, Queensland 4811 Australia Present address (for J. St John): Western Australian Marine Research Laboratories PO Box 20 North Beach, Western Australia 6020, Australia E-mail address (for J. SUohn): jstjohn@fish.wa.gov.au Ian W. Brown Southern Fisheries Centre PO. Box Deception Bay, Queensland 4508 Australia Lyle C. Squire 27 Barrett St. Bungalow, Queensland 4870 Australia Groupers (Serranidae) are an impor- tant fishery resource throughout the tropics (Ralston, 1987; Heemstra and Randall, 1993; Polunin and Roberts, 1996) and are a favored target species of fishermen (Bohnsack, 1982; Randall, 1987; Koslow et ah, 1988; Russ and Alcala, 1996). Like many other large piscine predators, the life history char- acteristics of groupers ( Bohnsack, 1982; Russ, 1991) make them vulnerable to overfishing (Sluka and Sullivan, 1998). Unlike most other multispecies trop- ical habitats, the Great Barrier Reef (GBR) supports a line fishery that tar- gets relatively few genera (Russ, 1991). The groupers of the genus Plectropo- mus support the most valuable commer- cial fin-fishery in Queensland (Trainor, 1991), with an annual value of $A10 million. Groupers bring a consistently high price (about $20/kilogram retail for fillet) on the Australian market and are highly valued in the “live ex- port” market to Asia (Miles1) Also, grou- pers are targeted by recreational line (Blarney and Hundloe2) and spear fish- ermen (Steven3), and recreational fish- ing is expanding with the rapid growth of the tourist industry on the GBR. Plectropomus leopardus 4 is the domi- nant component of the three main coral trout species caught on the GBR by all fishermen (Williams and Russ, 1994; Steven3), and by Australian standards, this species is subjected to considerable fishing pressure. Intense fishing pressure has been im- plicated in the drastic declines of grou- per populations off Florida and the Ca- ribbean (Sadovy, 1994; Bohnsack et al., 1994). In Australia, fishery managers have already detected a decline in abun- 1 Miles, A. 1997. Research, not rumors, needed for live fishing industry. Exploring Reef Science. May newsl. [Available from CRC Reef Research Centre, James Cook University, Townsville 4811, Australia.] 2 Blarney, R. K., and T. J. Hundloe. 1991. Characteristics of recreational boat fishing in the Great Barrier Reef region. Unpubl. report to the Great Barrier Reef Marine Park Authority (GBRMPA) PO Box 1379, Townsville, Queensland 4810, Australia. 3 Steven, A. 1988. An analysis of fishing activities on possible predators of the crown of thorns starfish (Acanthaster planci ) on the Great Barrier Reef. Unpubl. report to GBRMPA, PO Box 1379, Townsville, Queensland 4810, Australia. 131 p. 4 The official common name of Plectropomus leopardus is leopard coralgrouper, formerly known as bluedotted coraltrout (Heemstra and Randall, 1993). St John et al.: Diet of Plectropomus leopcirdus on the Great Barrier Reef 181 dance of P. leopardus in some parts of the GBR, particular- ly close to centers of human population (Craik, 1981). Fur- thermore, fishing pressure is expected to intensify owing to increases in both commercial and recreational demand for this species. Maintaining grouper stocks on coral reefs depends on careful management of these fisheries. Strategies of coral reef management include closing reefs to all forms of exploitation. This strategy is designed to protect reef-fish stocks and habitats (Williams and Russ, 1994), which enables populations of reef fishes to re- gain or maintain natural levels of abundance (see review by Roberts and Polunin, 1991). Since the establishment of the Great Barrier Reef Marine Park (GBRMP) in 1981, the major management strategy has involved partitioning of reefs into six main zones that permit different levels and types of fishing.5 Actual fishing pressure on reefs in these zones, however, has not been measured in many areas of the marine park (Williams and Russ, 1994). Most concerns about the impact of fishing have focused on the reduction of stock abundance (Russ, 1991). Both direct and indirect trophic effects on the structure of cor- al reef-fish communities by the removal of piscivores re- main poorly understood (Hixon, 1991; Russ, 1991; Steneck, 1998) . Some authors have argued that the removal of large piscivorous fishes leads to compensatory increases in over- all abundance or changes in relative abundance of prey of- ten termed “prey release” (Beddington and May, 1982; Go- eden, 1982; Beddington, 1984; Grigg et al., 1984; Koslow et al., 1988). Russ (1991) and Jennings and Lock (1996), however, argued that the evidence for “prey release” on coral reefs is limited and equivocal. In 1978 Goeden sug- gested that depleting the abundance of P. leopardus may irreversibly alter the structure of the community of coral- reef fishes on the GBR, and yet, 20 years later, nothing is known about the trophic impacts of this fishery. Coral reefs with different fishing histories represent valuable large scale manipulations of predator densities (Jennings and Polunin, 1997). In contrast to results of smaller-scale predator-removal experiments on site-at- tached coral reef fishes (Caley, 1993; Hixon and Beets, 1993; Carr and Hixon, 1995), large-scale “experiments” of predator removal by fishing have not demonstrated that predators play an important role in structuring fish com- munities. Several studies have compared prey populations on fished and unfished reefs but have found no evidence for corresponding changes in diversity or biomass of potential prey species (Bohnsack, 1982; Russ, 1985; Jennings and Po- lunin, 1997; Russ and Alcala, 1998a, 1988b). These studies, however, were hampered by a lack of dietary information necessary to determine exactly which prey fishes are im- portant in the diet of the fished species. For coral reef pisci- vores, commercially important or otherwise, quantitative studies on food consumption, feeding strategies, and com- prehensive lists of prey species are rare (but see Norris, 1985; Norris and Parrish, 1988; Sweatman, 1984; St John, 1999) . Studies of predator-prey relationships on coral reefs 5 1987. Central Zoning Information Pamphlet. GBRMPA, PO Box 1379, Townsville, Queensland 4810, Australia. Fold-out pam- hlet with no pagination. cannot progress further without knowledge of the prey con- sumed (Jennings and Polunin, 1997). Information on the trophic biology of groupers is impor- tant for protecting the stocks and ensuring a healthy fish- ery for the future. Knowledge of the breadth of the diet and the specific habitat of important prey will indicate the resilience of stocks to changes in prey abundance or habi- tat destruction. Furthermore, baseline information on the diet of unfished populations is needed as a comparison for diets of fished populations. Such information should in- clude some measure of spatial variation in regard to the diet. Fishing activities may alter the trophic ecology of the species targeted by altering the structure or the be- havior of the wild population or both. For example, remov- ing larger individuals of P. leopardus may alter the diet of the population because larger predators eat larger prey. St John (1999) found that the composition of the fish diet of P. leopardus on the GBR varied with size until preda- tors attained 35 cm (TL). If this result applies to other geographic areas, fishing could affect the trophic impact of P. leopardus populations where legal minimum size limits are less than 35 cm (TL) or do not exist (e.g. Okinawa, Ja- pan). On the GBR, however, such effects would be minimal because the legal minimum length is 38 cm (TL). Fishing may also affect the behavior of piscivores. Line fishermen use baited hooks and throw bait into the water (termed burleying) to attract leopard coralgroupers to their fishing sites on the GBR. The use of bait to catch these fish could affect feeding behavior or feeding-related patterns of movement of P. leopardus in fished reefs and may have short- or long-term effects on the feeding ecology of P. leopardus. Compared with commercially important fishes in many other ecosystems, relatively little is known about the diet of adult P. leopardus on the GBR, despite the huge num- bers of P leopardus fished from the reefs on the GBR annually (but see Choat, 1968; Goeden, 1978; Ivingsford, 1992; St John, 1995, 1999). In all the existing studies, most samples were collected by spear, presumably to avoid pos- sible dietary biases caused by line fishing methods. Thus, nothing is known about the trophic impacts associated with line fishing, which uses baited hooks at local fishing sites and which is by far the most common method used to catch P. leopardus. A comparison of stomach contents between P. leopardus caught by spear and those caught by hook and line at the same reefs will provide information on the selectivity of line fishing and trophic impact of this species. Our study is the first to examine the diet of a large com- mercial species of coral reef fish at reefs in open and closed fishing zones. The feeding ecology of P. leopardus , a major coral reef piscivore, was examined at reefs zoned open and closed to fishing for eight years in the Cairns section of the GBRMP. Although the primary aim of this study was to de- scribe and compare the diet off! leopardus on reefs in two fishing zones, we pooled reefs and compared the diet by lo- cation to assess the natural spatial variability in the diet of P. leopardus in this region. Also, we assessed the size struc- ture and diet of P. leopardus caught by two fishing meth- ods: nonselective spear fishing (described in the “Materials 182 Fishery Bulletin 99(1 ) and methods” section) and baited line fish- ing. Fishing pressure was assessed by com- paring differences in the size structure of populations of P. leopardus between fished and unfished reefs using size information from the catch at each reef as well as fishery independent estimates from visual surveys done just prior to sampling (Ayling and Ay- ling6; Brown et al.7). Materials and methods Study sites and sampling design Two pairs of mid-continental shelf reefs at the southern end of the Cairns section of the GBR Marine Park were sampled in Janu- ary and February 1992 (Table 1). Each pair comprised one reef that had been closed to fishing for eight years (Marine National Park (MNP) ‘B’) and another open to com- mercial and recreational fishing activities (General Use (GU) ‘B’). These reefs will sub- sequently be referred to as “closed” and “open” zones. The northern pair, Wardle Reef (closed) and Nathan Reef (open), and the southern pair, Noreaster Reef (closed) and Potter Reef (open), were approximately 30 km apart, and the reefs within pairs were 5 and 10 km minimum distance from each other respectively (Fig. 1). Figure t The southern part of the Cairns section of the Great Barrier Reef Marine Park showing Nathan and Potter Reefs (zoned GU ‘B’, all fishing allowed) and Wardle and Noreaster Reefs (zoned MNP ‘B\ closed to all fishing). Collection of P. leopardus Each reef was fished by line and spear simultaneously. Two teams of experienced fishermen collected fish by spear and line during daylight hours (0600-1800 h). Line fishermen worked in pairs from two small boats and used single hook rigs with pilchards (Sardinops neopilchardus ) as bait. Spear fishermen worked in two pairs and hunted each fish as soon as it was seen in an attempt to obtain a sample representative of the size distribution of the actual population of coral trout. If two or more grouper were spotted simultaneously, the choice of the target was not based on size, and spear fishermen were deliberately not size selective. Spear fishing, however, has some inherent biases. Larger fish will be sighted first by spear fisher- 6 Ayling, A. M. and A. L. Ayling. 1992. Effects of fishing pilot study: visual surveys on Cairns section closed reefs that will be opened under the new zoning plan. Unpub] . report to GBRMPA, PO Box 1379, Townsville, Queensland 4810, Austra- lia, 48 p. 7 Brown, I. W., L. C. Squire, C. Baltus, and M. Sellum. 1996. Effect of zoning changes on the fish populations of unexploited reefs — stage 2: post-opening assessment. Unpubl. report to GBRMPA, PO Box 1379, Townsville, Queensland 4810, Austra- lia, 40 p. men because they will approach a diver more readily than smaller fish and are more conspicuous at distances (Kul- bicki, 1998). These biases, however, should be similar at all reefs. Furthermore, the detection of camouflaged spe- cies, such as P. leopardus, increases with disturbance of the habitat when divers flush out these fish earlier at dis- turbed or fished reefs (Kulbicki, 1998), and this distur- bance may increase the speared catch. Amount of catch, however, was not the focus of this study. Samples were stored on ice in the small boats and frozen on board the mother ship (up to several hours later). Fork length (FL) of captured coral trout was measured to the nearest 0.5 cm. Stomach contents Only stomach contents were used in dietary analysis of P leopardus (St John, 1995). The mouth and gills of each specimen were examined for regurgitated prey (Parrish, 1987). The stomach was opened and any con- tents were removed. Stomach contents were fixed in 10% St John et al.: Diet of Plectropomus leopardus on the Great Barrier Reef 183 formal calcium acetate (FCA), a buffered formalin solu- tion (approximately 10 mL of FCA for every gram of stomach content) for a minimum of seven days and then stored in 70% ethanol. Stomach contents were classified broadly into natural prey or bait. All bait in our study were pilchards ( Sardinops neopilchardus), which does not occur naturally in the area. Every natural prey item was identified under low magni- fication to the lowest taxonomic group possible (St John, 1995). Fish were identified following Randall et al. (1990) by using several keys (Allen, 1975; Masuda et al., 1984; Smith and Heemstra, 1986; Myers, 1991). Families of prey fishes were classified by the area where they were most commonly found (Randall et al., 1990; se- nior author, personal obs.). Prey fishes from the “demer- sal reef substrata” habitat swam around and above coral, but used it for shelter (e.g. Pomacentridae and Scaridae). Fishes categorized as using the “benthic reef substrata” habitat were benthic dwellers that remained very close to the substrata (e.g. Blenniidae). Fishes categorized as in the “adjacent sands” habitat dwelled over, on, or within the sandy areas adjacent to reefs (e.g. Mullidae). Pelagic fishes in the “midwater” habitat were found from the wa- ter surface to a depth of approximately 1 m over the reef (e.g. Clupeidae and Caesionidae). Wet weights of the stomach contents were measured after preservation in 70% ethanol. Before weighing, the stomach contents were emptied into a sieve (St John, 1995) and shaken or sponged, or both shaken and sponged, to remove excess surface liquid (Parker, 1963). Wet weights of stomach contents were excluded from weight analyses when there was evidence of regurgitation of food from the stomach, (e.g. digested prey in the mouth or gills or an empty stomach that was stretched or everted), but this condition was rare. Generally, weights of digested prey reflected prey size and therefore were a useful measure of the diet (St John, 1995). Less than 14% of the prey was highly digested, and such fish could not be separated for weighing. In these cases, individual prey weights were estimated from total weights of the stomach sample, taking into account the di- gestion stage and the size (when known) of all individuals in the sample. The contents of the stomach were assumed to repre- sent daily feeding in P. leopardus because prey items were mostly digested after 24 hours (St John, 1995). Data analyses Kolmogorov-Smirnov (K-S) (Sokal and Rohlf, 1981) tests were used to compare size structures of leopard coral- grouper populations on reefs on both closed and open zones to fishing. Fisher’s exact tests were used to compare the frequen- cy of prey items in the diet because the categories (fish- ing zones, reefs, habitats, fishing methods and families of prey) were all nominal (Mehta and Patel, 1992). Because Fisher’s exact tests do not compare nested data sets, com- parisons within zones or locations used pooled data. Data from reefs were pooled when they did not differ signifi- cantly (i.e. P>0.05). Because the differences in number of prey per family in the diet was statistically marginal in reefs in the open zone (P=0.056), we used the P-0.01 level of significance to compare open and closed zones when reefs were pooled. Also, the P=0.01 level of signifi- cance was used when sample sizes were small (e.g. in the comparison between the natural diet of line-caught and speared fish). An independent Ptest was used to compare the mean number of families consumed in each fishing zone. Prior to analysis, the variances were tested for homogeneity by using Cochran’s test, which was not significant (Cochran’s test statistic=0.84, P>0.05). Because the null hypothesis was not rejected, the power of the test to detect specified differences was calculated following Cohen (1988). In the calculations of power, the sample means were assumed to be representative of the parametric means for each treat- ment group. A nonsignificant result was considered to be inconclusive unless the power of the test (1-/3) was >0.80. The index of relative importance (IRI) determined for the diet of P. leopardus was IRI = 0.5 x (% prey number + % prey weight). This measure was used in Schoener’s a index of dietary overlap (Schoener, 1970) for pair-wise comparisons of the diet of P. leopardus between reefs within each of the two fishing zones. Dietary overlaps were classified by using Langton’s ( 1982) scale: low 0-0.29, medium 0.30-0.59, and high >0.60. Results Size structure of P. leopardus In total, 672 P. leopardus, ranging in size from 13 to 58.5 cm FL, were collected by line and spear from the four reefs (Fig. 2) and fewer P. leopardus were caught by line (n=85) than by spear (n= 587) (Table 1). The speared catch was considered to represent popu- lations of P. leopardus on reefs because spear fishermen were deliberately not size selective in this study. The size- structure of the line catch was significantly larger than that of the speared catch (Dmax=0.243, K-S P<0.01) when samples from the four reefs were pooled (Fig. 2). Thus, line caught P. leopardus were excluded from further analyses of population size structure. When reef pairs within locations were compared, the size structure of the speared catch on the open reefs differed significantly from the closed reefs (north: Dmax=0.20, K-S, /2 = 143, P<0.01; south: Dmax=0.13, K-S, n=131 P<0.05; Fig. 2), but these zonal patterns differed between loca- tions. The size structure of the speared catch differed sig- nificantly between the two open reefs (Dmax=0.19, K-S, n = 155, PcO.Ol, Fig. 2); the proportion of larger fish ( >35 cm FL) was 45% at Nathan Reef and 64% at Potter Reef compared with 57% at each of the closed reefs. The results of our study did not demonstrate an effect of fishing on size structure between open and closed reefs. 184 Fishery Bulletin 99(1 ) Stomach contents of fish caught by the two methods Although the proportion of P. leopardus with empty stom- achs did not differ between the line-caught and speared catch (Fisher’s exact test(1|=0.12, P=0.81), the proportion of the catch with natural prey in their stomachs was sig- nificantly higher in speared fish (Fisher’s exact test[xp7.8, P=0.01) (Table 2). As expected, a much higher proportion of line-caught fish had consumed the bait S. neopilchardus (Table 2); however, five P. leopardus that were speared at sites where line fishing had not occurred, also had consumed bait. cl o © 0.20, 1-/3 < 0.33). Dietary overlap was high between reefs within fishing zones (open reefs: Schoener a =0.68; closed reefs: Schoener a=0.68) and between fishing zones (Schoener a=0. 65) when reefs were pooled. IRI values were higher in Pomacentri- dae and Labridae at the closed reefs than at the open reefs, whereas the opposite pattern occurred for Scaridae and Synodontidae (Table 6). The number of prey per family in the diet of P leopar- dus varied significantly among the four reefs (Table 8). This variation, however, could not be related to the zon- ing of reefs because the diet of P. leopardus did not differ St John et al.: Diet of Plectropomus leopardus on the Great Barrier Reef 187 Table 6 The index of relative importance (IRI) calculated for each fish family of prey and expressed as a percentage of the total for all identi fled fish families in the diet of P. leopardus, grouped by reef, fishing zone, and total. Small-size schooling fishes are underlined. Reefs open to fishing Reefs closed to fishing North South North South Prey families (Nathan) (Potter) Both (Wardle) (Noreaster) Both Total Pomacentridae 23.4 16.9 20.8 32.2 38.9 35.4 27.8 Labridae 16.6 13.3 15.4 26.4 25.9 26.0 20.6 Scaridae 15.8 14.3 15.0 9.0 4.9 9.8 Clupeidae 10.6 9/7 10.1 03 71 7/7 09 Caesionidae 20.7 8.5 1.4 13.4 8.1 8.9 Synodontidae 4.9 8.5 6.1 1.9 3.5 2.7 4.7 Blenniidae 0.9 1.4 1.1 15.1 7.0 4.4 Acanthuridae 3.9 5.8 4.7 3.0 1.4 3.2 Serranidae 4.7 2.0 3.5 2.0 1.1 2.2 Scorpaenidae 7.1 4.4 1.8 Apogonidae 1.0 5.4 3.0 1.6 Nemipteridae 5.3 3.3 1.3 Fisfulariidae 1.0 1.1 1.1 1.6 0.8 0.9 Engraulidae L9 TO T8 08 09 Plesiopidae 2.7 1.2 0.7 Gobiidae 1.9 1.0 0.5 Balistidae 1.5 0.7 0.4 Creedidae 1.4 0.7 0.3 Platycephalidae 1.3 0.7 0.3 Siganidae 1.3 0.6 0.3 Lutjanidae 1.0 0.5 0.2 Monacanthidae 0.9 0.5 0.2 No. of P. leopardus with prey 92 102 194 100 86 186 380 significantly between fishing zones (at P=0.01 level of sig- nificance, Table 8, see “Materials and methods” section for explanation). Furthermore, when pooled by location, the numerical composition of the diet of P. leopardus on south- ern and northern reefs differed significantly (Table 8). When categorized by their characteristic habitat (Table 9), the number of prey did not differ significantly among reefs (Fisher’s exact test|9|=14.7, P=0.10) nor between fish- ing zones (Fisher’s exact test[3j=3.3, P= 0.35, Table 10). Fishes in the majority of prey families (n=ll) lived in the demersal reef habitat (Table 9), which was the most im- portant source of prey for P leopardus (61% by number, IRI=67%, Table 10). Prey in three or four families lived in each of the benthic reef, midwater, and adjacent sandy habitats (Table 9). The large number of midwater prey at Potter Reef (Table 10) reflected the relatively large num- ber of Caesionidae consumed at this reef (Table 4). Discussion Overall, the diets of P leopardus from reefs in the two fishing zones were not dissimilar. Dietary overlap was high between P. leopardus from the reefs with open zones and reefs with closed zones for eight years. When com- pared to natural variation in diet among P. leopardus on the GBR, these values of overlap were very high because there was less similarity in the diet among regional popu- lations of P leopardus (Schoener a value of dietary over- lap ranged from 0.26 to 0.42, St John, 1995) and at one reef sampled over time (Schoener a value of dietary over- lap ranged from 0.44 to 0.84, St John, 1995). Generally, feeding behavior of P leopardus was similar between fish- ing zones. The lack of dietary differences between P leopardus on open and closed reefs is consistent with the lack of mean- ingful trends in comparing the size and age structures of these populations. Using the catch of both fishing meth- ods, Brown et al.7 found a slightly higher abundance of legal-size (>38 cm TL) leopard coralgrouper on the closed reefs, as well as a greater proportion of older fish (4+ year class). But, fishing affected the size and age structure of populations on the two open reefs very differently; Na- than Reef showed a large recruitment of the one year class (Brown et al.7) and differed from all other reefs. In visual surveys ofP leopardus at the four reefs, Brown et al.7 de- tected a greater density of larger P leopardus on closed reefs. In contrast, Ayling and Ayling6 found no differences between fishing zones in the density, average length, and recruitment of P leopardus when surveying the same reefs 188 Fishery Bulletin 99(1) just prior to Brown et al.7 Because decreases in abundance and size of populations are widely recognized as evidence of fishing pressure (Russ, 1991; Jennings and Lock, 1996), fishing pressure between the two zones in this study did not appear to vary. Eight years of effective protection from substantial fish- ing pressure should have produced detectable differences in the structure of populations of P. leopardus on these closed and open reefs. Generally, studies on the GBR have shown an increase in average size of P. leopardus popula- Table 7 Abundance (in number and percentage) of prey belonging to families in Families of fishes living in the midwaters are underlined. the diet of P. leopardus caught by the two fishing methods. Prey families Line Spear Total Number % Number % Pomacentridae 2 9.5 45 27.3 47 Labridae 1 4.8 36 21.8 37 Clupeidae n 52.5 20 12,1 31 Caesionidae 1 L8 n a6 12 Synodontidae 2 9.5 8 4.8 10 Scaridae 8 4.8 8 Blenniidae 7 4.2 7 Acanthuridae 1 4.8 5 3.0 6 Apogonidae 6 3.6 6 Serranidae 1 4.8 4 2.4 5 Engraulidae 3 L8 3 Fistulariidae 2 9A 1 06 3 Gobiidae 2 1.2 2 Balistidae 1 0.6 1 Creedidae 1 0.6 1 Lutjanidae 1 0.6 1 Monacanthidae 1 0.6 1 Nemipteridae 1 0.6 1 Platycephalidae 1 0.6 1 Plesiopidae 1 0.6 1 Scorpaenidae 1 0.6 1 Siganidae 1 0.6 1 Total 21 165 186 No. of P. leopardus with prey 38 342 380 Table 8 Results of Fisher exact tests for the number of prey per family. Information in brackets is Fisher’s exact test statistic, degrees of freedom (df), and exact probability levels ( P ). Significance of tests is denoted by ** for PcO.Ol and * for P<0.05, and ns is nonsig- nificant. Comparisons between fishing zones and fishing methods are tested at P =0.01 level of significance (see “Materials and methods” section for explanation). Dietary differences among P. leopardus Fishing zones Locations Fishing methods Open vs. closed (reefs pooled) ns (28.3, df= 21, P=0.046) North vs. south (reefs pooled) * (29.0, df=21, P=0.040) Speared vs. line caught (reefs pooled) ns (34.2, df=21, P=0.034) Open reefs Closed reefs North reefs South reefs Among 4 reefs * ns (22.5, df=16, P=0.056) (18.5, df=15, P=0.908) ns ns (23.2, df=20, P=0.135) (19.8, df=ll, P=0.155) ** (75.2, df=63, P=0.007) St John et al.: Diet of Plectropomus leopcirdus on the Great Barrier Reef 189 tions on reefs after closure (summarized in Williams and Russ, 1994). Also, population density of P. leopardus has differed in other closed-versus-open fishing zones in the Marine Park (Ayling et al., 1991). In contrast, the size structure of populations of P. leopardus on four reefs in the GBR showed no effect of protection from fishing after 3-4 years of reef closure (Ferreira and Russ, 1995). Possible reasons for an absence of the effects of fishing include mi- gration by P. leopardus among reefs and small differences in the actual fishing pressure between open and closed zones. Other studies on movement of P. leopardus indicate that migration of 5-10 km between reefs would be highly unlikely (Davies, 1995; Zeller, 1997). Of a tagged popula- tion of 4627 P. leopardus on five reefs on the GBR, only 1% moved between reefs in a period of 22 months and only 2% travelled distances of 5-7.5 km (Davies, 1995). Fishing pressure was not measured on any of our four reefs dur- ing the eight years of protection, so it is possible that fish- ing pressure was low on the open reefs; however it seems more likely that the “closed” reefs were fished illegally. Such violations are thought to be relatively common on the GBR, and illegal fishing of these prized food fishes is probably widespread throughout tropical waters. Another documented example of violations of a fishing regulation for a large grouper is found in the Florida Keys, where bans on harvesting Nassau grouper (Epinephelus striatus) appeared to be ineffective (Sluka and Sullivan, 1998). The diet of P. leopardus in our study did not differ be- tween fishing zones when families of prey were assessed by either their relative importance in the diet (dietary overlap) or their number of prey. Patterns between fish- ing zones occurred in four of the 22 families but were con- sidered weak because they were not detected statistically. Pomacentridae and Labridae, which ranked first and sec- ond respectively at all four reefs, were more important in the diet in the closed reefs, whereas Scaridae and Syn- odontidae were more important in the diet on open reefs. Information on prey availability on each reef may explain these results. Lastly, proportions of prey consumed from each of the four habitats were similar among reefs, sug- gesting that feeding behavior of P. leop- ardus did not differ among these reefs Two aspects of the diet of P. leopardus differed between the locations of the reef pairs. Rare prey families, which occurred only once in the diet in the entire study, were more common at the two northern reefs (eight families) than at the south- ern pair (one family). Also, P. leopardus from the two southern reefs consumed more of the large schooling Caesionidae. However, location of the reefs alone may not explain these patterns because reef location is confounded with time of sam- pling in our study. The southern pair of reefs was sampled one month after the northern pair. Similar to other dietary studies of adult P. leopardus on the GBR (Choat, 1968; Goeden, 1978; Kingsford, 1992; St John 1995, 1999), our study confirmed that leopard coralgroupers are highly Table 9 The 22 families of prey classified into four broad habitats on the reef: demer- sal reef (associated with substrata), benthic reef (strongly associated with reef substrata), midwater and adjacent sands. Demersal Benthic Midwater Adjacent sands Acanthuridae Blenniidae Caesionidae Creedidae Apogonidae Gobiidae Clupeidae Nemipteridae Balistidae Scorpaenidae Engraulidae Platycephalidae Labridae Lutjanidae Monacanthidae Pomacentridae Plesiopidae Scaridae Serranidae Siganidae Fistulariidae Synodontidae Table 10 Number of prey in the diet of P. leopardus on the four reefs in each of the four habitats (demersal reef, benthic reef, midwater, and adjacent sands). Total number and percentage of prey as well as importance of prey in the diet (IRI, index of relative importance! in each habitat are included. Habitat Open Closed Total IRI % North (Nathan) South (Potter) North ( Wardle) South (Noreaster) Number % Demersal reef 32 29 22 31 114 61.3 67.0 Benthic reef 4 1 5 0 10 5.4 6.6 Midwater 14 19 9 7 49 26.3 19.7 Adjacent sands 5 3 3 2 13 7.0 6.7 Total 55 52 39 40 186 100 190 Fishery Bulletin 99(1) piscivorous. Thus, unlike other less piscivorous coral reef serranids (Hobson, 1965; Randall, 1965; Harmelin-Vivien and Bouchon, 1976; Shpigel and Fishelson, 1989), adult P leopard us rely almost entirely on one general type of food (but juveniles consume crustaceans, St John, 1999). Also, nearly half of their diet (IRI=48.4%) comprised just two families of fishes, Pomacentridae and Labridae, but this finding does not suggest that the food supply of P. leop- ardus was limited. Plectropomus leopardus have been re- ported to consume more than 20 species in each of these two families (St John, 1995), which are highly diverse and abundant on coral reefs. On the GBR, Pomacentridae, with some 120 species, is the most numerically abundant fam- ily, and Labridae represents the second most speciose fam- ily (Randall et al., 1990). Therefore, the diet of P. leopardus was not dependent on a few species. The variety of prey fishes in the diet of P. leopardus reflects the groupers large home range (Samoilys, 1997; Zeller, 1997) that includes several habitats (Goeden, 1978; Kingsford, 1992; Samoilys, 1997). Plectropomus leopardus consumed fishes in families that lived in all four broad habitats on coral reefs; adjacent sands, midwater, benthic reef substrata, and demersal reef substrata. Yet, prey from the demersal reef environment and the midwaters were six times more important than prey in the other two habi- tats. The diverse and abundant families of fishes that live on or over adjacent sands, or dwell among the benthic reef substrata (e.g. Gobiidae and Blenniidae) were not impor- tant food for adult P. leopardus. Similarly, piscivores re- ported from coral reefs elsewhere have usually focused on prey in reef habitats. In the Caribbean, tethered prey con- sistently disappeared from sites close to areas of natural reef rather than from areas of adjacent sediment (Shul- man, 1985), and rates of encounter with predators for sur- geonfishes in bottles were highest at reef edges compared with other sediment habitats (Sweatman and Robertson, 1994). Lastly, the results of our study suggest several differenc- es in the catch and diet of P. leopardus caught by the two fishing methods. Such information is useful for managers when considering the impacts of different types of fishing methods on the fish population. Line fishing catches larg- er fish than nonselective spear fishing; however nonse- lective spear fishing probably never occurs in the real fishery. Based on the comparison of line catch and spear catch, baited lines appear to attract a higher proportion of hungry P. leopardus to common fishing sites, but preda- tors may leave the site after they have eaten bait. There- fore, line fishing can affect the trophic ecology of P. leop- ardus reef-wide, even when it is concentrated at just a few sites on reefs. Also, line fishing may alter patterns of movement of P. leopardus temporarily, and such op- portunistic behavior will be difficult to detect in move- ment studies on this species (e.g. Samoilys, 1997; Zeller, 1997). Thus, when reefs are partitioned into different fish- ing zones, line fishing could facilitate the movement of P. leopardus across reserve boundaries from protected ar- eas. In one study on the GBR, P. leopardus had low flux rates across reserve boundaries (Zeller and Russ, 1998). Another, more tentative, result of our study is that line- caught fish appear to eat more midwater pelagic fishes, which suggests that line fishermen or the methods used in line fishing target P. leopardus hunting this prey. Whether line fishermen chose sites where pelagic schooling fishes congregate (e.g. near the reef edge) or whether pelagic prey are attracted to a fishing site after fishermen hurley the waters, is uncertain. In conclusion, eight years after reefs have been closed to fishing, no differences that could be linked to zoning were detected in size structure of populations and diets of P. leopardus. Results of this dietary study, however, can offer some insights for the management of the P. leopardus fish- ery. Even though fish are the dominant food of this preda- tor, and the main fish prey are associated with the coral reef substrata, P. leopardus are not dependent on a narrow range of species for food. A diverse mixture of Pomacentri- dae and Labridae species represents nearly half of their di- et, and overall, the diet of P. leopardus is sufficiently broad to be resilient to the depletion of several species of prey. Acknowledgments We acknowledge and thank two anonymous referees for pertinent comments about the manuscript, Tony Ayling for providing his visual census data, the Squire family for accommodating J. St John, and the staff of the North- ern Fisheries Center for the use of their facilities. This research was supported by the Great Barrier Reef Marine Park Authority, a Freda Bage Fellowship (Australian Fed- eration of University Women, Queensland) to J. St John, and Australian Research Council funds to G. R. Russ. Literature cited Allen, G. R. 1975. Damselfishes of the South Seas. TFH Publications, Neptune City, NJ, 271 p. Ayling, A. M., A. L. Ayling, and B. D. Mapstone. 1991 Possible effects from protection by fishing pressure on recruitment rates of the coral trout (Plectropomus leo- pardus: Serranidae). In Recruitment processes (D. A. Hancock, ed.), p. 210-215. Australian Society for Fish Biology Workshop, Hobart, August 1991. Beddington, J. R. 1984. The responses of multispecies systems to perturba- tions. In Exploitation of marine communities (R. M. May, ed.), p. 209-225. Springer- Verlag, Berlin. Beddington, J. R., and R. M. May. 1982. The harvesting of interacting species in a natural eco- system. Sci. Am. 247:42-49. Bohnsack, J. A. 1982. Effects of piscivorous predator removal on coral reef fish community structure. In Gutshop '81: fish food habit studies (G. M. Caillet and C. A. Simenstad, eds.), p. 258-267. Wash. Sea Grant Publ., Seattle, WA. Bohnsack, J. A., D. E. Harper and D. B. McClellan. 1994. Fisheries trends from Monroe County, Florida. Bull. Mar. Sci. 54:982-1018. Caley, M. J. 1993. Predation, recruitment and the dynamics of commu- nities of coral-reef fishes. Mar. Biol. 117:33-43. St John et al.: Diet of Plectropomus leopardus on the Great Barrier Reef 191 Carr, M. H. and M. A. Hixon. 1995. Predation effects on early post-settlement survivor- ship of coral reef fishes. Mar. Ecol. Prog. Ser. 124:31-42. Choat, J. H. 1968. Feeding habits of and distribution of Plectropomus maculatus Serranidae at Heron Island. Proc. R. Soc. Qd. 80(21:13-18. Cohen, J. 1988. Statistical power analysis for the behavioral sciences ( 2nd ed. ). Lawrence Erlbaum Associates, Hillsdale, NJ, 567 p Craik, G. J. S. 1981. Underwater survey of coral trout Plectropomus leop- ardus Serranidae populations in the Capricornia Section of the Great Barrier Reef Marine Park. Proc. 2nd Int. Cor. ReefSymp. 1:54-58. Davies, C. 1995. Patterns of movement of three species of coral reef fish on the Great Barrier Reef. Unpubl. Ph. D. diss., Dept, of Marine Biology, James Cook University of North Queensland, Australia, 170 p. Ferreira, B. P., and G. R. Russ. 1995. Population structure of the leopard coralgrouper, Plec- tropomus leopardus , on fished and unfished reefs off Towns- ville, Central Great Barrier Reef, Australia. Fish. Bull. 93:629-642 Goeden, G. B. 1978. A monograph of the coral trout Plectropomus leopar- dus. Res. Bull. Qld. Fish. Ser. 1:1-42. 1982. Intensive fishing and a ‘keystone’ predator species: ingredients for community instability. Biol. Conserv. 22: 273-281. Grigg, R. W., J. J. Polovina, and M. J. Atkinson. 1984. Model of a coral reef ecosystem. Ill: Resource limi- tation, community regulation, fisheries yield and resource management. Coral Reefs 3:23-27. Harmelin-Vivien, M. L., and C. Bouchon. 1976. Feeding behavior of some carnivorous fishes (Serra- nidae and Scorpaenidae) from Tulear (Madagascar). Mar Biol 37:329-340. Heemstra, P, and J. Randall. 1993. FAO species catalogue. Vol 16: Groupers of the world family Serranidae, subfamily Epinephelinae: an annotated and illustrated catalogue of grouper, rockcod, hind, coral grouper and lyretail species known to date. FAO Fisher- ies Synopsis, FAO, Rome, 383 p. Hixon, M. A. 1991. Predation as a process structuring coral-reef fish com- munities. In The ecology of fishes on coral reefs (P. F. Sale, ed.), p. 475-508. Academic Press, San Diego, CA. Hixon, M. A. and J. P. Beets. 1993. Predation, prey refuges, and the structure of coral- reef fish assemblages. Ecol. Monogr. 63(11:77-101. Hobson, E. S. 1965. Diurnal-nocturnal activity of some inshore fishes in the Gulf of California. Copeia 1965:291-302. Jennings, S., and J. M. Lock. 1996. Population and ecosystem effects of reef fishing. In Reef fisheries (N. V. C. Polunin and C. M. Roberts, eds.1, p. 193-218 Fish and fisheries series. Chapman and Hall, London. Jennings, S., and N. V. C. Polunin. 1997. Impacts of predator depletion by fishing on the bio- mass and diversity of non-target reef fish communities. Coral Reefs 16:71-82. Kingsford, M. J. 1992. Spatial and temporal variation in predation on reef fishes by coral trout ( Plectropomus leopardus , Serranidae). Coral Reefs 11:193-198. Koslow, J. A., F. Hanley, and R. Wicklund. 1988. Effects of fishing on reef fish communities at Pedro Bank and Port Royal Cays, Jamaica. Mar. Ecol. Progr. Ser. 43:210-212. Kulbicki, M. 1998. How the acquired behaviour of commercial reef fishes may influence the results obtained from visual cen- suses. J. Exp. Mar. Biol. Ecol. 222:11-30. Langton, R. W. 1982. Diet overlap between Atlantic Cod, Gadus morhua, silver hake, Merluccius bilinearis, and fifteen other north- west Atlantic finfish. Fish. Bull. 804:745-759. Masuda, H., K. Amaoka, C. Araga, T. Uyeno, and T. Yoshino. 1984. The fishes of the Japanese archipelago. Tokai Univ. Press, Tokyo, 374 p. Mehta, C., and N. Patel. 1992. StatXact. CYTEL Software Corporation, Cambridge, 442 p. Myers, R. F. 1991. Micronesian reef fishes: a practical guide to the identi- fication of the inshore marine fishes of the tropical and cen- tral Pacific, 2nd ed. Coral Graphics, Barrigada, Guam. Norris, J. E. 1985. Trophic relationships of piscivorous coral reef fishes from the Northwestern Hawaian Islands. M.S. thesis, Univ. Hawaii, Honolulu, HI, 71 p. Norris, J. E. and J. D. Parrish 1988. Predator-prey relationships among fishes in pristine coral reef communities. Proc. 6th Inti. Coral Reef Symp. 2:107-113. Parker, R. R. 1963. Effects of formalin on length and weight of fishes. J. Fish. Res. Board Canada 20(6):1441-1455. Parrish, J. D. 1987. The trophic biology of snappers and groupers. In Tropical snappers and groupers: biology and fisheries man- agement (J. J. Polovina and S. Ralston, eds.), p. 405-463. Westview Press Inc., Boulder, CO. Polunin, N. V. C. and C. M. Roberts. 1996. Reef fisheries. In Fish and fisheries series, p. 193-218. Chapman and Hall, London. Ralston, S. 1987. Mortality rates of snappers and groupers. In Trop- ical snappers and groupers: biology and fisheries man- agement (J. J. Polovina and S. Ralston, eds.1, p. 375-404. Westview Press Inc., Boulder, CO. Randall, J. E. 1965. Food habits of the Nassau grouper ( Epinephelus stri- atus 1 6th meeting. Assoc Island Mar Labs Caribb. 6:13- 16. 1987. A preliminary synopsis of the groupers (Perciformes: Serranidae: Epinephelinae) of the Indo-Pacific region. In Tropical snappers and groupers: biology and fisheries man- agement (J. J. Polovina and S. Ralston, eds.1, p. 89-187. Westview Press Inc., Boulder, CO. Randall, J. E., G. R. Allen, and R. C. Steene. 1990. Complete diver’s and fisherman’s guide to fishes of the Great Barrier Reef and Coral Sea. Crawford House Press, Bathurst, 507 p. Roberts, C. M., and N. V. C. Polunin. 1991. Are marine reserves effective in management of reef fisheries? Rev. Fish Biol. Fisheries 1:65-91. Russ, G. R. 1985. Effects of protective management on coral reef fishes 192 Fishery Bulletin 99(1 ) in the central Philippines. Proc. 5th Int. Coral Reef Congr. 4:219-224. 1991. Coral reef fisheries: effects and yield. In The ecology of fishes on coral reefs (P. F. Sale, ed.), p. 601-636. Aca- demic Press, San Diego, CA. Russ, G. R. and A. C. Alcala. 1996. Marine reserves: rates and patterns of recovery and decline of large predatory fish. Ecol. Applic. 6:947-961. 1998a. Natural fishing experiments in marine reserves 1983-1993: community and trophic responses. Coral Reefs 17:383-397. 1988b. Natural fishing experiments in marine reserves 1983-1993: roles of life history and fishing intensity in family responses. Coral Reefs 17:399-416. Sadovy, Y. 1994. Grouper stocks of the Western Central Atlantic: the need for management and management needs. Proc. Gulf Caribb. Fish. Inst. 43:43-64. Samoilys, M. A. 1997. Movement in a large predatory fish: coral trout, Plec- tropomus leopardus (Pisces: Serranidae) on Heron Reef, Australia. Coral Reefs 16:151-158. Schoener, T. W. 1970. Nonsynchronous spatial overlap of lizards in patchy habitats. Ecology 513:408-418. Shpigel, M., and L. Fishelson. 1989. Habitat partitioning between species of the genus Cephalopholis (Pisces Serranidae) across the fringing reef of the Gulf of Aqaba (Red Sea). Mar. Ecol. Prog. Ser. 58:17-22. Shulman, M. J. 1985. Recruitment of coral reef fishes: effects of distribution of predators and shelter. Ecology 66(3):1056-1066. Sluka, R., and K. Sullivan. 1998. The influence of spear fishing on species composition and size of groupers on patch reefs in the upper Florida Keys. Fish. Bull. 96:388-392. Smith, M. M., and P. C. Heemstra. 1986. Smith’s sea fishes. Springer- Verlag, Berlin, 1047 p. Sokal, R. R., and F. J. Rohlf. 1981. Biometry, 2nd ed. W.H. Freeman, San Francisco, CA, 859 p. Steneck, R. S. 1998. Human influences on coastal ecosystems: Does over- fishing create trophic cascades? Tree 13(ll):429-430. St John, J. 1995. Feeding ecology of the coral trout, Plectropomus leop- ardus Serranidae on the Great Barrier Reef, Australia. Ph.D. diss., James Cook University of North Queensland, Townsville, Australia, 277 p. 1999. Ontogenetic changes in the diet of a large coral reef grouper Plectropomus leopardus (Serranidae): patterns in taxa, size and habitat of prey. Mar. Ecol. Prog. Ser. 233-246. Sweatman, H. P. A. 1984. A field study of the predatory behaviour and feeding rate of a piscivorous coral reef fish, the lizardfish Synodus englemani. Copeia 84:187-193. Sweatman, H. P. A., and D. R. Robertson. 1994. Grazing halos and predation on juvenile Caribbean surgeonfishes. Mar. Ecol. Prog. Ser. 111:1-6. Trainor, N. 1991. Commercial line fishing. The Queensland Fisher- man, March 9(3):17-24. Williams, D. M., and G. R. Russ. 1994. Review of data on fishes of commercial and recre- ational fishing interest on the Great Barrier Reef. Vol. I: Research publication 33. Great Barrier Reef Marine Park Authority, Townsville, Australia, 103 p. Zeller, D. C. 1997. Home range and activity patterns of the coral trout Plectropomus leopardus (Serranidae). Mar. Ecol. Prog. Ser. 154:65-77. Zeller, D. C., and G. R. Russ. 1998. Marine reserves: patterns of adult movement of the coral trout (Plectropomus leopardus (Serranidae)). Can. J. Fish Aquat. Sci. 55:917-924. 193 Estimating tag-shedding rates for skipjack tuna, Katsuwonus pelamis, off the Maldives M. Shiham Adam Renewable Resources Assessment Group T. H. Huxley School of Environment, Earth Sciences and Engineering Imperial College of Science Technology and Medicine, RSM Building Prince Consort Road London SW7 2BP, United Kingdom Present address Pelagic Fisheries Research Program Joint Institute of Marine and Atmospheric Research University of Hawaii at Manoa Honolulu, Hawaii 96822 E-mail address: msadam@soest hawaii.edu Geoffrey P. Kirkwood Renewable Resources Assessment Group T. H, Huxley School of Environment, Earth Sciences and Engineering Imperial College of Science, Technology and Medicine RSM Building, Prince Consort Road London SW7 2BP, United Kingdom One source of uncertainty in fishery assessments based on tag release and recapture data arises from tag shed- ding— the loss of tags from fish from the time of tagging until tag recovery. Inde- pendent estimation of tag-shedding rates from double tagging experiments is an integral part of well-designed tagging experiments. Failure to allow for tag shedding can result in biases in estimates of important parameters derived from tag-recapture data, such as fishery-induced and natural mortal- ity, and migration rates. A wide variety of methods have been proposed for esti- mating shedding rates using data from double tagging experiments, the vari- ety in part resulting from the nature of the data available from the experi- ments. The earlier literature has been well integrated and reviewed by Weth- erall (1982). More recently, Xiao (1996) has developed a general model that unifies the estimation of tag-shedding rates from double tagging experiments with exact and pooled times at liberty. Some recent examples of application of these methods include Xiao et al. (1999) for school gummy shark and Fabrizio et al. (1999) for lake trout. Tag shedding is of two types (Wether- all, 1982 ). Type-I shedding is a one-time event and occurs immediately after tag- ging, usually as a result of suboptimal placement of tags in the fish. Effective- ly, it reduces the number of tags ini- tially put out to sea. Type-II shedding is the loss of a tag or both tags over a period of time after the fish has been tagged and released back into the sea. For long-lived species, it may not occur at a constant rate because some tags are likely to have been applied more ef- fectively than others, and some may be- come firmly embedded (with growth of muscle tissues), such that they are very unlikely to be detached from the fish (Kirkwood, 1981). The Ministry of Fisheries and Agri- culture, Maldives, carried out two tag- ging experiments: one in 1990-91 (Ye- saki and Waheed, 1992) and the other between 1993 and 1995 (Anderson et al., 1996). The latter tagging experi- ment included a double tagging experi- ment in which 504 skipjack tuna, Kat- suwonus pelamis , were double tagged, by using the same type of tags and techniques of tagging as used in the single tagging experiment (Anderson et al., 1996). As of end 1996,atotal of53 of these fish had been recovered. These data are considered to be of imme- diate importance for designing large- scale experiments in the Indian Ocean, strategic tagging simulation studies (e.g. Bertignac, 1996), and for compar- ing the estimated rates with those ob- tained from similar tuna double tag- ging experiments conducted elsewhere. Our note reports estimates of tag-shed- ding rates carried out in the Maldives. Materials and methods Tagging methods Tagging was conducted on board local vessels fishing for skipjack tuna with pole-and-line gear using livebait. Plas- tic dart tags (100 mm x 1.5 mm diam- eter), manufactured by Hallprint™, Australia, were used throughout the experiment. The captured fish were gently placed on deck, quickly slipped onto a wet wooden measuring board, and held in place by biologists wear- ing cotton gloves. The first tag was inserted dorsally on the left-hand side, at an acute angle adjacent to the second dorsal fin so that the barb was caught under the fin-ray extension or the neural spine. The second tag was inserted about 1-2 cm posterior to the first on the right-hand side in the same manner. Consecutively numbered pairs of tags were used; even numbers were inserted on the left-hand side and odd numbers on the right. Where possible, fish were returned to (he water, facing the vessel’s bow, in a slightly head- down fashion. Tagging times (from hooking to release into the sea) ranged from 14 to 18 seconds. More details of the tagging program can be found in Yesaki and Waheed ( 1992) and Ander- son et al. ( 1996). Parameter estimation The method of parameter estimation used here is the maximum likelihood approach introduced by Kirkwood and Walker (1984) and later extended by Hampton and Kirkwood (1990) and Hampton (1997). This method was de- veloped for use with sets of data where recaptured fish are few, and with few fish that have shed a tag, provided that Manuscript accepted 21 July 2000. Fish. Bull. 99:193-196 (2001). 194 Fishery Bulletin 99(1 ) exact dates of recovery are known, as was the case with the data from the Maldives. Following Kirkwood and Walker (1984), consider an originally single tagged fish. Assuming that a is the type-I retention probability (1-type-I shedding probability) and A is the type-II shedding rate, assumed to be constant for short-lived species such as the skipjack tuna, then for an originally single tagged fish, the probability of a tag being retained at time t is given by Q(t) = a exp (-At). (1) Suppose that fish are double tagged with identical tags and released at time t = 0, and let pft) be the probability that the fish is alive and at liberty retaining i (z=0, 1, 2) tags at time t. Then, under the assumption that both tags are retained after immediate shedding and have indepen- dent and identical probabilities, p0(t) = [1 - Q(t)]2 zation routines in AD Model Builder (Otter Research Ltd., 1996). Results The numbers of recoveries of originally double tagged fish reported as retaining one (indicating left- or the right-side tag) or two tags on recapture are shown in Table 1, with their times at liberty. The maximum likelihood estimate of A was 0.22/yr (SE=0.13) and that of a was 0.97 (SE=0.03). If a = 1, the maximum likelihood estimate of A was 0.30/yr (SE=0.065/ yr). A likelihood ratio test (Cox and Hinkley, 1974) shows that the full model does not provide a significantly better fit to the data (at 5% level) than its special case (a=l) (P=0.146). Discussion p1(t) = 2Q(t)[l-Q(t)} (2) p2it ) = Q(t)2. In practice, identifiable recaptures will consist only of fish retaining either one tag or two tags. If P'j(t) is the prob- ability that an originally double-tagged fish, recaptured at time £■, is reported to have retained i tags (1=1, 2), then conditional on the retention of at least one tag, the prob- ability of capturing a fish retaining two tags at time t is P'(t) = P2(*> 1-PoU) (3) and the probability of capturing a fish retaining only one tag at time t is p;(t) = P i0.05). Harengula jaguana grow rapidly and live little more than a year (Table 2; Fig. 3, A-F). Observed ages varied from 141 to 370 days for males and from 149 to 335 days for females. Growth rates (slope) were not significantly different for males and females in either West Palm Beach (ANCOVA, F=0.29, df=l, 30, P=0.595) or in Tampa Bay (ANCOVA, F=3.24, df=l, 118, P=0.074). Intercepts were also not significantly different in either West Palm Beach (ANCOVA, F=2.71, df=l, 31, P=0.109) or in Tampa Bay (ANCOVA, F=0.98, df=l, 119, P=0.325); therefore the data were pooled for each coast. Pooled growth rates (immature and adult) for Tampa Bay were significantly different from West Palm Beach (ANCOVA, F=6.09, df=l, 249, P=0.014). Combined commercial purse-seine and cast-net length- frequency databases showed multiple juvenile recruitment peaks between June and November 1993 in Tampa Bay and between June and August 1992 from West Palm Beach (Fig. 4, A-B). Growth rates estimated from cohort length frequencies in Tampa Bay from February 1993 to May 1993 (spawned in September 1992) showed growth from 85 mm to 135 mm FL over 90 days, an average growth rate of 0.55 mm/day. A similar rate was obtained for fish at 45-mm mode in July 1993 (spawned in April 1993) that grew 125 mm FL by December 1993, a period of 150 days producing an average growth rate of 0.53 mm/day. Discussion Polished transverse sections of H. jaguana otoliths allowed enumeration of age in days. Houde et al. (1974) reported NOTE Pierce et al.: Age and growth of Harenguta jaguana in Florida waters 207 A West Palm Beach o >> o c t> 3 o* V 4> CL, 50 40 h 30 20 10 0 50 r- 40 - 30 20 10 0 50 40 30 20 10 0 50 40 30 20 10 50 (- 40 30 20 10 0 50 40 30 20 10 0 J l Feb 1992 « = 471 ...1 .J -1-1. -I JL May 1992 n = 188 1 1 i 1 f 1 -1 iJI »1li I I Jun 1992 n = 1298 Aug 1992 n = 253 Oct 1992 n = 93 l-l-l-I I, 1 Nov 1992 n = 196 ±... 1 I 1 I I I I _ at jan 1993 _ 92— I— « = 51 — IM Jun-/Jul. 92 Jm i i i i i i i i i oct. 92 Mar 1993 n = 42 LI -Lml— 1 I I I 1 1 I Sep 1992 n = 61 1 .1 1 J May 1993 n = 142 Nov. 92 Sept 92 .JL_J J Aug 1993 Apr. 93 n = 901 Mar. 93 Jan. 93 . Oct 1993 n = 379 l-l I I I I 40 60 80 100 120 140 160 40 60 80 100 120 140 160 Fork length (mm) Figure 4 Observed monthly fork length frequencies of H. jaguana collected from off West Palm Beach and in Tampa Bay, Florida. Cohort age was determined from the sagittal micro- structure and labeled with the estimated spawn date. (A) West Palm Beach 1992-93; (B I Tampa Bay 1993. that H. jaguana larvae <8.9-mm TL did not show any ossification and that maxillaries and dentaries in the head region were first to ossify at 14.0 mm TL; but these authors did not describe the formation of the otoliths (sagittae, lapilli, or asterisci) for laboratory hatched and reared larvae collected from south Florida. Therefore, the first identifiable increment about the core could form as early as 4-5 days (yolksac absorption) to 14 days (first ossification in the head region) after hatching but no later than transformation into juvenile (25 days after hatch- ing). Thus, growth from the daily counts could be underes- timated by 4-25 days. Hubold and Mazzetti ( 1982) calculated a von Bertalanffy growth curve for H. jaguana in Guanabara Bay, Brazil, based upon three values of length-at-age assigned from length-frequency analyses of fish caught in July 1979. They identified only two modal length-at-age groups — 105 mm (age 2) and 140 mm FL (age 3) — estimating the mean length at age 1 to be 40 mm FL. Applying either Houde and Palko’s (1970) (0.7 mm TL/day growth rate for labora- 208 Fishery Bulletin 99(1) B Tampa Bay 50 40 30 20 10 b 0 Aug. 92 _2uL22_ Jan 1993 n = 2659 Apr. 92 Apr. 93 Feb^Mir 93 10 0 50 « 40 c 30 119 cm LJFL) by using a metal tape mea- sure. No measurements were obtained for live or dead under-size swordfish (i.e. <119 cm LJFL based on visual estima- tion); however, their catch was recorded. Under-size swordfish were not brought aboard the vessel but were released in the water by cutting the gangion as close to the hook as possible. Other species were not measured because most of these were also released alive after capture. Chi-square (%2) was used to determine if the observed frequency of catches by gangion type for each species differed sig- nificantly (a<0.05) from an expected 1:1 ratio for individual and combined sets. Catches of commercial-size and under- size swordfish were combined for this analysis because the method for size de- termination of under-size fish was con- sidered too subjective for a separate anal- ysis by size category. Only cases where 10 or more observations were available for each species were used for within-set comparisons. Because very few captures of any species occurred on gangions with lightsticks attached, the influence of lightsticks on catch rates in this experiment were as- sumed to be negligible, even though they are often used by fishermen (at much higher deployment rates) to improve catches of swordfish. A two-way ANOVA was used to ex- amine differences in the mean lengths (LJFL) of commer- cial-size swordfish >119 cm LJFL) by set and gangion type after testing for homogeneity of variances. Results The primary species captured included swordfish (Xiphias gladius), blue shark ( Prionace glauca), shortfin mako shark (Isuj-us oxyrinchus), pelagic stingray (Dasyatis vio- lacea), loggerhead turtle ( Caretta caretta), white marlin ( Tetrapturus albidus), common dolphinfish (Coryphaena hippurus) and yellowfin tuna ( Thunnus albacares). A total of 1093 captures (all eight species combined) occurred from 10 pelagic longline sets, with 66.7% of captures occurring on monofilament and 33.3% on multifilament gangions. Catches of swordfish averaged 38.8 fish per set (based on 334 kept plus 54 released as under-size), followed by blue shark (34.1/set), shortfin mako shark (9.7/set), pelagic stingray (9.4/set), loggerhead turtle (6.6/set), white marlin (6.0/set), common dolphinfish (3.7/set), and yellowfin tuna (1.0/set). Commercial-size swordfish ranged from 120 to 242 cm LJFL (mean=154.4 ±21.14, n= 334). Swordfish captured on monofilament nylon and multifilament nylon gangions were very similar in mean length (154.7 ±21.66 and 153.8 ±20.18 cm LJFL, respectively) and overall size composi- tion (Fig. 4). There was no significant difference between mean lengths of commercial-size swordfish by set or gan- gion type, nor was there evidence of an interaction be- tween these factors (F=0.736, P=0.690, df=333, 19), indi- cating that there were no size-related preferences for one gangion type over the other for swordfish >119 cm LJFL. Observed swordfish captures by gangion type differed significantly from an expected 1:1 ratio in all but two of the 10 sets, with higher catches on monofilament gangions compared with the multifilament (Table 1). Similarly, ob- served captures of blue sharks differed significantly from expected in 3 of 10 sets, with higher catches on the mono- filament gangions. For all other species, fewer than 10 sets were available for x2 comparisons because less than 10 in- dividuals were captured for some sets. In the case of ma- ko shark, only 5 of 10 sets had sample sizes >10, but no significant difference occurred between the observed and expected catch frequency by gangion type, although more captures occurred on monofilament gangions for most sets. For white marlin and loggerhead turtle, only 2 of 10 sets had sample sizes >10 and showed no significant differ- ences in catch by gangion type, although most occurred on monofilament. However, for pelagic stingrays, 3 out of 4 sets available for comparison showed significant differ- ences between observed and expected catches by gangion type, with higher catches on the monofilament gangions. The catch per set of yellowfin tuna and common dolphin- fish was too low for statistical comparisons. Captures of all species combined were significantly higher on monofila- ment gangions for 9 out of 10 sets (Table 1). For all species, 60% or more of catches from combined sets occurred on monofilament gangions (Table 2). When the catches by species for all 10 sets were pooled, the NOTE Stone and Dixon: A comparison of catches of Xiphias gladius and other pelagic species with longline gear 213 Figure 3 Monofilament nylon (A) and tarred multifilament nylon ( B ) gangions used for ten pelagic longline sets con- ducted off Georges Bank from 22 July to 2 August 1999. observed captures by gangion type differed significantly from an expected 1:1, with the monofilament nylon outperforming the tarred multifilament nylon gangions for 6 of 8 spe- cies (Table 2). The catch ratio by gangion type (i.e. monofilament versus multifilament) for all species combined was 2:1 and was highest for yellowfin tuna (9:1) and lowest for mako shark ( 1.5:1). Discussion The purpose of our analysis was to examine differences in pelagic longline catch by spe- cies for two different types of gangion and was based on the premise that monofila- ment nylon gangions currently used by Cana- dian pelagic longline fishermen yield higher catches than the tarred multifilament nylon gangions used in the past. Although only a small data set from a limited geographic area was available for this analysis, it was appar- ent in the case of swordfish and blue shark that catches were significantly higher on monofilament gangions, which yielded double the catch of the multifilament gangions for 214 Fishery Bulletin 99(1 ) Table 1 Summary of pelagic longline catch by species and gangion type for sets 1 through 10. Chi-square statistics (%2) and corresponding B-values are presented for comparisons of catch by gangion type for species where total catch exceeded 10 individual per set. M = monofilament gangion; B = multifilament gangion. Species caught Set 1 Set 2 Set 3 Set 4 Set 5 M B X2 P M B X2 P M B X2 P M B X2 P M B X2 P Swordfish 22 6 9.14 0.003 31 26 0.44 0.508 23 9 6.13 0.013 27 14 4.122 0.042 44 23 6.58 0.010 Yellowfin tuna 0 0 — — 0 0 — — 1 0 — — 0 0 — — 2 0 — — Mako shark 5 1 — — 7 7 0.00 1.000 11 6 1.47 0.225 5 5 0.000 1.000 11 5 2.25 0.134 Blue shark 9 3 3.00 0.083 10 7 0.53 0.467 39 19 6.90 0.009 12 5 2.882 0.090 9 6 0.60 0.439 White marlin 4 1 — — 8 1 — — 4 0 — — 7 3 1.600 0.206 5 2 — — Dolphinfish 2 0 — — 4 3 — — 2 0 — — 6 1 — — 1 2 — — Stingray 13 4 4.77 0.029 7 9 0.25 0.617 3 3 — — 7 1 — — 4 1 — — Loggerhead turtle 3 4 — — 4 0 — — 6 4 0.40 0.527 3 1 — — 4 0 — — Total 58 19 39.00 0.000 71 53 18.00 0.106 89 41 48.00 0.000 68 30 38.0 0.000 80 39 41.00 0.000 Set 6 Set 7 Set 8 Set 9 Set 10 Species caught M B X2 p M B X2 P M B X2 P M B X2 p M B X2 p Swordfish 31 12 8.40 0.004 19 8 4.48 0.034 11 4 3.27 0.071 27 14 4.122 0.042 25 12 4.57 0.033 Yellowfin tuna 0 0 — — 1 1 — — 1 0 — — 2 0 — — 2 0 — — Mako shark 3 3 — — 6 2 — — 7 4 0.82 0.366 1 4 — — 2 2 — — Blue shark 30 20 2.00 0.157 51 26 8.12 0.004 47 19 11.88 0.001 13 5 3.556 0.059 5 6 0.09 0.763 White marlin 1 1 — — 5 0 — — 1 0 — — 7 3 1.600 0.206 5 2 — — Dolphinfish 2 1 — — 2 0 — — 0 0 — — 3 2 — — 5 1 — — Stingray 11 1 8.33 0.004 12 4 4.00 0.046 1 0 — — 1 4 — — 4 4 — — Loggerhead Turtle 4 4 — — 5 2 - — 2 2 — — 5 6 0.091 0.763 4 3 — — Total 83 42 41.00 0.000 101 43 58.00 0.000 70 29 41.00 0.000 59 38 21.00 0.033 52 30 22.00 0.015 some sets. Although only the upper half of the multifil- ament gangion was made from braided nylon material (based on a configuration used by fishermen in the past), the differences in catches for these two species between this gangion and one constructed entirely of monofilament nylon were striking. A similar trend for the other pelagic species was not as evident on a set-by-set basis owing to lower catches; however, for combined sets, more captures occurred on the monofilament gear overall. Although these results clearly indicate differences in catches between gan- gion types, the influence of oceanographic conditions off Georges Bank, such as water temperature and thermo- cline depth, likely influence the availability and catchabil- ity of all species. Therefore, it is important to point out that results could differ among geographic areas with dif- ferent oceanographic regimes. Reports of higher catches on monofilament gear have al- so been made for other species. Monofilament snoods (gan- gions) give higher catch rates for cod and haddock com- pared with multifilament snoods, and thinner snoods tend to give better catch rates than thicker ones (Bjordal and Lokkeborg, 1996). Under good light conditions, monofila- ment lines were observed to catch as much as three times more cod than multifilament lines (Bjordal and Lokke- borg, 1996). Over the past decade, Canadian bluefin tuna fishermen have gradually shifted to finer gauges of mono- filament nylon line (i.e. from 400 lb to 120 lb test) for use on rod and reel gear because it is their perception that bluefin tuna have learned to recognize and avoid the heavier monofilament nylon line. The lower visibility of monofilament gangions used on longline gear is common- ly used to explain why they give better catches than gan- gions made of multifilament nylon; however, the reason for the difference in catching power is unclear. Some pelagic fish species may be able to detect multifilament nylon gan- gions more readily because of their thicker diameter (5 mm versus 2 mm for monofilament) and darker color, and can make this distinction even during periods of darkness (i.e. when the pelagic longline gear is fishing). The higher visibility of multifilament lines may cause a restrained re- sponse towards attacking the baited hooks. Whether multifilament nylon gangions are more easily detected by some species, likely depends on the role that vision plays as the dominant sensory mechanism. Some pelagic species have large eyes (i.e. yellowfin tuna, sword- fish, white marlin, mako and blue shark) and are efficient visual predators even in dim light. Becuase visual acuity or resolution of detail improves with the size of the eye (Blaxter, 1980), some pelagic species may be better at de- tecting and avoiding the multifilament nylon gangions NOTE Stone and Dixon: A comparison of catches of Xiphias gladius and other pelagic species with longline gear 215 Table 2 Summary of pelagic longline catches by species and gangion type for all sets combined. Chi-square statistics ( x2 ) and corresponding P-values are presented for all species. M = monofilament gangion; B = multifilament gangion. Species Gangion n % by gangion type Ratio (M:B) X2 P % of total catch Swordfish M 260 67.0 2.03:1.00 44.90 0.000 35.7 B 128 33.0 35.2 Yellowfin tuna M 9 90.0 9.00:1.00 6.40 0.011 1.2 B 1 10.0 0.3 Mako shark M 58 59.8 1.49:1.00 3.72 0.054 7.8 B 39 40.2 10.7 Blue shark M 225 66.0 1.94:1.00 34.84 0.000 30.9 B 116 34.0 31.9 White marlin M 47 78.3 3.62:1.00 19.27 0.000 6.5 B 13 21.7 3.6 Dolphinfish M 27 73.0 2.70:1.00 7.81 0.005 3.7 B 10 2.8 Stingray M 63 67.0 2.03:1.00 10.89 0.001 8.6 B 31 33.0 8.5 Loggerhead turtle M 40 60.6 1.54:1.00 2.97 0.085 5.5 B 26 39.4 7.1 Total M 729 66.7 2.00:1.00 123.00 0.000 100.0 B 364 33.3 100.0 than others, as evidenced by the range of catch ratios be- tween monofilament and multifilament gangions for the various species encountered during our study (Table 2). Although no differences in the size of swordfish >119 cm LJFL occurred between gangion types, more swordfish were captured on monofilament gangions, along with oth- er bycatch species. Although catches of all species were higher on the monofilament gear, the percentage of total catch represented by each species for each gear type was very similar (Table 2, last column). Therefore, by fishing an extra one or two sets with the multifilament gear, fish- ermen would get the same amount of catch as they would have if monofilament gear were used, but at greater cost because more sets and days at sea would be required to yield the same catch. Furthermore, the absence of any spe- cies-specific trends between gangion types indicates that the use of monofilament gangions does not reduce the by- catch of other pelagic species. Catch-per-unit-of-effort (CPUE) indices based on com- mercial fishery statistics are often used in analytical mod- els to investigate trends in resource abundance, particu- larly in the stock assessments for swordfish and bluefin tuna conducted by the International Commission for the Conservation of Atlantic Tunas. Catch rates are generally standardized for the effects of gear, area, month, and oth- er factors by using general linear models (e.g. Hoey et al. 1997). The importance of gear changes and their effect on commercial catch rates is clearly evident in this study and underscores the need to detect and account for changes in gear technology in the development of any commercial catch rate series used in analytical stock assessments. Acknowledgments The expertise and assistance of the crew of the Nova Blue are greatly appreciated, for without their help this study would not have been possible. We also thank R. Halliday, J. Neilson, J. Hoey, and J. Porter for their many useful com- ments on an earlier draft of this manuscript. We also thank the Canadian Department of Fisheries and Oceans for per- mitting the experiment to be conducted in a closed area. Literature cited Beckett, J. S. 1974. Biology of swordfish, Xiphias gladius L., in the north- west Atlantic Ocean. In Proceedings of the international billfish symposium, Kailua-Kona, Hawaii, 9-12 August 1972. Part 2: Review and contributed papers (R. S. Shomura and F. Williams, eds.), p. 103-106. U.S. Dep. Commer., NOAA Technical Report NMFS SSRF-675. Berkeley. S. A., E. W. Irby, and J. W. Jolly. 1981. Florida’s commercial swordfish fishery: longline gear and methods. Florida Cooperative Extension Service, Univ. Miami, Miami, FL, Florida Sea Grant Marine Advi- sory Bull. MAP-14, 23 p. Bjordal, A., and S. Lokkeborg. 1996. Longlining. Fishing News Books, Oxford, England, 156 p. Blaxter, J.H.S. 1980. Vision and the feeding of fishes. In Fish behaviour and its use in the capture and culture of fishes (J. E. Bar- dach, J. J. Magnuson, R. C. May, and J ,M. Reinhart, eds.), p. 32-56. ICLARM. Manilla, Philippines. 216 Fishery Bulletin 99(1 ) Caddy, J. F. 1976. A review of some factors relevant to management of swordfish fisheries in the northwest Atlantic. Canada. Dep. of Environment. Fish. Mar. Serv. Res. Dev. Tech. Rep. 633, 36 p. Carey, F. G, and B. H. Robison. 1981. Daily patterns in the activities of swordfish, Xiphias gladius, observed by acoustic telemetry. Fish. Bull. 79(2): 277-292. Hoey, J. J., J. M. Mejuto J. M., Porter H. H. Stone, and Y. Uozumi. 1997. An updated biomass index of abundance for North Atlantic swordfish, 1963-1995. Int. Comm. Conserv. Atl. Tunas Col. Vol. Sci. Pap. 46 (3):354-361. Hurley, P. C. F., and T. D. lies. 1981. A review of the Canadian swordfish fishery. Int. Comm. Conserv. Atl. Tunas Col. Vol. Sci. Pap. 15 ( 1 ):348 — 360. Stone, H. H., and J. M. Porter. 1999. Updated age-specific CPUE for Canadian swordfish longline (1988-1997), with information on nominal CPUE for yellowfin, bigeye and albacore tuna bycatch. Int. Comm. Conserv. Atl. Tunas Col. Vol. Sci. Pap. 49 ( 1 ):407— 422. Fishery Bulletin 99(1 ) 217 Superintendent of Documents Publications Order Form *5178 1 1 YES, please send me the following publications: Subscriptions to Fishery Bulletin for $55.00 per year ($68.75 foreign) The total cost of my order is $ . Prices include regular domestic postage and handling and are subject to change. (Company or Personal Name) (Please type or print) (Additional address/attention line) (Street address) (City, State, ZIP Code) (Daytime phone including area code) (Purchase Order No.) Charge your order. ITS EASY! Please Choose Method of Payment: ] Check Payable to the Superintendent of Documents | GPO Deposit Account [ ] VISA or MasterCard Account To fax your orders (202) 512-2250 (Credit card expiration date) (Authorizing Signature) Mail To: Superintendent of Documents P.O. Box 371954, Pittsburgh, PA 15250-7954 Thank you for your order! Fishery Bulletin Guide for Contributors Content Articles published in Fishery Bulletin des- cribe original research in fishery marine science, engineering and economics, and the environmental and ecological sciences, including modeling. Articles may range from relatively short to extensive. Notes are reports of 5 to 10 pages without an abstract and describing methods or results not supported by a large body of data. Although all contributions are subject to peer review, responsibility for the contents of papers rests upon the authors and not upon the editor or the publisher. It is there- fore important that the contents of the manuscript are carefully considered by the authors. Submission of an article is understood to imply that the article is original and is not being considered for publication elsewhere. Manuscripts should be written in Eng- lish. Authors whose native language is not English are strongly advised to have their manuscripts checked by English-speaking colleagues prior to submission. Preparation Title page should include authors’ full names and mailing addresses, the corresponding author’s telephone, FAX number, and e-mail address, and a list of key words to describe the contents of the manuscript. Abstract should not exceed one double- spaced typed page. It should state the main scope of the research but emphasize its conclusions and relevant findings. Because abstracts are circulated by abstracting agencies, it is important that they repre- sent the research clearly and concisely. Text must be typed double-spaced through- out. A brief introduction should portray the broad significance of the paper; the remain- der of the paper should be divided into the following sections: Materials and meth- ods, Results, Discussion (or Conclusions), and Acknowledgments. Headings within each section must be short, reflect a logical sequence, and follow the rules of multiple subdivision (i.e. there can be no subdivision without at least two items). The entire text should be intelligible to interdisciplinary readers; therefore, all acronyms, abbrevia- tions, and technical terms should be spelled out the first time they are mentioned. The scientific names of species must be written out the first time they are mentioned; sub- sequent mention of scientific names may be abbreviated. Follow Scientific Style and Format: the CBE Manual for Authors, Edi- tors, and Publishers (6th ed.), for editorial style, and the most current issue of the American Fisheries Society’s Common and Scientific Names of Fishes from the United States and Canada, for fish nomenclature. Dates should be written as follows: 11 November 1991. Measurements should be expressed in metric units, e.g. metric tons as t; if other units of measurement are used, please make this fact explicit to the reader. The numeral one (1) should be typed as a one, not as a lower-case el (1). Use of appen- dices is discouraged. Text footnotes should be numbered with Arabic numerals and inserted at the bottom of text. Footnote all personal communica- tions, unpublished data, and unpublished manuscripts with full address of the com- municator or author, or, as in the case of unpublished data, where the data are on file. Authors are advised to avoid references to nonstandard (gray) literature, such as internal, project, processed, or adminis- trative reports. Where these references are used, please include whether they are available from NTIS (National Technical Information Service) or from some other public depository. Literature cited comprises published works and those accepted for publication in peer reviewed literature (in press). Follow the name and year system for citation format. In the text, cite as follows: Smith and Jones (1977) or (Smith and Jones, 1977). If there is a sequence of citations, list alphabeti- cally by author! s): Smith, 1987; Smith and Jones, 1985; Smith and Peterson, 1986. Abbreviations of serials should conform to abbreviations given in Serial Sources for the BIOSIS Previews Database. Authors are responsible for the accuracy and com- pleteness of all citations. Tables should not be excessive in size and must be cited in numerical order in the text. Headings within tables should be short but ample enough to allow the table to be intelligible on its own. All unusual symbols must be explained in the table legend. Other incidental comments may be footnoted with italic numerals. Use the asterisk only to indicate probability in sta- tistical data. Zeros should precede all deci- mal points for values less than one. Figures must be cited in numerical order in the text. The senior author’s last name and the figure number should be written on the back of each one. Hand-drawn illustra- tions should be submitted as originals and not as photocopies. Submit photographs as glossy prints or slides with good con- trast, otherwise we cannot guarantee a good final printed copy. Graphs should be submitted as laser-printed copies, not as photocopies. Label all figures with Helvet- ica typeface and capitalize the first letter of the first word in axis labels. Italicize species name and variables in equations. Use zeros before all decimal points. Use uppercase Times Roman bold typeface to label the parts of a figure, e.g. A, B, C, etc. Each figure legend should explain all sym- bols and abbreviations in the figure and should be double-spaced and placed at the end of the manuscript. Copyright law Government publications are not copyrighted; they fall within the public domain. If an author reproduces any part of a government publication in his work, reference to source is appreciated. Submission Send printed copies (original and three copies without staples) to the Scientific Editor: Dr. John V. Merriner Southeast Fisheries Science Center National Marine Fisheries Service 101 Pivers Island Road Beaufort, NC 28516 Once the manuscript has been accepted for publication, you will be asked to submit an electronic copy of your manuscript to the Managing Editor. The electronic copy should be submitted in WordPerfect or Microsoft Word text format and placed on a 3.5-inch disk that is double-sided, double or high density, and that is compatible with either DOS or Macintosh systems. A copy of page proofs will be sent to the author for final approval prior to publication. Reprints (copies of published articles and notes) are available free of charge to the senior author (50 copies) and to his or her laboratory (50 copies). Additional copies may be purchased in lots of 100 when the author receives page proofs. srf I I A £ rsr3 U.S. Department of Commerce Volume 99 Number 2 April 2001 Fishery Bulletin U.S. Department of Commerce Donald L. Evans Secretary National Oceanic and Atmospheric Administration Scott B. Gudes Acting Under Secretary for Oceans and Atmosphere National Marine Fisheries Service William T. Hogarth Acting Assistant Administrator for Fisheries Scientific Editor Dr. John V. Merriner Editorial Assistant Sarah Shoffler Center for Coastal Fisheries and Habitat Research, NOS 101 Pivers Island Road Beaufort, NC 28516 The Fishery Bulletin (ISSN 0090-0656) is published quarterly by the Scientific Publications Office, National Marine Fish- eries Service, NOAA, 7600 Sand Point Way NE, BIN C 15700, Seattle, WA 981 15-0070. Periodicals postage is paid at Seattle, WA, and at additional mailing offices: POST- MASTER: Send address changes for sub- scriptions to Fishery Bulletin, Superin- tendent of Documents, Attn.: Chief, Mail List Branch, Mail Stop SSOM, Washing- ton. DC 20402-9373. Although the contents of this publica- tion have not been copyrighted and may be reprinted entirely, reference to source is appreciated. The Secretary of Commerce has deter- mined that the publication of this peri- odical is necessary according to law for the transaction of public business of this Department. Use of funds for printing of this periodical has been approved by the Director of the Office of Management and Budget. For sale by the Superintendent of Documents, U.S. Government Printing Office, Washington, DC 20402. Subscrip- tion price per year: $55.00 domestic and $68.75 foreign. Cost per single issue: $34.00 domestic and $42.50 foreign. See back for order form. Managing Editor Sharyn Matriotti National Marine Fisheries Service Scientific Publications Office 7600 Sand Point Way NE, BIN C15700 Seattle, Washington 98115-0070 Editorial Committee Dr. Andrew E. Dizon Dr. Harlyn O. Halvorson Dr. Ronald W. Hardy Dr. Richard D. Methot Dr. Theodore W. Pietsch Dr. Joseph E. Powers Dr. Harald Rosenthal Dr. Fredric M. Serchuk National Marine Fisheries Service University of Massachusetts, Boston University of Idaho, Hagerman National Marine Fisheries Service University of Washington, Seattle National Marine Fisheries Service Universitat Kiel, Germany National Marine Fisheries Service Fishery Bulletin web site: fishbull.noaa.gov The Fishery Bulletin carries original research reports and technical notes on investigations in fishery science, engineering, and economics. It began as the Bulletin of the United States Fish Commission in 1881; it became the Bulletin of the Bureau of Fisheries in 1904 and the Fishery Bulletin of the Fish and Wildlife Service in 1941. Separates were issued as documents through volume 46; the last document was No. 1103. Beginning with volume 47 in 1931 and continuing through volume 62 in 1963, each separate appeared as a numbered bulletin. A new system began in 1963 with volume 63 in which papers are bound together in a single issue of the bulletin. Beginning with volume 70, number 1, January 1972, the Fishery Bulletin became a periodical, issued quarterly. In this form, it is available by subscription from the Superintendent of Documents, U.S. Government Printing Office, Washington, DC 20402. It is also available free in limited numbers to libraries, research institutions, State and Federal agencies, and in exchange for other scientific publications. a U.S. Department of Commerce Seattle, Washington Volume 99 Number 2 April 2001 Fishery Bulletin Contents The conclusions and opinions expressed in Fishery Bulletin are solely those of the authors and do not represent the official position of the National Marine Fisher- ies Service (NOAA) or any other agency or institution. The National Marine Fisheries Service (NMFS) does not approve, recommend, or endorse any proprietary product or pro- prietary material mentioned in this pub- lication. No reference shall be made to NMFS, or to this publication furnished by NMFS, in any advertising or sales pro- motion which would indicate or imply that NMFS approves, recommends, or endorses any proprietary product or pro- prietary material mentioned herein, or which has as its purpose an intent to cause directly or indirectly the advertised product to be used or purchased because of this NMFS publication. Articles 219-239 Baumgartner, Mark F., Keith D. Mullin, L. Nelson May, and Thomas D„ Leming Cetacean habitats in the northern Gulf of Mexico 240-246 Bjorndal, Karen A., Alan B. Bolten, Bruce Koike, Barbara A. Schroeder, Donna J. Shaver, Wendy G. Teas, and Wayne N. Witzell Somatic growth function for immature loggerhead sea turtles, Caretta caretta, in southeastern U.S. waters 247-253 Bulllimore, Blaise A., Philip B. Newman, Michel J. Kaiser, Susanne E. Gilbert, and Kate M. Lock A study of catches in a fleet of "ghost-fishing" pots 254-265 Burton, Michael L. Age, growth, and mortality of gray snapper, Lutjanus griseus, from the east coast of Florida 266-274 Carrasson, Maite, and Jesus MataSlanas Feeding ecology of the Mediterranean spiderfish, Bathypterois mecliterraneus (Pisces: Chlorophthalmidae), on the western Mediterranean slope 275-291 Choi, Jung HL, and Sung Y. Hong Larval development of the kishi velvet shrimp, Metapenaeopsis dalei (Rathbun) (Decapoda: Penaeidae), reared in the laboratory 292-302 Hobson, Edmund 5., James R. Chess, and Daniel F. Howard Interannual variation in predation on first-year Sebastes spp. by three northern California predators 303-308 Hooker, Sascha K., Robin W. Baird, Sa'ad Al-Omari, Shannon Gowans, and HaS Whitehead Behavioral reactions of northern bottlenose whales ( Hyperoodon ampullatus) to biopsy darting and tag attachment procedures ii Fishery Bulletin 99(2) 309-327 Jacobson, Larry D., Jon Brodziak, and Jean Rogers Depth distributions and time-varying bottom trawl selectivities for Dover sole (Microstomus pacificus), sablefish ( Anoplopoma fimbria), and thornyheads ( Sebastolobus alascanus and S. altivelis) in a commercial fishery 328-337 Jones, Cynthia M., and Brian K. Wells Yield-per-recruit analysis for black drum, Pogonias cromis, along the East Coast of the United States and management strategies for Chesapeake Bay 338-342 Salthaug, Are Adjustment of commercial trawling effort for Atantic cod, Gadus morhua, due to increasing catching efficiency 343-350 Schaefer, Kurt M. Assessment of skipjack tuna (Katsuwonus pelamis) spawning activity in the eastern Pacific Ocean 351-355 Smith, Peter J., and Peter G. Benson Biochemical identification of shark fins and fillets from the coastal fisheries in New Zealand 356-370 Yoneda, Michio, Muneharu Tokimura, Hitoshi Fujita, Naohiko Takeshita, Koji Takeshita, Michiya Matsuyama, and Shuhei Matsuura Reproductive cycle, fecundity, and seasonal distribution of the anglerfish Lophius litulon in the East China and Yellow seas 371-380 Zimmermann, Mark, Robin C. Harrison, and Anthony F. Jones Differential parasitism by Naobranchia occidentalis (Copepoda: Naobranchiidae) and Nectobrachia indivisa (Copepoda: Lernaeopodidae) on northern rock sole (Lepidopsetta polyxystra Orr and Matarese, 2000) and southern rock sole ( L . bilineata Ayres, 1855) in Alaskan waters 381-386 Note Davis, Tim L. O., and Jessica H. Farley Size distribution of southern bluefin tuna ( Thunnus maccoyii) by depth on their spawning ground 387 Subscription form 219 Abstract— Surveys were conducted in the northern Gulf of Mexico during the spring seasons of 1992, 1993, and 1994 to determine the distribution, abun- dance, and habitat preferences of oce- anic cetaceans. The distributions of bottlenose dolphins ( Tursiops trunca- tus ), Risso’s dolphins ( Grampus griseus ), Kogia spp. (pygmy [Kogia breviceps] and dwarf sperm whales [ Kogia s/mo]), pantropical spotted dolphins (Stenella attenuata ), and sperm whales ( Physe - ter maerocephalus)'were examined with respect to depth, depth gradient, surface temperature, surface temperature vari- ability, the depth of the 15°C isotherm, surface chlorophyll concentration, and epipelagic zooplankton biomass. Bottle- nose dolphins were encountered in two distinct regions: the shallow continen- tal shelf (0-150 m) and just seaward of the shelf break (200-750 m). Within both of these depth strata, bottlenose dolphins were sighted more frequently than expected in regions of high sur- face temperature variability which sug- gests an association with ocean fronts. Risso’s dolphins were encountered over the steeper sections of the upper con- tinental slope (200-1000 m), whereas the Kogia spp. were sighted more fre- quently in waters of the upper conti- nental slope that had high zooplankton biomass. The pantropical spotted dol- phin and sperm whale were similarly distributed over the lower continental slope and deep Gulf (>1000 m), but sperm whales were generally absent from anticyclonic oceanographic fea- tures (e.g. the Loop Current, warm-core eddies) characterized by deep occur- rences of the 15°C isotherm. Habitat partitioning, high-use areas, species accounts, environmental sampling lim- itations, and directions for future hab- itat work in the Gulf of Mexico are discussed. Manuscript accepted 11 October 2000. Fish. Bull. 99:219-239 (2001). Cetacean habitats in the northern GuSf of Mexico Mark F. Baumgartner Southeast Fisheries Science Center National Marine Fisheries Service Bldg. 1 103, Room 218 John C. Stennis Space Center, Mississippi 39529 Present address: College of Oceanic and Atmospheric Sciences Oregon State University 104 Ocean Administration Building Con/allis, Oregon 97331 E-mail address: mbaumgar@oce.orst edu Keith D. Mullin Southeast Fisheries Science Center National Marine Fisheries Service PO. Drawer 1207 Pascagoula, Mississippi 39568 L. Nelson May Thomas D. Leming Southeast Fisheries Science Center National Marine Fisheries Semce Bldg. 1103, Room 218 John C. Stennis Space Center, Mississippi 39529 Studies of cetacean distribution in the northern Gulf of Mexico have largely relied on stranding, opportunistic sight- ing, and limited survey data (Jefferson and Schiro, 1997) until recently (Mullin et ah, 1994; Davis and Fargion1; Davis et al.2). During the past decade, both aerial and shipboard assessment sur- veys in the oceanic (>200 m depth) northern Gulf have identified and char- acterized the abundance and distribu- tion of 20 species of cetaceans, all but one of which were odontocetes (Mullin et al., 1994; Mullin and Hansen, 1999; Hansen et al.3; Mullin and Hoggard4). Only two of these species, the bottle- 1 Davis, R. W., and G. S. Fargion. 1996. Distribution and abundance of cetaceans in the north-central and western Gulf of Mexico: final report, vol. I: executive summary. LIS. Department of the Inte- rior, Minerals Management Service, OCS Study MMS 96-007, 29 p. [Available from Public Information Office, MS 5034, Gulf of Mexico Region, Minerals Management Service, 1201 Elmwood Park Blvd., New Orleans, LA 70123-2394.] 2 Davis, R. W., W. E. Evans, and B. Wiirsig. 2000. Cetaceans, sea turtles and seabirds in the northern Gulf of Mexico: distribution, 2 ( continued ) abundance and habitat asso- ciations, vol. I: executive summary. U.S. Department of the Interior, Geological Sur- vey, Biological Resources Division, LTSGS/ BRD/CR- 1999-0006 and Minerals Man- agement Service, OCS (outer continental shelf) Study MMS 2000-003, 27 p. (Avail- able from Public Information Office, MS 5034, Gulf of Mexico Region, Minerals Management Service, 1201 Elmwood Park Blvd., New Orleans, LA 70123-2394.] 3 Hansen, L J , K. D. Mullin, T. A. Jefferson, and G. P. Scott. 1996. Visual surveys aboard ships and aircraft In Distribution and abundance of cetaceans in the north- central and western Gulf of Mexico: final report, vol. II: technical report (R. W. Davis and G. S. Fargion, eds.), p. 55-128. LIS. Department of the Interior, Minerals Management Service, OCS Study MMS 96-007. [Available from Public Information Office, MS 5034, Gulf of Mexico Region, Min- erals Management Service, 1201 Elmwood Park BlvcL, New Orleans, LA 70123-2394.] 4 Mullin, K. D., and W. Hoggard. 2000. Visual surveys of cetaceans and sea turtles from aircraft and ships. In Cetaceans, sea turtles and seabirds in the northern Gulf of Mexico: distribution, abundance and habitat associations, vol. II: technical report (R. W. Davis, W. E. Evans, and B. Wiirsig, eds. ), p. 11 1-171 U.S. Department of the Interior, footnote continued on next page 220 Fishery Bulletin 99(2) nose dolphin (Tursiops truncatus ) and Atlantic spotted dolphin (Stenella frontalis), occur regularly over the conti- nental shelf (Fritts et al., 1983; Mullin et ah, 1994; Davis et ah, 1998). In contrast, the oceanic Gulf supports a wide diversity of cetacean species by potentially supplying a large number of ecological niches. Although predator avoid- ance, interspecific competition, and reproductive strategies all affect cetacean distribution to some extent, energetic budget studies indicate that most cetaceans must feed every day (Smith and Gaskin, 1974; Lockyer, 1981; Kenney et ah, 1985; CETAP5) and thus habitat is assumed to be pri- marily determined by the availability of food (Kenney and Winn, 1986). The distribution of the oceanic species, then, is presumably linked to the rather dynamic oceanography of the Gulf of Mexico through physical-biological interac- tions and trophic relationships between phytoplankton, zooplankton, micronekton, and cetacean prey species. For most cetaceans in the Gulf of Mexico, specific prey species are not known but likely include epi- and mesopelagic fish and cephalopods (Fitch and Brownell, 1968; Perrin et ah, 1973; Wurtz et ah, 1992; Clarke, 1996). The physical and biological oceanography of the north- ern Gulf of Mexico is highly variable in both space and time. The eastern Gulf contains the Loop Current, an ex- tension of the Gulf Stream system that enters the Yucat- an Channel, turns anticyclonically, and exits through the Straits of Florida. The northward penetration of the Loop Current into the Gulf of Mexico normally varies between 24° and 28°N on a quasi-annual basis (Sturges and Evans, 1983). Cold, potentially biologically rich, upwelling fea- tures are frequently found at the edge of the Loop Current and often develop into cyclonic, cold-core eddies (Vukovich et ah, 1979; Maul et ah, 1984; Vukovich and Maul, 1985; Richards et ah, 1989). Large, anticyclonic, warm-core ed- dies can shed from the Loop Current during its maximum northerly penetration into the Gulf (Cochrane 1972; Hurl- burt and Thompson, 1982) after which they move slowly westward at an average speed of 5 km/day. More than one of these warm-core eddies can be found in the west- ern Gulf of Mexico because their translation (net) speed and decay are slow (Elliot, 1982). During their transit from the eastern to western Gulf of Mexico, these warm- core features can also have associated cyclonic features at their peripheries which are biologically productive (Biggs, 1992). Another major source of nutrients that can drive primary productivity in the oceanic Gulf is the Mississippi River. The Mississippi River Delta protrudes into the Gulf 4 ( continued from previous page) Geological Survey, Biological Resources Division, USGS/BRD/CR-1999-0006 and Minerals Management Service, OCS Study MMS 2000-003. (Available from Public Information Office, MS 5034, Gulf of Mexico Region, Minerals Management Service, 1201 Elmwood Park Blvd., New Orleans, LA 70123-2394.] ’ CETAP (Cetacean and Turtle Assessment Program). 1982. A characterization of marine mammals and turtles in the mid- and north Atlantic areas of the U.S. outer continental shelf. LIS. Department of the Interior, Bureau of Land Management, con- tract AA551-CT8-48. 584 p. [Available from National Techni- cal Information Service, LIS. Department of Commerce, 5285 Port Royal Road, Springfield, VA 22161.] in a region where the continental shelf is narrow and the continental slope is steep. The river’s nutrient-rich fresh water plume extends over the deep Gulf and supports high rates of primary productivity and large standing stocks of chlorophyll and zooplankton biomass (El-Sayed, 1972; Dagg et al., 1988; Ortner et al., 1989). Our study examines the distribution of five commonly encountered cetacean species or species groups in the northern Gulf of Mexico with respect to several physical, biological, and physiographic variables. These species are the bottlenose dolphin, Risso’s dolphin (Grampus griseus), Kogia spp. (pygmy [ Kogia breviceps | and dwarf sperm whale [Kogia sima]), pantropical spotted dolphin ( Stenel- la attenuata ) and sperm whale (Physeter macrocephalus) . The environmental and cetacean survey data for our study were collected by the U.S. National Marine Fish- eries Service. Subsets of these data have been analyzed by Baumgartner ( 1997 ) to characterize the distribution of Risso’s dolphins with respect to the physiography of the northern Gulf of Mexico and by Davis et al. (1998) to de- scribe cetacean habitats over the continental slope in the northwestern Gulf. One of the major objectives of these surveys was to help assess the impact of large-scale oil and gas exploration and development in the northern Gulf of Mexico on cetaceans. An understanding of the habitat preferences of each of these species will greatly improve management and conservation efforts by providing a con- text for interpreting future anthropogenic influences on cetacean distribution. Materials and methods Data collection and treatment We examined the distribution of each cetacean species with respect to seven environmental variables (Table 1) to char- acterize habitat. These variables were selected because they represent specific oceanographic or physiographic features or conditions. Depth and depth gradient (sea floor slope) were included to represent the physiography of the Gulf of Mexico because the distribution of some cetaceans has been associated with specific topographic features in the Gulf (Baumgartner, 1997; Davis et al., 1998) and elsewhere (Evans, 1975; Hui, 1979, 1985; Selzer and Payne, 1988; CETAP5; Dohl et al.6; Dohl et al.7; Green et al.8). A com- 6 Dohl, T. P., K. S. Norris, R. C. Guess, J. D. Bryant, and M. W. Honig. 1978. Summary of marine mammal and seabird sur- veys of the Southern California Bight area 1975-78, vol. Ill: Investigators’ Reports, part II: Cetacea of the Southern Cali- fornia Bight. U.S. Department of the Interior, Bureau of Land Management, Contract AA550-CT7-36, 414 p. [Available from National Technical Information Service, U.S. Department of Commerce, 5285 Port Royal Road, Spring-field, VA 22161.] 7 Dohl, T. P, R. C. Guess, M. L. Duman, and R. C. Helm. 1983. Cetaceans of central and northern California, 1980-1983. Status, abundance and distribution. U.S. Department of the Interior, Minerals Management Service, contract 14-12-0001-29090, 284 p. [Available from National Technical Information Service, U.S. Department of Commerce, 5285 Port Royal Road, Springfield, VA 22161.1 Baumgartner et at: Cetacean habitats in the northern Gulf of Mexico 221 Table 1 Environmental variables used in the habitat analyses. Variable Source Units Depth digital bathymetry m Depth gradient digital bathymetry m/1.1 km Surface temperature thermosalinograph °C Surface temperature standard deviation infrared satellite imagery °C Depth of 15°C isotherm CTD and XBT casts m Surface chlorophyll concentration surface samples mg/m3 Zooplankton biomass oblique bongo tows cc/100 m:i mon measure of bottom relief, contour index (Evans, 1975), was omitted because it does not distinguish between signifi- cantly different topographies in the northern Gulf of Mexico (Baumgartner, 1997). Many oceanographic features, such as eddies or river discharge, have strong sea surface temperature signatures, whereas areas where different water masses abut (frontal zones) are often charac- terized as regions of high surface temperature variability. Meso- scale warm-core eddies in the Gulf of Mexico are easily detected in hydrographic tran- sects by the deep occurrence of the 15°C isotherm. Finally, surface chlorophyll concentration and zooplankton biomass represent rough measures of the standing stocks on which higher trophic consumers might feed. Cetacean surveys were conducted during the spring sea- sons of 1992, 1993, and 1994 from NOAA Ship Oregon II in the Gulf of Mexico approximately north of a line con- necting Brownsville, Texas, and Key West, Florida, and primarily in waters deeper than 200 m (Fig. 1). Sighting data were collected with 25x binoculars and standard line- transect survey methods for cetaceans (e.g. Barlow, 1995; Hansen et al.3). Time and the ship’s position were recorded automatically every two minutes, and at regular intervals the survey team recorded ancillary data, such as sea state, sighting conditions, and effort status. These ancillary data were appended to the time and position records. Environ- mental data were extracted from the appropriate data sets (discussed below) and also appended to the time and po- sition records. These records comprise the effort data set which provides a complete history of the sighting condi- tions, survey effort, and environmental observations. The cetacean sighting records were also appended with the en- vironmental and ancillary data and collectively represent the cetacean sighting data set. Surface temperature was recorded at one-minute in- tervals with a flow-through thermosalinograph (SeaBird Electronics, Inc, Bellevue, WA). The temperature measure- ments were low-pass filtered to reduce high frequency and high wave number variability. The filter was a simple 5-min running mean which, at an average vessel speed of 5 m/s ( 10 knots), is equivalent to averaging over 1.5 km. Conductivity, temperature, and depth (CTD) or expend- able bathythermograph (XBT) casts were conducted every 55 km (30 nmi) along the survey transect. CTD casts were 8 Green, G. A., J. J. Brueggeman, R. A. Grotefendt, C. E. Bowlby, M. L. Bonnell, and K. C. Balcomb III. 1992. Ceta- cean distribution and abundance off Oregon and Washington, 1989-1990. In Oregon and Washington marine mammal and seabird surveys (J. J. Brueggeman, ed.), p. 1-100. U.S. Depart- ment of the Interior, Minerals Management Service, contract 14-12-0001-30426. [Available from National Technical Infor- mation Service, U.S. Department of Commerce, 5285 Port Royal Road, Springfield, VA 22161.] generally made to 500 m or just off the sea floor, whichever was shallower. XBT probes capable of operating to depths of 200 to 1000 m were used in appropriate depths. Surface water samples were collected every 55 km and chlorophyll a was measured in these samples by using fluorometric and spectrophotometric techniques described in Strickland and Parsons (1972) and Jeffery and Humphrey (1975). Plank- ton tows were also conducted at 55-km intervals by using a 61-cm diameter bongo equipped with 0.333-mm mesh nets and flowmeters. The nets were towed obliquely from 200 m or just off the sea floor, whichever was shallower. Samples from one of the bongos were analyzed by the Polish Sorting and Identification Center in Szczecin, Poland. Zooplankton biomass was computed as the ratio of the displacement vol- ume of the sample after organisms larger than 2.5 cm were removed (after Smith and Richardson, 1977) to the volume of water filtered during the tow. Remotely sensed sea surface temperature ( SST) data from the advanced very high resolution radiometer (AVHRR) carried aboard the National Oceanic and Atmospheric Ad- ministration (NOAA) polar orbiting environmental satel- lites were acquired from the U.S. National Environmental, Satellite and Data Information Service. The raw, level IB data from the NOAA 9, 10, and 11 satellites were warped to a 0.01° x0.01° linear latitude-longitude projection by using the supplied satellite navigation information, coregistered to a digital coastline and converted to sea surface tempera- tures by using separate day and night multichannel SST equations. Because of the lower accuracy and relative pau- city of the satellite-derived SST data, the in -situ surface temperature from the shipboard thermosalinograph was used in the analyses of cetacean habitat. However, these remotely sensed data are well suited to detecting horizon- tal gradients in SST due to their synoptic coverage. These gradients are often resolved by using digital image gradi- ent operators (e.g. Sobel, Prewitt, or Roberts operators), but we chose another approach after Smith et al. (1986). Because horizontal gradients in SST can be measured as horizontal variability, we computed the standard deviation of the remotely sensed SST within a 10-km radius of each transect and sighting position. Water depth was extracted from a digital bathymetric da- ta set compiled from NAVOCEANO's DBDB5 5-minute x 5 minute gridded bathymetry. National Ocean Service’s high 222 Fishery Bulletin 99(2) Map of shipboard surveys transects conducted by NOAA Ship Oregon 11 in the spring seasons of 1992, 1993, and 1994. Only transects conducted during active searches for cetaceans during adequate sighting conditions are shown. The 200-m and 2000-m isobaths are indicated in gray. resolution coastal bathymetric data set and Texas A&M University’s digitized bathymetric charts (Herring9). This depth data set was provided on a 0.01° x 0.01° linear latitudedongitude grid with a nominal resolution of 1.1 km for the entire Gulf of Mexico. Depth gradient or sea floor slope was derived from the depth grid by using a 3x3 pixel Sobel gradient operator. The resulting product had the same base resolution and spatial coverage as the bathymetry data set. For descriptive purposes, the fol- lowing physiographic terms will be used to denote spe- cific depth ranges or features: continental shelf (0-200 m), shelf break (-200 m), continental slope (200-2000 m), upper continental slope (200-1000 m), lower continental slope ( 1000-2000 m), and deep Gulf (>2000 m). A single descriptor of the vertical temperature structure in the upper ocean was selected to quantify the influence of mesoscale features such as eddies on cetacean distribu- tion. Reilly (1990) chose the depth of the 20°C isotherm 9 Herring, H. J. 1993. A bathymetric and hydrographic cli- matological atlas for the Gulf of Mexico (draft report). U.S. Department of the Interior, Minerals Management Service, con- tract 14-12-0001-30631, 191 p. [Available from National Tech- nical Information Service, U.S. Department of Commerce, 5285 Port Royal Road, Springfield, VA 22161.] as an approximate indicator of thermocline depth in his study of cetacean habitat in the eastern tropical Pacific. We used a similar approach by extracting the depth of the 15°C isotherm from each CTD and XBT profile. This vari- able is not intended to represent the depth of the thermo- cline, however. The low-frequency, large-scale temperature variability along this isotherm is associated with the me- soscale features of interest and it occurs deep enough that it never reaches the sea surface during the spring in the northern Gulf of Mexico. The discrete samples of the depth of the 15°C isotherm, surface chlorophyll concentration and zooplankton bio- mass from each cruise leg (9-17 days in duration) were interpolated on a regular 0.1° x 0.1° linear latitude-longi- tude grid by using the kriging method (Golden Software, 1994). Surface chlorophyll was log-transformed before in- terpolation because the observed chlorophyll concentra- tions had a log-normal distribution and spanned several orders of magnitude (0.02-13.02 mg/m3). The interpola- tion method provided consistent results when compared with other data sets (e.g. Fig. 2). Because no interpolation method will capture the true spatial structure of these variables, the accuracy of the interpolated values in the effort and sighting datasets is undoubtedly low. Despite these errors, however, the horizontal variability associated Baumgartner et al.: Cetacean habitats in the northern Gulf of Mexico 223 30°N 98 96 94 92 90 88 86 84 82°W Figure 2 Sea surface temperature of the northern Gulf of Mexico derived from remotely sensed AVHRR data collected between 21 and 23 May 1993. Image is a histogram-equalized, warmest-pixel composite of data derived from three satellite passes with some cloud contamination south of 24.5°N and also west of 93°W. CTD and XBT stations are indicated as filled circles and the contours represent the depth of the 15°C isotherm computed from the CTD and XBT casts collected between 19 May and 1 June 1993. The line along 27°N indicates the parallel from which data were extracted for Figure 3. The 200-m isobath is shown. 26 32 ' 30. 26 1 24 8 03 22 3 20 g co with mesoscale oceanographic features is much larger than these errors and therefore the interpolated fields rep- resent these features reasonably well (e.g. Fig. 2). The base unit of effort for this study was defined as 1 km of actively surveyed transect during adequate sighting conditions. To conform to this definition, each contiguous transect in the effort data set was broken into 1-km linear sections and all the environmental variables measured along each 1-km section were averaged. This provided a single set of observed environmental variables for each unit of effort. Only those 1-km sections that were actively surveyed (i.e. those where the observers were on-effort) during adequate sighting conditions (defined as Beaufort sea states of 3 or less) were used for analysis. Similarly, only those cetacean sightings that occurred while observ- ers were on-effort and in Beaufort sea states of 3 or less were used for analysis. All of the following analyses were conducted on cetacean group sightings and therefore do not account for group size. Some portions of the described data have been previ- ously published by Davis et al. (1998) and Baumgartner (1997). Davis et al. (1998) examined cetacean habitat in the northwestern Gulf of Mexico with respect to a variety of physical oceanographic and physiographic variables. We have included the sighting data and some of the environ- mental data from that study here (less than 40% of our total data set) to examine cetacean habitat throughout the entire northern Gulf of Mexico with an expanded set of environmental variables and new statistical analyses. With regard to Risso’s dolphin habitat, we have used the same sighting, depth, and depth gradient data presented in Baumgartner (1997). To these, we have added physical and biological oceanographic variables to test and extend the conclusions of Baumgartner ( 1997) and to strengthen the univariate and multivariate interspecies comparisons described below. Analytical methods The analysis of the sighting and effort data sets was conducted in two parts: 1) univariate and multivariate interspecies comparisons of the environmental variables measured at each cetacean sighting and 2) comparisons of each species’ distribution with respect to the environ- mental variables to that of the effort. The former analysis examined the null hypothesis that each species had simi- lar distributions with respect to each of the environmen- tal variables. This was tested with Mood’s median test (Conover, 1980) and the Kruskal-Wallis test (Sokal and Rohlf, 1981) as nonparametric substitutes for a one-way analysis of variance. Multivariate analysis of variance (MANOVA) and canonical linear discriminant function (LDF) analysis (Huberty, 1994; Johnson, 1998) with rank- transformed environmental variables were used to further examine interspecies differences. These analyses were con- ducted with the CANDISC procedure of the Statistical Analysis System (SAS, 1989), version 6.12. The MANOVA detects species group differences in multivariate space and 224 Fishery Bulletin 99(2) the canonical LDF analysis describes which environmen- tal factors contribute most to these group differences. The canonical LDF analysis is accomplished by finding a linear combination of the environmental variables that best dis- criminates between the species groups. These linear com- binations (canonical variables) are then examined by using the LDF structure correlations (Fluberty, 1994) to assess their ecological meaning and significance. The structure correlations are essentially the correlations between the canonical variables and the original environmental vari- ables and their interpretation is analogous to the interpre- tation of factor loadings in factor analysis. The second analysis uses univariate and bivariate chi- squared (^2) tests, Mann- Whitney tests, Monte Carlo tests, and equal-effort sighting rate distribution plots to deter- mine the specific relationships between the distribution of each species and each of the environmental variables. For the x2 analysis, the effort data were used to compute ex- pected uniform distributions for each species with respect to the individual environmental variables. Classes were chosen such that each contained an equal amount of effort (Kendall and Stuart, 1967). This approach "normalized” the sighting rates by creating class sizes of equal sighting prob- ability based on the effort and guaranteed that the anal- ysis would not be distorted by classes with exceptionally low or high amounts of effort. For a complete description of the methods used to compute the uniform distribution, see Baumgartner (1997). The actual distributions were then compared with the predicted uniform distributions by us- ing the x2 statistic. Equal-effort sighting rate distribution plots were constructed directly from the contingency tables used in the x2 analyses. In some cases, the sample size was lower than the minimum required for a conservative x2 test ( 72 =25 ), therefore the species’ and effort distributions were compared by using a Mann-Whitney test. Of the five species examined here, each had a distribu- tion with respect to depth that was significantly different from a uniform distribution. Further analyses with Monte Carlo (randomization) tests were conducted to determine if the distribution of a particular species with respect to the other environmental variables was an artifact of that spe- cies’ distribution with depth. For example, consider a hypo- thetical species that is only found on the continental shelf. The continental shelf in the northern Gulf of Mexico is char- acterized by low depth gradients, whereas the continental slope has high depth gradients and the abyssal plains of the deep Gulf have low depth gradients. Because this species occurs on the continental shelf, it would have distributions with respect to both depth and depth gradient that were significantly different from a uniform distribution. Howev- er, this species’ distribution with respect to depth gradient is merely an artifact of its distribution with respect to depth because of a correspondence between shallow depths and low depth gradients over the continental shelf. The Monte Carlo tests consisted of randomly choosing n transect sections from the effort data set that had the same depth distribution as the n sightings of the species of interest. These transect sections represent n “virtual” cetacean sightings that have the same depth distribution as the species of interest but have a random distribution with respect to all of the other environmental variables. A x2 analysis was then conducted to determine if the dis- tribution of the “virtual” sightings with respect to the par- ticular environmental variable of interest (e.g. depth gra- dient in the example above) was different from a uniform distribution predicted by the effort. The process of choos- ing ?i “virtual” sightings and of conducting the x2 analysis was performed 10,000 times. The proportion of the result- ing 10,000 x2 statistics that exceeded the x2 statistic as- sociated with the species’ actual distribution with respect to the environmental variable of interest was considered a P-value. This P-value represented the probability that the actual x2 statistic could have been observed by chance and was used to test the null hypothesis that the species’ distribution with respect to the environmental variable of interest was the same as a uniform distribution given its distribution with respect to depth. Results NOAA Ship Oi egon II completed 113 days of effort during the spring surveys from 1992 to 1994 and sampled the entire oceanic northern Gulf of Mexico once each year. A total of 9101 1-km transect sections (units of effort) were completed during adequate sighting conditions. The amount of environmental data available for each transect section was dependent on survey design, on instrument availability and performance, and, in the case of the re- motely sensed sea surface temperature variability, on sat- ellite orbital parameters and cloud conditions (Table 2). The Loop Current penetrated into the eastern Gulf to at least 27.5°N during each of the surveys and warm-core eddies could usually be found in the central and western Gulf (Fargion et al. 10 ). Both the Loop Current and the warm-core eddies were often accompanied by cold-core fea- tures at their peripheries. Examples of the major oceano- graphic features of the northern Gulf are shown in the composite AVHRR sea surface temperature image and the contoured depth of the 15°C isotherm (Fig. 2). The Loop Current is easily identifiable as the broad region in the eastern Gulf where the 15°C isotherm was at depths be- low 250 to 300 m and sea surface temperatures reached a local maximum. The remnants of a warm-core eddy (Eddy V) are evident in the northwestern Gulf centered at about 27.0°N, 95.5°W (Jockens et al., 1994; Fargion et al.10). Warm-core features like the Loop Current were characterized by depressed isotherms and were often ac- companied by warm surface temperatures and low zoo- plankton biomass (Fig. 3). Surface temperature gradients were high at the edge of these mesoscale features when 10 Fargion, G. S.. L. N. May, T. D. Leming, and C. Schroeder. 1996. Oceanographic surveys. In Distribution and abundance of cetaceans in the north-central and western Gulf of Mexico: final report, vol.I I: technical report (R.W. Davis and G.S. Far- gion, eds.), p. 207-269. U.S. Department of the Interior, Miner- als Management Service, OCS Study MMS 96-007. [Available from Public Information Office, MS 5034, Gulf of Mexico Region, Minerals Management Service, 1201 Elmwood Park Blvd., New Orleans, LA 70123-2394.1 Baumgartner et al.: Cetacean habitats in the northern Gulf of Mexico 225 Table 2 Number of 1-km transect sections (units of effort) with valid data for each environmental variable. Variable 1992 1993 1994 Total Depth 3454 2373 3274 9101 (100%) Depth gradient 3454 2373 3274 9101 (100%) Surface temperature 1414 2245 2915 6574 (72%) Surface temperature standard deviation 688 1084 498 2270 (25%) Depth of 15°C isotherm 2357 1939 2669 6965 (77%) Surface chlorophyll concentration 2844 2277 2859 7980 (88%) Zooplankton biomass 2127 1103 1419 4649 (51%) Table 3 Correlation matrix of environmental variables. Correlation coefficients for surface chlorophyll and zooplankton biomass were com- puted from the station samples (not from the interpolated fields). SD = standard deviation. Variable Depth Depth gradient Surface temperature Surface temperature SD Depth of 15°C isotherm Surface chlorophyll Depth gradient -0.003 Surface temperature 0.104 0.098 Surface temperature SD -0.067 0.032 0.019 Depth of 15°C isotherm 0.297 -0.139 0.199 0.365 Surface chlorophyll -0.341 0.013 -0.250 -0.165 -0.166 Zooplankton biomass -0.224 -0.064 -0.192 0.141 -0.380 0.710 indicates P < 0.05. indicates P < 0.01. Table 4 Number of group sightings in ), sighting rate (group sightings per 100 km) and mean, standard deviation (SD), minimum (Min) and maximum (Max) group size of the five most frequently encountered species or species groups. Group size Species n Sighting rate Mean SD Min Max Bottlenose dolphin 89 0.98 14.7 22.7 1 150 Risso’s dolphin 67 0.74 10.8 7.3 2 40 Kogia spp. 56 0.62 2.1 1.6 1 8 Pantropical spotted dolphin 107 1.18 51.8 42.4 3 180 Sperm whale 43 0.47 2.5 1.9 1 11 their surface temperature signa- tures were strong. Many of the environmental vari- ables in the effort data set were significantly correlated with one another (P<0.05), but correlation coefficients were less than 0.3 in most cases (Table 3). The high correlation between surface chlo- rophyll and zooplankton biomass was strongly influenced by stations on the continental shelf where both surface chlorophyll and zoo- plankton biomass were typically quite high. The surface chloro- phyll measurements collected on the shelf were significantly high- er than those from oceanic waters (Mann-Whitney, £7=4.44, P<0.0001), whereas the median zooplankton biomass measured on the shelf ( 10.1 cc/100 nr3) was almost twice as large as the median of the oceanic obser- vations (5.4 cc/100 m3). The correlation coefficient between the surface chlorophyll and zooplankton biomass measured in oceanic waters was not significantly different from zero (P>0.05). This contrast between the continental shelf and more oceanic waters was also manifested in the inverse re- lationships detected between depth and surface chlorophyll and between depth and zooplankton biomass. Of the 614 cetacean groups sighted between 1992 and 1994, the most frequently encountered species were the bottlenose dolphin, Risso’s dolphin, Kogia spp., pantropi- cal spotted dolphin, and sperm whale (Table 4). The Ko- 226 Fishery Bulletin 99(2) gici spp. group was made up of sightings of dwarf sperm whales (n=32), pygmy sperm whales (n= 7), and those small whales that could only be identified to the genus Kogia (n= 17). The spatial distributions of sightings suggest high use areas for each species (Fig. 4), but these are heavily in- fluenced by the distribution of the sighting effort. To better capture the true spatial distributions, the location of each sighting and 1-km transect section was projected onto the 200-m isobath (by using the minimum distance to this iso- bath) and equal-effort sighting distributions were gener- ated with respect to the distance along the 200-m isobath (Fig. 4). Chi-squared analyses indicated that all of the spe- cies’ distributions, except the Kogia spp. group, were sig- nificantly different from a uniform distribution (PcO.Ol). The northwestern Gulf of Mexico (west of the Mississippi River Delta) had much lower group sighting rates of each cetacean species when compared with the northeastern Gulf (east of the Mississippi River Delta). To the south of New Orleans, the Mississippi Canyon and just seaward of the Mississippi River Delta were regions of high group encounter rates for bottlenose dolphins, Risso’s dolphins, and sperm whales. Just to the east of this region and south of Mobile Bay, pantropi- cal spotted dolphin sighting rates reached a lo- cal maximum. Along the steep upper continen- tal slope of the Florida Escarpment between Tampa and Key West, very high relative abun- dances of Risso’s dolphins, pantropical spotted dolphins, and sperm whales were observed. Examination of the mean, median, first and third quartiles, and standard deviation of the environmental variables for each species sug- gested significant interspecies differences (Fig. 5). The null hypothesis of equal medians for each species was rejected for depth, depth gra- dient, surface temperature, and zooplankton biomass (Mood’s median test, .PcO.Ol). Similar- ly, the null hypothesis of equal “locations” was rejected with a Kruskal-Wallis test for depth, depth gradient, zooplankton biomass (PcO.Ol), and surface temperature (P<0.05). The bottle- nose dolphin had the lowest median depth, depth gradient, and surface temperature of all the species. The Risso’s dolphin had the high- est median depth gradient and surface tem- perature and the Kogia spp. had the highest median zooplankton biomass. The bottlenose dolphin’s median habitat was so different from the others that if this species was removed from each of the Mood’s median tests, the null hypothesis of equal medians between species would be rejected for only depth (PcO.Ol) and zooplankton biomass (Pc0.05). Despite clear heterogeneity of variances (Fig. 5A), a one-way analysis of variance indicated that the cetacean distributions with respect to depth were significantly different. Further- more, a Duncan’s multiple range test suggest- ed species groupings by depth (Pc0.05) that were qualitatively accurate and in agreement with earlier results (Mullin et al., 1994). These species groupings were 1 ) bottlenose dolphins, 2) Risso’s dolphins and Kogia spp., and 3) pan- tropical spotted dolphins and sperm whales. Bottlenose dolphins were encountered predom- inantly over the continental shelf and were never sighted seaward of the 750-m isobath. Risso’s dolphins and Kogia spp. were dis- tributed mostly over the upper continental slope, whereas pantropical spotted dolphins o CD £9 CD Cl co E ) o CO a> - Q_ CO 11 CD "O O J- co -1 w 2 CO 27 26 25 24 23 0.8 0.6 0.4 : 0.2- 0.0 8 E c o 6- o o c CD _cd Q- CO O CO O CD N £ 4 2 ■ 31 f o 100 f 200 4 -C a 300 400 500 V V V V Figure 3 Surface temperature (raw temperatures in gray, 10 km radius average in black), surface temperature variability (standard deviation in 10 km radius), epipelagic zooplankton biomass and vertical temperature struc- ture along 27°N observed between 19 May and 1 June 1993. Surface temperature and surface temperature variability were obtained from the composite satellite image shown in Figure 2. The inverted triangles indicate CTD/XBT station locations from which the temperature section was derived. Baumgartner et al.: Cetacean habitats in the northern Gulf of Mexico 227 Bottlenose Dolphin Q. 3 - o. 2 - 2 1 O o 1000 1500 2000 Distance Along 200m Isobath (km) 98 96 94 92 90 88 86 84 82 80 W Galveston New Orleans Tampa Brownsville Lake Charles Panama City Key West 0 500 1000 1500 2000 2500 3000 Distance Along 200m Isobath (km) 0 500 1000 1500 2000 2500 3000 Distance Along 200m Isobath (km) Galveston New Orleans Tampa Brownsville Lake Charles Panama City Key West Galveston New Orleans Tampa Brownsville Lake Charles Panama City Key West 0 500 1000 1500 2000 2500 3000 Distance Along 200m Isobath (km) Figure 4 Spatial distribution of group sightings and sighting rates for each species. Sightings and 1-km transect sections (effort) were projected onto the 200-m isobath and the group sighting rate distributions were computed by using equal-effort class sizes. The sighting rate distribution of the bottlenose dolphin only includes effort from 1000 m depth and shallower because no bottlenose dolphins were encountered seaward of the 750-m isobath. The sighting maps and sighting rate distribution plots are aligned geographically to facilitate comparison. The 200- and 2000-m isobaths are shown in the sighting maps. 228 Fishery Bulletin 99(2) °F 1000 2000 3000 : 89 67 56 E BD RD KS 107 PSD 43 SW 100 80 ^ 60 I 40 td m 6 20 0 - -20 L B S 89 BD 67 56 107 43 RD KS PSD SW 30 28 26 24 22 L 57 BD 39 RD 28 KS 62 PSD 14 SW 0.6 0.4 ra 0.2 a oo D 30 BD 14 RD 18 KS 31 PSD □ 11 SW ioo r 150 E O 200 250 3001 a 1 00 39 BD 52 RD 47 KS 99 PSD 34 SW 2 0.10 0.01 57 BD 49 RD 46 KS 0 102 PSD 39 SW 1 5 r G 10 - 40 BD 44 RD 35 KS 57 PSD 26 SW Mean + Stdev 3rd Quartile Mean -H Median 1st Quartile Mean - Stdev E = Effort BD = Bottlenose dolphin RD = Risso's dolphin KS = Kogia species PSD = Pantropical spotted dolphin SW = Sperm whale Figure 5 Mean, median, interquartile range, and standard deviation of (A) depth, (B) depth gradient, (C) surface temperature, (D) surface temperature variability, (E) depth of the 15°C isotherm, (F) surface chlorophyll, and (G) epipelagic zoo- plankton biomass for each species and the 1-km transect sections (effort). The sample size (n) is shown above each species abbreviation. Baumgartner et at: Cetacean habitats in the northern Gulf of Mexico 229 and sperm whales had distributions that extended from the upper continental slope to the deep Gulf. Mann-Whit- ney tests between Risso’s dolphins and Kogia spp. for each of the environmental variables indicated that only their distributions with respect to depth gradient ( {7=2.12, P<0.05) and zooplankton biomass ({7=1.69, P<0.05) were significantly different. Similar tests between pantropical spotted dolphins and sperm whales indicated that their distributions with respect to the depth of the 15°C iso- therm ({7=2.26, P<0.05) alone were significantly different. Differences between species were also detected with MANOVA and canonical linear discriminant function anal- ysis. Unfortunately, low sample size for both sea surface temperature and sea surface temperature variability pre- cluded their use in the multivariate analyses. Of the re- maining variables, the sample sizes for each species were as follows: bottlenose dolphins (n=18), Risso’s dolphins (n=35), Kogia spp. (n- 25), pantropical spotted dolphins (t?=51 ), and sperm whales all species except the bottlenose dolphin. The structure correlations associated with each canonical axis represent the approximate correlations between the canonical variables and depth (DP), depth gradient (DPG), depth of the 15°C isotherm (D15C), surface chlorophyll concentration (CHL), and epipelagic zooplankton biomass (PL). Species abbreviations are the same as those shown in Figure 5. 230 Fishery Bulletin 99(2) Q- D O a 5 - 3r 2-. 0 J 200 400 600 800 1000 1200 Depth (m) 20 22 24 26 28 Surface temperature (°C) 30 F6 - 4 •2 0.2 0 3 0.4 1.1 1.2 Surface temperature SD (°C) Figure 7 Sighting rate distributions of bottlenose dolphins with respect to (A) depth, (B) surface temperature and (C) surface temperature variability computed by using equal-effort, class sizes. again in the MANOVA (Wilks’ A=0.675, PcO.0001). The first two canonical variables in the canonical LDF analysis accounted for 94.2% of the total variability, and likelihood ratio tests indicated that only these first two canonical variables were significant (P<0.0001 for the first, P<0.05 for the second). The correlation structure suggested that high values of zooplankton biomass and deep occurrences of the depth of the 15°C isotherm were associated with positive values of the first canonical variable, whereas high values for depth and surface chlorophyll were associ- ated with positive values of the second canonical variable (Fig. 6B). Although there seems to be considerable overlap between the Risso’s dolphin, pantropical spotted dolphin, and Kogia spp. in the canonical space, the sperm whale is separated from the other species primarily along canoni- cal axis 1 (Fig. 6B). Bottlenose dolphin Because the bottlenose dolphin was never encountered seaward of the 750-m isobath, only the surveyed transect sections shallower than 1000 m were used in the compari- sons between the sightings and the effort. The distribu- tions of this species with respect to depth, depth gradient, surface temperature, and surface temperature variability were significantly different from a uniform distribution (Table 5). Monte Carlo tests suggested that the distribu- tion with respect to depth gradient may have been an artifact of the distribution with respect to depth (P>0.05; Table 5). The sighting rate distribution of the bottlenose dolphin with respect to depth (Fig. 7A) was bimodal as indicated by the peak in the sighting rate at the shal- lowest depth class ( <75 m) and another peak just sea- ward of the shelf break. Although no coherent pattern was apparent in the sighting rate distribution with respect to surface temperature (Fig. 7B), group sighting rates increased with increasing surface temperature variabil- ity (Fig. 70. Interpretation of the sighting rate distributions for sur- face temperature and surface temperature variability was confounded by this species’ bimodal distribution with re- spect to depth. To address this, the sightings were sepa- rated into a shelf group (<150 m) and a shelf break group (>150 m) by using the local minimum in the sighting rates with respect to depth as the separation criterion (Fig. 7A). The shelf dolphins (z? =24 ) were found in cooler surface waters in relation to that observed during the sighting effort ( Mann-Whitney test, 17=2.23, P<0.05), whereas the distribution of the shelf break dolphins {n= 33) with re- spect to surface temperature was not significantly differ- ent from the effort (Mann-Whitney test, [7=1.03, P>0.05; x2 test, ^2=9.7, df=5, P>0.05). Both the shelf 0? = 16, [7=3.23, P<0.01) and shelf break bottlenose dolphins (/i=14, [7=2.93, P<0.01) were encountered in regions of significantly higher surface temperature variability in relation to the effort. It should be noted that the bottlenose dolphin appears to have a distribution with respect to zooplankton biomass that is significantly different from the effort for all depths (Fig. 5G). In fact, a Mann-Whitney test supports this asser- tion ([7=5.42, P<0.0001). Once the analysis is restricted to the continental shelf and upper continental slope (0-1000 m), however, the distribution of the bottlenose dolphin with respect to zooplankton biomass is not significantly different from the effort ( [7=0.19,P>0.05). This apparent discrepancy is due to higher zooplankton biomass over the continental shelf than anywhere else in the northern Gulf of Mexico. Risso's dolphin The distribution of the Risso’s dolphin was significantly different from a uniform distribution for both depth and depth gradient (Table 5) and there was strong evidence Baumgartner et al.: Cetacean habitats in the northern Gulf of Mexico 231 Table 5 Results for univariate %2, Mann-Whitney, and Monte Carlo tests. Values of 0.0000 for P indicate P < 0.0001. X2 test Mann-Whitney Monte Carlo n *2 df P U P P Bottlenose dolphin Depth 89 57.6 11 0.0000 — Depth gradient 89 23.1 11 0.0174 0.7474 Surface temperature 57 26.8 9 0.0015 0.0224 Surface temperature standard deviation 30 17.3 4 0.0017 0.0018 Depth of 15°C isotherm 39 6.0 6 0.4184 0.4909 Surface chlorophyll 57 12.2 9 0.1999 0.4095 Zooplankton biomass 40 1.9 6 0.9268 0.9720 Risso’s dolphin Depth 67 53.9 12 0.0000 — Depth gradient 67 57.4 12 0.0000 0.0000 Surface temperature 39 9.9 6 0.1296 0.1527 Surface temperature standard deviation 14 1.69 0.0456 Depth of 15°C isotherm 52 16.3 9 0.0603 0.0971 Surface chlorophyll 49 9.2 8 0.3232 0.3227 Zooplankton biomass 44 7.3 7 0.3970 0.7056 Kogia spp. Depth 56 42.6 9 0.0000 — Depth gradient 56 20.4 9 0.0155 0.0690 Surface temperature 28 4.8 4 0.3038 0.3540 Surface temperature standard deviation 18 1.98 0.0238 Depth of 15°C isotherm 47 7.2 8 0.5125 0.5371 Surface chlorophyll 46 3.4 7 0.8441 0.8560 Zooplankton biomass 35 31.6 5 0.0000 0.0014 Pantropical spotted dolphin Depth 107 50.6 11 0.0000 — Depth gradient 107 25.1 11 0.0088 0.0687 Surface temperature 62 13.5 11 0.2614 0.3616 Surface temperature standard deviation 31 7.5 4 0.1096 0.1148 Depth of 15°C isotherm 99 19.1 11 0.0593 0.0748 Surface chlorophyll 102 20.6 11 0.0380 0.0985 Zooplankton biomass 57 16.4 9 0.0588 0.3522 Sperm whale Depth 43 14.5 7 0.0431 — Depth gradient 43 13.7 7 0.0566 0.0965 Surface temperature 14 0.27 0.3921 Surface temperature standard deviation 11 1.14 0.1268 Depth of 15°C isotherm 34 11.0 5 0.0508 0.0388 Surface chlorophyll 39 11.5 6 0.0741 0.0749 Zooplankton biomass 26 3.5 4 0.4729 0.7053 indicates P < 0.05. indicates P < 0.01. that the distribution with respect to depth gradient was not an artifact of the depth distribution (P<0.0001). The distribution with respect to surface temperature variabil- ity was significantly different from the effort (P<0.05; Table 5) and surface temperature variability at Risso’s dolphin sightings was generally higher than the effort (Fig. 5D). The sighting rate distribution with respect to depth was modal about the upper continental slope (Fig. 8A), whereas group sighting rates increased with increas- ing depth gradient (Fig. 8B). 232 Fishery Bulletin 99(2) Kogia spp. The distributions of the Kogia spp. with respect to depth, depth gradient, and epipelagic zooplankton biomass were sig- nificantly different from a uni- form distribution (Table 5); however, the distribution with respect to depth gradient may have been an artifact of the depth distribution (P>0.05; Table 5). The distribution with respect to surface temperature variability was significantly dif- ferent from the effort (P<0.05; Table 5), and the median value of surface temperature vari- ability for Kogia spp. was the highest of all the species exam- ined (Fig. 5D). Kogia spp. had a modal distribution about the upper continental slope (Fig. 9A) and group sighting rates increased with increasing zoo- plankton biomass (Fig. 9B). A bivariate x2 analysis (after Baumgartner, 1997) indicated that the distribution of Kogia spp. was significantly different from a uniform distribution with respect to both depth and zooplankton biomass (^2— 29.2, df=4, P<0.0001) and that in waters of high zooplankton bio- mass over the upper continen- tal slope, group sighting rates were 2.5 times the average. Pantropical spotted dolphin The distribution of the pan- tropical spotted dolphin was significantly different from a uniform distribution of depth, depth gradient, and surface chlorophyll (Table 5); however, the distributions with respect to depth gradient and surface chlorophyll may have been an artifact of the depth distri- bution (P>0.05;Table 5). This species was encountered only once near the shelf break (Fig. 10) and if the x2 analysis was limited to sightings and effort deeper than 500 m, then the depth distribution would not be significantly different from a uniform distribution (/2=14.5, df=ll, P>0.05). Sperm whale The distribution of the sperm whale with respect to depth was significantly different from a uniform distribution (Table 5; Fig. 11A). Only one group of sperm whales was encountered near the shelf break and if the x2 analysis was limited to sightings and effort deeper than 500 m, then the depth distribution would not be significantly dif- ferent from a uniform distribution (j2=2.2, df=7, P>0.05). There is evidence to suggest that the distribution with respect to the depth of the 15°C isotherm was significantly different from a uniform distribution when the depth dis- tribution is taken into consideration (Monte Carlo test, P<0.05; Table 5). In waters where the 15°C isotherm was deeper than 200 m, the group sighting rate of sperm Baumgartner et al.: Cetacean habitats in the northern Gulf of Mexico 233 whales was less than a quarter of the average and was one-sixth the sighting rate in waters where the depth of the 15°C isotherm was shallower than 200 m (Fig. 1 IB ) Discussion Previous studies have indicated that cetacean habitat within several hundred kilometers of the coast is most effectively partitioned by depth (Davis et al., 1998; CETAP5; Dohl et al6.; Dohl et al.7; Green et al.8; Davis et al.11). In the northern Gulf of Mexico, each of the five species examined in our study could be distinguished from at least three of the others by its distribution with depth alone. Although the distributions of Risso’s dolphin and the Kogia spp. with respect to depth overlapped on the upper continental slope, their distributions over the upper slope could be distinguished by using depth gradi- ent and zooplankton biomass. The distributions of pan- tropical spotted dolphins and sperm whales with respect to depth were very similar over the continental slope and deep Gulf, but their distributions differed with respect to the depth of the 15°C isotherm. These results suggest that cetaceans partition the north- ern Gulf of Mexico according to each species’ habitat pref- erences which are presumably based on different prey distri- butions. It is important to note that this partitioning does not necessarily imply spatial sep- aration. Given the right condi- tions (e.g. waters 500-750 m deep over a steep section of the continental slope with high zooplankton biomass in the upper 200 m), many of these cetaceans could be encoun- tered in the same area. Some of the environmental variables that are important descrip- tors of cetacean habitat (e.g. zooplankton biomass, depth of the 15°C isotherm, and surface temperature variability) vary over time, and therefore the locations and spatial extent of each species’ habitat may vary over time as well. Two regions of the northern Gulf of Mexico seem to be particularly important habitats for some of the more fre- 11 Davis, R. W., G. S. Fargion, W. E. Evans, L. N. May and T. D. Leming. 1996. Cetacean habitat. In Distribution and abundance of cetaceans in the north-central and western Gulf of Mexico: final report, vol. II: technical report (R. W. Davis and G. S. Fargion, eds.), p. 329-349. U.S. Department of the Interior, Minerals Management Service, OCS Study MMS 96-007. [Available from Public Information Office, MS 5034, Gulf of Mexico Region, Minerals Management Service, 1201 Elmwood Park Blvcl., New Orleans, LA 70123-2394.] quently encountered species during the spring (Fig. 4): the vicinity of the Mississippi River plume and just seaward of the southwestern Florida continental shelf. The Mis- sissippi River injects nutrients into an otherwise oligotro- phic oceanic Gulf in a region where the continental shelf is very narrow and the upper continental slope is quite steep. The rate of primary productivity and the standing stocks of chlorophyll and plankton associated with the nu- trient-rich, fresh-water plume are high in relation to other regions in the oceanic Gulf (El-Sayed, 1972; Dagg et al., 1988; Ortner et al., 1989; Miiller-Karger et al., 1991). Con- sequently, the plume region may provide feeding oppor- 1.0 0 8 0 4 Cl 3 0.2- o 0 0.0 J 1000 2000 3000 4000 Depth (m) B 1 0 -0 6 100 150 200 250 450 500 Depth of 15°C isotherm (m) Figure 11 Sighting rate distributions of sperm whales with respect to (A) depth and (B) the depth of the 15°C isotherm. 234 Fishery Bulletin 99(2) trinities for cetaceans through local trophic interactions. Likewise, the area west of the southwestern Florida shelf break may be another region of high productivity. The physical oceanography of this region is characterized by the formation of a cyclonic meander or eddy in the spring between the Loop Current to the west and the steep Flor- ida Escarpment to the east (Cochrane, 1972; Vukovich et ah, 1979; Vukovich and Maul, 1985). Maul et al. (1984) ob- served that bluefin tuna catch per unit of effort inside a cold-core meander in this region was three times higher than in the central Gulf the previous year. Between 83— 86°W and 24- 27°N in oceanic waters, the sighting rates of Risso’s dolphins, pantropical spotted dolphins, and sperm whales were 3.8, 2.6, and 2.8 times higher than the aver- age sighting rate and 4.9, 3.0, and 3.3 times higher than the sighting rate outside of this region, respectively. Bottlenose dolphin The bottlenose dolphin’s distribution in the northern Gulf of Mexico is markedly different from the other species examined in our study. This species and the Atlantic spot- ted dolphin are the only cetaceans that are routinely encountered on the continental shelf (Fritts et ah, 1983; Mullin et ah, 1994; Jefferson and Schiro, 1997; Hansen et al.3). Caution is warranted when interpreting the bimodal distribution of bottlenose dolphin sighting rates with respect to depth (Fig. 7A). Effort on the continental shelf was neither extensive nor distributed uniformly through- out the northern Gulf. During the CETAP study (Kenney, 1990; CETAP5), a distinct bimodal distribution of bottle- nose dolphins was observed north of Cape Hatteras. Bot- tlenose dolphins were concentrated during warm months in waters less than 25 m and year round near the 1000-m isobath and some groups were sighted in waters as deep as 4712 m (CETAP’). This bimodal distribution is sugges- tive of the inshore (coastal) and offshore forms of bottle- nose dolphins described by others (Norris and Prescott, 1961; Walker, 1975; Leatherwood and Reeves, 1982; Shane et al., 1986; Kenney, 1990; Walker1'-) and supported by mitochondrial DNA (Dowling and Brown, 1993; Curry and Smith, 1997), hematological (Duffield et al., 1983; Hersh and Duffield, 1990), and morphological (Hersh and Duff- ield, 1990) evidence. The spatial distribution of bottlenose dolphin group sightings from aerial surveys on the conti- nental shelf of the Gulf of Mexico (Blaylock et al., 1995) and off the southeast U.S. coast south of Cape Hatteras (Blaylock and Huggard, 1994), however, was not character- ized by any large-scale discontinuities in bottlenose dol- phin distribution similar to those observed north of Cape Hatteras. The shelf bottlenose dolphins were found in regions with cooler than expected surface waters and high surface tem- 12 Walker, W. A. 1981. Geographical variation in morphology and biology of bottlenose dolphins iTursiops) in the eastern North Pacific. Southwest Fisheries Center Administrative Report LJ-81-03C, 52 p. [Available from Southwest Fisheries Science Center, National Marine Fisheries Service, NOAA, P.O. Box 271, La Jolla, CA 92038.1 perature variability. These oceanographic characteristics are consistent with the cool and fresh water side of fronts associated with river plumes and, indeed, sighting rates of the shelf bottlenose dolphins were particularly high near the Mississippi River Delta. Sighting rates of the shelf break bottlenose dolphins were more evenly distributed in the central and eastern Gulf and the high surface temper- ature variability observed near these dolphins suggests a potential association with shelf break fronts. Risso's dolphin Baumgartner (1997) examined the same 1992-94 spring cruise data used in our study with the intent of defining Risso’s dolphin habitat in terms of the physiography of the northern Gulf of Mexico. Using both univariate and bivariate analyses, he determined that the sighting rate of Risso’s dolphin groups between the 350- and 975-m iso- baths and in depth gradients exceeding 24 m per 1.1 km was nearly 5 times the average. Of the groups encountered outside this region, 40% were sighted within 5 km of it. Aerial survey data collected during all seasons between 1992 and 1994 were used to independently assess this habitat model. Sighting rates from these surveys were nearly 6 times the average inside this core habitat, and of the groups encountered outside of this region, 73%< were sighted within 5 km of it. The distribution of Risso’s dolphin along the continen- tal slope has been noted in several studies (Wiirtz et al., 1992; CETAP5; Dohl et al.6; Dohl et al.7; Green et al.8; Da- vis et al.11) and some evidence exists to support this spe- cies’ association with the steeper sections of the upper con- tinental slope elsewhere. Off the Oregon and Washington coasts, Green et al.8 observed that Risso’s dolphin encoun- ter rates over the continental slope (200-2000 m) were seven times greater than on the shelf and that the groups sighted on the shelf were very close to the shelf break. Compared with the northern Gulf of Mexico, almost the entire Oregon-Washington continental slope can be con- sidered steep with depth gradients in excess of 22 m per 1.1 km (Fig. 11 in Green et al.8). Dohl et al.7 found a simi- lar distribution off central and northern California, where the majority of Risso’s dolphin sightings were between the 183- and 1830-m (100-1000 fathom) isobaths. As is the case off Oregon and Washington, virtually all of the con- tinental slope off central and northern California can be considered very steep (Fig. 1 in Dohl et al.7). The physiog- raphy of the northwestern Atlantic Ocean is much more like that found in the northern Gulf of Mexico and the CETAP study (Hain et al.. 1985; Kenney and Winn, 1986; CETAP5 ) found Risso’s dolphins concentrated at the shelf break (mode of 478 sightings was 183 m depth) and dis- tributed over the entire continental slope (average of 478 sightings was 1092 m). Baumgartner (1997) hypothesized that Risso’s dolphins aggregate along the upper continental slope because of the presence of a persistent ocean front separating the rela- tively cool and fresh waters of the continental shelf and the more warm and salty waters of the oceanic Gulf. This shelf break front may provide greater feeding opportuni- Baumgartner et al.: Cetacean habitats in the northern Gulf of Mexico 235 ties because of enhanced local productivity or because it forms a bolder between two separate, exploitable ecosys- tems. The observations of higher than expected surface temperature variability at Risso’s dolphin sightings seem to support this hypothesis, but the evidence is rather tenu- ous given the small sample size (n- 14). Kogia spp. Kogia spp. were predominantly encountered along the upper continental slope in regions with high epipelagic zooplankton biomass. Their distribution with respect to depth in the northern Gulf of Mexico is in agreement with inferences drawn from a stomach content study of stranded Kogia spp. in South Africa where cephalopods typical of the continental slope were identified as the largest constituent of the stranded whales’ stomach con- tents.13 Further stomach content and stable isotope analy- ses suggest that pygmy and dwarf sperm whales consume different prey species and therefore may occupy different habitats.14'15 No such separation was detectable in our study because of the low sample size for the individual species and the difficulty of positively identifying each at sea. Some diet overlap was observed between the two species off South Africa14 and therefore the association between Kogia spp. and high epipelagic zooplankton bio- mass in the northern Gulf of Mexico may be due to the uti- lization of zooplankton in the diet of one or more of their common prey species. Of all the cetaceans, Kogia spp. had the highest median value for surface temperature variability which suggests a similar association with ocean fronts as that observed for Risso’s dolphins. Sample size for this variable was unfortu- nately low (n-18) however; therefore it is difficult to accu- rately describe this potential association. Because the up- per continental slope can be a region of persistent frontal activity, it is conceivable that the distribution of both Ris- so’s dolphins and Kogia spp. with respect to surface tem- perature variability may have been a consequence of their distribution with depth. The low sample sizes for each of these species precludes any analysis that may have been able to further support or refute these hypotheses. Pantropicai spotted dolphin The distribution of the pantropicai spotted dolphin was not significantly different from a uniform distribution with respect to any of the environmental variables, except depth. Pantropicai spotted dolphins are rarely encoun- tered on the continental shelf in the northern Gulf of Mexico (Jefferson and Schiro, 1997) and from the results 1!Klages, N. 2000. Persona! commun. Port Elizabeth Museum at Bayworld, P.O. Box 13147, Humewood 6013. South Africa. 14 Plon, S. 2000. Personal commun. School of Biological Sci- ences, Thomas Building, Level 2, Univ. Auckland, Private Bag 92019, Auckland, New Zealand. 1:1 Barros, N. 2000. Personal commun. Center for Marine Mam- mal and Sea Turtle Research, Mote Marine Laboratory, 1600 Ken Thompson Parkway, Sarasota, FL 34236-1096. obtained in our study, are probably evenly distributed with depth over the continental slope and deep Gulf. These results were surprising in light of this species’ spatial dis- tribution in the northern Gulf of Mexico (Fig. 4). Local maxima in group encounter rates occurred southwest of Panama City and along the Florida Escarpment northwest of Key West. The coherent pattern in Figure 4 strongly suggests the existence of high-use areas for this species, but the characteristics that make these regions attractive to pantropicai spotted dolphins were not observed in the chosen set of environmental variables used in our study. Davis et al.16 reported that oceanic stenellids (pan- tropical spotted dolphins, striped dolphins [ Stenella coe- ruleoalba ], spinner dolphins [ Stenella longirostris] , and Clymene dolphins [ Stenella clymene ]) were more fre- quently encountered in cyclonic, cold-core eddies and less frequently encountered in anticyclonic, warm-core eddies than expected based on the distribution of the GulfCet program sighting effort in the northern Gulf of Mexico. Although no such relationship was detected in our study, species grouping, confounding by other environmental or behavioral factors or temporal variability in habitat associ- ations (or both) or prey availability could easily account for the apparent discrepancy between these two studies. Sperm whale Like the pantropicai spotted dolphin, the sperm whale was never encountered on the continental shelf and appears to have a roughly even distribution with respect to depth over the continental slope and deep Gulf. The distribu- tions of these two species with respect to the depth of the 15°C isotherm were significantly different, however, and the canonical LDF analysis suggested that this vari- able contributed to the separation between sperm whales and the other oceanic cetaceans (Fig. 6B). Sperm whales were encountered much less frequently in regions where the depth of the 15°C isotherm was quite deep (Fig. 11B), which suggests that this species avoids the interior of anti- cyclonic, warm-core features such as the Loop Current or warm-core mesoseale eddies. Waring et al. ( 1993) and Grif- fin (1999) described similar results from studies of sperm whale distributions in and around the periphery of warm- core eddies associated with the Gulf Stream in the north- west Atlantic Ocean. Davis et al.16 and Biggs et al. (2000) reported that sperm whales were not only encountered outside of anticylonic features in the northeastern Gulf of Mexico, but most of the visual and acoustic contacts with 16 Davis, R. W„ J. G. Ortega-Ortiz, C. A. Ribic, W. E. Evans, I). C. Biggs, P. H Ressler, J. H. Wormuth. R. R. Leben, K. D. Mullin, and B. Wursig. 2000. Cetacean habitat in the northern Gulf of Mexico. In Cetaceans, sea turtles and seabirds in the north- ern Gulf of Mexico: distribution, abundance and habitat asso- ciations, vol. II: technical report (R. W. Davis, W. E. Evans, and B. Wursig, eds.), p. 217-253. U.S. Department of the Interior, Geological Survey, Biological Resources Division, LTSGS/BRD/ CR- 1999-0006 and Minerals Management Service, OCS Study MMS 2000-003. [Available from Public Information Office, MS 5034, Gulf of Mexico Region, Minerals Management Ser- vice, 1201 Elmwood Park Blvd., New Orleans, LA 70123-2394.] 236 Fishery Bulletin 99(2) sperm whales during the GulfCet II focal cruises were in regions characterized by cyclonic mesoscale features. Jaquet ( 1996) reviewed a variety of sperm whale habitat studies that seemed to have contradictory conclusions re- garding the primary oceanographic processes influencing sperm whale distribution (namely upwelling and down- welling). Jaquet attributed these discrepancies to a prob- lem of defining the appropriate spatial and temporal scales, and she and others illustrated this point by demonstrating a varying but positive correlation between historical sperm whale catches and surface chlorophyll over increasing tem- poral and spatial scales in the equatorial Pacific (Jaquet et al., 1996). These results seem to indicate that upwelling, which contributes to increased surface phytoplankton bio- mass, is a predominant factor in influencing sperm whale distribution in the equatorial Pacific. Historical catches in temperate waters, however, are not at all correlated with surface chlorophyll (see Fig. 1 of Jaquet, 1996 and Fig. 1 of Jaquet et ah, 1996) which suggests that other oceano- graphic processes or physiographic influences may be im- portant (e.g. downwelling or biological-physical processes associated with continental slopes). At comparatively short time scales and small spatial scales, we found no evidence to suggest a relationship between the distribution of sperm whales and surface chlorophyll in the northern Gulf of Mexico. Even at longer temporal and larger spatial scales, we would expect this same result because the oceanic Gulf is persistently oligotrophic both in time and space (Miillei- Karger et ah, 1991; Longhurst, 1998). Berzin ( 1971) examined harvest records from the world- wide sperm whale fishery and suggested that sperm whale distribution was closely linked to processes that support- ed the meso- and bathypelagic food webs. Because sperm whales feed almost exclusively on mesopelagic or demer- sal cephalopods (Clarke, 1986, 1996), they probably aggre- gate in areas where these prey are abundant. These deep- water prey species are entirely dependent on the rain of organic matter from the surface for their sustenance and so these species will be found in regions where the export of detritus from the surface is enhanced. This process oc- curs in convergence zones where downwelling forces sur- face biomass and oxygen into the deep ocean, such as in the middle of anticyclonic eddies, at the peripheries of cy- clonic eddies, to the right (left) of surface ocean currents in the northern (southern) hemisphere, in the middle of the large-scale anticyclonic ocean gyres (e.g. the Sargasso Sea), or along fronts where surface water masses abut. The global sperm whale distribution maps provided by Townsend ( 1935 ) and Berzin ( 197 1 ) do indeed suggest that this species was frequently harvested in or near large- scale oceanic convergence zones, especially along the sub- tropical convergence zones and the Antarctic polar front. The distribution of sperm whales in the northern Gulf of Mexico and northwestern Atlantic Ocean (Waring et al., 1993; Griffin, 1999) seems contradictory to Berzin’s hy- pothesis, however. Features such as the Loop Current or warm-core eddies rotate anticyclonically and have conver- gent centers in which downwelling occurs. According to Berzin’s hypothesis, the interior of these features would be favorable to sperm whales because of the enhanced export of surface biomass to the deep ocean and the resultant in- crease in prey species. The interior of anticyclonic eddies in the northern Gulf of Mexico are, however, low in surface zooplankton biomass (Biggs, 1992). Although the rate of detrital export to the deep is enhanced by increased verti- cal velocities within these features, the amount of biomass actually exported may be too small to support large popu- lations of deep-water prey. Another possible explanation for the distribution of sperm whales with respect to the depth of the 15°C iso- therm is related to the availability of prey. Berzin (1971) characterized cephalopods as thermophilic and thus in- dicated that they are distributed within a narrow range of ocean temperatures according to their species-specific thermal requirements or to the thermal requirements of their prey. These requirements not only govern the hori- zontal distribution of cephalopods, but their vertical dis- tribution as well. Because warm-core features are charac- terized by depressed isotherms (e.g. Fig. 3), cephalopods within these features may be hundreds of meters deeper in the water column than in the waters outside these fea- tures. Despite their well-known ability to dive to great depths, foraging continuously at greater depths under warm-core features would be much more energetically ex- pensive than foraging outside these features. Thus, when prey abundance inside and outside of warm-core eddies are equivalent, sperm whales may feed on prey distribut- ed at shallower depths outside of these features to reduce their energy expenditure. Caveats It is important to remember that this study was limited to surveys conducted during the spring season. The spa- tial distribution of cetaceans may be different in other sea- sons because the oceanographic conditions of the northern Gulf of Mexico change over the course of the year. The northward penetration of the Loop Current into the Gulf varies on a quasi-annual basis (Vukovich et al., 1979; Stur- ges and Evans, 1983; Vukovich, 1995) and the variability in the position of the Loop Current affects the generation and positions of both anticyclonic and cyclonic eddies. This variability may, in turn, greatly influence the productiv- ity and availability of prey species in the eastern Gulf of Mexico. In the northwestern Gulf, the slow march of warm- core eddies from east to west toward the “eddy graveyard” over the continental slope also varies with time and may affect the seasonal distribution of cetaceans. Hansen et al.3 observed seasonal differences in cetacean abundance in the western and central regions of the northern Gulf of Mexico that may have been influenced by temporal changes in the local oceanography. Another potential limitation of our study was the rather coarse environmental sampling. Although the CTD/XBT sampling strategy was sufficient to identify the large-scale oceanographic features, some of the most biologically sig- nificant processes in the oceanic Gulf of Mexico occur on smaller spatial scales. In particular, the outer edge of the Loop Current is frequently a site of upwelling and these divergent features often develop into cyclonic meanders Baumgartner et al.: Cetacean habitats in the northern Gulf of Mexico 237 and eddies at the northern and eastern sides of the Loop Current (Vukovich and Maul, 1985). These cyclonic fea- tures are usually much smaller than the Loop Current itself or the warm-core features of the central and west- ern Gulf (Cochrane, 1972). Along the latitude of 27°N and within 40 km of the Loop Current (near 88.8°W) in late spring, 1993, for instance, the satellite-borne AVHRR de- tected surface temperatures 1.5°C cooler than the Loop Current itself, but the vertical temperature structure, sur- face chlorophyll and zooplankton biomass associated with this narrow feature were not captured because of the coarse sampling strategy (Figs. 2 and 3). The recent Gulf- Cet II program (Davis et al.16) examined cetacean habitat associations in the vicinity of cyclonic-anticyclonic eddy pairs in the northeastern Gulf of Mexico and demonstrat- ed that these mesoscale hydrographic features can indeed influence cetacean distribution. In addition to the cyclonic features associated with the Loop Current, the CTD and XBT sampling strategy of our study did not resolve other potentially productive features that occur on smaller spa- tial scales, such as nutrient-rich Mississippi River plume waters entrained at the edge of the Loop Current (Maul, 1977; Miiller-Karger et al., 1991) and shelf break fronts. Future studies of cetacean habitats in the Gulf of Mexi- co should continue to consider these smaller scale features as potential sites of large cetacean aggregations because of their high levels of biological activity. These features could potentially 1 ) have high rates of primary productiv- ity that is converted into prey biomass over short tempo- ral and spatial scales, 2) concentrate prey through solely physical mechanisms or through physical-biological inter- actions or 3) make local prey more accessible to surface- bound cetaceans. Although investigating these processes is undoubtedly a challenge, it is important to elucidate what processes affect cetacean distribution and at what spatial and temporal scales these processes operate if we are to understand how oceanographic conditions affect ce- tacean ecology. Acknowledgments This work was supported by the National Marine Fisher- ies Service’s Marine Mammal Program. Additional fund- ing was provided by the Minerals Management Service under contract 14-35-0001-30619 and Interagency Agree- ment 16197 (GulfCet Program). Portions of the environ- mental data and all of the zooplankton data were collected under the Southeast Area Monitoring and Assessment Program (SEAMAP) and were generously provided by J Shultz. We would like to acknowledge the hard work and dedication of the marine mammal observers, the officers and crew of the NOAA Ship Oregon II, and the staff of the Southeast Fisheries Science Center. Special thanks go to C. Roden for managing the marine mammal surveys and M. McDuff for technical guidance and support. S. Ander- son and R. Weller of the Woods Hole Oceanographic Insti- tution graciously provided computing resources. We also thank three anonymous reviewers for criticizing earlier drafts of this manuscript. Literature cited Barlow, J. 1995. The abundance of cetaceans in California waters. Part I: ship surveys in summer and fall of 1991. Fish. Bull. 93:1-14. Baumgartner, M. F. 1997. The distribution of Risso’s dolphins (G)-ampus gri- seus) with respect to the physiography of the northern Gulf of Mexico. Mar. Mamm. Sci. 13:614-638. Berzin, A. A. 197 1 The sperm whale. Pischevaya Promyshlennost, Mos- cow, 394 p. [Translated from Russian by E. Hoz and Z. Blake, Israel Program for Scientific Translations, Ltd.] Biggs, D. C. 1992. Nutrients, plankton, and productivity in a warm-core ring in the western Gulf of Mexico. J. Geophys. Res. 97:2143-2154. Biggs, D. C., R. R. Leben, and J. G. Ortega-Ortiz. 2000. Ship and satellite studies of mesoscale circulation and sperm whale habitats in the northeast Gulf of Mexico during GulfCet II. Gulf of Mexico Science 18:15-22. Blaylock, R. A., J. W. Hain, L. J. Hansen, D. L. Palka, and G. T. Waring. 1995. U.S. Atlantic and Gulf of Mexico marine mammal stock assessments. U.S. Dep. Commer., NOAA Tech. Memo. NMFS-SEFSC-363, 211 p. Blaylock, R. A., and W. Hoggard. 1994. Preliminary estimates of bottlenose dolphin abun- dance in southern U.S. Atlantic and Gulf of Mexico con- tinental shelf waters. U.S. Dep. Commer., NOAA Tech. Memo. NMFS-SEFSC-356, 10 p. Clarke, M. R. 1986. Cephalopods in the diet of odontocetes. In Research on dolphins (M. M. Bryden and R. Harrison, eds.), p. 281-321. Oxford Univ. Press, Oxford. 1996. Cephalopods as prey. III: Cetaceans. Philos. Trans. R. Soc. Lond. (ser B.) Biol. Sci. 351:1053-1065. Cochrane, J. D 1972. Separation of an anticyclone and subsequent develop- ments in the Loop Current (1969). In Contributions on the physical oceanography of the Gulf of Mexico (L. R. A. Capurro and J. L. Reid, eds.), p. 91-106. Gulf Publishing, Houston, TX. Conover, W. J. 1980. Practical nonparametric statistics. John Wiley and Sons, Inc, New York, NY, 493 p. Curry, B. E., and J. Smith. 1997. Phylogeographic structure of the bottlenose dolphin ( Tursiops truncatus): stock identification and implications for management. In Molecular genetics of marine mam- mals (A. E. Dizon, S. J. Cbivers and W. F. Perrin, eds.), p. 227-247. Society for Marine Mammalogy. Dagg, M. J., P B. Ortner, and F. Al-Yamani. 1988. Winter-time distribution and abundance of copepod nauplii in the northern Gulf of Mexico. Fish. Bull. 86:219- 230. Davis, R. W., G. S. Fargion, L. N. May, T. D. Leming, M. F. Baumgartner, W. E. Evans, L. J. Hansen, and K. Mullin. 1998. Physical habitat of cetaceans along the continental slope in the north-central and western Gulf of Mexico. Mar. Mamm. Sci. 14:490-507. Dowling, T. E., and W. M. Brown. 1993. Population structure of the bottlenose dolphin ( Tur- siops truncatus) as determined by restriction endonuclease 238 Fishery Bulletin 99(2) analysis of mitochondrial DNA. Mar. Mamm. Sci. 9:138- 155. Duffield, D. A., S. H. Ridgway, and L. H. Cornell. 1983. Hematology distinguishes coastal and offshore forms of dolphins ( Tursiops ). Can. J. Zool. 61:930-933. Elliot, B. A. 1982. Anticyclonic rings in the Gulf of Mexico. J. Physical Oceanography 12:1292-1309. El-Sayed, S. Z. 1972. Primary productivity and standing crop of phyto- plankton in the Gulf of Mexico. In Chemistry, primary productivity and benthic algae of the Gulf of Mexico, Ser. Atlas Mar. Environ., folio 22 (V. C. Busnell, ed.), p. 8-13. Am. Geogr. Soc. Evans, W. E. 1975. Distribution, differentiation of populations and other aspects of the natural history of Delphinus delphis Lin- naeus in the northeastern Pacific. Ph.D. diss., LTniv. Cali- fornia, Los Angeles, CA, 145 p. Fitch, J. E., and R. L. Brownell -Jr. 1968. Fish otoliths in cetacean stomachs and their impor- tance in interpreting feeding habits. -J. Fish. Res. Board Can. 25:2561-2574. Fritts, T. H., A. B. Irvine, R. D. Jennings, L. A. Collum, W. Hoffman, and M. A. McGehee. 1983. Turtles, birds and mammals in the northern Gulf of Mexico and nearby Atlantic waters. U.S. Fish and Wildl. Serv., Div. Biol. Serv., report FWS/OBS-82/65, 455 p. U.S. Fish. Wildl. Serv., Washington, D.C. Golden Software. 1994. SLIRFER for Windows users’s guide. Golden Soft- ware, Boulder, CO, 420 p. Griffin. R. B. 1999. Sperm whale distributions and community ecology associated with a warm-core ring off Georges Bank. Mar. Mamm. Sci. 15:33-51. Hain, J. H. W., M. A. M. Hyman, R. D. Kenney, and H. E. Winn. 1985. The role of cetaceans in the shelf-edge region of the northeastern United States. Mar. Fish. Rev. 47:13-17. Hersh, S. L., and D. A. Duffield. 1990. Distinction between northwest Atlantic offshore and coastal bottlenose dolphins based on hemoglobin profile and morphometry. In The bottlenose dolphin (S. Leather- wood and R. R. Reeves, eds.), p. 129-139. Academic Press, San Diego, CA. Huberty, C. J. 1994. Applied discriminant analysis. John Wiley and Sons, Inc, New York, NY, 496 p. Hui, C. A. 1979. Undersea topography and distribution of dolphins of the genus Delphinus in the southern California Bight. J. Mammal. 60:521-527. 1985. Undersea topography and the comparative distribu- tions of two pelagic cetaceans. Fish. Bull. 83:472-475. Hurlburt, H. E., and J. D. Thompson. 1982. The dynamics of the Loop Current and shed eddies in a numerical model of the Gulf of Mexico. In Hydro- dynamics of semi-enclosed seas (J. C. -J. Nihoul, ed.), p. 243-297. Elsevier Scientific Publishing Co., Amster- dam, The Netherlands. Jaquet, N. 1996. How spatial and temporal scales influence under- standing of sperm whale distribution: a review. Mammal Rev. 26:51-65. Jaquet, N., H. Whitehead, and M. Lewis. 1996. Coherence between 19th century sperm whale dis- tributions and satellite-derived pigments in the tropical Pacific. Mar. Ecol. Prog. Ser. 145:1-10. Jefferson, T. A., and A. J. Schiro. 1997. Distribution of cetaceans in the offshore Gulf of Mexico. Mammal Rev. 27:27-50. Jeffery, S. W., and G. F. Humphrey. 1975. New spectrophotometric equations for determining chlorophylls a, b, c and c2 in higher plants, algae and nat- ural phytoplankton. Biochem. Physiol. Pflanz (Bpp) 167: 191-194. Jockens, A., G. S. Fargion, and R. Leben. 1994. Observation of the fate of Eddy V: April-June 1993. EOS 75:212. Johnson, D. E. 1998. Applied multivariate methods for data analysts. Dux- bury Press, Pacific Grove, CA, 567 p. Kendall, M. G., and A. Stuart. 1967. The advanced theory of statistics, vol. 2: inference and relationship, 2nd ed. Griffon and Co. Limited, London, 690 p. Kenney, R.D. 1990. Bottlenose dolphins off the northeastern LTnited States. In The bottlenose dolphin (S. Leatherwood and R. R. Reeves, eds.), p. 369-386. Academic Press, San Diego, CA. Kenney, R. D., M. A. M. Hyman and H. E. Winn. 1985. Calculation of standing stocks and energetic require- ments of cetaceans of the northeast United States outer continental shelf. U.S. Dep. Commer., NOAA Tech. Memo. NMFS F/NEC-41, 99 p. Kenney, R D., and H. E. Winn. 1986. Cetacean high-use habitats of the northeast United States continental shelf. Fish. Bull. 84:345-357. Leatherwood, S., and R. R. Reeves. 1982. Bottlenose dolphin ( Tursiops tnincatus) and other toothed cetaceans. In Wild mammals of North America ( J. A. Chapman and G. A. Feldhamer, eds.), p. 369-414. Johns Hopkins LIniv. Press, Baltimore, MD. Lockyer, C. 1981. Growth and energy budgets of large baleen whales from the Southern Hemisphere. In FAO Advisory Com- mittee on Marine Resource Research, Mammals in the Sea, vol. Ill, General papers and large cetaceans, p. 379-487. FAO, Rome. Longhurst, A. 1998. Ecological geography of the sea. Academic Press, San Diego, CA, 398 p. Maul, G. A. 1977. The annual cycle of the Gulf Loop Current, part 1: observations during a one-year time series. J. Mar. Res. 35:29-47. Maul, G. A., F. Williams, M. Roffer, and F. M. Sousa. 1984. Remotely sensed oceanographic patterns and variabil- ity of bluefin tuna catch in the Gulf of Mexico. Oceanol. Acta 7:469-479. Miiller-Karger, F. E., J. J. Walsh, R. H. Evans, and M. B. Meyers. 1991. On the seasonal phytoplankton concentration and sea surface temperature cycles of the Gulf of Mexico as deter- mined by satellites. J. Geophys. Res. 96:12,645-12,665. Mullin, K. D., W. Hoggard, C. L. Roden, R. R. Lohoefener, C. M. Rogers, and B. Taggart. 1994. Cetaceans on the upper continental slope in the north- central Gulf of Mexico. Fish. Bull. 92:773-786. Mullin, K. D., and L. J. Hansen. 1999. Marine mammals of the northern Gulf of Mexico. In The Gulf of Mexico large marine ecosystem: assessment, sus- Baumgartner et al Cetacean habitats in the northern Gulf of Mexico 239 tainability and management (H. Kumpf, K. Steidinger, and K. Sherman, eds.), p. 269-277. Blackwell Science, London. Norris, K. S., and J. H. Prescott. 1961. Observations on Pacific cetaceans of Californian and Mexican waters. Univ. Calif. Publ. Zool. 63:291-402. Ortner, P. B., L. C. Hill, and S. R. Cummings. 1989. Zooplankton community structure and copepod spe- cies composition in the northern Gulf of Mexico. Conti- nental Shelf Research 9:387-402. Perrin, W. F., R. R. Warner, C. H. Fiscus, and D. B. Holts. 1973. Stomach contents of porpoise, Stenella spp., and yel- lowfin tuna, Thunnus albacai'es, in mixed-species aggrega- tions. Fish. Bull. 71:1077-1092. Reilly, S. B. 1990. Seasonal changes in distribution and habitat differ- ences among dolphins in the eastern tropical Pacific. Mar. Ecol. Prog. Ser. 66:1-11. Richards, W. J., T. Leming, M. F. McGowan, J. T. Lamkin, and S. Kelley-Fraga. 1989. Distribution of fish larvae in relation to hydrographic features of the Loop Current boundary in the Gulf of Mexico. Rapp. P-V. Reun. Cons. Int. Explor. Mer 191:169-176. SAS Institute. 1989. SAS/STAT user’s guide, version 6, 4th ed., vol. 1. SAS Institute, Inc., Cary, NC, 943 p. Selzer, L. A., and P. M. Payne. 1988. The distribution of white-sided ( Lagenorhynchus acutus) and common dolphins ( Delphinus delphis ) vs. envi- ronmental features of the continental shelf of the north- eastern United States. Mar. Mamm. Sci. 4:141-153. Shane, S. H., R. S. Wells, and B. Wiirsig. 1986. Ecology, behavior and social organization of the bot- tlenose dolphin: a review. Mar. Mamm. Sci. 2:34-63. Smith, G. J. D., and D. E. Gaskin. 1974. The diet of harbor porpoises (Phocoena phocoena [L. ] ) in coastal waters of eastern Canada with special reference to the Bay of Fundy. Can. J. Zool. 52:777-782. Smith, P. E. and S. L. Richardson. 1977. Standard techniques for pelagic fish egg and larva surveys. FAO Fish. Tech. Pap. 175, 100 p. Smith, R. C., P. Dustan, D. Au, K. S. Baker, and E. A. Dunlap. 1986. Distribution of cetaceans and sea-surface chlorophyll con- centrations in the California Current. Mar. Biol. 91:385- 402. Sokal R. R., and F. J. Rohlf. 1981. Biometry: the principles and practice of statistics in biological research, 2nd ed. W.H. Freeman, San Francisco, CA, 859 p. Strickland, J. D. H. and T. R. Parsons. 1972. A practical handbook of seawater analysis. Bull. Fish. Res. Board Can. 167: 1-31 1 Sturges, W., and J. C. Evans. 1983. On the variability of the Loop Current in the Gulf of Mexico. J. Mar. Res. 41:639— 653. Townsend, C. H. 1935. The distribution of certain whales as shown by logbook records of the American whaleships. Zoologica 19:1-50. Vukovich, F. M. 1995. An updated evaluation of the Loop Current’s eddy- shedding frequency. J. Geophys. Res. 100:8655-8659. Vukovich, F. M., B. W. Crissman. M. Busnell, and W. J. King. 1979. Some aspects of the oceanography of the Gulf of Mexico using satellite and in situ data. J. Geophys. Res. 84: 7749-7768. Vukovich, F. M., and G. A. Maul. 1985. Cyclonic eddies in the eastern Gulf of Mexico. J. Physical Oceanography 15:105-117. Walker, W. 1975. Review of the live capture fishery for smaller ceta- ceans taken in southern California waters for public dis- play, 1966-73. J. Fish. Res. Board Can. 32:1197-1211. Waring, G. T., C. P. Fairfield, C. M. Ruhsam, and M. Sano. 1993. Sperm whales associated with Gulf Stream features off the north-eastern USA shelf. Fish. Oceanogr. 2:101-105. Wiirtz, M., R. Poggi, and M. R. Clarke. 1992. Cephalopods from the stomachs of a Risso’s dolphin (Grampus griseus) from the Mediterranean. J. Mar. Biol. Assoc. U.K. 72:861-867. Somatic growth function for immature Boggerhead sea turtles, Caretta caretta, in southeastern U.S. waters Karen A. Bjorndal Alan B. Bolten Archie Carr Center for Sea Turtle Research and Department of Zoology University of Florida P.O.Box 118525 Gainesville, Florida 32611 E-mail address (for K. A. Bjorndal); kab@zoo.ufl.edu Biological Resources Division U.S. Geological Survey— Padre Island National Seashore PO. Box 181300 Corpus Christi, Texas 78480 Abstract-The Sea Turtle Stranding and Salvage Network, coordinated by the National Marine Fisheries Service through a network of state coordinators, archives data on sea turtles that strand along the U.S. coast. We conducted length-frequency analyses, using MUL- TIFAN software, to generate somatic growth functions for loggerhead sea turtles, Caretta caretta , that stranded along the Atlantic coast of Florida (n = 1234) and along the U.S. coast of the Gulf of Mexico (n= 570) between 1988 and 1995. In both regions, the size range of loggerhead sea turtles between the size at which they begin to recruit in substantial numbers from pelagic to neritic habitats (46 cm curved cara- pace length [CCL] ) and minimum size at sexual maturity (87 cm CCL) was composed of 20 year classes and had similar von Bertalanffy growth func- tions. Our estimates of 20 year classes fall within the range of estimates cal- culated from previous studies (9 to 29 years) for this life stage. Because survi- vorship in this size range has been iden- tified as critical for population recovery, an accurate estimate of life stage is essential for developing effective man- agement plans. Manuscript accepted 3 November 2000. Fish. Bull. 99:240-246 (2001). Bruce Koike Aquarium of the Americas 1 Canal Street New Orleans, Louisiana 70130 Barbara A. Schroeder Office of Protected Resources, National Marine Fisheries Service, NOAA 1315 East West Highway Silver Spring, Maryland 20910 Donna J. Shaver Wendy G. Teas Wayne N. Witzell Southeast Fisheries Science Center National Marine Fisheries Service, NOAA 75 Virginia Beach Drive Miami, Florida 33149 Somatic growth functions are critical parameters for understanding the life history of a species and for developing management plans for wild popu- lations. Substantial effort has been invested in — and considerable infor- mation on growth rates has been gained from — mark-recapture studies of sea turtle populations (Chaloupka and Limpus, 1997; Chaloupka and Musick, 1997; Limpus and Chaloupka, 1997; Bjorndal et ah, 2000). Although growth studies based on mark and recapture of individual animals yield direct measures of growth rates, sea turtles have characteristics that make them relatively poor candidates for mark-recapture studies. Sea turtles are relatively slow growing; consequently mark-recapture studies are long-term labor-intensive efforts. The probability of recapturing marked individuals is low in many populations because of the long-range movements and obscure migratory patterns in some lifestages (e.g. posthatchlings in a pelagic habi- tat), high natural mortality in young lifestages, and high human-induced mortality in juvenile and adult life- stages. Bjorndal et al.: Somatic growth function for immature Caretta carettci 241 Length-frequency analysis has been used for many years to estimate growth rates, age structure, and mortality in marine fish and invertebrate populations (Ricker, 1975; Pauly and Morgan, 1987; Hilborn and Walters, 1992). More recently, length-frequency analyses have been used to eval- uate somatic growth rates in sea turtles. The accuracy of four length-frequency analysis programs (ELEFAN I, [Holden and Bravington, 1992 1 Shepherd’s length compo- sition analysis [SLCA, Holden and Bravington, 1992], pro- jection matrix method [Holden and Bravington, 1992], and MULTIFAN) for predicting growth rates was tested in a population of green turtles, Chelonia mydas , in the south- ern Bahamas for which growth rates had been measured in a long-term mark and recapture study (Bjorndal and Bol- ten, 1995; Bjorndal et al., 1995). MULTIFAN successfully estimated growth rates in this population, SLCA was par- tially successful, and ELEFAN I and the projection matrix method were not successful. In young, pelagic-stage logger- head sea turtles (Caretta caretta), estimates of growth rates generated by MULTIFAN were consistent with results from recaptures of tagged turtles (Bjorndal et al., 2000). In our study, we generated a growth model for imma- ture loggerhead sea turtles in southeastern U.S. waters between the size at which they begin to recruit in substan- tial numbers to neritic habitats (46 cm curved carapace length [CCL]) and minimum size at sexual maturity (87 cm CCL). The duration of the growth interval between 46 and 87 cm CCL is critical information for developing management plans and demographic models for this sea turtle, which is listed as a threatened species in the U.S. Endangered Species Act of 1973. This size range includes the large juvenile and subadult lifestages defined in the stage-based population model developed for North Atlan- tic loggerhead sea turtles (Crouse et al., 1987; Crowder et al., 1994). This stage-based population model has identi- fied survivorship in the large juvenile lifestage as the most critical for population recovery. We based our length-fre- quency analyses on data collected from hundreds of log- gerhead sea turtle carcasses that were measured by the Sea Turtle Stranding and Salvage Network from Florida, Alabama, Mississippi, Louisiana, and Texas between 1988 and 1995. Methods Length-frequency data The Sea Turtle Stranding and Salvage Network (STSSN) is an organized network of individuals who monitor the shoreline and record data on each stranded sea turtle, including date and location of stranding, species, and cara- pace length (curved or straight carapace length, or both). Carapace length is measured from the anterior point at midline (nuchal scute) to the posterior tip of the supra- caudals. The stranding data are compiled and verified by state coordinators and archived at the Southeast Fish- eries Science Center (SEFSC) Miami Laboratory (Teas, 1993). We received data on stranded turtles for 1988 through 1995 for Alabama, Mississippi, Louisiana, and Texas from SEFSC and data for 1988 through 1995 for Florida from the Florida Department of Environmental Protection, Florida Marine Research Institute. All turtles known to have been “head-started” (that is, raised in cap- tivity before being released into the wild) were excluded from the analyses because growth rates in captivity, and therefore length-at-age, may be quite different for head- started turtles. We divided the data into two geographic regions: the Atlantic coast of Florida and the U.S. coast of the Gulf of Mexico (Florida, Alabama, Mississippi, Louisi- ana, and Texas). The Florida coast was divided between Atlantic and Gulf by the STSSN at 80.5°W. Because most of the data on carapace length were over- the-curve measurements, straight carapace lengths (SCL) were converted to curved carapace lengths (CCL) by using the conversion equation (n= 932, r2=0.97, P<0.001) in Teas (1993) SCL = (0.948 x CCL ) - 1.442. After conversions were completed, all CCL data were rounded to the nearest cm. We wanted to limit our analyses to the subadult, nerit- ic lifestage that inhabits the coastal waters of the south- eastern U.S.; therefore we limited the size range of logger- head sea turtles from 46 to 87 cm CCL. The lower value was based on length-frequency distributions (Bolten et al., 1993; Bjorndal et al., 2000) that indicated that 46 cm CCL is the size at which these sea turtles begin to recruit to neritic habitats in substantial numbers. The largest sub- adult size was taken as 87 cm CCL based on Withering- ton (1986), who reported that 88 cm CCL was the size of the smallest nesting loggerhead sea turtle at Melbourne Beach, Florida. This value is a very conservative division between subadults and adults; many with CCLs greater than 87 cm are still immature. For length-frequency anal- yses, however, it is better to exclude some subadults than to include many adult animals. Any factor that acts to obscure the modal structure of the sample — such as ces- sation or near-cessation of growth in older age classes — will decrease the potential for successful length-frequen- cy analysis. If older age classes cannot be distinguished, K (intrinsic growth rate) will be overestimated and the number of age classes underestimated (Terceiro et ah, 1992). Because loggerheads essentially stop growing at sexual maturity and because they attain sexual maturity at a range of sizes (Frazer and Ehrhart, 1985), the age classes — or modes — above the minimum size at sexual ma- turity are obscured and cannot be distinguished in length- frequency analyses. Kolmogorov-Smirnov analyses were conducted with SPSS software (version 9.0 SPSS, 1996). Length-frequency analysis We used MULTIFAN (version 32(f), Otter Research Ltd., 1992) modified to include 30 age classes by Fournier ( Otter Research Ltd., 1992). MULTIFAN simultaneously ana- lyzes multiple samples of length-frequency data (Otter Research Ltd., 1992) and uses nonlinear statistical mod- 242 Fishery Bulletin 99(2) eling and robust parameter estima- tion to estimate the parameters of the von Bertalanffy growth function (Fournier et ah, 1990, 1991). Log-like- lihood objective functions are com- pared by using maximum likelihood analyses to identify the parameter set for the von Bertalanffy model with the best fit. The form of the von Bertalanffy equation that was used in the MUL- TIFAN program is u , = m, +(rnN -m,) 1 -P"" 1-P' where hm = the mean length of the age class j turtles in the orth length frequency data vset; mx = the mean length of the first age class; rn N - the mean length of the last age class; p = the Brody growth coef- ficient; /??(«)- 1 = the number of months after the presumed birth month of the turtle in the orth length-frquency data set; and N = the number of age classes in the data set. among years by month so that samples were sufficiently large for length-frequency analyses. This parameterization of the von Bertalanffy growth equa- tion is derived in Schnute and Fournier ( 1980). The MLTLTIFAN length-frequency program has the fol- lowing assumptions: 1) growth is described by a von Ber- talanffy growth curve; 2) samples represent the structure of the population; 3) recruitment occurs in seasonal puls- es, 4) the lengths of animals in each age class are normally distributed; and 5) the standard deviations of the lengths are a simple function of the mean length-at-age. MULTIFAN requires that initial values for the follow- ing parameters be designated as starting points for the iterations: expected number of age classes; expected ini- tial K values; mean length of the mode representing the youngest age class; standard deviation of a distinct mode; and month in which youngest animals recruit to the popu- lation. We estimated initial values for expected number of age classes as varying between 2 and 30 years, and for K as 0.01, 0.05, 0.1, and 0.5/yr. The initial estimate for mean length of the youngest age class was 47 cm, and the ini- tial standard deviation of mode width was estimated as 1.5 cm. Because there was a significant trend in standard deviation of length-at-age with increasing length, this pa- rameter was included in the models reported here. April was designated as the month in which youngest turtles recruit into the population because the April samples had the smallest individuals. The CCL data were combined Results The length-frequency distributions of loggerhead sea tur- tles within the size range of 46 to 87 cm CCL that stranded from 1988 through 1995 along the Atlantic coast of Florida ( /? = 1 234 ) and along the U.S. coast of the Gulf of Mexico (Gulf coast of Florida, Alabama, Mississippi, Louisiana, and Texas, n=570) are shown in Figures 1 and 2, respectively. The two distributions are significantly different (Kolmogorov-Smirnov test, Z=3.934, PcO.OOl), although the relative patterns are similar. We assumed that the length-frequency distributions of stranded sea turtles are representative of the length-frequency distri- butions of sea turtles in the two regions, although there is potential for sampling bias from incidental capture in commercial fisheries. For loggerhead sea turtles in both the Florida Atlantic and the Gulf of Mexico, the MULTIFAN analysis estimat- ed that the 46 to 87 cm CCL size range comprises 20 year classes (Table 1, Fig. 3). For both geographic regions, vi- sual inspection revealed that the models fit the length- frequency data well (an example is shown in Fig. 4). We were unable to assess annual variation because we com- bined data for each month from different years owing to small sample sizes. Also, combining data among years B|orndal et a!.: Somatic growth function for immature Caretta caretta 243 Table 1 Number of age classes (years) in Atlantic loggerhead sea turtle populations within the size range of 46 to 87 cm curved carapace length. Estimates were calculated from von Bertalanffy growth models presented in the referenced studies. Method is type of data (SC=skeletochronology, MR=mark-recapture, LF=length frequency); n = sample size. Location Number of age classes n Method Reference Chesapeake Bay 13 83 SC Klinger and Musick, 1995 Cumberland Island, Georgia IF 25' SC Parham and Zug, 1997 9- 26- SC Parham and Zug, 1997 Cape Canaveral, Florida 29,J 5D MR Schmid, 1995 264 \14 MR Schmid, 1995 Mosquito Lagoon, Florida 15 28 MR Frazer and Ehrhart, 1985 Florida, Atlantic Coast 20 1234 LF This study Gulf of Mexico 20 570 LF This study ' Based on “1979-new” data set in which correction-factor protocol was used (Parham and Zug, 1997). 2 Based on “1980” data set in which correction-factor protocol was used (Parham and Zug, 1997). 3 Based on “all recaptures” data set (Schmid, 1995). 4 Based on “contract vessel” data set (Schmid, 1995). could have obscured age-class modes, but because MULTIFAN was able to distinguish modes and the models fit- ted the data well, this procedure was apparently not a problem. The von Bertalanffy parameter es- timates generated by MULTIFAN for asymptotic length (LJ) and intrinsic growth rate ( K ) for the Florida At- lantic were L„= 118.5 ±0.7 cm and A"=0.044 ±0.001/yr and for the Gulf of Mexico were Lx= 113.0 ±0.4 cm and A=0.051 ±0.001/yr. The maxi- mum CCL value recorded for logger- heads nesting in the southeastern United States is 124 cm (Dodd, 1988); therefore the estimates of asymptotic length are reasonable. Discussion Duration of the neritic lifestage Most loggerhead sea turtles in the North Atlantic recruit to neritic habi- tats over a size range of 46 to 64 cm CCL (Bjorndal et al., 2000). Size at recruitment will affect duration of the neritic lifestage. From our model, log- gerhead sea turtles that recruit at 46 cm CCL will require 20 years to reach 87 cm CCL, whereas those that recruit at 64 cm CCL will require 13 years to grow to 87 cm CCL (Fig. 3). Because loggerhead sea turtles usually remain in neritic habitats after recruiting to these habitats, total duration of the neritic lifestage will include the years subsequent to attaining 87 cm CCL. The demo- graphic ramifications of this variation in size at recruit- ment should be evaluated. 244 Fishery Bulletin 99(2) Other studies have used von Bertalanffy models to eval- uate growth in immature loggerhead sea turtles in the At- lantic (Table 1) based on either mark and recapture data (Frazer and Ehrhart, 1985; Schmid, 1995) or skeletochro- nology data (Klinger and Musick, 1995; Parham and Zug, 1997). Schmid ( 1995) and Parham and Zug ( 1997) present- ed data for a number of data sets and types of calculation. For both studies, the data presented in Table 1 are for the two data sets that Schmid (1995) and Parham and Zug (1997) considered the most representative. Our estimates of 20 year classes in the size range from 46 to 87 cm CCL fall within the range of values of 9 to 29 years that we calculated from the von Bertalanffy growth equations presented in other studies (Table 1). There is no geographic trend in these estimates. Von Bertalanffy growth model MULTIFAN uses a von Bertalanffy growth model as de- rived by Schnute and Fournier (1980). Concerns about the application of the von Bertalanffy model to sea turtle populations have been discussed elsewhere (Bjorndal and Bolten, 1988; Chaloupka and Musick, 1997). A major con- cern is the use of the von Bertalanffy model to extrapolate outside the size range of a study (Day and Taylor, 1997), which has been a common practice in estimating age at sexual maturity in sea turtle populations (reviewed in Cha- loupka and Musick, 1997). However, within a growth phase that exhibits monotonic decline — that is, for a given size range of a population with similar habitats and diet and that has declining growth rates with increasing body size — the von Bertalanffy model may provide reasonable esti- mates of growth rates and number of age classes. Green turtles in the southern Bahamas have a monotonic non- linear declining function for SCL-specific growth rates in a size range of 30 to 70 cm SCL as determined from non- linear regression of mark-recapture data (Bjorndal et ah, 2000). Length-frequency analyses with MULTIFAN yielded the same estimates as the nonlinear regression analysis for growth rates and number of age classes between 30 and 70 cm SCL for that population (Bjorndal et ah, 2000). Similar trends of decreasing growth rates with increasing size have been reported from mark-recapture studies in Atlantic log- gerhead sea turtles (summarized in Parham and Zug, 1997). In addition, use of the von Bertalanffy model within the studied size range is supported by the similarity between growth rates generated from MULTIFAN and those calcu- lated from recaptures of tagged pelagic-stage loggerhead sea turtles (Bjorndal et ah, 2000). As growth data accu- mulate for sea turtle populations, appropriate growth func- tions— which may well be polyphasic (Chaloupka and Zug, 1997; Chaloupka, 1998) as a result of habitat shifts — can be incorporated into the length-frequency software. Age at sexual maturity An estimate of the number of years to grow to 87 cm CCL may be gen- erated by combining our results (20 year classes between 46 and 87 cm CCL) with those from a growth study in pelagic-stage loggerhead sea tur- tles in the North Atlantic (Bjorndal et ah, 2000). In that study, based on length-frequency analyses and mark-recapture data, the age of log- gerheads with a CCL of 46 cm was estimated to be 6.5 years. Therefore, an 87-cm loggerhead sea turtle would be approximately 26.5 years old. This estimate of 26.5 years at 87 cm CCL should not be used as an es- timate of age at sexual maturity. Al- though we used 87 cm CCL to rep- resent the largest subadult in our study, as explained above, we used a conservative estimate to exclude al- most all adults from the sample to avoid obscured modes in the length- frequency distributions. Many logger- head sea turtles will reach sexual ma- turity at lengths much greater than 87 cm CCL, and because growth rates are slow in these large subadults, the average age at sexual maturity would be substantially greater than the av- erage age at 87 cm CCL. Figure 3 Age-specific von Bertalanffy growth model generated by MULTIFAN for loggerhead sea turtles stranded along the Atlantic coast of Florida from 1988 through 1995 within the size range of 46 to 87-cm curved carapace length. The growth curve for loggerhead sea turtles stranded in the Gulf of Mexico is not plotted because it is indistinguishable from the line plotted for the Atlantic coast of Florida. Age is defined as years since recruitment to neritic habitats at a size of 46 cm curved carapace length. Bjorndal et al.: Somatic growth function for immature Caretta caretta 245 Conclusion With length-frequency analyses, we estimated that growth from 46 to 87 cm CCL in Atlantic loggerhead sea turtles requires 20 years. This estimate falls within the range of estimates of 9 to 29 years that we calculated from other studies based on mark-recapture and skeletochronology. Because survivorship in this size class has been identified as a critical parameter for population recovery (Crouse et al., 1987; Crowder et al., 1994), accurate estimates for the duration of the stage are essential for developing suc- cessful management plans. Research must be continued to refine this estimate. Conservation of loggerhead sea turtles that spend an ex- tended period of time in nearshore habitats prior to reach- ing sexual maturity is compromised. Numerous and signif- icant threats — including incidental capture in commercial fishing operations, collisions with motorized vessels, dredg- ing operations, exposure to pollutants and biotoxins, and habitat degradation — are present in nearshore develop- mental habitats. There is a high probability that these tur- tles will encounter one or more of these threats during their maturation period (National Research Council, 1990; Eckert, 1995; Lutcavage et al., 1997). To be successful, re- covery activities must be sustained for long periods of time, and long-term monitoring programs to assess the status of populations of juvenile loggerhead sea turtles in U.S. wa- ters must be established. Because current loggerhead pop- ulation assessments depend upon the numbers of nesting females or nests, two or more decades must pass before re- sults of recovery activities aimed at the earliest age classes in nearshore waters can be evaluated. Acknowledgments This project could not have been conducted without the long hours invested by the many participants in the Sea Turtle Stranding and Salvage Network in Alabama, Flor- ida, Louisiana, Mississippi, and Texas. In particular, we thank Gary Hopkins and Robert Shipp, the coordinators for Mississippi and Alabama, respectively, and Carrie Crady, Allen Foley, and Ron Mezich. Support for this proj- ect was provided by the MARFIN program of the National Marine Fisheries Service (project NA57FF0063 to KAB and ABB). We thank Nancy Thompson for her assistance with this project. David Fournier generously provided advice and modification of the MULTIFAN software. Peter Eliazar assisted with data compilation, and Jennifer Pias- cik drafted Figure 4. 246 Fishery Bulletin 99(2) Literature cited Bjorndal, K. A., and A. B. Bolten. 1988. Growth rates of immature green turtles, Chelonia mydas , on feeding grounds in the southern Bahamas. Copeia 1988:555-564. 1995. Comparison of length-frequency analyses for estima- tion of growth parameters for a population of green turtles. Herpetologica 51:160-167. Bjorndal, K. A., A. B. Bolten, and M. Y. Chaloupka. 2000. Green turtle somatic growth model: evidence for den- sity dependence. Ecol. Applic. 10:269-282. Bjorndal, K. A., A. B. Bolten, A. L. Coan Jr., and P. Kleiber. 1995. Estimation of green turtle ( Chelonia mydas) growth rates from length-frequency analysis. Copeia 1995:71-77. Bjorndal, K. A., A. B. Bolten, and H. R. Martins. 2000. Somatic growth model of juvenile loggerhead sea tur- tles Caretta caretta-. duration of pelagic stage. Mar. Ecol. Prog. Ser. 202:265-272. Bolten, A. B., H. R. Martins, K. A. Bjorndal, and J. Gordon. 1993. Size distribution of pelagic-stage loggerhead sea tur- tles (Caretta caretta) in the waters around the Azores and Madeira. Arquipelago llA:49-54. Chaloupka, M. Y. 1998. Polyphasic growth in pelagic loggerhead sea turtles. Copeia 1998:516-518. Chaloupka, M. Y., and C. J. Limpus. 1997. Robust statistical modelling of hawksbill sea turtle growth rates (southern Great Barrier Reef). Mar. Ecol. Prog. Ser. 146:1-8. Chaloupka, M. Y., and J. A. Musick. 1997. Age, growth, and population dynamics. In The biol- ogy of sea turtles (P. L. Lutz and J. A. Musick, eds.), p. 233- 276. CRC Press, Boca Raton, FL. Chaloupka, M. Y., and G. R. Zug. 1997. A polyphasic growth function for the endangered Kemp’s ridley sea turtle, Lepidochelys kempt. Fish. Bull. 95:849-856. Crouse, D. T., L. B. Crowder, and H. Caswell. 1987. A stage-based population model for loggerhead sea turtles and implications for conservation. Ecology 68: 1412-1423. Crowder, L. B., D. T. Crouse, S. S. Heppell, and T. H. Martin. 1994. Predicting the impact of turtle excluder devices on log- gerhead sea turtle populations. Ecol. Applic. 4:437-445. Day, T., and P. D. Taylor. 1997. Von Bertalanffy’s growth equation should not be used to model age and size at maturity. Am. Nat. 149:381-393. Dodd, C. K., Jr. 1988. Synopsis of the biological data on the loggerhead sea turtle Caretta caretta (Linnaeus 1758). U.S. Fish Wildl. Serv. Biol. Rep. 88:1-110. Eckert, K. L. 1995. Anthropogenic threats to sea turtles. In Biology and conservation of sea turtles, revised ed. (K. A. Bjorndal, ed.), p. 611-612. Smithsonian Institution Press, Washington, D.C. Fournier, D. A., J. R. Sibert, J. Majkowski, and J. Hampton. 1990. MULTIFAN, a likelihood-based method for estimat- ing growth parameters and age composition from multiple length frequency data sets illustrated using data for south- ern bluefin tuna (Thunnus maccoyii ). Can. J. Fish. Aquat. Sci. 47:301-317. Fournier, D. A., J. R. Sibert, and M. Terceiro. 1991. Analysis of length frequency samples with relative abundance data for the Gulf of Maine northern shrimp ( Pandalus borealis) by the MULTIFAN method. Can. J. Fish. Aquat. Sci. 48:591-598. Frazer, N. B., and L. M. Ehrhart. 1985. Preliminary growth models for green, Chelonia mydas , and loggerhead, Caretta caretta, turtles in the wild. Copeia 1985:73-79. Hilborn, R., and C. J. Walters. 1992. Quantitative fisheries stock assessment: choice, dy- namics and uncertainty. Chapman and Hall, New York, NY, 570 p. Holden, S., and M. Bravington. 1992. Length frequency distribution analysis, the LFDA package, version 3.10. Marine Resources Assessment Group Ltd., London. Klinger, R. C., and J. A. Musick. 1995. Age and growth of loggerhead turtles (Caretta caretta ) from Chesapeake Bay. Copeia 1995:204-209. Limpus, C., and M. Chaloupka. 1997. Nonparametric regression modelling of green sea turtle growth rates (southern Great Barrier Reef). Mar. Ecol. Prog. Ser. 149:23-34. Lutcavage, M. E., P. Plotkin, B. Witherington, and P. L. Lutz. 1997. Human impacts on sea turtle survival. In The biol- ogy of sea turtles (P. L. Lutz and J. A. Musick, eds.), p. 387- 409. CRC Press, Boca Raton, FL. National Research Council. 1990. Decline of the sea turtles: causes and prevention. National Academy Press, Washington, D.C., 259 p. Otter Research Ltd. 1992. MULTIFAN 3 user’s guide and reference manual. Otter Research Ltd., Nanaimo, B.C., Canada, 70 p. Parham, J. F., and G. R. Zug. 1997. Age and growth of loggerhead sea turtles (Caretta caretta) of coastal Georgia: an assessment ofskeletochrono- logical age-estimates. Bull. Mar. Sci. 61:287-304. Pauly, D., and G. R. Morgan (eds.). 1987. Length-based methods in fisheries research. ICLARM Conference Proceedings, No. 13, ICLARM, Manila, Philip- pines, 468 p. Ricker, W. E. 1975. Computation and interpretation of biological statis- tics of fish populations. Fish. Res. Board Can., Bull. 191. Schnute, J., and D. A. Fournier. 1980. A new approach to length frequency analysis: growth structure. J. Fish. Res. Board Can. 37:1337-1351. Schmid, J. R. 1995. Marine turtle populations on the east-central coast of Florida: results of tagging studies at Cape Canaveral, Florida, 1986-1991. Fish. Bull. 93:139-151. SPSS, Inc. 1996. SPSS user’s guide. SPSS Inc., Chicago, IL, 564 p. Teas, W. G. 1993. Species composition and size class distribution of marine turtle strandings on the Gulf of Mexico and south- east United States coasts, 1985-1991. U.S. Dep. Commer., NOAATech. Memo. NMFS-SEFSC-315, 43 p. Terceiro, M., D. A. Fournier, and J. R. Sibert. 1992. Comparative performance of MLTLTIFAN and Shep- herd’s length composition analysis (SLCA) on simulated length-frequency distributions. Trans. Am. Fish. Soc. 121: 667-677. Witherington, B. E. 1986. Human and natural causes of marine turtle clutch and hatchling mortality and their relationship to hatchling production on an important Florida nesting beach. Mas- ters thesis, Univ. Central Florida, Orlando, FL, 141 p. 247 Abstract —Pots are a form of trap used to capture fishes, crustaceans, or gastro- pod mollusks. Occasionally, these traps are lost at sea, where they have the potential to fish for many years because they are constructed of robust man- made materials. The present study quantified the mortality and number of animals caught by a fleet of crusta- cean pots (12 pots) that were set on the seabed and left to fish continually in a manner designed to simulate ghost- fishing off the coast of Wales, UK. The bait originally placed in the pots was consumed within 28 days of the begin- ning of the experiment. Spider crabs and brown crabs dominated the catches within the pots throughout the experi- ment. The CPUE of spider and brown crabs declined as an inverse function of time and reached a minimum between 125 to 270 days after initial deploy- ment in August 1995. After this period, CPUE increased again, although it did not attain the rates associated with the beginning of the experiment. The fleet of twelve pots caught a minimum of 7.08 spider and 6.06 brown crabs per pot per year and killed a minimum of 6.06 brown crabs and 0.44 lobsters per pot per year. Other species caught in the traps included velvet swimming crab, lobster, ballan wrasse, dogfish, and triggerfish. The pots continued to catch animals into the second year of the experiment. These results suggest that pots have the potential to fish for extended periods. The wider use of biodegradable escape panels is rec- ommended because currently there is no national legislation in the UK to enforce such escape measures. Manuscript accepted 6 October 2000. Fish. Bull. 99:247-253 (2001). A study of catches in a fleet of "ghost-fishing" pots Blaise A. Bullimore Philip B. Newman Countryside Council for Wales Skomer Marine Nature Reserve Fisherman's Cottage Martin's Haven Dyfed, SA62 3BJ, United Kingdom Michel J. Kaiser School of Ocean Sciences University of Wales-Bangor Menai Bridge Gwynedd, LL59 5EY, United Kingdom E-mail address (for M. J Kaiser, contact author): m.j kaiser@bangor ac.uk Susansie E. Gilbert Kate M. Lock Countryside Council for Wales Skomer Marine Nature Reserve Fisherman's Cottage Martin's Haven Dyfed, SA62 3BJ, United Kingdom Static fishing gear, such as gill and tram- mel nets and pots or traps, are consid- ered to be highly selective for target species. In addition, these gears might be considered to be environmentally friendly because they cause relatively little disturbance of seabed communi- ties when compared with towed bot- tom-fishing gears (Jennings and Kaiser, 1998). However, set nets or pots can be lost as a result of bad weather, ice chafing and cutting mooring ropes, pots being snagged on seabed obstructions, or pots being inadvertently towed away by mobile fishing gears. The lost gear may then continue to fish indiscrimi- nately, which is a phenomenon known as “ghost-fishing.” In contrast with the numerous records of bird, reptile, and cetacean entanglement in set gears (see Dayton et al., 1995 and references therein), little is known about the fre- quency of static gear loss or for how long such gears continue to fish. The pau- city of information relating to this phe- nomenon results from the reluctance of fishermen to report such incidents and the difficulty in undertaking long-term studies in a realistic manner. Estimates of the proportion of nets lost from commercial fleets appear to be substantial. Approximately 7000 km of drift nets (20-30% of the total nets set each day) were lost per year in a North Pacific fishery (Eisenbud, 1985). Considerable numbers of pots are also lost each year from some fisheries, al- though estimates vary greatly between different studies. For example, Kruse and Kimker1 estimated that that in 1990 and 1991, 31,600 pots per year were lost in the North American Bris- tol Bay king crab ( Pciralithodes camt- schaticus) fishery, whereas Paul et al. (1994) and Stevens (1996) estimated that losses from the same fishery were respectively 20,000 and 7000 pots per year. Breen (1987) estimated that 11% of the traps used in the Dungeness crab Cancer magister fishery of British Co- lumbia are lost each year. Modern com- 1 Kruse, G. H., and A Kimker. 1993. De- gradable escape machanisms for pot gear: a summary report to the Alaska Board of Fisheries. Regional Information Report 5J93-01, 23 p. Alaska Depart ment of Fish and Game (ADFG),211 Mission Rd. Kodiak, AK 99615, Alaska. 248 Fishery Bulletin 99(2) mercial set nets and pots are made of nonbiodegradable man-made materials. As a result, lost nets and pots have the potential to persist and continue to fish in the marine environment for several years depending upon the pre- vailing environmental conditions (Breen, 1987; Carr et ah, 1990; Kaiser et ah, 1996). Set nets that are lost in areas exposed to large swell and storm activity (e.g. off the North west coast of Spain) are rapidly destroyed (Puente2). Nets lost in shallow, clear water are rapidly overgrown with encrusting biota that makes them more visible and reduc- es their fishing capabilities (Erzini et ah, 1997). However, when static gear becomes snagged on rocks that hold it in place, or is lost in deep water in a relatively stable en- vironment, it may continue to fish for more than a year (Carr et ah, 1990; Kaiser et ah, 1996). Lost pots are likely to continue fishing for longer than stat- ic nets because they are constructed either entirely of metal or of thick net t ing attached to a rigid frame. Exact ly how long lost pots are likely to continue fishing remains unquantified. The present study was undertaken in grounds where pot gear is typically used — at a site within the Skomer Ma- rine Nature Reserve (MNR), Wales, UK. MNR staff have recorded the occurrence of lost pots over a number of years within the waters of the reserve. Most observations were of single pots in locations sheltered from strong water move- ment. Such pots were usually heavily overgrown with ses- sile biota, indicating that they had been submerged for at least 6 months. Lost pots within the MNR have been reported by recreational divers who have described occa- sional large resident catches of Crustacea and fish, such as ballan wrasse ( Lcibrus bergylta) and conger eels (Conger conger). The objectives of our study were 1) to quantify the number of organisms removed by a fleet of lost pots; 2) to describe changes in catch rate over time; and 3) to record any deterioration in the integrity of the fishing gear. Methods Site selection The study site was located in the Skomer MNR off the coast of Pembrokeshire, Wales (Fig. 1). This site was rep- resentative of those fished by locally based inshore pot fishermen. Maximum tidal streams ranged from 2 to 3 knots on spring tides. The site was sheltered from heavy wave action and the seabed comprised mainly boulders to a depth of 15 m below the lowest recorded astronomical tidal height. Below this depth, the substratum was com- posed of mixed coarse sediments, cobbles and boulders, and occasional bedrock outcrops. Gear and deployment The gear comprised 12 parlor pots of approximate dimen- sions 750 mm x 500 mm x 500 mm with a top-mounted cylindrical entrance ring made of plastic (Fig. 1). Two sizes 2 Puente, E. 1997. Personal commun. Isla de Txatxarramendi, s/n 48395 Sukarrieta (Bizkaia), Spain. of entrance ring were used, either 200 mm (n= 7) or 240 mm (n=5) diameter. In our analysis we made no attempt to differentiate catches based on entrance ring size because size of trap entrance was not a major consideration in our experiment. The pots were attached to a mainline by using 3.5-m lengths of polypropylene rope at intervals of 18 m. This configuration of the gear is the normal fishing prac- tice used in the UK (Fig. 1). The fleet of pots was anchored at each end with a 75-kg weight and marked with surface marker buoys (Fig. 1). Each pot was baited with the corpse of a thornback ray ( Raja clavata ) (its wings having been removed) and attached to the entrance ring with a rubber band. On each sampling occasion the position of the fleet of pots was determined by a differential global positioning system. The pots were deployed on 4 August 1995. During deployment, the mainline was checked to see that it was tight between each of the pots when shot from the boat. Gear deployment was undertaken in consultation with a local professional pot fisherman. Divers surveyed the fleet of pots immediately after it had been shot away to confirm that the gear was deployed on the seabed correctly and to record the depth at which each pot had settled. It was thought important to record pot depth during the study because pots lost in shallow water might be more prone to destruction by wave action, whereas progressive movement of the pots into deeper water might prolong their fishing capabilities. Although it would have been preferable to have deployed replicate fleets of pots in different habitats, the logistics of diver-based observations and the need to mini- mize the adverse effects of the experiment on local popula- tions of animals meant that this was neither possible nor considered ethically acceptable. Indeed, the ethical consid- erations of undertaking such an experiment in an MNR meant that 3 pots were removed from the experiment after 88 days and a further 3 pots removed after 270 days of fish- ing to minimize animal deaths associated with the study. Data recording Observations on each of the pots were recorded by divers 1, 4, 12, 27, 40, 69, 88, 101, 125, 270, 333, 369, and 398 days after initial deployment. Divers undertook the following tasks: 1 recorded the depth (adjusted for tidal height) of pots throughout the experiment. 2 recorded the identity of the catch in each pot. 3 tagged newly captured crustaceans (brown crab [Cancer pcigurus ] spider crab [ Maja squinado } and the lobster Homarus gammarus ) on each sampling occasion with coded and colored cable ties on their appendages so that they could be distinguished from newly captured ani- mals (without tags) on subsequent occasions. The identification of previously caught individual lob- sters, brown crabs, and spider crabs caught in the pots was checked by inspection of the species-specific colored and coded tags placed on different body parts of each ani- mal during a previous sampling routine. It was not pos- sible to tag smaller crustaceans such as the velvet swim- Bullimore et al.: Study of catches in a fleet of "ghost-fishing" pots 249 B Pot entrance One way entrance to parlor Parlor c (A) Map of Skomer Island MNR, Pembrokeshire, Wales, UK, showing the location of the ghost fishing experiment, (B) a schematic diagram of the construction of a parlor pot, and (C) the configuration of a deployed fleet of pots. ming crab ( Liocarcinus puber) and fish were never tagged. However, fish were readily identifiable on consecutive oc- casions from the injuries sustained while resident in the pots. These injuries most frequently occurred as wounds on the head region that were sustained as the fish tried to escape through the pot meshes. Records were made of the presence and condition of bait on each occasion. Pot depth was estimated from divers’ depth gauges and corrected for the state of tide from appropriate tide-tables. Statistical analyses Variation in the total resident catch (all species) within the pots at each time interval was analyzed by using a GLM (general linear modeling) ANOVA with time as the independent variable. Catch rates were calculated as the number of newly captured animals caught per pot and recorded by divers on each consecutive sampling occasion. It was not possible to standardize the sampling time inter- val because weather conditions and water visibility during the winter period were not conducive to regular sampling. Because the intersample period and number of pots varied throughout the experiment, the catch data were expressed as catch-per-unit-of-effort (CPUE) data with the following formula: CPUE = Nil(Ep(tl-t,)), where N t - the number of newly caught animals; E/( = number of pots fishing; and tj — tt = the time interval since the previous observa- tion (tt). The existence of a relationship between the decline in catch rate with time was determined by using nonlinear regression analysis in the SPSS statistical software pack- age (SPSS, 1998). These analyses were performed only for brown and spider crabs because other animals were caught in insufficient numbers for meaningful analyses. Preliminary examination of the data suggested that the capture rate for the crab species differed with season; hence the regression analyses were undertaken by using only the data for the first 125 days of the experiment. Fur- thermore, there was a long period without direct obser- vations from 125 to 270 days after initial deployment of the gear. Catch rate does not provide an indication of the total actual mortality of individuals associated with the ghost-fishing pots. Mortality was confirmed when divers observed the remains of individuals and their tags either at the bottom of the pot, or on the seabed nearby, because both the tags and dead animal fragments were small 250 Fishery Bulletin 99(2) Table II Resident catches (all species) in each of the twelve pots in the experimental fleet observed after initial deployment of gear at time intervals from August 1995 to September 1996. Mean resident catch per pot is also shown. “R” indicates those pots that were removed at each time interval Pot number Month Day 1 2 3 4 5 6 7 8 9 10 11 12 Mean ± 95% Cl Aug 1 1 2 0 0 1 4 4 1 4 2 2 0 1.75 ±0.87 Aug 4 2 2 1 0 2 4 4 1 4 3 4 1 2.33 ±0.81 Aug 13 3 2 2 2 3 4 4 4 6 0 4 0 2.83 ±0.99 Aug 27 2 2 3 3 4 7 4 6 5 1 3 2 3.50 ±1.01 Sep 40 0 2 3 3 4 5 4 6 5 1 3 1 3.08 ±1.04 Oct 69 0 2 6 3 5 7 5 6 7 1 1 3 3.83 ±1.40 Nov 88 0 2 9 4 5 7 3 5 7 1 1 2 3.83 ±1.60 Nov 101 R 3 8 R 6 6 3 R 7 1 1 2 4.11 ±1.74 Dec 125 R 3 5 R 6 2 5 R 7 2 2 1 3.67 ±1.39 May 270 R 1 0 R 5 2 5 R 2 0 0 0 1.67 ±1.65 Jul 333 R 5 2 R 3 R 4 R 4 0 R R 3.00 ±1.17 Aug 369 R 7 0 R 2 R 5 R 3 2 R R 3.17 ±1.62 Sep 398 R 7 0 R 2 R 5 R 3 2 R R 3.17 ±1.62 enough to pass through the meshes of the pots. Thus we were able to quantify the minimum mortality of animals captured per pot per year for the data up to the first 369 days (approximately 12 months) after initial deployment. The effect of substratum type on the movement of pots on the seabed was analyzed for the first 7 observation pe- riods by using one-way ANOVA. Data from the remaining observation periods were not used because some of the pots were removed after the seventh sampling date. Three different substratum categories were used: rock and large boulders; medium boulders and cobbles; and mixed sedi- ments. Results The bait within the pots remained secure during their deployment. The bait was consumed rapidly, and by day 13 only remnants of the bait were observed in 9 of the pots, and by day 27 only 1 pot had any remnants of bait. Throughout the experiment , none of the pots became detached from the mainline between the sinkers. Pot movement was greatest in the first few weeks after deploy- ment and was related to seabed topography. The entire fleet of pots tended to move in a north-westerly direction in line with the prevailing tidal currents as determined from the position of the surface marker buoys. Substratum type did not affect the extent to which the individual pots moved on average during the first 88 days of the experi- ment (ANOVA, i\59=0.62, P=0.56). Throughout the experiment seven different species were captured in the pots. Divers observed that some animals were always resident in the pots because previously ob- served animals were frequently observed on subsequent sampling dates. The number of animals resident in the pots did not differ significantly between different observa- tion periods (Table 1, GLM FV2 112=1.07,P=0.39).This find- ing suggests that as animals died or escaped, they were re- placed by new catches. Only spider and brown crabs were captured in sufficient numbers for more detailed analyses. Analyses of the CPUE up to 125 days after initial deploy- ment of the pots revealed that the CPUE decreased sig- nificantly for spider crab and brown crab (Fig. 2). The best relationship between the decline in CPUE with time was given by the following relationships: CPUE spider crab = 0.0216 + (0.377/day) [r=0.97, Fl 7=236.7, P<0.0001], CPUE broum crab - 0.0205 + (0.244/day) [/-0.92, P17=77.5, PcO.0001]. It is clear from Figure 2 that the CPUE for spider crab and brown crab declined to a minimum rate between 125 and 270 days after initial deployment and then increased rap- idly before decreasing again. The regression equation used to calculate the total predicted catch for the first 125 days of the experiment gives a catch of 4.74 spider crabs and 3.88 brown crabs per pot. Catch rate, however, does not provide an indication of the total mortality of animals associated with the ghost- fishing pots. We were able to confirm annual mortality on- ly per pot for lobster and brown crabs, of which 100% died because divers retrieved all the tags deployed on individu- al lobsters and brown crabs in the pots (Table 1). A conser- vative estimate of total mortality indicates that each pot killed on average 6.06 brown crabs and 0.44 lobsters per year (Table 2). This finding takes no account of animals Bullimore et al.: Study of catches in a fleet of "ghost-fishing" pots 251 Table 2 Newly captured animals for the entire fleet of pots at various time intervals after first deployment of gear. For ethical reasons, after day 88, the number of pots was reduced to 9 and after day 270 the number of pots was further reduced to 6. Mean catch per pot per year was calculated from the sum of the total new catch divided by the number of pots fishing at each time interval. Days from original deployment 1 4 12 27 40 69 88 101 125 270 333 369 Number of pots fishing Mean catch Species Common name 12 12 12 12 12 12 12 9 9 9 6 6 per pot per year Maja squinado spider crab 15 3 9 10 6 7 3 2 0 1 8 6 7.08 Cancel' pagurus brown crab 3 5 5 3 3 9 4 0 3 8 7 6 6.06' Necora puber velvet swimming crab 1 0 1 0 0 0 0 2 2 0 0 0 0.61 Homarus gammarus lobster 2 1 1 0 0 0 0 0 1 0 0 0 0.44' Labrus bergylta ballan wrasse 0 0 0 1 0 0 2 0 1 1 4 5 1.97 Bctllistes carolinensis triggerfish 0 0 0 0 0 0 1 0 0 0 0 0 0.08 Scyliorhinus canicula lesser-spotted dogfish 0 0 0 0 0 0 1 0 0 0 0 0 0.08 1 indicates those animals for which all observed catches died. that may have entered the pots and died without being ob- served by the divers. Discusssion The results of this experiment demonstrate that lost pots can continue to fish for many months after the bait within them has been consumed. There was no sign of deterio- ration in the integrity of the pots throughout the study. Several “ghost-fishing” pots were found in close proximity to the experimental fleet. This discovery emphasized the practicality of our study and confirmed that location of our experiment was representative of sites used by pot fisher- men and that pot losses do occur in this area. Spider and brown crabs dominated the catches, whereas lobster, velvet swimming crab, and fishes were caught less frequently. The rate of decline of the CPUEs for spider and brown crabs both followed similar patterns. We extrapolated the CPUE for only the first 125 days of the experiment be- cause our own observations indicated that spider crabs mi- grate into shallower waters around Skomer to breed in the late spring and early summer. Consequently, catches of spi- der crab were high in August and tailed off as the experi- ment extended into autumn and winter and the spider crabs migrated farther offshore. In contrast, brown crabs are resi- dent throughout the year in the Skomer MNR and although the CPUE was low during the winter period, it is possible that they continued to be caught sporadically between days 125 and 270. A second pulse of spider crab and brown crab catches occurred 333 days after initial deployment of the pots, which may be linked to rising water temperatures that occur in late spring and early summer (Fig. 2). We were able to confirm mortality only for lobster and brown crabs (0.44 lobster and 6.06 brown crab per pot per 1.25 04 0 35 0.3 0.25 ^ 0 100 200 300 400 500 LL o 0 100 200 300 400 500 Days after gear deployment Figure 2 The catch per unit of effort (CPUE) for (A) spider crabs and (B) brown crabs on each of the sampling dates (open circles). The closed circles represent the predicted values for CPUE as determined from the regression relationships (see text). A 252 Fishery Bulletin 99(2) Table 3 Bylaws currently in operation in UK waters pertaining to the use of escape gaps and pot design. Cumbria Sea Fisheries Committee Bylaw No. 25: No person shall use or cause to be used for the purpose of fishing for seafish or Crustacea any pot, creel or trap constructed of whatever material unless: • it has at least one unobstructed escape gap located in the lowest part of the pot, creel or trap or in the case of a parlour pot the parlour area; and • is so designed and constructed that each escape gap is of sufficient size that there may be easily passed through the escape gap and completely passed in to the pot, creel or trap a rigid boxed shaped gauge which shall be a gauge 74 mm wide, 44 mm high and 100 mm long. States of Jersey Requirements of escape gaps in parlour pots: No person shall use or cause to be used for the purpose of fishing for seafish any parlour pot, of whatever material constructed, unless: • it has a least one unobstructed escape gap which shall be located in the lowest part of the parlour areas on a side or sides of the parlour or the bottom of the parlour pot; and • is so designed and constructed that each escape gap is of sufficient size that there may be easily passed through the escape gap and completely inserted into the parlour pot, whether the parlour pot is wet or dry, a rigid boxed shaped gauge which - • in the case of an escape gap located on a side of the parlour pot, shall be a gauge 79 mm wide, 44 mm high and 100 mm long; and • in the case of an escape gap located on the bottom of the parlour pot, shall be a gauge 199 mm wide, 44 mm high and 100 mm long. year). Currently, there is no information on the frequency of pot loss in the UK; hence it is presently not possible to speculate about the annual death rate of crabs and lob- sters from ghost-fishing in the UK. Financial losses to the industry are potentially high because lobsters command high prices (up to U.S. $12 per kg) whereas brown crab market prices have fluctuated between $0.60 and $1.60 per kg in recent years. However, there is no market for spi- der crab in the UK and only a sporadic overseas market for this species in countries such as Portugal and Spain. All of the initial bait was exhausted after day 27. Pre- sumably, as pot residents die, they act as bait luring more animals into the traps. We suspect that the fish that died would provide a ready source of food for trapped Crus- tacea, an observation that concurs with those in related studies (Kaiser et al., 1996). Recent studies have shown that a similar pattern of capture is observed in lost set nets. Over the first few days, catches decline almost expo- nentially. Then, for the next few weeks, the decaying bod- ies of fishes and Crustacea attract large number of scav- enging crustaceans that also become trapped in the gear. Thereafter, there appears to be a continuous cycle of cap- ture, decay, and attraction for as long as the gear remains intact (Carr et al., 1990; Kaiser et al., 1996). It is interesting to note that ballan wrasse were caught in greater numbers in the traps during the last 65 days of the experiment, perhaps when there were fewer Crustacea in the pots compared with the beginning of the experi- ment. Wrasse seemed to develop wounds on their heads that resulted from their attempts to push through the mesh of the pot. These became severely infected and we believe contributed to their eventual death. Wrasse may become trapped within pots when they seek shelter. Our estimates of capture rate are probably conservative because some animals may escape the traps. Divers found several spider crabs that had been tagged in one pot, but on the subsequent sampling occasion, they were re- corded in a different pot. These animals had escaped one trap only to be captured in another. Similarly, Guillory ( 1993) found that up to 42% of blue crabs ( Callinectes sapi- dus) escaped traps. The “ghost-fishing” potential of pots also varies for different fisheries and pot designs. For ex- ample, Parrish and Kazama (1992) found that the major- ity of Hawaiian spiny lobster ( Palinurus marginatus) and slipper lobster (Scyllarides squammosus ) were able to es- cape traps, whereas parlor-type traps lead to mortalities of 12-25% for American lobster ( Homarus americanus) (Smolowitz, 1978). Catches in pots are also affected by the identity of the initial occupants. For example, Miller and Addison (1995) found that the presence of American lob- sters within a pot deters entry of smaller crab species. Similarly, the presence of recently molted brown crabs within pots deterred entry of conspecifics (Addison, 1995). Potential losses from the brown crab fishery due to ghost-fishing gear could be large. These losses are unde- sirable both from a conservation and economic point of view. Pot fisheries target species with high individual val- ue, hence each loss is expensive. Fishermen in set-net fish- eries have taken their own steps to reduce this phenom- enon leading to a grapnel survey of the seabed on Georges Bank and retrieval of 341 actively fishing ghost nets from 286 tows (Brothers3). In some fisheries in North America, fishermen must fit their pots with escape gaps or escape panels that either biodegrade and fall out of the pot after a certain length of time, or that have degradable escape panel clips. In contrast, there is no UK legislation to force 3 Brothers, G. 1992. Lost or abandoned fishing gear in the Newfoundland aquatic environment. Report of the Symposium on Marine Stewardship in the Northwest Atlantic, Department of Fisheries and Oceans, St. Johns, Newfoundland, Canada. Bullimore et al.: Study of catches in a fleet of "ghost-fishing" pots 253 fishermen to use these conservation measures in conjunc- tion with the traps, with the exception of bylaws in two areas (Table 3). Conservation measures such as these are relatively inexpensive to introduce and would greatly re- duce losses from the fishery of commercially important species, such as lobster and brown crab, as well as species of conservation interest such as ballan wrasse. Acknowledgments The authors gratefully acknowledge the help of David Bray, a local professional fisherman, and the South Wales Seafisheries Committee for the loan of pots. The input of several referees has greatly improved previous versions of this manuscript. This study was funded in part by Euro- pean Commission study contract 94/076. Literature cited Addison, J. T. 1995. Influence of behavioural interactions on lobster dis- tribution and abundance as inferred from pot-caught sam- ples. ICES Mar. Sci. Symp. 199:294-300. Breen, P. A. 1987. Mortality of Dungeness crabs caught by lost traps in the Fraser River Estuary, British Columbia. N. Am. J. Fish. Manage. 7:429-435. Carr, H. A., E. H. Amaral, A.W. Hulbert, and R. Cooper. 1990. Underwater survey of simulated lost demersal and lost commercial gill nets off New England. In Marine debris: sources, impacts and solutions ( J. M. Coe and D. B. Rogers, eds.), p. 171-186. Springer, New York, NY. Dayton, P. K., S. F. Thrush, M. T. Agardy, and R. J. Hofman. 1995. Environmental effects of marine fishing. Aquat. Cons. Mar. Freshwater Ecosyt. 5:205-232. Eisenbud, R. 1985. The pelagic driftnet. Salt Water Sportsman. May :65- 72. Erzini, K., C. Monteiro, J. Ribeiro, M. Santos, M. Caspar, P. Monteiro, and T. Borges. 1997. An experimental study of gill net and trammel net ‘ghost-fishing’ off the Algarve (southern Portugal). Mar. Ecol. Prog. Ser. 158:257-265. Guillory, V. 1993. Ghost fishing by blue crab traps. N. Am. J. Fish. Manage. 13:459-466. Jennings, S., and M. J. Kaiser. 1998. The effects of fishing on marine ecosystems. Adv. Mar. Biol. 34:201-352. Kaiser, M. J., B. A. Bullimore, P B. Newman, K. M. Lock, and S. E. Gilbert. 1996. Catches in ‘ghost-fishing’ set nets. Mar. Ecol. Prog. Ser. 145:11-16. Miller, R. J., and J. T. Addison. 1995. Trapping interactions of crabs and American lobster in laboratory tanks. Can. J. Fish. Aquat. Sci. 52:315-324. Parrish, F. A., and T. K. Kazama. 1992. Evaluation of ghost fishing in the Hawaiian lobster fishery. Fish. Bull. 90:720-725 Paul, J. M., A. J. Paul, and A. Kimker. 1994. Compensatory feeding capacity of two Brachyuran crabs. Tanner and Dungeness, after starvation periods like those encountered in pots. Alaska Fish. Res. Bull. 1:184-187. Smolowitz, R. J. 1978. Trap design and ghost, fishing: an overview. Mar. Fish. Rev. 40:2-8. SPSS, Inc. 1998. SPSS user’s quide, version 9.0.0. SPSS, Inc, Chicago, IL. Stevens, B. G. 1996. Crab bycatch in pot fisheries. In Solving bycatch: considerations for today and tomorrow, p. 151-158. Alaska Sea Grant Program Report 96-03. Univ. Alaska Fair- banks, Juneau, AK. 254 Abstract-Gray snapper, Lutjanus gri- seus , were sampled from recreational headboat and commercial landings along the east coast of Florida, 1994-97. Fish were weighed (g) and measured (total length, TL, in mm), and sagittal oto- liths were removed for aging. Marginal increment analysis on sectioned oto- liths (n=1243) confirmed annulus for- mation in June and July. The oldest fish examined was 24 years old and mea- sured 760 mm TL. Weight-length rela- tions were not significantly different by sex. Weight-length relations were sig- nificantly different (F= 39. 198, P<0.001, df=10,705) for fish measured from the headboat survey from 1982-97 between north Florida /t + 0.4634 x log10T, where Lm = the asymptotic length; K = the Brody growth coefficient from the von Ber- talanffy ( 1938) growth equation; and T = the mean annual seawater temperature (°C). I derived the latter from sea surface temperature readings from buoys operated by NOAA’s National Oceanographic Data Center during 1998. Finally, I estimated M by using the regression of Ralston ( 1987); M = 0.0189 + 2.06 x K, where K - the Brody growth coefficient. Observed ages at length for all years combined were used to develop age-length keys (ALK) for each area (Rick- er, 1975). I assigned aged fish (my samples) to 25-mm-TL intervals and calculated age distribution (as a percentage) for each size interval. Area-specific age-length keys were used to convert length frequencies from each area and fishery, weighted by the corresponding annual landings, into age frequencies by assigning ages to unaged fish from the length frequencies. Length-frequency data and annu- al landings data were acquired from the South Atlantic headboat survey, the MRFSS, and the Trip Interview Pro- gram (TIP)5 survey. Total instantaneous mortality rate, Z, was estimated by the absolute value of the slope of the descending right limb of the plot of logt, age frequency on age (catch curves) (Beverton and Holt, 1957). Only fully recruited ages (age groups to the right of the top of the dome of the curve) were used to estimate Z because the age group at the top of the catch dome may not be fully vulnerable to the fishing gear (Everhart et al., 1975). Results Age determination and validation of annuli A total of 98% (1243 of 1260) of gray snapper sampled had legible cross-sectioned otoliths. Opaque rings were distinct and easily counted. Otolith radius was correlated with fish length across all ages: North Florida: TL = ( 10.02 x OR) - 52.98 (rI 2=0.90, n=519), South Florida: TL = (9.90 x OR) - 91.68 (r2=0.78, n= 724), Areas combined: TL = (11.05 x OR) - 130.32 (r2=0.89, n=1243). Marginal increment analyses showed one minimum per year in June, validating the annual periodicity of otolith in- crements (Fig. 1A). The monthly percentage of fish with a marginal increment equal to zero (Fig. IB) showed a single maximum and provided further evidence that annulus for- mation occurred yearly in June or July. To satisfy Beamish and McFarlane’s (1983) assertion that individual ages need to be validated, I analyzed marginal increments by age for 5 Trip Interview Program. 1998. Administered by Southeast Fisheries Science Center, NMFS, NOAA, 75 Virginia Beach Dr., Miami, FL 33149 Burton: Age, growth, and mortality of Lutjonus griseus 257 CT> CO E c CO Month of capture Figure 2 Mean monthly marginal increments of gray snapper otoliths by age, areas pooled. all areas pooled. Annulus formation occurred in the summer months, with minima in June, for ages 2-9 (Fig. 2). Sample size was inadequate for analyses of older age classes. Weight-length relationship The relationship between W (kg) and TL (mm) for all gray snapper measured by the headboat survey from 1982 to 1997 was estimated by using a direct nonlinear fit with SAS PROC NLIN and the Marquardt algorithm software (SAS Institute, Inc., 1987). Examination of the residuals indicated an additive error term, and I concluded that the nonlinear fit was more appropriate than a linearized log- transform fit of the data. Area-specific regressions (see Table 1 for parameters and statistics) were North Florida: W = 8.4 x 10-9 TL3 08, South Florida: W = 5.4 x 10-9 TL315, and Pooled areas: W = 7.22 x 10~9 TL311. In addition, nonlinear regressions by sex, derived from the subset of aging samples for which I had sex information, were Males: W = 7.13 x 1CF9 TL3 n, and Females: W = 6.95 x 1CF9 TL3 W. Although regression coefficients were significantly differ- ent by area (f=-8.024, PcO.OOl, df=10,704, and f=8.159, PcO.OOl, df=10,704 for intercept and slope, respectively), predicted weights for gray snapper at 300 mm, 400 mm, and 500 mm TL from north and south Florida for the respective length-weight regressions were similar: 0.35 vs. 0.34, 0.87 vs. 0.85, and 1.73 vs. 1.71 (kg). This result is likely due to the ability to detect statistically significant differences with an extremely large sample size when the actual differences may be very small and mean little biologically. Regression coefficients by sex were not significantly different, as indi- cated by overlapping 95% confidence intervals. 258 Fishery Bulletin 99(2) Table 1 Parameters and associated statistics for weight-length relationships of gray snapper by area and sex. SE = standard error; MSE = mean squared error. Parameter (SE) North Florida South Florida Areas pooled Males Females a 8.4 x 1(L9 5.45 x 10 -9 7.22 x 10-9 7.13 x 10-9 6.95 x 10-9 (SE) (6.25 x 10-10) (2.76 x 10“10) (2.8 x 10±10) (1.26 x 10-9) (1.88 x 10 9) b 3.08 3.15 3.11 3.106 3.097 (SE) (0.012) (0.008) (0.006) (0.029) (0.041) n 4034 6670 10,704 262 212 MSE 0.078 0.020 0.042 0.019 0.034 Table 2 Mean observed total lengths at age of gray snapper by area. Age (yr) North Florida South Florida n Mean T1 ±SE (mm) Range (mm) n Mean TL ±SE (mm) Range (mm) 1 13 218 ±24 181-255 — 2 7 220 ±24 185-257 20 284 ±36 167-327 3 17 347 ±43 212-415 138 302 ±33 182-397 4 69 388 ±42 307-505 260 325 ±29 231-447 5 138 432 ±42 332-555 177 358 ±45 265-525 6 75 467 ±46 347-565 81 397 ±60 300-644 7 71 525 ±43 370-615 32 474 ±46 390-578 8 26 549 ±40 480-615 11 468 ±60 342-542 9 36 589 ±42 525-702 3 417 ±141 330-581 10 23 595 ±31 537-657 3 414 ±86 317-481 11 14 608 ±32 545-645 1 452 12 9 623 ±30 580-685 — 13 8 660 ±35 600-705 2 501 ±166 383-618 14 7 652 ±43 600-725 — 15 3 682 ±34 660-722 1 504 504-504 16 1 677 17 1 635 18 2 649 ±68 600-697 19 1 707 20 1 630 21 2 710 ±25 692-727 22 — 23 2 748 ±18 735-760 24 2 740 ±28 720-760 Growth Mean observed lengths at age were larger for fish from north Florida than those for fish from south Florida for all ages except age 2 (Table 2), but sample size may have affected estimates for age-2 fish. North Florida fish ranged from 181 mm at age 1 to 760 mm at age 24. South Florida fish ranged from 167 mm TL at age 2 to 618 mm at age 13. The oldest fish from south Florida was age 15 and measured 504 mm. Age-1 fish were difficult to acquire in south Florida because of minimum-size catch regula- tions (305 mm TL in federal waters, and 250 mm TL in Florida waters). All age-1 fish from north Florida were collected through fishery-independent hook-and-line sam- pling. In both areas, most fish attain the federal minimum size limit (305 mm TL) by age 3. Burton: Age, growth, and mortality of Lutjanus griseus 259 cn c 0) To ,o Figure 3 Theoretical growth curves for gray snapper by (A) area and < B ) sex, areas pooled. Back-calculated sizes at age of gray snapper by area are shown in Tables 3 and 4. Tbe mean total back- calculated lengths for gray snapper from north Florida, back-calculated to the last annulus, were larger than those for fish from south Florida for all ages except age 2. Given the ob- vious differences in size at age be- tween the two areas, it seemed in- appropriate to pool the data for the purposes of analyzing growth. The linear regressions of the mea- surements to the first and second an- nuli on age for headboat and commer- cial specimens from north Florida were significantly different from zero (77=519, P=0.0001, r2=0.04; and 77 = 506, P=0.0012, r2=0.02). Although this result indicates the presence of Lee’s phenomenon for samples from these fisheries (Potts et al., 1998), the model explains very little of the variation, so that Lee’s phenomenon was weak and likely masked by other environmental variables. The lineal- regressions of the measurements to the first and second annuli on age for headboat and commercial specimens from south Florida were not signif- icantly different from zero (77=720, P= 0.23, r2=0.002 and 77=720, P=0.81, 7-2=0.0001), indicating that size-se- lective mortality was not detected. The von Bertalanffy ( 1938) growth equations, fitted to back-calculated lengths at age for the last annulus (Fig. 3A), were North Florida: Lt = 717 (1 _ e-° 17 o + 0.0251), ancj South Florida: Lt= 625 (1 -e-°-13,' + 133)). Parameters and associated statistics for these equations as well as for equations fitted to the subset of data by sex, are listed in Table 5. Esti- mates of growth parameters were not significantly different between sexes [Hotelling's T 2 test, F{ 0.025,3, ~>= 1.0 (Bernard, 1981)] (Fig. 3B). How- ever, estimates of growth parameters were significantly different between areas (Pl0025 3r =84.1). Mortality Estimates of the instantaneous rate of natural mortality (M) varied considerably by method but were similar across areas for a given method. Estimates of M assume a con- stant natural mortality and so apply to all ages of fish. Hoenig’s (1983) longevity-mortality relation returned the lowest estimates, 0.18 and 0.29 (north and south Florida), and his equation adjusted for sample size returned similar values between areas, 0.33 and 0.35. Pauly’s (1980) equa- tion, with growth parameters and mean seawater temper- 260 Fishery Bulletin 99(2) Burton: Age, growth, and mortality of Lut/anus griseus 261 atures of 25.0°C and 26.1°C for north and south Florida, estimated M at 0.43 and 0.38 for the two areas. The regres- sion method of Ralston (1987) estimated M = 0.37 and M = 0.29 for north Florida and south Florida, respectively. I used a variety of estimation methods to give the reader a sense of the variation associated with estimating M, a parameter we often know little about but which is a very important variable in stock assessments. Estimates of total mortality derived from catch curves differed substantially by area (Fig. 4). Estimates of Z for gray snapper from north Florida averaged 0.35 during 1986-97 for all fisheries combined. Gray snapper were ful- ly recruited to the headboat fishery between ages 5 and 6, to the commercial fishery between ages 7 and 8, and to the private recreational fishery between ages 4 and 5. The av- erage value of Z for 1986-97 for gray snapper from south Florida was 0.94, almost three times that of north Florida. Full recruitment occurs between ages 4 and 5 in all fisher- ies in south Florida. Discussion Differences in growth rates and size at age between areas were great enough to argue against pooling the data for growth analysis. However, the pooled data set does provide the opportunity for comparing my results against a previ- ous similar study for the purpose of validating my aging estimates. Back-calculated lengths at age were in close agreement with lengths estimated previously by Manooch and Matheson (1981). Estimated lengths (mm) for ages 1, 5, 10, and 15 with measurements to the last annulus, were 87 vs. 89, 370 vs. 370, 559 vs. 556, and 630 vs 680 (pres- ent study vs. previous study). The close agreement in size Table 4 Back-calculated total lengths (mm) of gray snapper aged by sectioned otoliths for South Florida. Annulus number Obs. age n 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 1 2 20 100 227 3 136 89 195 272 4 258 90 184 253 305 5 174 94 190 254 303 341 6 81 97 192 256 305 346 383 7 32 96 212 283 337 382 424 461 8 10 99 203 262 313 355 391 422 451 9 3 71 179 248 286 317 342 367 392 414 10 3 77 164 210 253 281 311 336 358 380 404 11 1 70 152 197 243 279 316 343 370 397 425 443 12 — 13 2 84 198 244 280 307 335 358 385 408 431 454 473 491 14 — 15 1 73 138 213 260 288 307 335 354 382 401 420 438 457 475 495 No. of calculations 721 721 701 565 307 133 52 20 10 7 4 3 3 1 1 Weighted means Annual mean 92 191 258 306 346 389 432 412 398 414 443 461 480 475 495 growth increment 92 99 67 48 40 43 43 (20) (14) 16 29 18 19 (5) 20 Table 5 Von Bertalanffy growth parameters and associated statistics for gray snapper by area and sex, 1994-1997. Data set L«, SE (LJ 95% Cl K SE (K) 95% Cl to SE (tQ) 95% Cl n North Florida 716 11.33 693-738 0.17 0.01 0.16-0.19 -0.001 0.11 -0.22-0.22 520 South Florida 625 56.33 515-736 0.13 0.02 0.08-0.18 -1.33 0.41 -2.12-0.53 721 Males 697 22.93 652-742 0.18 0.017 0.15-0.21 0.49 0.21 0.08-0.89 339 Females 768 35.7 697-838 0.15 0.017 0.1 -0.18 0.16 0.26 -0.36-0.67 272 262 Fishery Bulletin 99(2) at age from these two studies, both of which used otoliths and comprised samples from the same geographic area, validates the aging of gray snapper determined in the present study. Back-calculated lengths at age of gray snapper determined from scales by Starck and Schro- eder ( 1970) were generally smaller than those determined from otolith sections, reasons for which are unclear. My samples, as well as those of Manooch and Matheson (1981), were from Florida’s east coast, whereas Starck and Schroeder (1970) used fish from only one loca- tion in the Florida Keys. A geographic bias may account for the difference in ages between the two studies. Marginal increment analysis demonstrated that gray snapper ages 2-9 deposit one annu- lus per year, in June. These ages account for the majority of age classes in the fishery; thus I posit that the critical evaluations stressed by van Oosten (1929) and Beamish and Mc- Farlane (1983) were met. Annual deposition of growth increments on otoliths of gray snapper was not val- idated by previous investigators and thus was an essential part of my study. I constructed area-specific age-length keys using Sebas- tian Inlet, FL (27.8°N latitude), as the north-south divid- ing line. In addition to being the traditional division point for sample coverage for the south Atlantic headboat sur- vey, it also approximates the nearshore-offshore break in distribution of reef habitat. I did not have enough data to construct annual ALKs as recommended by Ricker (1975) and Westrheim and Ricker ( 1978), and my decision to pro- duce area-specific ALKs was based on a priori informa- tion gained as a port agent in Florida from 1982 to 1987. Gray snapper from headboats in north Florida were larger than gray snapper from headboats in south Florida. More- over, Manooch and Matheson (1981) found differences in length-frequency distributions and estimates of total mortality by area (smaller fish and higher mortality, Z, in south Florida). The difference in growth of gray snapper between north and south Florida is readily apparent . Mean observed and back-calculated sizes at age were largest for fish from north Florida, and these fish achieved a much greater maximum size and age than did their south Florida coun- terparts; significant differences in theoretical maximum size between areas were observed (Fig. 3A). Johnson et al. ( 1994) found similar results for gray snapper collected from the east coast of Florida and the Gulf of Mexico. They compared back-calculated size at age for fish cap- tured from north and south of 27°N latitude and conclud- ed that fish from the northern region were significantly larger than fish from the southern region for ages 1-13. Their latitude was proximal to the headboat survey divi- sional line used in the present study. The differences observed in size at age become greater in older ages of gray snapper, although estimates for south Florida are affected by small sample sizes (few fish >age 8). The small sample size of older fish in south Florida probably reflect a lack of abundance rather than sam- pling deficiencies. Commercial and headboat data (Fig. 5) showed larger and presumably older fish taken from north Florida than from south Florida. The headboat fish- ery modal length intervals were 400-424 mm TL for north Florida and 300-324 mm TL for south Florida. Modal val- ues of commercial length frequencies showed a greater dif- ference between areas: 550-574 mm TL for north Florida compared with 325-349 TL mm for south Florida. Given the efficiency of modern fishing fleets (all using hook-and- line gear), this finding is strong evidence that the south Florida population of gray snapper has a truncated size distribution. Manooch and Matheson (1981) found a sim- ilar disparity in size distribution of gray snapper from headboats from north and south Florida, with respective modes at 450-499 mm TL versus 300-349 mm TL. One possible explanation for the lack of older, larger fish in south Florida is the much greater fishing pressure there. Manooch and Matheson (1981) estimated the in- stantaneous rate of fishing mortality (F) to be 0.17 and 0.38 for north and south Florida, respectively. I estimated F = 0.16 and F = 0.66 for the respective areas for all fish- eries combined, using Hoenig’s (1983) estimates of M - 0.18 and M = 0.29 and the area-specific estimates of Z (F=Z-M). The demography and geography of the Florida peninsu- la probably affect fishing pressure on gray snapper. South Florida is more densely populated than north Florida and thus has many more potential anglers. Specifically, an- glers fished an average total of 186,687 days from south Florida headboats annually from 1982 to 1997, compared with an annual average total of 82,325 days from north Florida headboats. Shorter distance to the fishing grounds in south Florida (5-8 km) than to those in north Florida (40-50 km) also leads to increased exploitation. Increased pressure on younger inshore fish could lead to growth overfishing, whereas easier access to the mature adults offshore may contribute to recruitment overfishing. The Burton: Age, growth, and mortality of Lut/anus griseus 263 £ LL 25 20 15 10 oil I Commercial I NFL DSFL All 150 225 300 375 450 525 600 675 750 25 20 15 10 0 +- Headboat I NFL DSFL LIIli 1 1 ■ 150 225 300 375 450 525 600 675 750 Length interval starting point Figure 5 Length frequencies for gray snapper measured by fishery and area, 1986-97. latter scenario would be aggravated by the sus- pected tendency of gray snapper to undertake spawning migrations to offshore reefs (Domeier et ah, 1997). This behavior, resulting in higher fish densities on these reefs, would increase their vulnerability to fishery harvest during summer months, a time of maximum fishing effort due to favorable weather conditions. Another effect of long-term heavy fishing pres- sure could be a genetic shift in growth charac- teristics of the fish. Size-selective mortality could result in slower growing individuals in the popu- lation. Buxton ( 1993) found growth rates for Chry- solephus cristiceps (Sparidae) to be significantly lower in exploited than in protected populations. Zhao et al. (1997) found that mean back-calculat- ed lengths at age for vermilion snapper, Rhom- boplites aurorubens, declined from 1979 to 1987, concluding that this result was a true change in growth, posssibly caused by overfishing. Har- ris and McGovern (1997) attributed decreases in growth and maturity rates of red porgy ( Pagrus pagrus ), over time, to sustained heavy fishing pressure. Zhao and McGovern (1997) found sim- ilar decreases in size and age at maturity for vermilion snapper over time, attributing the de- clines to increasing fishing pressure. Other inves- tigators, however, have hypothesized that these results were caused by size selectivity charac- teristics of different gears used during different sampling periods (e.g. Potts et al., 1998). An alternate explanation for the lack of larger, older gray snapper in south Florida is emigration. An argument might be made that the inshore to offshore spawning migrations (Domeier et al., 1997) previously mentioned might take them be- yond the range of the fisheries and make gray snapper less vulnerable to fishing gear in the south Florida area. This hypothesis seems highly unlikely given the range and technology of modern fisher- ies. Moreover, most reef fishes are thought to exhibit a sed- entary lifestyle as adults, staying close to the same general reef area (Ehrlich, 1975; Heemstra and Randall, 1993; Sam- oilys, 1997). However, Moe (1969) found that red grouper (. Epinephelus morio ) showed some migratory behavior — 22 individuals moved 29 km in 50 days, and another indi- vidual moved 76 km. Most of these movements were from inshore to offshore. A comprehensive tagging study could be designed to address the question of whether significant numbers of gray snapper migrate out of the south Florida area. Management of gray snapper is the responsibility of the South Atlantic Fishery Management Council (SAFMC), whose current strategy is to manage this species, and most others in the snapper-grouper complex, as a single stock throughout their management area. It is very unlikely that gray snapper in north and south Florida are geneti- cally distinct because of the northward flow of the Gulf Stream and the resulting widespread distribution of prog- eny. However, this study in conjunction with that of Ma- nooch and Matheson (1981) strongly suggests that gray snapper have been exploited at higher rates of F in south Florida than in north Florida for at least two decades. As a result, gray snapper reach a smaller maximum size and younger maximum age in the population, as well as smaller sizes at most ages. These biological features carry implications for overall population health because fecun- dity is usually proportional to size or age, or to both. Fish- ery managers attempting to assess stocks of gray snapper should perform area-specific analyses in order to manage this species in the most effective manner. Given the re- sults of this study, it seems that managing gray snapper as a stock unit could worsen overfishing conditions in south Florida. Acknowledgments I gratefully acknowledge the many port samplers whose efforts made this study possible: headboat samplers Dan Theisen, Pamela Washnock, and Peggy Kirwin, NMFS 264 Fishery Bulletin 99(2) Beaufort Laboratory; and commercial port samplers Charles Schaefer, Tim Brandt, Claudia Dennis, and Renee Roman, NMFS Miami Laboratory. Charles S. Manooch III and Jennifer Potts, NMFS Beaufort, provided guidance in the techniques of aging otolith sections. Doug Vaughan and Jim Waters, NMFS Beaufort, and Potts provided valu- able advice on statistical analyses, and. C.S. Manooch, J. Potts, and Joseph Smith, NMFS Beaufort, as well as three anonymous reviewers, provided constructive critical reviews of the manuscript. Literature cited Beamish, R. J., and G. A. McFarlane. 1983. The forgotten requirement for age validation in fish- eries biology. Trans. Am. Fish. Soc. 112:735-743. Bernard, D. R. 1981 . Multivariate analysis as a means of comparing growth in fish. Can. J. Fish. Aquat. Sci. 38:233-236. Beverton, R. J. H., and S. J. Holt. 1957. On the dynamics of exploited fish populations. Fish. Invest, ser. II Mar. fish., G. B. Minist. Agrie. Fish. Food. 19: 1-533. Buxton, C. D. 1993. Life-history changes in exploited reef fishes on the east coast of South Africa. Environ. Biol. Fishes 36:47-63. Croker, R. A. 1962. Growth and food of the gray snapper, Lutjanus gri- seus, in Everglades National Park. Trans. Am. Fish. Soc. 91:379-383. Domeier, M. L., C. Koenig, and F. Coleman. 1997. Reproductive biology of the gray snapper ( Lutjanus gri- seus) with notes on spawning for other western Atlantic snap- pers. In Biology, fisheries and culture of tropical snappers and groupers ( Arreguin-Sanchez, F., J. L. Munro, M. C. Balgos, and D. Pauly, eds.), p. 189-201. ICLARM Conf. Proc. 48. Ehrlich, P. R. 1975. The population biology of coral reef fishes. Ann. Rev. Ecol. Syst. 6:211-247. Everhart, W. II. , A. W. Eipper, and W. L). Youngs. 1975. Principles of Fishery Science. Cornell Lhiiv. Press, Ithaca, NY, 288 p. Francis, R. I. C. C. 1990. Back-calculation of fish length: a critical review. J. Fish. Biol. 36:883-902. Harris, P. J., and J. C. McGovern. 1997. Changes in the life history of red porgy, Pagrus pagrus, from the southeastern United States, 1972-1994. Fish. Bull. 95(4):732-747. Heemstra, P. C., and J. E. Randall. 1993. FAO Species catalogue. Vol. 16: Groupers of the world (family Serranidae, subfamily Epinephelinae): an annotated and illustrated catalogue of the grouper, rockcod, hind, coral grouper and lyretail species known to date. FAO, Rome, 382 p. Hettler, W. F.. and D. L. Barker. 1993. Distribution and abundance of larval fishes at two North Carolina inlets. Estuarine, Coast Shelf Sci. 37:161- 173. Hoenig, J. M. 1983. Empirical use of longevity data to estimate mortality rates. Fish. Bull. 82:898-903. Johnson, A. G., L. A. Collins, and C. P. Keim. 1994. Age-size structure of gray snapper from the south- eastern United States: a comparison of two methods of back-calculating size at age from otolith data. Proc. Annu. Conf. Southeast. Assoc. Fish Wildl. Agencies 48:592-600. Manooch, C. S., III. 1984. Fisherman’s guide to the fishes of the southeastern United States. North Carolina State Museum of Natural History, Raleigh, NC, 362 p. Manooch, C. S., Ill, and R. H. Matheson III. 1981. Age, growth and mortality of gray snapper collected from Florida waters. Proc. Annu. Conf. Southeast. Assoc. Fish Wildl. Agencies 35:331-344. Miller, G. C., and W. J. Richards. 1980. Reef fish habitat, faunal assemblages, and factors de- termining distributions in the South Atlantic Bight. Proc. Gulf Caribb. Fish. Inst. 32:114-130. Moe, M. A. 1969. Biology of the red grouper, Epinephelus mono (Valen- ciennes), from theeastern Gulf of Mexico. Fla. Dep. Natur. Resour. Mar. Res. Lab. Prof. Pap. Ser. 10, 95 p. Pauly, D. 1980. On the interrelationships between natural mortality, growth parameters, and mean environmental temperature in 175 fish stocks. J. Cons. Int. Explor. Mer 39:175-192. Potts, J. C., and C. S. Manooch III. 1995. Age and growth of red hind and rock hind collected from North Carolina through the Dry Tortugas, Florida. Bull. Mar. Sci. 56(3):784-794. Potts, J. C., C. S. Manooch, III, and D. S. Vaughan. 1998. Age and growth of vermilion snapper from the southeastern United States. Trans. Am. Fish. Soc. 127: 787-795. Ralston, S. 1987. Mortality rates of snappers and groupers. In Trop- ical snappers and groupers: biology and fisheries man- agement (J. J. Polovina and S. Ralston, eds.), p. 375-404. Westview Press, Boulder, CO. Ricker, W. E. 1975. Computation and interpretation of biological statis- tics of fish populations. Bull. Fish. Res. Board Canada 191, 382 p. Robins, C. R., G. C. Ray, J. Douglas, and R. Freund. 1986. A field guide to Atlantic coast fishes. Houghton Mif- flin Company, Boston, MA, 354 p. Rutherford, E. S., E. B. Thue, and D. G. Buker. 1983. Population structure, food habits, and spawning activity of gray snapper, Lutjanus griseus , in Everglades National Park. South Florida Research Center Report SFRC-83/02. National Park Service, Everglades National Park, Homestead, Florida, 41 p. Samoilys, M. A. 1997. Movement in a large predatory fish: coral trout, Plec- tropomus leopardus (Pisces: Serranidae), on Heron Reef, Australia. Coral Reefs 16:151-158. SAS Institute, Inc. 1987. SAS/STAT guide for personal computers, vers. 6 eel. SAS Institute, Cary, NC, 1028 p. Starck, W. A. II, and R. E. Schroeder. 1970. Investigations on the gray snapper, Lutjanus griseus. Stud. Trop. Oceanogr. 10, Univ. Miami Press, Coral Gables, FL, 224 p. Sumner, F. B., R. C. Osburn, and L. J. Cole. 1911. A biological survey of the waters of Woods Hole. Bull. U. S. Bur. Fish. 31:549-794. van Oosten, J. 1929. Life history of the lake herring ( Leucicthys artedi LeSeuer), of Lake Huron as revealed by its scales with a Burton: Age, growth, and mortality of Lut/anus gnseus 265 critique of the scale method. Bull. U.S. Bur. Fish. 44:265- 428. Vaughan, D. S., and M. L. Burton. 1994. Estimation of von Bertalanffy growth parameters in the presence of size-selective mortality: a simulated exam- ple with red grouper. Trans. Am. Fish. Soc 123:1-8. von Bertalanffy, L. 1938. A quantitative theory of organic growth. Hum. Biol. 10:181-243. Westrheim, S. J., and W. E. Ricker. 1978. Bias in using an age-length key to estimate age-fre- quency distributions. J. Fish. Res. Board Canada 35:184— 189. Zhao, B., and J. C. McGovern. 1997. Temporal variation in sexual maturity and gear- specific sex ratio of the vermilion snapper, Rhomboplites aurorubens, in the South Atlantic Bight. Fish. Bull. 95:837-848. Zhao, B., J. C. McGovern, and P. J. Harris. 1997. Age, growth, and temporal change in size at age of the vermilion snapper from the South Atlantic Bight. Trans. Am. Fish. Soc. 126:181-93. 266 Abstract— The feeding habits of the Mediterranean spiderfish, Bathypterois mediterraneus , the most abundant fish below 1000 m on the deep slope of the Catalan Sea (western Mediterranean), were studied. Samples were obtained at depths between 1000 and 2250 m. Diet was analyzed for two different size classes (immature and mature speci- mens) and three different bathymetric strata. The most important food items found were benthopelagic planktonic calanoid copepods. In juveniles from 1800 to 2250 m, benthic tanaidaceans were secondary, whereas in the adults, mysids were secondary. At other depths, there were no secondary prey: calanoid copepods were consumed almost exclu- sively; other items were very scarce. Adults ingest larger amounts and sizes of prey than juveniles. The scarcity of resources below 1200-1400 m diversi- fied the diet, although it still primarily consisted of elements from the ben- thopelagic plankton. Manuscript accepted 12 September 2000. Fish. Bull. 99:266-274 (2001 ). Feeding ecology of the Mediterranean spiderfish, Bathypterois mediterraneus (Pisces: Chlorophthalmidae), on the western Mediterranean slope Maite Carrasson Jesus Matallanas Dpto. Biologia Animal, Biologia Vegetal y Ecologla Universidad Autonoma de Barcelona Bellaterra, E-08193 Barcelona, Spain E-mail address (for M Carrasson): Maite.Carrasson@uab.es The genus Bathypterois Gunther, 1878 comprises a small group of benthic chlo- rophthalmid species adapted for life in the deep sea. The genus is circum- global at temperate latitudes (Sulak, 1984a). Two species of Bathypterois are found in the Mediterranean: Bathypter- ois dubius and Bathypterois mediterra- neus (Bauchot, 1962). The presence of a third, Bathypterois gi'allator, based on underwater photographs, has been suggested by Sulak (1984a). The Medi- terranean spiderfish, Bathypterois med- iterraneus, is the only endemic species of the genus, and undoubtedly the most abundant in the Mediterranean (Sulak, 1977; Bauchot, 1987). Some species of the genus Bathyp- terois, such as Bathypterois dubius, Bathypterois grallator, Bathypterois atricoloi', and Bathypterois viridensis, have been observed from submersibles (Church, 1971; Heezen and Hollister, 1971; Saldanha, 1977; Jones and Su- lak, 1990; Chave and Mundy, 1994; Chave and Malahoff, 1998). The fish are benthic, rest on the bottom, and touch the sediment with their longest pelvic and caudal rays (Heezen and Hollister, 1971; Saldanha, 1977), while the long pectoral rays are directed for- wards over the head (Saldanha, 1977). Pectoral rays are likely used as sensory devices to detect the presence of plank- tonic prey, by both direct contact and chemoreception (Sulak, 1977). According to Sulak ( 1984a ), Bathypter- ois mediterraneus is benthic on the con- tinental slope and rise, at 260-2800 m. However, in the Catalan Sea (western Mediterranean), the species is restricted in its bathymetric distribution to depths greater than 748 nr (Stefanescu et al., 1992a, 1994; Morales-Nin et al., 1996). Of the 31 different species of deep-wa- ter fish in the Catalan Sea (Stefanescu et al., 1992a), Bathypterois mediterra- neus is subdominant between 1000 and 1 500 m, and dominant below this depth. At 1000-2250 m Bathypterois mediter- raneus is the fourth in importance for biomass of all fish (Stefanescu et al., 1992a). This species aggregates (Stefan- escu et al., 1992a; Morales-Nin et al., 1996), contrary to Sulak’s ( 1984a) obser- vation that it is usually solitary, but may aggregate occasionally. Knowledge of the biology of deep-sea fauna in the Mediterranean is limited. The feeding habits of some deep-water fish and decapod species in the Cat- alan Sea have been studied recently (Cartes and Sarda, 1989; Carrasson et al., 1992; Cartes and Abello, 1992; Cartes, 1993a, 1993b, 1993c). Informa- tion on Bathypterois mediterraneus bi- ology is limited and fragmentary. Data regarding growth and depth-size trends are given by Morales-Nin (1990), Ste- fanescu et al. (1992b), and Morales-Nin et al. (1996). The morphological and morphometric characteristics of the al- imentary tract and some generic data on the diet of this species have been reported by Carrasson and Matallanas (1990, 1994). Carrasson and Matallanas: Feeding ecology of Bathypterois mediterraneus 267 Sampling data for Bathypterois mediterraneus of specimens sampled. Table 1 in the present study. N lat = North latitude; E long = East longitude, n = number Station Date Depth (m) (initial final) Final situation n N lat E long BII-4 Jul 29 1987 1432 1419 40 44.7 1 52.6 2 BII-5 Jul 30 1987 1753 1715 40 25.4 1 56.9 60 BII-6 Jul 31 1987 1287 1329 40 54.7 2 11.5 55 BII-8 Aug 1 1987 1295 1357 41 02.6 2 27.8 59 BIII-3 Jun 25 1988 1774 1783 40 18.5 1 57.2 33 BIII-4 Jun 26 1988 2163 2039 40 37.7 3 06.2 16 BIII-5 Jun 26 1988 2256 2239 40 32.3 3 44.7 80 The object of this paper is to provide new and detailed information on the diet of Bathypterois mediterraneus from depths of 1000-2250 m in the western Mediterra- nean. Its feeding habits are analyzed in different bathy- metric strata and within groups of immature and adult specimens. The influence of these factors on diet, and pat- terns of dietary overlap, are also discussed. Materials and methods Samples were collected from the continental slope (1000-2250 m) of the Catalan Sea (western Mediterra- nean), during two cruises (BATHOS II— III ), on board the RV Garcia del Cid (Table 1), with a semi-balloon otter- trawl (OTSB14) towed from a single warp (cf. Merret and Marshall, 1981). All specimens were fixed in 10% formalin immediately after capture. Once in the laboratory, they were measured (standard length: SL, to the nearest millimeter) and dis- sected to analyze the gut contents. A total of 305 speci- mens were dissected to examine feeding activity. The linear and poorly differentiated stomach of Bathyp- terois mediterraneus is almost always found empty; prey are usually found in the intestine. Therefore, intestinal contents were analyzed, which makes identifying prey more difficult, because of advanced digestion. Food items were identified to the lowest taxonomic lev- el possible. Numbers and weights were registered to the nearest 0.1 mg, after items were dried with blotting paper to remove surface moisture. The quantitative importance of each prey group in the diet was determined by the index of relative importance (IRI) (Pinkas et ah, 1971), defined as IRI = %F(%N + %V), where %F = frequency of occurrence of the food item; %N = numerical percentage of a food item in the stomachs; and %V = percentage by volume of the food item in the stomachs (Hureau, 1970). In our study, weight was used (%W) instead of volume (%M). This modified index has been expressed as %IRI - (IRI/URI) x 100 (Rosecchi and Nouaze, 1987). To analyze the diet of Bathypterois mediterraneus, indi- viduals from all trawls were grouped according to capture depth (three bathymetric strata: 1000-1425 m; 1425-1800 m; and 1800-2250 m) and size of individuals (two catego- ries: immature or size 1, standard length <113 mm; and mature or size 2, standard length >113 mm). The %IRI of the main prey items was determined for each of the six combinations of depth and size by pooling diet data from the individuals included in each combination. The affinity of these six combinations was computed by using a hier- archical analysis (weighted-pair groups methods analysis, WPGMA). Trophic diversity (H') was calculated, in terms of mean %W of prey items, by using the Shannon index. Degree of overlap in the diet of Bathypterois mediterraneus, by different sizes and bathymetric strata, was determined, based on mean %W results, by using the quantitative Schoener index (Schoener, 1974). Results Of the 305 specimens of Bathypterois mediterraneus ana- lyzed, 23 had an empty gut. Forty-nine categories of prey items were identified from the 282 guts containing food (Table 2). Calanoid copepods were the most numeri- cally abundant prey (%N=81. 1 1%), and mysids were the most abundant by mass (%W=37.59%). Calanoid cope- pods (%IRI=91. 63), according to Wishner ( 1980) and Smith ( 1982 ), are planktonic elements of the benthopelagic fauna. Benthic fauna are accidental prey. From the cluster analysis of the size and depth combina- tions, we identified four groups (Fig. 1): juveniles collected at depths of 1800-2250 m (group A), adults collected at 268 Fishery Bulletin 99(2) this depth (group B), adults and juveniles from other two depth categories (groups C and D). Group A, juveniles (size 1) from 1800-2250 m In 68 specimens analyzed, benthopelagic calanoid cope- pods (Table 2) were the most important prey {%IRI= 56.26). Benthic tanaidaceans were secondary (Fig. 2A), despite having a low overall weight, because they were frequently captured ( W/specbnen=0.0Q01 g). Amphipods were the most abundant accidental prey item (%7f?/=15.27) owing mainly to the weight of Rhachotropis sp. and other amphi- pods (Fig. 2A). Group B, adults (size 2) from 1800-2250 m Calanoid copepods predominated in the 25 adult specimens analyzed (%1RI- 79.52), whereas suprabenthic mysids were Figure 1 Dendrogram of disimilarity between diets of different groups (bathymetric and ontogenetic) of Bathypterois mediterraneus . l=immature (size 1) individuals of 1000-1425 m depth; 2 = mature (size 2) individuals of 1000-1425 m depth; 3 = immature (size 1) individuals of 1425-1800 m depth; 4 = mature (size 2) individuals of 1425-1800 nr depth; 5 = immature (size 1) individuals of 1800-2250 m depth; 6 = mature (size 2) individuals of 1800-2250 m depth. Groups identified: A = 5; B = 6; C = 3, 4; and IJ = 1,2. secondary because of their high weight (Fig. 2B). Benthic cumaceans were the major accidental prey (%IRI= 9.6) because of high numerical abundance (Fig. 2B). Group C, 1425-1800 m In 95 specimens analyzed, the most important prey were benthopelagic calanoid copepods; their high frequency and abundance (Fig. 3) resulted in a %IRI of 95.49%. Incidence of other prey was minimal, %IRI <1.5. Group D, 1000-1425 m In 94 specimens analyzed, planktonic benthopelagic cala- noid copepods were the most important prey (%1RI= 88.72). Suprabenthic mysids were the major accidental prey because of their high mass (%W=43.11 %) (Fig. 4). Trophic diversity values ( H\ Table 2) were lowest at depths of 1425-1800 m, highest in adults from 1800 to 2250 m, and higher in adults than juveniles for all the bathymetric zones analyzed. Dietary overlap (Table 3) was higher between groups 1000-1425 m and 1425-1800 m. Affinity between the diets of juveniles from 1800 to 2250 m and all other groups was low. Discussion According to Sulak (1977), Bathypterois species feed on benthopelagic plankton. Our study shows that Bathypter- ois mediterraneus, on the deep slope in the Catalan Sea, feed mostly on benthopelagic plankton (calanoid copepods), although occasionally on benthic resources (mainly supra- benthic form; occasionally endobenthic cumaceans, tanaid- aceans, etc; or epibenthic amphipods). The long and thin laminar gill rakers of Bathypterois mediterraneus (Car- rasson and Matallanas, 1994) are highly adapted for retain- ing planktonic prey and indicate filter feeding. Munk (1965) noted that Bathypterois have minute eyes, probably of lim- ited use in feeding. However, Collin and Partridge (1996) suggest that Bathypterois dubius may use two retinal area specializations in feeding. Bathypterois mediterraneus may exhibit similar behavior. However, there are no photographs of Bathypterois with an open mouth that might confirm filter feeding. The small size of Bathypterois mediterraneus is important in its feeding because Sulak (1984b) postu- lated that small-bodied components of the abyssal fauna are microphagous (e.g. Bathypterois longipes). Bathymetric differences were found in the diet oi Bathy- pterois mediterraneus. The prey consumed changed with depth, mainly at 1800-2250 m, where juveniles, in addi- tion to benthopelagic calanoid copepods, also fed on endo- benthic and epibenthic prey, mostly tanaids (80% of them found in individuals from a catch at a depth of 2250 m). Although a sample taken at 1800 m did not indicate a greater abundance of tanaids than in shallower waters (800-1000 m) (Cartes1), it is probable that tanaid abun- 1 Cartes, J. 1998. Personal commun. Institut Ciencies del Mar (CSIC), P. Joan de Borbo s/n, 08039 Barcelona, Spain. Carrasson and Matallanas: Feeding ecology of Bathypterois mediterraneus 269 Table 2 Composition of the diet of Bathypterois mediterraneus in the four groups (bathymetric and ontogenetic) established. IRI = index of relative importance; %IRI = percentage of IRI. Unid. = unidentified. 1800-2250 m 1000-1425 m 1425-1800 m Size 1 Size 2 Diversity (H') No. of specimens with food Composition of diet (Sizes 1 and 2) 2.69 94 IRI %IRI (Sizes 1 and 2) 2.33 95 IRI %IRI 2.87 (2.93) 25 IRI %IRI 3.16 68 IRI %IRI Foraminifera — — 0.3 0.0 — — — — Polychaeta 75.0 0.6 4.6 0.0 29.4 0.3 42.4 0.3 Polychaeta unid. 41.5 0.4 1.2 0.0 — — 25.0 0.2 Aphroditimorfa 4.2 0.0 1.1 0.0 29.4 0.3 4.2 0.0 Crustacea Crustacea unid. 42.0 0.4 — — — — — — Copepoda 9966.3 77.4 13195.5 90.0 5644.8 52.2 8550.2 67.6 Copepoda unid. — - 8.4 0.1 - — — — Calanoid Copepoda 9966.3 88.7 12531.0 95.5 5014.3 56.3 8550.2 79.5 Harpacticoid Copepoda — — — - 37.3 0.4 — — Ostracoda 20.0 0.2 3.8 0.0 — — 3.4 0.0 Ostracoda unid. 0.9 0.0 0.1 0.0 — — 0.4 0.0 Cipridin a sp. — — 1.0 0.0 — — 1.5 0.0 Cipridinidae 12.4 0.1 0.4 0.0 — — — — Amphipoda 609.5 4.7 117.1 0.8 1651.2 15.3 516.5 4.1 Amphipoda unid. — — 0.4 0.0 — — 0.8 0.0 Amphipoda Gammaridea 551.2 4.3 95.2 0.7 1651.2 15.3 477.5 3.8 Amphipoda Gammaridea unid. 138.1 1.2 30.2 0.2 151.8 1.7 122.8 1.1 Orchomene humilis 15.5 0.1 — — — — — — Orchomene sp. — — 0.1 0.0 13.6 0.2 — — Lyssianasidae 17.8 0.2 3.0 0.0 26.6 0.3 — — Harpini a sp. — — 1.0 0.0 — — — — Bruzelia typica — — 2.5 0.0 — — 4.7 0.0 Pseudotiton bouvieri — - — — — — 0.7 0.0 Rhachotropis caeca 0.7 0.0 - — - — — — Rhachotropis sp. 14.7 0.1 — — 438.3 4.9 30.6 0.3 Monoculode s sp. 0.9 0.0 — — — — 0.9 0.0 Oediceridae — — — — — — 3.9 0.0 Amphipoda Hyperiidea 1.3 0.0 0.2 0.0 — — — — Isopoda 13.1 0.1 5.0 0.0 22.6 0.2 8.8 0.1 Isopoda unid. 1.3 0.0 2.2 0.0 5.7 0.1 2.9 0.0 Gnathi a sp. 1.0 0.0 — — — — 0.4 0.0 Anthuridae 1.0 0.0 0.6 0.0 5.7 0.1 0.3 0.0 Ilyarachna sp. 0.2 0.0 — — — — — — Tanaidacea 0.9 0.0 7.7 0.1 2664.1 24.6 49.5 0.4 Tanaidacea unid. 0.9 0.0 — — — — — — Tanaidae — — 7.7 0.1 2664.1 29.9 41.7 0.4 Paratanaidae — — — — — — 0.3 0.0 Cumacea 15.8 0.1 194.6 1.3 751.8 7.0 1214.7 9.6 Cumacea unid. 7.3 0.1 99.3 0.8 401.9 4.5 596.0 5.5 Cyclaspis longicaudata — — 0.6 0.0 — — 14.0 0.1 Leucon longirostris 0.6 0.0 — — — — — — Campylaspis glabra 0.2 0.0 — — — — — — Campylaspis sp. — — 0.1 0.0 — — — — continued 270 Fishery Bulletin 99(2) Table 2 (continued) Diversity (H') No. of specimens with food Composition of diet 1000- 1425 m 1425- 1800 m 1800-2250 m Size 1 Size 2 (Sizes 1 and 2) 2.69 94 IRI %IRI (Sizes 1 and 2) 2.33 95 IRI %IRI 2.87 (2.93) 25 IRI %IRI 3.16 68 IRI %IRI Nannastacidae 1.6 0.0 _ Platysympus typicus — — 2.2 0.0 65.6 0.7 14.7 0.1 Diastylis sp. — — 0.2 0.0 7.6 0.1 9.1 0.1 Makrokylindrus sp. — — — — — - 9.4 0.1 Mysidacea 2112.3 16.4 1078.8 7.4 50.3 0.5 2269.5 17.9 Mysidacea unid. 785.5 7.0 196.0 1.5 50.3 0.6 1117.3 10.4 Boreomysis arctics. 108.4 1.0 47.0 0.4 — — 2.2 0.0 Boreomysis sp. 49.3 0.4 148.9 1.1 — — 198.2 1.8 Parapseudomm a sp. 1.2 0.0 — — — — — — Decapoda 4.3 0.0 27.9 0.2 — — — — Decapoda unid. 4.3 0.0 — — — — — — Larval Decapoda — — 0.4 0.0 — — — — Decapoda Natantia — - 17.4 0.1 — — — — Acanthephyra eximia — — 7.2 0.1 — — — — Pontophilus norvegicus - — 1.5 0.0 — — — — Osteichthyes — — 27.0 0.2 — ■ — — — Scales 16.0 0.1 — — — — — — dance increases in the environment at 2250 m. The mouth position of Bathypterois makes it improbable that they can extract tanaids from the substratum, in which these crea- tures tend to live. Holdich and Jones (1983) observed that some tanaids can swim very fast for short periods and it is probable that Bathypterois mediterraneus capture them during these periods. At 1800-2250 m, adult Bathypterois mediterraneus consume mysids as secondary prey, which they capture swimming over the bottom. The scarcity of resources, which decrease even more below 1200-1400 m (Cartes and Sorbe, 1993), may force Bathypterois to diversify their diet at 1800-2250 m (H'=2. 93), extending to endobenthic prey, such as cu- maceans and tanaidaceans. This stenophagic decrease co- incides with Dayton and Hessler’s (1972) deep predator prototype. Less intense specialization in a type of resource leads deep predators to be more adaptable, whether due to a general scarcity of resources or to the abundance of an occasional single resource. Ontogenetic differences in the diet were small. Adult intestines mainly contained more and larger prey than did the intestines of juveniles, from all depths. Only at 1800-2250 m were there clear differences in the prey captured by juveniles and adults, possibly reflecting pro- nounced qualitative changes in the decapod fauna of the western Mediterranean recorded around 2000 m by Cartes ( 1993d). Bathypterois grail ator , the largest member of the genus (Sulak, 1977) from the Bahamas, also has an onto- genetic shift in diet (Crabtree et al., 1991), but Bathypter- Table 3 Diet overlap (Schoener index) among the different bathy- metric and ontogenetic groups established. 1425-1800 m 1800- -2250 m Sizes 1 and 2 Size 1 Size 2 1000-1425 m 0.70 0.50 0.69 1425-1800 m — 0.47 0.65 1800-2250 m. (sizel ) — — 0.60 ois longipes and Bathypterois phenax do not significantly shift their diet ontogenetically. Our study goes beyond the preliminary data of Car- rasson and Matallanas (1990), indicating that benthope- lagic plankton, namely calanoid copepods, play an impor- tant dietary role in Bathypterois mediterraneus. Further, greater importance for mysids over amphipods (both ac- cidental prey), coincided with their relative abundance in the area (Cartes and Sorbe, 1993). Tanaids were found in the diet for the first time and seem important at depths of 1800-2250 m. Bathypterois dubius, a related species found in the North Atlantic has a diet broadly similar to that of Bathyp- terois mediterraneus feeding exclusively on planktoben- thic copepods (Marshall and Merrett, 1977). However, Sal- Carrasson and Matallanas: Feeding ecology of Bathypterois mediterraneus 271 A i 800-2250 m size 1 %W %N B size 2 CoC %W %N Figure 2 Percentage by weight (%W) and by number (%N) of prey species in the diet of (A) immature (size 1) and (B) mature (size 2) individuals at 1800-2250 m depth. CoC = Copepoda Calanoida; Bor = Boreomysis sp.; Mys = Mysidacea unidentified.; Pla = Platysympus typicus\ Cum = Cumacea unidentified.; CUM = total Cumacea; Amp = Amphipoda Gammaridea unidentified.; Rha = Rhachotropis sp.;Tan = Tanaidae; Aph = Aphroditimorfa; OTH = Others. 1425-1800 m CoC %w CoC %N Figure 3 Percentage by weight (%W) and by number (%N) of prey species in the diet of individuals at 1425-1800 m depth. CoC = Copepoda Calanoida; BoA = Bo?'eomysis arctica; Bor = Boreomysis sp.; Mys = Mysidacea unidentified.; MYS = total Mysidacea; Aca = Acanthephira eximicv, Ost = Osteicht.hyes; CUM = total Cuma- cea; AMG = total Amphipoda Gammaridea; OTH = Others. 272 Fishery Bulletin 99(2) 1000-1425 m %W POL OTH Figur e 4 Percentage by weight (%W) and by number (%N) of prey species in the diet of individuals at 1000-1425 m depth. CoC = Copepoda Calanoida; BoA = Boreomysis arctica\ Bor = Boreo- mysis sp.; Mys = Mysidacea unidentified.; MYS = total Mysidacea; Amp = Amphipoda Gam- maridea unidentified.; AMG =- total Amphipoda Gammaridea; Rha = Rhachotropis sp.; Dec = Decapoda unidentified.; POL = Polychaeta; OTH = Others. danha ( 1988) reported that Bathypterois chibius consumes some amphipods and mysids in addition to calanoid cope- pods, a finding similar to our data for Bathypterois medi- terraneus. In the western North Atlantic (Crabtree et ah, 1991) and eastern North Atlantic (Marshall and Merret, 1977; Merret, 1987) Bathypterois longipes primarily feed on copepods, amphipods, and decapods, but the difference is the greater importance of the benthic prey for Bathyp- terois mediterraneus, especially at depths between 1800 and 2250 m. In the Atlantic, Bathypterois dubius is restricted to depths of less than 2000 m (Sulak, 1984a); Bathypterois longipes inhabits depths below 2000 m. Sulak (1977) con- sidered the bathymetric and geographic distribution of these species to be a consequence of competitive interspe- cific exclusion, and this is supported by the results of Mar- shall and Merrett (1977). Bathypterois mediterraneus is the only abundant Mediterranean species of Bathypter- ois, with a lower bathymetric limit of 2800 m (Bauchot, 1987). In the Catalan Sea, it ranges down to the maxi- mum depth of 2250 m and exclusively exploits a trophic niche unoccupied by any other species. This strategy en- ables Bathyptei'ois mediterraneus to be the dominant spe- cies below 1425 m, where resources are more limited. The relatively higher metabolism as a result of small size in Bathypterois mediterraneus , like other Bathypter- ois (Heezen and Hollister, 1971; Saldanha, 1977), is com- pensated by reduced mobility and coupled with a micro- phagous filtering diet based on organisms distributed more uniformly in space, such as calanoid copepods. This trophic strategy makes Bathypterois mediterraneus espe- cially fit for a nutrient-limited environment, such as the Mediterranean, particularly at greater depths (Carpine, 1970; Tluel, 1983; Peres, 1985). Bathypterois species are dominant on the middle and lower slope and in the abyssal depths ofoligotrophic regions or similar less productive water masses. Sulak (1984b) and Merrett (1987) have noted the major importance of Chlorophthalmidae in marine fauna of oligotrophic envi- ronments. In more productive water masses, species with energetically expensive life history patterns dominate the ichthyofauna. In the Mediterranean, changes in biomass are evident so that the larger and middle-size dominant species between 1000 and 1200 m ( Mora moro, Phycis blen- noides, Trachyrhynchus trachyrhynchus, and Alepoeepha- lus rostratus ) are progressively replaced at greater depth by other smaller species such as Bathypterois mediterra- neus, Coryphaenoides guentheri, and Chalinura mediterra- nea (Stefanescu et al., 1993). The relative importance of Bathypterois mediterraneus increases as depth increases (Stefanescu et al., 1993), parallel to an increasing scarcity of trophic resources (Cartes, 1991; Cartes and Sorbe, 1993). This species is the most abundant fish of the lower slope assemblage (Morales-Nin et al., 1996) and exemplifies a species adapted to a food scarce environment by energeti- cally conservative feeding strategies. Acknowledgments We thank all the members of the BATIMAR research pro- gram for their collaboration, especially J. Rucabado and D. Lloris. We are grateful to J. E. Cartes for his collaboration in identifying several species and for his comments. The BATIMAR project was financed by CSIC and CAICYT ( PAC 86-008/ID 821). We also thanks the anonymous reviewers for their valuable comments on the manuscript. Literature cited Bauchot, M.-L. 1962. Description d'un nouveau Bathypterois mediter- raneen (Poisson Clupeiforme Bathypteroidae). Affinites et Carrasson and Matallanas: Feeding ecology of Bathypterois mediterraneus 273 remarques sur plusieurs especes du genre. Vie Milieu. 13:613-647. 1987. Famille Chlorophthalmidae. In Fiches FAO identifi- cation des especes pour les besoins de la peche (revision 1). Mediterranee et mer Noire. Zone de peche 37: Verte- bres (W. Fischer, M.-L. Bauchot, and M. Schneider, eds.), p. 1043-1046. FAO, Rome. Carpine, C. 1970. Ecologie de l’etage bathyal dans la Mediterranee occi- dentale. Bull. Inst. Oceanogr. (Monaco) 2: 1-146. Carrasson, M., and J. Matallanas. 1990. Preliminary data about the feeding habits of some deep-sea Mediterranean fishes. J. Fish Biol. 36:461-463. 1994. Morphometric characteristics of the alimentary tract of deep-sea Mediterranean teleosts in relation to their feed- ing habits. Mai'. Biol. 118:319-322. Carrasson, M., C. Stefanescu, and J. E. Cartes. 1992. Diets and bathymetric distributions of two bathyal sharks of the Catalan deep sea (western Mediterranean). Mar. Ecol. Prog. Ser. 82:21-30. Cartes, J. E. 1991. Analisis de las comunidades y estructura trofica de los crustaceos decapodos batiales del mar Catalan. Ph. D. diss., Universidad de Barcelona, 627 p. Cartes, J. E. 1993a. Diets of deep-sea brachyuran crabs in the Western Mediterranean Sea. Mar. Biol. 117:449-457. 1993b. Feeding habits of pasiphaeid shrimps close to the bottom in the Western Mediterranean slope. Mar. Biol. 117:459-468. 1993c. Diets of deep-water pandalid shrimps on the West- ern Mediterranean slope. Mar. Ecol. Prog. Ser. 96:49-61. 1993d. Deep-sea decapod fauna of the Western Mediterra- nean: Bathymetric distribution and biogeographic aspects. Crustaceana 65(1): 29-40. Cartes, J. E., and P. Abello. 1992. Comparative feeding habits of polychelid lobsters in the western Mediterranean deep-sea communities. Mar. Ecol. Prog. Ser. 84:139-150. Cartes, J. E., and F. Sarda. 1989. Feeding ecology of the deep-water aristeid crustacean Aristeus antennatus. Mar. Ecol. Prog. Ser. 54:229-238. Cartes, J. E. and J. C. Sorbe. 1993. Les communautes suprabenthiques bathyales de la Mer Catalane ( Mediterranee occidentale): donnees prelimi- naires sur la repartition bathymetrique et labondance des crustaces peracarides. Crustaceana 64( 2 ): 1 55—17 1 . Chave, E. H., and A. Malahoff. 1998. Deeper waters. Univ. Hawaii Press, Honolulu, Hawaii, 128 p. Chave, E. H., and B. C. Mundy. 1994. Deep-sea benthic fish of the Hawaiian Archipielago, Cross Seamount, and Johnston Atoll. Pac. Sci. 48(41:367-409. Church, R. 1971. Deepstar explores the ocean floor. Natl. Geogr. Mag. January 1971:110-129. Collin, S. P. and J. C. Partridge. 1996. Retinal specializations in the eyes of deep-sea tele- osts. J. Fish Biol. 49 (suppl. A):157-174. Crabtree, R. E., J. Carter, and J. A. Musick. 1991. The comparative feeding ecology of temperate and tropical deep-sea fishes from the western North Atlantic. Deep-Sea Res. 38( 10):1277-1298. Dayton, P. K., and R. R. Hessler. 1972. Role of biological disturbance in maintaining diver- sity in the deep sea. Deep-Sea Res. 19:199-208. Heezen, B. C., and C. D. Hollister. 1971 . The face of the deep. Oxford Univ. Press, New York, NY, 659 p. Holdich, D. M., and J. A. Jones. 1983. Tanaids. Cambridge Univ. Press, New York, NY, 98 p. Hureau, J.-C. 1970. Biologie comparee de quelques poissons antarctiques (Nototheniidae). Bull. Inst. Oceanogr. (Monaco). 68(1391): 1-250. Jones, A. T., and K. J. Sulak. 1990. First central Pacific plate and Hawaiian record of the deep-sea tripod fish Bathypterois grallator (Pisces: Chlo- rophthalmidae). Pac. Sci. 44(3):254-257. Marshall, N.B., and N. R. Merrett. 1977. The existence of a benthopelagic fauna in the deep sea. Deep-Sea Res. 24 ( suppl. ):483-497. Merret, N. R. 1987. A zone of faunal change in assemblages of abyssal demersal fish in the eastern north Atlantic: a response to seasonality in production? Biol. Oceanogr. 5:137-151. Merret, N. R., and N. B. Marshall. 1981 . Observations on the ecology of deep-sea bottom living fishes collected northwest Africa (08°27 N). Prog. Ocean- ogr. 9:185-244. Morales-Nin, B. 1990. A first attempt at determining growth patterns of some Mediterranean deep-sea fishes. Sci. Mar. 54(31:241-248. Morales-Nin, B., E. Massutf, and C. Stefanescu. 1996. Bathymetric distribution and growth patterns of Bathypterois mediterraneus from the north-western Medi- terranean Sea. J. Fish Biol. 49( suppl. A):276-288. Munk, O. 1965. Ocular degeneration in deep-sea fishes. Galathea Rep. 8:21-31. Peres, J. M. 1985. History of the Mediterranean biota and the coloni- zation of the depths. In Western Mediterranean (R. Mar- galef, ed.), p. 198-232. Pergamon Press, New York, NY. Pinkas, L., M. S. Oliphant , and I L. K. Iverson. 1971. Food habits of albacore, bluefin tuna and bonito in Cali- fornia waters. Calif. Dep. Fish Game Fish. Bull. 152:1-105. Rosecchi, E., and Y. Nouaze. 1987. Comparaison de cinq indices alimentaires utilises dans l’analyse des contenus stomacaux. Rev. Trav. Inst. Peches Marit. 49(3-4): 111-123. Saldanha, L. 1977. Poissons captures et photographies lors de plongees du bathyscaphe Archimede aux Azores, 1969. Arq. Mus. Bocage 6(3):35-50. 1988. Notes sur le regime alimentire de deux poissons de profondeur: Alepocephalus bairdii et Bathypterois dubius. Cybium 12(2): 169-171. Sehoener, T. W. 1974. Resource partitioning in ecological communities. Sci- ence (Wash. D.C.) 185:27-39. Smith, K. L. Jr. 1982. Zooplankton of a bathyal benthic boundary layer: in situ rates of oxygen consumption and ammonium excre- tion. Limnol. Oceanogr. 27(3): 461-471. Stefanescu, C., D. L. Loris and J. Rucabado. 1992a. Deep-living demersal fishes in the Catalan Sea (western Mediterranean) below a depth of 1000 m. J. Nat. Hist. 26:197-213. Stefanescu, C., J. Rucabado and D. Lloris. 1992b. Depth-size trends in western Mediterranean demer- sal deep-sea fishes. Mar. Ecol. Prog. Ser. 81(3): 205-213. 274 Fishery Bulletin 99(2) Stefanescu, C., D. Lloris, and J. Rucabado. 1993. Deep-sea fish assemblages in the Catalan Sea (west- ern Mediterranean) below a depth of 1000 m. Deep-Sea Res. 40:695-707. Stefanescu, C., B. Morales-Nin, and E. Massuti. 1994. Fish assemblages on the slope in the Catalan sea (western Mediterranean): influence of a submarine canyon. J. Mar. Biol. Assoc. U.K. 74:499-512. Sulak, K. J. 1977. The systematics and biology of Bathypterois (Pisces, Chlorophthalmidae) with a revised classification of benthic myetophiform fishes. Galathea Rep. 14:49-108. 1984a. Family: Chlorophthalmidae (including Bathyptero- idae, Benthosauridae, Ipnopidae). In Fishes of the north- eastern Atlantic and the Mediterranean, vol. 1 (P. J. P. Whitehead, M.-L. Bauchot, J.-C. Hureau, J. Nielsen, and E. Tortonese, eds.), p. 412-420. UNESCO, Bungay. 1984b. A comparative ecological analysis of temperate and tropical demersal deep-sea fish faunas in the Western North Atlantic. Adaptation of Ph.D. diss., Univ. Miami, Florida, 211 p. Thiel, H. 1983. Meiobenthos and nanobenthos of the deep sea. In Deep-sea biology: the sea, vol. 8. (G.T. Rowe, ed.), p. 167-230. John Wiley & Sons, New York, NY. Wishner, K. F. 1980. The biomass of the deep-sea benthopelagic plankton. Deep-Sea Res. 27:203-216. 275 Larval development of the kishi velvet shrimp, Metapenaeopsis dales (Rathbun) (Decapoda: Penaeidae), reared in the laboratory Jung H. Choi Sung Y. Hong Department of Marine Biology Pukyong National University Pusan 608-737, Korea E-mail (for J H Choi): choijh@mail1.pknu ac.kr Abstract— The complete larva! devel- opment of Metapenaeopsis daiei (Rath- bun) is described from laboratory- reared larvae. The larvae were reared in a plastic container (20 L). Larvae from the first zoeal stage to the third zoeal stage were fed algae; larvae from the first mysis stage to the first postlar- val stage were fed newly hatched nau- plii of Artemia. The nauplii of M. daiei hatched about 20 hours after spawning. The larvae passed through six naupliar, three zoeal, and three mysis stages before the postlarval stage. About 15 days (about 347 hours) were required from hatching to the first postlarval stage. Larval morphology of M. daiei is described and compared with those of other Metapenaeopsis spp. The first nauplius stage larva of M. daiei bears a chitinous conical protuberance on the mediodistal margin of the body. The third protozoeal stage larva has five dorsomedian abdominal spines on the abdomen. The morphological char- acters of the first postlarva are 43 scaphognathite setae on the maxilla, three scaphocerite spines on the outer margin of antenna, and sternal plate spines: 2, 2, 0,1,1. Manuscript accepted 7 November 2000. Fish. Bull. 99:275-291 (2001). The kishi velvet shrimp, Metapenae- opsis daiei (Rathbun), inhabits coastal waters of Korea (Kim, 1977) and Japan (Kubo, 1949). Of the 26 penaeid genera comprising 215 species, the genus Metapenaeopsis is the largest with 76 species (Perez Farfante and Kensley, 1997). However, studies on the com- plete larval development oi Metapenae- opsis spp. are few. The mysis and postlarval stages of Metapenaeopsis mogiensis (Ratbun) and Metapenaeopsis andamanensis (Wood- Mason) and postlarval stages of Metap- enaeopsis barbata (De Haan) have been partially described by Paulinose (1988) from plankton samples. Jackson et al. (1989) reared Metapenaeopsis palmen- sis (Haswell) in the laboratory and de- scribed all the larval stages of the spe- cies except the naupliar stages. From previous works on Indo-west Pacific penaeid larvae, they also described the important morphological characters for identification. Chong and Sasekumar (1994) studied the larval development of Metapenaeopsis stridulans (Alcock) from the egg to the first postlarval stage reared in the laboratory and prepared descriptions of all the larval stages. Ronquillo and Saisho (1997) provided descriptions and illustrative figures of the complete larval development of M. barbata and compared the morphologi- cal characters of the larval stages with those of several penaeid species. The purpose of the present study is to describe the larval development of M. daiei from the egg to the postlarval stage and to compare its larval devel- opment with that of other known Meta- penaeopsis larvae. Materials and methods On 22 August 1996, gravid females of Metapenaeopsis daiei were caught by shrimp trawl in Oeyondo, Korea, and transported to the laboratory of the Taean Fish Hatchery, National Fisher- ies Research and Development Insti- tute, Korea (NFRDI). Each of them was kept in a 0.5 t container, containing filtered seawater (22.0°C; 32.2%o), until they spawned. On the night of 22 August 1996, a gravid female spawned, and the eggs hatched on the next day. The larvae were reared in a plastic 20-L contain- er. Half the total volume of rearing wa- ter was changed daily with fresh sea- water (24.0-26.0°C; 32.2-33.2%D. The larvae were fed with a mixture of algae ( Chaetoceros calcitrans and Isochrysis galbana) from the first zoeal stage to the third zoeal stage and with newly hatched nauplii of Artemia from the first mysis stage to the first postlarval stage. For size measurement and morpho- logical observation, eggs were sampled from the bottom of the rearing tank im- mediately after having been spawned; the developmental processes of the eggs were observed every 10 minutes, the nauplii were sampled at four hour in- tervals after hatching, and subsequent larval stages were sampled at least twice a day. The following measurements were taken from preserved larvae by using an ocular micrometer: total length (TL), from the tip of the rostrum to the tip of the caudal end or telson; carapace length (CL), from the anterior margin of 276 Fishery Bulletin 99(2) Table 1 Metapenaeopsis dalei. Chronology of embryonic devel- opment under laboratory conditions (24.0-26.0°C; 32.2- 33.2 %o) Developmental stages Duration (hour: min.) Cumulative (hour: min.) 1 cell 00: 23 00: 00 2 cell 00: 25 00: 23 4 cell 00: 26 00: 48 8 cell 00: 28 01: 14 16 cell 00: 20 01: 42 32 cell 00: 28 02: 02 64 cell 00: 50 02: 30 Gastrula stage 04: 50 03: 20 Early embryonic nauplius 05:30 08: 10 Late embryonic nauplius 06:30 13: 40 Embryonic nauplius just before hatching 20: 10 Table 2 Metapenaeopsis dalei. Chronology of larval development under laboratory conditions (24.0-26.0°C; 32.2-33.2%D Duration Cumulative Larval stages (hour) (hour) First naupliar stage 06 006 Second naupliar stage 07 013 Third naupliar stage 07 020 Fourth naupliar stage 08 028 Fifth naupliar stage 08 036 Sixth naupliar stage 12 048 First protozoeal stage 48 096 Second protozoeal stage 60 156 Third protozoeal stage 50 206 First mysis stage 46 252 Second mysis stage 26 278 Third mysis stage 49 327 First postlarval stage 119 446 the carapace to the midposterior margin of the carapace for zoeae, from the postorbital margin to the midposterior margin of the carapace for mysis and first postlarvae; body width (BW), the greatest width across the body for nauplii. Results Development of larvae and duration of the larval stage A gravid female of Metcipenaeopsis dalei spawned for two hours from 22:00 to 24:00. The egg was spherical and greenish grey, with a mean diameter of 0.35 mm (n=20). The nauplius hatched about 20 hours after spawning (Table 1). The larvae passed through six naupliar, three zoeal, and three mysis stages, before the postlarval stage. About 15 days (about 347 hours) were required from the hatching to the first postlarval stage (Table 2). Descriptions of larvae First naupliar stage (Fig. 1, A and B) Size TL: 0.37 mm (0.35—0.41 mm; SD=0.02; n = 12) BW: 0.23 mm (0.21-0.25 mm; SD=0.02; n=12) Body (Fig. 1, A and B ): Slightly oval in dorsal or ventral view, convex dorsally; anterodorsal median ocellus; posterior mar- gin rounded; one chitinous conical protuberance on postero- dorsal part of body; caudal end with one pair of setae; body unsegmented; three pairs of appendages with naked setae. Antennule (Fig. 1A): Uniramous; two terminal setae and four lateroventral setae. Antenna (Fig. 1A): Biramous; endopod with three termi- nal and two lateroventral setae; exopod with three termi- nal and three lateroventral setae. Mandible (Fig. 1A): Biramous; endopod and exopod each with one lateroventral and two terminal setae. Second naupliar stage (Fig. 1, C and D) Size TL: 0.43 mm (0.37-0.53 mm; SD=0.04; n=15) BW: 0.23 mm (0.16-0.27 mm; SD=0.03; n=15) Body (Fig. 1, C and D): Elongate oval in dorsal or ven- tral view; anterodorsal median ocellus; posterior margin rounded; posterodorsal protuberance absent; caudal end with two pairs of setae; three pairs of appendages with bi- pinnate setae; body unsegmented. Antennule (Fig. 1C): Unchanged. Antenna (Fig. 1C): Unchanged. Mandible (Fig. 1C): Unchanged. Third naupliar stage (Fig. 1, E and F) Size TL: 0.44 mm (0.42-0.45 mm; SD=0.02; n=6) BW: 0.24 mm (0.21-0.27 mm; SD=0.03; n= 6) Body (Fig. 1, E and F): Form unchanged; anteroventral median ocellus; posterior margin rounded; caudal end with two pairs of setae; body unsegmented. Antennule (Fig. IE): Uniramous; three terminal setae, outermost short and four lateroventral setae. Antenna (Fig. IE): Biramous; endopod with three termi- nal and three lateroventral setae; exopod with three ter- minal and three lateroventral setae. Mandible (Fig. IE): Unchanged. Fourth naupliar stage (Fig. 2, A and B) Size TL: 0.43 mm (0.40-0.47 mm; SD=0.03; n- 4) BW: 0.23 mm (0.22-0.24 mm; SD=0.01; 72=4) Body (Fig. 2, A and B): Slightly elongate; anteroventral median ocellus; medial notch dividing posterior end into two symmetrical lobes each with 3+3 setae; thoracic part with some rudimentary segmentation ventrally. Choi and Hong: Larval development of Metapenaeopsis dale/ 277 278 Fishery Bulletin 99(2) Choi and Hong: Larval development of Metapenaeopsis dale i 279 Antennule (Fig. 2A): Uniramous; three terminal and three lateroventral setae. Antenna (Fig. 2A): Biramous; endopod with three termi- nal and two lateroventral setae; exopod with three termi- nal and four lateroventral setae. Mandible (Fig. 2A): Biramous; protopodal part swollen; endopod and exopod each with two terminal and one lat- eroventral setae; coxal part enlarged. Fifth naupliar stage (Fig. 2, C and D) Size TL: 0.41 mm (0.34-0.54 mm; SD=0.07; 72 = 10) BW: 0.18 mm (0.14-0.23 mm; SD=0.03; n = 10) Body (Fig. 2, C and D): Slightly more elongate, tapering towards telson lobes; anteroventral median ocellus; cara- pace rudimentary; buds of the oral appendages appeared. Antennule (Fig. 20: Uniramous; segmentation rudi- mentary; two terminal and two lateroventral setae. Antenna (Fig. 20: Biramous; segmentation rudimentary; endopod with three terminal and two lateroventral setae; exopod with four terminal and five lateroventral setae. Mandible (Fig. 20: Biramous; protopod bulbous; seta- tion unchanged. Telson (Fig. 20: Bilobed with 6+6 setae. Sixth naupliar stage (Fig. 2, E and F) Size TL: 0.56 mm (0.53-0.58 mm; SD=0.01; /? = 15 ) BW: 0.24 mm (0.23-0.26 mm; SD=0.01; 22=15) Body (Fig. 2, E and F): Elongate with distinct waist; an- teroventral median ocellus; carapace developed; rudimen- tary maxillae and maxillipeds bearing numerous short setae. Antennule (Fig. 2E): Uniramous; 11-segmented with 0, 0, 0, 0, 0, 1, 0, 1, 0, 2, 4 setae. Antenna (Fig. 2E): Biramous; peduncle two-segmented; endopod four-segmented with 0, 1, 1, 4 setae; exopocl 10-segmented with 0, 0, 1, 2, 1, 1, 1, 1, 1, 3/4 setae. Mandible (Fig. 2E): Unchanged. Telson (Fig. 2E): Bilobed with 7+7 setae of varying sizes. First protozoeal stage (Fig. 3, A-l) Size TL: 1.07 mm (0.97-1.23 mm; SD=0.09; n=10) Cephalothorax (Fig. 3A): Incompletely and loosely cov- ered by unarmed carapace; carapace round to oval, with shallow posterior median notch, anterior margin with one pair of orbital spines; rostrum absent; paired sessile com- pound eyes. Antennule (Fig. 3B): Uniramous; six-segmented with 0, 0, 0, 0, 4, 6 (three aesthetascs) setae. Antenna (Fig. 30: Biramous; protopod two-segmented; endopod two-segmented with 5, 5 setae; exopod 10-seg- rnented with 0, 0, 1, 2, 1, 2, 1, 1, 1, 3 setae. Mandible (Fig. 3D): Endopod and exopod absent; asym- metrical; masticatory surface with lower molar processes and upper curved incisor processes; molar with many small teeth; incisor with 2-4 larger teeth; left and right mandible each with one denticulated tooth. Maxillule (Fig. 3E): Biramous; protopod two-lobed, coxal endite with eight setae, basal endite with two stout setae and two setae; endopod three-segmented with 3, 2, 5 setae; small exopod with four plumose setae. Maxilla (Fig. 3F): Biramous; protopod five-lobed, endites with 8, 3, 3, 3, 3 setae; endopod four-segmented with 2, 2, 2, 3 setae; exopod with five plumose setae. First maxilliped (Fig. 3G): Biramous; protopod partially divided, coxa with six setae, basis with nine setae; endo- pod four-segmented with 3, 2, 2, 5 setae; exopod with seven plumose setae. Second maxilliped (Fig. 3H): Biramous; protopod two- segmented, coxa without setae, basis with four setae; en- dopod four-segmented with 2, 2, 2, 5 setae; exopod with five plumose setae. Third maxilliped (Fig. 31): Undeveloped biramous; exo- pod with two plumose setae. Post-carapacial region (Fig. 3A): Six thoracic and one free abdominal somites; unsegmented abdomen with forked telson. Telson (Fig. 3A): Broadly bifurcated in two lobes with seven pairs of plumose setae. Second protozoeal stage (Fig. 4, A-l) Size TL: 1.58 mm (1.44-1.83 mm; SD=0.14; 22=17) CL: 0.50 mm (0.40-0.60 mm; SD=0.05; 22=17) Cephalothorax (Fig. 4A): Loosely covered by carapace; carapace armed with long acute rostrum and two pairs of supraorbital spines; rostrum present and exceeding eye; eyes stalked. Antennule (Fig. 4B): Uniramous; seven-segmented with 0, 0, 0, 0, 1, 4, 6 (three aesthetascs) setae. Antenna (Fig. 40: Biramous; protopod two-segmented with 0, 1 setae; unchanged; exopod 10-segmented with 0, 1,2, 1,2, 1, 1, 1, 1,3 setae. Mandible (Fig. 4D): Form unchanged; left mandible with four denticulate setae; right mandible with one denticu- late seta. Maxillule (Fig. 4E): Biramous; coxal endite with eight setae, basal endite with five stout setae and one seta; en- dopod three-segmented with 2, 2, 5 setae; exopod with four plumose setae. Maxilla (Fig. 4F): Biramous; endites with 10, 4, 4, 5, 3 setae; endopod four-segmented with 3, 2, 2, 3 setae; exopod unchanged. First maxilliped (Fig. 40: Biramous; coxa with six se- tae, basis with 12 setae; endopod with 3, 2, 2, 4 setae; exo- pod unchanged. Second maxilliped (Fig. 4H): Biramous; coxa with one seta, basis with nine setae; endopod four-segmented with 2, 2, 2, 5 setae; exopod with six plumose setae. Third maxilliped (Fig. 41): Undeveloped biramous; exo- pod with three plumose setae. Post-carapacial region (Fig. 4A): Six thoracic and six abdominal somites; unsegmented abdomen with forked telson. Telson (Fig. 4A): Unchanged. Third protozoeal stage (Fig. 5, A-K) Size TL: 2.22 mm ( 1.80-2.43 mm; SD=0. 18; 72 = 11 ) CL: 0.60 mm (0.50-0.70 mm; SD=0.06: 77 = ] 1 ) Cephalothorax (Fig. 5A): Incompletely covered by cara- pace; carapace armed with long rostrum and two pairs of supraorbital spines; rostrum long. 280 Fishery Bulletin 99(2) Figure 3 Metapenaeopsis dalei. First zoeal stage: (A) dorsal view; (B) antennule; (C) antenna; < D ) mandible: (E) maxillule; (F) maxilla; (G> first maxilliped; (II) second maxilliped; (I) third maxilliped. Scale bars = 0.2 mm. Antennule (Fig. 5B): Uniramous; four-segmented with 1, 1, 2, 8 (three aesthetascs). Antenna (Fig. 5C): Unchanged. Mandible (Fig. 5D): Left mandible with six denticulate setae; right mandible with two denticulate setae. Maxillule (Fig. 5E): Biramous; coxal endite with eight setae, basal endite with seven stout setae and three setae; Choi and Hong: Laiwal development of Metcipenaeopsis dale/ 281 Figure 4 Metcipenaeopsis dcilei. Second protozoeal stage: (A) dorsal view; 1B1 antennule; (C) antenna; * D ) mandible; (E) maxillule; (F) maxilla; (G) first maxilliped; pleopod; ( J ) telson and uropod. Scale bars = 0.2 mm. Fifth pereiopod (Fig. 10E): Similar to 4th pereiopod. Abdomen (Fig. 6F): Similar to 2nd mysis stage. Pleopods (Fig. 91); Two-segmented with 0, 2 minute ter- minal setae. Uropod (Fig. 9J): Protopod with one spine; endopod with 19 plumose setae; exopod with one fused spine and 20 plu- mose setae. Telson (Fig. 9J): Posterior margin straight, with seven pairs of plumose setae and one posteromedian spine. First postlarval stage (Fig. 6 , G and H; Fig. 11, A-J; Fig. 12, A-J) Size TL: 5.07 mm (4.68-5.64 mm; SD=0.31; n= 20) CL: 1.21 mm (1.03—1.51 mm; SD=0.14; n= 20) 288 Fishery Bulletin 99(2) Figure 10 Metctpenaeopsis dalei. Third mysis stage: (A) first pereiopod; (B) second pereiopod; (C) third pereiopod; (D) fourth pereiopod; (E) fifth pereiopod. Scale bar = 0.2 mm. Body contour (Figs. 6G, 111): Adult-like; pleopod devel- oped; thorax completely covered by carapace; last five tho- racic sternal plates with 2, 2, 0, 1, 1 spines. Carapace (Fig. 6, G and H): Long rostrum with one epi- gastric tooth and three dorsal teeth; one pair of supraor- bital, antennal, pterygostomial and hepatic spines antero- ventral and posteroventral spines absent. Antennule (Fig. 1 1 A): Peduncle three-segmented, 1st segment with one large ventral spine, one small proximal spine, 40 plumose setae and a statocyst, 2nd segment with 10 plumose setae, 3rd segment with seven plumose setae; outer ramus two-segmented with 2, 6 aesthetascs; inner ramus two-segmented with 1. 3 plumose setae. Antenna (Fig. 11B): Protopod unsegmented with two spines and one seta; endopod 23-segmented, 1st segment with two spines; scaphocerite with 34 marginal plumose se- tae, four spines on outer margin and 21 setae on surface. Mandible (Fig. 1 1C ): Denticulate setae absent; mandib- ular palp three-segmented, distal segment ovate; left man- dible with 1, 10, 20 plumose setae; right mandible with 2, 11,21 plumose setae. Maxillule (Fig. 1 1 D ): Protopod two-lobed, coxal endite with nine plumose setae, basial endite with 22 naked se- tae; endopod four-segmented with 2, 2, 2, 3 setae. Maxilla (Fig. 11E): Protopod four-lobed, endites with 11, 4, 10, 8 setae; endopod two-segmented with 8, 3 setae; scaphognathite with 43 plumose setae. First maxilliped (Fig. 1 IF ): Protopod two-lobed, coxal endite with nine setae and wo spinules, basial endite with 36 setae; endopod four-segmented with 2, 0, 1, 5 setae; exo- pod with 10 plumose setae. Second maxilliped (Fig. 11G): Naked podobranch, basis with 16 setae; endopod five-segmented with 8, 15, 1,9, 10 setae; exopod with six plumose setae. Third maxilliped (Fig. 11H): Coxa with three setae, ba- sis with six setae; endopod five-segmented with 13, 12, 9, 13, 9 setae; exopod with eight plumose setae. First pereiopod (Fig. 12A): Coxa with four setae, basis with four setae; endopod 5-segmented and fully chelate with numerous setae; exopod with 8 plumose setae. Second pereiopod (Fig. 12B): Coxa with one seta, basis with two setae; endopod five-segmented and fully chelate with numerous setae; exopod with 8 plumose setae. Third pereiopod (Fig. 120: Coxa naked, basis with five setae; endopod five-segmented and fully chelate with nu- merous setae; exopod with 10 plumose setae. Fourth pereiopod (Fig. 12D): Coxa naked, basis with four setae; endopod five-segmented with numerous fine setae; exopod with 11-12 plumose setae. Fifth pereiopod (Fig. 12E): Similar to 4th pereiopod in form; exopod with 7-8 plumose setae. First pleopod (Fig. 12F): Uniramous; two-segmented, proximal segment with 3-5 setae, distal segment with 10-14 plumose setae. Choi and Hong: Larval development of Metapenaeopsis dalei 289 Figure 11 Metapenaeopsis dalei. First postlarval stage: (A) antennule; 2 individuals of one species, or indistinguishable as species, only the largest and smallest were measured. These samples were analyzed to characterize food habits and the intent of the analysis was simply to determine range of sizes consumed. Also, records of S. paiicispinus are excluded because this species grows much faster than the others, and several individuals of the subject species were excluded because it was evident they were in their second year. - Two distinct groups are included here < /=9. 129, 4 df, P<0.001). Nineteen ranged from 4,0 to 6.0 cm SL, whereas the other five ranged from 7.0 to 8.0 cm SL. Table 3 Number ofYOY Sebastes spp. consumed by each of the three predator species during June, July, August, and during all other months, 1977-87 (values in parentheses=number predators examined that had gut contents). Sebastes species June July August All other S. melanops 313 (57) 44 (53) 40 (31) 1 (37) S. mystinus 32 (100) 5 (37) 0 (95) 1 (286) H. decagrammus 81 (75) 12 (51) 4 ( 65 ) 4 (142) 298 Fishery Bulletin 99(2) CD TD O CD O) 03 C 0) o CD CL 60 r 50 40 30 20 10 0 Fish lH S. melanops n = 178, 20.0 to 47.5, x = 31.9 (SE 0.39) cm SL □ S. mystinus n = 518, 15.1 to 39.2, x = 26.5 (SE 0.23) cm SL | | H. decagrammus n = 333, 22.8 to 37.5, x = 29.9 (SE 0.1 5) cm SL Foods Figure 4 Foods consumed by the three predator species ( Sebastes melanops, S. mystinus, and S. decagrammus) 1977-87 (n = individuals with identifiable food in gut). tion in YOY abundance (Fig. 6). Evidence of predation was strongest during years when YOY were most numerous in the habitat (1977, 1979, 1985, and 1987), and there was virtually no evidence of predation during the years when YOY were least numerous in the habitat (1983, 1984, and 1986). Furthermore, it was only during the years that YOY were more numerous that we found them among the prey of S. mystinus — 1985 being the year we found the most. We saw no indication that distributions of adult S. mela- nops, S. mystinus, or H. decagrammus were influenced by the presence of YOY Sebastes. Even when YOY were most numerous, attackers seemed limited to residents of the im- mediate area. This finding should have been expected be- cause each year’s relative abundance was in effect coast wide. Discussion Much predation is opportunistic; thus predators often are drawn to concentrations of organisms that would not oth- erwise be their prey. Consider, for example, that when the squid Loligo opcdescens deposits great masses of eggs on sediment off central California, various fishes, marine mammals, and birds converge from surrounding habitats to forage on what is for them an unusual food (Morejohn et ah, 1978). Similarly, when the herring Clupea pallasi deposits eggs in great abundance on the seabed in San Francisco Bay, these become food for the brown rockfish, Sebastes auriculatus, which otherwise feeds mainly on decapod Crustacea and fishes (Ryan, 1986). Another exam- ple is an incident during 1962 in the Gulf of California, where a 1-m moray eel (Muraenidae) was observed thrash- ing at the water’s surface in a vigorous attempt to feed from a dense swarm of larval fishes, each no more than 1 cm long (Hobson, 1968). Certainly this predator was behaving in a manner that was contrary to its usual mode of feeding. So perhaps it should have been expected that YOY Sebastes in exceptionally large numbers would draw attacks from predators not otherwise prone to show them interest. But the level of predation on YOY Sebastes off Mendoci- no involved more than relative abundance of prey. The attacks were concentrated during June, declined sharply through July and August, then remained at low levels dur- ing the rest of the year. This pattern did not follow the number of YOY present. These YOY gained abundance through most of June to attain maximum numbers late that month or during July, and then remained abundant after predation had subsided to low levels at the end of August. Often our counts of YOY were higher during Sep- tember or October than during June or July, and although to a considerable extent this higher court was related to increased visibility in clear water, it nevertheless argues Hobson et al.: Predation on first-year Sebastes spp. 299 70 60 - 50 Sebastes YOY a) T3 40 CL 30 20 10 melanops mystinus decagrammus n = 141 n = 232 n = 191 June-August B Other fish melanops mystinus decagrammus n= 37 n = 286 n=142 September-May Figure 5 Fish in diet of the three predators ( Sebastes melanops, S. mystinus , and S. decagrammus ) June through August com- pared with September through May. YOY Sebastes spp. are distinguished from other fish. « individuals with identifi- able food in gut. against attributing the precipitous decline in predation through late August simply to shortages of prey. It has been widely reported that mortality among YOY of a variety of fishes is greatest during and immediately after settlement (e.g. Doherty and Sale, 1985; Victor, 1986; Shulman and Ogden, 1987), but there has been virtually no attempt to explain this finding, other than to implicate predation. We propose that the intense predation of early summer, and its subsequent sharp decline, mirrored a pat- tern of vulnerability among the YOY. Abilities critical for survival are quickness in responding to attacks, speed in attaining shelter, and the ability to use shelter that is available. And increasingly important with time is the abil- ity to acclimate to novel situations — the capacity to learn. These abilities are based on inherent characteristics that can be expected to vary widely among individuals, with the more deficient being more vulnerable to predators. It fol- lows that individuals most deficient are likely to be among those consumed during or shortly after settlement, where- as individuals progressively less deficient would be con- sumed in diminishing numbers over the following months. Certainly other factors contributed to the decline in pre- dation during the summer. That YOY decreased in number and increased in size must have influenced the intensity of predation. There is evidence that by summer’s end most YOY had grown too large for predation by S. mystinus and H. decagrammus. The largest YOY Sebastes among the gut contents of either species were 5.5 cm SL (Table 2), and by September most in the environment were larger than this (based on growth evident in the representative S. mysti- nus sampled for study of food habits). Other considerations, however, argue against the impor- tance of prey-size in shaping the observed pattern of pre- dation. Sebastes melanops, which is morphologically better suited than either S. mystinus or H. decagrammus to prey on fishes (Fig. 1), was able to feed on YOY Sebastes of up to at least 8.0 cm SL (Table 2), yet ate very few of them af- ter August. Also, the great variation in size among preda- tors, as well as in growth among YOY would be expected to dampen the effects of prey-size on population-level feeding intensity. Another consideration is the continued presence of YOY as small as 4.0 cm SL (Table 2) during the late- summer switch to other prey. Although probably a combination of factors contributed to the decline in predation during the summer, we contin- ue to consider that the concentrated attacks immediately after settlement are mostly likely elicited by the presence of inherently less-adaptive individuals. This is classic nat- 300 Fishery Bulletin 99(2) 100 80 60 40 20 n 6 6 3 freq. 10030.7 8 36 4 00 00 00 4 0 30 0 0 0.0 0 0 ■ S. melanops □ S. mystinus □ H. decagrammus ft 1 34 9 11 38 23 000 00.0 0 7 040.8 115 10 8 18 4 000001 090201 <13 O c 03 T3 C 13 JD 03 03 > CD cc 1977 1978 1979 1980 1981 1982 1983 1984 1985 1986 1987 Figure 6 Inter-annual variation in occurrence of YOY Sebastes spp. in predator diets in relation to abundance of YOY in the habitats, June through August. Measures of YOY abundance in habitat are based on information presented in Figure 3. n = Individuals examined that had identifiable food in gut. ural selection. That predators studied in our study had es- sentially resumed their regular diets by the end of sum- mer indicated to us that YOY vulnerable to their attacks had by that time become scarce. There are other predators that follow similar temporal patterns in feeding on YOY Sebastes off northern Cali- fornia. Pacific hake, Merluccius productus, for example, have been reported attacking YOY S. jordani so close to shore that many are driven onto the beach — but only dur- ing June and early July of years when these prey were particularly abundant (Hobson and Howard, 1989). Simi- larly, king salmon. Oncorhynchus tshawytscha, regularly switch to YOY Sebastes from other prey during late May and June, then switch back before the end of July (Adams et al.2). Other examples include Sebastes spp. studied by Hallacher and Roberts (1985) in kelp forests off Carmel, CA. In a chronicle that followed activities of these fishes over one year (with limited data from July and August of two other years), it was determined that S. atrovirens, S. carnatus, S. chrysomelas, and S. melanops fed primarily on YOY Sebastes during the “upwelling season” (April to Au- gust), but all except S. carnatus switched to invertebrates during the “non-upwelling season” (September to March). Hallacher and Roberts attributed this pattern to seasonal differences in YOY abundance. 2 Adams, P. B., W. M. Samiere, and C. J. Ryan. 1986. Unpubl. manuscript. Prey selection and diet of marine chinook salmon, Oncorhynchus tshawytscha, 17 p. Natl. Mar. Fish. Ser., NOAA, Tiburon CA 94920. The decline in attacks during the summer does not lead to essentially an end to predation on YOY, however. At least one resident piscivore, the ling cod, Ophiodon elon- gatus, remains a regular threat (Miller and Geibel, 1973). In fact, Adams and Howard (1996), studying in part the same series of assessments used in our study, considered predation the major cause of natural mortality in YOY S. mystinus from late summer through early spring. Their estimates were higher during years that YOY were more abundant, indicating persistent density-dependent preda- tion, but were much lower than most other published rates of natural mortality for juvenile fishes, probably because they missed the intense predation of late spring and early summer. And although Hallacher and Roberts (1985) re- ported that most Sebastes spp. off Carmel switched to pre- dation on invertebrates from September to March, they noted that S. carnatus continued to prey on YOY Sebastes during that period. There is evidence that density-related predation con- tinues to dampen interannual variation in year-class size through entry into the fishery, which for most Sebastes spp. is at about age 3-4 years (Ralston and Howard, 1995). Arguments for the importance of postsettlement mortal- ity in establishing ultimate year-class size have empha- sized the extended effect of this mortality (e.g. Sissenwine, 1984) but do not recognize the extent that mortality from predation is concentrated during a relatively brief period immediately after settlement. We suggest that manage- ment needs would be most effectively met by measuring year-class size soon after this period of intense predation. Hobson et al.: Predation on first-year Sebastes spp 301 Although a measure based on recruits entering the fish- ery would more accurately define the resource, there have been no effective methods developed to determine abun- dance at that point. And even if available, the lateness of such a measure would limit its effectiveness. An as- sessment made at the end of the first summer based on direct visual counts by underwater observers, as described above, is an effective compromise. The YOY are readily counted, much of the postsettlement mortality is included, and there is time to use the results in planning manage- ment strategy. Acknowledgments We thank Mickey Eldridge and Bruce McFarlane for con- structive criticism of the manuscript. Literature cited Adams, P. B., and D. F. Howard. 1996. Natural mortality of blue rockfish, Sebastes mystinus, during their first year in nearshore benthic habitats. Fish. Bull. 94:156-162. Bakun, A. 1973. Coastal upwelling indices, west coast od North Amer- ica, 1946-71. U. S. Dep. Commer., NOAATech Rep. NMFS SSRF-671, 103 p. Bailey, K. M., and S. M. Spring. 1992. Comparison of larval, age-0 juvenile and age-2 recruit abundance indices of walleye pollock, Theragra chalco- gramma , in the western Gulf of Alaska. ICES J. Mar. Sci. 49:297-304. Bertram, D. F., and W. C. Leggett. 1994. Predation risk during the early life history periods of fishes: separating the effects of size and age. Mar. Ecol. Prog. Ser. 109:105-114. Carlson, H. R., and L. Barr. 1977. Seasonal changes in spatial distribution and activity of two species of Pacific rockfishes, Sebastes flavidus and S. ciliatus, in Lynn Canal, southeastern Alaska. Mar. Fish. Rev. 39:23-24. Cushing, D. H. 1973. Recruitment and parent stock in fishes. Univ. Wash- ington, Seattle, WA, 197 p. Doherty, P. J., and P. F. Sale. 1985. Predation on juvenile coral reef fishes: an exclusion experiment. Coral Reefs 4:225-234. Gotshall, D. W., J. G. Smith, and A. Holbert. 1965. Food of the blue rockfish, Sebastodes mystinus. Calif. Fish Game 51:147-162. Hallacher, L. E., and D. A. Roberts. 1985. Differential utilization of space and food by the inshore rockfishes (Scorpaenidae: Sebastes ) of Carmel Bay, California. Environ. Biol. Fish. 12:91-110. Hixon, M. A., and M. H. Carr. 1997. Synergistic predation, density dependence, and popu- lation regulation in marine fish. Science (Wash. D.C.) 277: 946-949. Hjort, J. 1926. Fluctuations in the year classes of important food fishes. J. Cons. Int. Explor. Mer 1:5-38. Hobson, E. S. 1968. Predatory behavior of some shore fishes in the Gulf of California. U. S. Fish Wildl. Serv. Res. Rep. 73, 92 p. Hobson, E. S., and J. R. Chess. 1988. Trophic relations of the blue rockfish, Sebastes mysti- nus, in a coastal upwelling system off northern California. Fish. Bull. 86:715-743. Hobson, E. S., and D. F. Howard. 1989. Mass stranding of juvenile shortbelly rockfish and Pacific hake along the coast of northern California. Calif. Fish Game 75:169-183. Laroche, W. A., and S. J. Richardson. 1980. Development and occurrence of larvae and juveniles of the rockfishes Sebastes flavidus and Sebastes melanops (Scorpaenidae) off Oregon. Fish. Bull. 77:901-923. Lockwood, S. J. 1980. Density-dependent mortality in 0-group plaice iP/eu- ronectes platessa L. ) populations. J. Cons. Int. Explor. Mer 39: 148-153. Love, M. S., M. H. Carr, and L. -J. Haldorson. 1991. The ecology of substrate-associated juveniles of the genus Sebastes. Environ. Biol. Fish. 30:225-243. Miller, D. -I. and J. J. Geibel. 1973. Summary of blue rockfish and lingcod life histories: a reef ecology study; and giant kelp, Maerocystis pynfera, experiments in Monterey Bay, California. Calif. Dep. Fish Game, Fish Bull 158, 137 p. Moorejohn, G. V., J. T. Harvey, and L. T. Krasnow. 1978. The importance of Loligo opalescens in the food web of marine vertebrates in Monterey Bay, California. In Bio- logical, oceanographic and acoustic aspects of the market squid, Loligo opalescens Berry (C. W. Recksiek and H. W. Frey, eds.), p. 67-98. Calif. Dep. Fish Game Fish Bull. 169. Moulton, L. L. 1977. An ecological analysis of fishes inhabiting the rocky nearshore regions of northern Puget Sound, Washington. Ph.D. diss., Univ. Washington, Seattle. WA, 181 p. Myers, R. A., and N. G. Cadigan. 1993. Density-dependent juvenile mortality in marine de- mersal fish. Can. J. Fish. Aquat. Sci. 50:1576-1587. Ralston. S., and D. F. Howard. 1995. On the development of year-class variability in two northern California rockfishes. Fish. Bull. 93:710-720. Ricker, W. E. 1954. Stock and recruitment. J. Fish. Res. Board Can. 11: 559-623. Rosenthal, R. J., V. Moran-O’Connell, and M. C. Murphy. 1988. Feeding ecology of ten species of rockfishes (Scorpae- nidae) from Gulf of Alaska. Calif. Fish Game 74:16-37. Ryan, C. 1986. Feeding habits of brown rockfish, Sebastes auricula- tus , associated with a dock in San Francisco Bay, California. M. A. thesis, San Francisco State Univ., San Francisco, CA, 88 p. Sale, P. K, and D. J. Ferrell. 1988. Early survivorship of juvenile coral reef fishes. Coral Reef 7:117-124. Sano, M. 1997. Temporal variation in density dependence: recruit- ment and postrecruitment demography of a temperate zone sand goby. J. Exp. Mar. Biol. Ecol. 214:67-84. Shulman, M. -J., and J. C. Ogden. 1987. What controls tropical reef populations: recruitment 302 Fishery Bulletin 99(2) or benthic mortality? An example in the Caribbean reef fish Haemulon flavolineatum. Mar. Ecol. Prog. Ser. 39: 233-242. Sissenwine, M. P. 1984. Why do fish populations vary? In Exploitation of marine communities (R. M. May, ed.), p. 59-94. Springer- Verlag, Berlin. Steele, M. A. 1997. Population regulation by post-settlement mortality in two temperate reef fishes. Oecologia 112:64-74. Stephens, J. S., Jr., P. A. Morris, K. Zerba, and M. Love. 1984. Factors affecting fish diversity on a temperate reef: the fish assemblage at Palos Verdes Point, 1974-1981. Environ. Biol. Fish. 11:259-275. Veer, H. W. Van der. 1986. Immigration, settlement and density-dependent mor- tality of a larval and early postlarval 0-group plaice (Pleu- ronectes platessa) population in the western Wadden Sea. Mar. Ecol. Prog. Ser. 29:223-236. Victor, B. C. 1986. Larval settlement and juvenile mortality in a recruit- ment-limited coral reef fish population. Ecol. Monogr. 56: 145-160. Witting, D. A., and K. W. Able. 1995. Predation by sevenspine bay shrimp Crangon septem- spinosa on winter flounder Pleuronectes ame/'icanus during settlement: laboratory observations. Mar. Ecol. Prog. Ser. 123:23-31. 303 Behavioral reactions of northern bottlenose whales (Hyperoodon ampul/atus ) to biopsy darting and tag attachment procedures Sascha K. Hooker Robin W. Baird Sa'ad Al-Omari Shannon Gowans Ha! Whitehead Biology Department Dalhousie University Halifax, Nova Scotia B3H 4J1, Canada Present address (for S. K. Hooker): British Antarctic Survey High Cross, Madingley Road Cambridge, CBE OET, United Kingdom E mail address (for S. K. Hooker). skh@bas.ac.uk Abstract— The effects of invasive or intrusive research techniques need to be thoroughly documented in order to satisfy appropriate standards of animal care. How cetaceans react to either biopsy darting or tag attachment, pro- cedures has been studied for several species, and considerable interspecific variability in responses has been dem- onstrated; however, few studies have compared reactions to both techniques. In the family Ziphiidae (the beaked whales) nothing has been previously reported on responses to either tech- nique. We examined and compared the reactions of northern bottlenose whales ( Hyperoodon ampullatus) to biopsy darting and tagging. Reactions to both these procedures were gener- ally low-level and short-lived; stronger responses were given to hits than to misses. There was no statistical differ- ence in observed response to tag versus biopsy hits. The prior behavioral state of the whales appeared to influence the magnitude of reaction to both hits and misses and thus may be an impor- tant factor to consider in such impact assessment. Whales lying still at the surface showed stronger reactions than traveling or milling animals. Sea state appeared to affect whether there was a reaction to misses. Whales were more likely to respond to a miss in calm sea conditions. No avoidance of the research vessel was observed following a tag or biopsy attempt, and in most cases whales approached the research vessel again within several minutes. Manuscript accepted 16 November 2000. Fish. Bull. 99:303-308 (2001 ). The nonlethal firing of projectiles at whales and dolphins is increasingly being used, both in order to obtain skin and blubber samples (e.g. Lambertsen, 1987) and to attach data-recording or transmitting devices (e.g. Mate and Harvey, 1983; Goodyear, 1993; Baird, 1998; Mate et al., 1998). Data collected with these techniques are important for management and conservation pur- poses but may come at some cost (usu- ally a behavioral disturbance) to the individuals involved. This cost may vary for different species or populations (see e.g. Schneider et ah, 1998), therefore the impacts should be assessed each time a study is conducted. Reactions of various species of ce- taceans to biopsy darting have gener- ally been mild (e.g. International Whal- ing Commission, 1989; Whitehead et al., 1990; Brown et al., 1991; Weinrich et al., 1991, 1992; Barrett-Lennard et al., 1996; Weller et al., 1997). The most common response is a “startle” reac- tion, although the level of reaction var- ies slightly between species, and also between populations and individuals. In contrast, the reaction of cetaceans to tagging with suction-cup-attached tags has been found to vary dramatically. Al- though reactions of killer whales ( Or- em us area) and Dali’s porpoises iPho- coenoides dalli) to the technique were minor (Baird, 1994; Hanson and Baird, 1998), those of bottlenose dolphins (Tur- siops sp.) were strong and relatively long-lasting (Schneider et al., 1998). The family Ziphiidae (the beaked whales) is the second largest family of cetaceans, yet no studies have reported their reactions to tagging or biopsy tech- niques. In this paper we compare the react ions of northern bottlenose whales ( Hyperoodon ampullatus ) to both tech- niques and investigate factors affect- ing the behavioral reactions observed. These results are particularly relevant to agencies that grant research permits (e.g. the National Marine Fisheries Ser- vice in the USA), which often require some discussion of the implications of research techniques in terms of animal care. Additionally, assessment of the magnitude and duration of any behav- ioral response caused by the process of attaching a tag is vital in ensuring that the attachment of the tag does not con- found the behavioral data it records. Materials and methods Field research took place off eastern Canada, approximately 300 km east of Halifax, Nova Scotia, over a subma- rine canyon termed the “Gully" (approx- imate position: 44°N, 59°W ) during June-August 1996-98. All tagging or biopsy attempts were made opportunis- 304 Fishery Bulletin 99(2) Figure 1 Photograph of the suction-cup-attached time-depth recorder and VHF radio tag ready to be deployed from a crossbow. CROSSBOW ticallv from a 12-m sailing vessel, operated under power at speeds of 1-4 knots. The biopsy dart had a 2.5-cm-long, 0.6-cm-diameter cy- lindrical punch fitted with a dental broach (a barbed fila- ment to hold a sample in place) (as illustrated in Barrett- Lennard et ah, 1996), attached to the end of a standard crossbow bolt (total weight 28.5 g). A cylindrical stopper, set 2.5 cm back from the tip of the punch, caused the bolt to rebound after impact with the whale. Bolts were fired from a 67-kg-draw crossbow (Barnett WildCat XL) at a range of 5-15 m. Samples were usually taken from the flank near the dorsal fin. The floating dart was recovered and the skin and blubber sample was removed. The tis- sue was then subsampled for various analyses requiring either skin or blubber or both. The gender of the biopsied whales in this study was determined genetically (Gowans et ah, 2000). The tag measured 20 x 4 x 5 cm and had a 40-cm flex- ible antenna. The foam housing of the tag contained a time- depth recorder (Wildlife Computers, Redmond, WA, or AGO Environmental Electronics, Victoria, BC) and a VHF radio- transmitter (Advanced Telemetry Systems, Isanti, MN, or Telonics, Mesa, AZ). An 8-cm-diameter rubber suction cup (designed for automobile roof-racks, Canadian Tire) was used to attach the tag. The total unit weighed approxi- mately 340 g in air. Tags were attached to a modified cross- bow bolt (weight approximately 25 g) and were deployed with the same crossbow as described above (Fig. 1). The group size and behavioral state of the whales prior to the biopsy darting or tagging attempt were noted. Group size was defined as the number of animals at the surface within five body lengths of each other (chain rule, see Smolker et ah, 1992). Behavioral state was assigned as either logging (lying still or moving slowly in one direc- tion at the surface) or milling or traveling (milling — mov- ing slowly in no consistent direction; traveling — moving in a consistent direction at greater than 2 knots). Whenever possible, each tag or biopsy attempt was videotaped and this recording was used to confirm the consistency of be- havioral categories assigned by different observers in the field for both before tagging attempt behavior and reac- tion. Attempts were classified as a hit or a miss; a hit was defined as contact with the whale and hits were further subdivided as to whether they were successful, i.e. wheth- er biopsies obtained a sample or whether tags remained attached to the whale for more than 30 seconds. Sea state (Beaufort scale) was recorded every hour; sea state at the time of the biopsy or tag attempt was interpolated from these hourly logs. Categories of reaction types were de- fined following Weinrich et al. (1991): 1 No reaction: whale continued to show the same behav- ior as before the biopsy or tagging attempt; 2 Low-level reaction: whale modified its behavior slightly, e.g. dived rapidly or flinched; 3 Moderate reaction: whale modified its behavior in a more forceful manner but gave no prolonged evidence of behavioral disturbance, e.g. tad slap, acceleration, and rapid dive; 4 Strong reaction: whale modified its behavior in a succes- sion of forceful activities, e.g. successive percussive be- haviors (breaches, tail slaps). Hooker et al.: Behavioral reactions of Hyperoodon ampullatus to biopsy darting and tag attachment procedures 305 Table 1 Number of whales showing different reaction-types to tag- ging and biopsy deployment attempts (percentages shown in parentheses). Reaction level Total Low- no. of Event None level Moderate whales Tag-hit 2 (7%) 19 (65%) 8 (28%) 29 Biopsy-hit 3 (11%) 20 (74%) 4(15%) 27 Tag-miss 33 (60%) 21 (38%) 1 (2%) 55 Biopsy-miss 16 (80%) 3 (15%) 1 (5%) 20 Goodness-of-fit G-tests were used to compare reactions to different techniques and under different conditions. Small sample sizes often necessitated pooling between categories. Because reactions to misses tended to be lower- level they were pooled between none and low to moderate, whereas reactions to bits tended to be higher-level and were pooled between none (or low) and moderate. Results Forty-seven biopsy attempts were made on northern bot- tlenose whales in 1996 and 1997 (Table 1). Of these, 27 attempts hit the whale and 20 obtained a skin and blub- ber sample. Six attempts hit, but did not retain a sample (primarily resulting from low hits at the water line). One dart sank after hitting a whale. Video footage was taken of 18 biopsy hits (15 successful) and 11 biopsy misses. Eighty-four tagging attempts were made between 1996 and 1998 (Table 1). Twenty-nine attempts hit the whale but only six successfully attached (one for only 30 sec- onds). Video footage was taken of 15 tag hits (three suc- cessful attachments) and 34 tag misses. The majority of whales showed no or low-level reactions to both tag and biopsy attempts (Table 1). No strong re- actions were observed during our study. The whales react- ed to 93% of the tag hits and 89%- of the biopsy hits, but reacted to only 40% of the tag misses and 20%> of the bi- opsy misses (Table 1). Reactions (categorized as none, low, or moderate) were significantly greater for hits than for misses (G=50.3, PcO.OOl, 2 df). Reactions to tag hits were not significantly different from the reactions to biopsy hits (G=1.51, P=0.47, 2 df). Of the 6 successful tag hits, 2 ani- mals gave a hard tail flick, accelerated and dove (moder- ate reaction), 1 animal accelerated and dove, and 3 dove rapidly or flinched (low-level reactions). The type of reaction to a hit was significantly related to the animal’s behavioral state prior to the tagging or biopsy attempt (comparison for all hits of behavioral state (log- ging vs. milling or traveling) and reaction (none [or low] vs. moderate) G=4.04, P=0.044, 1 df). Low-level reactions were most common for traveling or milling whales, where- as logging whales were more likely to show stronger reac- Table 2 Gender of biopsied whales in= 20) and respective reactions. Reaction level Total no. of whales Gender None Low- level Moderate Male 0 5 2 7 Female 2 9 2 13 tions, especially to tag hits (Fig. 2). There was also a simi- lar relationship between behavior and reaction to misses, although whales rarely showed a moderate reaction to a miss (comparison of behavioral state [as above] and reac- tion [none vs. low or moderate] G=4.02, P=0.045, 1 df. Fig. 2). Reactions to hits of animals in groups were similar to those for lone individuals (G=0.767, P= 0.38, 1 df). There was an effect of sea state on reaction type, but only for missed shots (comparison of sea state [>Beaufort 2 with Beaufort <2], G=4.38, P=0.036 1 df. Fig. 3). This effect was greater for tag misses (G=6.80, P=0.009, 1 df) than for bi- opsy misses (G=2.41, P=0.12, 1 df). No avoidance of the research vessel was observed after tagging or biopsy attempts. For the majority of attempts, animals remained at the surface. If the animals made a shallow dive, they usually returned to the research vessel within a few seconds or were photographed again within a few minutes of the tagging or biopsy attempt (for 82% of all attempts, whales returned to the surface and remained with the research vessel for at least five minutes). There was little difference in the likelihood of encounters ending within five minutes between tagging and biopsy attempts or between hits and misses. Furthermore, encounters in- cluding tagging or biopsy attempts were not shorter in du- ration than encounters without any tagging or biopsy at- tempt (one-tailed f -tests, P>0.05). No significant difference between the reactions of males and females was found (G=2.1, P= 0.35, 2 df, Table 2), but samples sizes were small and did not adequately repre- sent the population. Discussion The response rate of northern bottlenose whales to biopsy hits (89%) was greater than that found for baleen whales (right whales, Eubalaena glacialis, 19%; Brown et ah, 1991; humpback whales, Megaptera novaeangliae, 50%, Weinrich et ah, 1991), but was similar to that recorded for other odontocetes. A 100%’ response rate was found for sperm whales ( Physeter macrocephalus) (Whitehead et ah, 1990) and for bottlenose dolphins (Weller et ah, 1997), and an 81% response rate was found for killer whales ( Barrett- Lennard et ah, 1996). Reactions of bottlenose dolphins consisted of an observable short-term change in behavior (Weller et ah, 1997). Momentary shakes or accelerations were observed for killer whales (Barrett-Lennard et ah, 306 Fishery Bulletin 99(2) 100 80 0) 2 60 o 40 (D 20 0 Tag hits _ 15 - 5 4 1 1 t nuz 3 HL logging milling/traveling Reaction type □ none □ low-level CD moderate 100 1 Reaction type 80 - 60 - □ none □ low-level 40 - ta moderate 20 - 0 Biopsy misses 15 logging milling/trausling Predeployment behavior Figure 2 Relationship between behavioral state and reaction type for tag and biopsy deployments. Reactions during logging behavior were stronger than for other behaviors for both hits (P=0.044) and misses (P=0.045). Percentage of each reaction-type is displayed for each prebiopsy behavioral category (number of reactions of each type are shown above each bar). 1996), whereas sperm whales showed strong startle reac- tions, occasionally involving defecation (Whitehead et al., 1990). Reactions of bottlenose whales seemed most similar to killer whale reactions, i.e. relatively “low-level.” Gen- der-related differences in the levels of reaction have pre- viously been noted for humpback whales (Brown et al., 1994), although no such effects were found in this study. Reactions of northern bottlenose whales to suction-cup tag deployment were also low-level. Bottlenose whales re- acted to 93% of tag hits and 40% of tag misses. Baird (1994) documented only minor reactions by killer whales to crossbow-deployed suction-cup-attached tags: 52% re- action to hits and 26% reactions to misses. Reactions of short-finned pilot whales ( Globicephala macrorhynchus) to the same tags deployed by crossbow consisted of a tail flick and rapid dive (Baird1). Similar variation in reaction has been observed from suction-cup-attached tags deployed by pole. Little reaction was observed to tags deployed on Dali’s porpoise (Hanson and Baird, 1998) or pantropical spotted dolphins ( Stenella attenucitcr, Baird1), whereas bot- tlenose dolphins showed prolonged and intense reactions 1 Baird, R.W. 1999. Unpubl. data. Biology Department, Dal- housie University, Halifax, NS, B3H 4J1 Canada. to tagging attempts, reacting to 100% of hits and 71% of misses (Schneider et al., 1998). The scale of reaction for northern bottlenose whales appears to lie in the middle of this range. As noted, reactions to tag hits were similar to reactions to biopsy hits, despite the fact that tags weigh substan- tially more than biopsy darts. A potential explanation for this is the variation in “tag hit” under our definition: some of the tag hits were glancing blows and so may have had little striking impact, whereas others were direct hits. In contrast, the impact of biopsy darting was more consis- tent (i.e. there were virtually no glancing hits for biopsy darts). In addition, the greater weight of the tags result- ed in a slower delivery speed, thus the force of a heavier, slower-moving tag may have been similar to that of the lighter, faster-moving biopsy dart. Alternatively, animals may have a set reaction to any impact, such that the same response will be elicited unless some threshold is exceeded (perhaps through excessive or repeated impact). As many authors have pointed out, evaluating short and long-term reaction to tag attachment is important in de- termining whether the presence of tags affects the record- ed behavior (White and Garrott, 1990; Walker and Boveng, 1995; Croll et al., 1996). Data were recovered from two of Hooker et al: Behavioral reactions of Hyperoodon ampullotus to biopsy darting and tag attachment procedures 307 Tag misses Reaction type □ none □ low-level a moderate Biopsy misses Reaction type □ none □ low-level moderate Figure 3 Relationship between sea state and reaction type for tag and biopsy misses. Reactions to misses were greater in calm (Beaufort 2) sea states (P=0.036). Per- centage of each reaction-type is displayed for these two Beaufort sea state categories (number of reactions of each type are shown above each bar). the six successful tag attachments and one of these tags was equipped to record velocity (Hooker and Baird, 1999). The whale’s initial reaction to the at- tachment of the tag was a rapid acceleration and dive. The velocity record showed that this initial in- crease dropped within the first two minutes to levels observed for the rest of the deployment ( Hooker and Baird, 1999). The general behavior of all six tagged whales (in terms of surfacing intervals and dive du- rations) was also consistent with that observed from nontagged whales. Thus, although based on a small sample size, it appears that the target animals’ be- havior is modified for only a few minutes in a short- term reaction to the tagging procedure. Previous studies that examined responses to biopsy darting have been criticized because of the potential- ly confounding effect of the research vessel approach (Brown et al., 1994). In our study, the behaviors of the whales when first sighted did not change in any no- ticeable way during the approach of the research ves- sel for photo-identification, and immediately prior to the tagging or biopsy attempt. Thus we are relatively confident that the reactions we observed were due to the tagging or biopsy darting, rather than to the prox- imity of the research vessel. However, the approach of the research vessel may have caused subtle changes in behavior that we did not observe. The major cue to which bottlenose whales react appears to be the physical impact of the tag or bi- opsy, because reaction to hits was much greater than reaction to misses (Table 1). Whales also appear to react to an acoustic or other sensory cue, evidenced by their stronger reactions to tag misses in calm sea conditions (when the hit would be more audible or the splash more likely to be detected above back- ground levels; Fig. 3). The primary factor affecting the reaction of bottle- nose whales to either tag or biopsy stimulus appears to be the behavior of the whale at the time of the stimulus. The relative stimulus of a biopsy or tag is less for whales involved in traveling or milling than for whales lying still at the surface (Fig. 2). Similarly, differ- ences in the level of reaction have been observed for hump- back whales involved in different activities. Reactions of migrating humpback whales were generally lower than those of whales on feeding or breeding grounds (Brown et al., 1994), and mothers or primary escorts of humpback whales (thought to be involved in breeding activity) showed less reaction than other whales on the breeding grounds (Clapham and Mattila, 1993). Such results have wide im- plications for monitoring the effect of various activities (e.g. noise pollution) on cetaceans because the likelihood of reaction may vary depending on behavioral state. Acknowledgments We are grateful to all the crew members of Bctlaena who participated in fieldwork in the Gully. Research in the Gully was funded by the Natural Sciences and Engi- neering Research Council (NSERC), World Wildlife Fund Canada, and the Canadian Federation of Humane Societ- ies. Mere] Dalebout (University of Auckland) determined the gender of biopsied whales. S.K.H. was supported by a Canadian Commonwealth Scholarship, R.W.B. by an NSERC postdoctoral fellowship, and S.G. by an NSERC and an Izaak Walton Killam Memorial Scholarship. Lisa Balance, David Maehr, and Per Palsboll made helpful com- ments on the manuscript. Literature cited Baird, R. W. 1994. Foraging behaviour and ecology of transient killer whales ( Orcinus orca). Ph.D. diss., Simon Fraser Univ., Burnaby, British Columbia, 157 p. 1998. Studying diving behavior of whales and dolphins using suction-cup attached tags. Whalewatcher: J. Am. Cetacean Soc. 3 1 ( 1 1:3-7. 308 Fishery Bulletin 99(2) Barrett-Lennard, L. G., T. G. Smith, and G. M. Ellis. 1996. A cetacean biopsy system using lightweight pneu- matic darts, and its effect on the behavior of killer whales. Mar. Mamm. Sci. 12:14-27. Brown, M. R., P. J. Corkeron, P. T. Hale, K. W. Schultz, and M. M. Bryden. 1994. Behavioral responses of east Australian humpback whales Megaptera novaeangliae to biopsy sampling. Mar. Mamm. Sci. 10:391-400. Brown, M. W., S. D. Kraus, and D. E. Gaskin. 1991 Reaction of north Atlantic right whales ( Eubalaena glacialis) to skin biopsy sampling for genetic and pollutant analysis. Rep. Int. Whal. Comm. Spec. Issue 13:81-89. Clapham, P. J., and D. K. Mattila. 1993. Reactions of humpback whales to skin biopsy sam- pling on a West Indies breeding ground. Mar. Mamm. Sci. 9:382-391. Croll, D. A., J. K. Jansen, M. E. Goebel, P. L. Boveng, and J. L Bengtson. 1996. Foraging behavior and reproductive success in chin- strap penguins: the effects of transmitter attachment. J. Field Ornith. 67:1-9. Goodyear, J. D. 1993. A sonic/radio tag for monitoring dive depths and underwater movements of whales. J. Wild], Manage. 57: 503-513. Gowans, S., M. L. Dalebout, S K. Hooker, H. Whitehead. 2000. Reliability of photographic and molecular techniques for sexing northern bottlenose whales (Hyperoodon ampul- latus). Can. J. Zool. 78:1224-1229. Hanson, M. B., and R. W. Baird. 1998. Dali’s porpoise reactions to tagging attempts using a remotely-deployed suction-cup tag. Mar. Tech. Soc. J. 32(2): 18-23. Hooker, S. K., and R. W. Baird. 1999. Deep-diving behaviour of the northern bottlenose whale, Hyperoodon ampullatus (Cetacea: Ziphiidae). Proc. Roy. Soc. London B 266:671-676. International Whaling Commission. 1989. Report of the sub-committee on small cetaceans. Rep. Int. Whal. Comm. 39:117-129. Lambertsen, R. H. 1987. A biopsy system for large whales and its use for cyto- genetics. J. Mammal. 68:443-445. Mate, B. R., and J. T. Harvey. 1983. A new attachment device for radio-tagging large whales. J. Wildl. Manage. 47:868-872. Mate, B. R., R. Gisiner, and J. Mobley. 1998. Local and migratory movements of Hawaiian hump- back whales tracked by satellite telemetry. Can. J. Zool. 76:863-868. Schneider, K., R. W. Baird, S. Dawson, I. Visser, and S. Childerhouse. 1998. Reactions of bottlenose dolphins to tagging attempts using a remotely-deployed suction-cup tag. Mar. Mamm. Sci. 14:316-324. Smolker, R. A., A. F. Richards, R. C. Connor, and J. W. Pepper. 1992. Sex differences in patterns of association among Indian Ocean bottlenose dolphins. Behaviour 123:38-69. Walker, B. G., and P. L. Boveng. 1995. Effects of time-depth recorders on maternal foraging and attendance behavior of Antarctic fur seals ( Arctocepha - lus gazella). Can. J. Zool. 73:1538-1544. Weinrich, M. T., R. H. Lambertsen, C. S. Baker, M. R. Schilling, and C R. Belt. 1991. Behavioural responses of humpback whales (Megap- tera novaeangliae ) in the southern gulf of Maine to biopsy sampling. Rep. Int. Whal. Comm. Spec. Issue 13:91-97. Weinrich, M. T., R. H. Lambertsen, C. R. Belt, M. R. Schilling, H. J. Iken, and S. E. Syrjala. 1992. Behavioral reactions of humpback whales Megaptera novaeangliae to biopsy procedures. Fish. Bull. 90:588-598. Weller, D. W., V. G. Cockcroft, B. Wursig, S. K. Lynn, and D. Fertl. 1997. Behavioral responses of bottlenose dolphins to remote biopsy sampling and observations of surgical biopsy wound healing. Aquat. Mammals 23:49-58. White, G .C., and R. A. Garrott. 1990. Analysis of wildlife radio-tracking data. Academic Press, San Diego, CA, 383 p. Whitehead, H., J. Gordon, E. A. Mathews, and K. R. Richard. 1990. Obtaining skin samples from living sperm whales. Mar. Mamm. Sci. 6:316-326. 309 Abstract— We estimated size-specific depth distributions and commercial bottom trawl fishery selectivities for Dover sole (Microstomus pacificus), shortspine thornyhead (Sebastolobus alascanus), longspine thornyhead (S. altivelis), and sablefish ( Anoplopoma fimbria) along the U.S. west coast. Depth distributions are size-specific be- cause fish migrate ontogenetically to deep water. With ontogenetic migra- tion, fishery selectivities of commercial bottom trawls depend on depth of fish- ing because large fish are most common in deep water. Depth distributions were similar for northern and southern areas and for males and females. Results show ontogenetic migration in sablefish, sug- gest a possible weak ontogenetic mi- gration in longspine thornyhead, and confirm ontogenetic migration patterns already reported for Dover sole and shortspine thornyhead. Fishery selec- tivities varied among species, between areas, and changed dramatically over time for most species as fishing effort moved into deep water. Our approach used biological data collected during research bottom trawl surveys but was generally not affected by size selectiv- ity of bottom trawl survey gear. Uncer- tainty in our commercial bottom trawl selectivity estimates was mostly from length-specific capture probabilities (or vulnerabilities) for fish in the path of commercial bottom trawls. Our esti- mates complement selectivity estimates from stock assessment models. The approach may be useful whenever the geographic distribution of fish depends on size or age, fishing effort is not ran- domly distributed geographically, and survey estimates of fish density, bathy- metric data, and commercial fishing effort information are available. Manuscript accepted 11 October 2000. Fish. Bull. 99:309-327 (2001 ). Depth distributions and time-varying bottom trawl selectivities for Dover sole {Microstomus pacificus), sablefish {Anoplopoma fimbria), and tliornyheads {Sebastolobus alascanus and 5. altivelis) in a commercial fishery Larry D. Jacobson Southwest Fisheries Science Center National Marine Fisheries Sen/ice, NOAA P.O Box 271 La Jolla, California 92038 Present address: Northeast Fisheries Science Center National Marine Fisheries Sen/ice, NOAA 166 Water Street Woods Hole, Massachusetts 02543-1026 E-mail address: larry.|acobson@noaa gov Jon Brodziak Jean Rogers Northwest Fisheries Science Center National Marine Fisheries Service, NOAA 2030 SE Marine Science Drive Newport, Oregon 97365-5296 In our study, we estimated depth dis- tributions and fishery selectivities for four demersal fish species taken in com- mercial bottom trawls: Dover sole ( Micros- tomus pacificus), shortspine thorny- head ( Sebastolobus alascanus), long- spine thornyhead (S. altivelis ), and sable- fish (Anoplopoma fimbria). The fishes in our study were all valuable com- ponents of the deep-water commercial bottom trawl fishery off Washington, Oregon, and California (Pacific Fishery Management Council, 1998). Depth dis- tributions for many fishes in the deep- water fishery depend on length and age because of ontogenetic migration (move- ment to deep water as fish grow and age, Jacobson and Hunter, 1993; Jacobson and Vetter, 1995). Depth distributions and ontogenetic migration are impor- tant because they affect many aspects of the deep-water fishery, including se- lectivity of commercial bottom trawls, which are the primary fishing gear. Fishery selectivities measure the rela- tive intensity of fishing mortality on fish of different size or age (Megrey, 1989). Fishery selectivities depend on size for fishes in the deep-water bottom trawl fishery (Perez-Comas, 1996) because of factors that include size and shape of mesh, size and shape of fish, orientation of netting, twine material (Wileman et al., 1996), and (as shown below) depth of fishing. In many length-structured stock assessment models, for example, the size- specific fishing mortality rate (FyL) in year y for fish in length class L is sep- arated into the product of year-specific fishing mortality (F ) and size-specific selectivity parameters (sL), so that F L= Fv sL (Megrey, 1989). Selectivities are typically scaled so that the selectivity for a reference size or age is one (Deriso et al., 1985; Methot, 1990). By convention, we scaled selectivities so that the length group with the highest fishing mortality rate had a selectivity of one. Selectivities determine how fishing af- fects the size and age structure of a fish stock. They are used in stock assessment models to relate length and age compo- sition data from catch samples to length and age composition of the stock. They are important in predicting effects of harvest rates (Legault, 1998) and in cal- culating biological reference points (e.g. *0.1> Fr*p> *35*. Fn,a.V See Clark’ 1991 > used to recommend catch levels. At the policy and legal levels, they are often 310 Fishery Bulletin 99(2) involved in defining overfishing and rebuilding overfished stocks as required under U.S. law (Restrepo et al., 1998). Estimating selectivity patterns for commercial fishing is a central issue in use of most stock assessment models based on forward simulation calculations (e.g. Deriso et ah, 1985; Methot, 1990; Fournier and Archibald, 1982; Jacobson et ah, 1994). Changes in fishery selectivity patterns over time may be difficult to measure if length or age composi- tion data are not available for some years. When fishery length or age composition data are available, they can often be explained equally well by many different assumptions about fishery selectivity and population length or age com- position. To understand this, consider the catch in number (CL) of a single size group (length L) from a population. If the fishing mortality rate (F) is low and the selectivity for the size group is sL, then CL~ NrFv sL. Even if Fv is known, the resulting catch CL could be from a high NL and low sL, low Nl and high sL, or an infinite number of intermediate combinations. Problems are compounded if the operation of the commercial fishery and selectivity parameters have changed over time (Sampson, 1993; Brodziak et al., 1997; Rogers et al., 1997) or if natural mortality is also a function of size or age. For example, Tagart et al. ( 1997) found that scarcity of large female fish in fishery length-composition data was explained equally well by two models. One model had constant natural mortality and fishery selectivity de- creased with size. The other model had constant fishery se- lectivity and natural mortality increased with size. In our study, we estimated fishery selectivities for the commercial bottom trawl fishery using a new approach that complements estimates from stock assessment models. Our approach is based on information available in many fish- eries, including data from bottom trawl surveys, informa- tion about bathymetry of fishing grounds, fishing effort data from logbooks, and length- or age-specific vulnerabil- ities to commercial fishing gear from field experiments. First, we used Jacobson and Hunter’s (1993) method with our bottom trawl survey and bathymetric data to estimate depth distributions for fish of different lengths. Next, we used a new method based on commercial fishing logbook data, bathymetric information, length-specific vulnerabili- ties (from field experiments with commercial fishing gear) and depth distributions to estimate fishery selectivities in the commercial bottom trawl fishery. Our approach may be useful whenever the geographic distribution of fish de- pends on size or age, when fishing effort is not randomly distributed geographically, and when both survey densities and commercial fishing effort data are available. Our results show clear differences in commercial fish- ery selectivities among species, areas, and over time. In addition, our analysis provides new information on depth distributions of sablefish and more precise understanding about depth distributions of Dover sole and shortspine and longspine thornyheads. MateriaSs and methods All depths in this study are measured in fathoms (fm). Our study area was the continental shelf and upper continen- tal slope at depths of 100-700 fm (equivalent to 183-1280 m) along the west coast of the U.S. between 36°00' and 48°30'N (Fig. 1). We divided the study area near the Ore- gon-California border into southern (36°00'N to 43°00'N) and northern (43°00'N to 48°30'N) subareas to account for geographic differences in groundfish habitat, bottom trawl fishery and logbook data, and to accommodate areas defined for management of the groundfish fishery. The boundary 43°00'N separates the Eureka and Columbia INPFC (International North Pacific Fisheries Commis- sion) management areas. Areas (km2) of each 100-fm stratum (estimated from spherical projections at sea level) were the same as those used by the National Marine Fisheries Service (NMFS) to estimate fish density and swept area abundance (Lauth1). The shallowest depth stratum in our study (100-199 fm) was relatively larger in the northern subarea (24%) than in the southern subarea (16%, Fig. 1; Tables 1 and 2). Fishing effort shifted into deep water earlier in the south (Tables 1 and 2 ). Fishing effort data from the southern subar- ea were collected mostly from California logbooks, whereas fishing effort data from the northern subarea were mostly collected from Oregon and Washington logbooks. Survey data Data from eight NMFS bottom trawl surveys on the upper continental slope in our study area were used to estimate depth distributions (Table 3). Each survey was conducted during October-December from the National Oceanic and Atmospheric Administration (NOAA) ship Miller Freemcm (e.g. Lauth, 1997a, 1997b; Lauth, 1999). As a group, the surveys covered the entire study area (Fig. 1). A NMFS standard Nor’eastern otter trawl net with a 27.2-m headrope, 37.4-m groundgear, 89-mm codend mesh and a 32-mm mesh liner was used in each bottom trawl survey. In each survey, bottom trawl stations were allo- cated roughly in proportion to the area of 100-m depth strata (100-199, 200-299, 300-399, 400-499, 500-599, and 600-699 fm). Tows with poor gear performance, out- side the study area, and at depths greater than 699 fm or less than 100 fm were excluded from our study. Lengths of Dover sole captured in surveys were recorded as total length (TL) in mm. Lengths of other species were recorded as fork length (FL) in mm. Fishing effoit data Bottom trawl fishing effort data (hours towed) for the northern (Table 1) and southern area (Table 2) were obtained from logbooks submitted by commercial vessels operating out of ports in Washington (1985-97), Oregon ( 1978-97), and California ( 1978-96). The fishing effort data in our study were nominal (as reported) hours towed for bottom trawl tows in which the catch of Dover sole, thorny- heads, or sablefish was greater than zero. 1 Lauth, R. 1998. Personal commun. Alaska Fisheries Sci- ence Center, National Marine Fisheries Service, 7600 Sand Point Way, BIN C15700, Seattle, WA 98115-0070. Jacobson et al : Depth distributions and time-varying selectivity for various bottom fishes 311 Figure 1 (A-C): study areas, 100-199, 200-299, 300-399, 400-499, 500-599, and 600-699 fm depth contours, and location of National Marine Fisheries Service bottom trawl survey tows used to estimate gear selectivities in the commercial bottom trawl fishery. Letters A-H are map symbols defined in Table 3 that identify locations of tows from different bottom trawl surveys. Depth distributions All calculations were based on fish length (two centimeter length groups), rather than fish age, because insufficient survey age data (see below ) were available. The smallest and largest length groups in our analysis were “plus” groups. For example, a plus group of 20 cm at the low end of length com- position would include fish 20 cm FL and smaller. We chose the largest and smallest length groups to use the widest pos- sible range in lengths and to achieve reasonable precision and smoothness in commercial bottom trawl selectivity and depth distribution estimates for large and small fish. For each species, depth distributions in the total popu- lation were estimated by conditional probabilities *p(cl \L) which gave the odds, based on data from bottom trawl sur- vey s, of finding a fish of length L at depth cl in the popula- tion (Jacobson and Hunter, 1993). Following Jacobson and Hunter (1993), we used Bayes’s theorem and data from a single bottom trawl survey in the estimator: ^ i r > _ Ps(L,d) _ ps(L | d) p s(d) pKCl Lj) — — where the joint probability distribution pjL,d) gives the probability that a randomly selected fish taken in bottom trawl survey s was length L and from depth stratum cl. PjL) is the probability that a randomly selected fish taken in the survey was length L. Other terms are defined below. It is important to note that sp(d\L) refers to an esti- mate for the total population based on data from survey s (leading superscript notation), and terms on the right- hand side of the equation refer to the portion of the popu- lation selected by the gear used for survey s (trailing sub- script notation). The total and surveyed populations differ because survey bottom trawls tend to select fish of cer- tain size or ages and, depending on a variety of conditions, length composition data from survey catches will differ 312 Fishery Bulletin 99(2) R from the length composition of the population (Gunderson, 1993). Our estimates of depth distributions for the total population sp(d\L) were generally unaffected by bottom trawl survey gear selectivity because selectivity of the sur- vey gear affects both PJL,d) and PJL) equally and “can- cels out.” This important point is explained further below, after other terms in Equation 1 are defined. For each bottom trawl survey, species, and depth stra- tum, length composition of the surveyed population pjL ) d) was calculated as a weighted average of length composi- tion data from each tow in the stratum: ^psl(L\d)wsdl p (L\d) = — where pst(L \ d) = the length distribution from tow t, and wsdt = the tow catch rate (fish/m2). Tow catch rates were computed as ws d t = nsd tl asd t, where nsdt is the total number of fish caught and asd t is the area swept (width of the net times distance towed). Jacobson et al.: Depth distributions and time-varying selectivities for various bottom fishes 313 The marginal distribution pjd) gives the proportion of the surveyed population (all sizes and ages) in depth stra- tum d based on data from the bottom trawl survey: The length distribution of the surveyed population pjL) was calculated by summing the joint distribution for depth and length across depth strata: pSd) P s ( Zy ) - ^ ps(L,d). (4) where Ws_rf = the average (weighted by area swept) catch rate in stratum d\ and As :l = the total area (km2) of the survey stratum. Data collected from the surveyed population on the right-hand side of Equation 1 can be used to estimate depth distributions for the total population because bot- 314 Fishery Bulletin 99(2) Table 1 Depth stratum area (in 1000 km2), percentage of total area, and nominal fishing effort (h/yr) by depth stratum for bottom trawls in the northern subarea (43°00'N-48°30'N) during 1978-96. Nominal fishing effort was calculated from Oregon and Washington bottom trawl logbook data as total hours trawled for trips catching any thornyheads, Dover sole, or sablefish. Depth (fm) Year 100-199 Area 5.213 % total area 24 200-299 4.159 19 300-399 3.131 15 400-499 2.970 14 500-599 3.055 14 600-699 2.925 14 1978 4224 1059 269 6 0 0 1979 4808 2991 1744 117 0 0 1980 1910 1277 811 46 5 0 1981 3669 1725 1215 118 0 0 1982 6955 5172 2252 263 16 0 1983 5942 4211 2089 354 23 0 1984 4845 4542 2026 235 0 0 1985 9086 6568 3017 1224 14 0 1986 6541 5680 1934 237 0 0 1987 9083 7639 2864 425 0 0 1988 12,762 12,874 5293 706 31 6 1989 15,125 17,458 7609 1792 1793 45 1990 13,820 14,070 8571 7020 4674 314 1991 19,346 20,148 13,346 7547 2976 221 1992 15,063 15,191 12,977 12,233 5467 595 1993 22,571 20,027 14,144 13,202 10,498 2262 1994 13,531 13,569 10,239 12,773 10,531 1634 1995 13,318 10,225 8502 11,117 15,580 2578 1996 13,539 10,867 8745 9831 12,831 1673 tom trawl survey gear selectivity cancels out. To prove this important point algebraically, note that and p{L,d) Id (L,d)- 25 30 35 40 45 50 55 60 Fork length (cm) Figure 6 Commercial bottom trawl (472-inch mesh codend) fishery selectivity estimates for sablefish (sexes combined) in the northern and southern subareas during 1978 and 1996. Selectivity curves for other years were intermediate. Esti- mates for 1978 are biased owing to commercial trawls in the fishery with smaller mesh codends. Smooth lines show trends and were fitted to estimates by locally weighted regression smoothing (LOESS). 1.0 0.8 ^ 0.6 > O Q) CD 0.2 0.0 20 25 30 35 40 45 Total length (cm) Figure 7 Commercial bottom trawl (472-inch mesh codend) fishery selectivity estimates for Dover sole (sexes combined) in the northern and southern subareas during 1978 and 1996. Selectivity curves for other years were intermediate. Esti- mates for 1978 are biased due to commercial trawls in the fishery with smaller mesh codends. Smooth lines show trends and were fitted to estimates by locally weighted regression smoothing (LOESS). as Bayesian priors for selectivity parameters estimated by stock assessment models (Metliot, 1990). Our estimates of depth distributions reflect conditions during the autumn (October-December). Our study did not include data collected during other seasons that could be used to test hypotheses about seasonal migrations be- tween deep and shallow water (Alverson, 1960). Addition- al bottom trawl survey data collected at different times of the year with a variety of vessels and trawl gears are available (Lauth3) and could be used to measure seasonal differences in depth distributions. Our analysis used data from bottom trawl surveys to estimate depth distributions. Our approach might be ap- plicable to other types of surveys as long as densities of organisms in each strata can be calculated on an relative or absolute basis. Although survey gear selectivities can- cel out in calculations, it is important that the survey gear be relatively efficient for length groups used in cal- culations. Otherwise, density estimates used to calculate depth distributions may be too variable. The assumption that survey bottom trawl selectivities are constant with depth is important in calculating depth distributions p(d\L) because the proof that survey selec- tivities cancel out depends on the assumption. It is possi- ble for example, that small fish might evade bottom trawls by hiding in rubble or depressions. If there were more rubble or depressions in deep water than in shallow wa- ter, then bottom trawl survey selectivities would change with depth, and depth distribution estimates (as well as commercial fishery bottom trawl selectivity calculations) would be affected. The magnitude of any possible problem would depend on a variety of factors (e.g. the relative abundance of small fish at depths with more or less rub- ble) and cannot be predicted in general. However, factors that affect selectivity of survey bottom trawls (including herding, escapement under the footrope, escapement over the top of the net, and escapement through meshes) are not important in calculating depth distributions, even if they depend on fish size, as long as they remain the same for all depth strata. Sensitivity of selectivity estimates to errors in fishing effort data Our estimator for fishery selectivities does not depend on knowing total effective commercial fishing effort (Ey d) for each depth stratum. It does depend on knowing the pro- portion of total effective fishing effort (which is propor- tional to fishing mortality) in each stratum. This means that unreported fishing effort would not affect our calcu- lations unless there were differences in depth of fishing among fishermen who did and did not turn in log data, differences among states in depth of fishing and propor- tion of fishermen who submit log data, or differences in logbook reporting rates among fishermen who fish in dif- ferent areas or at different depths. :l Lauth, R. 1998. Personal commun. Alaska Fisheries Sci- ence Center, National Marine Fisheries Service, 7600 Sand Point Way, BIN C 15700, Seattle, WA 98115-0070. Jacobson et al.: Depth distributions and time-varying selectivity for various bottom fishes 323 Our approach to calculating fishery selectivities for commercial bottom trawls is based on the assumption that nominal fishing effort data in each stratum from log- books is a relative measure of commercial fishing mortali- ty. Bias in fishing effort data as a measure of relative fish- ing mortality (e.g. due to differences among depth strata in average fishing power) would affect our estimates of fishery selectivities for commercial bottom trawls. It would be better to use a standardized measure of fishing effort for each stratum adjusted for differences in season, gear, vessel size, engine size, skipper skill, target spe- cies, bycatch, or other operational characteristics of the vessels (Hilborn and Walters, 1992). We hypothesize that this is a minor problem in interpretation of our results, however, because the nominal fishing effort data we used showed clear and substantial shifts towards deep water (Tables 1-2). Sensitivity of selectivity estimates to vulnerability estimates from field studies Vulnerability estimates were the most important uncer- tainty in our estimates of fishery selectivities for com- mercial bottom trawls (but did not affect estimates of depth distributions because they were not used to calcu- late depth distributions). We conducted extensive sensitiv- ity analysis for each species and determined that a 1-cm change in the assumed length at 50% vulnerability (L50 in Table 4) shifted the commercial bottom trawl selectivity curves in the same direction by about 1 cm. The vulnerability parameters used in our analysis were for 4V2-inch mesh which is the current legal minimum mesh size in bottom trawls along the west coast, but smaller mesh was used by some vessels during earlier years (Pacific Fishery Management Council, 1998). Chang- es over time in commercial mesh size would affect our calculations for early years and small fish. The extent of the potential problem is unknown because no vulnerabil- ity parameters were available for small mesh gear (e.g. 3-inch mesh) and because there is no information about proportions of total fishing effort by trawls with different mesh size during early years. The potential problem is not important in estimating selectivity of commercial bottom trawls during recent years because recent mesh size regu- lations make 4V2-inch mesh standard in the fishery. Results from Perez-Comas (1996) were valid estimates of selectivities (as defined in his study) but likely underes- timate vulnerability (as defined in our analysis) and bias our estimates for small fish. Precise definition of terms is important in this regard. According to Perez-Comas ( 1996), his estimates measured “the differential retention of certain sizes of fish after they come in contact with the gear” (Gulland, 1983). In our words, Perez-Comas measured “the relative probability of capture given that a fish entered the mouth of the trawl.” These definitions are similar but not identical to our definition of vulnerabilities as “the relative probability of capture given that a fish is in the path of the trawl.” In particular, Perez- Comas’s ( 1996 ) definition differs from ours to the extent that fish of different sizes have different probabilities of moving « Southern 1978 10 15 20 25 30 35 40 45 50 Fork length (cm) Figure 8 Commercial bottom trawl (4V2-inch mesh codend) fishery selectivity estimates for shortspine thornyhead (sexes com- bined) in the northern and southern subareas during 1978 and 1996. Selectivity curves for other years were interme- diate. Estimates for 1978 are biased owing to commercial trawls in the fishery with smaller mesh codends. Smooth lines show trends and were fitted to estimates by locally weighted regression smoothing (LOESS). 0.0 Iiiiii — i i i 10.0 12.5 15.0 17.5 20.0 22.5 25.0 27.5 30.0 Fork length (cm) Figure 9 Commercial bottom trawl (4'/2-inch mesh codend) fishery selectivity estimates for sablefish (sexes combined) in the northern and southern subareas during 1978 and 1996. Selectivity curves for other years were intermediate. Esti- mates for 1978 are biased owing to commercial trawls in the fishery with smaller mesh codends. Smooth lines show trends and were fitted to estimates by locally weighted regression smoothing (LOESS). 324 Fishery Bulletin 99(2) away from or into the path of the net and to the extent that escape after entering the net depends on fish size. Perez-Comas ( 1996) used a sophisticated estimation pro- cedure but, conceptually, his vulnerability estimates were proportional to the length-specific ratios of catch rates in 4V2-inch and 3-inch mesh (i.e. VL proportional to CL/KI, where CL and KL are catch rates with 4V2 and 3-inch gear). It is likely that catch rates for small fish in 3-inch mesh Table 7 Average depth distributions, i.e. probabilities of depth given length or p(d \L), and CVs for shortspine thornyhead (sexes combined) between 36°00'N and 48°30'N lat and between 100 and 699 fm during October-December based on eight National Marine Fisheries Service bottom trawl surveys. Length groups and depth intervals defined as in Table 5. All eight surveys took 14-50 cm shortspine thornyhead. The symbol “ — “ means that the CV could not be calculated because the average depth distribution was zero. Depth intervals (fm) Fork length (cm) 100-199 200-299 300-399 400-499 500-599 600-699 Depth distributions 14 0.3863 0.6127 0.0011 0.0000 0.0000 0.0000 16 0.4221 0.5752 0.0027 0.0000 0.0000 0.0000 18 0.4607 0.5330 0.0061 0.0000 0.0003 0.0000 20 0.4969 0.4816 0.0185 0.0006 0.0024 0.0000 22 0.4349 0.5056 0.0567 0.0002 0.0025 0.0002 24 0.3920 0.4944 0.1079 0.0028 0.0025 0.0006 26 0.3506 0.4383 0.1989 0.0075 0.0028 0.0019 28 0.2780 0.3779 0.3271 0.0117 0.0032 0.0021 30 0.2011 0.3174 0.4373 0.0405 0.0020 0.0016 32 0.1718 0.2694 0.4564 0.0824 0.0130 0.0070 34 0.1214 0.2042 0.4773 0.1309 0.0404 0.0259 36 0.0826 0.1874 0.4169 0.2025 0.0662 0.0445 38 0.0291 0.1544 0.3592 0.2522 0.1168 0.0884 40 0.0190 0.0797 0.3410 0.2175 0.1868 0.1560 42 0.0164 0.0906 0.1919 0.1709 0.2713 0.2589 44 0.0039 0.0518 0.2034 0.1698 0.2908 0.2804 46 0.0056 0.0437 0.1539 0.1608 0.3453 0.2907 48 0.0080 0.0440 0.1429 0.1342 0.3108 0.3602 50 0.0034 0.0386 0.1970 0.1095 0.3033 0.3482 CV 14 0.18 0.11 0.45 — — — 16 0.14 0.10 0.44 — — — 18 0.06 0.05 0.25 — 1.00 - 20 0.08 0.08 0.26 0.45 0.91 — 22 0.09 0.07 0.25 0.66 1.00 1.00 24 0.10 0.06 0.21 0.27 0.76 0.74 26 0.11 0.09 0.11 0.15 0.43 0.65 28 0.10 0.07 0.03 0.20 0.75 0.52 30 0.12 0.07 0.04 0.22 0.67 0.69 32 0.15 0.08 0.05 0.17 0.31 0.35 34 0.19 0.14 0.09 0.14 0.23 0.59 36 0.18 0.16 0.10 0.12 0.34 0.27 38 0.23 0.21 0.10 0.13 0.27 0.25 40 0.43 0.23 0.12 0.09 0.11 0.17 42 0.57 0.25 0.15 0.17 0.11 0.17 44 0.69 0.11 0.12 0.15 0.11 0.13 46 0.66 0.28 0.13 0.16 0.07 0.12 48 0.41 0.26 0.18 0.15 0.11 0.09 50 0.46 0.26 0.15 0.12 0.08 0.13 Jacobson et al : Depth distributions and time-varying selectivities for various bottom fishes 325 (K[ ) were reduced by escapement of small fish through 3-inch meshes so that vulnerabilities estimates for small fish were biased high as well. The extent of the potential bias is unknown but bias was probably low for large fish. It might be possible to refine estimates of vulnerability parameters by carrying out field studies similar to those analyzed by Perez-Comas, but by using small mesh liners in the codends of commercial bottom trawls as the ref- erence standard. Small mesh can cause problems under commercial fishing conditions (Erickson et al., 1996) and it might therefore be preferable to use codend covers to capture small fish as they escape through codends with commercial-size mesh instead. As described above, fish may move away from or into the path of a commercial bottom trawl (herding) to an ex- tent that depends on size (Gunderson, 1993). Herding is complex and depends on net design, size, towing speed, and other factors (Ramm and Xiao, 1995). These variables would affect our calculations of size-specific selectivities in commercial bottom trawls to the extent that the vulner- ability estimates used in our study would fail to measure the relatively probability of capture for fish of different size in front of commercial bottom trawls. We have little information on this topic for west coast groundfish. Loss of small fish through 3-inch codends during the paired bottom trawl experiments used to estimate vulner- abilities likely biased our estimates of bottom trawl selec- tivities in commercial fisheries. Length-composition data for sablefish, Dover sole, and thornyheads from commer- cial gear with 3-inch codends include fewer small fish than length-composition data from survey bottom trawls with small mesh (3.2 cm) liners (Lauth3). We have no other infor- mation about size-specific probabilities of escape of small fish, but studies with survey bottom trawls and a variety of species (e.g. Engas and Godp, 1989a; 1989b; Walsh, 1992) show that small fish do escape. We recommend field studies with commercial bottom trawls and video equipment. Acknowledgments J. A. Perez-Comas (School of Fisheries, Univerisity of Washington) provided advice, information, and unpub- lished estimates of vulnerability parameters for sablefish. R. Lauth and M. Wilkins (Alaska Fisheries Science Center, National Marine Fisheries Service) provided data and advice. D. Sampson (Oregon State University) stimulated our sensitivity analyses. R. Methot (Northwest Fisheries Table 8 Average depth distributions, i.e. probabilities of depth given length or p(cl | L), and CVs for longspine thornyhead (sexes combined) between 36°00'N and 48°30'N lat and between 200 and 699 fm (longspine thornyhead were seldom taken at 100-199 fm) during October-December based on eight National Marine Fisheries Service bottom trawl surveys. Length groups and depth intervals defined as in Table 5. All eight surveys took 12-30 cm longspine thornyhead Depth intervals (fm) Fork length ( cm ) 200-299 300-399 400-499 500-599 600-699 Depth distributions 12 0.0684 0.2512 0.3288 0.1845 0.1671 14 0.0701 0.2082 0.3252 0.2028 0.1938 16 0.0817 0.2024 0.3155 0.2078 0.1925 18 0.0652 0.1831 0.2875 0.2289 0.2353 20 0.0401 0.1701 0.2801 0.2590 0.2509 22 0.0334 0.1545 0.2970 0.2957 0.2195 24 0.0250 0.1344 0.2894 0.3488 0.2024 26 0.0112 0.1081 0.2784 0.4152 0.1871 28 0.0080 0.1225 0.3151 0.3610 0.1934 30 0.0034 0.2403 0.2711 0.2777 0.2076 cv 12 0.34 0.09 0.09 0.07 0.09 14 0.38 0.11 0.13 0.09 0.13 16 0.37 0.14 0.10 0.12 0.13 18 0.33 0.14 0.09 0.10 Oil 20 0.25 0.12 0.09 0.11 0.11 22 0.27 0.13 0.07 0.07 0.13 24 0.34 0.09 0.06 0.04 0.14 26 0.26 0.13 0.08 0.03 0.17 28 0.69 0.15 0.15 0.07 0.23 30 0.46 0.19 0.23 0.16 0.25 326 Fishery Bulletin 99(2) Science Center, National Marine Fisheries Service) pro- vided travel funds for the senior author and ideas. Three anonymous reviewers, the scientific editor, and the man- aging editor made useful suggestions that improved pre- sentation and explanations. Literature cited Alverson, D. L. 1960. A study of annual and seasonal bathymetric catch patterns for commercially important groundfishes of the Pacific northwest coast of North America. Pac. Mar. Fish. Comm. Bull. 4, 66 p. Brodziak, J., L. Jacobson, R. Lauth, and M. Wilkins. 1997. Assessment of the Dover sole stock for 1997. In Pacific Fishery Management Council. Appendix: status of the Pacific coast groundfish fishery through 1997 and recommended biological catches for 1998: stock assessment and fishery evaluation, 135 p. [Available from Pacific Fishery Manage- ment Council, 2130 SW Fifth Avenue, Suite 224, Portland, Oregon 97201.] Clark, W. G. 1991. Groundfish exploitation rates based on life history parameters. Can. J. Fish. Aquat. Sci. 48:734-750. Deriso, R. B., T. J. Quinn II, and P. R. Neal. 1985. Catch-age analysis with auxiliary information. Can. J. Fish. Aquat. Sci. 42:815-824. Efron, B. 1982. The jackknife, the bootstrap and other resampling plans. Society for Industrial and Applied Mathematics, Philadelphia, PA, 92 p. Engas A., and R. Godo. 1989a. The effect of different sweep lengths on the length composition of bottom-sampling trawl catches. J. Cons, int. Explor. Mer 45:263-268. 1989b. Escape of fish under the fishing line of a Norwegian sampling trawl and its influence on results. J. Cons. int. Explor. Mer 45:269-276. Erickson, D. L., J. A. Perez-Comas, E. K. Pikitch, and J. R. Wallace. 1996. Effects of catch size and codend type on the escape- ment of walleye pollack ( Theragra chalcogramma) from pelagic trawls. Fish. Res. 28:179-196. Fournier, D., and C. P. Archibald. 1982. A General theory for analyzing catch at age data. Can. J. Fish. Aquat. Sci. 39:1195-1207. Gulland, J. A. 1983. Fish stock assessment: a manual of basic methods. FAOAViley Ser. Food Agric., vol. 1. Wiley-Interscience, Chi- chester, UK, 223 p. Gunderson, D. R. 1993. Surveys of fisheries resources. John Wiley & Sons, New York, NY, 248 p. Hilborn, R., and C. J. Walters. 1992. Quantitative fisheries stock assessment. Routledge, Chapman and Hall Inc., New York, NY, 570 p. Jacobson, L. D., and J. R. Hunter. 1993. Bathymetric demography and management of Dover sole. N. Am. J. Fish. Manage. 13:405-420. Jacobson, L. D., N. C. H. Lo, and J. T. Barnes. 1994. A biomass based assessment model for northern anchovy Engraulis mordax. Fish. Bull. 92:711-724. Jacobson, L. D., and R. D. Vetter. 1995. Bathymetric demography and niche separation of thornyhead rockfish: Sebastolobus alascanus and Sebastol- obus altivelis. Can. J. Fish. Aquat. Sci. 53:600-609. Lauth, R.R. 1997a. The 1995 Pacific west coast upper continental slope trawl survey of groundfish resources off southern Oregon and northern California: estimates of distribution, abun- dance, and length composition. U.S. Dep. Commer., NOAA Tech. Memo. NMFS-AFSC-81, 156 p. 1997b. The 1996 Pacific west coast upper continental slope trawl survey of groundfish resources off southern Oregon and northern California: estimates of distribution, abun- dance, and length composition. U.S. Dep. Commer., NOAA Tech. Memo. NMFS-AFSC-80, 110 p. 1999. The 1997 Pacific west coast upper continental slope trawl survey of groundfish resources off southern Oregon and northern California: estimates of distribution, abun- dance, and length composition. U.S. Dep. Commer., NOAA Tech. Memo. NMFS-AFSC-98, 284 p. Lauth, R. R., M. E. Wilkins, and P. R. Raymore Jr. 1997. Results of trawl surveys of groundfish resources of the West Coast upper continental slope. U.S. Dep. Commer., NOAA Tech. Memo. NMFS-AFSC-79, 342 p. Legault, C. M. 1998. Linking total catch quotas and allocation schemes. N. Am. J. Fish. Manage. 18: 454~457. Megrey, B. A. 1989. Review and comparison of age-structured stock assess- ment models from theoretical and applied points of view. In Mathematical analysis of fish stock dynamics (E. F. Edwards and B. A. Megrey eds.), p. 8-48. An. Fish. Soc. Symp. 6, Bethesda, MA. Methot, R. D. 1990. Synthesis model: an adaptable framework for analy- sis of diverse stock assessment data. Int. N. Pac. Fish. Comm. Bull. 50:259-277. Methot, R., P. Crone, R. Conser, J. Brodziak, T. Builder, and D. Kamikawa. 1998. Status of the sablefish resource off the U.S. Pacific coast in 1998. In Pacific Fishery Management Council. Appendix: status of the Pacific coast groundfish fishery through 1998 and recommended biological catches for 1999: stock assessment and fishery evaluation, 146 p. [Avail- able from Pacific Fishery Management Council, 2130 SW Fifth Avenue, Suite 224, Portland, Oregon 97201.] Norris, J. G. 1997. Adaptive radiation and sablefish, Anoplopoma fim- bria. In Biology and management of sablefish, Anoplo- poma fimbria (M. E. Wilkins and M. W. Saunders, eds.), p. 99-114. U.S. Dep. Commer., NOAA Tech. Rep. NMFS 130. Parks, N. B., and F. R. Shaw. 1990. Changes in relative abundance and size composition of sablefish in coastal waters of Washington and Oregon, 1979-1989. U.S. Dep. Commer., NOAA Tech. Memo. NMFS F/NWC-188, 38 p. Pacific Fishery Management Council. 1998. Status of the Pacific coast groundfish fishery through 1998 and recommended biological catches for 1999: stock assessment and fishery evaluation, 31 p. [Available from Pacific Fishery Management Council, 2130 SW Fifth Avenue, Suite 224, Portland, OR 97201.] Paloheimo, J. E., and L. M. Dickie. 1964. Abundance and fishing success. Rapp. P.-C. Reun. Cons. Perm. Int. Explor. Mer 155:152-163. Perez-Comas, J. A. 1996. Assessment of mesh size selectivity under commer- Jacobson et al : Depth distributions and time-varying selectivity for various bottom fishes 327 cial fishing conditions. Ph.D. diss., Univ. WA, Seattle, WA, 229 p. Ramm, D. C., and Y. Xiao. 1995. Herding in groundfish and effective path width of trawls. Fish. Res. 24:243-259. Restrepo, V. R., G. G. Thompson, R M. Mace, W. L. Gabriel, L. L. Low, A. D. MacCall, R. D. Methot, J. E. Powers, B. L. Taylor, P. R. Wade, and J. F. Witzig. 1998. Technical guidance on the use of precautionary approaches to implementing National Standard 1 of the Magnuson-Ste- vens Fishery Conservation and Management Act. U.S. Dep. Commer., NOAATech. Memo. NMFS-F/SPO-31, 54 p. Ricker, W. 1975. Computation and interpretation of biological statistics of fish populations. Fisheries Research Board of Canada, Bulletin 191, 382 p. Rogers, J., T. Builder, P. Crone, J. Brodziak, R. Methot, R. Conser, and R Lauth. 1997. Status of the shortspine thornyhead (Sebastolobus alas- canus) resource in 1997. In Pacific Fishery Management Council. Appendix: status of the Pacific coast groundfish fishery through 1997 and recommended biological catches for 1998: stock assessment and fishery evaluation, 122 p. [Available from Pacific Fishery Management Council, 2130 SW Fifth Avenue, Suite 224, Portland, Oregon 97201.] Sampson, D. B. 1993. The assumption of constant selectivity and the stock assessment for widow rockfish Sebastes entomelas. Fish. Bull. 91:676-689. Saunders, M. W., B. M. Leaman, and G. A. McFarlane. 1997. Influence of ontogeny and fishing mortality on the interpretation of sablefish, Anoplopomci fimbria, life history. U.S. Dep. Commer., NOAATech. Rep. NMFS 130:81-92. Sigler, M. F., S. A. Lowe, C. R. Kastelle. 1997. Area and depth differences in the age-length rela- tionship of sablefish, Anoplopoma fimbria , in the Gulf of Alaska. In Biology and management of sablefish, Ano- plopoma fimbria (M. E. Wilkins and M. W. Saunders, eds.), p. 55-63. U.S. Dep. Commer., NOAA Tech. Rep. NMFS 130. Tagart, J. V., J. N. Ianelli, A. Hoffman, and F. R. Wallace. 1997. Status of the yellowtail rockfish resource in 1997. In Pacific Fishery Management Council. Appendix: status of the Pacific coast groundfish fishery through 1997 and rec- ommended biological catches for 1998: stock assessment and fishery evaluation, 146 p. [Available from Pacific Fishery Management Council, 2130 SW Fifth Avenue, Suite 224, Portland, Oregon 97201.] Walsh, S. J. 1992. Size-dependent selection at the footgear of a ground- fish survey trawl. N. Am. J. Fish. Manage. 12:625-633. Wileman, D. A., R. S. T. Ferro, R. Fonteyne, and R. B. Millar. 1996. Manual of methods of measuring the selectivity of towed fishing gears. ICES Coop. Res. Rep. 215, 126 p. 328 Yield-per-recruit analysis for black drum, Pogonias cromis, along the East Coast of the United States and management strategies for Chesapeake Bay Abstract— Black drum, Pogon ias crom is, along the U.S. East Coast is subject to commercial and recreational harvest. However, prior to this study no model- ing had been undertaken to examine the potential for overfishing in the Chesapeake Bay region. We present evidence from yield-per-recruit models that growth overfishing of black drum is unlikely under current fishing prac- tices in this region. Particular attention was given to fishing practices in the Chesapeake Bay region where old, large fish predominate in the commercial and recreational catches (mean age=26 years; mean total length= 108.4 cm; mean weight 22.1 kg). Yield-per-recruit model results showed that growth over- fishing was unlikely in the Chesa- peake Bay region under all but the lowest estimates of natural mortality (M=0.02-0.04). Such extreme low values of M predict potential life spans of 200 years and were dismissed as improb- able— the oldest age recorded for this species is 59 years. Additionally, bio- mass-per-recruit model results indi- cated a 42-59% decrease to current biomass from the unfished stock. The apparent age-specific migration of this stock argues for protection of young fish that have dominated the catch in Northeast Florida. Modeling indicated that growth overfishing could result from heavy fishing on these young ages and would all but eliminate this resource of the northern fishery. Manuscript accepted 3 November 2000. Fish. Bull. 99:328-337 (2001). Cynthia M. Jones Brian K. Wells Old Dominion University Department of Biological Sciences Norfolk, Virginia 23529-0456 E-mail address (for C. M. Jones): qones@odu.edu Black drum ( Pogonias cromis', family Sciaenidae) range in U.S. waters pri- marily from Massachusetts to Florida along the East Coast and, in the Gulf of Mexico, from the west coast of Florida along the northern Gulf to Texas. They form at least three populations, at least two in the Gulf of Mexico (Gold et al., 1995) and one along the U.S. East Coast (Gold and Richardson, 1998; Gold1). This population structure is seen as “isolation by distance” (Gold and Rich- ardson, 1998). East Coast black drum undertake long-range migrations with a general pattern of movement to the north and inshore in spring, and south and offshore in the fall (Richards, 1973; Murphy and Taylor, 1989; Jones and Wells, 1998 (.These seasonal migrations are age-specific and influence exploi- tation patterns differently along the coast. Although the East Coast stock shows a maximum age of 59 years, which indicates low total annual mor- tality of 8-11% (Jones and Wells, 1998), a greater proportion of old fish are found north of Cape Hatteras, and the potential exists for different age-spe- cific mortalities along the range. Along the East Coast, fisheries for black drum predominantly target small, young fish in the south (Music and Pafford, 1984; Murphy and Muller2 and Wenner3) and large, old fish in the north (Jones and Wells, 1998). Although small fish are targeted in the south, large fish are present and are caught occasionally in the recreational fisheries there. In contrast, small fish are rarely present north of Cape Hatteras besides young of the year fish that leave the bays after their first summer. Hence, little fishing mortality occurs to young fish in the northern part of the range. Compared with other exploited sciae- nids, black drum do not support large recreational or commercial fisheries. Along the East Coast between 1979 and 1994, commercial catches aver- aged only 99,510 kg yearly (218,923 pounds I2'4 5 6-5 6-7’8 Virginia and Florida have the greatest average annual land- 1 Gold, J. R. 1995. Personal commun. Center for Biosystematics and Biodiver- sity, Department of Wildlife and Fisheries Sciences, Texas A&M University, College Station, TX 77843. 2 Murphy, M. D., and R. G. Muller. 1995. A stock assessment of black drum, Pogonias cromis, in Florida. Unpubl. manuscript, 20 p. Department of Environmental Pro- tection, Florida Marine Research Institute, St. Petersburg, FL, 33701-5095. 3 Wenner, C. 1995. Personal commun. South Carolina Department of Nat ural Re- sources, P.O. Box 12559, Charleston, SC 29422. 4 Virginia Marine Resources Commission. 1993. Mimeo: Virginia black drum land- ings by gear type 1973-1993. Virginia Marine Resources Commission, 2600 Wash- ington Ave, Newport News, VA, 23607. 5 North Carolina Division of Marine Fisher- ies. 1980-94. Mimeos: North Carolina landings reports ( 1979-1993). North Car- olina Division of Marine Fisheries, More- head City, NC, 28557-0769. 6 Werner, C. 1996. Personal written com- mun. Preliminary South Carolina land- ings reported to the Fisheries Statistics Program 1979-1995. South Carolina De- partment of Natural Resources, P.O. Box 12559, Fisheries Statistics Program, Charleston, SC, 29422-2559. 7 Pafford, J. 1995. Fisheries Statistics Proj- ect: Georgia black drum landings 1972-94. Georgia Department of Natural Resouixes, footnote continued on next page Jones and Wells: Yield modeling for Pogonias cromis 329 ings (averaging 37,000 and 26,000 kg respectively), where- as New Jersey, Delaware, Maryland, North Carolina, South Carolina, and Georgia have average landings of less than 18,000 kg. In contrast, between 1981 and 1994 recreation- al landings averaged 315,000 kg (693,000 pounds)I 2 * * * * * *'9 annu- ally, 300% higher than the commercial catch. Recreational landings north of Cape Hatteras vary from 0.4% to 78% of the annual East Coast recreational catch, reflecting varia- tions in abundance of older fish and in their seasonal mi- gration patterns. With its location at the northern end of the range, Ches- apeake Bay fisheries target black drum that are primar- ily old (26 yr), and large (108.4 cm; 22.1 kg) during a short season; most of the catches occur from April to June (Jones and Wells, 1998). Large fish enter the Bay in April and are caught by the commercial fishery with 33-cm stretch mesh anchored and drifted gill nets. Historically, the commercial market is local, and fillets and roe are a spring treat for residents of the eastern shore of Virginia and Maryland during April and May (Jones et al., 1990). Because of this limited market that becomes saturated, the price drops in late spring and commercial fishermen turn from black drum to pursue more profitable fishes. The recreational fishery usually begins and ends a month later than the commercial fishery, from May to June, and anglers target large trophy fish with hook and line. Although the recre- ational season is short, it occurs before more popular fish enter the Bay, and the fishery supports local business at that time. Thus, the black drum fishery is important to the economies of two of the poorest counties in Virginia, which are located on the eastern shore (Jones et al., 1990). In the mid-Atlantic region, the lack of accurate catch and effort data from the commercial and recreational black drum fisheries makes it difficult to evaluate whether the long-term fluctuations in population abundance and the current decline in abundance of citation-size fish result from natural patterns of dominant year classes or from ex- cess exploitation and subsequent population decline. Re- ports of catch and effort in the commercial fishery have been based, generally, on voluntary reporting. Likewise, the difficulty in sampling this short-season and charter-based recreational fishery has led to estimates of catch and effort that are characterized by extremely broad confidence lim- its. Even so, in response to the concerns of Virginia’s recre- ational anglers to supposed population decline, commercial harvest quotas were imposed on these fisheries in 1992 in the absence of any stock assessment ( Commonwealth of Vir- ginia, 1992, VMRC regulation 4 VAC 20-320-10 et seq.). Yield-per-recruit models can provide the benchmarks for assessing growth overfishing (Gulland, 1983; King, 1995). I ( continued ) Coastal Resources Division, One Conservation Way, Brunswick, GA 31523-8600. s Sutherland, D. 1995. 1979-94 Black drum commercial land- ings for Atlantic Coast states. Fisheries Statistics Division, National Marine Fisheries Service, 1315 East West Highway, Silver Spring, MD 20910. II Fisheries Statistics Division and Economics Division. 1996. Personal commun. Fisheries Statistics Div. and Economics Div., National Marine Fisheries Service, 1315 East West High- way, Silver Spring, MD 20910. Specifically, yield-per-recruit modeling provides reference points to theoretically maximize yield from a cohort ( FMAX ), or increase the number of trophy-size fish in the population. Because FMAX frequently results in unsustainable harvests, an ad hoc benchmark (F01) is calculated to provide more conservative harvest recommendations. However as impor- tant as this modeling is to science-based management, no published application of yield-per-recuit models exists for black drum from the Chesapeake Bay region. In our study we used data from Chesapeake Bay (Jones and Wells, 1998) and Florida (Murphy and Taylor, 1989) to evaluate the effect of fishing mortality and age at first capture on yield-per- recruit models of these fish, especially for management in the Bay. Although more accurate stock assessments result when catch-age or age-structured models such as ADAPT are used, the absence of a time series of aged-catch data precludes their use for this fishery. Until such data become available, the results of yield-per-recruit models can be used now to determine whether regulations such as size limits, catch quotas, and effort limitations, which are already in ef- fect, are necessary to manage this fishery. Materials and methods Yield-per-recruit analysis The Beverton-Holt yield-per-recruit model (Beverton and Holt, 1957) was used to calculate yield-per-recruit curves following the formula Y / R - Fe T T -hKU -lu > w Y-Li ^ F + M + nK (1) where Y/R F M tc tr W,, *o K yield-per-recruit in weight (kg); instantaneous fishing mortality coefficient; instantaneous natural mortality coefficient; summation parameter ( C70=l, Uy=— 3, U2- 3, U3=- 1); mean age (years) at first capture; mean age (years) at recruitment to the fishing area; and asymptotic weight; hypothetical age the fish would be zero length; and the Brody growth coefficient. Computations were performed by using a modification of the computer program B-H3 available in the Basic Fisher- ies Science Programs package (Saila et al., 1988). Parame- ters used in these simulations are summarized in Table 1. The first two parameters, tQ and K, are derived from the von Bertalanffy growth equation for black drum (Jones and Wells, 1998): 1, = 117.3(1- e"(nor"'+2:i’). (2) Because fish aged 1-5 were not available in Chesapeake Bay, our estimate of K( 0.105) was smaller than that obtained by 330 Fishery Bulletin 99(2) Table 1 Parameter estimates or range of values used in yield-per-recruit and biomass simulations for black drum, Pogonias cromis, on the east coast of the U.S. Data taken from Jones and Wells (1998) for the Chesapeake Bay and Murphy and Taylor (1989). Parameter Chesapeake Bay NE Florida Method tc 5-25 yr age composition of catches tr 5 yr lyr life history information to -2.3 yr -1.3 yr growth curve K 0.105 yr 0.124 yr growth curve 117.3 cm 117.2 cm growth curve 27.5 kg 25.5 kg converted from Lx z 0.08-0.12 catch curves and longevity M 0.00-0.12 longevity P 3.11 length-weight regression Murphy and Taylor, (A'=0. 124; 1989 ) for black drum sampled from the north- east coast of Florida. Hence, we also used estimates of K from the north- east coast of Florida in our modeling to ensure that results would reflect the available scientific data from the U.S. East Coast. For both areas, asymp- totic mean weight, Wm, was converted from an allometric weight-length rela- tionship (6=3.11; Jones and Wells, 1998). This slight deviation from iso- metric growth (6=3.0) may result in a small overestimation of yield (less than 7%) which Ricker (1975) dismissed as inconsequential to further calculations. Because we focused on the relative yields that result from varying t0 and F at different levels of M, differences in yield should be even less than this absolute level (Barbieri et ah, 1997). Age of recruitment to the fishing area, tr, was unknown for this fishery and was set to age 1 for the Florida fishery and age 5 for the Chesapeake Bay fishery, a year less than the youngest adult black drum caught in the Bay during our three-year study. Fisheries-based data included Z, F , M, and f . Estimates of the instantaneous total mortality, Z, for fully recruited black drum were obtained from catch-curve anal- ysis and maximum age procedures, and ranged from 0.08 to 0.12 (Murphy and Taylor, 1989; Jones and Wells, 1998). Although we had direct estimates of total mortality, Z, we lacked independent estimates of instantaneous fishing mor- tality, F, and instantaneous natural mortality, M. However, the estimate of Z allowed us to estimate current levels of fishing mortality, FCUR(i), for different values of M, as Fcvru, = Z~M„ (3) where Mt = 0.02-0.12. We estimated the most probable value of M by extrapolat- ing to maximum age estimates realistic for an unfished fishery. This range of M was lower than that predicted with a multiple regression developed by Pauly ( 1980; M= 0.16-0.30). Using our lower estimate of M, we made our modeling more sensitive to potential growth overfishing; therefore management strategies would be conservative. Ricker yield model Ricker’s yield model is used to simulate the potential for growth overfishing over the life of a cohort by mea- suring available biomass at age under various levels of F (King, 1995). Mortality and growth are opposing effects that result in a maximum biomass during the lifetime at the age of maximum biomass, t critical ■ The model equa- tion is from Saila et al. (1988): where Ye = estimated lifetime equilibrium yield referenced to an arbitrary recruitment biomass of 1000 g; Bi = biomass at age; Ft = instantaneous fishing mortality at age; Z( = total mortality at age; G( = growth in weight-at-age; and t- = age where ti is calculated from the age of first capture, tc, to the last fishable age, tL. When calculated at F= 0, the model produces estimates of equilibrium yield for the unfished stock. Computations were performed by using the computer program Ricker modified from the Basic Fisheries Science Programs pack- age (Saila et al., 1988). Parameter values used in simulations are summarized in Table 1. Estimates of growth parameters Bj} and Gl for Chesapeake Bay and Florida black drum were ob- tained from Jones and Wells (1998) and Murphy and Tay- lor (1989). Because of the long life of black drum, we grouped parameters into 5-year intervals to increase com- putation efficiency. Simulations used six values of M (0.02, 0.04, 0.06, 0.08, 0.10, and 0.12) and six levels of F (0.0, 0.02, 0.04, 0.06, 0.08, and 0.10). This model is not used to calculate optimum yield as is the Beverton-Holt yield-per- recruit model. By integrating the area under the curves, reduction in stock biomass at a given level of F can be compared with biomass of the unfished stock, thus demon- strating the loss of trophy-size fish that are prized in rec- reational fisheries. Simulations were done to model two scenarios of fish- ing mortality and their effect on biomass: 1) uniform low F over the life span, and 2) very high F in the first 5-year interval and uniform low F over the remaining lifetime. In the first scenario the chosen level of F was partitioned equally over 12 age intervals. (Because we lacked age-spe- cific estimates of F, the most straightforward approach was to equally partition F across age intervals.) In the sec- ond scenario fish in the first 5-year interval were given an F='2.0 and thereafter experienced the chosen level of F 2 Jones and Weils: Yield modeling for Pogonias cromis 331 partitioned equally over the remaining 11 age intervals. Hence, in the second scenario the lifetime Z was greater than 2.0. The second scenario was chosen to model ex- tremely severe F on young fish that could be experienced from both directed fisheries and bycatch where young fish predominate. Cohort biomass and harvesting time The maximum possible yield for a year class occurs at the age ^tCRITICAL) when the biomass of the cohort is at its maximum in the absence of fishing. For comparison with the Beverton-Holt and Ricker yield-per-recruit modeling results, we estimated t critical f°r black drum following Quinn and Deri so (1999) with the following equation: ^ CRITICAL = ^0 + 1 ^ + ~ j > ) where m = MIK, P = the length-weight allometry coefficient, and t0, K , and M are defined as in Equation 1. Parameter estimates or the range of values used in calcu- lations are listed in Table 1. Age at maximum biomass can be compared to mean age in the catch to indicate whether further juvenesence is possible. To calculate the proportion of potential growth span (Pa) remaining when black drum enter the exploited phase of life (Beverton and Holt, 1957), we used the quantity (Be- verton, 1963): pg = a-ic/Lj, (6) where L^, the asymptotic length, was obtained from Jones and Wells (1998); and Zc, the average length at first cap- ture, was obtained by converting tc to length with the von Bertalanffy growth curve reported for black drum in Ches- apeake Bay (Jones and Wells, 1998) and Florida (Murphy and Taylor, 1989). Both parameters are based on total length in cm. Results Modeling with parameters from Chesapeake Bay Yield-per-recruit curves on F showed that the yield of black drum in Chesapeake Bay could be maximized by decreasing tc to 10-15 yr over most of the range of M (0.06-0.12) and F used in our simulations (Fig. 1; Table 2). The gains in yield-per-recruit could be substantial. For example, at the estimated current levels of fishing mortal- ity for black drum in Chesapeake Bay (FCUR- 0.04-0.06), yields could be increased 58% at M= 0.06 and 89% at M=0.08 by decreasing current tc from 25 yr to 15 yr. Yield-per-recruit curves showed marked peaks only at the lowest levels of M (0.02; 0.04) when tc<10-15 or at higher levels of M when fc<10 (Fig. 1). Otherwise, curves were asymptotic or rising, and FMAX was reached only at the highest fishing mortalities (F MAX> 2.0; Table 2). When M was 0.02, curves peaked for tc up to 20 yr, resulting in Fmax <0.4. However, because an M of 0.02 predicts a maximum age of over 200 yr in an unexploited stock and because there is no indication of such longevity in black drum, we rejected this scenario as improbable. When M was 0.04, curves peaked for fc<15, for ages constituting less than five percent of the catch and well below the mean age (25 y) in the catch in the Chesapeake Bay fishery. At higher values of M when t(,>10, curves were asymptotic or rising and FMAX occurred only at the highest levels of F. Although yields increased continuously with F for M> 0.04, increases in yield beyond F- 0. 1-0.3 were very small. For M> 0.06 and t> 5, estimates of FCUR were below the levels giving maximum potential yield-per-recruit ^ MAX* and F0 l (Fig. 1; Table 2). For M- 0.06, FCUR equals 0.06 at most and F0 l equals 0.07, indicating that, although below the maximum potential yield-per-recruit, estimated cur- rent levels of harvest are only slightly below this more conservative benchmark of F. When M>0.06, F0 1 is great- er than 0.08 and always above FCUR, indicating that cur- rent levels of harvest are below this conservative bench- mark. In contrast, if M<0.04 and t(<10, F0 1 is higher than Fcur (Table 2) indicating that there is some justification for decreasing F. However, as mentioned previously, we be- lieve these levels of M<0.04 to be unrealistically low for this species. Curves of biomass on age showed that biomass de- creased with increases in M or F (Table 3). Lifetime co- hort biomass of an unfished stock decreased by 85% from M= 0.02 to M=0.12. Within a given M, increased F resulted in decreased lifetime cohort biomass. For example, when the most credible combinations of M and FCUR were mod- eled (M= 0.06, Fcur= 0.06; M=0.08, FCUR = 0.04), biomass declined 59% and 42%, respectively, from that of the un- fished stock (Fig. 2). Similar patterns were shown when we modeled heavy fishing in the first 5 years (F=2.0), and uniform low mor- tality was evident thereafter. Curves of biomass on age showed a much larger decrease in biomass with increasing M and F (Fig. 3; Table 4). Maximum biomass at minimum fishing mortality (F=0.02; M=0.02-0.12) was 81-67% less than seen without heavy early mortality. For example, un- der the most likely combinations of M and FCUR for the Chesapeake Bay fishery, biomass was reduced approxi- mately 82-87% (M=0.06 FCUR= 0.06; M=0.08 FCUR=0M). Values of tCRiTCAL estimated by using different values of M were relatively high for black drum in Chesapeake Bay. Increasing M resulted in a decrease in tCRmCAL from 25 yr at M-0.02 to 10 yr at M= 0.12. This finding indicates that, for the range of M considered in our study, maxi- mum theoretical cohort biomass, in the absence of fishing, is achieved before black drum reach age 25. This occurs at the lowest value of M, approximately the mean age of capture in Chesapeake Bay. For the most likely combina- tions of M and F (M= 0.06 FCUR= 0.06; M= 0.08 FCUR= 0.04), t critical declined from 13(M=0.08)-15(M=0.06) yr in the unfished stock to 10 yr in the fished stock. In this example, t critical below the mean age of capture in the Bay, 26 yr, and potential yield is lost to natural mortality. 332 Fishery Bulletin 99(2) K= 0.105 Figure 1 Beverton-Holt yield per recruit curves on F for black drum, estimated tc= 5-25 and M=0.02-0.12 under A'=0.105. The dotted line ( ) in each panel (t =25) represents the estimated current level oft for black drum in the Chesapeake Bay region. FUIV is represented bv the symbol • and F„ ; is represented by the symbol O. Estimated values of Pn were also low for black drum caught in Chesapeake Bay. For L^=117 cm, and the cur- rent estimated / . ( 1 10 cm, corresponding to f =25), P =0.06, i.e. on average, only 6% of their potential growth still re- mains when black drum in Chesapeake Bay enter the ex- ploited phase at age 25. For alternative values of tr equal to 5, 10, 15, 20, and 30 years, values of P„ are 0.46, 0.27, 0.16, 0.09, and 0.03, respectively. Modeling with /C=0.124 Yield modeling was also done to encompass an alterna- tive estimate of growth based on the Brody coefficient ( K ) determined from the northeast Florida fishery (Murphy and Taylor, 1989). Because Chesapeake Bay region catches did not include fish aged 1-5, the estimate of K (0.105; Jones and Wells, 1998) differed slightly from that esti- Jones and Wells: Yield modeling for Pogomas cromis 333 Table 2 Estimates of F0 l and FMAX compared to FCUH from Beverton-Holt yield-per-recruit modeling for various levels of K, M, and tr. The symbol *** indicates that F MAX occurs at the highest values of F. FCUR(i) was calculated from the upper-bound estimate of Z (as F cuR(i))=Z~M i ) and represents the upper-bound estimate of current F. K M tc n.i Fmax Fcur K M tc Fn Fmax Fcur 0.105 0.02 5 0.03 0.06 0.10 0.124 0.02 5 0.05 0.07 0.10 10 0.05 0.09 10 0.06 0.11 15 0.06 0.17 15 0.07 0.23 20 0.07 0.40 20 0.08 25 0.09 *** 25 0.09 *** 0.04 5 0.06 0.08 0.08 0.04 5 0.06 0.10 0.08 10 0.07 0.17 10 0.07 0.24 15 0.09 0.51 15 0.09 1.38 20 0.10 *** 20 0.12 25 0.12 *** 25 0.12 0.06 5 0.07 0.13 0.06 0.06 5 0.08 0.15 0.06 10 0.09 0.31 10 0.10 0.47 15 0.11 *** 15 0.13 20 0.13 *** 20 0.14 25 0.15 25 0.14 0.08 5 0.09 0.17 0.04 0.08 5 0.09 0.21 0.04 10 0.12 0.61 10 0.14 1.11 15 0.15 15 0.17 20 0.17 20 0.19 25 0.19 25 0.20 0.10 5 0.11 0.23 0.02 0.10 5 0.12 0.28 0.02 10 0.15 1.51 10 0.18 15 0.19 15 0.20 20 0.22 20 0.23 25 0.25 25 0.27 0.12 5 0.13 0.30 0.00 0.12 5 0.14 0.39 0.00 10 0.19 10 0.20 15 0.23 15 0.25 20 0.27 20 0.29 25 0.311 25 0.30 Table 3 Lifetime cohort biomass ( g ) from the Ricker biomass model (Saila et al. 1988) under M=0. 02-0. 12, and uniform F=0-0.12. Integra- tion was by rectangular approximation. Simulations were based on an arbitrary starting biomass of 1000 g. F M 0 0.02 0.04 0.06 0.08 0.1 0.12 0.02 180,730 107,915 70,195 49,170 36,595 28,595 23,210 0.04 107,915 70,195 49,170 36,595 28,595 23,210 19,420 0.06 70,195 49,170 36,595 28,595 23,210 19,420 16,675 0.08 49,170 36,595 28,595 23,210 19,420 16,675 14,600 0.10 36,595 28,595 23,210 19,420 16,675 14,600 13,040 0.12 28,595 23,210 19,420 16,675 14,600 13,040 11,750 mated for northeast Florida (K= 0.124) which did include these ages. The values for L were virtually identical from both studies Model results based on this faster growth rate produced similarly shaped yield-per-recruit curves but with slightly higher yields and benchmark values (Fig. 4). At the most probable values of M (0.06; 0.08), yield-per- recruit curves peaked only at tc <10 yr. Otherwise, curves were asymptotic or rising. FMAX at t =5 yr was 0.15 for 334 Fishery Bulletin 99(2) F 0.00 0.02 — — 0.04 0.06 0.08 Figure 3 Ricker biomass curves under conditions of high instanta- neous fishing mortality, F= 2.0, in the first 5 years with low, uniform F=0. 0-0.1 thereafter and M=0.06-0.10. The dash- dot-dot-dash ( — ) and dash-dot-dash ( — — ) lines rep- resent the most likely range of current fishing mortality, Fcur=® .04 and 0.06, respectively. Note that recruit biomass is arbitrarily set at 1000 g. Table 4 Lifetime cohort biomass (g) from the Ricker biomass model (Saila et al. 1988) under M=0.02-0.12, and F=‘2.0 over the first 5 years and low uniform F=0-0. 12 thereafter. Integration was by rectangular approximation. Simulations were based on an arbitrary starting biomass of 1000 g. F M 0 0.02 0.04 0.06 0.08 0.1 0.12 0.02 180,730 20,395 15,785 13,075 11,390 10,270 9495 0.04 107,915 14,750 12,305 10,775 9760 9055 8560 0.06 70,195 11,610 10,225 9315 8675 8225 7875 0.08 49,170 9730 8895 8325 7910 7600 7370 0.10 36,595 8525 8000 7630 7360 7145 6990 0.12 28,595 7730 7380 7130 6940 6785 6660 M= 0.06 and 0.21 for M=0.08 (Table 2), greater than our estimate of Z and FCUR. At tc>10, FMAX occurred at the highest levels of F. At the most probable levels of M, F0 x was greater than FCUR (Table 2). Hence under either FMAX or F0 : and larger K , FCUR was still below that needed to obtain maximum yields from the fishery in Chesapeake Bay. Similarly, model results from this faster growth 0.28. Likewise, this tc was too young for the fishery, and Z and FCUR were consider- ably smaller than the lowest values of FMAX or F0 l Hence, 336 Fishery Bulletin 99(2) even when the full range of M and faster growth were con- sidered, growth overfishing was not likely for the Chesa- peake Bay region fishery. Overall, yield-per-recruit curves showed that a sixfold increase in M resulted in a 50% decrease in yield for both growth rates (Figs. 1 and 4). As M increased, yield-per-re- cruit decreased. For a given M, yield-per-recruit increased to a maximum at an intermediate level of tc. Increases in yield slowed from 5-15 yr and decreased from maxima thereafter at older ages. In only one case (M= 0.02 and A'=0.105), yield-per-recruit increased with increased tc up to 20 yr. In all other cases, yield decreased for fc>20 yr, indicating that, beyond 20 yr, biomass was lost to natural mortality. Discussion Our results indicate that yield-per-recruit for black drum in the Chesapeake Bay region is below its maximum for all but the lowest values of M used in our simulations. For MAO. 04, current fishing mortality was below FMAX. Only when M=0.02 and ft.<15 does the upper bound of FCUR fall above FMAX . We discounted this case of extreme low M because of the unusually long lifetime that it predicts — some 200 years. Yield-per-recruit and economic efficiency could be maximized for black drum in Chesapeake Bay by decreasing tr to 5 years along with higher rates of fishing mortality. However, this may not be the most viable man- agement option for this species for several reasons. First, because the relation between yield-per-recruit and F is essentially asymptotic, harvesting black drum in the Bay at or near FMAX would require a huge increase in fishing effort, making harvest of this species economically inef- ficient, especially with the current low demand for these fish. Besides, benchmarks such as FMAX are no longer thought to provide a sustainable measure of long-term maximum yield from a fishei'y. Second, the current tc may reflect the mean age of migrating adults that are recruited to the fishery. If so, decreasing tc may not be possible because young fish may not undertake migration along the coast, and a decrease in mesh size may result in fail- ure of the net to “gill” the larger fish, with the result that catches would be diminished. Large reductions in biomass, especially of older fish, were shown in biomass modeling. Biomass decreases 42-59%' under the most likely values of mortality (M=0.08, FCur= 0-04; M=0.06, FCUR~- 0.06, respectively) more than that of the unfished stock. Reductions in biomass (up to 87%) are exacerbated when heavy fishing mortality is con- centrated on young fish. Concurrent with these reductions in biomass, is a rapid and dramatic loss of older fish from the stock. This juvenescence occurs quickly — tCRITICAl is reduced from 15 in the unexploited stock to 10 at F=0.02 for M- 0.06, and from 13 to 10 at M- 0.08. At greater F, the decrease in tCRITICAL is even greater and the abundance of older fish diminishes further. Altogether these modeling results show no indication of growth overfishing in the Chesapeake Bay region where old fish are predominantly targeted. Moreover, it is diffi- cult to growth overfish a stock when fishing concentrates on capturing primarily older, larger fish. For example, black drum have already obtained 58% of their lifetime growth in length, and 22% of their lifetime weight when they first recruit to the Chesapeake Bay region at age six (Jones and Wells, 1998). By their mean age of capture in this region, they have obtained 90% of their lifetime growth in length and 51% in weight. Exploited cohorts have already surpassed their maximum growth by the time they enter the Bay region, and thereafter, natural mortality predominates. Cohort biomass has already de- clined from its optimum by the age fish enter the exploited stock in the Bay region. Although these modeling results show no indication of growth overfishing in Chesapeake Bay, they do indicate that black drum are vulnerable when heavy fishing is di- rected to young fish in the southern portion of their range along the U.S. East Coast. We chose a high level of F in the first five years of life to dramatically illustrate the effect of targeted fishing on small fish and the potential effects of bycatch from other fisheries. These simulations clearly indicate the importance of limiting fishing mortality in re- gions where young fish occur. Prior to 1989, black drum landed in the Florida east coast commercial fisheries aver- aged 320 mm (Murphy and Muller2), and 80% of the catch was 4 yr or younger (Murphy and Taylor, 1989), raising the potential of growth overfishing at that time. Capture at this young age also raises concern for recruitment over- fishing, which our modeling does not address, especially when fish are targeted before they can reach sexual ma- turity (age 5). The potential for recruitment overfishing is minimal in areas, such as Chesapeake Bay, where the fishery targets older fish that have reproduced for many years before capture. Moreover, recent bans on gillnetting in Florida and other regulations on black drum fishing since 1989 should preclude recruitment overfishing and help preserve the stock. Models are typically used in management to regulate fishing mortality in order to obtain sustainable harvests from a stock. These regulations have historically resulted in harvests with large biomass that are valued in commer- cial fisheries. In contrast, recreational anglers are not as interested in obtaining maximum biomass as they are in catching fewer, but larger fish. Moreover, increased produc- tion of larger fish occurs when fishing mortality is below F MAX and when recruitment is high. Hence, in the black drum fishery, which is targeted by both commercial and recreational fishermen, management objectives are at cross purposes. The commercial fishery benefits when yields are maximized to the detriment of survival and growth for the trophy-size fish desired by recreational anglers. In the Chesapeake Bay region, fishing mortality is low and sup- ports the objectives of managing the recreational fishery. However, the most influential fishing mortality is on young fish and is not under the control of the Bay region manage- ment agencies, but is controlled by states farther south. The long-range migrations of the East Coast black drum stock argue for a coast-wide management strategy. Through our modeling, we have shown that fishing prac- tices in the Bay region have little impact on the production Jones and Wells: Yield modeling for Pogonias cromis 337 of harvestable biomass and that mortality on young fish drives eventual production available to the Chesapeake Bay region black drum fishery. The supply of fish to the Bay region depends on mortality during the first ten years of life, years when these fish are found off the coasts of the South Atlantic states. Hence, management practices by states south of Cape Hatteras will determine the sup- ply of fish to this coast-wide stock. Acknowledgments We would like to thank Barbara McClellan for her assis- tance with modeling and graphics, and Douglas Vaughan and Michael Murphy for their review of and suggestions for improving this manuscript. This research was funded by a Wallop/Breaux Program Grant for Sport Fish Resto- ration from the U.S. Fish and Wildlife Service through the Virginia Marine Resource Commission, Project F-88-R3 to C. Jones and Mark E. Chittenden Jr. Literature cited Barbieri, L. R., M. E. Chittenden Jr., and C. M. Jones. 1997. Yield-per-recruit analysis and management strate- gies for Atlantic croaker, Mic-ropogonias undulatus, in the Middle Atlantic Bight. Fish. Bull. 95:637-645. Beverton, R. J. H. 1963. Maturation, growth, and mortality of clupeid and engraulid stocks in relation to fishing. Rapp. Reun. P.-V. Cone. Int. Explor. Mer 154:44-67. Beverton, R. J. H., and S. J. Holt. 1957. On the dynamics of exploited fish populations. U.K. Min. Agric. Fish. Food., Fish. Invest, (ser. 2) 19:1-533. Gold, J. R., and L. R. Richardson. 1998. Mitochondrial DNA diversification and population structure in fishes from the Gulf of Mexico and Western Atlantic. J. Heredity 89:404-414. Gold, J. R., L. R. Richardson, C. Furman, and F. Sun. 1995. Mitochondrial DNA diversity and population struc- ture in marine fish species from the Gulf of Mexico. Can. J. Fish. Aquat. Sci. 51 (suppl. 1 ):205-214. Gulland, J. A. 1983. Fish stock assessment: a manual of basic methods. FAOAViley series on food and agriculture, vol 1. John Wiley and Sons, New York, NY, 223 p. Jones, C. M., K. H. Pollock, A. Ehtisham, and W. Hinkle. 1990. Assessment of the black drum recreational fishery, 1989, in Virginia. Old Dominion Univ. Res. Found. Tech. Rep. 90-2, Norfolk, VA, 100 p. Jones, C M„ and B. K. Wells. 1998. Age, growth, and mortality of black drum, Pogonias cromis, in the Chesapeake Bay region. Fish. Bull. 96:451- 461. King, M. 1995. Fisheriesbiology.assessmentandmanagement. Fish- ing News Books, Blackwell, Cambridge, MA, 341 p. Murphy, M. D., and R. G. Taylor. 1989. Reproduction and growth of black drum, Pogonias cromis, in northeast Florida. Northeast Gulf Sci. 10:127- 137. Music, Jr., J. L., and J. M. Pafford. 1984. Population dynamics and life history aspects of major marine sportfishes in Georgia’s coastal waters. Contrib. Ser. 38, Georgia Dep. Nat. Res., 1200 Glynn Ave, Bruns- wick, GE 31523, 382 p. Pauly, D. 1980. On the interrelationships between natural mortality, growth parameters and mean environmental temperatures in 175 fish stocks. J. Cons. Int. Explor. Mer 39:195-212. Quinn, II, T. J., and R. B. Deriso. 1999. Quantitative Fish Dynamics. Oxford Univ. Press, New York, NY, 542 p. Richards, W. E. 1973. Age, growth and distribution of black drum ( Pogonias cromis ) in Virginia. Trans. Am. Fish. Soc. 102:584-590. Ricker, W. E. 1975. Computations and interpretation of biological statis- tics of fish populations. Bull. Fish. Res. Board Canada 191, 382 p. Saila, S. B., C. W. Recksieck, and M. H. Prager. 1988. Basic fisheries science programs. Elsevier, New York, NY, 230 p. 338 Abstract— A learning curve analysis is applied to model the change in effort units with time by the commercial fish- ery for Atlantic cod (Gadus morhua ) in the Barents Sea. The results indicate that fishing efficiency has increased and that the adjusted commercial catch per unit of effort (CPUE) is more strongly related with stock abundance. Manuscript accepted 1 1 October 2000. Fish. Bull. 99:338-342 (2001). Adjustment of commercial trawling effort for Atlantic cod, Gadus morhua, due to increasing catching efficiency Are Salthaug Institute of Marine Research Nordnesgaten 50 PO Box 1870 N-5817 Bergen Norway E-mail address: ares@imr.no Fishing fleets and individual fishing vessels are expected to increase their efficiency with time owing to techno- logical improvements. This is a ser- ious problem when a time series of com- mercial catch per unit effort (CPUE) is used to measure trends in fish stock abundance ( Gulland, 1983 ). A standard effort unit will gradually remove a greater proportion of the stock, and the CPUE corresponding to a given fish stock abundance will increase. Commercial effort units should, therefore, be adjusted to account for gradual changes in catch- ability (Gulland, 1983). A learning curve describes how unit costs decline as organizations gain experience in production (Argot and Epple, 1990). The general form of the learning curve is given by y = ax~b, ( 1 ) where y = the number of direct labor hours required to produce the xth unit; a = the number of labor hours required to produce the first unit; x = the cumulative number of units produced; and b = the parameter measuring the rate at which labor hours are reduced as cumu- lative output increases (Ar- got and Epple, 1990). If a learning curve is used to describe a fishery, the labor hours required to pro- duce the xth unit can be translated into the effort required to catch a certain fraction of the fish stock, termed effort per stock fraction ( epsf ). Time, instead of cumulative catch, is assumed to be a better measure of x because fisher- men learn even if stock abundance and catches are low. In this study, appro- priate learning curves are fitted to the decrease in epsf with time for vessels in the Norwegian bottom trawler fleet in the fishery for Atlantic cod ( Gadus morhua). Effort is then adjusted by using the fitted learning curve, to gen- erate CPUE-indices that better reflect fish stock abundance. Materials and methods To calculate the effort per stock fraction ( epsf) for a given vessel, effort is divided by the ratio of catch and fishable stock biomass (stock fraction caught) in a given time period: where B ? = the estimated weight of fishable biomass; and e = the total effort and c is the total catch (weight) during the time period. The catch and effort data should be independent from the estimate of Bf. Some catch composition criteria should also be introduced to increase the prob- ability that the vessels were part of the targeted fishery during the period. Note that epsf is the inverse of catch- ability. By inserting epsf for y and time (t) for x in Equation 1, then Salthaug: Adjustment of commercial effort for Gadus morhua 339 \n(epsf) = ln(a) - b • ln(f). (3) By fitting a regression line to In (epsf) versus In (t), esti- mates of the parameters ln(al and b in the learning curve are obtained. An unbiased estimator for a is exp(ln(a)+ MSE/2) (Casella and Berger, 1990), where ln(a) is the intercept of the regression line and MSE is the mean squared error from the regression. A learning curve can be estimated at the fleet level by using average epsf of the vessels, and at the vessel level by using an individual vessel’s epsf. These two approaches are somewhat differ- ent because the efficiency of a fleet is increased both by improvement of existing vessels and by addition of new and better vessels. The effort units in a time series of CPUE can be adjust- ed to the level of one of the elements (index time) in the time series as follows: ePsf \ (4) epsf, ) where ef = the adjusted effort at time z; and epsf index = the effort per stock fraction modelled and epsft with the estimated learning curve at the index time and time i, respectively. Note that the learning curve model is a continous function and CPUE is normally given as a discrete time series. Time i should, therefore, correspond to the mid- point of the time interval over which the corresponding CPUE observation is calculated. The catch and effort data were taken from a logbook database collected by the Norwegian Directorate of Fish- eries. Logbooks from the Norwegian bottom trawler fleet have been recorded since 1971 and each individual record includes vessel, species, date, and summarized duration (in hours) and summarized catch (in kilograms) of the trawl hauls each day. Estimates of trawlable biomass (Bf of Atlantic cod in the Barents Sea were taken from VPA (virtual population analysis) estimates of stock biomass at the beginning of each year (ICES1). Only 20% of the 3-year-old cod and 50% of the 4-year-old cod were consid- ered to belong in the trawlable part of the biomass because of mesh selection in the trawls. This selection of young cod corresponded broadly to the retention probabilities in the beginning of the year based on a trawl selection curve. The effort ( e ) is hours of trawling. Because the VPA estimates of stock size are given for the start of the year, the sum- mation of catch and effort for a given year (t ) are over the period from July in year t — 1 to June in year t. In this way the stock estimates from VPA are given for the mid- dle of the time period. Only records with more than 20%> 1 ICES (International Council for the Exploration of the Sea). 2000. Report of the Arctic Fisheries Working Group, Copenha- gen, 23. August-1. September 1999. ICES CM 2000/Assess:3, 311 p. ICES, Palaegade 2-4 DK-1261 Copenhagen K, Denmark. (in weight) cod for a recorded day were used to increase the probability that cod was the target species. Varying the minimum accepted proportion of a species in catches lias been shown to significantly affect estimates of CPUE (Ketchen, 1964). To avoid vessels that did not take active part in the cod fishery, only vessels with at least 10 ob- servations where more than 50% cod were present in a given year were used in the calculation of the fleet’s aver- age epsf. Only cod records from north of 67°N were used because this latitude is the limit of the distributional area for Atlantic cod along the Norwegian coast. Changes in yearly epsf were analyzed and estimates of the parameters in the learning curve were made at the vessel level and at the fleet level (average epsf) in the pe- riod 1971-99. Only one vessel was active during the whole period without being rebuilt, and this vessel was chosen for analysis at the vessel level. The trawl fishery for Atlantic cod in the Barents Sea has existed since about 1920, but the Norwegian trawler fleet did not significantly participate in this fishery until after the end of the Second World War. Year one in the learning process for the Norwegian trawler fleet is therefore set at 1946. In the present study, sufficient data for estimating learning curves exist only from 1971, which is a limited part of the time period in which the Norwegian fishery has existed. This causes the year 1972, the first year where it is possible to estimate epsf , to become year 27 in the learn- ing process. A time series of CPUE for Atlantic cod of the Barents Sea was calculated by using the same catch and effort da- ta as above, and yearly effort was adjusted (Eq. 4) to the level of 1972 by using the estimated learning curve on the fleet level. Only records containing more than 20% cod (in weight) were used. CPUE was set as total catch divided by total effort in the period from July in year t — 1 to June in year t. Results The estimates of annual epsf showed a decreasing trend with time both at the fleet level and at the vessel level (Fig. 1, A and C). The slope and intercept of the fitted regression line at the fleet level (Fig. IB) were 1.742 and 14.272, respectively, and the correlation given as the r2- value was 0.50. At the vessel level the values of the slope and intercept (Fig. ID) were 1.538 and 13.334, respectively and r2 was 0.55. The estimated a was 1895039.044 at the fleet level and 714722.761 at the vessel level. The slopes of the two regression lines were significantly different from zero (P<0.0001), but the two regression lines were not sig- nificantly different from each other (P>0.05) . The time series of annual CPUE became much more similar to the biomass estimate from VPA when effort was adjusted (Fig. 2), especially in the first and last part of the time period. In the middle of the period (after 1980) the trend in adjusted CPUE suddenly became different from that of biomass estimates from VPA, and CPUE values were much higher than biomass in relation to the ratio in the first and last part of the period for some years. 340 Fishery Bulletin 99(2) 8000 A 9,2 B 7000 9 9 9 8,8 9 6000 9 8,4 • 5000 '' - ^ 9 9 9 CO g 4000 © & 8,0 CD 3000 9 9 * . * • 7,6 • * - 9 2000 . * * * • • . * 9 • • 7,2 m • 1000 9 6,8 1970 1976 1982 1988 1994 2000 3,3 3,4 3,5 3,6 3,7 3,8 3,9 4,0 Year Ln(f) 7500 c 9,0 D 6500 • 8,6 9 5500 8,2 9 9 4500 cr * - * . • "co Cl • 1 7,8 9 ^ 3500 9 9 C 9 9 9 9 9 7,4 9 * 9 9 * • 2500 • . 9 9 1500 9 • 9 9 9 ® • * *9 • 9 7,0 9 6,6 1970 1976 1982 1988 1994 2000 3,3 3,4 3,5 3,6 3,7 3,8 3,9 4,0 Year Ln(f) Figure 1 Relation between average effort per stock fraction (epsf) per vessel in the fleet and year (A), epsf for an individual vessel and year (C), log- transformed epfs and time (in ln( years)) for the average vessel in the fleet (B) and for an individual vessel (D ). Note that the first observed time value in B and D is ln( 27 ) because 1946 is assumed to be the first year in the learning process. The slope and intercept for the fitted lines and the correlation between the van ables in B and D are given in the text. Discussion A learning curve ideally requires data points from the init ial phase of the process. For many fish st ocks, catch and effort data, together with an independent estimate of total biomass, exist only for recent years. Given the approximate year the fleet entered the fishery, it is necessary to assume that epsf follows the learning curve pattern from that year onwards. Under this assumption, it is possible to adjust effort backwards in time to the level of one year in the time series by using the estimated learning curve. Tech- nical revolutions, such as the introduction of hydraulic wires, may cause a very dramatic change in the fishery, and if this happens the learning curve should be esti- mated from the time when the new technology appeared. Good knowledge about the history of the fishery is thus necessary. There are also alternative methods for dealing with increases in efficiency with time in a fishery. Gulland (1983) suggested constant monitoring of the changes in the fishing gears by conducting experiments. This solution may, however, be very expensive. The question whether the effort should be adjusted with a learning curve from the fleet level or from the vessel level depends on the resolution of the catch and effort da- ta and on other standardization of effort. If effort is first adjusted within the fleet because of the individual differ- ences in fishing power, adjustment of the effort due to learning should be based on a learning curve from the standard vessel or the group of standard vessels to avoid double standardization. It is important to be aware that individual vessels also may show different learning rates and these may differ from the average learning rate of the fleet. An individual vessel may improve its efficiency by in- creasing the cooperation with other vessels, by buying bet- ter searching equipment and more efficient fishing gear, and by hiring more skilled crew. Salthaug: Adjustment of commercial effort for Godus morhua 341 CD O' 3 U) in 2,4e6 2e6 1 ,6e6 1 ,2e6 8e5 4e5 Year Figure 2 Yearly CPUE in kilograms per hour and trawlable biomass from VPA estimates for original (A) and adjusted (B) CPUE. In B the effort is adjusted to 1972-level bv using the learning curve estimated in Figure IB. An important assumption in this work is that a constant fraction of the stock is caught when the fleet lias a constant efficiency. Many fish stocks, including the Atlantic cod stock in the Barents Sea, have a seasonal cycle in availability to the fishing fleet because of regular spawning and feeding migrations that cause large peaks in the den- sity of fish. The assumption above is therefore violated if the spatial and temporal distribution of effort in the fleet during the year chang- es between years. Correction for changes in the seasonal distribu- tion of effort between years are, however, described by e.g. Gulland (1983) and Gavaris (1980). Around 1979 quotas and fishery regula- tions were introduced in the Nor- wegian fisheries. As seen from the outliers in Figure 1, it is obvious that something affected the epsf af- ter 1980. Dramatic responses are expected when a fishery is regu- lated (Hilborn and Walters, 1992). Respones to regulations may in- clude changes in the effort distri- bution during the year, changes in the patterns of competition and cooperation with other vessels, increased misreporting, and in- creased amounts of discards. In addition to uncertainties in the reported catch and effort from the vessels, there are also uncertainties in the VPA estimates of stock size (see Ulltang, 1977). The trawlable fraction of the youngest year class- es also changes within and among years due to individual growth and yearly differences in growth. The constant retention probabili- ties used in this work to select the youngest age classes from the VPA estimates, and the use of only one single stock estimate per year is thus, perhaps, an over simplifica- tion. To get a better estimate of the learning curve, epsf can be estimat- ed based on averaging effort and catch over shorter time periods than in this work. This requires, however, advanced modelling of the selection process, and possibilities for obtain- ing estimates of total abundance more than once a year. The VPA estimates and the catch and effort data used in this work are not totally independent from each other. A time series of commercial CPUE from the Norwegian trawler fleet was used, together with other CPUE time se- ries, to calibrate the most recent VPA estimates of the old- 342 Fishery Bulletin 99(2) est fish (owing to lack of survey data for old fish). Total catch data from the fleet were also used in the VPA proce- dure. Converged VPA were, however, fairly independent of CPUE, and the limited dependence was not considered to have a large effect on the learning curve estimation. It seems reasonable, as seen from the plots in Figure 1, to assume that epsf followed a learning curve pattern with time. The outliers around 1980 affected the regres- sion lines in Figure 1 (B and D) considerably. If these out- liers were caused by the fishermens’ response to the new regulations, they should perhaps be removed. Another so- lution is to develop learning curve models that account for such strong interventions. The sudden change in the relation between CPUE and VPA estimates in Figure 2 may have been caused by the introduction of fishery regulations, but there are other im- portant factors affecting the use of CPUE as an index of abundance. Some of these factors are similar to those mentioned above, under the discussion of the assumptions concerning the estimation of epsf. Another explanation for the sudden change of the ratio between CPUE and bio- mass estimates from VPA in Figure 2B and for the outliers in Figure 1 around 1980 is that CPUE and epsf (or equiva- lently the catchability) are functions of stock abundance. An increase in catchability at low stock size is suggested in Figure 2 by the relatively high catch rates (given the low biomass) in the early to mid 1980s. This increase may, however, be contradicted by the similar trend in biomass and adjusted CPUE in Figure 2B at the end of the time pe- riod, when stock abundance became relatively low again. If catchability is shown to be density dependent for Atlan- tic cod in the Barents Sea, it should be incorporated into the model when the learning curve is fitted. Both long- term changes (e.g. Gordoa and Hightower, 1991; Swain et ah, 1994) and density-dependent changes (Crecco and Overholtz, 1990; Rose and Kulka, 1999) in catchability of gadoids have been indicated when commercial CPUE is being used as an index of abundance. Conclusion Modeling of epsf using the learning curve seems be a prom- ising method for solving the old and well-known problem of learning when time series of commercial CPUEs are used as indices of abundance. Care needs to be taken when selecting data and observations for estimation of learning curves, especially in multispecies fisheries. Acknowledgments I am grateful to Michael Pennington for improving the manuscript and to Christopher Harmon for interdisci- plinary help. The work received financial support from the Research Council of Norway. The Norwegian Directorate of Fisheries is thanked for providing me with data. Literature cited Argot, L., and D. Epple. 1990. Learning curves in manufacturing. Science (Wash., D.C.) 247:920-924. Casella, G., and R. L. Berger. 1990. Statistical inference. Duxbury Press, Belmont, CA. 650 p. Crecco, V., and W. J. Overholtz. 1990. Causes of density-dependent catchability for Georges Bank haddock Melanogrammus aeglefinus. Can. J. Fish. Aquat. Sci. 47:385-394. Gavaris, S. 1980. Use of a multiplicative model to estimate catch rate and effort from commercial data. Can. J. Fish. Aquat. Sci. 37:2272-2275. Gordoa, A., and J. E. Hightower. 1991. Changes in catchability in a bottom-trawl fishery for Cape hake (Merluccius capensis ). Can. J. Fish. Aquat. Sci. 48:1887-1895. Gulland, J. A. 1983. Fish stock assessment: a manual of basic methods. John Wiley and Sons, New York, NY, 223 p. Hilborn, R., and C. J. Walters. 1992. Quantitative fisheries stock assessment. Chapman and Hall, New York, NY, 570 p. Ketchen, K. S. 1964. Measures of abundance from fisheries for more than one species. Rapp. P-V. Reun. Cons. Int. Explor. Mer 155: 113-116. Rose, G. A., and D. W. Kulka. 1999. Hyperaggregation of fish and fisheries: how catch- per- unit-effort increased as the northern cod iGadus morhua) declined. Can. J. Fish. Aquat. Sci. 56(suppl. 11:118-127. Swain, D. P, G. A. Nielsen , A. F. Sinclair, and G. A. Chouinard. 1994. Changes in catchability of Atlantic cod ( Gadus mor- hua) to an otter trawl fishery and research survey in the southern Gulf of St Lawrence. ICES J. Mar. Sci. 51:493- 504. Ulltang, 0. 1977. Sources of errors in and limitations of virtual pop- ulation analysis (cohort analysis). J. Cons. int. Explor. Mer 37:249-260. 343 Assessment of skipjack tuna C Katsuwonus pelamis ) spawning activity in the eastern Pacific Ocean Abstract— An investigation of skip- jack tuna, Katsuwonus pelamis, spawn- ing activity was conducted to test an earlier established and widely accepted hypothesis that significant spawning of skipjack does not occur in the east- ern Pacific Ocean (EPO). Skipjack tuna ovaries were collected, from fish greater than 50 cm in length, from commercial landings of purse seiners fishing in the EPO during 1995. A total of 76 sam- ples, each consisting of approximately 25 females, were collected, processed, and analyzed. The characteristics used to classify the reproductive condition of the individual fish were the mean dia- meter of the oocytes of the most advanced mode and the presence or absence of residual hydrated oocytes. Results indi- cate that significant spawning of skip- jack tuna, 50 cm or greater in length, occurs in areas of the EPO where sea surface temperatures are equal to or greater than 24°C. Manuscript accepted 3 November 2000. Fish. Bull. 99:343-350 (2001). Kurt M. Schaefer Inter-American Tropical Tuna Commission 8604 La Jolla Shores Drive La Jolla, California 92037-1508 E-mail address: kschaefer@iattc.org Skipjack tuna, Katsuwonus pelamis, are distributed throughout the Pacific Ocean in tropical and subtropical waters, where they are a primary target species of large-scale purse-seine fisher- ies. Total estimated catches now exceed one million metric tons. In the eastern Pacific Ocean (EPO ), skipjack tuna have been fished from 34°N, off southern California, to 27°S, off northern Chile (Collette and Nauen, 1983; Matsumoto et ui.. 1984; Wild and Hampton, 1994). The fishery in the EPO was historically concentrated in two areas, a southern area at about 5°N to 10°S and east of 110°W, and a northern area at about 10°N to 30°N and west of 105°W to about 125°W. Estimated annual catches of skipjack tuna during the 1988 to 1997 period in the EPO averaged about 97 thousand metric tons, rang- ing from 63 to 159 thousand metric tons (Table 3 in Bayliff, 1999). The catches in the southern area have gen- erally been twice as large as those in the northern area, and there have been large annual fluctuations in both areas. Since 1995, the greatest catches have been made by purse seiners setting on floating objects, including fish-aggre- gating devices (FADs), between about 10°N and 15°S from the coast of the Americas to about 140°W. Schaefer (1963), reporting on pop- ulation structure of skipjack tuna in the eastern Pacific, stated that “some, at least, of the West Coast population range far to the westward.” He further stated that, “most important, there is little evidence of skipjack tuna spawn- ing in the eastern Pacific fishing region, or near to it; most reproduction occurs farther westward.” Rothschild (1965), however, hypothesized that “most skip- jack taken by the eastern Pacific skip- jack fisheries originate in the central Pacific.” His hypothesis was based on available evidence, at that time, from larval distributions, gonad indices, size distributions, tag recoveries, catch pre- dictions, and immunogenetic studies. This hypothesis, however, regarding the origin of skipjack tuna that occur in the eastern Pacific may not be completely valid. The results of research on the re- productive biology of skipjack tuna in the EPO (Schaefer and Orange, 1956; Orange, 1961 ) indicated some spawning off Central America and in the vicinity of the Revillagigedo Islands, Mexico. In addition, larval surveys have indicated that skipjack tuna spawning occurs in offshore waters and, to a lesser extent, in coastal waters of the EPO (Matsu- moto, 1975; Matsumoto et al., 1984; Ni- shikawa et al., 1985). In order to achieve a better under- standing of the stock structure of skip- jack tuna in the Pacific Ocean, an in- vestigation of the spawning potential of skipjack tuna in the EPO was under- taken. The objective of this study was to test the hypothesis that significant spawning of skipjack tuna does not oc- cur in the EPO. Materials and methods Field sampling Skipjack tuna specimens were selected by staff members of the Inter-Ameri- can Tropical Tuna Commission ( IATTC ) from the landings of purse seiners at processing plants in Manta (Ecuador), 344 Fishery Bulletin 99(2) Ensenada and Mazatlan (Mexico), Mayaguez and San Juan (Puerto Rico), and Cumana (Venezuela). Sampling was restricted to catches from east of 150°W from Janu- ary through December 1995. The sampling program was designed to select specimens greater than 50 cm in length, in order to efficiently sample sexually mature fish (Orange, 1961), if there were any. The area sampling strata (Fig. 1) used by the IATTC for length data (Tomlinson et al., 1992) were also used for sampling skipjack tuna specimens for the present study. The sampling was designed to collect 25 recognizable ova- ries from a single purse-seine set from each area and month stratum for one year. A quota was stipulated of no more than three such samples, for a total of 75 ovaries, from each area and month stratum from which skipjack tuna were caught and unloaded. A total of 76 samples (in- cluding 1 1 replicates), each consisting of approximately 25 female skipjack tuna, for a total of 1822 ovaries, were col- lected and processed. Skipjack tuna specimens were not available for sampling from various area and month stra- tum during this sampling period because of the distribu- tion of fishing effort. Each fish was measured to the nearest centimeter with calipers. The gonads were removed, sex was determined, and the ovaries placed in a plastic bag with identification labels and frozen. The capture locations, dates, set speci- fications, and sea surface temperatures (SSTs) were ob- tained from records of scientific observers or abstracts of the vessel logbooks made by IATTC employees, or from both. Analysis of ovaries Chi-square contingency tests were employed to compare distributions of oocyte diameters from t issue samples from the rostral, medial, and caudal regions of both ovaries from 30 fish. Because no significant differences were found (P>0.5), oocytes from an approximately 1-g tissue sample from the medial region of the right or left ovary were measured to the nearest 0.03 mm at 27x magnification with an ocular micrometer in a dissecting microscope. The mean diameter (random axis) was determined from 10 oocytes per fish (with oocytes >0.21 mm) present in the most advanced mode of oocytes. Ovaries were slit longitu- dinally and the ovarian lumen examined microscopically for the presence of residual hydrated oocytes, indicative of recent spawning activity. Female skipjack tunas can be classified as immature, maturing, reproductively active, or resting. Skipjack have asynchronous oocyte development and are considered to he multiple or batch spawners (Schaefer, 1998; Schaefer, in press). Histological and morphological descriptions of oo- genesis in the ovary of skipjack tuna (Cayre and Farrugio, 1986; Goldberg and Au, 1986; Stequert and Ramcharrun, 1996) were used for classification of reproductive condi- tion of individuals in the present study. The characteris- tics used to classify the reproductive condition of the indi- vidual fish were 1) the mean diameter of the oocytes of the most advanced mode and 2) the presence or absence of re- sidual hydrated oocytes. A fish is classified as immature if Figure 1 Areas of the eastern Pacific used for sampling skipjack tuna gonads. the mean diameter of the most advanced group of oocytes is less than or equal to 0.2 mm and if there are no residual hydrated oocytes present. Results Sex ratios There were 65 purse-seine sets on schools of skipjack tuna in which a minimum of 24 fish and a maximum of 87 fish, equal to or greater than 50 cm in length, were sampled (mean=52.5 cm, SE=1.44 cm). The percentage of males in the samples ranged from 26.5 to 71.3 (mean=51.5, SE=1.21). A total of 3284 fish, of which 1737 (52.9%) were males and 1547 (47.1%) were females, were sampled. The overall sex ratio for the pooled data from the 65 samples was signifi- cantly different (/^'o.o.5.i= 10.99 ) from the expected 1:1 ratio. Chi-square tests of males and females grouped into 5-cm length classes for the pooled data (Table 1) indicated a sig- nificant deviation in the 50-54.9 cm class (56.0% males), and the 65—69.9 cm class (57.0% males). Replicate samples For the 11 replicate samples collected, chi-square tests indicated no significant differences in the proportions of reproductively active female skipjack tuna specimens among the paired replicates (P>0.05). Thus, the sample sizes of approximately 25 female skipjack tuna from a single purse-seine set appeared to be adequate for assess- ing the reproductive status of the fish from the sets sam- pled. The data from the replicate samples were not used in any of the subsequent analyses. Schaefer: Assessment of spawning activity of Kcitsuwonus pelcimis in the eastern Pacific Ocean 345 150 140 130 120 110 100 90 80 70 Distributions of reproductively-active (solid squares) and inactive (open squares) female skipjack tuna sampled during 1995. Spatiotemporal patterns in spawning Reproductive activity is certain for the females with ad- vanced yolked oocytes that are equal to or greater than 0.55 mm in diameter. Of these, 429 (28%) were classified as reproductively active, and reproductively active females were found in 27 of the samples (42%) (Table 2). Of the 429 skipjack tuna with advanced yolked oocytes, 232 (54%) also had residual hydrated oocytes in the lumen of their ovaries, indicative of recent spawning. Based upon the dis- tribution of these samples, skipjack tuna spawning in the EPO appears to be fairly widespread from around 15°N to 10°S and from the coast to about 130°W (Fig. 2). Reproduc- tively active skipjack tuna were present north of the equa- tor throughout the year and south of the equator during the first three quarters of the year, and there were no apparent seasonal peaks in either stratum (Fig. 3). No samples of mature fish were collected during the fourth quarter south of the equator. The length-frequency distri- bution of the females from the 65 independent samples is shown in Figure 4. Based upon the SST data collected in conjunction with catch information for each of these samples, it appears that skipjack tuna are sexually inactive at SSTs less than 24°C. Of the 65 samples, 20, or 31%, were taken from skip- jack tuna captured at SSTs less than or equal to 24°C (Fig. 5). In other words, just over half of the 38 samples that did not contain reproductively active fish were obtained from skipjack tuna captured at SSTs below those at which spawning occurs. Table 1 Sex ratios of skipjack tuna, Katsuwonus pelamis. Number observed Percent male Length (cm) Male Female Chi-square 50.0-54.9 873 687 56.0 22. IS 55.0-59.9 365 387 48.5 0.64 60.0-64.9 252 285 46.9 2.03 65.0-69.9 226 170 57.0 7.92 70.0-74.9 21 18 53.8 0.23 Total 1737 1547 52.9 10.99' *=P< 0.05. The frequency distributions of the estimated sizes, in metric tons, of the skipjack tuna schools sampled, along with the subset of those that were classified as spawning, are shown in Figure 6. The prominent mode in each dis- tribution was around 25 metric tons; the ranges in each mode were from about 5 to 270 tons. Discussion It is evident from the results of this investigation that sig- nificant spawning of skipjack tuna, 50 cm or greater in 346 Fishery Bulletin 99(2) Table 2 Reproductive status of female skipjack tuna caught in the eastern Pacific Ocean and sampled at boundaries are shown in Figure 1. canneries during 1995. Area 2 4 5 6 Area 7 10 11 13 Total No. of samples collected 2 8 10 8 20 4 6 7 65 No. of ovaries collected 29 170 232 196 493 100 150 177 1547 No. of samples with reproductively active females 0 4 4 0 7 3 6 3 27 Percentage of samples with reproductively active females 0 50 40 0 35 75 100 43 42 No. of active females 0 44 72 0 67 55 117 74 429 Percentage of active females 0 26 31 0 14 55 78 42 28 length, occurs in areas of the EPO where the SSTs are equal to or greater than 24°C. Skipjack tuna spawning occurs throughout the year in tropical waters and season- ally in subtropical waters in all major oceans (Matsumoto et ah, 1984; Nishikawa et al., 1985; Schaefer, in press). The latitudinal range in the spawning distribution of skipjack tuna has been shown for the Pacific and Atlantic Oceans to coincide with the area delineated on the north and south by the 24°C isotherm (Ueyanagi, 1969; Cayre and Farru- gio, 1986). In the western part of the equatorial Indian Ocean skipjack tuna spawning also occurs throughout the year (Stequert and Ramcharrun, 1996; Timohina and Romanov, 1996). Earlier research on the reproductive biology of skipjack tuna in the EPO indicated spawning off Central America and off Baja California, Mexico, near the Revillagigedo Is- lands (Schaefer and Orange, 1956; Orange, 1961). It was concluded from these studies that skipjack tuna spawn- ing occurs mainly offshore in the EPO, and the estimated minimum size at maturity in the vicinity of the Revilla- gigedo Islands is about 55 cm and about 40 cm off Cen- tral America. The present study also indicates spawning of skipjack tuna in the EPO appears to be more concentrated offshore, west of 95°W longitude (Fig. 2). It should be noted that during the 1950s and early 1960s the fishery for tu- nas in the eastern Pacific occurred within a few hundred miles of the mainland and in the vicinity of a few offshore islands and banks (Alverson, 1960, 1963). Thus, skipjack tuna gonad sampling for those earlier studies was con- fined to those areas. Larval surveys have also indicated that skipjack tuna spawning occurs in offshore areas of the EPO and may be restricted in coastal areas (Klawe, 1963; Matsumoto, 1975; Matsumoto et al., 1984). One of the objectives of the EAST- ROPAC expeditions of 1967 was to look at the distribu- tion of scombrid larvae over a vast expanse of the EPO. During EASTROPAC I (Ahlstrom, 1971), larvae of skip- jack tuna (17 occurrences, 214 larvae) were collected pre- dominantly in the offshore southern portion of the EPO at about 7°S and 120°W. Likewise, few yellowfin tuna larvae (19 occurrences, 40 larvae) were collected during EAST- ROPAC I, even though the spawning distribution of yel- lowfin tuna has been shown to be widespread throughout the EPO (Schaefer, 1998). Scombrid larvae were markedly less abundant during EASTROPAC II; few skipjack tuna (2 occurrences, 2 larvae) and yellowfin tuna (2 occurrenc- es, 2 larvae) were collected (Ahlstrom, 1972). Catch rates of larval skipjack tuna adjusted for different size nets and different towing methods in a band of water from 10°S to 20°N indicated that skipjack tuna larval abundance was highest between 160°E and 140°W, moderate between 100°W and 140°W and between 120°E and 160°E, and low in the EPO east of 100°W (Matsumoto, 1975). The exten- sive compilation of data on larval scombrids by Nishika- wa et al. (1985) also indicates that east of 150°W between KPS and KPN there is widespread spawning of skipjack tuna, mostly west of 1KPW. Although there is also a dis- tinct tendency of increasing abundance of skipjack tuna larvae from the EPO to the WPO, these authors caution that this tendency is perhaps overstressed due to a poten- tial bias in the data. Tag release and recapture studies have shown that con- siderable mixing of skipjack tuna exists in the Pacific Ocean (Hunter et al., 1986). There is a large volume of informa- tion on skipjack tuna movements in the western Pacific Ocean (WPO). Although there were numerous long-distance movements of individual tagged skipjack tuna observed, the overall percentage of recoveries having displacements of greater than 200 nmi was only 17% and there are few probable migration routes revealed from the recovery of tagged skipjack tuna (Wild and Hampton, 1994). Further- more, when considering skipjack tuna on a Pacific-wide ba- sis, particularly the areas of tagging operations in the WPO, it was concluded that skipjack tuna did not appear to mi- grate toward specific areas for feeding or spawning but ap- peared to move in more or less random directions within broad limits ( Hunter et al., 1986). There have been no recov- eries in the EPO from skipjack tuna tagged in the central or WPO. The assumed eastward migration routes of juve- niles described in the skipjack tuna migration model (Roth- schild, 1965) lack validation, and hypotheses about the en- ergetic advantages of migration to the EPO using the North Schaefer: Assessment of spawning activity of Katsuwonus pelomis in the eastern Pacific Ocean 347 E E a> E aj TD CD o o O 1.0 - 0.9 - 0.8 - 0.7 - 0.6 - 0.5 - 0.4 - 0.3 A o o n =228 O 1.0 - B 0.9 - O n = 265 0.8 - 0.7 - 0.6 - 0.5 — 0.4 - 0.3 T 1 Quarter Figure 3 Seasonal variation in oocyte diameters from the most advanced modal group of oocytes from (A) skipjack tuna collected from north of the equator and (B) skipjack tuna collected from south of the equator during 1995. Immature fish were excluded from this analysis. The width of the box is proportional to the number of samples in the group. The horizontal line within the box is at the median. The bottom of the box is at the first quartile (Qp and the top is at the third quartile (Q3) value. The lines extend from the top and bottom of the box to the adjacent lowest and highest observations within the lower and upper limits defined as Q, ±1.5 (Q3 - Q,). Outliers are points outside the lower and upper limits and are plotted as circles. Equatorial Countercurrent and the Equatorial Undercur- rent (Williams, 1972) are unsubstantiated. Numerous tagging studies have also been conducted in the EPO to investigate movements of skipjack tuna (Fink and Bayliff, 1970; Bayliff, 1984). It appears from these stud- ies that skipjack tuna show some consistency of directed movement in the nearshore regions off Central America and northern South America. In the northern region around the Revillagigedo Islands and the west coast of Baja Cali- fornia, there is a northern and then southern movement of the fish between 20°N and 30°N in response to the season- al movements of the 20°C surface isotherm between about 348 Fishery Bulletin 99(2) May and December. There is also some movement of fish between the northern and southern areas of the fishery in the EPO. However, from well over 100,000 skipjack tuna tagged in the EPO and several thousand returns, only 27 skipjack tuna have been recovered in the central Pacific or the WPO, and 21 of those were recaptured around the Ha- waiian Islands (Bayliff, 1988). Of those fish recovered, 19 were tagged off Baja California, 4 fish off the Revillagigedo Islands, 2 fish off Clipperton Island, and 1 fish well offshore at about 4°N and 119°W. Only 1 skipjack tuna tagged in the near shore waters off Central America, within the area of the primary fishery, has been recovered in the central Pa- cific around Hawaii. Before the recovery of this tagged skip- jack tuna there was no evidence from tagging that fish of the southern group migrate to the central Pacific. Although the information has been largely ignored, mor- phometries research has shown significant statistical dif- ferences between skipjack tuna from the EPO and the cen- tral Pacific (Hawaii and French Polynesia) (Hennemuth, 1959). These differences could indicate a low level of mix- ing of skipjack tuna populations between the central Pa- cific and the EPO. Reliable estimates of the skipjack tuna spawning bio- mass in the EPO are presently unavailable. However, some indication of the relative abundance of skipjack tuna spawners in the EPO can be derived from considering the estimated total catch of skipjack tuna by length intervals from the surface fishery in the EPO (Fig. 7). It is appar- ent for 1993 to 1998 that there were significant amounts of skipjack tuna greater than 50 cm in length present in the EPO. These data for skipjack tuna catch by length, combined with the data presented in the present study of spawning activity of skipjack tuna females greater than 50 cm in length, provide strong evidence to support the concept of a significant skipjack tuna spawning biomass in the EPO. It also seems reasonable to consider that the sub- Degrees (°C) Figure 5 Sea surface temperatures at which (A) all skipjack tuna schools were sampled and (B) spawning skipjack tuna schools were sampled in the eastern Pacific Ocean during 1995. sequent survivors, resulting from skipjack tuna spawning in the EPO, constitute a significant component of the skip- jack tuna population in the EPO. Skipjack tuna do not appear to be a migratory species such as Pacific, Atlantic, and southern bluefin tunas (Thun- 7ius orientalis , T. thynnus, and T. maccoyii) that have dis- crete spawning locations (Schaefer, in press). Too much emphasis has been placed on long-range movements of a few tagged skipjack. The tagging data for skipjack tuna only supports extensive offshore-onshore movements as well as north-south movements in the EPO. Based upon what is known about skipjack tuna life history (Matsumo- to et ah, 1984; Wild and Hampton, 1994), it would seem that skipjack tuna need to be opportunistic in their repro- ductive strategy, taking advantage of the suitable environ- ment in the EPO. Once skipjack tuna reach sexual matu- rity, they probably spawn throughout their distribution, whenever water temperatures rise above 24°C. The long-standing premise that skipjack tuna in the EPO are merely short-term transients, originating from spawning in the central Pacific or WPO, should be seri- ously reconsidered. This issue has important international management implications. Although the stock in the EPO is not a closed unit, as a result of the highly mobile and op- Schaefer: Assessment of spawning activity of Katsuwonus pelcimis in the eastern Pacific Ocean 349 portunistic nature of this species, assessments should nev- ertheless be conducted for the skipjack tuna stock within this geographical region. Ideally, stock assessment models for skipjack tuna should incorporate spatial and temporal variability in the size-specific parameters for age, growth, movement, mortality, and reproduction. A more compre- hensive investigation of the reproductive biology of skip- jack tuna in the EPO is in progress. Acknowledgments I gratefully acknowledge the previous Director of IATTC, Jim Joseph, the present Director, Robin Allen, and Chief Scientist of the Tuna-Billfish Program, Rick Deriso, for their encouragement and support of this investigation. I thank numerous members of the IATTC staff, particularly those stationed at field offices, whose assistance in sam- pling skipjack specimens for this project was invaluable. I also wish to thank Bill Bayliff and two anonymous review- ers for constructive comments on drafts of the manuscript. A 93 lllllh..^ . m Miiiiiii. _r jA 95 hill. : ..mu 96 IIMul i\ 97 lIllilEli.= jk 98 lllllllll... 20 30 40 50 60 70 80 Length in centimeters 90 Figure 7 Estimated length compositions of the total skipjack tuna catch, by 2-cm intervals, in the eastern Pacific Ocean. These histograms are based on the catch statistics and length- frequency samples of skipjack tuna, for each year of the 1993-98 period, compiled by the Inter-American Tropical Tuna Commission. Literature cited Ahlstrom, E. H. 1971. Kinds and abundance of fish larvae in the eastern tropical Pacific, based on collections made on EASTROPAC I. Fish. Bull. 69( 1 ):3-77. 1972. Kinds and abundance of fish larvae in the eastern tropical Pacific based on the second multivessel EASTRO- PAC survey, and observations on the annual cycle of larval abundance. Fish. Bull. 70(41:1153—1242. A1 verson, F G. 1960. Distribution of fishing effort and resulting tuna catches from the eastern tropical Pacific Ocean by quar- ters of the year, 1951-1958. Inter-Am. Trop. Tuna Comm.. Bull. 4(61:319-446. 1963. Distribution of fishing effort and resulting tuna catches from the eastern tropical Pacific Ocean, by quar- ters of the year, 1959-1962. Inter-Am. Trop. Tuna Comm., Bull. 8(61:317—379. 350 Fishery Bulletin 99(2) Bayliff, W. H. 1984. Migrations of yellowfin and skipjack tuna released in the central portion of the eastern Pacific Ocean, as determined by tagging experiments. In ter- Am. Trop. Tuna Comm., Intern. Rep. 18, 107 p. 1988. Growth of skipjack, Katsuwonus pelamis, and yellow- fin, Thunnus albacares, tunas in the eastern Pacific Ocean, as estimated from tagging data. Inter-Am. Trop. Tuna Comm., Bull., 19( 4 ):307 — 385. Bayliff, W. H. (editor). 1999. Inter-Am. Trop. Tuna Comm., Ann. Rep. for 1997, 310 p. IATTC, La Jolla, CA. Cayre, P., and H. Farrugio. 1986. Biologie de la reproduction du listao ( Katsuwonus pelamis) de l’ocean Atlantique. In Proceedings of the ICCAT conference on the international skipjack year pro- gram (P. E. K. Symons, P. M. Miyake, and G. T. Sakagawa (eds.), p. 252-272. ICCAT (International Commission for the Conservation of Atlantic Tunas), Madrid. Collette, B. B., and C. E. Nauen. 1983. FAO species catalogue. Vol. 2: Scombrids of the world: an annotated and illustrated catalogue of tunas, macker- els, bonitos and related species known to date. FAO Fish. Synop. 125, 137 p. FAO, Rome. Fink, B. D„ and W. H. Bayliff. 1970. Migrations of yellowfin and skipjack tuna in the eastern Pacific Ocean as determined by tagging experi- ments, 1952-1964. Inter-Am. Trop. Tuna Comm., Bull. 15( 1 ): 1—227. Goldberg, S. R.. and D. W. K. Au. 1986. The spawning of skipjack tuna from southeastern Brazil as determined from histological examination of ova- ries. In Proceedings of the ICCAT conference on the inter- national skipjack year program (P. E. K. Symons, P. M. Miyake, and G. T. Sakagawa, eds.), p. 277-184. ICCAT, Madrid. Hennemuth. R C. 1959. Morphometric comparison of skipjack from the cen- tral and eastern tropical Pacific Ocean. Inter-Am. Trop. Tuna Comm., Bull. 3(6):239-304. Hunter, J. R., A. W. Argue, W. H. Bayliff, A. E. Dizon, A. Fonteneau, D. Goodman, and G. Seckel. 1986. The dynamics of tuna movements: an evaluation of past and future research. FAO Fish. Tech. Pap. 277, 78 p. FAO, Rome. Klawe, W. L. 1963. Observations on the spawning of four species of tuna, Neothunnus macropterus, Katsuwonus pelamis, Auxis thaz- ard , and Euthynnus lineatus, in the eastern Pacific Ocean, based on the distribution of their larvae and juveniles. Inter-Am. Trop. Tuna Comm., Bull. 6(9):447-540. Matsumoto, W. M. 1975. Distribution, relative abundance, and movement of skipjack tuna, Katsuwonus pelamis, in the Pacific Ocean based on Japanese tuna longline catches, 1964-67. U.S. Dep. Commer., NOAA Tech. Rep. NMFS Spec. Sci. Rep. Fish. Ser. 695, 30 p. Matsumoto, W. M., R. A. Skillman, and A. E. Dizon. 1984. Synopsis of biological data on skipjack tuna, Kat- suwonus pelamis. U.S. Dep. Commer., NOAA Tech. Rep. NMFS Circ. 451, 92 p. Nishikawa, Y., M. Honma, S. Ueyanagi, and S. Kikawa. 1985. Average distribution of larvae of oceanic species of scombrid fishes, 1956-1981. Far Seas Fish. Res. Lab. S series 12, 99 p. Orange, C. J. 1961. Spawning of yellowfin and skipjack in the eastern tropical Pacific, as inferred from studies of gonad develop- ment. Inter-Am. Trop. Tuna Comm., Bull. 5(6): 459-526. Rothschild, B. J. 1965. Hypothesis on the origin of exploited skipjack tuna ( Katsuwonus pelamis) in the eastern and central Pacific Ocean. U.S. Fish. Wildl. Serv., Spec. Sci. Rep. Fish., 512, 20 p. Schaefer, K. M. 1998. Reproductive biology of yellowfin tuna ( Thunnus alb- acares) in the eastern Pacific Ocean. Inter-Am. Trop. Tuna Comm., Bull. 21(5): 201-272. In press. Reproductive biology of tunas. In Tuna: physio- logical ecology and evolution (B. A. Block and E. D. Stevens, eds.). Academic Press, San Diego, CA. Schaefer, M. B. 1963. Report on the investigations of the Inter- American Tropical Tuna Commission for the year 1962 Inter-Am. Trop. Tuna Comm., Ann. Rep. 1962:35-149. Schaefer, M. B., and C. J. Orange. 1956. Studies on the sexual development and spawning of yellowfin ( Neothunnus macropterus) and skipjack ( Katsu- wonus pelamis) in three areas of the eastern Pacific Ocean, by examination of gonads. Inter-Am. Trop. Tuna Comm., Bull. 1( 6 ):281— 349. Stequert, B., and B. Ramcharrun. 1996. La reproduction du listao ( Katsuwonus pelamis) dans le bassin ouest de l’ocean Indien. Aquat. Living Resourc. 9( 3 ):235— 247. Timohina, O. I., and E. V. Romanov. 1996. Characteristics of ovogenesis and some data on matu- ration and spawning of skipjack tuna, Katsuwonus pelamis (Linnaeus, 1758), from the western part of the equatorial zone of the Indian Ocean. In Proceedings of the expert consultation on Indian Ocean tunas, 6th session, Colombo, Sri Lanka (A. A. Anganuzzi, K. A. Stobberup, and N. J. Webb (eds.). p. 247-257. Indo-Pacific Tuna Development and Management Programme, Colombo, Sri Lanka. Tomlinson, P. K., S. Tsuji, and T. P. Calkins. 1992. Length-frequency estimation for yellowfin tuna (Thun- nus albacares) caught by commercial fishing gear in the eastern Pacific Ocean. Inter-Am. Trop. Tuna Comm., Bull. 20(61:359-398. Ueyanagi, S. 1969. Observat ions on the distribution of tuna larvae in the Indo-Pacific Ocean with emphasis on the delineation of the spawning areas of albacore, Thunnus alalunga. Bull. Far Seas Fish. Res. Lab. 2:177-256. Wild, A., and J. Hampton. 1994. A review of the biology and fisheries for skipjack tuna, Katsuwonus pelamis, in the Pacific Ocean. FAO Fish. Tech. Pap. 336(2):1-51. Williams, F. 1972. Consideration of three proposed models of the migra- tion of young skipjack tuna ( Katsuwonus pelamis ) into the eastern Pacific Ocean. Fish. Bull. 70(31:741-762. 351 Abstract— The protein fingerprints of shark fillets and fins taken from com- mercial landings in northern New Zea- land waters were compared with the protein fingerprints from control sam- ples of ten species of coastal sharks. Isoelectric focusing (IEF) in agarose gels revealed species-specific protein profiles in the ten control species. The fillets and fins were identified as school shark ( Galeorhinus galeus ), rig (Mustelus lenticulatus), hammer- head shark (Sphyrna zygaena), and bronze whaler ( Careharhinus brachy- urus). Around 40% of fillets from cartons labelled as lemon fish (M. lenticulatus) were from other species. Shark fins were identified from four species, two of which are prohibited target species in northern New Zealand. The large number of mislabelled shark products necessitates the use of a simple bio- chemical technique for identification of shark species in commercial shark products. With IEF, around 100 speci- mens can be identified by a laboratory technician each working day from small amounts (<0.5 g) of white muscle. Manuscript accepted 16 November 2000. Fish. Bull. 99:351-355 (2001). Biochemical identification of shark fins and fillets from the coastal fisheries in New Zealand Peter J. Smith Peter G. Benson National Institute of Water and Atmospheric Research Ltd 301 Evans Bay Parade Wellington, New Zealand E-mail address (for P J. Smith): p.smith@mwa.cri,nz Several shark species are common in New Zealand coastal waters and form the basis of fisheries producing fillets for the domestic market and fins for the Asian market. Like most sharks, the New Zealand species are vulnera- ble to over exploitation owing to their slow growth rate, large size at maturity and low fecundity (Castro et al., 1999). Three shark species are currently man- aged under the New Zealand Quota Management System (QMS): school shark (Galeorhinus galeus), rig (Muste- lus lenticulatus), and the Callorhinchi- dae elephant fish (Callorhinchus milii) (Annala et al.1). Quotas were intro- duced for these species in 1986 fol- lowing dramatic increases in catches. Two species, G. galeus and M. lenticula- tus, are caught in set-net fisheries that target coastal sharks, and since the introduction of the QMS, the fisheries have stabilized at around 5000 tons per annum (Annala et al.1). A competitive quota has been established for spiny dogfish (Squalus acanthias), but other species of shark, either because of low abundance or assumed low value, have been excluded from the QMS. Some of these shark species, such as bronze whaler ( Carcharin us brachyurus), ham- merhead shark (Sphyrna zygaena ), and blue shark (Prionace glauca ), are taken as bycatch in other fisheries but are prohibited as target species. The shark species in the New Zea- land coastal fisheries are large, and whole fish are readily identified by ex- ternal characteristics. However most sharks are processed at sea: specimens are “trunked” — the head, guts, and fins are removed — and the product chilled or frozen. Further processing may take place on land and the product is sold as either trunks or fillets and fins. It is dif- ficult to identify fillets and fins to the species level and landed catches may include several species. Management and enforcement of multispecies shark fisheries require that landed products can be correctly identified to the spe- cies level. Thus a robust test is needed that will allow identification of fillets and fins. Several molecular genetic methods have been applied to fisheries-related taxonomic problems to identify and dis- tinguish closely related species. Allo- zymes have been the primary tool for taxonomic problems in fishes (e.g. Day- ton et al., 1994; Lacson and Bassler, 1992) and have revealed cryptic species in teleosts (e.g. Lacson, 1994; Smith et al., 1996). More sophisticated and ex- pensive molecular methods, based on DNA extraction, followed by sequenc- ing or restriction enzyme digestion, are increasingly used for similar identifi- cation problems (Bartlett and David- son 1992), including shark species (e.g. Martin, 1993; Heist and Gold, 1998). Isoelectric focusing (IEF) of muscle pro- teins has been the preferred method for identification of teleost fillets and products (Lundstrom, 1980; Rehbein, 1990); the method is used in legal cas- es to identify mislabeled fish products. Polyacrylamide IEF (Lundstrom, 1977, 1980) has been adopted by the U.S. Food and Drug Administration for iden- 1 Annala, J. H., K. J. Sullivan, C. J. O’Brien, and S .D. Ibal 1 . 1998. Report from the Fishery Assessment Plenary, May 1998: stock assessments and yield estimates, 409 p. Unpublished report held in NIWA Library, 301 Evans Bay Parade, Welling- ton, New Zealand. 352 Fishery Bulletin 99(2) Table 1 Common and scientific names of shark control samples used in the isoelectric focusing analyses. Common name Other common names Scientific name Rig gummy shark, smoothhound, spotted dogfish, spotted smoothhound. lemon fish, pioke Mustelus lenticulatus School shark grey shark, tope, flake, makohuarau, tupere Galeorhinus galeus Spiny dogfish southern spiny dogfish, spurdog, spineback, spikey dog Squalus acanthias Northern spiny dogfish grey spiny dogfish, shortspine spurdog, green-eyed dogfish Squalus mitsukurii Hammerhead shark hammerhead Sphyrna zygaena Bronze whaler whaler Carcharhinus brachyurus Blue shark blue pointer, blue whaler Prionace glauca Porbeagle mackerel shark Lamna nasus Mako mackerel shark Isurus oxyrinchus Carpet shark swell shark, cat shark Cephaloscyllium isabellum tification of fish product (Tenge et al., 1993) and applied to shark species (Weaver et al., 1999); it provides a finer sep- aration of proteins than conventional starch and cellulose acetate electrophoresis, and the muscle proteins exhibit little intraspecific variation. In addition, muscle proteins are stable, withstanding repeated freezing and thawing, and provide a species profile in one gel, unlike allozyme methods, where several gels are required in order to iden- tify a range of species. However, some closely related spe- cies of teleosts share protein profiles (Bartlett and David- son, 1991; Smith et al., 1994; Smith et al., 1996) and thus IEF is not always amenable for distinguishing closely re- lated species. Our study was undertaken to evaluate aga- rose IEF (Lundstrom, 1981, 1983) of shark muscle pro- teins in order to provide a quick and robust biochemical method to identify shark fins and fillets from commercial vessels and the market place. Materials and methods Samples Control samples were taken from shark specimens caught on research and commercial vessels around New Zealand (Table 1). The specimens were frozen whole at sea and returned to the laboratory where the identity of the species was confirmed through the use of identification keys (Paulin et al., 1989). Samples of white muscle were removed from up to ten specimens of each species. For Mustelus lenticulatus and Galeorhinus galeus, ten speci- mens were tested from the Bay of Plenty, North Island, and ten specimens from the east coast South Island. Muscle tissue samples were stored in separate, labelled bags at -70°C. In addition, samples of body muscle from the head, mid, and tail regions were compared with sam- ples of muscle from the base of the pectoral and dorsal fins taken from the same specimens for both M. lenticulatus and Sphyrna zygaena. Mustelus lenticulatus and G. galeus are in the same fam- ily of smoot hhounds (Triakidae) and lack the characteristic dorsal spine of the spiny dogfishes. Whole specimens of M. lenticulatus are distinguished from G. galeus by the size of the second dorsal fin, which is nearly as large as the first dorsal fin in M. lenticulatus , but much smaller in G. galeus, and by their teeth; there are small teeth or grind- ing plates in M. lenticulatus and distinctive large triangu- lar teeth in G. galeus (Paulin et al., 1989). Both species are widely distributed in temperate coastal waters and total recorded landings in New Zealand are around 3200 tons for school shark and 1800 tons for M. lenticulatus (Annala et al.1). Two species of spiny dogfish (Squalidae) are com- mon in New Zealand waters, the shortspine dogfish ( Squa - lus mitsukurii ) in northern waters and the spiny dogfish ( Squalus acanthias) in southern waters. The hammerhead shark (Sphyrna zygaena, family Sphyrinidae) was included because small specimens are caught in northern waters of New Zealand. The species has a characteristic black mar- gin to the fins, but we have noted that juvenile, but not adult, G. galeus also have a black margin on the dorsal fin. Two species of requiem sharks (family Carcharhinidae), Carcharhinus brachyurus and Prionace glauca, are also common in coastal waters but are prohibited as target spe- cies in northern New Zealand. Lamna nasus, Isurus oxy- rinchus, and Cephaloscyllium isabellum, which are caught in northern New Zealand, were also included as controls (Table 1). Three-hundred and eighty shark fins were supplied by Ministry of Fisheries staff from commercial shark fisher- ies in the Bay of Plenty in northern New Zealand. In the laboratory, a small piece of muscle tissue was removed from each fin and stored at -70°C. The color and shape of each fin was noted and the maximum length between the flesh area and the fin tip was recorded; fins were stored frozen at -20°C. Eight 10-kg cartons of shark fillets, all labelled as lemon fish (=M. lenticulatus ), were tested. A small piece of muscle tissue was removed from each indi- vidual fillet, labelled, and stored at -70°C prior to isoelec- tric focusing. Smith and Benson: Biochemical identification of shark fins and fillets 353 Isoelectric focusing Small samples (about 0.5 g) of white muscle were removed from each of the tissue samples taken from the fins and fillets and from control samples taken from known speci- mens of New Zealand shark species. The muscle samples were homogenized individually in two volumes of cold (4°C) deionised water and centrifuged at 12,000 g for five minutes at 4°C. The clear supernatants were placed on filter paper wicks that were placed directly onto aga- rose IEF gels (Amersham Pharmacia Biotech, Uppsala) on a flat bed IEF system (Amersham Pharmacia Biotech, Uppsala). The 1-mm 1 7c agarose gels were made up in wide range pharmalyte, pH 3-10 (Amersham Pharmacia Bio- tech, Uppsala), and focused at 1500 volts for 90 minutes. After focusing, the proteins were fixed, washed, stained with coomassie brilliant blue (BDH Laboratory Supplies, Poole), destained, and dried (Benson and Smith, 1989). An initial gel was run with control samples only to ensure that each species produced a unique protein fingerprint. In addition, samples of body muscle from the head, mid, and tail regions were compared with samples of muscle from the base of pectoral and dorsal fins taken from the same specimens of M. lenticulatus and S. zygaena . Twenty- five suspect and seven control samples were run on subse- quent gels. The same control samples were used in each IEF gel to avoid mismatches between gels. Results The muscle tissue samples from the control specimens produced different protein fingerprint patterns in each species (Fig. 1). Tests of samples of body muscle and fin muscle from the same specimen produced the same IEF pattern, demonstrating that the muscle control samples were suitable for identification of fillet or fin samples. Samples of M. lenticulatus and G. galeus from the Bay of Plenty and east coast South Island showed no intraspe- cific variation in protein profiles. The fins could be grouped into three types based on shape: dorsal, pectoral, and caudal (or tail). There was a wide range of sizes from 7 to 28 cm. It is possible that some small dorsal fins may have been anal fins or second dorsal fins. Most fins were of a pale gray color, but some were a darker gray color and had a narrow black margin; most of these latter fins had a narrower shape than the pale gray dorsal fins. A few pectoral fins were dark gray on one sur- face and light gray on the other surface. None of the fins had a spine or showed sign of a spine having been cut out. The protein fingerprints of most of the suspect shark fins matched with one of the fingerprint patterns of M. len- ticulatus, G. galeus, S. zygaena, or C. brachyurus (Table 2). Four of the 392 fins produced a weak and indistinct finger- print pattern that could not be matched to any of the con- trol samples. Around 10r/( of the fins were from two spe- 354 Fishery Bulletin 99(2) Table 2 Number of shark fins and fillets identified to species level with isoelectric focusing of muscle proteins. All of the fillets were from commercial cartons labeled as lemonfish ( Mustelus lenticulatus). No ID = not identified because of a weak protein fingerprint. Fin or fillet M. lenticulatus G. galeus S. zygaena C. brachyurus No ID Pectoral fin 127 69 0 7 0 Dorsal fin 81 46 25 3 3 Tail fin 0 6 22 1 1 Unidentified 1 0 0 0 0 Fillet 113 27 28 26 1 cies that are prohibited as target species, S. zygaena and C. brachyurus (Table 2). None of the protein fingerprints of the fin samples matched with the control samples from S. mitsukurii, S. acanthias, P. glauca,L. nasus, I. oxyrinchus, or C. isabellum. Protein fingerprints of fillets from cartons labelled as lemon fish (M. lenticulatus ) showed that the fillets were from four shark species, two of which are prohibited as tar- get species, S. zygaena and C. brachyurus (Table 2), and that 41.8% of the fillets were not the species shown on the carton labels. Discussion To identify the suspect shark fillets and fins, it is essen- tial to use a test that can distinguish closely related spe- cies of sharks. Occasionally closely related pairs of species are found with very similar protein fingerprints, e.g. the teleosts tarakihi and king tarakihi, in which case addi- tional biochemical tests are sought to identify specimens (Smith et ah, 1996). The muscle tissue samples from the shark control specimens produced different protein finger- print patterns in each species (Fig 1.). Samples of body muscle and fin muscle from the same specimen produced the same IFF pattern. In addition, samples of M. lenticu- latus and G. gcileus from the Bay of Plenty in the North Island and east coast South Island showed no intraspe- cific variation in protein profiles. These observations dem- onstrate that muscle protein profiles are an appropriate tool for the identification of shark fillets or fins taken in the New Zealand coastal fishery. It is not possible to quantify the results and state how many fish specimens have been sampled in the fins and fillets. Each fish may yield two pectoral, two dorsal, one caudal, and one anal fin, but fishermen may discard small or damaged fins. Likewise with fillets, each fish may yield a minimum of two fillets, but four or more fillets may be taken from large specimens. Nevertheless the results indi- cate that both quota and nonquota species are being land- ed for the fillet and fin markets, and that the domestic market has cartons of mislabelled species. Around 40% of the fillets tested in our study were not the species on the label — M. lenticulatus (Table 2). Such observations dem- onstrate that shark landings recorded in New Zealand wa- ters may be inaccurate, which will not only confound catch statistics but may compromise assessments upon which regulatory decisions are made. The mislabeled fillets iden- tified in our study suggest that effort is targeting non- ITQ species or that prohibited target species such as Car- charhinus brachyurus (prohibited in area 1), and Sphyima zygaena (prohibited in all areas) are being landed. Unlike other biochemical techniques, such as allozyme and DNA markers, the protein fingerprints revealed by IEF show little intraspecific variation (Lundstrom, 1981). Most individuals from the same species have identical pro- tein fingerprints. When protein fingerprints vary among individuals from the same species, the differences are re- stricted to the presence or absence of one or a few of the protein bands; the majority of bands are shared among all individuals. Isoelectric focusing is a relatively quick and cheap iden- tification technique (Lundstrom, 1981) compared with DNA-based extraction methods. One operator is able to process and identify up to 100 specimens in one working day. The IEF technique works well with fresh and frozen material and produces clear protein profiles. However, some proteins denature when they are heat-treated (Keen- an and Shaklee, 1985). Therefore, shark products, such as sun-dried fins, may require alternative methods, such as polyacrylamide gel electrophoresis of parvalbumins (Keen- an and Shaklee, 1985) or DNA-based methods (Martin, 1993; Heist and Gold, 1998), for species identification. Acknowledgments We are grateful to Mark Scott and Brent Lincoln from the Ministry of Fisheries, Tauranga, for supplying fishery samples and bringing the issue of shark fin identification to our attention. We thank three anonymous referees for constructive comments on the manuscript. This project was supported by the New Zealand Ministry of Fisheries, project number MOF802, species identification. Literature cited Bartlett, S. E., and W. S. Davidson. 1991. Identification of Thunnus tuna species by the poly- Smith and Benson: Biochemical identification of shark fins and fillets 355 merase chain reaction and direct sequence analysis of their mitochondrial cytochrome b genes. Can. J. Fish. Aquat. Sci. 48:309-317. 1992. FINS (forensically informative nucleotide sequenc- ing): a procedure for identifying the animal origin of bio- logical specimens. BioTechniques 12:408-411. Benson, P. G., and P. J. Smith. 1989. A manual of techniques for electrophoretic analysis of fish and shellfish tissues. New Zealand Fisheries Techni- cal Report 13, 32 p. Castro, J. I., C. M. Woodley, and R. L.Brudek. 1999. A preliminary evaluation of the status of shark spe- cies. FAO fisheries technical paper, 72 p. FAO Rome. Dayton, C., A. C. Santayana, and J. M. Lacson. 1994. Genetic evidence for reproductive isolation of the recently described unicornfish Naso caesius and its sibling Nhexacanthus. Mar. Biol. 118:551-554. Heist, E. J., and J. R. Gold. 1998. Genetic identification of sharks in the U.S. Atlantic large coastal shark fishery. Fish. Bull. 97:53-61. Keenan, C. P., and J. B. Shaklee. 1985. Electrophoretic identification of raw and cooked fish fillets and other marine products. Food Tech. Australia 37:117-128. Lacson, J. M. 1994. Fixed allele frequency differences among Paluan and Okinawan populations of the damselfishes Chry- siptera cyanea and Pomacentrus coelestis. Mar. Biol 118:359-365. Lacson, J. M., and C. P. Bassler. 1992. Biochemical systematics of fishes of the genus Steg- astes (Pomacentraidae) from the Southern Marianas. Mar. Biol. 113:457-462. Lundstrom, R. C. 1977. The identification of fish species by thin-layer poly- acrylamide gel isoelectric focusing. Fish. Bull. 75:571- 576. 1980. Fish species identification by thin-layer polyacryl- amide gel isoelectric focusing: a collaborative study. -J. Assoc. Off. Anal. Chem. 63:69-73. 1981. Rapid fish species identification by agarose gel iso- electric focusing of sarcoplasmic proteins. J. Assoc. Off. Anal. Chem. 64:38-43. 1983. Fish species identification by agarose gel isoelectric focusing: a collaborative study. J. Assoc. Off. Anal. Chem. 66:123-127. Martin, A. P 1993. Application of mitochondrial DNA sequence analysis to the problem of species identification of sharks. In Con- servation biology of elasmobranchs, p. 53-59. U.S. Dep. Commerce, NOAA Technical Report NMFS 115. Paulin, C., A. Stewart, C. Roberts, and P McMillan. 1989. New Zealand fish: a complete guide. National Mu- seum of New Zealand Miscellaneous Series 19, 279 p. Rehbein, H. 1990. Electrophoretic techniques for species identification of fishery products. Z. Lebensum. Unters. Forsch. 191:1-10. Smith, P. J.. A. M. Conroy, and P. R. Taylor. 1994. Biochemical-genetic identification of northern bluefin tuna Thunnus thynnus in the New Zealand fishery. NZ J. Mar. Freshwater Res. 28: 1 13-118. Smith, P. J., C. D. Roberts, S. M. McVeagh, and P. G. Benson. 1996. Genetic evidence for two species of tarakihi (Teleostei. Cheilodactylidae Nemadactylus) in New Zealand waters. NZ J. Mar. Freshwater Res. 30:209-220. Tenge, B., N-L. Dang, F. Fry, W. Savary, P. Rogers, J. Barnett, W. Hill, S. Rippey, J. Wiskerchen, and M. Wekell. 1993. The regulatory fish encyclopedia: an internet-based com- pilation of photographic, textural and laboratory aid in spe- cies identification of selected fish species. U.S. Food and Drug Administration, http.llvm.cfscan.fda.govl~frflrfe0.html. Weaver, L. A., R. C. Lundstrom, and A. Colbert. 1999. Identification of shark species by polyacrylamide gel isoelectric focusing of sarcoplasmic proteins. J. AOAC (Association of Official Analytical Chemists) International 82:1163-1170. 356 Abstract— The reproductive cycle, sex- ual maturity, fecundity, and seasonal distribution of the anglerfish Lophius litulon were studied from a total of 989 specimens collected in the East China and Yellow seas between March 1991 and July 1997. Males and females reached 50% sexual maturity at 362 mm TL (age 5.4) and at 567 mm TL (age 6.2), respectively. The histology of the gonads showing seasonal changes in both the gonadosomatic index (GSI) and hepa- tosomatic index (HSI) are described. Mean GSI of females increases rapidly with ovarian development, whereas mean HSI decreases from the middle of vitellogenesis until the ovaries have fully matured. Segregation of oocytes by size within the ovary suggests that L. litulon spawns in batches, which was confirmed by observation of a captive specimen. Batch fecundity (BF) in 15 females with secondary yolk stage ova- ries was related to total length (TL, mm) by the equation BF = (-1.64 x 106) + 3688.13 TL (546100 pm in yolk vesicle stage ovaries were measured; in other ovarian stages, only oocytes >200 pm were measured. The size-frequency distribution of oocytes at each stage rep- resented the ovary from one randomly selected specimen. Estimates of size at sexual maturity were based on the examination of males (193-692 mm TL [n=236]) and fe- males (174-1013 mm TL |n=246]) collected during the spawning season between February and May. Of these, 38 females were defined as sexually immature, based on mac- roscopic observations (Afonso-Dias and Hislop, 1996). The remaining 444 specimens were classified into each gonad- al stage on the basis of histological observations. Sexually mature individuals were defined as males with testes in stages 3 and 4, and females with ovaries in mid-stage 2 (oocytes at the secondary yolk stage) or at more advanced stages (see Results). Males with testes in stages 3 and 4 collected in the spawning season were thought to have the potential to spawn because our preliminary observations indicated that milt ran from their genital pores on slight pressure. Females with ovaries with secondary yolk stage oocytes were considered sexually mature because an ad- vanced group of secondary yolk stage oocytes forms an iso- lated batch and increases in size in tandem with ovarian development. Females with ovaries in stages 2 (late), 3, 4, or 5 were also considered sexually mature, because these individuals were thought to be pre- or postspawning fish. Similarly, to estimate age at sexual maturity, the age of individual fish collected in the spawning season was deter- mined by counting the annual ring marks on the surface of the vertebral centrum (Yoneda et al., 1997). Of 482 spec- Yoneda et ai.: Reproductive cycle, fecundity, and seasonal distributions of Lophius litulon 359 Table 2 Summary of results of oocyte density (the number of secondary yolk stage oocytes per unit sam female Lophius litulon from various locations and results of two-way analysis of variance, int. pie weight [g] ) = interior; ext. in the ovaries of five = exterior. TL (mm) Position of sample in ovary* Right ovary Left ovary Anterior (int. or ext.) Middle (center) Posterior (int. or ext.) Anterior (int. or ext. ) Middle (center) Posterior (int. or ext.) 546 28.32ext 2711 2802ext 2768'nt 2960 3000int 647 3008ext 3113 3067,nt 3059int 3160 2980"“ 662 2917ext 3009 2759int 2913lnt 2635 3216"“ 796 2540int 2771 2691ext 2662ext 2990 2780"“ 847 2607int 2572 2821,nt 2680ext 2800 2800exl Mean 2781 2835 2828 2816 2909 2955 Two-way analysis of variance Source of variation df SS MS F Right vs. left ovary 1 23130 23130 0.73 Position within ovary 2 105500 52730 1.66 Interaction 2 48140 24070 0.76 Error 24 764700 31860 imens collected in the spawning season, 410 specimens (187 males and 223 females) were used to estimate age at sexual maturity. The vertebral centra of the other 72 spec- imens were either damaged in preparation or lost. To esti- mate the total length (L50) and age at which 50% of males and females are sexually mature, the fraction of mature fish in each interval ( 10-mm length or year of age) was fit- ted with a logistic function with the Marquardt method (Draper and Smith, 1966). All specimens TL > L50 were used to determine the monthly changes in gonadosomatic index (GSI) and hepa- tosomatic index (HIS) for adult males and females. The GSI and HSI were calculated in the following manner: GSI = (GW / (BW - VW)) x 100. HSI = {LW KBW- VW))x 100. The Kruskal-Wallis test (one-way analysis of variance, ANOVA) followed by Dunn’s multiple comparison test were used to test for significant differences between the GSI and HSI values of gonadal stage groups of fish. Estimation of batch fecundity followed Yoneda et al. (1998a). Batch fecundity was estimated only from spec- imens with ovaries containing oocytes in the secondary yolk stage and no gelatinous material. Only 15 females contained such ovaries during our study; these fish were thought to be ready to spawn for the first time during that spawning season because their ovaries contained nei- ther postovulatory follicles nor atretic oocytes. To deter- mine whether secondary yolk stage oocytes were randomly distributed throughout the ovary densities (no. oocytes/g ovary wt.) of secondary yolk stage oocytes from the six locations within the ovaries of five fish were compared (Ta- ble 2): two samples from the center of the right and left ovarian lobes, two samples from the posterior section (ei- ther from the interior or exterior) and two samples from the anterior section (either from the interior or exterior). A two-way ANOVA was performed to test for the effect of sample location on oocyte density within each ovary. There was no significant change in oocyte density by location of oocytes within the ovaries (Table 2). Advanced yolked (secondary yolk stage) oocytes were randomly distributed within the ovary and samples could be taken from any lo- cation without bias. Fecundity samples were collected from six different parts of the ovary in the anterior, middle, and posterior portion of each ovarian lobe. Ovarian tissue sam- ples (30-120 mg), each containing approximately 100-350 oocytes, were placed on a slide in water and covered with a cover slip. The most advanced oocytes were counted with a profile projector (50-100x). Batch fecundity for each female was calculated as the product of the number of secondary yolk stage oocytes per unit of weight (of each of the six sam- pling sites) multiplied by the total ovarian weight. Linear regression analysis was used to examine the relationship between batch fecundity and the total length of the fish. To examine the seasonal distribution of fish, the num- bers of specimens collected at each sampling station dur- ing each of the three study periods (September, Novem- ber-January, and February— May) were compared. The September samples were collected in the 1993 SNFRI trawl survey. Samples for the other two periods came from 360 Fishery Bulletin 99(2) the trawl surveys conducted by SNFRI between January and February in 1991, 1995, 1996, and 1997, and from the commercial trawl fishery in 1991-97. Additional samples for February— May were collected in the trawl survey con- ducted by Nagasaki University in May 1995. Sexually ma- ture individuals collected in September and November- January were defined as those larger than the L50 for that sex because most had recrudescent gonads. Between Feb- ruary and May, sexually mature individuals were defined as males with testes in stage 4 and females with ovaries in stages 2 (late), 3, 4, or 5 (see “Results” section). These indi- viduals were thought to be pre- or postspawning fish, pos- sibly collected at or near the spawning grounds. Sexually immature individuals were defined as males with testes in stages 1 or 2 and females with ovaries in stage 1. These in- dividuals were considered to be nonspawning fishes pres- ent during the spawning season. Results Structure of the testis The paired testicular lobes are located in the posterior por- tion of the abdominal cavity. The main longitudinal sperm duct is located ventral to the testicular groove (hilus) in each testis. The seminiferous tubules radiate towards and terminate blindly from the testicular periphery of the main sperm duct. The germinal cysts of spermatogonia, spermatocytes, and young spermatids are arranged ran- domly on the walls of the seminiferous tubules (Fig. 2A). As young spermatids mature, they are released from their cysts into the lumina of the seminiferous tubules (Fig. 2B). Spermatids and spermatozoa are both found in the lumina of the seminiferous tubules and sperm ducts (Fig. 2, B and C). Structure of the ovary The right and left ovarian lobes of L. litulon are con- nected to each other at their posterior ends, forming a single organ. Stalk-like ovigerous lamellae protrude from the ovarian wall and each contains many oocytes at dif- ferent stages of development. In developing and matur- ing ovaries, one or two of the most advanced oocytes are located in the terminal portion of each ovigerous lamella, and previtellogenic oocytes are found near the base of the ovigerous lamella throughout the year. The ovarian lumen is lined with both ovarian wall epi- thelium and ovigerous lamella epithelium. These epithelia undergo morphological changes accompanying the ovar- ian maturation cycle (Fig. 3). As ovarian development con- tinues in the secondary and tertiary yolk stages, gelati- nous material is secreted from both the ovigerous lamellar epithelium and ovarian wall epithelium, and fills the ovar- ian lumen. The early-stage postovulatory follicles are convoluted, with many folds, and contain a follicular lumen (Fig. 4A). The granulosa cells are either columnar or cuboidal and are arranged in a regular manner together with thecal A sc Figure 2 Photomicrographs of sections of the testis of Lophius litulon. (A) Transverse sections of the seminiferous tubules during spermatogenesis, showing that the germinal cysts are arranged ran- domly on the walls of the seminiferous tubules. ( B ) Transverse sections of the seminiferous tubules during spermatogenesis, showing that spermatids are released into the lumina of the seminiferous tubules. ( C ) Transverse section of the main sperm duct during spermatogenesis, showing that both spermatids and spermatozoa are present in the main sperm duct. Bar = 25 pm; sg = spermatogo- nia; sc = spermatocyte; st = spermatid; sz = sper- matozoon; de = main duct epithelium. cells (Fig. 4B). The nuclei are located in the basal or mid- dle portion of the granulosa cells. The late-stage postovu- latory follicles are much smaller than those in the previ- ous stage, and the follicular lumen continues to decrease Yoneda et al.: Reproductive cycle, fecundity, and seasonal distributions of Lophius litulon 361 in size until it disappears (Fig. 40. Ultimately, the granulosa cell layer becomes indistinct and the thecal cell layer is much regressed. Maturity stages of testes The testes can be classified into four stages of matu- rity according to their histological characteristics. No spent specimens, defined as those in which spermato- genesis has ceased and residual spermatozoa remain in the testis, were found during the study. Stage 1 immature (Fig. 5A): Germinal cysts containing spermatogonia, spermatocytes, and spermatids are observed along the wall of the seminiferous tubules. Sper- matids and spermatozoa are not present in the lumina of the seminiferous tubules and the small main duct. All specimens with testes at this stage were <306 mm TL. Stage 2 early spermatogenesis (Fig. 5B ): Germ cells at al! stages of spermatogenesis are present. Spermatids and a few spermatozoa are observed in the lumina of the seminiferous tubules and main sperm duct. Stage 3 late spermatogenesis (Fig. 50: Active sper- matogenesis occurs in the testes. Spermatids and sper- matozoa are more abundant in the lumina of the seminiferous tubules and in the main sperm duct than in the previous stage. Stage 4 mature (Fig. 5D): Large quantities of sperma- tozoa and a few spermatids are present in the lumina of the seminiferous tubules and main sperm duct. Sper- matogenesis and spermatogonial division also occurs in the seminiferous tubules, though few, if any, germinal cysts containing spermatogonia or spermatocytes are found around the main sperm duct. Maturity stages of ovaries The ovaries can be classified into six stages of maturity according to their histological characteristics and the de- velopment of the most advanced oocytes. The classification of stage-4, -5, and -6 ovaries is based on a modification of the atretic states of Hunter and Macewicz (1985). Stage 1 immature (Fig. 6A): Only previtellogenic oocytes are present and the epithelia of both the ovigerous lamel- lae and ovarian wall are thin. Stage 2 developing (Fig. 6B ): Most advanced oocytes have reached the primary to tertiary yolk stages. All of the ovig- erous lamellae have vitellogenic oocytes. This stage can be subdivided into early and late stages. The early stage is defined by the presence of primary or secondary yolk stage oocytes and the late stage by the presence of tertiary yolk stage oocytes with gelatinous material. Stage 3 mature (Fig. 60: The most advanced oocytes are in the migratory nucleus or mature stages. The ovu- A B Figure 3 Photomicrographs of the ovigerous lamellar epithelium and ovarian wall epithelium at various stages of ovarian matu- ration in Lophius litulon. (A) Ovigerous lamellar epithelium (ole) at the previtellogenic stage. (B) Ovarian wall epithelium (owe) at the previtellogenic stage. (A) and (B) show the epithe- lial cells of both the ovigerous lamella and ovarian wall are squamous or cuboidal in shape and contain a small nucleus. (C) Ovigerous lamellar epithelium at the tertiary yolk stage. (B) Ovarian wall epithelium at the tertiary yolk stage. (C) and (D) show that gelatinous material is actively secreted from the apical surfaces of the epithelia of both the ovigerous lamellae and ovarian wall. Bar = 25 pm; gm = gelatinous material. lated oocytes are found in the gelatinous material form- ing within the ovarian lumen just before spawning. A part of the remaining smaller oocytes contains early yolk stage oocytes. Stage 4 spawning (Fig. 6D): Vitellogenic oocytes with no sign of atresia and postovulatory follicles are present. Degenerating residual mature oocytes are frequently observed. More than 50% of the ovigerous lamellae have vitellogenic oocytes. Stage 5 spent (Fig. 6E): Vitellogenic oocytes are degener- ating (early atretic stage) and postovulatory follicles are observed. More than 50% of the ovigerous lamellae have only previtellogenic oocytes or atretic oocytes (or both). Stage 6 resting (Fig. 6F): Late atretic stage oocytes and previtellogenic oocytes are present, and the epithelia of both the ovigerous lamellae and ovarian wall are thin. To determine how frequently ovigerous lamellae were found with yolked oocytes after spawning, 30-50 oviger- ous lamella samples from the ovaries of 18 females with 362 Fishery Bulletin 99(2) Figure 4 Photomicrographs of atretic postovulatory follicles of Lophius litulon. (A) and (B) Early-stage postovulatory follicles. (B) shows that the granulosa cells are arranged in an orderly manner together with the thecal cells. (C) Late-stage postovulatory follicles. Bar = 75 pm; 1 = follicular lumen; g = granulosa cell layer; t = thecal cell layer. St Figure 5 Photomicrographs of testes in the four different stages of maturity in Lophius litulon. (A) Stage 1 (immature). < B i Stage 2 (early spermatogenesis). (C) Stage 3 (late spermatogen- esis). ( D) Stage 4 (mature). Bar = 100 pm; st = spermatid; sz = spermatozoon. Yoneda et al.: Reproductive cycle, fecundity, and seasonal distributions of Lophius litulon 363 Figure 6 Photomicrographs of ovaries at the six different stages of maturity in Lophius litulon. (A) Stage 1 (imma- ture), (B) Stage 2 (developing). (C) Stage 3 (mature). (D) Stage 4 (spawning). (E) Stage 5 (spent). (F) Stage 6 (resting). Bar = 250 pm; ow = ovarian wall; gm = gelatinous material; n = nucleus; o = oil droplet; pof = postovulatory follicle; ao = atretic oocyte. postovulatory follicles were examined. The frequency of ovigerous lamellae with yolked oocytes present clearly dif- ferentiates two ovarian conditions (Table 3). The ovaries of females in which more than 75.0% of the ovigerous la- mellae have secondary yolk stage oocytes are in stage 4, regardless of the degenerative stage of the postovulatory follicles. The ovaries of females in which less than 36.4% of the ovigerous lamellae have primary yolk stage oocytes are in stage 5. Some yolked oocytes in the process of be- coming atretic are found in specimens with primary yolk oocytes. No females were found with ovaries containing both postovulatory follicles and tertiary yolk or a more ad- vanced stage of oocytes during our study. Length and age at sexua! maturity There were clear differences between males and females in the size and age at sexual maturity (Fig. 7). The minimum size and age at sexual maturity for males were 325 mm TL and age 4; for females, minimum size and age at sexual maturity were 546 mm TL and age 5. The size and age at 50% sexual maturity for males and females were 362 mm TL and age 5.4, and 567 mm TL and age 6.2, respectively. All males >390 mm TL and age 7, and all females >630 mm TL and age 8, were mature. Annual reproductive cycle Mature males were found over a longer period of the year than mature females (Tables 4 and 5). Spermatogenesis occurs throughout most of the year; therefore males with mature testes are frequently collected. In females, early stages of oocyte development occur between November and March, and the later stages from February through April. Females in stages 3 and 4 were collected from Feb- ruary to May, which is therefore considered the spawning season. Between May and November, most females had stage 1, 5, or 6 ovaries. 364 Fishery Bulletin 99(2) labile 3 Results of histological examination of stage-4 and stage-5 ovaries of Lophius litulon collected between February and June. Stage-4 ovaries are defined as those in which more than 50% of the ovigerous lamellae have yolked oocytes in females with secondary yolk stage oocytes with no sign of atresia, regardless of the degenerative stage of the postovulatory (POF) follicles. Stage-5 ovaries are defined as those in which less than 50% (or none) of the ovigerous lamellae have yolked oocytes with frequent signs of atresia. GSI = gonadosomatic index. Date Total length (mm) GSI Oocyte stage7 POF stage2 Ovigerous lamellae with yolked oocytes (%) Atretic stage2 Ovarian stage 27 Feb 93 594 9.14 sy 1 75.0 4 20 Mar 95 634 7.00 sy 1 89.7 — 4 20 Mar 95 738 7.50 sy 1 93.0 — 44 18 Mar 96 622 5.31 py e 30.4 — 5 18 Mar 96 639 4.50 py i 17.6 e 5 18 Mar 96 711 6.96 sy l 97.7 — 4 18 Mar 96 1013 6.02 sy i 92.1 — 4 17 Mar 97 821 6.81 py e 20.6 — 5 17 Mar 97 853 5.76 py e 25.0 — 5 26 Apr 94 645 8.30 py i 20.0 e 5 14 Apr 95 857 8.60 py i 16.4 e 5 23 Apr 96 702 6.38 sy e 92.3 — 4 26 Apr 97 868 8.45 py i 30.8 e 5 5 May 93 626 6.48 sy e 77.6 — 4 28 May 94 891 7.70 py i 18.2 e 5 28 May 94 981 9.20 py i 36.4 e 5 7 May 97 825 3.47 yv i 0 e, 1 5 4 7 Jun 97 830 4.20 yv i 0 e, 1 5 ' Most advanced oocytes are py = primary yolk stage; sy = secondary yolk stage; and yv = yolk vesicle stage. - Postovulatory follicles are e = early stage; 1 = late stage. 1 Atretic oocytes are e = early stage; 1 = late stage. J Photographs of histological sections of stage-4 and stage-5 ovaries are shown in Figure 6 (D and E). Gonadosomatic and hepatosomatic index The mean GSI for males increased from September and peaked in January after which it declined (Fig. 8A). The mean HSI for males was highly variable between Sep- tember and December but gradually decreased from Jan- uary through July (Fig. 8B). The mean GSI for females increased gradually throughout the early part of the year, peaking sharply in May, after which it dropped rapidly (Fig. 8C). The mean HSI for females started to increase in August and peaked in December (Fig. 8D). It remained low between March and July. With testicular development, the GSI increased and reached a maximum when the testes were in stage 4 (Ta- ble 6). There were significant differences in the value of these indices in the stage-4 testes, compared with those in the other three stages (P< 0.05). The mean HSI for males peaked in stage- 1 testes, but there were no statistical dif- ferences in the value of the median HSI between the four stages (ANOVA, P> 0.05). In females, the mean GSI gradually increased and peaked in stage-3 ovaries (Table 7), but there were no statistical differences in the value of the median GSI between the early stage-2 and stage-5 ovaries (P>0.05). The mean HSI peaked in early stage-2 ovaries and then dropped rapidly with ovarian development. The median HSI during the early and late stage-2 ovaries was signifi- cantly higher than in stage- 1, stage-3, and stage-5 ovaries (P<0.05). Size-frequency distribution of oocytes The size of all the oocytes in a group gradually increased in tandem with ovarian development (Fig. 9). When an advanced group of the oocytes reached the secondary yolk stage, they formed an isolated batch that separated almost completely from adjacent groups of smaller oocytes. Between the tertiary yolk and mature ovary stages, only the oocytes in the advanced batch increased in size, and the remainder of the group remained smaller than 550 pm. This finding would suggest that L. litulon is a batch spawner (a conclusion which has been supported by aquar- ium observations — see “Discussion” section). Batch fecundity The relationship between batch fecundity (BF) and total length (TL), based on 15 specimens with secondary yolk Yoneda et al.: Reproductive cycle, fecundity, and seasonal distributions of Lophius litulon 365 stage ovaries collected between December and February, was described by the equation BF = (-1.64 x 106) + 3688.13 TL for total lengths between 546 and 846 mm (r-=0.86; P<0.001) (Fig. 10). Batch fecundity ranged from 0.31 x 106 eggs in a 578-mm-TL fish, to 1.54 x 101’ eggs in a 796-mm- TL fish. Seasonal distribution In September, most of the specimens from both sexes were collected in the Yellow Sea (Fig. 11). Between November and January their distribution extended from the Yellow Sea to the East China Sea. At this time, sampling sites showed a clear difference in the distribution of sexually mature males and females. Males were collected mainly in the East China Sea, whereas females were collected only in the Yellow Sea. During the spawning season, from Table 4 Number of males Lophius litulon in the East China and Yellow seas at the various maturity stages of the testes by month. Only data from specimens larger than the mini- mum size at sexual maturity for males (TL=325 mm) have been used in this table. Stage 2 = early spermatogenesis; stage 3 = late spermatogenesis; stage 4 = mature. Maturity stage of the testes Month 2 3 4 Jan 22 5 12 Feb 40 14 7 Mar 1 4 17 Apr 3 3 5 May 4 24 18 Jun 3 Jul 2 3 Sep 1 3 Oct 8 15 Nov 14 Dec 2 17 Table 5 Number of female Lophius litulon in the East China and Yellow seas at the various maturity stages of the ovaries by month. Only data from specimens larger than the min- imum size at sexual maturity for females (TL=546 mm) have been used in this table. Stage 1 = immature; stage 2 = developing; stage 3 = mature; stage 4 = spawning; stage 5 = spent; stage 6 = resting. Maturity stage of the ovary Month 1 2 (early) 2 (late) 3 4 5 6 Jan 5 Feb 18 4 1 Mar 3 4144 Apr 2 3 13 May 1 3 2 Jun 1 1 Jul 1 2 Aug 1 Sep 3 Nov 3 3 Dec 5 February throughout May, immature individuals were collected throughout the East China and Yellow seas, whereas mature individuals were caught only in the East China Sea and the coastal waters off Kyushu, and none were caught in the Yellow Sea. 366 Fishery Bulletin 99(2) Month Figure 8 Monthly changes in the mean gonadosomatie index (GSI) and hepatosomatic index (HSI) for mature male (n=306) and female (n= 67 ) Lophius litulon in the East China and Yellow seas. Only specimens larger than the size at 50% sexual maturity for males (L50=362 mm) and females (Lso=567 mm ) were used in our study. Vertical lines indicate standard error. Discussion The testicular structure of L. litulon is similar to that of other teleosts with unrestricted spermatogonial (Grier et al., 1980; Grier, 1981) or lobular type testes (Billard et ah, 1982; Billard, 1986). Although the process of sper- matogenesis conformed to that of other teleosts, it was not completed within the germinal cysts. Rather, spermatids were released into the lumina of the seminiferous t ubules and did not differentiate synchronously. This form of sper- matogenesis first described in Lepadogaster lepadogaster (Mattei and Mattei, 1978), and later termed “semi-cystic” type (Mattei et ah, 1993), has subsequently been reported in Neoceratiidae ( Jespersen, 1984), blennioid fishes (Lahn- steiner and Patzner, 1990), Ophidion sp. (Mattei et ah, 1993), Opistognathus whitehurstii (Manni and Rasotto, 1997), and L. setigerus (Yoneda et ah, 1998c). The ovarian structure of L. litulon is similar to that reported in other Lophiiformes such as L. piscatorius (Afonso-Dias and Hislop, 1996), Antennarius scaber, His- trio histrio, and Ogcocephalus vespertilio (Rasquin, 1958), L. americanus (Armstrong et ah, 1992), and L. setigerus (Yoneda et ah, 1998a). Most female Lophiiformes are thought to spawn gelatinous egg masses, within which in- dividual eggs float in separate chambers (Rasquin, 1958; Armstrong et ah, 1992; Afonso-Dias and Hislop, 1996; Yoneda et ah, 1998a, 1998c). In L. litulon, as in other lophii- form fishes (Rasquin, 1958; Armstrong et ah, 1992; Yone- da et ah, 1998a, 1998c), gelatinous material was secreted from the epithelium of both the ovigerous lamellae and the ovarian wall. Rasquin (1958) compared the structure of the ovary of H. histrio with that of the released egg mass and concluded that the shape of the egg mass was a replica of the internal surface of the ovary. This is expected to be the case in other fishes of the family Lophiiformes. Each stalk-like ovigerous lamella is thought to serve as a “mold,” forming a separate chamber within the gelati- nous egg mass. The arrangement of oocytes, with the most advanced oocytes at the margins of the ovigerous lamel- lae, may facilitate the release of mature oocytes into each chamber. Our examination of the gonadal condition of both sexes indicates that L. litulon spawns during the period from February through May. Most females with late-develop- ing, mature, or spawning ovaries are found in March and April. This finding agrees with previous reports. The peak of the spawning season of L. litulon occurred between Feb- ruary and March in inshore waters off Kyushu Island (Mi- to, 1963) and in March and April in the East China and Yellow seas (Yamada, 1986). In Sendai Bay, off the north- east coast of Honshu Island, Japan, the spawning season of L. litulon occurs between May and July (Kosaka, 1966). Yoneda et al.: Reproductive cycle, fecundity, and seasonal distributions of Lophius litulon 367 This difference in spawning season depending on latitude would indicate that the spawning season of L. litulon in Japanese waters occurs progressively later, with increas- ing latitude. This also occurs with L. americanus (Bigelow and Schroeder, 1953) in American waters and with L. pis- catorius (Afonso-Dias and Hislop, 1996) in northern Euro- pean waters. After spawning, ovaries in two conditions (stages 4 and 5) were observed during our study. Females had postovula- tory follicles and yolked oocytes at the primary or second- ary yolk oocyte stages. Secondary yolk stage oocytes were more often present in the ovaries of females collected in the Table 6 Gonadosomatic index ( GSI, mean ± SE ) and hepatosomatic index (HSI, mean ± SE) at each stage of maturity for male and female Lophius litulon. Values with different super- scripts are significantly different (Dunn’s multiple com- parison test, P<0.05). See tables 5 and 6 for definitions of maturity stages. Maturity stage of the gonads n GSI HSI Male 1 47 1.17 ±0.25" 5.57 ±0.77 2 143 2.00 ±0.22" 4.70 ±0.44 3 63 2.09 ±0.10" 4.80 ±0.37 4 114 2.48 ±0.07'’ 4.50 ±0.24 Female 1 206 1.34 ±0.08" 5.40 ±0.25" 2 (early) 24 5.73 ±0.42'’ 10.38 ±0.70'' 2 (late) 10 15.27 ±2.02'’ 11.21 ±1.30'’ 3 4 51.78 ±5.49'’ 3.58 ±0.23" 4 7 7.07 ±0.39'' 6.03 ±0.84"' 5 11 6.58 ±0.61'’ 4.34 ±0.53" 6 5 2.85 ±0.37" 6.37 ±0.94"' first half of the spawning period ( February-March ). It seems likely that these normally develop into yolked oocytes because no signs of oocyte atresia were observed in histo- logical examination. Specimens with primary yolk oocytes were collected between March and May. However, there was evidence of atresia in half of the fish collected during the latter half of the spawning period (April-May). 1 20° E 125°E 130°E 120°E 125°E 130°E 120°E 125°E 130°E 368 Fishery Bulletin 99(2) A solitary female L. litulon, in Oarai Aquarium, released an infertile egg mass on 19 April 1994 and 35 days later extruded another mass.2 Similarly, another captive female spawned three times, on 15 February, 2 June, and 21 July c ^ ■5 .s S 0J a h m 5 " 2 p h CD ^ cd 6 fe P ~Q c g i> 5 LO P 2 g O P c o ° 0 O S* n a; c ~ II M c s ® s £ a 13 to c « £ ^ a -d T '3 (D ^ -m ^ ■t— i O • ^ 4-h c/3 5-< - -rt ^ 9 1998. These data indicate that L. litulon may have the po- tential to spawn more than once a year, although the five spawnings observed in the aquarium were not accompanied by normal spawning behavior. Recently, we reported a case of repeated spawning in L. setigerus, which has a long spawning period from May to No- vember (Yoneda et al., 1998a). In contrast, L. americanus (Feinberg, 1984) and L. piscato- lius (Afonso-Dias and Hislop, 1996) are be- lieved to spawn only once a season. Future field studies are required to determine the spawning frequency for L. litulon , which will be important for estimating its reproductive potential. Both sexes of L. americanus inhabiting northern waters were larger and a little younger at 50% sexual maturity than fish from more southern waters (Almeida et ah, 1995). In L. litulon, there also appears to be a size difference at sexual maturity between fish from the East China and Yel- low seas (our study) and those from Sendai Bay off northeast Honshu Island (Kosaka, 1966); the minimum size at sexual matu- rity for males and females was 340 mm body length (BL) and 600 mm BL for Ko- saka’s report and 325 mm TL and 546 mm TL for our results, respectively. The study from Sendai Bay, however, was carried out about 30 years ago and did not identify the age at sexual maturity; therefore the dif- ferences by area for size at sexual maturi- ty may be attributed to different sampling times of the year or to different growth rates. In the East China and Yellow seas, females L. litulon reached sexual maturity at larger than 500 mm BL, and the min- imum size at sexual maturity was about 350 mm BL in male, as previously report- ed by Yamada (1986). The size at sexual maturity for female L. litulon in the 1980s (Yamada, 1986) is fairly close to our re- sults, whereas that for males is higher than that found in our study. Although the reason for the size difference found in males at sexual maturity is unclear, it may be due to a difference in criteria for ma- turity. Our histological criteria for maturi- ty in males were based on the characteris- tics of testicular development with unique spermatogenesis; therefore our determina- tion of sexual maturity is more reliable than macroscopic methods. There was a noticeable inverse correla- tion between the development of the ovary and the weight of the liver (HSI). The de- 03 JJ TL i; p f 03 C CD •a 5 % d CD -2 P C/3 03 P q 1 _ CD 8 O £ 0) lO « c 1 ^ (D D w ^ 'h 3 P ® -e g a > -2 Pc1' .5 -S m Sac « 5 S d £ £ O >> w 03 C/3 P cc X m 03 l; C/3 ^ C/3 d 03 ^ P d cp I >> p p 03 P rC bb d) § | ° P 9" S oi cp £ S d ^ Sq ■p 03 ° g § S3 O 32 £ H LO CO O CO 2 Kofuji, K. 1998. Personal commun. Oarai Aquarium, Isohama, Oarai, Higashi-Ibaraki, Ibaraki 311-1301, Japan. Yoneda et al.: Reproductive cycle, fecundity, and seasonal distributions of Lophius litulon 369 crease in HSI that is associated with increasing GSI is probably due to materials that have been stored in the liv- er becoming mobile and being transferred to the gonads. In teleosts, as in most other vertebrates, the precursor protein of yolk (vitellogenin) is synthesized in the liver. The secreted vitellogenin is selectively removed from the bloodstream by the developing oocytes (Wallace and Sel- man, 1981; Nagahama, 1987). The rapid accumulation of yolk probably accounts for the decrease in the weight of the liver. In many fish, batch fecundity is estimated by using mi- gratory nuclei or hydrated oocytes, which can be easily distinguished from the less advanced oocytes: e.g. Engrau- lis mordax (Hunter and Goldberg, 1980; Hunter et al., 1985), Thunnus albacares (Schaefer, 1996), and Rhombop- lites aurorubens (Cuellar et al., 1996). We found that dur- ing and after the tertiary yolk stage, a large amount of ge- latinous material was rapidly secreted and accumulated in the ovarian lumen. Hence, counts of advanced oocytes from a small portion of the ovary, when extrapolated to the total weight of the gelatinous material, may give more variable estimates. These findings were also evident in L. setigerus (Yoneda et al., 1998a). However, the oocyte size- frequency profiles indicated that when the most advanced oocytes reached the secondary yolk stage, they formed a batch that was almost completely separated from the ad- jacent group of smaller oocytes. These ovarian character- istics of L. litulon imply that estimates of batch fecundity can be made only by using oocytes that have attained the secondary yolk stage. Both the immature and mature distribution of L. litu- lon ranged into the East China and Yellow seas, as pre- viously reported by Yamada (1986) and Tokimura.1 This species is caught mainly at depths between 50 and 100 m and at temperatures ranging from 6 to 13°C (Yamada, 1986; Tokimura1). In the Yellow Sea, the Yellow Sea Cen- tral Cold Water, cooler than 10°C, is found throughout the year and there are few seasonal changes in water temper- ature (±2-3°C). The water of the East China Sea remains lower than 13°C owing to the influence of the Continental Coastal Cold Water in the winter and spring, whereas in summer it increases higher than 20°C (Kondo, 1985; To- kimura1). These oceanographic conditions in the East Chi- na and Yellow seas may influence the migration of L. li- tulon from area to area throughout the year. A seasonal movement of L. litulon has also been reported in Sendai Bay (Kosaka, 1966; Omori, 1979). Lophius litulon are most abundant in shallow waters between February and June. From August through October, they disperse toward deep- er waters. This seasonal movement in Sendai Bay is ob- served mainly in immature fish and is thought to be as- sociated with their feeding activities (Kosaka, 1966). In L. arnericanus, a seasonal migration, thought to occur in response to changes in hydrographic conditions, has been observed along the northeastern coast of the United St ates (Jean, 1965; Almeida et al., 1995). During the February-May spawning season, mature males and females with ovaries in a condition that sug- gests they are either about to spawn, or have just spawned, are found in the East China Sea and the coastal waters off Kyushu. In contrast, immature individuals were distribut- ed throughout the East China and Yellow seas during this same period. This indicates that the spawning grounds of L. litulon cover a large area, extending from the East Chi- na Sea to inshore waters off Kyushu. Furthermore, our study reveals the migratory pattern of both sexes of L. li- tulon in relation to the spawning grounds in the period before the spawning season. In February, with the onset of the spawning season, females collected in the northern East China Sea had developing ovaries (secondary or ter- tiary yolk stage), whereas those collected in the Yellow Sea had immature or primary yolk stage ovaries. This finding implies that the start of the migration of females seems to be more dependent on ovarian development than ocean- ographic conditions. Different migratory patterns of the two sexes before spawning have also been reported in Eu- ropean plaice, Pleuronectes platessa, in the Straits of Do- ver off England (Arnold and Metcalfe, 1995). Our study has identified the spawning grounds and migratory pat- tern of L. litulon in broad terms. Further research is need- ed to identify specific spawning grounds and details of the migratory behavior of this species. Acknowledgments We are grateful to the officers and crew of the RV Kcuho Maru, Okinawa Prefecture, and the training ship Naga- saki Maru , Nagasaki University, for allowing us to board their vessels and to collect specimens, and to many fisher- men, especially members of the Yamada Suisan Company Ltd. and the Murayama Suisan Company Ltd., for helping us to obtain samples. We wish to express our gratitude to K. Kofuji for his report on spawning activity observed at Oarai Aquarium, and to R. G. Bakkala and B. N. Campbell for offering valuable suggestions and for critically reading this manuscript, and to anonymous reviewers for review- ing the manuscript and suggesting many improvements. Literature cited Afonso-Dias, I. R, and J. R. G. Hislop. 1996. The reproduction of anglerfish Lophius piscatorius Linnaeus from the north-west, coast of Scotland. J. Fish Biol. 49 (suppl. A ): 18—39. Almeida, F. P., D. Hartley, and J. Burnett. 1995. Length-weight relationships and sexual maturity of goosefish off the northeast coast of the United States. N. Am. J. Fish. Manage. 15:14—25. Armstrong, M. R, J. A. Musick, and J. A. Colvocoresses. 1992. Age, growth, and reproduction of the goosefish Lo- phius arnericanus (Pisces: Lophiiformes). Fish. Bull. 90: 217-230. Arnold, G. R, and J. II. Metcalfe. 1995. Seasonal migrations of plaice ( Pleuronectes platessa) through the Dover Strait. Mar. Biol. 127:151-160. Bigelow, H. B., and W. C. Sehroeder. 1953. Fishes of the Gulf of Maine. Fish. Bull. 53:532-541. Billard, R. 1986. Spermatogenesis and spermatology of some teleost fish species. Reprod. Nutr. Develop. 26:877-920. 370 Fishery Bulletin 99(2) Billard, R., A. Fostier, C. Weil, and B. Breton. 1982. Endocrine control of spermatogenesis in teleost fish. Can. J. Fish. Aquat. Sci. 39:65-79. Caruso, J. H. 1983. The systematics and distribution of the lophiid angler- fishes. II: Revisions of the genera Lophiomus and Lophius. Copeia 1983:11-30. Cuellar, N., G. R. Sedberry, and D. M. Wyanski. 1996. Reproductive seasonality, maturation, fecundity, and spawning frequency of the vermilion snapper, Rhombop- lites aurorubens , off the southeastern United States. Fish. Bull. 94:635-653. Draper, N. R., and H. Smith. 1966. Applied regression analysis. Morikita Shuppan, To- kyo, 378 p. [In Japanese.] Feinberg, M. N. 1984. All head and mouth. Nat. Hist. 93:28—33. Grier, H. J. 1981. Cellular organization of the testis and spermatogen- esis in fishes. Am. Zool. 21:345-357. Grier, H. J., J. R. Linton, J. F. Leatherland, and V. L. de Vlaming. 1980. Structural evidence for two different testicular types in teleost fishes. Am. J. Anat. 159:331-345. Hunter, J. R., and S. R. Goldberg. 1980. Spawning incidence and batch fecundity in northern anchovy, Engraulis mordax. Fish. Bull. 77:641-652. Hunter, J. R., N. C. H. Lo, and R. J. H. Leong. 1985. Batch fecundity in multiple spawning fishes. In An egg production method for estimating spawning biomass of pelagic fish: application to the northern anchovy, Engrau- lis mordax (R. Lasker, ed.), p. 67-77. LIS. Dep. Commer., NOAA Tech. Rep. NMFS 36. Hunter, J. R., and B. J. Macewicz. 1985. Measurement of spawning frequency in multiple spawning fishes. In An egg production method for estimat- ing spawning biomass of pelagic fish: application to the north- ern anchovy, Engraulis mordax (R. Lasker, ed.), p. 79-94. U.S. Dep. Commer., NOAA Tech. Rep. NMFS 36. Jean, Y. 1965. Seasonal distribution of monkfish along the Canadian Atlantic mainland. J. Fish Res. Board Can. 22: 621-624. Jespersen, A. 1984. Spermatozoans from a parasitic dwarf male of Neoc- eratias spinifer Pappenheim, 1914. Vidensk. Meddr dansk naturh. Foren. 145:37-42. Kosaka, M. 1966. Feeding habits of angler fish, Lophius litulon. J. Fac. Oceanogr. Tokai Univ. 1:51-70. [In Japanese.] Kondo, M. 1985. Oceanographic investigations of fishing grounds in the East China Sea and the Yellow Sea. I: Characteristics of the mean temperature and salinity distributions mea- sured at 50 m and near the bottom. Bull. Seikai Reg. Fish. Res. Lab. 62: 19-66. [In Japanese.] Lahnsteiner, F., and R. A. Patzner. 1990. Spermiogenesis and structure of mature spermato- zoa in blenniid fishes (Pisces, Blenniidae). J. Submicrosc. Cytcl . Pathol. 22:565-576. Manni, L., and M. B. Rasotto. 1997. Ultrastructure and histochemistry of the testicular efferent duct system and spermiogenesis in Opistognathus whitehurstii (Teleostei,Trachinoidei). Zoomorphology 117: 93-102. Mattei, C., and X. Mattei. 1978. La spermiogenese d’un poisson teleosteen ( Lepado - gaster lepadogaster ). II: Le spermatozoide. Biol. Cellulaire 32: 267-274. [In French.] Mattei, X., Y. Siau, O. T. Thiaw, and D. Thiam. 1993. Peculiarities in the organization of testis of Ophidion sp. (Pisces Teleostei): evidence for two types of spermato- genesis in teleost fish. J. Fish Biol. 43:931-937. Mito, S. 1963. Pelagic fish eggs from Japanese waters. X: Gadida and Lophiida. Japanese J. Ichthyol. 11:103-113. [In Japanese.] Nagahama, Y. 1987. Gonadotropin action on gametogenesis and steroido- genesis in teleost gonads. Zool. Sci. 4:209-222. Omori, M. 1979. Studies on the production ecology of demersal fishes in Sendai Bay. Ill: Comparison of fish fauna between the areas of different abiotic environmental condition. Bull. Seikai Reg. Fish. Lab. 52:91-129. [In Japanese.] Rasquin, P. 1958. Ovarian morphology and early embryology of the pediculate fishes Antennarius and Histrio. Bull. Am. Mus. Nat. Hist. 114:327-372. Schaefer, K. M. 1996. Spawning time, frequency, and batch fecundity of yel- lowfin tuna, Thunnus albacares, near Clipperton Atoll in the eastern Pacific Ocean. Fish. Bull. 94:98-112. Wallace, R. A., and K. Selman. 1981. Cellular and dynamic aspects of oocyte growth in tele- osts. Am. Zool. 21:325— 343. Yamada, U 1986. Lophiomus setigerus and Lophius litulon. In Fishes of the East China Sea and the Yellow Sea (Okamura, ed.), p. 106-109. Seikai Reg. Fish. Res. Lab., Nagasaki. [In Japanese.] Yamamoto, K. 1956. Studies on the formation of fish egg. I: Annual cycle in the development of ovarian eggs in the flounder, Liop- setta obscura. J. Fac. Sci Hokkaido Univ. Ser. IV, Zool. 12: 362-373. Yoneda, M., M. Tokimura, H. Fujita, N. Takeshita, K. Takeshita, M. Matsuyama, and S. Matsuura. 1997. Age and growth of anglerfish Lophius litulon in the East China Sea and the Yellow Sea. Fisheries Sci. 63:887- 892. 1998a. Ovarian structure and batch fecundity in Lophio- mus setigerus. J. Fish Biol. 52:94-106. 1998b. Age and growth of the anglerfish Lophiomus setigerus in the East China Sea. Fisheries Sci. 64:379-384. 1998c. Reproductive cycle and sexual maturity of the angler- fish Lophiomus setigerus in the East China Sea with a note on specialized spermatogenesis. J. Fish Biol. 53:164-178. 371 Differentia! parasitism by Naobranchia occidentals (Copepoda: Naobranchiidae) and Nectobrachia indivisa (Copepoda: lernaeopodidae) on northern rock sole ( Lepidopsetta polyxystra Orr and Matarese, 2000) and southern rock sole (£. bilineata Ayres, 1855) in Alaskan waters Abstract— Northern rock sole ( Lepi- dopsetta polyxystra Orr and Matarese, 2000) and southern rock sole (L. bilin- eata Ayres, 1855) from the Gulf of Alaska and northern rock sole from the Aleutian Islands were examined for gill parasites. Four species of copepod parasites were identified: Naobranchia occidentalis and Nectobrachia indivisa were the most common. Both parasites were more prevalent on northern rock sole (22% and 15%, respectively) than on southern rock sole (5% and 1%, respectively) in the Gulf of Alaska sam- ples. Northern rock sole tended to have a greater mean intensity of Naobran- chia occidentalis than southern rock sole but there was not a significant dif- ference because of the high variance about the means; too few southern rock sole were infested by Nectobrachia indi- visa for comparison. Northern rock sole from the Aleutian Islands region had a significantly greater prevalence (36%>) and mean intensity (10.2/infested fish) of Naobranchia occidentalis than north- ern rock sole from the Gulf of Alaska (22%, and 4.4, respectively) but did not differ significantly in prevalence and mean intensity of Nectobrachia indi- visa. Parasitized male northern rock sole from the Gulf of Alaska had a significantly reduced weight at length, indicating a possible effect of para- sitism. Naobranchia occidentalis selec- tively infested larger northern rock sole and only the largest southern rock sole. Nectobrachia indivisa also were found on larger northern rock sole but did not infest enough southern rock sole to describe a trend. Southern rock sole males were not infested by either parasit e. Naobranchia occidentalis pre- ferred to infest the middle gill arches of hosts and Nectobrachia indivisa pre- ferred to infest the exterior gill arches of hosts. Manuscript accepted 14 November 2000. Fish. Bull. 99:371-380 (2001). Mark Zimmermann Robin C. Harrison Anthony F. Jones Alaska Fisheries Science Center National Marine Fisheries Semce, NOAA 7600 Sand Point Way NE Seattle, Washington 98115-0070 Email: Mark.Zimmermann@noaa.gov The rock soles (Lepidopsetta spp.) are important commercial flatfish species that inhabit the continental shelf of the north Pacific Ocean (Hart, 1973). Stock assessment bottom trawl surveys con- ducted by the Alaska Fisheries Science Center (AFSC) historically recognized only one species of rock sole occurring in four survey areas: eastern Bering Sea, Aleutian Islands, Gulf of Alaska, and western U.S. coast. Recent larval mor- phometries by Matarese et al. ( 1989 ) and a generic revision by Orr and Matarese (2000) have demonstrated that there are actually two species of rock sole that overlap in the AFSC survey areas. The northern rock sole (L. polyxystra Orr and Matarese, 2000) ranges throughout the Bering Sea and Aleutian Islands to Puget Sound, whereas the southern rock sole (L. bilineata Ayres, 1855) ranges from the Islands of Four Moun- tains in the eastern Aleutian Islands to Mexico (Orr and Matarese, 2000). With the knowledge of this recent research, field biologists began separating the two rock sole species in AFSC bottom trawl surveys starting in 1996. The northern rock sole is distinguished by higher gill- raker counts and a whiter blind side than southern rock sole occurring in the same trawl hauls (Orr and Matarese, 2000). While examining gill rakers to iden- tify specimens of rock soles captured near the Shumagin Islands during the 1996 Gulf of Alaska survey, we noticed that several live, apparently healthy, northern rock soles had pink or white gill filaments, a condition normally as- sociated with dead fish. On close exam- ination, we found small copepod para- sites attached to the gill filaments of these fish. Southern rock soles from the same trawl hauls usually had red, apparently healthy gill filaments and were less frequently parasitized. There- fore we surveyed northern and south- ern rock soles in selected hauls from the Gulf of Alaska in 1996 for the pres- ence of these parasites and saved infest- ed heads for parasite identification and enumeration in the laboratory. In 1997, we conducted the same parasite inves- tigation in several hauls in the Aleu- tian Islands region. Our main objective was to quantify differences in the prev- alence (percent fish infested) and mean intensity (average number of parasites per infested fish) of gill parasites in- festing the northern and southern rock soles (see Margolis et al., 1982; Bush et al., 1997, for complete definitions of eco- logical terms in parasitology). Second- ary objectives were to test for regional differences in prevalence and mean in- tensity, to describe and compare site preferences of the parasites, and to de- termine if parasites influenced weight at length for parasitized fish. 372 Fishery Bulletin 99(2) 178° 176° 174° 172° 170° 168° 166° 164° 162° 160° 158° 156° 154° 152° Location of trawl hauls in the Gulf of Alaska and Aleutian Islands where northern and southern rock sole were examined for gill parasites. Materials and methods Between 8 and 18 June 1996 (Table 1), northern and southern rock soles captured in a series of trawl hauls from the western Gulf of Alaska were examined for para- sites (Fig. 1). Between 25 and 27 June 1997 (Table 1), we examined northern rock soles from a series of trawl hauls around Seguam Island and the east end of Amlia Island in the central Aleutian Islands. The two sampling areas are approximately 850 km apart. We scanned the interior and exterior surfaces of gill filaments on all gill arches for parasites. All rock soles were examined in hauls with few fish (n<50), and a random selection of fish in larger hauls were subsampled in a manner similar to that used for obtaining a random subsample for lengths (see Martin, 1997, for Gulf of Alaska bottom trawl survey methods). For each parasitized fish, we measured the fork length (FL) in centimeters, weight (±2 grams), determined the sex, and removed and froze the entire head, including gill arches, at sea. Lengths and weights were also collected from unin- fested northern and southern rock soles in the Gulf of Alaska survey for comparisons with parasitized fish. Frozen heads were thawed slowly in a refrigerator; thaw- ing heads quickly in warm water produced gill filaments with an excess of mucus, making it difficult to locate para- sites. Host species identification was confirmed by examin- ing the gill rakers on the first gill arch on the blind side, and by counting supraorbital pores (Orr and Matarese, 2000). The internal and external surfaces of all gill filaments were examined for parasites with a dissecting microscope. The parasites were removed from gill filaments by washing with water into a filter, or pulling with tweezers, or by removing the gill filament if the parasite was well-anchored. All par- asites were preserved in 10% formalin for 3-5 days and stored in 70% ethanol with a small amount of glycerol (Ka- bata and Cousens, 1972). Fish heads were fixed in 10% for- malin for up to one week and stored in 70% ethanol. We identified the parasites by using the key of Kabata (1988), and Kabata verified our identifications. Generally, only female parasites, which measured at least 2 millime- ters in length, were enumerated. Males were found occa- sionally, sometimes attached to females. However, males were too small (<0.3 mm in length) to identify and quanti- fy; therefore males were not included in any summaries or Zimmermann et at Parasitism by Naobranchia ocadentahs 373 Table 1 Summary of field collection information for northern and southern rock sole from the two sampling areas of the western Gulf of Alaska and the central Aleutian Islands. Area Dates Hauls Depth range (m) Temperature range (Celcius) Gulf of Alaska 8 to 18 Jun 1996 18 46-149 4. 0-7. 4 Aleutian Islands 25 to 27 Jun 1997 8 143-232 3. 9-4. 5 Table 2 Summary of fish examined at sea and parasite collections. Mean intensity = average number of parasites for each infested fish. Range is from minimum to maximum numbers of parasites for each infested fish. Naobranchia occidentalis Nectobrachia indivisa Area and rock sole species Fish examined at sea for gill parasites Fish infested with either gill parasite Prevalence (%) Mean intensity Range Prevalence (%) Mean intensity Range Gulf of Alaska Northern 225 58 22 4.4 1-22 15 3.8 1-21 Southern 282 15 5 2.9 1-13 1 2.0 1-3 Aleutian Islands Northern 237 89 36 10.2 1-45 9 3.1 1-8 analyses. Differences in parasite prevalence between the northern and the southern rock sole were determined with chi-square tests, and differences in mean intensity were determined with Welch’s approximate 1-test (Zar, 1984). To test for significant differences in the length-weight relationship between parasitized (containing Naobranchia occidentci/is or Nectobrachia indivisci , or both) and unpar- asitized fish, natural log-transformed length and weight data were compared by linear regression. The slopes and T-intercepts were compared by using F-ratios. Results Parasite species Four species of copepod parasites were identified from the gills of the infested fish. The most common parasite was Naobranchia occidentalis Wilson, 1915, which attaches to the gill filaments by its modified second maxillae (Kabata, 1988). This parasite was easily removed in the laboratory; freezing and thawing the specimens may, therefore, have loosened the parasite’s grip. The next most common para- site, Nectobrachia indivisci Fraser, 1920, firmly anchors in the filament and could only be removed by cutting the filament or breaking off the bulla. Nectobrachia inch- visa was almost always found attached to the tips of the gill filaments. Two specimens of Acanthochondria vancou- verensis were found unattached in the gill chambers of two Gulf of Alaska northern rock sole, constituting the first record of parasitism in wild rock sole for this species (Kabata1). A single specimen of Haemobciphes diceraus was found attached on the third gill arch, eyed side, of a Gulf of Alaska northern rock sole (Kabata2). Kabata ( 1988) reported that all of these parasites occur on rock sole, except for A. vancouverensis. Small, highly mobile leeches were also frequently noted on the gills of fish caught in the Gulf of Alaska survey, but because of their mobility, their presence could not be reliably documented. We did not attempt to enumerate or identify this species. Gulf of Alaska A total of 225 northern and 282 southern rock soles were visually examined for parasites at sea during the 1996 Gulf of Alaska survey, and 78 rock sole heads were collected because they appeared to have at least one copepod gill parasite (Table 2). Naobranchia occidenta- lis or Nectobrachia indivisa (or both) were found on 73 of the collected heads during laboratory dissection (58 north- ern and 15 southern rock sole). Only one juvenile female or Kabata, Z. 1998. Personal commun. Department of Fisher- ies and Oceans, Pacific Biological Station, Nanaimo, British Columbia, Canada, V9R 5K6. Kabata suggested that we consider this incidence of parasitism by A. vancouverensis to be the first record of parasitism in wild rock sole by this species. A former record of parasitism by this species was for aquarium fish only. Kabata, Z. 1998. Personal Commun. Department of Fish- eries and Oceans, Pacific Biological Station, Nanaimo, British Columbia, Canada, V9R 5K6. Kabata confirmed our tentative identification of Haemobaphes and identified it with fair cer- tainty as H diceraus. 374 Fishery Bulletin 99(2) \ Gulf of Alaska Comparative parasite prevalence B Gulf of Alaska Comparative parasite intensity Southern Northern rock sole rock sole Southern Northern rock sole rock sole Northern rock sole Comparative parasite prevalence 40 Aleutian Islands Gulf of Alaska Aleutian Islands Gulf of Alaska Figure 2 (A) Prevalence of Naobranchia occidentalis and Nectobrachia indivisa on northern and southern rock sole in the Gulf of Alaska. (B) Mean intensity of parasite infestation on northern and southern rock sole in the Gulf of Alaska. (C) Prevalence of Naobranchia occidentalis and Nectobrachia indivisa on northern rock sole in the Gulf of Alaska and Aleutian Islands. (D) Mean intensity of parasite infestation on northern rock sole in the Gulf of Alaska and Aleutian Islands. postreproduetive adult Naobranchia occidentalis was found in the Gulf of Alaska samples. The original at-sea identifica- tion of the two rock sole species was confirmed for all but one fish in the laboratory (a northern rock sole was incor- rectly identified at sea as a southern rock sole). Both Naobranchia occidentalis and Nectobrachia indivi- sa were more prevalent (chi square test, a=0.05, P<0.001) on the northern rock sole (22% and 15%' of the 225 fish examined, respectively) than on the southern rock sole (5% and 1% of the 282 fish examined, respectively, Fig. 2A). Mean intensity was not significantly different be- tween the northern (4.4/fish, range 1-22) and southern rock sole (2.9/fish, range 1-13) for Naobranchia occiden- talis (Welch’s approximate t-test, a=0.05, df=28, P=0.08, Fig. 2B) owing to a single southern rock sole with 13 parasites. Mean intensity also was not significantly differ- ent between the northern and the southern rock sole for Nectobrachia indivisa (Welch’s approximate f-test, a=0.05, df=2, P>0.05); however only two southern rock soles were infested with N. indivisa. A chi-square test showed that infestation of northern rock sole by N. indivisa was not independent of infestation by Naobranchia occidentalis (a=0.05, PcO.OOl). Thus, northern rock sole infested by one species of parasite had an increased likelihood of also being infested by the other. However, infestation of south- ern rock sole by Nectobrachia indivisa was independent of infestation by Naobranchia occidentalis (P> 0.05). Aleutian Islands Only northern rock soles were identified at sea in sampled hauls from the Aleutian Islands region, and laboratory analysis confirmed their species identifica- tion. A total of 237 northern rock soles from 8 trawl hauls was examined at sea for gill parasites: 90 fish appeared to have at least one gill parasite and their heads were frozen Zimmermann et al.: Parasitism by Naobranchia occidentalis 375 at sea (Table 2). None of the fish had abnormally pale gill filaments. Naobranchia occidentalis or Nectobrachia indi- visa (or both) were found on all but one of the northern rock soles examined in the laboratory. No other parasite spe- cies were observed. Approximately 17% of the Naobranchia occidentalis were juvenile females. Northern rock soles infested by one species of parasite had an increased likeli- hood of also being infested by the other species of parasite (chi square test, a=0.05, P<0.001), similar to our observa- tions for northern rock soles from the Gulf of Alaska. Area and species infestation comparisons Naobranchia occidentalis was more prevalent (chi square test, a=0.05, PcO.Ol) on the northern rock soles in the Aleutian Islands (36%) than on fish from the Gulf of Alaska (22%) (Fig. 2C). The northern rock soles from the Aleutian Islands also had a significantly higher mean intensity of N. occidentalis (Welch’s approximate f-test, a=0.05, df=127, P<0.001; 10.2/fish) than northern rock soles from the Gulf of Alaska (4.4/fish, Fig. 2D). There were no differences in either prevalence (chi-square test, a=0.05, P >0.05) or mean intensity (Welch’s approximate f-test, a=0.05, df=50, P>0.05) of Nectobrachia i?idivisa on the northern rock soles from the Aleutian Islands and Gulf of Alaska. Trends in parasite infestations— arches and sides Heads of fish from the 1996 Gulf of Alaska survey had been frozen for a year before examination and the gill fila- ments were in poor condition. As a result, only a small per- centage of Naobranchia occidentalis were still attached to a gill arch, and it was not possible to detect preferences of N. occidentalis for individual gill arches, or for eyed-side or blind-side arches, on northern rock sole (Table 3). Ninety- one percent of the observed Nectobrachia indivisa were still firmly anchored in the gill filaments on northern rock sole, were evenly distributed per side (60), and showed a prefer- ence for the outer arches over the inner arches on both sides of the fish (Table 3). All Naobranchia occidentalis on the southern rock sole from the Gulf of Alaska were unattached. Four Nectobrachia indivisa were found still attached on the blind side of southern rock sole from the Gulf of Alaska, but this small sample was insufficient for analysis. Aleutian Islands samples were frozen for only a few weeks prior to examination in the laboratory and were in much better condition than samples from the Gulf of Alas- ka survey. The infested gill arch was determined for 83%' of the Naobranchia occidentalis and for 97% of the Nec- tobrachia indivisa (Table 3). For northern rock sole from the Aleutian Islands, the total number of Naobranchia oc- cidentalis was higher on the blind side (452) than the eyed side (414) and equal on both sides for Nectobrachia indi- visa (33); these results nearly duplicated those observed for northern rock sole from the Gulf of Alaska. Naobran- chia occidentalis showed a strong preference for the mid- dle and inner gill arches on both eyed and blind sides (Ta- ble 3). The results were mixed for Nectobrachia indivisa ; parasites on the blind side showed a preference for the outer arches, whereas parasites on the eyed side showed 376 Fishery Bulletin 99(2) Figure 3 Length frequency of male and female northern rock sole examined for parasites from the Gulf of Alaska. Numbers of fish infested with one parasite species added to those infested with the other parasite species may exceed the total because several fish were infested with both parasite species. a preference for middle gill arches, and overall, there was a slight preference for outer arches (Table 3). Although there were numerous juvenile Naobranchia occidentalis parasites on northern rock soles from the Aleutian Is- lands, their individual positions were not recorded sepa- rately from the adults. Trends in parasite infestations— species, sex, and size Gulf of Alaska A wide size range of northern rock sole males (15-42 cm FL) and females (14-54 cm FL) were examined for parasites during the Gulf of Alaska survey (Fig. 3), but Naobranchia occidentalis was found only on larger males (28-40 cm FL) and females (29-51 cm FL). Similarly, Nectobrachia indivisa was found only on larger males (28-40 cm FL) and females (29-51 cm FL), with the exception of a single, small (21 cm FL) infested female. None of the male southern rock soles (19-40 cm FL) were infested with either parasite species, but male south- ern rock soles were uncommon in our survey catches and only 28 fish were examined (Fig. 4). A total of 254 female southern rock soles ( 14-58 cm FL) were examined for par- asites during the survey and Naobranchia occidentalis in- fested only 14 of the larger females (42-52 cm FL). Only two relatively large (37 and 47 cm FL) female southern rock soles were infested with Nectobrachia indivisa. Aleutian Islands Male (19-37 cm FL) and female (20-46 cm FL) northern rock soles from the Aleutian Islands were examined, and Naobranchia occidentalis was found only in larger individuals (24-37 cm FL males and 28-44 cm FL females, Fig. 5). The size range of fish infested with Nectobrachia indivisa was slightly more restricted (24-33 cm FL males and 28-43 cm FL females) than the size range of fish infested with Naobranchia occidentalis. Length-weight relationships Parasitized (range 28-40 cm FL) and unparasitized (range 18-36 cm FL) male northern rock soles from the Gulf of Zimmermann et al.: Parasitism by Naobranchia occidentalis 377 C 0.05) but had significantly different intercepts (df=l, F-ratio=13.74, PcO.GOl). For example, a 33-cm-FL parasitized male northern rock sole from the Gulf of Alaska had an estimated weight of 407.5 grams, whereas an unparasitized fish of the same length weighed an estimated 451.4 grams, a difference of more than 10%. The length-weight slopes and intercepts were not significantly different for female northern or southern rock soles from the Gulf of Alaska. No comparison could be made between parasitized and unparasitized Aleutian Islands fish because individual weights were not taken for unparasitized fish. Discussion Laboratory examination confirmed our original observa- tions that copepod parasites were more prevalent on the northern rock soles from the Gulf of Alaska than on south- ern rock soles. There could also be a higher mean inten- sity for both species of parasites on the northern rock sole, but our analysis was limited by a small sample size and high variability within the sample. Of the two gill par- asites examined in detail, Naobranchia occidentalis was more prevalent and caused more intense infestations than did Nectobrachia indivisa. Our sampling efforts in the Aleutian Islands occurred beyond the geographic range of southern rock sole and indicated that the northern rock sole had a higher prevalence and greater mean intensity of Naobranchia occidentalis than the northern rock sole in the Gulf of Alaska. Differences in parasite prevalence and mean intensity between the Gulf of Alaska and Aleutian Island samples were confounded by sampling in different years, limited areal coverage, limited temperature range, and greater depths in the Aleutian samples. Also, the sig- nificance of the large number of juvenile N. occidentalis in the Aleutian Island samples is not known. Despite these limitations, it is noteworthy that none of the northern rock soles heavily infested with Naobranchia occidentalis 378 Fishery Bulletin 99(2) 12 10 8 6 4 3 2 0 I I Total fish examined I | Fish with Naobranchia occidentalis B Fish with Nectobrachia indivisa 10 15 20 25 I I 30 35 40 Male “I 1 1 1 45 50 Figure 5 Length frequency of male and female northern rock sole examined for parasites from the Aleutian Islands. Numbers of fish infested with one parasite species added to those infested with the other parasite species may exceed the total because several fish were infested with both parasite species. from the Aleutian Island samples had white gill filaments. Because of this observation and the apparent preference of Nectobrachia indivisa for northern rock sole, we specu- late that the white gill filaments in the northern rock soles from the Gulf of Alaska survey may have been caused by heavy infestation with N. indivisa. The decreased weight at length of parasitized male northern rock soles from the Gulf of Alaska indicates a possible effect of infestation. It is interesting to note that northern rock soles from both areas, which were infested with one parasite, had an increased chance of also being infested by the other parasite. Either some northern rock soles are more susceptible to infestation by both para- sites, or infestation by one parasite increases the likeli- hood of infestation by the other species. Because infesta- tion of southern rock soles by Nectobrachia indivisa was independent of infestation by Naobranchia occidentalis, it is possible that southern rock sole may be resistant to infestation by both parasite species. Apparently Naobraiichia occidentalis infests smaller northern than southern rock sole; thus, much of the south- ern rock sole population may be unavailable to successful infestation by this parasite. This size-dependent preva- lence may be related to the grasping method of attachment of N. occidentalis ; thus, its ability to infest rock sole may be related to the diameter of the gill filaments. Cressey et al. (1983) proposed a similar mechanical limitation for fe- male pseudocycnids that grasp the gill filament in scom- brids by partially encircling it with the lateral lobes of their cephalon. They suggested that hosts must reach a mini- mum size before the gill filament is large enough for the parasite to achieve a firm grip. Roubal and Graham (1999) noted that the smallest fish in their study were not infest- ed with Naobranchia variabilis , which attaches to gill fila- ments in a similar manner as N. occidentalis, and suggested it was because those fish simply had not been encoun- tered by a parasite. They did not suggest any sort of me- chanical limitations to successful infestation because the Zimmermann et ai.: Parasitism by Naobranchia occidentalis 379 parasite larvae are apparently sufficiently small (Roubal3). Both Cressey et al. (1983) and Roubal and Graham (1999) noted that the largest adult parasites were found on the largest fish — a finding that suggests that the gill filaments of smaller fish may be less suitable for infestation. Nectobrachia indivisa also selectively infested larger northern rock sole but rarely infested southern rock sole, suggesting an apparent species preference. Because N. in- divisa attaches by means of a bulla permanently embed- ded near the end of a gill filament, it seems less likely that this parasite would be limited by mechanical restrictions such as the diameter of the gill filament. The parasite Sal- mincola californiensis, which also attaches to its host by means of a bulla (Kabata and Cousens, 1972), seemed to be dependent on its sockeye salmon (Oncorhynchus nerka) host reaching a large size before the gill filaments became the preferred site of attachment (Kabata and Cousens, 1977). Perhaps N. indivisa is able to infest northern rock sole successfully only after the fish reach a certain size. We did not perform histological analysis of gill filaments to determine differential damage caused by infestation of Naobranchia occidentalis and Nectobrachia indivisa. Rou- bal (1999) determined that damage caused by Naobran- chia variahilis was restricted to the infested gill filament and was minor. Kabata (1984) noted that naobranchids only partially compress the gill filament, do not completely restrict blood flow, and do not cause whitening of gill fila- ments. Roubal (1999) also determined that N. variahilis live by feeding on the blood supply and not by grazing the tissue, as do lernaeopodids. Kabata and Cousens (1977) reported macroscopic observations of significant atrophy and tissue reaction to as much as one-third of the gill fila- ment surface area of juvenile sockeye salmon due to the presence of S. californiensis. This type of damage is consis- tent with our macroscopic observations of the white gills of northern rock soles seen early in the Gulf of Alaska sur- vey. Kabata and Cousens (1977) also reported on micro- scopic observations of gill filament damage, sometimes in- cluding proliferation of the gill epithelium, hypertrophy of epithelial cells, fusion of adjacent filaments, thickening of filament walls, blood blisters, and erosion of epithelial cells as the oral apparatus of the S. californiensis scrapes them for ingestion. Margolis and Arthur ( 1979), Kabata and Whitaker ( 1984), and Kabata (1988) all reported that both Naobranchia occidentalis and Nectobrachia indivisa occurred on rock soles in Pacific Canadian waters, prior to the determina- tion that there are two species of rock sole in this area (Orr and Matarese, 2000). Neither Moles (1982) nor Love and Moser (1983) reported these parasites from rock soles from Alaskan waters and U.S. west coast waters, respec- tively. Our study reported new host records for Acantho- chondria vancouverensis and Haemobaphes diceraus on northern rock sole. The apparent preference of both parasites for the north- ern rock soles is supported by the conclusion of Orr and 3 Roubal, F. R. 1999. Personal commun. Department of Para- sitology, The University of Queensland, Brisbane 4072, Queens- land, Australia. Matarese (2000) that northern and southern rock soles are two distinct species. Possible differences in the ecology of these newly described rock sole species, such as food hab- its, growth rates, habitats, spawning seasons and locations, nursery grounds, and seasonal and ontogenetic migrations might also account for the differential parasitism that we observed. It is not known if parasite prevalence and inten- sity is related to differences in the ecology and behavior of these closely related species, or related to differences in their anatomy or physiology. As more research is done on these species, based on the work of Orr and Matarese (2000), more potential differences will be determined. Acknowledgments The laboratory research of this project was performed by A. Jones, as an internship in partial fulfillment of a Bach- elor of Science degree from Western Washington Univer- sity. Steve Syrjala provided statistical advice. Jay Orr taught A. Jones the subtle differences between the rock sole species and how to identify the structures of inter- est. Discussions with Frank Morado improved our labora- tory methods and manuscript. Comments and suggestions from several coworkers greatly improved the quality of this manuscript and an anonymous reviewer provided helpful criticism of the manuscript. We thank Frank Roubal for sharing additional results from his studies and for hypothesizing on the infection dynamics of Nao- branchia. Teresa Turk, Steven Hochberg, Michelle Arm- strong, Dick Haight, and Ben Page examined fish out at sea for parasites. We are extremely grateful to Z. Kabata for examining samples of all four copepod parasites and confirming our species identifications. Literature cited Bush, A. O., K. D. Lafferty, J. M. Lotz, and A. W. Shostak. 1997. Parasitology meets ecology on its own terms: Margolis et al. revisited. J. Parasitol. 83(4)575-583. Cressey, R. F., B. B. Collette, and J. L. Russo. 1983. Copepods and scombrid fishes: a study in host-para- site relationships. Fish. Bull. 81(2)227-265. Hart, J. L. 1973. Pacific fishes of Canada. Fish. Res. Board Can. Bull. 180, 740 p. Kabata, Z. 1984. Diseases caused by metazoans: crustaceans. In Dis- eases of marine animals. Vol IV, part 1: Introduction: Pisces (Kinne, ed.), p. 321-399. Biologische Anstalt Helgoland, Hamburg. 1988. Copepoda and Branchiura. In Guide to the parasites of fishes of Canada. Part II: Crustacea (L. Margolis and Z. Kabata, eds. ), p. 3-127. Can. Spec. Publ. Fish. Aquat. Sci. 101. Kabata, Z., and B. Cousens. 1972. The structure of the attachment organ of Lernaeopodi- dae (Crustacea: Copepoda). J. Fish. Res. Board Canada 29: 1015-1023. 1977. Host-parasite relationships between sockeye salmon, Oncorhynchus nerka, and Salmincola californiensis (Co- 380 Fishery Bulletin 99(2) pepoda: Lernaeopodidae). J. Fish. Res. Board Can. 34:191— 202. Kabata, Z., and D. J. Whitaker. 1984. Results of three investigations of the parasite fauna of several marine fishes of British Columbia. Fish. Res. Board Can., Tech. Rep. 1303, 19 p. Love, M. S., and M. Moser. 1983. A checklist of parasites of California, Oregon, and Washington marine and estuarine fishes. U.S. Dep. Commer., NOAA Tech. Rep. NMFS SSRF-777, 576 p. Margolis, L., and .1 R. Arthur. 1979. Synopsis of the parasites of fishes of Canada. Fish. Res. Board Can. Bull. 199, 269 p. Margolis, L , G. W. Esch, J. C. Holmes, A. M. Kuris, and G. A. Schad. 1982. The use of ecological terms in parasitology (report of an ad hoc committee of the American Society of Parasitologists). .J. Parasitol., 68t 1 ): 131-133. Martin, M H. 1997. Data report: 1996 Gulf of Alaska bottom trawl survey. U.S. Dep. Commer., NOAA Tech. Memo. NMFS-AFSC-82, 235 p. Matarese, A. C., A. W. Kendall -Jr., D. M. Blood, and B. M. Vinter. 1989. Laboratory guide to early life history stages of North- east Pacific fishes. U.S. Dep. Commer., NOAA Tech. Rep. NMFS 80, 651 p. Moles, A. 1982. Parasite-host records of Alaskan fishes. U.S. Dep. Commer., NOAA Tech. Rep. NMFS SSRF-760, 41 p. Orr, J. W., and A. C. Matarese. 2000. Revision of the genus Lepidopsetta Gill, 1862 (Tele- ostei: Pleuronectidae) based on larval and adult morphol- ogy, with a description of a new species from the North Pacific Ocean and Bering Sea. Fish. Bull. 98:539-582. Roubal, F. R. 1999. Extent of gill pathology in the toadfish Tetractenos hamiltoni caused by Naobranchia variabilis (Copepoda: Naobranchiidae). Dis. Aquat. Org. 35:203-211. Roubal, F. R., and D. Graham. 1999. Monthly variation in recruitment, infection, size, fecundity and mating of Naobranchia variabilis (Copepoda: Naobranchiidae) parasitic on the gills of toadfish Tetracte- nos hamiltoni from Moreton Bay, Australia. Mar. Fresh- water Res. 50:291-298. Zar, J. H. 1984. Biostatistical analysis, second ed. Prentice-Hall Inc., Englewood Cliffs, NJ, 718 p. 381 Size distribution of southern bluefin tuna (Thun nus maccoyii ) by depth on their spawning ground Tim L. O. Davis Jessica H. Farley CSIRO Division of Marine Research PO Box 1538, Hobart Tasmania 7001, Australia E-mail address (for T L. O Davis): tim.davis@manne.csiro.au Indonesian and Japanese longline ves- sels catch different-size southern blue- fin tuna ( Thunnus maccoyii) on their spawning ground in the Indian Ocean south of Bah. The length distributions of southern bluefin tuna (SBT) caught by Japanese longline are markedly smaller than those caught by the Indo- nesians (Davis et al. 1 ; Itoh2). Both measurement error and misidentifi- cation of small SBT as bigeye tuna (Thunnus obesus ) in the Indonesian catch data have been suggested as causes of this discrepancy (Suzuki and Nishida3), but neither has been substantiated. Japanese vessels target bigeye tuna by using deep longline sets (Itoh2), whereas most Indonesian ves- sels target yellowfin tuna (Thunnus a! bacares) by using shallow longline sets (Davis et ah, 1995). The differ- ence in types of longline sets raises the possibility that SBT on the spawn- ing ground are segregated by size with depth. Three types of boats operate in the Indonesian fishery (Davis et ah, 1995). Deep longline boats (generally >50 tonnes) use multifilament mainlines that are set deep. Mini (<20 tonnes gross weight) and regular longline boats (20-50 tonnes) use monofilament mainlines and generally make shal- low longline sets. However, the depth at which the lines fish varies consid- erably because they carry live or fro- zen baits according to different phases of the moon, and both the number of hooks and their placement on the cat- enary between floats changes. Predic- tion of fishing depth based on catenary geometry, line length, and distance be- tween floats (Yoshihara, 1954) differs significantly from actual depth fished (Saito, 1973; Nishi, 1990; Boggs, 1992). In this fishery, the number of hooks be- tween floats is recorded (Davis et ah, 1999), but this parameter alone is a poor indicator of the depth of fishing. Using hook timers, Boggs (1992) de- termined depth at the time of hook- ing. He found that bigeye catch rates peaked at 360-400 m and 8-10°C (temperature), but were still high at 200-360 m. Bigeye tuna have a shal- lower distribution at night (modal depth of 80 m) than during the day (220 m) (Holland et ah, 1990). However, on the SBT spawning ground, longline setting starts at about 06:00 h and hauling starts at about 14:00 h (Davis4); therefore most bigeye tuna would be caught during the day when they are deeper. The preferred depths of bigeye tuna vary regionally depending on thermo- cline structure, but lie within 10° and 15°C (Hanamoto, 1986; Mohn et ah, 1996) and where 0., > 1 mL/L (Hana- moto, 1986). These temperatures occur at 180-400 m on the SBT spawning ground (Yukinawa and Miyabe, 1984; Yukinawa and Koido, 1985; Yukinawa, 1987). Yellowfin tuna, on the other hand, are found in warmer waters and are mainly caught at depths of 40-230 m (Suzuki and Kume, 1982; Yang and Gong, 1988; Boggs, 1992). The propor- tion of bigeye to yellowfin tuna might therefore be used as a proxy for the depth of fishing in the Indonesian long- line fishery. In our study we used this depth proxy to investigate whether there is size partitioning by depth of SBT on the spawning ground, and what underlying biological processes might be involved. Methods We used catch data obtained from 15,882 Indonesian longline landings monitored at export processing facto- ries at the Port of Benoa, Bali, from 1992 to 1999 (Davis et ah, 1995; 1999). About 65% of the SBT in these land- ings were measured (fork length in cm). Fewer high-grade export tuna (30%) were measured than low-grade tuna (89%) because the former were immersed in an ice slurry immediately after grading, leaving little opportu- nity for measurement. Grading, how- ever, was not dependent on size. There was no significant difference in the length distributions of 102 export tuna and 102 low-grade tuna from 20 land- ings in which all tuna were measured (Kolmogorov-Smirnov two sample test, P= 0.22). 1 Davis, T. L. O., J. H. Farley, and S. Bahar. 1996. Catch monitoring of the fresh tuna caught by the Bali-based longline fishery. Commission for the Conservation of South- ern Bluefin Tuna scientific meeting, 26 August-5 September 1996, Hobart, Aus- tralia, Rep. CCSBT/SC/96/6, 26 p. CSIRO Marine Laboratories, PO Box 1538, Hobart, Tasmania 7001, Australia. - Itoh, T. 1997. Longline survey in south- ern bluefin tuna spawning ground. Com- mission for the Conservation of Southern Bluefin Tuna scientific meeting, 28 July-8 August 1997, Canberra, Australia, Rep. CCSBT/SC/97/12, 4 p. CSIRO Marine Laboratories, PO Box 1538, Hobart, Tas- mania 7001, Australia. ;i Suzuki, Z., and T. Nishida. 1997. Com- parison of information on the catch and size of fish in the spawning ground of southern bluefin obtained from Indone- sian and Japanese longline fisheries. Com- mission for the Conservation of Southern Bluefin Tuna scientific meeting, 28 July-8 August 1997, Canberra, Australia, Rep. CCSBT/SC/97/13, 8 p. CSIRO Marine Laboratories, PO Box 1538, Hobart, Tas- mania 7001, Australia. 4 Davis, T. L. O. 1999. Unpubl. data. CSIRO Marine Laboratories, PO Box 1538. Hobart, Tas 7001, Australia. Manuscript accepted 12 September 2000. Fish. Bull. 99:381-386 (2001). 382 Fishery Bulletin 99(2) Distribution square=516, Table 1 (%) of length groups (10-cm intervals) of southern bluefin tuna across bigeye tuna »=8416, df=24, <0.001). (BE) indices (Pearson chi- Length (cm) BE indices Total no. Total 0.0-0. 2 0.2-0. 4 0. 4-0.6 0.6-0. 8 O cq o 140-149 13.3 6.7 13.3 26.7 40.0 15 100.0 150-159 2.8 10.1 17.0 18.6 51.4 247 100.0 160-169 8.7 15.4 19.5 24.5 31.8 1019 100.0 170-179 12.7 25.7 20.8 20.8 20.0 2442 100.0 180-189 17.9 26.6 18.5 19.3 17.8 3520 100.0 190-199 27.0 23.7 18.9 14.7 15.7 990 100.0 200-209 35.9 21.8 19.6 10.9 12.0 184 100.0 No. of landings 2100 3585 3876 4421 1900 For each landing we calculated a bigeye (BE) tuna index as BE index = Weight of bigeye / (weight of bigeye + weight of yellow fin) . This equation was used as a proxy for the depth of fish- ing, with an index of 1 = deep and 0 = shallow. Landings were grouped into one of five levels of this index, i.e. 0-0.2, 0.2-0. 4, etc., and then the length-frequency distributions of SBT within landings at each level were compared. In order to investigate patterns of distribution of fish size with depth, we grouped fish into 10-cm length class- es and calculated their relative abundance across the five levels of the BE index. Because of uneven sampling with depth, the number of fish in each BE index were first weighted inversely by the effort (number of landings) at each level of the index. The ovaries of 475 SBT were collected during monitor- ing from 1992 to 1995. These were examined histologically for evidence of recent or imminent spawning (Farley and Davis, 1998). Spawning fish were classed as those having spawned less than 24 hours previously (postovulatory fol- licles present in ovary), or about to spawn that day (ova- ries containing oocytes at migratory nucleus or hydrated stage). Postspawning SBT were identified by the propor- tion and type of atretic oocytes present (see details in Far- ley and Davis, 1998). Nonspawning SBT were mature fish on the spawning ground that were neither spawning nor postspawning individuals. Chi-square contingency analyses were used to test for differences in length classes of SBT, and for differences in the proportion of spawning and nonspawning SBT at dif- ferent levels of the BE index (the proxy for depth). Results The length-frequency distribution of SBT caught at five levels of the BE index shows a trend of increased propor- tions of small SBT with an increase in this index (Fig. 1 ). Fish <165 cm ranged from 3.3% of catch at an index <0.2 to 15.7% at a index >0.8. Chi-square contingency analyses indicated significant differences in the proportion of length classes with the BE index (Table 1, Fig. 2). The chi-square test ignores the ordered and continuous nature of the categories, making it less powerful than it could be. However, we obtained a highly significant test result despite this weakness, re- flecting how strong the size-with-depth patterns are. The smaller length classes (150-169 cm) were better repre- sented in the deep catches (BE index >0.8) than they were in the shallow catches (BE index <0.2). Conversely, the larger length classes (190-209 cm) were better represent- ed in the shallow catches (BE index <0.2) than they were in the deep catches (BE index >0.8). Smaller fish were more likely to be caught in the deepest sets, which target bigeye, whereas the bigger fish were more likely to be caught in the shallow sets. Significantly, there is a system- atic change in depth distribution with size over the whole size range of SBT that occur on the spawning ground. This pattern is very clear when comparing the proportion of fish caught in shallow (BE index of 0.0-0. 2 or 0.0-0. 4) ver- sus deep (BE index of 0. 8-1.0 or 0. 6-1.0) sets for each length class. The proportion of SBT caught at the surface increases with size (Fig. 3). The proportion of spawning and nonspawning fish (based on the subset of histological data) was then deter- mined for each level of BE index (Fig. 4). Chi-square con- tingency analyses indicated significant differences in the proportions (Table 2). Spawning fish were better repre- sented in the shallow catches than in the deep catches. Conversely, nonspawning fish were better represented in the deep catches than in the shallow catches. There were insufficient numbers of SBT in the smaller size classes (only seven SBT<160 cm) to use the histology data to ex- amine directly the relation between size and proportion of spawning fish or spawning frequencies. Because spent fish were rarely encountered on the spawning ground, Farley and Davis ( 1998) concluded that they move south soon af- NOTE Davis and Farley: Size distribution of Thunnus maccoyii on their spawning ground 383 Fork length (cm) Figure 1 Length-frequency distribution (2-cm intervals) of southern bluefin tuna in landings by bigeye tuna index — a proxy for fishing depth. ter spawning. However, the two spent fish detected were in landings with a BE index >0.9. Discussion There is a systematic change in depth distribution with size over the whole size range of SBT caught on the spawn- ing ground. This pattern is clear, even though the BE index may only represent a crude approximation of depth. Deep longline catches are often contaminated by surface catches — 10% of bigeye tuna are caught when hooks are not at settled depths (Boggs, 1992). Also, both SBT (Gunn et ah ’; Davis and Stanley6) and bigeye tuna (Holland et ah, 1990) might be caught outside their preferred depth as they regularly traverse the water column. 5 Gunn, J. S.. T. Polacheck. T. L. Davis, M. Sherlock, and A. Betlehem. 1994. The application of archival tags to study the movement, behaviour and physiology of southern bluefin tuna, with comments on the transfer of the technology to groundfish research. ICES CM 1994/Mini: 21, 23 p. (Mimeo.] 6 Davis, T. L. O., and C. A. Stanle. 2001 In prep. Vertical and horizontal movements of southern bluefin tuna, Thunnus maccoyii , in the Great Australian Bight observed bv ultrasonic telemetry. 384 Fishery Bulletin 99(2) The pattern of size distribution with depth is mirrored by the pattern of spawning and nonspawning with depth. Both smaller and nonspawning SBT are more abundant at depth, whereas both larger and spawning SBT are more abundant near the surface. The vertical distribution of SBT larvae suggests that SBT spawn at the surface (Da- vis et ah, 1990), as do caged Atlantic bluefin tuna (Th un- nus thynnus ) (Fushimi et ah, 1998). Surface-water tem- peratures on the spawning ground usually exceed 24°C (Yukinawa and Miyabe, 1984; Yukinawa and Koido, 1985; Yukinawa, 1987). These warm surface waters may be nec- essary for the survival of their eggs and larvae, but adult SBT normally feed in colder water (often as low as 5°C [Ol- son, 1980] ). Temperatures of 10°-15°C preferred by bigeye tuna (Hanamoto, 1986; Mohri et ah, 1996) may offer more favorable conditions for nonspawning SBT and explain their strong association with high BE indices on the spawning ground. Previous studies have shown that yellowfin tuna caught by purse seine and handline have higher go- nadosomatic indices than yellowfin caught by long- line (Hisada, 1973; Suzuki, 1988; Koido and Suzuki, 1989). Histological studies have found that yellow- fin tuna catches from purse-seine sets and shallow (Taiwanese-style) longline sets have a higher pro- portion of actively spawning fish than catches from deep (Japanese-style) longline sets (Itano7). Thus, spawning fish are more likely to be caught near the surface and nonspawning fish are more likely to be caught in deeper water. The biological basis for size partitioning with depth could be that large fish spawn more frequent- ly than small fish and, therefore, bigger fish will be caught at the surface more often than smaller ones. Spawning frequency is known to increase with size in female yellowfin tuna (Schaefer, 1998) but could not be determined for SBT. The pattern of size distribution may reflect recruitment into spawning. However, this hypothesis is unlikely because histo- logical examination of ovaries indicated that all SBT caught on the spawning ground were mature i.e. had advanced yolked oocytes (Farley and Davis, 1998), although this does not preclude the possibility that they might not be ready to spawn. The most likely reason for size partitioning is that the spawning fre- quency or the proportion of time spent spawning to time spent in a nonspawning condition increases with size. If the ability to tolerate higher than preferred wa- ter temperatures improved with fish size, then this would facilitate longer spawning episodes or more extensive feeding in shallow waters, both of which would produce the observed pattern of size distri- bution with depth. Although the ability to conserve heat in cold waters may increase with size in SBT, it is not clear what size-dependent processes might be involved in avoiding overheating at high ambient temperatures. We do not understand the temporal and spatial scale of vertical movements of SBT on the spawn- ing grounds in relation to spawning and feeding, nor Itano, D. G. 2000. The reproductive biology of yellow- fin tuna ( Thunnus albacares ) in Hawaiian waters and the western tropical Pacific Ocean: project sum- mary. SOEST (School of Ocean and Earth Science and Technology) 00-01, JIMAR (Joint Institute for Marine and Atmospheric Research) Contribution 00-328, 69 p. Univ. Hawaii, 1000 Pope Road, MSB 312, Honolulu, HI 96822-2336, US. NOTE Davis and Farley: Size distribution of Thunnus maccoyn on their spawning ground 385 Table 2 Percentage of spawning and nonspawning southern bluefin tuna caught at different bigeye indices (Pearson chi-square=24.1, «=326, df=4, PcO.OOl). BE index 0.0-0. 2 0. 2-0.4 0.4-0. 6 0.6-0. 8 © bo o Total no. Spawning 85.5 71.4 80.8 56.4 56.3 227 Nonspawning 14.5 28.6 19.2 43.6 43.7 99 how these might change with fish size. This behav- ioral information is needed in order to interpret the patterns presented in our study and might best be achieved by pop-up satellite archival tagging. Because SBT aggregate by size and depth on the spawning ground, it is necessary to account for their distribution when determining the age and size structure of the spawning stock. This is espe- cially important when evaluating time series of size and age distributions in a fishery where there have been shifts in targeting between yellowfin and big- eye tuna. In the absence of reliable information on the depth of fishing, the most practical way of doing this in the Indonesian fishery would be to inversely weight the effort directed at the different levels of the BE index. The determination of spawning fre- quency should also take into account longline fishing strategies because it is likely that spawning frequen- cy is affected by fish size and because samples will be caught within or outside the spawning depth. If the increase in the proportion of SBT at the sur- face with size is due to spawning activity, then this feature will affect the contribution different size fish make to total annual egg production. A lower spawn- ing frequency, coupled with an exponential relation- ship between length and batch fecundity ( Farley and Davis 1998), would mean that individual small, but mature, fish make a relatively small contribution to total annual egg production. When making stock projections, it may therefore be more appropriate to adopt a parameter that reflects size at mean annual egg production rather than the currently accepted parameter of mean size at first maturity. Further histological research on the reproductive dynamics of small fish is required to better define these pa- rameters. Small fish were rarely caught when the histological work of Farley and Davis (1998) was carried out in 1992-95 but they have become more abundant in recent years ( Davis et al.8) making such a study possible. H Davis, T. L. O., S. Bahar, N. Naamin, and J. H. Farley- 1998. Catch monitoring of the fresh tuna caught by the Bali-based longline fishery. Commission for the Conser- vation of Southern Bluefin Tuna scientific meeting, 23—31 July 1998, Shimizu, Japan, Rep. CCSBT/SC/9807/6, 17 p. CSIRO Marine Laboratories, PO Box 1538, Hobart, Tas- mania 7001 . Australia. 100-j 8(H 60 — j 40-1 t v? ere 20- C o 0^ r o u_ o 100-1 CL 80- 60- 40- 20- Shallow ft Spawning Deep Nonspawning i i ; i i 0.0-0. 2 0. 2-0.4 0.4-0. 6 0.6-0. 8 0.8-10 Bigeye index Figure 4 Proportion of spawners and nonspawners in landings by the big- eye tuna index. 386 Fishery Bulletin 99(2) Acknowledgments We thank the managers at PT. Perikanan Samodra Besar, PT. Sari Segara Utama, and PT. Bandar Nelayan for facilitating catch sampling at their processing plants in Benoa. We are grateful to Waluyo Suharto, Kiroan Siregar, Mashar Machmud, and Labuhan Siregar for monitoring catches at the various plants; Sofri Bahar at the Research Institute of Marine Fisheries, Indonesia, for coordinating the monitoring program; and Duyet Le for laboratory assistance. We thank Kurt Schaefer, Bill Hearn, and John Stevens for their reviews of the manuscript and Vivienne Mawson for editing. This research was supported by Fish- eries Resources Research Fund Grants from the Austra- lian Fisheries Management Authority. Literature cited Bailey, B. J. R. 1980. Large sample simultaneous confidence intervals for multinomial probabilities based on transformations of the cell frequencies. Technometrics 22:583-589. Boggs, C. H. 1992. Depth, capture time, and hooked longevity of long- line-caught pelagic fish: timing bites of fish with chips. Fish. Bull. 90:642-658. Davis, T. L. 0., S. Bahar, and J. H. Farley. 1995. Southern bluefin tuna in the Indonesian longline fish- ery: historical development, composition, season, some bio- logical parameters, landing estimation and catch statistics for 1993. Indonesian Fish. Res. J. l(l):68-86. Davis, T. L. O., S. Bahar, N., Naamin, J. H. Farley, and D. Le. 1999. Trends in tuna and billfish landings by the longline fishery operating out of Benoa, Bali. Indonesian Fish. Res. J. 5 (l):50-60. Davis, T. L. O., G. R Jenkins, and J. W. Young. 1990. Diel patterns of vertical distribution in larvae of southern bluefin Thunnus maccoyii , and other tuna in the East Indian Ocean. Mar. Ecol. Prog. Ser. 59:63-74. Farley, J. H., and T. L. O. Davis. 1998. Reproductive dynamics of southern bluefin tuna, Thunnus maccoyii. Fish. Bull. 96:223-236. Fushimi, H., K. Kani, H. Nhhala, S. Nakamura, A. Abrouch, K. Chebaki, and A. Berraho. 1998. Attempts at resource enhancement of Atlantic bluefin tuna: present status on, and future prospects for, a Japa- nese-Moroccan co-operative project for aquaculture of Atlan- tic bluefin tuna. ICCAT collective volume of scientific papers, vol.L(2), p. 479-489. International Commission for the Conservation of Atlantic Tunas, Madrid, Spain. Hanamoto, E. 1986. Distribution of bigeye catch in the Pacific Ocean. Bull. Jpn. Soc. Fish. Oceanogr. 51:9-15. Hisada, K. 1973. Investigations on tuna hand-line fishing ground and some biological observations on yellowfin and bigeye tunas caught in the northwestern Coral Sea. Bull Far Seas Fish. Lab. 8: 35-69. Holland, K. N„ R. W. Brill, and R. K. C. Chang. 1990. Horizontal and vertical movements of yellowfin and bigeye tuna associated with fish aggregating devices. Fish. Bull. 88:493-507. Koido, T., and Z. Suzuki. 1989. Main spawning season of yellowfin tuna, Thunnus albacares , in the western tropical Pacific Ocean, based on gonad index. Bull Far Seas Fish. Lab. 26:153-163. Mohri, M., E. Hanamoto, and S. Takeuchi. 1996. Optimum water temperatures for bigeye tuna in the Indian Ocean as seen from tuna longline catches. Nippon Suisan Gakkaishi 62 (5):761-764. Nishi, T. 1990. The hourly variations of the depth of hooks and the hooking depth of yellowfin tuna ( Thunnus albacares), and bigeye tuna ( Thunnus obesus), of tuna longline in the region of the Indian Ocean. Mem. Fac. Fish. Kagoshima Univ. 39:81-98. Olson, R. J. 1980. Synopsis of biological data on the southern bluefin tuna, Thunnus maccoyii (Castlenau 1872). Inter-Am. Trop. Tuna Comm. Special Rep. 2:151-212. Saito, S. 1973. Studies on fishing of albacore, Thunnus alalunga (Bonaterre) by experimantal deep-sea tuna long-line. Mem. Fac. Fish. Hokkaido Univ. 21:107-185. Schaefer, K. M. 1998. Reproductive biology of yellowfin tuna (Thunnus alb- acares) in the eastern tropical Pacific Ocean. Inter-Am. Trop. Tuna Comm. Bull. 21 (5):205-272. Suzuki, Z. 1988. Study of interaction between long-line and purse seine fisheries on yellowfin tuna, Thunnus albacares, (Bonnat.). Bull. Far Seas Fish. Res. Lab. 25:73-139. Suzuki, Z., and S. Kume. 1982. Fishing efficiency of deep longline for bigeye tuna in the Atlantic as inferred from the operations in the Pacific and Indian Oceans. Int. Comm. Conserv. Atl. Tunas Col- lect. Vol. Sci. Pap. 17:471-486. Yang, W. S., and Y. Gong. 1988. The vertical distribution of tunas and billfishes, and fishing efficiency between Korean regular and deep long- lines in the Atlantic Ocean. Bull. Natl. Fish. Res. Dev. Agency [Pusan, Korea) 42:39-42. Yoshihara, T. 1954. Distribution of catch of tuna longline. IV: On the rela- tion between k and