U.S. Department of Commerce Volume 103 Number 1 January 2005 Fishery Bulletin U.S. Department of Commerce Donald L Evans Secretary National Oceanic and Atmospheric Administration Vice Admiral Conrad C. Lautenbacher Jr., USN (ret.) Under Secretary for Oceans and Atmosphere National Marine Fisheries Service William T. Hogarth Assistant Administrator for Fisheries ^nT0Fc% •^TES 0* *" The Fishery Bulletin (ISSN 0090-0656) is published quarterly by the Scientific Publications Office, National Marine Fish- eries Service, NOAA, 7600 Sand Point Way NE, BIN C15700, Seattle, WA98115-0070. Periodicals postage is paid at Seattle, WA, and at additional mailing offices. POST- MASTER: Send address changes for sub- scriptions to Fishery Bulletin, Superin- tendent of Documents, Attn.: Chief. Mail List Branch, Mail Stop SSOM, Washing- ton, DC 20402-9373. Although the contents of this publica- tion have not been copyrighted and may be reprinted entirely, reference to source is appreciated. The Secretary of Commerce has deter- mined that the publication of this peri- odical is necessary according to law for the transaction of public business of this Department. Use of funds for printing of this periodical has been approved by the Director of the Office of Management and Budget. For sale by the Superintendent of Documents, U.S. Government Printing Office, Washington, DC 20402. Subscrip- tion price per year: $55.00 domestic and $68.75 foreign. Cost per single issue: $28.00 domestic and $35.00 foreign. See back for order form. Scientific Editor Norman Bartoo, PhD Associate Editor Sarah Shoffler National Marine Fisheries Service, NOAA 8604 La Jolla Shores Drive La Jolla, California 92037 Managing Editor Sharyn Matriotti National Marine Fisheries Service Scientific Publications Office 7600 Sand Point Way NE, BIN C15700 Seattle, Washington 98115-0070 Editorial Committee Harlyn O. Halvorson, PhD Ronald W. Hardy, PhD Richard D. Methot, PhD Theodore W. Pietsch, PhD Joseph E. Powers, PhD Harald Rosenthal, PhD Fredric M. Serchuk, PhD George Watters, PhD University of Massachusetts, Boston University of Idaho, Hagerman National Marine Fisheries Service University of Washington, Seattle National Marine Fisheries Service Universitat Kiel, Germany National Marine Fisheries Service National Marine Fisheries Service Fishery Bulletin web site: www.fishbull.noaa.gov The Fishery Bulletin carries original research reports and technical notes on investigations in fishery science, engineering, and economics. It began as the Bulletin of the United States Fish Commission in 1881; it became the Bulletin of the Bureau of Fisheries in 1904 and the Fishery Bulletin of the Fish and Wildlife Service in 1941. Separates were issued as documents through volume 46: the last document was No. 1103. Beginning with volume 47 in 1931 and continuing through volume 62 in 1963, each separate appeared as a numbered bulletin. A new system began in 1963 with volume 63 in which papers are bound together in a single issue of the bulletin. Beginning with volume 70, number 1, January 1972, the Fishery Bulletin became a periodical, issued quarterly. In this form, it is available by subscription from the Superintendent of Documents, U.S. Government Printing Office, Washington, DC 20402. It is also available free in limited numbers to libraries, research institutions, State and Federal agencies, and in exchange for other scientific publications. U.S. Department of Commerce Seattle, Washington Volume 103 Number 1 January 2005 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 12.7cm Totals This study Millionaire net 2 3 0 0 5 Large box net 0 14 19 7 40 Totals 2 17 19 7 45 <6.35-cm 6.35-10.16 cm 11.43-cm Study Gear mesh mesh mesh Unknown Totals NMFS Millionaire net 9 5 3 0 17 Large box bet 5 4 2 7 18 Totals 14 9 5 7 35 Black-sea bass-targeted tows 6.35-10.16 cm 11.43-cm 10.16+11.43 cm Study Gear mesh mesh composite Totals This study Millionaire net 0 0 0 0 Large box net 0 3 9 12 Totals 0 3 9 12 <6.35-cm 6.35-10.16 cm 11.43-cm Study Gear mesh mesh mesh L'nknown Totals NMFS Millionaire net 0 0 0 0 0 Large box net 0 6 0 0 6 Totals 0 6 0 0 6 Bochenek et al.: Assessment of Stenotomus chrysops and Centropnstas striata discards in the Mid-Atlantic Bight fell between 22.9 cm (25th percentile) and 25.0 cm (75th percentile). For the NMFS observer data, the mean size of scup discards was 17.2 cm and ranged from 13.6 to 20.6 cm. Fifty percent of the scup discarded fell between 16.2 cm (25th percentile) and 18.2 cm (75th percentile). For scup landed, the mean size was 22.8 cm and ranged from 19.4 to 28.9 cm. Fifty percent of the scup landed fell between 21.3 cm (25th percentile) and 23.8 cm (75th percentile). Codend and gear Vessels participating in this study and in the NMFS observer program used either millionaire or box nets (Table 1). More tows were made with the box net in both data sets (Table 1). Most tows in our study were made with the composite and 11.43-cm codends because com- parison of these two codends was one focus of our study. Most scup-targeted tows in the NMFS observer database were made with codends <10.16 cm mesh (Table 1). Codend mesh size did not have a significant effect on catch length frequencies when data from our study and the NMFS observer data were analyzed separately or combined. We deleted codends with the smallest meshes (meshes slO.16 cm) and re-analyzed the data for the remaining larger codend meshes. Again, catch length frequencies were not significantly different for any of the codend mesh sizes. Finally, we considered the landed and discarded scup separately. For landings, codend mesh size had a moderately significant effect on median length (P=0.0220) and a stronger significant effect on mean length (P=0.0062). Codends with some meshes 212.7 cm caught more of the landed size fraction than the composite and slightly more than the 11.43-cm mesh codend; however, the actual difference in mean length between the three mesh-size groups was small, approximately one cm (Fig. 2). The efficiency of the codend may change with the amount of fish caught such that selectivity declines with high catches. Accordingly, scup catches were divided into two groups: those above and those below the me- dian catch for all tows. For catches above the median, codend mesh size had a significant effect (P=0.0441) on the 25th percentile of size for scup discarded in our study. The 25th percentile size was largest for codends with some meshes ^12.7 cm and smallest for the com- posite codend. For landed scup, the 25th percentile sizes were about 22.0 cm regardless of codend mesh size. No significant differences existed between codend mesh sizes for scup length frequency in tows with catches below the median. We examined the composition of the catch by weight. Codend mesh size had a limited effect on the ratio of scup discarded to landed, the total catch and to- tal discards of all species, and total scup landed and discarded. Scup discards were greater with codends having some meshes 212.7 cm (P=0.0211). More scup were landed from these tows as well (P=0.0034) and therefore this codend style may contribute to a greater catch rate (Table 2). Very likely, this trend in increased I Landed, Composite 1 Landed 11.43 cm jnik'ii > \11 tin Figure 2 Mean scup {Stenotomus chrysops) length fractions for those tows with landed scup with a composite (10.16 + 11.43 cm) codend, 11.43-cm mesh codend, and a sl2.7-cm mesh codend. l = smallest size. 100 = larg- est size. catch is produced by the small number of tows (n=l) in this mesh size category rather than a real improvement in net performance. No significant gear effects existed for any of the length-frequency fractions in the combined data set (our study and NMFS observer study). The only significant effect of scope (P= 0.0175) was on total scup discarded in our study. This effect was not present in the NMFS observer data set. Discards-to-landings ratio Of the 62 tows completed in our study, 39 targeted scup. The NMFS observer program, from 1997-mid 2000, included 35 scup-targeted tows (Table 3). Overall, mean catch per tow for scup-targeted tows was 972.6 kg in our study and 945.3 kg for NMFS observed tows. In our study, the discards-to-landings ratio for all species combined ranged from 1.77 with the composite codend to 2.91 with a codend with some meshes al2.7 cm. In the NMFS observer data set, the discards-to-landings ratio for all species combined in scup-targeted tows ranged from 0.47 with codends having meshes of 6.35-10.16 cm to 3.43 with codends with meshes of 11.43 cm (Table 2). The mean discards-to-landings ratio for scup ranged from 1.1 for the NMFS database to 2.4 for our study (Table 3). Ratios varied from a low of 0.35 to a high of 5.72 among the various gear and mesh-size combina- tions (Table 4). We analyzed cases where scup discards exceeded or were less than landings. When our data and the NMFS observer data were combined, the 25th (P=0.0219), 50th Fishery Bulletin 103(1) Table 2 Mean weight (in kilograms) of scup discarded, scup landed, total discards of all species, total catch of all species cards-to-landings ratio of all species per tow by codend mesh size for this study and the NMFS observer data. and total dis- Scup Scup Total Total Total discards- Total number Study Codend discarded landed discards catch to-landings ratio of tows This study Composite 659.7 329.3 1078.1 1686.0 1.77 16 This study 11.43 cm 607.8 210.7 1060.2 1437.7 2.81 14 This study al2.70 cm 1020.7 404.9 1973.4 2652.8 2.91 7 NMFS <6.35 cm 615.5 493.4 949.1 1530.1 1.63 14 NMFS 6.35-10.16 cm 230.5 510.5 321.1 999.2 0.47 9 NMFS 11.43 cm 535.2 260.0 1015.7 1311.6 3.43 5 This study NMFS Table 3 Mean catch and landings per tow for scup and black sea bass-targeted tows. Scup Study Tow type Total no. of tows Mean catch (kg) Mean landed (kg) Mean discarded (kg) Ratio of scup discards to landings This study NMFS Target Target 39 35 972.6 945.3 286.3 461.3 686.3 484.0 2.40 1.05 Black sea bass Study Tow type Total no. of tows Mean catch (kg) Mean landed (kg) Mean discarded (kg) Ratio of black sea bass discards to landings Target Target 10 6 365.0 171.9 278.7 170.1 86.3 1.8 0.31 0.01 (P=0.0085), and 75th (P=0.0038) percentile sizes and the mean length (P= 0.0001) were significantly lower for tows in which most scup were discarded (Fig. 3). When the data sets were analyzed separately, our study found that the 50th (P=0.0133) and 75th (P=0.0040) percentile sizes and the mean length (P= 0.0338) were significantly lower for tows where discards exceeded landings. Not surprisingly, when fishermen caught larger scup, fewer scup were discarded. In the NMFS observer data set, no significant size effects were found for any of the percentile fractions. When the discards and landings were analyzed sepa- rately, the lengths of fish discarded did not differ be- tween tows for which discards exceeded landings and tows for which landings exceeded discards. However, for the landed fish, the 50th (P=0.0034) and 75th (P=0.0018) percentile sizes and the mean length (P=0.0033) were larger for tows where landings exceeded discards (Fig. 4). Discards decline when larger scup are propor- tionately more abundant in the catch. We examined the influence of total catch (all species combined) on the length-frequencies of scup in tows where scup landings exceeded or did not exceed scup discards. For those tows with total catches below the median catch, a significant effect was noted for the median (P=0.0039), the 75th percentile (P=0.0006), and the mean (P=0.0288) length of scup. In those tows where total catch weight was relatively low, the median, mean, and 75th percentile lengths were larger in tows where scup landings exceeded discards. No significant effects on the length-frequency distribution of scup were observed for total catches that were above the median. The analysis identifies a strong trend towards the land- ing of larger-size scup in tows yielding total catches below the median for all tows. Both landings and discards were affected in those tows in which total catch fell below the median. For landed scup, the median (P= 0.0062), the 75th percentile (P=0.0051), and the mean (P=0.0113) length were higher in tows with total catches below the median when scup landings exceeded discards. For those scup that were discarded from tows with total catches below the me- dian, a significant size effect was observed for the 75th percentile (P=0.0265). Discarded scup were larger in Bochenek et al.: Assessment of Stenotomus chrysops and Centropnstas striata discards in the Mid-Atlantic Bight Table 4 Synopsis of catch and landings data (kg) by study, the NMFS observer database. gear, and codend mesh size for scup-targeted tows in our study and those in dear Mesh size Total scup/tow Scup landed Scup discarded Ratio of scup discards to landings Number of tows This study Large box 10.16+11.43 cm composite 1332.3 519.3 813.0 1.57 9 11.43 cm 1572.8 431.6 1141.2 2.64 5 al2.70 cm 2609.7 624.1 1985.6 3.18 2 Millionaire 6.35-10.16 cm 32.7 16.3 16.3 1.00 1 10.16+11.43 cm composite 547.5 85.0 462.5 5.44 7 11.43 cm 399.5 88.1 311.5 3.54 9 212.70 cm 951.9 317.2 634.8 2.00 5 Unknown 636.8 94.8 542.0 5.72 1 NMFS Large box <6.35 cm 1076.7 196.0 880.8 4.49 5 6.35-10.16 cm 20.8 3.4 17.4 5.12 4 11.4.3 cm 176.9 131.5 45.4 0.35 2 Unknown 988.2 477.6 510.6 1.07 7 Millionaire <6.35 cm 1126.6 658.6 468.1 0.71 9 6.35-10.16cm 1317.1 916.2 401.0 0.44 5 11.43 cm 1207.3 345.5 861.8 2.49 3 Table 5 Total discards and total catch of all fish species (in kg) and scup discarded and landed (in by having more or less discards of scup than the median catch per tow. kg) for only those tows characterized Study More or less discards Scup discarded Scup landed Total discards Total catch Total number of tows This study Less This study More NMFS Less NMFS More 145.4 898.9 235.1 815.9 355.6 259.0 521.1 381.5 565.4 1451.6 426.3 1242.8 1027.5 11 1982.3 28 1058.3 20 1761.5 15 these tows, reflecting the overall larger size of the scup catch in tows where total catch was relatively low. Finally, for tows in which scup discards exceeded landings, total catch of all species and total discards of all species were also high. This trend was significant for total catch (P=0.0273) and total discards (P=0.0038) in our study (Table 5) and for total discards (P= 0.0017) and total catch (P=0.0112) in the NMFS observer data set (Table 5). Therefore, scup discards tended to increase with respect to landings as total catch increased. Time and effort For our study, effort significantly affected the 25th (P=0.0247) and 50th (P=0.0466) percentiles of the size- frequency distribution of discards. The size frequencies for landings were not similarly affected. In both former cases, the 25th and 50th percentile sizes were larger when effort was less (shorter tows). No significant effects were observed in the NMFS observer data set. Because the length frequency of the entire catch did not change sig- nificantly, this is likely an effect of processing onboard the boat. Given trip limits, one might anticipate discards to increase in tows made at the end of the trip. We ex- amined the amount of scup caught either in the first half of the tows or in the last half of the tows on each trip. For this study, more scup were landed (P= 0.0008) and discarded (P=0.0001) in tows that occurred during the last half of the trip. Total catch and total discards were unaffected. For the NMFS observer data set, more scup were landed (P=0.0001) and discarded (P=0.0001) Fishery Bulletin 103(1) £ 25- 20 10 25 Mean 50 75 100 Discarded 1 25 Mean 50 Percentile 75 100 1 M landed, less disc. □ m. landed, more disc. □ N landed, less disc. n^ landed, more disc. ■ M. discarded, less disc. □ M discarded, more disc □ N, discarded, less disc. □ N. discarded, more disc. Figure 3 Percentiles of scup {Stenotomus chrysops) length frequency for those tows in which discards exceeded or failed to exceed landings for landed and discarded scup. "M" represents this study and "N" represents NMFS observer data. Landed, less disc. = for scup landed, tows with less discarded scup than landed scup. Discarded, less disc. = for scup discarded, tows with less discarded scup than landed scup. Landed, more disc. = for scup landed, tows with more discards of scup than landed scup. Discarded, more disc. = for scup discarded, tows with more discards of scup than landed scup. l = smallest size. 100=largest size. and the total catch of all species (P=0.0195) and total discards of all species (P= 0.0004) were higher in tows taken during the last half of the trip (Table 6). More scup being landed and discarded in the last half of the trip indicates that captains learn where to fish for scup during the trip and CPUE rises as a consequence. No evidence exists that discards increased with respect to landings during the trip. We anticipated that reduction of the trip limit from 4536 kg to 454 kg on 24 January would influence the total weight of discards. Time did influence total weight of scup discards (P=0.0056) in our study. More discards per tow occurred on trips taken prior to 24 January, likely because of the larger trip limit (weight limit per species for each trip) for the 1-24 January period. With a larger trip limit, more scup can be caught per tow and therefore more scup will be dis- carded. The discards-to-landings ratio, however, was not significantly affected — indicating that captains controlled total scup catch in proportion to the land- ing limit. The present study versus the NMFS observer study We compared trends in our data with those in the NMFS observer data. The subset of directed scup tows in the two data sets rarely disagreed, despite the disparity in codend mesh sizes reported (Table 1). Bochenek et ai.: Assessment of Stenotomus chrysops and Centropnstas striata discards in the Mid-Atlantic Bight Landed Landed, less ihs^ D Landed, more disc 25 Mean 50 75 100 Discarded o 25 I Discarded, less disc. I Discarded, more disc. Figure 4 Percentiles of scup {Stenotomus chrysops) lengths (FL) for those tows in which discards exceeded or failed to exceed landings for landed and discarded scup. Codends included the composite (10.16 + 11.43 cm), 11.43 cm, and &12.7 cm meshes. Landed, less disc. = for scup landed, tows with less discarded scup than landed scup. Discarded, less disc. = for scup discarded, tows with less discarded scup than landed scup. Landed, more disc. = for scup landed, tows with more discards of scup than landed scup. Discard, more disc. = for scup discarded, tows with more discards of scup than landed scup. l = smallest size. 100 = largest size. Table 6 Mean weight (in kg) of scup discarded and landed and the total of all fish tows in the first half of the trip and the second half of the trip. species landed and discarded per tow for scup-targeted Study First half or second half of trip Scup discarded Scup landed Total discards Total catch Total number of tows This study First 253.3 125.2 970.2 1279.7 22 This study Second 1246.8 494.7 1501.2 2273.8 17 NMFS First 43.2 23.2 463.0 670.3 19 NMFS Second 1007.5 981.5 1148.3 2178.3 16 10 Fishery Bulletin 103(1) Table 7 Mean weight (in kg) of black sea bass discarded, black sea bass landed, total discards of all species, total catch of all species, and total discards-to-landings ratio of all species per tow by codend mesh size and gear for this study and the NMFS observer data. Study Gear Codend Black sea bass discarded Black sea bass landed Total discards Total catch Total discards to landings Ratio of Total number of tows This study Large box 10.16+11.43 cm composite 119.2 366.7 845.6 1306.9 1.83 7 This study Large box 11.43 cm 5.0 25.9 1201.8 1250.3 24.78 2 This study Millionaire 11.43 cm 18.6 168.3 368.1 683.8 1.17 1 NMFS Large box 6.35-10.16 cm 1.8 170.1 23.7 224.2 0.12 6 Catch statistics — black sea bass During the winter scup season, black sea bass are legally caught with 10.16-cm mesh codends in offshore waters. A boat captain often will target scup and black sea bass on the same trip, but will use different mesh codends. A total of 12 black-sea-bass-targeted tows were observed in our study and 6 black-sea-bass-targeted tows were documented in the NMFS observer data set (Table 1). Length frequency — black sea bass Black sea bass length-frequency distributions were highly significantly different (often P=0.0001) between those fish landed and those discarded. The mean size of discarded black sea bass from our study was 22.9 cm and ranged from 18.4 to 25.4 cm. Fifty percent of the black sea bass discarded fell between 22.1 cm (25th percentile) and 24.3 cm (75th percentile). In contrast, the mean size of landed black sea bass was 31.1 cm and ranged from 25.4 to 40.9 cm. For black sea bass landed, fifty percent of the fish were found between 28.6 cm (25th percentile) and 33.3 cm (75th percentile). For the NMFS observer data, the mean size of black sea bass discarded was 23.4 cm and ranged from 20.7 to 27.0 cm. Fifty percent of the black sea bass discarded fell between 22.3 cm (25th percentile) and 24.7 cm (75th percentile). The mean size of landed black sea bass was 28.5 cm and ranged from 24.5 to 34.0 cm. For landed black sea bass, fifty percent fell between 25.0 cm (25th percentile) and 31.5 cm (75th percentile). Codend and gear Nine tows were made with the composite codend and three tows were made with the 11.43-cm legal mesh codend in our study. For the NMFS observer data, all six targeted tows fell into the 6.35-10.16 cm mesh-size group that included the legal mesh size of 10.16 cm (Table 1). We found no significant effects of codend mesh size on the percentile length-frequency fractions of black sea bass. We considered landed and discarded black sea bass separately for those tows with total catches above and below the median and, once again, no significant codend mesh-size effects were observed. The total num- ber of tows, however, was small. A significant codend mesh-size effect (P=0.0389) was observed for black sea bass landed. Landings were higher with the larger mesh codends (composite 10.16+11.43 cm codend and the 11.43-cm codend) rather than with the sl0.16-cm mesh codend (Table 7). The small number of total tows with the larger codend mesh sizes (10.16+11.43 cm and the 11.43 cm) is probably responsible for this difference in landings rather than differences in net performance. Gear effects (net types) were not determined because only the box net was used. Discards-to-landings ratio Total mean catch per tow was 365 kg and total mean landings per tow was 279 kg for the 10 tows in our study. For the six directed tows in the NMFS observer data, average total catch was 172 kg and average total land- ings were 170 kg (Table 3). In black-sea-bass-targeted tows, the black sea bass catch comprised 34.2% of the total catch. The discards-to-landings ratio for black sea bass was 0.230. Relatively few black sea bass were dis- carded. Scup comprised 0.9% of the total catch in black sea bass targeted tows. Less than one percent (0.4%) of the scup catch in black-sea-bass-targeted tows was discarded. We analyzed cases where black sea bass discards ex- ceeded or were less than landings in tows where total catch (all species combined) was above or below the me- dian. For total catches above the median, a significant size effect was noted for the median length (P=0.0040), the 75th percentile size (P= 0.0007), and the mean length (P= 0.0026). Larger fish were present in tows where dis- carding was lower (Fig. 5). No significant effects on the size distribution of black sea bass were observed in tows with total catches below the median. We further divided the catches above the median into discards and land- ings. For those black sea bass that were landed from tows with total catches above the median, a significant size effect was observed for the 25th (P=0.0199), the Bochenek et al.: Assessment of Stenotomus chrysops and Centropnstas striata discards in the Mid-Atlantic Bight Landed I Landed, less disc. I anded, more disc Discarded I Discarded, less disc. Discarded, more disc I 25 Mean 50 Percentile Figure 5 Percentiles of black sea bass (Centropristas striata) length frequencies for those tows in which discards of black sea bass exceeded or failed to exceed landings for landed or discarded black sea bass. Landed, less disc. = for black sea bass landed, tows with less discards of black sea bass than what was landed. Discarded, less disc. = for black sea bass discarded, tows with less discards of black sea bass than what was landed. Landed, more disc: for black sea bass landed, tows with more discards of black sea bass than what was landed. Discarded, more disc. = for black sea bass discarded, tows with more discards of black sea bass than what was landed. l = smallest size. 100 = largest size. 50th (P=0.0280), and the 75th (P=0.0090) percentile size fractions and the mean length (P=0.0133). The size of landed black sea bass was larger in tows where discard- ing was low (Fig. 5). Time and effort Because of trip limits, discards could increase in tows taken near the end of a trip. Therefore, we compared the catch of black sea bass in the first and the last half of the tows. The quantity caught and the length-frequency percentiles were not significantly different between the first and last half of the tows. In contrast, for scup trips, discards and landings tended to be higher in tows made in the last half of the trip. Effort significantly affected the 25th (P=0.0010), 50th (P= 0.0003), and 75th percentile (P= 0.0153) size fractions of black sea bass for the combined data sets (NMFS study and our study). In these cases, higher effort was associated with more smaller fish. When the two data sets were analyzed independently, most of the effort effects were no longer present. 12 Fishery Bulletin 103(1) Discussion Scup The type of net (gear) and the size of codend mesh had only a minor effect on the length frequencies of scup caught. Although variations in codend mesh size nor- mally influence catch in other studies (Hastie, 1996; Petrakis and Stergiou, 1997; Stergiou et al., 1997; Broad- hurst et al., 1999), a wide range in codend mesh sizes produced similar results for scup. Codends with some meshes al2.7 cm appeared to catch more of the size classes of fish chosen for landing than the composite codend and just slightly more than the legal 11.43-cm mesh codend; therefore the al2.7-cm mesh condend may reduce discards. The actual difference in scup lengths between the three codends was only about one cm. In terms of kilograms caught, more scup were caught in tows with the larger codend mesh. Landings increased, but so did discards, so that the discards-to-landings ratio remained unchanged. This finding indicates that the small upward bias in sizes caught did not significantly reduce total catch. In general, the smaller mesh codends (6.35-10.16 cm) and the composite codend (10.16+11.43 cm) performed similarly to the current legal mesh design (11.43 cm). Overall, discards of scup remained high regardless of the type of gear (nets) and codends used. In our study, more larger scup were caught in longer tows. When a boat encounters a large school of scup, the mean length of the catch tended to be smaller. In addition, the larger-size scup tended to be caught more often in those tows with total catches below the me- dian. This trend is probably a biological effect, but an effect of mesh size or gear cannot be excluded. Most populations contain relatively few larger fish and, there- fore, more smaller individuals. Morse6 (in Steimle et al., 1999) noted that scup schools are size-structured. When larger scup are less common in schools, then schools with these larger individuals most likely would be smaller and more effort would be required to achieve the same catch of these individuals. The same result would occur if larger scup tended to be on the outside or above smaller scup in schools. Little is known about scup behavior. However, any spatial size structure in the population could promote a direct relationship be- tween effort and the mean length of fish caught and an inverse relationship between total catch (all species) and mean length of fish caught. As an alternative explanation for the lower catch rate of larger individuals, clogging of the codend may occur when catch rates are high and, as a consequence, size- selectivity would decline. Different codend mesh sizes do not seem to affect the number of discarded scup as much as one might anticipate because codends clog dur- 6 Morse, W. W. 1978. Biological and fisheries data on scup, Stenotomus chrysops (Linnaeus). NMFS, NEFSC, Sandy Hook Lab. tech. ser. rep. no. 12, 41 p. James J. Howard Marine Sciences Laboratory, Northeast Fisheries Science Center, 74 Magruder Rd., Sandy Hook, NJ 07732. ing the interception of large schools. Accordingly, lower CPUE could produce greater size selectivity resulting in increased mean length when catches are relatively low. However, the trends observed in length frequency with effort and total catch were not significantly influenced by codend mesh size. Accordingly, the observed trend is likely a direct consequence of fishing on size-structured populations. In general, more scup were landed and discarded in the last half of a trip. This finding indicates that the captain learns where to fish for scup by the second half of the trip and CPUE increases as a consequence. We had expected an increase in discards as the trip limit was reached towards the end of the trip. However, no effect of trip limits on the total weight of discards could be discerned in our data set or the NMFS observer data set. More scup were discarded per tow in tows observed during the first half of the 2001 season, namely 1-24 January, than in the second half of the season, 25 January-February, but the discards-to-landings ratio did not vary for either half of the season. The fact that the ratio did not differ indicates that more scup are discarded per tow when fishermen are allowed a larger trip limit (4536 kg). The higher discards of scup per tow during the first half of the season are likely due to the increased total catch per tow that might be anticipated when allowable landings are higher. Accordingly, cap- tains are able to reduce catch rate and, thus, discards when landing limits are low. We compared the NMFS observer database to our observer data. Despite a substantial variation in the distribution of codend mesh sizes between the two data sets, the discards-to-landings ratio was not significantly different. Concerns raised by the high discards-to-land- ings ratio observed in the NMFS observer data were supported by our study. The discards-to-landings ratio for the directed scup fishery consistently exceeded 1.0. In summary, the objective of our study was to evalu- ate the effect of codend mesh size on the amount of scup discards and to identify mechanisms to reduce scup discards. Although we observed a number of trends in discards in our study, neither the current legal mesh nor any of the experimental codends seem to adequately filter out scup smaller than 22.86 cm. Neither did trip limits seem to influence the total weight of scup dis- cards. In fact, the only consistent trends produced by variations in effort and total catch seem most likely due to biological effects not easily controlled for by the captain of a fish vessel. Overall, the total weight of dis- cards seems to be primarily a function of the regulated size limit, abetted by the tendency for smaller fish to be captured when encounter rates are high. The present study found that the length of the median discards was about 17.78 cm FL (19.83 cm TL based upon a conver- sion factor of Hamer4 [in MAFMC, 1996]). O'Brien et al. (1993) and NEFSC (1993) reported that 50% of both male and female scup reach maturity at 15.49 cm FL (17.27 cm TL). Therefore, lowering the scup minimum size limit to 17.78 cm FL (19.83 cm TL) would greatly Bochenek et al .: Assessment of Stenotomus chrysops and Centropnstas striata discards in the Mid-Atlantic Bight 13 reduce scup discards, yet permit the majority of scup to attain sexual maturity. Kilograms discarded might be reduced by more than half. Fishermen would reach their trip limit sooner and thus stop fishing earlier. As a result, fishing mortality rate even on larger scup would be reduced. This single change would reduce discards more than any change in net or codend design tested to date and would not result in any increase in fishing-induced mortality for scup. the percentage in scup-targeted tows. This finding indi- cates that there is considerable discrimination between the two species at the level of the fishery. The black sea bass fishery is currently regulated under the small- mesh fishery GRA plan in which fishing is prohibited in some areas to reduce scup mortality. This investigation finds no evidence to support the efficacy of this manage- ment approach. Scup discards do not appear to be an important attribute of the black sea bass fishery. Black sea bass Estimates of discards of black sea bass are low in the black-sea-bass-targeted fishery, based on the few observed tows in our study and data from the NMFS observer database. Regardless of which codends were used, the same size fractions of black sea bass were caught. The composite codend (10.16+11.43 cm mesh) caught more black sea bass than were landed. Discards was also higher. As with scup, mesh size and gear type had minor effects on the size frequency, the discards- to-landings ratio, and the kilograms of black sea bass caught. The majority of tows where black sea bass were caught had ratios of black sea bass discarded to landed of less than 0.3, indicating that few discards occur in this fishery. In contrast, most of the scup tows were charac- terized by discards-to-landings ratios greater than one. The differences in discards-to-landings ratios between black sea bass and scup may be due to a combination of biological factors controlling the average size of scup in the larger schools and to regulatory factors that do not match well with the size range of scup in schools. Unlike scup, black sea bass size frequencies and total weight caught were similar in tows taken during the first and last half of the trip. Trip limits are in effect for both black sea bass and scup. The difference between the two species in the distribution of catch through the time course of the trip may be the result of biological effects in that the schooling of scup would tend to pro- duce higher catches during the middle or latter part of the trip as the captain finds schools of fish. Powell et al.3 showed that black sea bass and scup are caught simultaneously more frequently than expected by chance in tows in the Atlantic mackerel (Sco?nber sco?nbrus), Loligo squid, scup, and silver hake fisheries and suggested that they should be regulated together. Our analysis also showed this pattern in that the two species were frequently caught in the same tows (39 out of 40 scup-targeted tows and seven out of 10 black- sea-bass-targeted tows caught both scup and black sea bass). In addition, Shepherd and Terceiro (1994), Musick et al., (1985), and Musick and Mercer (1977) also found that both scup and black sea bass were caught in the same tow. Use of a common codend mesh size regulation for both fisheries may prove useful. The failure to find significant differences between mesh sizes suggests that the 10.16+11.43 cm composite bag might be a reasonable choice for both fisheries. However, scup discards were a small fraction of black sea bass landings in black-sea- bass-targeted tows (0.4%) — very small in comparison to Conclusions Because fishermen catch both scup and black sea bass in the same tow and because the current regulations require fishermen to use an 11.43-cm mesh codend when targeting scup, and, a 10.16-cm mesh codend when tar- geting black sea bass, two different codend mesh sizes are used on the same trip. The composite codend was designed to retain the smaller black sea bass catches and some scup when catch rates are low but permits more scup to escape at higher catch rates. The composite codend (10.16+11.43 cm mesh) performed as well as the other codends used in our study, including the 11.43-cm legal-size codend. The composite codend with 10.16-cm mesh followed by the 11.43-cm or 12.7-cm mesh codends should be further evaluated on both black sea bass and scup-directed tows. If this composite codend works equally as well as the legal 11.43-cm mesh codend cur- rently in place for scup (and the data presented here sug- gest that it does), consideration should be given to using this codend because it permits the retention of smaller black sea bass without negatively influencing scup. This change would eliminate the need to carry two codends onboard and thus would reduce overall trip costs without impacting the number of scup discards. However, neither codend successfully addresses the need to significantly reduce scup discarding in the scup-directed fishery. Codends with some 12.7-cm meshes tended to reduce discards by reducing the catchability of smaller scup, but the trends were often not significant, possibly due to the small sample size, but possibly also because nets were clogged by schools of smaller-size scup. The data indicate that further studies with 12.7-cm or greater mesh composites may identify codend configurations that will produce fewer discards. DeAlteris and La Valley (1999) have documented that scup can survive capture in a trawl net and subsequent escapement. Therefore, optimizing codend mesh size could reduce discard mortality. Larger scup were caught in tows where the total catch weight was low. Large catches tended to accompany the interception of scup schools. These large catches can clog the nets and thus reduce size selection even at larger mesh sizes. Alternatively, larger scup may not be associated with smaller scup in schools. We cannot discriminate between the two explanations. Regard- less of the reason, the tendency of the largest catches to contain proportionately more smaller fish will likely minimize the positive influence of net management in 14 Fishery Bulletin 103(1) reducing scup discards. Rather, the tendency of the largest catches to contain proportionately more smaller fish suggests that fisheries managers may want to lower the legal-size limit for scup from 22.86 cm to 17.78 cm FL. The median size of scup discards in our study was 17.78 cm FL. Setting the size limit at 17.78 cm FL (19.83 cm TL) would greatly reduce discards and thus overall discard mortality. This management change would likely have a much greater effect in reducing scup discards than any other single management mea- sure directed at gear modification or area closure and would not endanger the stock (most discarded scup fail to survive); thus, any approach significantly reducing discards must significantly increase overall survival of the population. Acknowledgments We would like to thank the National Fisheries Institute, Scientific Monitoring Committee, for providing support for this project. We also thank the captain and crew for the use of the four commercial fishing vessels from Cape May that cooperated in the project. Without their assistance, this project would not have been possible. We also thank NMFS-NEFSC for providing the NMFS observer data used in our analysis. Literature cited Alverson, D. L. 1999. Some observations on the science of bycatch. MTS (Marine Technology Society) Journal 33(2):6-12. Broadhurst, M. K., S. J. Kennelly, and S. Eayrs. 1999. Flow-related effects in prawn-trawl codends: poten- tial for increasing the escape of unwanted fish through square-mesh panels. Fish. Bull. 97:1-8. DeAlteris, J., and K. J. La Valley. 1999. Physiological response of scup, Stenotomus chrysops, to a simulated trawl capture and escape event. MTS Journal 33(2):25-34. Glass, C. W., B. Sano, H. O. Milliken, G. D. Morris, and H. A. Carr. 1999. Bycatch reduction in Massachusetts inshore squid (Loligo pealei) trawl fisheries. MTS Journal 33(2):35-41. Hastie, L. C. 1996. Estimation of trawl codend selectivity for squid (Loligo forbesi), based on Scottish research vessel survey data. ICES J Mar Sci 53:741-744. Howell, W. H., and R. Langdon. 1987. Commercial trawler discards of four flounder spe- cies in the Gulf of Maine. North Am. J. Fish Manag. 7:6-117. Kennelly, S. J. 1999. Areas, depths and times of high discard rates of scup, Stenotomus chrysops, during demersal fish trawl- ing off the northeastern United States. Fish. Bull. 97:185-192. MAFMC (Mid-Atlantic Fishery Management Council). 1996. Amendment 8 to the summer flounder, scup, and black sea bass fishery management plan: fishery manage- ment plan and final environmental impact statement for the scup fishery. January 1996, 353 p. Mid-Atlantic Fishery Management Council, Dover, DE. Mooney-Seus, M. 1999. Formula for bycatch reduction. MTS Journal 33(21:3-5. Musick, J. A., J. A. Colvocoresses, and E. J. Foell. 1985. Seasonality and the distribution, availability and composition of fish assemblages in Chesapeake Bight. In Fish community ecology in estuaries and coastal lagoons: towards an ecosystem integration (A. Y. Arancibia, ed.), p. 451-474. OR(R) UNAM (Uni- versidad Nacional Autonoma de Mexico) Press, Mexico. [ISBN 9688376183.] Musick, J. A., and L. P. Mercer. 1977. Seasonal distribution of black sea bass, Centropris- tis striata, in the Mid-Atlantic Bight with comments on the ecology and fisheries of the species. Trans. Am. Fish. Soc. 106:12-25. NEFSC (Northeast Fisheries Science Center). 1993. Status of the fishery resources off the Northeastern United States. NOAA Tech. Mem. NMFS-F/NEC-101, 140 p. O'Brien, L., J. Burnett, and R. K. Mayo. 1993. Maturation of nineteen species of fin fish off the northeast coast of the United States, 1985-1990. NOAA Tech. Rep. NMFS 113, 66 p. Petrakis, G., and K. I. Stergiou. 1997. Size selectivity of diamond and square mesh codends for four commercial Mediterranean fish species. ICES J Mar Sci. 54:13-23. Powell, E. N., A. J. Bonner, B. Muller, and E. A. Bochenek. 2004. Assessment of the effectiveness of scup bycatch- reduction regulations in the Loligo squid fishery. J. Environ. Manag. 71:155-167. Roel, B. A., K. L. Cochrane, and J. J. Field. 2000. Investigation into the declining trend in Chokka squid Loligo vulgaris reynaudii catches made by South African trawlers. S. Afr. J. Mar. Sci. 22:212-135. Shepherd, G. R., and M. Terceiro. 1994. The summer flounder, scup and black sea bass fishery of the Middle Atlantic Bight and southern New England waters. NOAA Tech. Rept NMFS 122, 13 p. Steimle, F. W., C. A. Zetlin, P. L. Berrien, D. L. Johnson, and S. Chang. 1999. Essential fish habitat source document: scup, Stenotomus chrysops, life history and habitat char- acteristics. NOAA Tech. Mem. NMFS-NE-149, 48 p. Stergiou, K. I., C-Y. Politou, E. D. Christou, and G Petrakis. 1997. Selectivity experiments in the NE Mediterranean: the effect of trawl codend mesh size on species diversity and discards. ICES J Mar Sci 34:774-786. Suuronen, P., J. A. Perez-Comas, L. Lehtonen, and V. Tschernij. 1996. Size-related mortality of herring (Clupea harengus L.) escaping through a rigid sorting grid and trawl codend meshes. ICES J Mar Sci 53:691-700. 15 Abstract — Fecundity was estimated for shortspine thornyhead (Sebas- toiobus alascanus) and longspine thornyhead (S. altivelis) from the northeastern Pacific Ocean. Fecun- dity was not significantly different between shortspine thornyhead off Alaska and the West Coast of the United States and is described by 0.0544 xFL3978, where FL=fish fork length (cm). Fecundity was esti- mated for longspine thornyhead off the West Coast of the United States and is described by 0.8890 xFL3249. Contrary to expectations for batch spawners, fecundity estimates for each species were not lower for fish collected during the spawning season compared to those collected prior to the spawning season. Stereological and gravimetric fecundity estimation techniques for shortspine thornyhead provided similar results. The stereo- logical method enabled the estimation of fecundity for samples collected ear- lier in ovarian development; however it could not be used for fecundity esti- mation in larger fish. Fecundity of shortspine thornyhead (Sebastoiobus alascanus) and longspine thornyhead (5. altivelis) (Scorpaenidae) from the northeastern Pacific Ocean, determined by stereological and gravimetric techniques* Daniel W. Cooper Katherine E. Pearson Donald R. Gunderson School of Aquatic and Fishery Sciences University of Washington 1122 NE Boat Street Seattle, Washington 98105 Present address (for D W Cooper, contact author): Alaska Fisheries Science Center, F/AKC2 7600 Sand Point Way NE Seattle, Washington 98115-0700. E-mail address (for D W Cooper) dan.cooper@noaa.gov Manuscript submitted 15 July 2003 to the Scientific Editor's Office. Manuscript approved for publication 20 September 2003 by the Scientific Editor. Fish. Bull. 103:15-22 12005). Shortspine thornyhead {Sebastoiobus alascanus) is distributed from the Bering Sea to Baja California (Orr et al., 2000). Longspine thornyhead (S. altivelis) is distributed from the Gulf of Alaska to Baja California (Orr et al., 2000), and a few specimens have recently been collected in the eastern Bering Sea (Hoff and Britt, 2003). Both species are commercially impor- tant (Piner and Methot, 2001; Gaichas and Ianelli, 2003) and inhabit deep waters over the continental shelf and slope. Both shortspine and longspine thornyhead are determinate spawn- ers (Wakefield, 1990; Pearson and Gunderson, 2003), and spawn pelagic, gelatinous egg masses (Pearcy, 1962; Best, 1964; Wakefield, 1990; Wake- field and Smith, 1990). Shortspine thornyhead spawn between April and July in Alaska, and between Decem- ber and May along the West Coast of the United States, whereas longspine spawn between January and April along the West Coast (Pearson and Gunderson, 2003). Annual fecundity is used as a mea- sure of reproductive output in fishery population models and life history studies. Accurate annual fecundity estimates require identifying oocytes to be spawned in the current spawn- ing season. For iteroparous spawn- ers, developing oocytes are often distinguished from reserve oocytes by diameter or yolk presence (Macer, 1974). Collection date for samples is important. If samples are collected too early in oocyte development, some developing oocytes will be indistin- guishable from reserve oocytes, and fecundity will be underestimated. In shortspine thornyhead, oo- cyte stages 4-8 are maturing to be spawned in the current spawning sea- son, whereas oocyte stages 1-3 are reserve oocytes to be spawned in fu- ture spawning seasons (Pearson and Gunderson, 2003). Early vitellogenic oocytes (stage 4) overlap in size with late perinucleus (stage 3) reserve oo- cytes (Pearson and Gunderson, 2003). Late vitellogenic oocytes (stage 5) are easily distinguished from reserve oo- cytes. In whole oocytes, neither oo- cyte size nor appearance can be relied on to distinguish stage-3 and early stage-4 oocytes; however stage-3 and stage-4 oocytes can be visually dis- tinguished from histological samples (Pearson and Gunderson, 2003). Em- erson et al. (1990) developed a stereo- logical method to estimate fecundity : Contribution 929 from the Joint Insti- tute for the Study of the Atmosphere and Ocean (JISAO), 4909 25th Ave NE, Seattle, WA. 16 Fishery Bulletin 103(1) from histological sections. Unlike gravimetric methods (e.g., Hunter et al., 1992) where whole oocytes are used to estimate fecundity, stereological methods do not rely on oocyte diameter or other proxies for vitellogenesis. A collection of shortspine thornyhead ovaries from Alas- ka contained few specimens considered suitable for a gravimetric fecundity method because too few of the specimens contained all developing oocytes in stage 5 or beyond. However, enough samples were suitable for the stereological method. This study provides a fecundity estimate based on stereological and gravimetric techniques for shortspine thornyhead off Alaska. Benefits and limitations of the stereological method in this case are discussed. A gravi- metric technique is also used to estimate fecundity for longspine thornyhead and shortspine thornyhead from samples off the West Coast of the United States. In addition, we examine the hypothesis that thornyheads are batch spawners, and that fecundity consequently declines over the course of the spawning season (Wake- field, 1990). Materials and methods Ovaries were collected from a large geographic area in Alaska, including the Gulf of Alaska, the Aleutian Islands, and the Bering Sea. National Marine Fisheries Service (NMFS) observers aboard commercial fishing vessels collected ovaries from April through June 2000. Length and somatic weight (ovaries and stomach con- tents removed) (±5 g) were recorded at sea. Ovaries were excised and placed in 10% formalin solution buffered with sodium bicarbonate. Ovaries from shortspine thornyhead and longspine thornyhead were also collected during the 1999 NMFS West Coast trawl survey. Samples were collected be- tween Northern California and Washington (34°57'N lat. 121°33'W long, to 48°04' lat. 125°58'W long.). Length and somatic weight (±2 g) were recorded at sea. Additional West Coast longspine and shortspine thornyhead ovaries were collected from commercial fishing vessels by the Oregon Department of Fish and Wildlife in Astoria. Ovaries were collected off Oregon and Washington from February through May 2000, during December 2000, and during January 2001. Af- ter shipment to the NMFS Alaska Fisheries Science Center in Seattle, length, somatic weight (±2 g), and ovary weight (±0.001 g) were recorded. Ovaries were excised and placed in 10% formalin buffered with so- dium bicarbonate. A cross section was removed from one ovarian lobe (middle or middle posterior region) for histological pro- cessing. When a whole cross section was too large to fit on a microscope slide, a wedge was cut from the cross section that included both the ovarian wall and the center of the ovary. Samples were processed through a dehydration series, embedded in paraffin, and sec- tioned at 4 um. Slides were stained with hematoxylin and eosin. Gravimetric fecundity estimation Histological ovary sections were examined at 100 x mag- nification to select samples for the gravimetric method. Oocytes were identified to one of eight developmental stages as described by Pearson and Gunderson (2003). To differentiate between oocytes to be spawned in the current year and reserve oocytes for future years, only ovaries with all maturing oocytes in stage-5 (late vitel- logenesis) and beyond were used. By definition, yolk fills more than 50% of the cytoplasm within stage-5 oocytes, and the dark yolk made it easy to distinguish these oocytes. Stage-4 oocytes would also be spawned in the current year but overlapped significantly in size with nonmature stage-3 oocytes, and early stage-4 oocytes did not always have enough yolk (0-50%) to differentiate them from stage-3 oocytes with the gravimetric method. Specimens containing any stage-4 oocytes were omitted as a result. Ovaries with stage-8 oocytes were also omit- ted because the increased amount of gelatinous material which surrounds the oocytes in Sebastolobus could not be contained within the ovaries during subsampling. Ovaries were weighed (±0.001 g) after they had been stored in formalin. Subsamples were cut from the ova- ries and weighed (±0.001 g). For smaller ovaries, an entire cross section was taken. For larger ovaries, a pie-piece-shaped wedge was cut from the cross section to ensure a representative sample of outer ovarian wall. When cut correctly, a wedge starting at the center of the cross section would have the same weight ratio of ovarian wall to wedge subsample as the original cross section. Subsamples usually contained approximately 1000 oocytes (mean=1133), but this number varied ac- cording to stage of development and the amount of ge- latinous material in the ovary (range: 108-3711). Gelatinous material could not be subsampled by cut- ting at room temperature; therefore ovaries were briefly frozen before subsampling. This procedure enabled the gelatinous material to be cut, and also made it easier to obtain a representative sample of the ovarian wall. Ini- tially, parts of three ovaries were frozen, and no effects of the freezing were detected with a light microscope. Only samples for gravimetric fecundity estimates were briefly frozen. No difference in oocyte density was found among the different regions of the ovaries (see "Results" sec- tion); however, gravimetric subsamples were still taken randomly along the length of the ovaries to minimize potential bias from any location. The oocytes in the subsamples were counted under a stereomicroscope, and fecundity was estimated by W Fec = —N, w where Fee = estimated fecundity; W = total ovary weight; w = subsample weight; and n = number of oocytes in the subsample. Cooper et al.: Fecundity of Sebastolobus alascanus and Sebastolobus altivelis 17 Stereological fecundity estimation The majority of oocytes within an ovary were found to be at the same developmental stage; however develop- ment was not completely synchronous. Some ovaries containing stage-5 and -6 oocytes (late vitellogenesis to migratory nucleus) also contained a few stage-4 oocytes, which although unsuitable for fecundity esti- mation with the gravimetric method, could be used with the stereological method described by Emerson et al. (1990). Fecundity was estimated from ten of these samples by using the stereological method to complete the shortspine thornyhead collection from Alaska. Fecundity was estimated per unit of volume and then multiplied by the volume of both ovaries. The formula used to estimate fecundity per unit of vol- " PV?' where N = the number of oocytes per unit of volume; k = an oocyte size correction coefficient; /3 = an oocyte shape correction coefficient; N„ = the average number of vitellogenic oocytes per unit of area; and V = the average fractional volume of vitel- logenic oocytes per unit of area. The method for estimating the parameter k is given in Emerson et al. (1990) and the parameter k was estimated for six shortspine thornyhead samples. The resulting k values had a small range (1.0088-1.022), and a small standard deviation (0.0066), and a mean k value of 1.017 was used for all samples as a result, ft was calculated by using the method given in Weibel and Gomez (1962). The ft parameter was calculated from one shortspine thornyhead sample (53 oocytes) to be 1.565. Exact volume of sample ovaries was impossible to determine because portions of the ovaries had already been removed for histological study (Pearson and Gunderson, 2003). Volume was estimated by dividing whole ovary weight by an average density of 1.052 g/ mL. This was the average density from six samples (SD = 0.0297) estimated by water displacement in a graduated cylinder. Values for Na and V, were estimated by using a sim- plified Weibel grid for particulate structures (Weibel et al., 1966) instead of a Weibel multipurpose grid. A square containing 13 rows of 13 points was created and printed out on a clear acetate sheet. This overlay was taped to the front of a monitor. A video camera mounted to a stereomicroscope sent the image of the histology section to the computer monitor. The num- ber of vitellogenic oocytes per grid and the number of points falling on vitellogenic oocytes were recorded and used to estimate Na and V,, respectively. The Weibel grid was used at 25x magnification, and 50x magnifica- NV V 0.5 cm Figure 1 Partial cross section of shortspine thornyhead rockfish {Sebastolobus alascanus) ovary showing bands of vitel- logenic (V) and nonvitellogenic (NV) oocytes. tion was used to help distinguish borderline vitellogenic oocytes. A sampling grid was placed under the ovary histologi- cal section. The corner of the Weibel grid was aligned with corners of the sampling grid in order to systemati- cally sample the ovary cross section. Two histological sections were sampled per ovary. The number of Weibel grid counts per ovary depended on the size of the ovary cross section. An average of 55.9 (range: 29-103) Weibel grid counts were taken per ovary. This number was greater than the average num- ber of Weibel grid counts used by Emerson et al. (1990), but the extra counts were made because shortspine thornyhead vitellogenic oocytes develop on peduncles (Erickson and Pikitch, 1993; Pearson and Gunderson 2003) and are distributed in a band around the central part of the ovary (Fig. 1). Because the vitellogenic oo- cytes are not uniformly distributed, the Weibel grid was applied systematically at more points across the entire ovary, and the counts were averaged. Because the whole cross section could not be systematically sampled and averaged, cross sections of larger fish were not used for stereological estimates. Statistical methods Length-fecundity relationships were estimated by using the following equation: Fee = alb, 18 Fishery Bulletin 103(1) Table 1 ""ecundity estimates (number of oocytes) by ovary location and method. Species Stereological method Gravimetric method Sample locatior in ovary Sample location in ovary Mid Posterior Anterior CV Mid Posterior Anterior CV Shortspine 122,180 87,504 111,758 0,166 131,934 110,456 111,425 0.103 Shortspine 313,131 257,378 304,348 0.103 269.453 230,992 257,427 0.078 Shortspine 184,802 199,572 203,014 0.049 Shortspine 474,432 458,877 0.024 Longspine 38,061 26,179 28,424 0.204 38,968 33,207 33,653 0.091 Longspine 36,152 23,127 19,411 0.335 Mean CV 0.147 Mean CV 0.091 where Fee = estimated fecundity; I = fork length; and parameters a and b were estimated by nonlinear regres- sion with SPSS software (version 11.0, SPSS Inc., Chi- cago, ID. Weight-fecundity relationships were estimated by using the following equation Fee = mWsomatic) + bl, where Fee = estimated fecundity; Wtsomallc = somatic weight; and m and 61 were estimated by using linear regression in EXCEL (Microsoft, Redmond, WA). Reduction in variance F tests (Quinn and Deriso, 1999) were used to compare fecundity relationships between areas, studies, and before and during spawn- ing season. the gravimetric versus stereological estimates showed that they follow a 1:1 trend line (Fig. 2). The gravimet- ric method gave a somewhat lower coefficient of varia- tion than the stereological method, based on multiple samples of the same ovaries (Table 1). An F test (Quinn and Deriso, 1999) did not show a significant difference (P=0.84) between the gravimetric (n=16) and stereologi- cal (??=10) methods in the length-fecundity relationships obtained for Alaskan shortspine thornyhead, and the data were therefore combined (Fig. 3). Shortspine thornyhead Shortspine thornyheads from Alaska (/!=26) and the West Coast (n = 30) had similar fecundity at length (Fig. 3). An F test did not indicate fecundity at length for the two areas was significantly different (P=0.53); therefore the data were combined to obtain the relation- ships (Figs. 3 and 4): Fee = 0.0544(Fork Lengthicm )) (r2 = 0.792, 7i=56) Results Ovary location differences We tested for difference in oocyte density between middle, posterior, and anterior sections of six ovary pairs with the stereological method (ovaries from the migratory nucleus to late hydration phase) and did not find a significant difference in ovary location (two-way ANOVA, P=0.148) (Table 1). Stereological method versus gravimetric method The gravimetric method and the stereological method provided similar results. For shortspine thornyhead, the average ratio of gravimetric to stereological estimates for ten pairs of data was 0.993 (Table 2), and a plot of Fee = 0.223(Wtsomatic(g))- 63.079 (r2 = 0.781, n=53). A majority of the shortspine thornyhead fecundity at length data points obtained in this study fell below the regression line reported by Miller (1985) (Fig. 3). The raw data from Miller (1985) were not published; there- fore no statistical test was possible. The data were also separated into months preced- ing the start of spawning and those after the start of spawning (Pearson and Gunderson, 2003) to look for evidence of batch spawning. Shortspine collected between October and November were grouped as speci- mens before the start of spawning. Shortspine collected from April through June in Alaska and from March through May off the West Coast were grouped as speci- mens after the start of spawning. Fish collected after spawning had begun (/;=41) did not show a significant Cooper et al.: Fecundity of Sebastolobus alascanus and Sebastolobus altivelis 19 Table 2 Paired fecundity estimates (number of oocytes) by method and by section of the ovary (middle, posterior, anterior) where oocyte samples were taken. Specimen Shortspine 1 Shortspine 2 Shortspine 3 Shortspine 4 Shortspine 5 Shortspine 6 Position in the ovary Gravimetric Stereological Ratio of gravimetric to stereological Middle Middle Middle Middle Middle Posterior Anterior Middle Posterior Anterior 150,448 195,356 427,717 414,594 131,934 110,456 111,425 269,453 230,992 257,427 184,853 187,037 307,771 561,258 122,180 87,504 111,758 313,131 257,378 304,348 O.K14 1.044 1.390 0.739 1.080 1.262 0.997 0.861 0.897 0.846 Mean ratio 0.993 w 600 i tt) ♦ to .E 500 • w en CD CD & 0 400 ' /^ 1 o yS 3 W g -D 300 ■ O/^ ♦ "~ ra ♦X^ ra « Jr y g 200 ■ ♦/♦ o **▼ £ 100 ■ S^ a> S^ W 0 100 200 300 400 500 Gravimetric fecundity estimates (thousands of eggs) Figure 2 Plot of gravimetric versus stereological fecundity estimates for ten shortspine thornyhead rockfish [Sebastolobus alascanus) data pairs. Line = 1:1 ratio. 2500 2000 ■ £ 1500- 1000 ■ O West Coast x Alaska gravimetric • Alaska stereological Combined data regression - - • Miller (1985) regression (r>=60) 0 *J 500 20 40 60 Fork length (cm) Figure 3 Shortspine thornyhead {Sebastolobus alascanus) fecundity-at-length estimates by location and method, and regression of combined data (our study) and by regression of data from Miller's study (1985). decrease in fecundity at length when compared to fish collected before spawning had begun (n = ll) (F test, P=0.71) (Fig. 5). Longspine thornyhead Longspine thornyhead fecundity data conformed more closely to a linear regression on somatic weight (Fig. 6): Fee = 183.8l(Wt8omatic(g))- 4617 (/-2 = 0.536, n=29) than to a nonlinear regression on length (Fig. 7): Fee = 0.889Q(Fork Length(cm)) (r2=0.442, n=29). A majority of the predicted fecundity values at somatic weight were higher than those derived from Wakefield's (1990) regression line on somatic weight (Fig. 6), but Wakefield's (1990) raw data were not published. Wakefield (1990) estimated spawning to begin in Feb- ruary and created separate fecundity-at-weight relation- ships for fish collected in October-November and in February-March). He noted a decline in fecundity as the spawning season progressed but did not test this fecundity difference for statistical significance. Similar groupings (October-December, n = \l; and February- March, n=ll) in our study did show a statistically sig- nificant difference in fecundity as the spawning sea- son progressed (F test, P=0.004) (Fig. 7); however, the 20 Fishery Bulletin 103(1) 2500 2000 ■ £ 1500 1000 500 Alaska and West Coast combined Linear regression (Alaska and West ♦ Coast combined) 0 2000 4000 6000 Somatic weight (g) 8000 Figure 4 Fecundity at somatic weight for combined Alaska and West Coast shortspine thornyhead rockfish iSebastolobus alascanus). 90 - ♦ West Coast (this study) "w 80 - Wakefield (1990) Oct- Nov (n=11) a 70 - co Wakefield (1990) Feb- Mar (n=22) o 60 - In 50 - CD §f 40- ° 30- CD | 20- ♦ Z 10 - ♦ *i>^^ ♦ 0 100 200 300 400 Somatic weight (g) Figure 6 West Coast longspine thornyhead rockfish iSebas- tolobus altivelis) fecundity data at somatic weight lour study*, compared to fecundity data of Wake- field (1990). 2500 - ♦ Oct - Nov to I 2000 - to 3 Oct - Nov regression Mar - Jun o §. 1 500 - — — - Mar - Jun regression ' to ® 1000 - o ♦ ,' CD E 500 - 3 z ^ 0 20 40 60 80 Fork length (cm) Figure 5 Shortspine thornyhead rockfish (Sebastolobus alascanus) fecundity at length separated by October-November and March-June collection dates. 90 - ♦ Oct - Dec "w 80 i ID Oct - Dec regression ra 70 - cn Feb - Mar ,' o 60 - sz w 50- O. 40 o 30 ® on E =i 10 ■ z — — Feb - Mar regression ,' ♦ / -L ♦ ♦/' 0 J 1 0 10 20 30 40 Fork length (cm) Figure 7 Longspine thornyhead rockfish {Sebastolo- bus altivelis) fecundity at length separated by October-December and February-March collec- tion dates. regression lines intersected, and the February-March group was not lower than the October-December group. The February-March group did have lower fecundity than the October-December group for lengths smaller than 27 cm; however the sample size was very small. No significant difference existed between the two groups when the single, large fecundity observation late in the spawning season was ignored (P=0.34). Discussion Emerson et al. (1990) cited the ability to distinguish borderline vitellogenic oocytes from nonvitellogenic oocytes as an advantage of the stereological method, and this was a clear benefit in our study. The stereological method allowed us to differentiate between vitellogenic and nonvitellogenic oocytes at an earlier stage of ovary development than was possible with the gravimetric method. However, the use of ovaries in earlier stages of development increases the potential magnitude of fecun- dity overestimates due to atresia. Atresia, or the resorp- tion of oocytes, is a potential source of error for fecundity estimates (Hunter et al., 1992). Although atretic oocytes can be identified with the stereological method, oocytes that are destined for atresia will be counted, causing fecundity to be overestimated. The amount of atresia will determine the magnitude of this overestimate. Samples Cooper et al.: Fecundity of Sebastolobus alascanus and Sebastolobus altivelis 21 collected at later ovarian development stages would avoid this potential error (Tuene et al., 2002). Because of a nonrandom distribution of vitellogenic and nonvitellogenic oocytes in the ovary, it was neces- sary to average Weibel grid counts over an entire ovary cross section. Larger ovaries that did not fit on a single slide could not be used, so that fecundity of larger fish had to be determined with the gravimetric method. This was a major limitation because few fish greater than 60 cm had ovaries small enough to be suitable for the ste- reological method. This limitation, however, might not apply to fish species with vitellogenic oocytes randomly distributed throughout the ovary. The number of Weibel grid counts required was larger in our study than in Emerson et al. (1990), and the extra counts increased the amount of time involved with com- putation of fecundity estimates. In addition to the time required to prepare histological sections, the time to obtain stereological estimates took approximately twice as long as those obtained with the gravimetric method. Our estimates of shortspine thornyhead fecundity at length (Fig. 3) appeared lower than the regression pub- lished by Miller (1985), but our longspine thornyhead fecundity estimates were higher than those published by Wakefield (1990) (Fig. 6). Several potential explanations exist for the differences. Temporal or geographic differ- ences in fecundity could exist. Samples from different decades were used in the two studies, and Wakefield (1990) used longspine samples taken from off Point Sur, California, whereas we used samples collected off Oregon and Washington. However, the differences may also be explained by methodological differences between authors, including different criteria to include oocytes in fecundity estimates, and differences in the ovarian development of samples. Relatively small sample sizes from our study and from Wakefield (1990) may add un- certainty to these fecundity estimates. The length range of samples could also affect comparisons for shortspine thornyhead fecundity. The fecundity estimates from Miller (1985) did not include any fish greater than 60 cm, whereas we used fish approaching 80 cm. Wakefield (1990) grouped fecundity data by date, that is to say before the start of spawning and after the start of spawning. His data indicated a decline in fecundity after spawning begins, which he attributed to batch spawning. Similar temporal groupings in our study did not necessarily show a decrease in fecundity that was indicative of batch spawning in longspine or shortspine thornyhead. An important caveat regarding these comparisons is that the combination of small sample sizes and high variability in fecundity at length would cause only large differences in fecundity to be detected. However, the sample sizes used for compari- son before and during spawning season (shortspine thornyhead n=ll, 41) (longspine thornyhead n = 17,ll) were close to the sample sizes Wakefield (1990) used as evidence for batch spawning (rc=ll,22). Larger sample sizes for both species would help answer the question of whether these are batch-spawning species. Pearson and Gunderson (2003) did not find any hydrated oocytes or postovulatory follicles co-occurring with vitellogenic oocytes in histological sections of either species used in our study. They concluded that batch spawning does not occur from off Northern California to Alaska for short- spine thornyhead, and from off Northern California to Washington for longspine thornyhead, and the results of the present study support this conclusion. Ovaries are often opportunistically collected dur- ing commercial fishing seasons or scheduled fisheries surveys and may not provide oocyte samples from the optimum time of year for estimating fecundity with gravimetric techniques. Nevertheless, the stereological technique enabled us to make fecundity estimates for a greater number of the available samples. The technique could be used in similar instances where the logistics of sampling require collections to be made earlier than the optimal date for gravimetric estimates. Acknowledgments Dave Douglas of the Oregon Department of Fish and Wildlife collected many samples, as did numerous NMFS RACE and REFM division scientists and the following NMFS observers: C. Colway, A. Hayward, W. Mitchell, E. White, N. Spang, K. Redslob, M. Waters, and D. Tran. We thank Frank Morado, Lisa Appesland, and Dan Nichol of the NMFS Alaska Fisheries Science Center (AFSC) for use of equipment and equipment instruc- tion. We also thank Marcus Duke of the UW SAFS for creating a Weibel grid. Jim Ianelli and Rebecca Reuter of the NMFS Alaska Fisheries Science Center provided quantitative assistance. Cathy Schwartz of the UW SAFS assisted with the figures and tables. We thank two anonymous reviewers for providing useful comments. This research was supported by the Joint Institute for the Study of the Atmosphere and Ocean (JISAO) under NOAA cooperative agreement no. NA17RJ1232. Literature cited Best, E. A. 1964. Spawning of longspine channel rockfish, Sebastolo- bus altivelis Gilbert. Calif. Fish Game 50:265-267. Emerson L. S., M. G. Walker, and P. R. Witthames. 1990. A stereological method for estimating fish fecundity. J. Fish Biol. 36:721-730. Erickson, D. L., and E. K. Pikitch. 1993. A histological description of shortspine thornyhead, Sebastolobus alascanus, ovaries: structures associated with the production of gelatinous egg masses. Environ. Biol. Fishes 36:273-282. Gaichas, S., and J. N. Ianelli. 2003. Assessment of thornyheads iSebastolobus spp.) in the Gulf of Alaska. In Stock assessment and fishery evaluation report for the groundfish resources of the Gulf of Alaska, p. 659-698. North Pacific Fishery Management Council, Anchorage, AK. Hoff, G. R., and L. L. Britt. 2003. The 2002 eastern Bering Sea upper continental slope 22 Fishery Bulletin 103(1) survey of ground fish and invertebrate resources. NOAA Tech. Memo. NMFS-AFSC-141, 261 p. Hunter, J. R., B. J. Macewicz, N. C. Lo, and C. A. Kimbrell. 1992. Fecundity, spawning, and maturity of female Dover sole, Microstomias pacificus, with an evaluation of assumptions and precision. Fish. Bull. 90:101-128. Macer, C. T. 1974. The reproductive biology of the horse mackerel Trachurus trachurus (L.) in the North Sea and English Channel. J. Fish Biol. 6:415-438. Miller, P. P. 1985. Life history study of the shortspine thornyhead, Sebastolobus alascanus, at Cape Ommaney, southeastern Alaska. M.S. thesis, 76 p. Univ. of Alaska, Juneau, AK. Orr, J. W., M. A. Brown, and D. C. Baker. 2000. Guide to rockfishes (Seorpaenidae) of the genera Sebastes, Sebastolobus, and Adelosebastes of the North- east Pacific Ocean, 2nd ed. NOAA Tech. Memo. NMFS- AFSC-117, 48 p. Pearcy, W. G. 1962. Egg masses and early developmental stages of the scorpaenid fish, Sebastolobus. J. Fish. Res. Board Can. 19:1169-1173. Pearson, K. E„ and D. R. Gunderson. 2003. Reproductive biology and ecology of shortspine thornyhead rockfish, Sebastolobus alascanus, and longspine thornyhead rockfish, S. altivelis, from the northeastern Pacific Ocean. Environ. Biol. Fishes 67:11-136. Piner, K., and R. Methot. 2001. Stock assessment and fishery evaluation report of shortspine thornyhead off the Pacific West Coast of the United States 2001. In Status of the Pacific coast groundfish fishery through 2001 and acceptable biological catches for 2002. Pacific Fishery Manage- ment Council. Portland, OR. Quinn, T J., and R. B. Deriso. 1999. Quantitative fish dynamics, 542 p. Oxford Univ. Press, New York, NY. Tuene. S., A. C. Gundersen. W. Emblem, I. Fossen, J. Boje, P. Steingrund, and L. H. Ofstad. 2002. Maturation and occurrence of atresia in oocytes of Greenland halibut tReinhardtius hippoglossoides W.) in the waters of East Greenland, Faroe Islands and Hatton Bank. In Reproduction of West-Nordic Greenland hali- but: studies reflecting on maturity, fecundity, spawning, and TEP (A. C. Gundersen, ed.), p. 39-69. TemaNord 2002:519. Wakefield, W. W. 1990. Patterns in the distribution of demersal fishes on the upper continental slope off central California with studies on the role of ontogenetic vertical migration in particle flux. Ph.D. diss., 281 p. Univ. California, San Diego, CA. Wakefield, W. W., and K. L. Smith Jr. 1990. Ontogenetic vertical migration in Sebastolobus altivelis as a mechanism for transport of particulate organic matter at continental slope depths. Limnol. Oceanogr. 35:1314-1328. Weibel, E. R„ and D. M. Gomez. 1962. A principle for counting tissue structures on random sections. J. Appl. Physiol. 17:343-348. Weibel, E. R., G. S. Kistler, and W. Scherle. 1966. Practical stereological methods for morphometric cytology. J. Cell Biol. 30:22-38. 23 Abstract — Body size at gonadal matu- rity is described for females of the slip- per lobster (Scyllarides squammosus) (Scyllaridae) and the endemic Hawaiian spiny lobster (Panulirus marginatus) (Palinuridae) based on microscopic ex- amination of histological preparations of ovaries. These data are used to validate several morphological metrics (relative exopodite length, ovigerous condition) of functional sexual maturity. Relative exopodite length ("pleopod length"! pro- duced consistent estimates of size at maturity when evaluated with a newly derived statistical application for esti- mating size at the morphometric matu- ration point IMMP) for the population, identified as the midpoint of a sigmoid function spanning the estimated bound- aries of overlap between the largest immature and smallest adult animals. Estimates of the MMP were related to matched (same-year) characterizations of sexual maturity based on ovigerous condition — a more conventional measure of functional maturity previously used to characterize maturity for the two lobster species. Both measures of functional maturity were similar for the respective species and were within 5% and 2% of one another for slipper and spiny lob- ster, respectively. The precision observed for two shipboard collection series of pleopod-length data indicated that the method is reliable and not dependent on specialized expertise. Precision of matu- rity estimates for S. squammosus with the pleopod-length metric was similar to that for P. marginatus with any of the other measures (including conven- tional evidence of ovigerous condition) and greatly exceeded the precision of estimates for S. squammosus based on ovigerous condition alone. The two measures of functional maturity aver- aged within 8f» of the estimated size at gonadal maturity for the respective spe- cies. Appendage-to-body size proportions, such as the pleopod length metric, hold great promise, particularly for species of slipper lobsters like S. squammosus for which there exist no other reliable conventional morphological measures of sexual maturity. Morphometric propor- tions also should be included among the factors evaluated when assessing size at sexual maturity in spiny lobster stocks; previously, these proportions have been obtained routinely only for brachyuran crabs within the Crustacea. Manuscript submitted 2 September 2003 to the Scientific Editor's Office. Manuscript approved for publication 26 August 2004 by the Scientific Editor. Fish. Bull. 103:23-33 (20051. Relative pleopod length as an indicator of size at sexual maturity in slipper (Scyllarides squammosus) and spiny Hawaiian (Panulirus marginatus) lobsters Edward E. DeMartini Marti L. McCracken Robert B. Moffitt Jerry A. Wetherall Pacific Islands Fisheries Science Center National Marine Fisheries Service, NOAA 2570 Dole Street Honolulu, Hawaii 96822-2396 E mail address (for E. E. DeMartini) edward demartinianooa gov Estimates of body size and age at sexual maturity provide key informa- tion for stock assessments and hence for managing sustainable fisheries. Characterizations of size at matu- rity are relatively straightforward in lobsters and most other crustaceans. One presently accepted standard is to regress percentage mature against classes of some body size metric and to fit a logistic model to predict the size class in which 50% of the population is mature. A necessary prerequisite is accurate data on the maturation state of individuals. In spiny lobsters of the family Palinuridae, female matura- tion is usually deduced from "berried" (ovigerous) condition (Groeneveld and Melville-Smith, 1994), the presence of external morphological indicators such as changes in the number of pleo- pod setae (Gregory and Labisky, 1981; Montgomery, 1992), relative lengths of abdominal and thoracic segments (Jayakody, 1989), or proportional lengths of segments of walking or egg- bearing appendages at the pubertal molt (George and Morgan, 1979; Grey, 1979; Juinio, 1987; Plaut, 1993; Evans et al., 1995; Hogarth and Barratt, 1996; Minagawa and Higuchi, 1997). A major complication arises, however, when the percentage mature within size classes cannot be accurately described. Such is the case for Scyl- larides squammosus, a species of slip- per lobster (family Scyllaridae) that prior to closure of the fishery in 2000 had become an increasingly important target of the Northwestern Hawaiian Island (NWHI) commercial trap fish- ery. In S. squammosus, unberried but mature females are indistinguishable, based on gross external morphology, from immature females. In this spe- cies, the additional variance intro- duced by combining falsely classified "immature" with truly immature females inflates requisite sample sizes enough (given the sampling effort fea- sible on annual research surveys) to prevent characterization of possible changes in size at maturity with data pooled from less than several surveys. Combining unberried adults with true immature individuals also introduces an overestimation bias (DeMartini et al., 2003). To date only one study has provided a description of the use of a morpho- logical measure of maturity in a slip- per lobster (Hossain, 1978). Morphol- ogy-based maturity measures have been described for numerous spiny lobsters of the genus Panulirus, but such measures for the endemic Ha- waiian spiny lobster (Panulirus mar- ginatus) have not been fully described (Prescott, 1984). Our objectives are to describe the development and use of an external body metric for accurately and pre- cisely characterizing body size at mor- phological (functional) sexual maturi- ty in female Scyllarides squammosus. We likewise use this external metric 24 Fishery Bulletin 103(1) to estimate size at maturity of females of the Hawai- ian spiny lobster, for which functional maturity can be accurately described by using a combination of other, more apparent external features. We also estimate body size at gonadal maturity by microscopic examination of histological preparations of ovaries of each species and use these results to validate the functional maturity characterizations. We contrast the benefits of the dif- ferent approaches for estimating functional maturity in these two lobsters and discuss the potential importance of applying efficient measures of maturation for manag- ing the NWHI lobster fishery. Materials and methods Specimen collection A research vessel was used to set and retrieve lobster traps. All specimens of spiny lobster used in this study were taken from Necker Bank surrounding Necker Island (23°34'N, 164°42'W), NWHI. All the slipper lobsters used were taken from Maro Bank, located about 600 km to the northwest of Necker at 25°25'N, 170°35'W. Lobsters were caught from bank terraces at median depths of 15 fm (slipper lobsters, Maro) and 17 fm (spiny lobsters, Necker) with molded plastic (Fathoms Plus", San Diego, CA) traps baited with 1 kg of mackerel (Scomber japoni- cus) and left for a standard (overnight) soak. Shipboard processing All specimens were processed alive within minutes of trap retrieval. Tail width (TW), as defined for slip- per lobster by DeMartini and Williams (2001) and for Hawaiian spiny lobster by DeMartini et al. (2003), was measured with 0.1 mm accuracy. Berried females were scored by egg-development stage with a gross visual proxy (brooded eggs noted as either orange or brown in color to the unaided eye). Female spiny lobsters were scored by the presence or absence and by condi- tion ("smooth"=unused, "rough"=partly used) of sper- matophoric (sperm) mass (Matthews, 1951; Berry and Heydorn, 1970) on the sternum. Female S. squammosus in almost all cases lack a sperm mass and the presence- absence of this feature provides no useful information. In 1998-2000, ovaries were dissected from a maximum of two living specimens for each 1-mm TW class of the two species and fixed in 10% (sea water buffered) for- malin for subsequent histological analyses. Egg-bear- ing "tails" (abdominal segments) were flash-frozen at -20 C. During 1997-99, pleopods of each species were mea- sured aboard ship to evaluate measurement accuracy under field conditions. Maxima of 10 live individuals per 1-mm TW class of each species were measured as described below. Two independent measurements of each specimen were made by each of two measurers (one inexperienced and one experienced). In 2000-01, pleopods for a larger series of morphometries were simi- larly measured aboard ship to evaluate production- scale numbers (500-1000 specimens per species on each cruise) based on a single measurement per specimen taken by one measurer. Laboratory measurements Beginning with specimens collected in 2000, the lengths of exopodites on first pleopods were measured for a representative sample of berried and unberried tails of each species, after the tails were thawed overnight in a refrigerator at 3 C. Preliminary observations indicated that the first pleopod was disproportionately large in berried females; measurements of the first pleopod of all (berried and unberried) females moreover were the most precise, i.e. the measurements were more likely to be obtained again — probably because the first pleopod was the easiest to measure. The straightline distance between base and tip of exopodite on the first pleopod (exopodite length=EL) was measured with dial calipers to 0.01 mm. An analogous measurement of exopodite width (EW) was taken perpendicular to the EL axis at the structure's widest point. The left exopodite in ven- tral aspect (Fig. 1) was routinely measured because the ventral aspect was easier to measure for live animals aboard ship. Measurements of the right exopodite (of the same specimen) in dorsal aspect were taken for a range of body sizes to evaluate the possible influence of aspect (dorsal vs. ventral) or body side (left vs. right) on the measurement that was taken. Replicate measure- ments (independent, with calipers reset to zero between measurements) were used to assess inter-measurer and inherent measurement error. Formalized ovaries were weighed (blotted damp-dry) to the nearest 0.01 g after fixation for at least a month. Histological validation Fixed ovary specimens of each species were dehydrated, imbedded, and sectioned by using standard techniques, and were stained with hematoxylin and counter-stained with eosin to differentiate protein and yolk materials within oocytes. Histological slides were viewed under a compound microscope at 150x magnification. For each specimen, the diameters (average of major and minor axis) were measured for 10 oocytes (randomly chosen) within the largest size class of oocytes present. The median diameter was used to characterize oocyte size for that specimen; the median diameter based on 10 measurements yielded CVs (100% x standard error/mean) <10% (DeMartini et al., 2003). Developmental staging followed Minagawa (1997) and Minagawa and Sano (1997): females were scored as mature 1) if unberried in developing or ripe ovarian stages II and III, respec- tively; 2) if berried in ripe and redeveloping stages IV and V, respectively; or 3) if recently spent (stage VI) with heavily setose pleopods (P. marginatus only). Inactive females in stage I were scored as immature. A gonad index, calculated as GI = (OWx 105/TW3), where OW = ovary weight in g, was used to complement his- DeMartini et al.: Validated morphological metric for lobster size at maturity 25 Female, Scyllandes squammosus, 56.0 mm TW 29.8 mm exopodite length 5mm exopodite B 5mm Female, Panulirus marginalus, 51 .2 mm TAW 36.1 mm exopodite length endopodite egg-bearing setae exopodite endopodite egg-bearing setae Figure 1 Schematic diagram of the left first exopodite (ventral aspect! of (A) slipper lobster (Scyllarides squammosus) and (B) the Hawaiian spiny lobster I Panulirus marginatus), showing axis of measurement. TW = tail width. tological scores in assessing gonadal maturity (Minagawa and Sano, 1997). Gonadal maturity was used as a means of validating, as well as ref- erencing, estimates of size at functional maturity (Ennis, 1984). Statistical analyses Data for EL and EW (as response variables) and TW (regressor) for the same specimen were first plotted for all specimens of each species. Prelimi- nary evaluations of these data (both raw and log- transformed) with least-squares linear regression (REG procedure; SAS vers. 8, SAS Institute, Inc., Cary, NC) indicated allometric relationships for which double-log functions provided approximate fits. Identification of join points by iteration based on minimizing the total residual sums of squares of pairs of joined regression equations (Somerton, 1980), however, resulted in linear spline fits that, although significant, had obviously nonrandom residuals. Simple linear fits with log-log plots, how- ever, were useful for selecting the most appropriate metric: the regressions of EW on TW, qualitatively similar to those for EL regressed on TW, had con- sistently lower r2 values, likely because pleopod width was more difficult to measure than pleopod length. The EL metric was therefore chosen for all further analyses. Because lobsters, like most biological populations, are composed of individuals that differ in the size at which first maturity occurs, we fitted a curve to the EL-TW re- lation that included a sigmoid segment bridging the re- gion between the estimated sizes of the smallest adults (0O) and the largest immature individuals (0j) (Fig. 2). The curve was fitted by using iterative reweighted least CO Q. >. "O o n o "5 E o "ro o lab; RCB ANOVA; both P=0.001) for slipper lobster and spiny lobster (Table 1). For each species, however, the mean difference between venues was trivial (0.2-0.4 mm or 0.6-1.4%). Differences between ship and laboratory were detectable despite the consistently lower precision pro- vided by shipboard measurements (shipboard CVs were 47% and 39% larger for slipper and spiny lobster, respec- tively; RCB ANOVA: both P<0.001; Table 1). Absolute differences between shipboard and lab CVs were small for the respective species (0.2% and 0.7%; Table 1). Measurer effects An extensive series of shipboard inter- measurer comparisons between pleopod length mea- surements taken by one experienced (A) and a second inexperienced (B) measurer indicated trivial systematic differences between measurers (0.2%; RCB ANOVA; P=0.25). Precision also was unaffected by measurer (P=0.31; Table 1). Standardized metric It follows from the above that the best measure available for use was the length (in ven- tral aspect) of the left first exopodite. This metric was used in all quantitative comparisons among maturity assessment methods and is recommended for future applications with these species. Estimated sizes at functional maturity Slipper lobster Pleopod-to-TW relations for S. squammo- sus did not differ meaningfully between 2000 and 2001 (ANCOVA; accept HQ: slopes equal, P=0.11; intercepts only 0.5% different) and both years' data were pooled for further analyses. The estimated MMP (95% CI) for the TW at which 50% of the female S. squami7iosus exhibit a disproportionately long first left exopodite was 47.6 mm (45.1-49.4 mm; Fig. 3). Estimated median body size at functional maturity based on presence or absence of ber- ried eggs, using the same series of 2000-01 specimens, was 55.5 (52.7-58.3) mm TW (Fig. 4). DeMartini et al.: Validated morphological metric for lobster size at maturity 27 Table 1 Results of tests of potential effects of various criterion variables on the accuracy (bias of delta-barsl and precision (CVs of deltas) for measured lengths of first pleopod exopodites for slipper lobster (Scyllarides squammosus) and Hawaiian spiny lobster (Panu- lirus marginatus) caught from Necker Bank, Hawaii. Delta-bar = mean paired-difference; samples sizes are n paired observa- tions. Variable Slipper lobster Body side (left vs. right) Measurement aspect (ventral vs. dorsal) Measurement venue (shipboard vs. lab) Measurement venue (shipboard vs. lab) Measurer (A vs. B) Measurer (A vs. B) Spiny lobster Body side (left vs. right) Measurement aspect (ventral vs. dorsal) Measurement venue (shipboard vs. lab) Measurement venue (shipboard vs. lab) Measurer (A vs. B) Measurer (A vs. B) Criterion Test statistic Delta-bar accuracy paired ?=-4.0 0.9 mm accuracy RCB Anova Pi, 62=202.7 0.9 mm accuracy RCB Anova Fl 62=23.6 0.4 mm precision RCB Anova *Y 62=21-6 0.7% accuracy RCB Anova ^1.62 = 213 0.3 mm precision RCB Anova Fi, 62=1-54 0.2% accuracy paired t=-5.7 0.7 mm accuracy RCB Anova Fi. 32=31-7 0.7 mm accuracy RCB Anova F187=11.62 0.2 mm precision RCB Anova Fi, 87=11-74 0.4% accuracy RCB Anova Fi. 87=1.37 <0.1 mm precision RCB Anova F!, 87=l-04 <0.2 % 0.001 1)1)01 0.001 0.001 0.001 0.22 0.001 0.001 0.001 0.001 0.25 0.31 74 63 63 63 63 63 135 33 88 Spiny lobster Year effects on pleopod-to-TW relations for P. marginatus were likewise insignificant ( ANCOVA; accept H0: slopes equal, P>0.67; intercepts only 0.2% dif- ferent) and data for both years were pooled for further analyses. The MMP for the TW at which 50% of the P. marginatus females exhibit a disproportionately long pleopod was 36.4 mm (34.1-38.0 mm; Fig. 5). Figure 6 illustrates the corresponding estimate of median size at functional maturity, 35.4 (33.7-37.1) mm TW, based on the combined criteria of sperm mass and berried egg presence, for P. marginatus. Estimated sizes at physiological maturity Gonadal maturity determined from microscopic staging of histological ovary preparations indicated matura- tion stages ranging from oogonial to fully vitellogenic (Table 2; Minagawa and Sano, 1997) for the females of each species. For both species, gonad indices (GIs) and median oocyte diameters generally increased over the cycle of development even though berried specimens exhibited lower GIs and oocyte sizes than unberried adults of the respective species (Table 2). The ovaries of mature females contained a preponderance of fully yolked oocytes whose average minimum diameter (fol- lowing dehydration and staining) was 0.24 mm and 0.30 mm for S. squammosus and P. marginatus, respec- tively. The maximum observed diameter of fully yolked oocytes was 0.60 mm (in S. squammosus) and 0.58 mm {P. marginatus). The proportions of observed immature individuals ranged from 32% to 38% of total female specimens (depending on species) and were sufficient to construct logistic curves relating percentage gonadal maturity to body size for each species. Estimated median TWs at gonadal maturity were 51.1 (48.6-53.5) mm and 40.5 (37.9-43.1) mm TW for S. squammosus (Fig. 4) and P. marginatus (Fig. 6), respectively. 28 Fishery Bulletin 103(1) 50 Seyllarides squammosus • E E Ol c 0> "O o Q. O CD CL e CD _l 40 30 20 / oo„° • 1 1 59 > ([e(l+e,]/2) 0 2 0 o 29 < ([e(1+e,]/2) 0 30 40 50 60 70 80 90 Tall width (mm) Figure 3 Scatterplots of the relation between exopodite length and tail width for slipper lobster (Seyllarides squammosus). The moi phometric maturation point (MMP; indicated by the verti- cal line) represents the allometric threshold coincident with sexual maturity ([0Q+0J1/2). The outlier indicated by an arrow was not used in estimating the MMP. Table 2 Stages of ovarian development in 197 slipper lobster {Seyllarides squammosus ginatus) caught from Necker Bank, Hawaii. There were no stage-VI S. squam ) and 122 Hawaiian spiny lobster (Panulirus mar- 710SUS. Ovarian stage Characteristics of ovaries and oocytes Gonad index mean ±SD (range) n Most advanced oocyte substage (median diameter) Slipper lobster oogonia and previtellogenic oocytes conspicuous; 0.43+0.25 60 preyolk platelet I (inactive) ovary white (0.02-1.08) (0.18 mm) II— III (developing and ripe) unberried; developing moderately to fully vitellogenic oocytes; ovary pale orange to orange 1.63 ±1.45 (0.25-5.47) 75 prematuration or maturation (0.28 mm) IV-V (ripe and redeveloping) berried; developed fully yolked oocytes; ovary dark orange 1.12 ±0.66 (0.15-3.30) 62 maturation (0.26 mm) Spiny lobster I (inactive) oogonia and previtellogenic oocytes; ovary white 0.85 ±0.67 (0.16-2.59) 30 preyolk platelet (0.12) II— III (developing and ripe) unberried; developing moderately to fully vitellogenic oocytes; ovary pale orange to orange 14.03 ±4.46 (3.32-22.69) 42 prematuration or maturation (0.49) IV-V (ripe and redeveloping) berried; developed fully yolked oocytes; ovary dark orange 5.84 ±4.31 (0.56-17.06) 47 maturation (0.30) VI (spent) residual unspawned mature oocytes; ovulation traces 14.85 ±6.38 (7.5-18.9) 3 yolk platelet but atretic (0.48) DeMartini et al.: Validated morphological metric for lobster size at maturity 29 Discussion Properties of the EL-TW model In order to determine morphometric maturity, we first attempted to use a method developed by Watters and Hobday (1998). With this method splines were used to model the relationship between the morphometric char- acter and body size; then the morphometric size at which the second derivative of the fitted curve is maximal is computed. At first this technique is alluring in that it makes no allometric or other assumption as to the shape of the relationship between the morphometric character and body size. It instead assumes that maturation cor- responds to the maximum of the second derivative. This assumption is likely invalid even if we assume that the relationship between the morphometric character and body size changes abruptly at maturation for each indi- vidual (as at the pubertal molt in crustaceans) because individuals in the population mature at different sizes. When we applied the Watters and Hobday method, the resulting body size estimate appeared to character- ize the minimum, not the median, size at attainment of sexual maturity in the population and was clearly inappropriate for our needs. Our method generated fitted splines that were comfortingly similar in shape to the parametric logistic (sigmoidal function) models that we used to estimate maturation with berried and histological criteria. The magnitude of the difference between the sizes at maturity estimated by our and the Watters and Hobday (1998) model should vary in proportion to the magni- tude of the difference between the minimum (0O) and median ([0o+0j]/2) body sizes at maturity and therefore be case-dependent. In our slipper lobster case, the 80 and 0l estimates differed by about 6.6 mm; hence, the two model estimates differed by about 6.6/2 = 3.3 mm or approximately 7% of the [{6^+6-^)12} median. Because other cases certainly include those in which immature and adult sizes overlap even more greatly, we suggest that our more general and accurate model be adopted. Functional versus physiological measures of maturity Morphological features can provide adequate if imperfect measures of functional sexual maturity, as can physi- ological evidence for gonadal maturity (Ennis, 1984). Morphological features such as ovigerous condition can underestimate the incidence of mature individuals, but the degree to which they do so depends on numerous fac- tors including species and population. Physiological met- rics in some cases can provide more accurate estimators of both body size and age at maturity because they reveal the reproductive readiness of individuals at the time of collection. Individual body size and age at maturity can be decoupled from functional maturity metrics in Crustacea, however. For example, some crustaceans like majid crabs exhibit determinate growth following a ter- minal, pubertal molt (Hartnoll, 1982). For such species, size at attainment of sexual maturity is synonymous 100 50 Tail width (TW, mm) Figure 4 Scatterplots and fitted curves of the relations between body size (tail width, TW) and percent sexual maturity based on functional maturity gauged by presence-absence of berried condition (dotted curve), overlaid on gonadal maturation gauged by microscopic examination of ovaries (dark-line curve); the pleopod length-based morphometric maturation point (MMP) estimate of size at functional maturity is indicated by the large circle with cross-hairs (©), for slipper lobster iScyllarides squammosus). A 3-parameter logistic equation was necessary to fit the dotted curve; a 2-parameter logistic was sufficient to fit the dark-line curve (see text). with the median body size of adults. These two attributes are not synonymous for lobsters with indeterminate growth. It is further obvious that the pleopods and other allometric body parts of Crustacea like lobsters reflect an array of gonadal maturities ranging from developing immature to fully mature, which can be problematic because some or many females might abort and resorb developing gonadal eggs after the pubertal molt (Aiken and Waddy, 1980) or may not become inseminated (Hey- dorn, 1969). By attributing maturity to specimens that either have not matured physiologically or that will not reproduce although capable of doing so, appendage-to- body proportions can underestimate the age at maturity in Crustacea. The degree of underestimation should be proportional to the incidence of gonadal resorption during the intermolt period following the pubertal molt, 30 Fishery Bulletin 103(1) Panulirus marginatus 10 20 30 40 50 60 70 80 90 Tail width (mm) Figure 5 Scatterplots of the relation between pleopod length and tail width for the Hawaiian spiny lobster (Panulirus marginatus). The morphometric maturation point (MMP: indicated by a vertical line) represents the allometric threshold coincident with sexual maturity ([Oo+fJ 12). as well as the duration of the intermolt. These specific topics deserve future study. The above caveats notwithstanding, it is helpful to compare estimates of body sizes at sexual maturity based on various morphological and physiological evi- dence and to ascertain the degree of agreement among the estimates (Fernandez-Vergaz et al., 2000). The es- timate of MMP (47.6 mm) indicated by the pleopod length-to-TW relation for S. squammosus, for example, was about 16% smaller than the median size at matu- rity (55.5 [±1.35 SE] mm) estimated by using simple presence-absence of berried eggs for the same series of specimens. The latter estimate, however, is imprecise and an overestimate. The long-term mean TW at 50% maturity based on berried condition for the period from 1986 to 2001, indistinguishable among component years, was 50.0 ±0.83 mm, more precise than the single-year estimate although still biased high (DeMartini et al., 2002). If this 50.0 value is used for reference, the pleo- pod length-based estimate of the MMP falls within <5% of the long-term mean. For P. marginatus, the analogous MMP = 36.4 mm value was within 3% of the estimated median size at maturity (35.4 mm) based on the com- bined criteria of berried eggs and sperm mass presence. All the various estimates of functional maturity for the two species were within 2.0-12.6% (mean=7.9%) of the best respective estimate of gonadal maturity. These close similarities, despite the inherent biases of the two methods, indicate that maturity metrics such as relative pleopod length can provide highly satisfactory proxies of true functional maturity that are closely related to gonadal maturity in certain cases. Pleopod length as a maturity metric In some Crustacea (once again, not lobsters, as far as is known), allometries are not fixed at the pubertal molt; and, in a minority of these, allometric growth is seasonally cyclic and allometries disappear when mature instars molt during nonreproductive periods (Hartnoll, 1974. 1982). And body proportions may not be strong predictors of sexual maturity for clawed lobsters (Comeau and Savoie, 2002). In many, if not most, deca- pods such as spiny lobsters (e.g., George and Morgan, 1979; Groeneveld and Melville-Smith, 1994), however, relative appendage-to-body sizes, as well as obvious morphological criteria such as the presence of berried eggs and a sperm mass, indicate functional sexual matu- rity. Body part allometries in some cases can be better predictors of maturity than more obvious characters like berried eggs. An incomplete measure such as per- centage berried, exemplified by the slipper lobster (S. squammosus) in the present study, can falsely fail to detect reproductively inactive adult females. Appendage- to-body size proportions thus have one major advantage over other morphometries in that they permit reproduc- tively inactive adult females to be correctly classified as mature. This advantage is relatively unimportant in other species like P. marginatus for which additional gross morphological indicators such as the presence- absence of a sperm mass complement the information provided by berried condition. Even so, proportional appendage lengths can be used in such cases as another fairly inexpensive and independent measure that could contribute to a multivariate assessment of maturity. DeMartim et al.: Validated morphological metric for lobster size at maturity 31 100 Panulirus marginatus curve: histological criteria Px= 100/(1 +exp-(-9.415+0.233TW)) l2= 0.907 curve: berried + sperm mass criteria Px = 1 00 / (1 + exp-(-1 8.836+0.532 TW)) ^=0.895 50 60 Tail width (TW, mm) 80 Figure 6 Scatterplots and fitted curves for the relations between body size (tail width, TW) and percent sexual maturity based on functional maturity gauged by presence-absence of sperm mass and berried condition (dotted curve), overlaid on gonadal maturation gauged by microscopic examination of ovaries (dark-line curve); the pleopod length-based morphometric maturation point (MMP) estimate of size at functional maturity is indicated by the large circle with cross-hairs (©), for Hawaiian spiny lobster (Panulirus marginatus). Two-parameter logistic equations were sufficient to fit both the dotted and dark-line curves. truly immature from mature, but reproductively inac- tive, females generates an inflated "immature" class, and the estimates of median size at sexual maturity thus obtained with logistic equation fits are biased high. Variances of median-size estimates based on sample sizes available on single research surveys are often so large that 3-parameter logistic applications (necessary to scale maturity to 100%) fail to converge, and reliable individual-year estimates are impossible (DeMartini et al., 2002). Unfortunately, the temporal dynamics of targeting species by fishermen in the NWHI trap fishery and the rapid phenotypic responses in fecundity and maturation size to harvesting, fluctuating natural productivity, and changing population densities that have been observed in P. marginatus (DeMartini et al., 2003), require that size at maturity be re-estimated at short (one-to-several-year) intervals for this species at least and possibly for S. squammosus as well. The accurate and precise estimates of median body size at sexual maturity made possible by using the pleopod length metric enable such yearly re-evaluations for S. squammosus and provide a second reliable and independent estimator for P. marginatus. Our success- ful applications for a scyllarid as well as a palinurid, together with prior observations for numerous other spiny lobster species, indicate that easily measured appendage length-to-body size relations are generally suitable for assessing functional sexual maturity in lobsters and other decapods. We recommend that these relations be explored for other commercially exploited crustacean stocks and wherever possible routinely ap- plied to provide cost-effective and timely information on size at maturity for stock assessments. Managers responsible for the assessment of lobster and other crus- tacean stocks will then have a more complete toolbox of methods generally available for assessing the size at maturity and harvestability of stocks, particularly for species like S. squammosus in which conventional morphological measures are inadequate. Acknowledgments We thank D. Yamaguchi for assistance with Figure 1 and G. DiNardo and J. Polovina for constructive criti- cisms of the manuscript. unconstrained by a conspicuous but perhaps inaccurate feature like berried condition. Literature cited Management implications Estimates of body size at sexual maturity can provide key information to various stock assessment models, but only if the estimates are accurate and sufficiently pre- cise. For the slipper lobster (S. squammosus), DeMartini et al. (2002) have shown that estimates made by using percent berried as the lone maturity criterion, the only morphological metric previously available, are both inaccurate and imprecise. The inability to distinguish Aiken, D. E., and S. L. Waddy. 1980. Reproductive biology. /« The biology and manage- ment of lobsters, vol. I, physiology and behavior (J. S. Cobb and B. F. Phillips, eds.), p. 215-276. Academic Press, New York, NY. Berry, P. F., and A. E. F. Heydorn. 1970. A comparison of the spermatophoric masses and mechanisms of fertilization in Southern African spiny lobsters (Palinuridae). S. Afr. Assoc. Mar. Biol. Res., Oceanogr. Res. Inst. Invest. Rep. 25, 18 p. 32 Fishery Bulletin 103(1) Comeau, M., and F. Savoie. 2002. Maturity and reproductive cycle of the female Amer- ican lobster, Homarus americanus, in the southern Gulf of St. Lawrence, Canada. J. Crust. Biol. 22:762-774. Davison, A. C, and D. V. Hinkley. 1997. Bootstrap methods and their application, 582 p. Cambridge University Press, New York, NY. DeMartini, E. E., G. T. DiNardo, and H. A. Williams. 2003. Temporal changes in population density, fecundity and egg size of the Hawaiian spiny lobster, Panulirus marginatus, at Necker Bank, Northwestern Hawaiian Islands. Fish. Bull. 101:22-31. DeMartini, E. E., P. Kleiber, and G. T. DiNardo. 2002 . Comprehensive ( 1986-2001 ) characterization of size at sexual maturity for Hawaiian spiny lobster (Panulirus marginatus) and slipper lobster (Scyllarides squammo- sus) in the Northwestern Hawaiian Islands. NOAA Tech Memo NMFS-SWFSC-344, 12 p. DeMartini, E. E„ and H. A.Williams. 2001. Fecundity and egg size of Scyllarides squammosus (Decapoda: Scyllaridae) at Maro Reef, Northwestern Hawaiian Islands. J. Crust. Biol. 21:891-896. Ennis, G. P. 1984. Comparison of physiological and functional size- at-maturity relationships in two Newfoundland popu- lations of lobsters Homarus americanus. Fish. Bull. 82:244-249. Evans, C. R., A. P. M. Lockwood, A. J. Evans, and E. Free. 1995. Field studies of the reproductive biology of the spiny lobster Panulirus argus (Latreille) and P. gutta- tus (Latreille) at Bermuda. J. Shellfish Res. 14:371- 381. Fernandez-Vergaz, V., L. J. Lopez Abellan, and E. Balguerias. 2000. Morphometric, functional and sexual maturity of the deep-sea red crab Chaceon affinis inhabiting Canary Island waters: chronology of maturation. Mar. Ecol. Prog. Ser. 204:169-178. George, R. W„ and G. R. Morgan. 1979. Linear growth stages in the rock lobster {Panu- lirus versicolor) as a method for determining size at first physical maturity. Rap. P.-V. Reun. Cons. Int. Explor. Mer 175:182-185. Gregory, D. R. Jr, and R. F Labisky. 1981. Ovigerous setae as an indicator of reproduc- tive maturity in the spiny lobster, Panulirus argus (Latreille). Northeast Gulf Sci. 4:109-113. Grey, K. A. 1979. Estimates of the size at first maturity of the west- ern rock lobster, Panulirus cygnus, using secondary sexual characteristics. Aust. J. Mar. Freshw. Res. 30:785-791. Groeneveld, J. C, and R. Melville-Smith. 1994. Size at onset of sexual maturity in the south coast rock lobster, Panulirus gilchristi (Decapoda: Palinuridae). S. Afr. J. Mar. Sci. 14:219-233. Hartnoll, R. G. 1974. Variation in growth pattern between some sec- ondary sexual characters in crabs (Decapoda: Brach- yura). Crustaceana 27:131-136. 1982. Growth. In The biology of Crustacea, vol. I, em- bryology, morphology, and genetics (D. E. Bliss, ed.), p. 111-196. Academic Press, London. Heydorn, A. E.F. 1969. The rock lobster of the South African west coast Jasus lalandii (H. Milne-Edwards). 2. Population studies, behavior, reproduction, moulting, growth and migration. S. Afr. Div. Sea Fish. Invest. Rep. 7:1- 52. Hogarth, P. J., and L. A. Barratt. 1996. Size distribution, maturity and fecundity of the spiny lobster Panulirus penicillatus (Oliver 1791) in the Red Sea. Trop. Zool. 9:399-408. Hossain, M. A. 1978. Appearance and development of sexual characters of sand lobster Thenus orientalis (Lund) (Decapoda: Scyllaridae) from the Bay of Bengal. Bangladesh. J. Zool. 6:31-42. Jayakody, D. S. 1989. Size at onset of sexual maturity and onset of spawning in female Panulirus homarus (Crustacea: Decapoda: Palinuridae) in Sri Lanka. Mar. Ecol. Prog. Ser. 57:83-87. Juinio, M. A. R. 1987. Some aspects of the reproduction of Panulirus penicillatus (Decapoda: Palinuridae). Bull. Mar. Sci. 41:242-252. Matthews. D. C. 1951. The origin, development, and nature of the sper- matophoric mass of the spiny lobster, Panulirus penicil- latus (Oliver). Pac. Sci. 5:359-371. Minagawa, M. 1997. Reproductive cycle and size-dependent spawning of female spiny lobsters {Panulirus japonicus) off Oshima Island, Tokyo, Japan. Mar. Freshw. Res. 48:869- 874. Minagawa, M., and S. Higuchi. 1997. Analysis of size, gonadal maturation, and func- tional maturity in the spiny lobster Panulirus japonicus (Decapoda: Palinuridae). J. Crust. Biol. 17:70-80. Minagawa, M., and M. Sano. 1997. Oogenesis and ovarian development cycle of the spiny lobster Panulirus japonicus (Decapoda: Palinuridae). Mar. Freshw. Res. 48:875-887. Montgomery, S. S. 1992. Sizes at first maturity and at onset of breeding in female Jasus verreauxi (Decapoda: Palinuridae) from New South Wales waters, Australia. Aust. J. Mar. Freshw. Res. 43:1373-1379. Plaut, I. 1993. Sexual maturity, reproductive season and fecun- dity of the spiny lobster Panulirus penicillatus from the Gulf of Eilat (Aqaba), Red Sea. Aust. J. Mar. Freshw. Res. 44:527-535. Prescott, J. H. 1984. Determination of size at maturity in the Hawaiian spiny lobster Panulirus marginatus, from changes in relative growth. Proc. Res. Inv. NWHI. UNIHI-SEA GRANT-MR-84-01, p. 345. Univ. Hawaii, Honolulu, HI. Ratkowsky, D. A. 1983. Nonlinear regression modeling: a unified practical approach, 276 p. Marcel Dekker, New York, NY. Somerton, D. A. 1980. A computer technique for estimating the size of sexual maturity in crabs. Can. J. Fish. Aquat. Sci. 37:1488-1494. Watters, G, and A. J. Hobday. 1998. A new method for estimating the morphometric size at maturity of crabs. Can. J. Fish. Aquat. Sci. 55:704-714. DeMartini et al.: Validated morphological metric for lobster size at maturity 33 Appendix Method for estimation of maturation with pleopod metrics To model the allometry we used the power function Y=(5X'1' and assumed multiplicative error. The logarithmic transformation of this function leads to a linear regres- sion model. Specifically, we defined ln(Y) = f1(X)+e1 where fi(X)=a1+l\\n{X) and a, =ln((3j), as the allometric relationship for juvenile lobsters and ln(Y) = f2(X)+e2, where f2(X)=a2+fi2\n(X) and a2 =ln(d2), as the allometric relationship for adult lobsters. The errors, el and e2, were assumed to be independent and normally distributed with mean 0 and variance ( (i) When y=0, Pim\x) increases linearly from 0 to 1 over the interval [80, 6l ]. For y>0, the curves are sigmoidal, symmetrical, and the rate that the probability changes with respect to tail width is bell shaped (the sigmoidal curve first accelerates, then decelerates). The point of inflection, {6^+0^12, is the tail width at which 50% of the lobsters are expected to be mature. For both species, we assumed that yaO. Defining the allometry model and the probability of maturity as above, we expressed the model relat- ing pleopod length to tail width as ln( Y)=/1(X)(1- P(m\x))+f2iX)Pim\x)+e, where f are independent normal variates with mean 0 and covariance Vm. Assuming (.t,,y() i=l, ... ,n independent paired obser- vations and cP- = o^=o22, V!H=Io2+M, where / is the (n x n) identity matrix, M is the diagonal matrix MU=A2 (xt) P(m,l.v,) (1-P(m,l.v,)), and AU,)^*,)-/^*,). Hence, we have a weighted least squares problem with weights (o'+Mj if xl s 00 or xl a 6X ife00), 80 was bounded such that 0oaexp(-cc3//33), and if /33<0, 61 was bounded such that 0jsexp(-a3//33). The curve was fitted by using iteratively reweighted least squares. The weights were recomputed at each iteration. While fitting the lobster data to the specified model, we observed that one or more of the parameters in- volved in defining the sigmoidal curve departed from linear behavior. Under these circumstances, the con- fidence interval derived by assuming the asymptotic properties of maximum likelihood estimates may be invalid (Ratkowsky, 1983). Therefore, we computed ap- proximate 95% confidence intervals for the point of inflection using the bootstrap method. Specifically, we used case resampling with 1000 bootstrap replications. Confidence intervals were derived by using the studen- tized bootstrap confidence limits (Davison and Hinkley, 1997). J4 Abstract — In this study we present new information on seasonal variation in absolute growth rate in length of coho salmon (Oncorhynchus kisutch ) in the ocean off Oregon and Washington, and relate these changes in growth rate to concurrent changes in the spacing of scale circuli. Average spac- ing of scale circuli and average rate of circulus formation were significantly and positively correlated with average growth rate among groups of juvenile and maturing coho salmon and thus could provide estimates of growth between age groups and seasons. Regression analyses indicated that the spacing of circuli was proportional to the scale growth rate raised to the 0.4-0.6 power. Seasonal changes in the spacing of scale circuli reflected seasonal changes in apparent growth rates offish. Spacing of circuli at the scale margin was greatest during the spring and early summer, decreased during the summer, and was lowest in winter or early spring. Changes over time in length offish caught during research cruises indicated that the average growth rate of juvenile coho salmon between June and Septem- ber was about 1.3 mm/d and then decreased during the fall and winter to about 0.6 mm/d. Average growth rate of maturing fish was about 2 mm/d between May and June, then decreased to about 1 mm/d between June and September. Average appar- ent growth rates of groups of matur- ing coded-wire-tagged coho salmon caught in the ocean hook-and-line fisheries also decreased between June and September. Our results indicate that seasonal change in the spacing of scale circuli is a useful indicator of seasonal change in growth rate of coho salmon in the ocean. Seasonal changes in growth of coho salmon (Oncorhynchus kisutch) off Oregon and Washington and concurrent changes in the spacing of scale circuli Joseph P. Fisher William G. Pearcy College of Oceanic and Atmospheric Sciences Oregon State University 104 Ocean Admin. Building Corvallis, Oregon 97331-5503 E-mail address (for J. P. Fisher) |fisheng>coasoregonstateedu Manuscript submitted 20 September 2003 to the Scientific Editor. Manuscript approved for publication 8 September 2004 by the Scientific Editor Fish. Bull. 34-51(2005). Large interannual and decadal varia- tions occur in the abundance and pro- ductivity of North Pacific salmonids. These fluctuations, which affect har- vestable biomass, are influenced by survival rates, ages at maturity, and somatic growth (Beamish and Bouil- lon, 1993; Mantua et al., 1997; Hare et al. 1999; Pyper et al., 1999; Hobday and Boehlert, 2001). The growth of smolts after ocean entry — growth that is critical to production — is also thought to be an important determinant of their survival. As for juvenile and larval fishes in general, size-selective mor- tality may occur (Miller et al., 1988; Bailey and Houde, 1989; Litvak and Leggett, 1992; Sogard, 1997) with the result that faster growing sal- monids experience less mortality from predators than slower growing salmonids (Parker, 1971; Bax, 1983; Fisher and Pearcy, 1988; Holtby et al., 1990; Jaenicke et al., 1994; Wil- lette, 1996, 2001). This size-selective mortality may explain much of the interannual variability in survival of juvenile salmonids and the sub- sequent abundance of different year classes. However, other investigators have not found a strong relationship between growth of juvenile salmon and mortality (Fisher and Pearcy, 1988; Mathews and Ishida, 1989; Blackbourn, 1990). Intercirculus spacing of scales has been used to estimate early ocean growth rate of juvenile salmon and has been linked to differential sur- vival rates. For example, Healey (1982) used the spacing of the first five circuli to demonstrate intensive size-selective mortality in juvenile chum salmon (Oncorhynchus keta) as they migrated offshore. Holtby et al. (1990) correlated early ocean growth, based on intercirculus spacing, with marine survival of age 1+ coho (O. kisutch) smolts. The spacing of early ocean circuli from the scales of ma- turing Atlantic salmon (Salmo salar) has been used to estimate juvenile growth rates, which are correlated with survival and age at maturity, and to identify stocks (Friedland et al., 1993; Friedland and Haas, 1996; Friedland and Reddin, 2000; Fried- land et al., 2000). Correlation between circulus spac- ing and growth rate was reported by Fisher and Pearcy (1990) for age 0.0 coho smolts reared for 60 days in salt water tanks. In addition, posi- tive correlations between the spacing of scale circuli and fish growth rate have been observed for rainbow trout (O. mykiss) (Bhatia, 1932), and sock- eye salmon (O. nerka) (Fukuwaka and Kaeriyama, 1997), and between the spacing of circuli and feeding ration and growth for sockeye salmon (Bil- ton and Robins, 1971; Bilton, 1975). Bigelow and White (1996) were able to manipulate the spacing of scale circuli of cutthroat trout (O. clarkii) Fisher and Pearcy: Seasonal changes in growth of Oncorhynchus kisutch off Oregon and Washington 35 Table 1 Main sources of coho salmon data used in this study. Source Numbers of fish Scale samples CWT maturing fish caught in the Oregon ocean sport and troll fisheries 1982-92 (see Table 2) 687 687 Maturing coho salmon caught in the ocean during research cruises 1981-85 1391 352 1998-2002 714 236 Juvenile fish caught in the ocean during research cruises 1981-85 1798 1998-2002 3684 1052 CWT maturing coho salmon caught in the sport and troll ocean fisheries (all catch areas) 149,718 — and released between northern Oregon and northern Washington' 1 FL data in the Pacific States Marine Fisheries Commission, Regional Mark Information System online CWT data base http //ww w.rmis.org/. [Accessed 1 April 2003.] in the hatchery by varying the feeding levels: the group that was fed the most also grew the most and had the most widely spaced scale circuli. Positive correlations between circulus spacing and growth also have been ob- served for nonsalmonid fishes including Tilapia (Doyle et al., 1987; Matricia et. al., 1989; Talbot and Doyle, 1992), and walleye (Stizostedion vitreum) (Glenn and Mathias, 1985). Circulus spacing is potentially useful for comparing ocean growth rates of salmon in the ocean. Spacing of the first few ocean scale circuli may indicate relative growth rates of juvenile fish immediately after ocean entry. However, in order for spacing of scale circuli to be a practical indicator of fish growth rate, the relationship between the two must be consistent and significant. The relationship between circulus spacing and fish or scale growth rate is determined by the relative rates of growth and circulus formation. If circuli (like tree rings) are formed at a constant rate, then there would be a directly proportional relationship between spacing and growth rate (e.g., a doubling of growth rate would result in a doubling of spacing). Conversely, if the rates at which circuli are formed are directly proportional to growth rates (e.g., a doubling of growth rate would result in a doubling of circulus formation rate), then the spacing of circuli would be constant. Our earlier study of growth rate, circulus formation, and circulus spacing among 82 individually marked juvenile coho salmon growing for a period of 63 days in saltwater tanks indi- cated that neither of these two extremes is the case, but that both circulus formation rate and circulus spacing are positively correlated with fish growth rate (Fisher and Pearcy, 1990). Our main objectives in this study are to further as- sess the reliability of circulus spacing as an indicator of growth rate in FL of coho salmon in the ocean, to investigate how growth of coho salmon changes season- ally, and to compare any seasonal changes in growth rate with seasonal changes in the spacing of scale cir- culi. If circulus spacing is a reliable indicator of growth rate, then seasonal changes in growth rate should be tracked by changes in the spacing of circuli laid down at the scale margin. We investigated relationships be- tween scale growth rate, fish growth rate, circulus spac- ing, and circulus formation rate for coded-wire-tagged (CWT) adult coho salmon collected in the ocean fisher- ies in years when ocean growth varied widely, including year classes affected by the 1982-83 El Nino, and for juvenile and maturing coho salmon caught in the ocean off Oregon and Washington in research cruises 1981-85 and 1998-2002. Materials and methods Scale and FL data Fish fork length (FL) and scale data from a variety of sources were used in this study (Table 1). During research cruises on the Oregon and Washington coastal shelf we collected juvenile and maturing coho salmon in the upper 20-40 m of the water column with purse seines from 1981-85 (Pearcy and Fisher, 1988, 1990) and with a rope trawl from 1998-2002 (Emmett and Brodeur, 2000). Scales samples were removed from the fish from an area equivalent to area "A" described in Scarnnechia (1979). When scales were not available from area "A," we took scales from between areas "A" and "B" in Scarnnechia (1979). (See also Clutter and Whitesel, 1956). We also examined scales from the same area from 687 maturing CWT Columbia River and northern coastal Oregon coho salmon caught in the Oregon ocean fisheries between 1982 and 1992. Changes over time in FLs of maturing coho salmon caught in research nets and of CWT hatchery coho salmon originating between northern Oregon and north- ern Washington and caught in the ocean fisheries be- 36 Fishery Bulletin 103(1) 1982-1983 1983-1984 Figure 1 Scales from the 1982-83 and 1983-84 (smolt year through adult year) year classes of coho salmon (Oncorhynchus kisutch) showing the axis of measurement, the scale focus (F), ocean entry (OE), the annulus (A) at the end of the annual ring and the scale margin (Ml. tween 1975 and 2002 were used to estimate growth rates of maturing fish (Table 1). Scale measurements We measured the distances (mm) along the anterior-pos- terior scale axis from the focus (F) to the last circulus of the freshwater zone (ocean entry, OE), to the outside edge of the winter annual ring (the "winter annulus," A) when present, and to the margin (M), and also deter- mined the total numbers and average spacing of circuli in the ocean growth zone (Fig. 1). For certain scale samples we also determined the spacing of every circulus in the ocean growth zone of the scales or of the last few circuli at the scale margin. Measurements of scales from juvenile fish caught during research cruises 1981-85 were taken from im- ages projected by a microfiche reader at a magnifica- tion of about 88x and measurements of scales from all other fish were acquired with image analysis software (Optimas, vers. 5.1, Optimas, Inc., Seattle, WA, and Image-Pro Discovery, vers. 4.5, Media Cybernetics, Sil- ver Spring, MD) by using a CCD camera coupled to a Leica compound microscope. All measurements were calibrated from images of a stage micrometer. Circulus spacing and formation rate versus growth rate We used correlation and regression analyses to relate average circulus spacing and formation rate to average scale and fish growth rate among year classes of juvenile coho salmon during their first four or five months in the ocean and among groups of maturing CWT coho salmon during their entire ocean life (Table 2). We described the relationships between the scale characteristics and growth rate as power functions by using natural log (In) transformed variables in linear regressions. Geo- metric mean (GM) regression (Ricker, 1973, 1992; Sokal and Rohlf, 1995) was used to relate the In-transformed variables because they were subject to both natural variability and measurement error and because our pur- pose in the present study was to describe the functional relationships between the variables and not to predict one from the other. For each fish, rates of scale growth, fish growth, and circulus formation in the ocean were estimated as (SR-SR0E)/Ad, (FL-FL0E)IAd, and CIRC/Ad, re- spectively, where SR = scale radius at capture, SR0E= scale radius at ocean entry (F to OE in Fig. 1), FL = fork length at capture, FL()A=estimated fork length at ocean entry, C/.RC=the total number of circuli in the ocean growth zone of the scale, and Ad = estimated days between ocean entry and capture. Average spacing of circuli was calculated as (SRLAST-SR0E)/CIRC, where SRj ,lsr=the scale radius to the last circulus before the scale margin. For juvenile fish, FL0E was estimated by using the Fraser-Lee back-calculation method (Ricker, 1992) and the intercept from the FL-SR regression for ocean- caught juvenile fish (34.16 mm. Fig. 2). However, be- Fisher and Pearcy: Seasonal changes in growth of Oncorhynchus kisutch off Oregon and Washington 37 cause of allometry in the FL-SR relationships of juvenile and maturing fish (Fig. 2), which a ln-ln transformation of the data failed adequately to correct, the Fraser-Lee method was not used to estimate FLOE of the maturing fish caught in the ocean. Instead, FLOE of maturing fish was estimated by direct substitution of (SROE) into the GM regression relationship between FL and SR for juvenile coho salmon caught in the ocean 1981-85 and 1998-2001 (gray regression line. Fig. 2). For juvenile fish caught in August or September, Ad was estimated as the capture date minus 25 May, a date near the peak of coho salmon smolt migration in the Columbia River estuary (Dawley et al., 1985a). Because we used a single date of ocean entry for all fish, errors in estimated growth rates of some individual juvenile coho salmon probably were quite large; the timing of ocean entry of fish can vary by as much as two months. However, for the correlation and regression analyses we used growth rates averaged by year class, which were probably quite accurate, if the average date of ocean entry of the fish in the samples is assumed to be similar across years. In the Columbia River, the major source of juvenile coho salmon on the Oregon and Washington coasts, ocean entry was concentrated between late April and early June and the timing of ocean entry varied little between years (Dawley et al., 1985a). Dates of ocean entry of the maturing CWT Sandy and Cowlitz hatchery coho salmon (Table 2) were esti- mated from the hatchery release dates and the rates of downstream migrations of these fish observed during extensive sampling of migrating smolts at rkm 75 in the upper Columbia River estuary (Dawley et al., 1985b). To estimate dates of ocean entry of the Fall Creek hatch- ery fish, for which data on downstream migration were lacking, we assumed that smolts migrated to the ocean from the different release sites at the same average rate of downstream migration as that of Cowlitz Hatchery fish released in late April (5.7 km/d). Potential errors in estimated growth rates of matur- ing CWT coho salmon caused by inaccurately estimat- ing size of fish at ocean entry, or date of ocean entry, were proportionally very small when compared to the total amount or duration of ocean growth. At a typical SR0E of around 0.7 mm, the 95% prediction limits for FL from the SR-FL regression of juvenile fish (Fig. 2) are about ±31mm. An error in size at OE of 15-30 mm would only be 2-10% of the estimated total growth in FL in the ocean of the maturing fish (320 mm-610 mm). Similarly, an error in estimated date of ocean entry of 30 days would equal only about 6-10% of the total time that the fish was in the ocean (336-535 d). Errors for the group-averaged data used in our correlation and regression analyses were probably much lower. Seasonal changes in spacing of circuli To investigate whether circulus spacing and growth rate were correlated seasonally, we first described the patterns of seasonally changing circulus spacing of juvenile and maturing coho salmon in the ocean and Table 2 Nine year classes of juvenile coho salmon caught in research nets in August or September and 17 groups of CWT maturing coho salmon caught in the Oregon ocean fisheries used in the correlation and regression analyses of scale characteristics and growth rate. CWT maturing fish were from three hatcheries (Fall Creek "F" on the northern Oregon coast and Sandy "S" and Cowlitz "C" in the lower Columbia River basin) and were released from hatcheries during three periods. Capture year Hatcheries Numbers offish CWT maturing fish released late April or early May (days 119-127) 1982 F, S 1983 F, S, C 1984 S, C 1985 S, C 1986 S 1987 S 1989 S 1990 S CWT maturing fish released in March (days 74-76 1984 F 31 1985 F 21 CWT maturing fish released in late May or early Juneldays 151-157) 1991 S 30 1992 S 77 11, 15 34, 17, 51 52,35 12,26 67 94 57 18 Juvenile fish 1981 1982 1983 1984 1998 1999 2000 2001 2002 99 95 81 88 13 60 75 67 123 then compared these patterns of changing circulus spacing to changing fish growth rates. Because the widths of the pre-annulus and postannulus scale zones and the numbers of circuli in each zone varied greatly among individual fish and among groups of fish, we described circulus spacing in each of 25 equally spaced intervals between OE and the annulus and in each of 25 equally spaced intervals between the annulus and the scale margin, rather than on a circulus by circulus basis. Specifically, the pre-annulus and postannulus ocean zones of scales were each divided into 25 equal intervals, and the radial distance from OE to the upper bounds of each of the intervals was determined. Next, the numbers of ocean circuli between OE and the upper bounds of each of the 50 intervals were interpolated. For example, if a boundary fell 25% of the distance 38 Fishery Bulletin 103(1) — 400 200 between the 38th and 39th ocean circulus, the circulus number 38.25 was assigned to that boundary. We calculated the circulus spac- ing in each interval as 4mm/Acirc, where 4mm = the width in mm of the interval, and 4circ = the difference between the interpo- lated circulus numbers at the upper and lower bounds of the interval. The circulus spacing in each of the 50 intervals was aver- aged across all the scales from the fish in a group. This produced a profile of the average spacing of circuli at 50 different positions in relation to OE (lower bound of interval 1), the annulus (upper bound of interval 25) and the scale margin (upper bound of interval 50). Finally, the group-average cir- culus spacing in each of the 50 intervals was plotted against the group-average radial dis- tance from OE to the upper bounds of each of the 50 intervals. For juvenile fish caught in trawls in September 1999-2002, circu- lus spacing was described at 25 intervals in relation to OE (lower bound of interval 1) and the scale margin (upper bound of interval 25). Seasonal changes in the spacing of circuli at the growing edge of the scale may reflect similar seasonal changes in the growth rate of the juvenile and maturing coho salmon. To investigate this possible correlation, we measured the spacing of the last two circu- lus pairs at the scale margin of juvenile fish caught in early and late summer in 1982 and 1999 through 2002 and of maturing fish caught in research nets 1981-83 and 2000-2002 and in the ocean fisheries 1982-92 (Table 1). Mean spacing of the last two circulus pairs was summarized by cruise for the fish caught in research nets, and by 10-day catch intervals for the fish caught in the ocean fisheries. The seasonal trends in spacing at the scale margin were then compared with the seasonal trend in apparent growth rates of fish. Seasonal changes in fish growth rate Seasonal trends in growth rates of juvenile and matur- ing coho salmon caught in research cruises 1981-83 and 1998-2002 were estimated from the changes between cruises in average FL. We also estimated average growth rates (pooled across years) of juvenile and adult coho salmon during different seasons by fitting regressions to the FL versus catch date data. Changing stock composition of the juvenile (Teel et al., 2003) or maturing coho salmon caught in research nets over the course of the summer could potentially have a strong effect, independent of growth, on the size distributions of fish caught at different times. Therefore, changes over time in average FLs of mixed stocks of fish, such as in our research collections, may not ac- curately indicate actual fish growth rates. 1000 800 600 - 0 - o Adults, May-Sept. 1981-1983 ° Adults, June and Sept, 2001 , 2002 • Adults, June 2000 and Table 2 • Juveniles, 1981-1985, 1998-2001 FL (mm) = 1 50-94 SR* 34.1 6, n=2834. r2 = 0.94 Scale radius (mm) Figure 2 Fork length (FL) versus scale radius (SR) for juvenile and matur- ing coho salmon I O. kisutch) caught in research trawls and GM regressions of FL versus SR fitted to juvenile and adult fish sepa- rately. Note the allometry in the FL-SR relationship of juvenile and adult fish. Because of the potential for error when inferring seasonal changes in growth rate from changes over time in average FLs of mixed stocks of fish, we also examined temporal changes in FL of maturing CWT coho salmon of known origin caught in the ocean hook- and-line fisheries (sport and troll fisheries). Using data available from the Pacific States Marine Fisheries Com- mission1 we investigated changes over the summer in FLs of maturing CWT coho salmon originating from six areas (north Oregon coast, lower Columbia River basin-Oregon, lower Columbia River basin-Washington, Willapa Bay basin, Grays Harbor basin, and the north- west Washington coast). Because the date that a smolt is released from a hatchery (e.g., March vs. June) could affect its size the following year, we also grouped the fish by release periods of 25-46 days duration. Da- ta were available on FLs of maturing CWT fish from 1975-2002. For each group in each year we calculated the average FL of CWT fish at 10-day intervals in the hook-and-line fisheries (sport and troll fisheries) pooled for all catch areas between California and Alaska. Data were discarded when there were fewer than 5 fish mea- 1 Regional Mark Information System CWT database (http:// www.rmis.org). [Accessed on: 1 April 2003.1 Fisher and Pearcy: Seasonal changes in growth of Oncorhynchus kisutch off Oregon and Washington 39 Table 3 Summary statistics of lus formation, and avc and during the entire average estimated fish growth rate, average estimated scale growth rate, average rage circulus spacing between ocean entry and late summer for nine year classes ocean growth period for the 17 groups of CWT maturing coho salmon (see Table 2) estimated rate of circu- of juvenile coho salmon Statistic Average fish growth rate (mm/d) Average scale growth rate (mm/dl Average circulus formation rate (circuli/d) Average circulus spacing (mm) Juvenile fish, n=9 Grand average 1.33 0.0087 0.188 0.0460 Minimum 1.18 0.0080 0.175 0.0428 Maximum 1.52 0.0101 0.202 0.0494 SD 0.10 0.0007 0.008 0.0023 CV 7.6% 8.3% 4.1% 4.9% Maturing fish, n = 17 Grand average 1.11 0.0060 0.131 0.0463 Minumum 0.94 0.0048 0.110 0.0426 Maximum 1.23 0.0066 0.144 0.0511 SD 0.07 0.0005 0.009 0.0020 CV 6.7% 8.1% 6.8% 4.4% sured in any 10-day catch period. The average FLs were averaged across all years of data, yielding grand- average FLs for each 10-day catch period. The grand average FL for each 10-day catch interval comprised 1-27 years of data, but those periods with fewer than 5 years of data were discarded. In all, FLs from 149,718 fish were used in the analysis. Grand average FLs and the apparent growth rates in FL between each 10-day catch period were plotted against date and compared with the seasonal changes in circulus spacing at the scale margin of the fish in our scale sample. Results Growth and scale statistics for juvenile and maturing fish Average growth rates and circulus formation rates were greater for juvenile fish during their first ocean summer than for maturing fish during their entire ocean life probably because maturing fish experience slow growth in the winter (Table 3). During their first summer in the ocean, juvenile fish grew an average of 1.33 mm/d and formed circuli at the rate of 0.188/d (one every 5.3 days): whereas, during their entire ocean life maturing fish grew an average of 1.11 mm/d and formed circuli at the rate of 0.131/d (one every 7.6 days). The highest average growth rate (1.52 mm/d) among the eight year classes of juvenile coho salmon was about 28% higher than the lowest average growth rate (1.18 mm/d). The percentage range in growth rate of maturing fish was similar (31%). Average spacing of circuli was similar for both juvenile and maturing coho salmon (0.0460 mm vs. 0.0463 mm), probably because scales from the maturing fish contained both more narrowly spaced circuli formed during the winter and more widely spaced circuli formed during the second ocean summer (see below). The varia- tion among groups in average circulus spacing (CV=4.9% and 4.4%) was lower than the variation in fish or scale growth rates (CV=6.7% to 8.3%), although estimation error may have increased the coefficients of variation of the growth rates. Correlations between scale characteristics and growth rate Circulus spacing was strongly correlated (r=0.89 and 0.82, respectively) with scale and fish growth rates among the nine year classes of juvenile coho salmon (Table 4). Circulus spacing was also significantly cor- related with scale and fish growth rates among the 17 groups of maturing fish, but the correlations were weaker (r=0.57 and 0.55, respectively) than those for the juvenile fish. Conversely, correlations between the rate of circulus formation and the scale and fish growth rates were slightly higher for the maturing fish (r=0.85 and 0.75, respectively) than for the juvenile fish (r=0.76 and 0.81, respectively). These results suggest that when growth is averaged over several seasons, during which growth rate varies greatly and may even cease for vary- ing periods of time, differences in growth among year classes or groups may be reflected more clearly by dif- ferences in the numbers of circuli laid down on the scale than by differences in the average spacing of circuli. Although the average spacing of circuli and the aver- age rate at which circuli form were both correlated with scale and fish growth rates, they were not correlated with each other (Table 4). This finding indicates that 40 Fishery Bulletin 103(1) circulus spacing and circulus formation rate are inde- pendent indicators of growth rate — both tending to in- crease with increasing growth rate but not necessarily together in the same fish or in the same group or year class. At least when averaged over periods of months or more than a year, differences in average growth rate may be expressed by differences in average spacing of circuli, differences in average rate of circulus formation, or differences in both. Regressions of circulus spacing and formation rate on growth rate We expressed average spacing of circuli and rates of circulus formation as power functions of the scale growth rates, equivalent to linear regressions of ln-ln trans- formed data. These regressions are shown in Figures 3 and 4 for year classes of juvenile fish and groups of maturing fish, respectively. Because scale growth rate and fish growth rate were very strongly corre- lated (Table 4), we show only the regressions with scale growth rate. Change in average spacing of circuli and in average rate at which circuli form was proportionally smaller than the change in average scale growth rate. Aver- age spacing of circuli was proportional to the average scale growth rate raised to the 0.6 power (juvenile fish, Fig. 3A) or the 0.5 power (maturing fish, Fig. 4A). If these relationships hold over a wider range of scale growth rate and circulus spacing, then a doubling of scale growth rate would be associated with only a 1.5- fold (20-6) or 1.4-fold (205) increase in circulus spac- ing. Similarly, average rate of circulus formation was proportional to the average scale growth rate raised to the 0.5 power (juvenile fish, Fig. 3B) or the 0.8 power (maturing fish, Fig. 4B). Seasonal changes in circulus spacing and fish growth rate Seasonal changes in average circulus spacing were con- sistent among the different year classes and release times of CWT coho salmon (Fig. 5, A-E). During the first year in the ocean, average spacing of scale circuli increased rapidly after OE (usually in May) to aver- age peak values of about 0.050 mm-0.055 mm, then gradually decreased to average minimum values of about 0.031 mm-0.040 mm in the annual ring. By late September 1999-2002, spacing at the margin of scales from juvenile fish had decreased from peak values (Fig. 5E), indicating that the gradual decrease in spacing of circuli which forms the annual ring begins as early as the late summer of the first ocean year. For some year classes (e.g., 82-83, 85-86, 90-91, 91-92) the annual ring was a distinct narrow zone of very closely spaced circuli (Fig. 5, A and C), whereas in other years the annual ring was broad and subtle, with more widely spaced circuli (e.g., 83-84, 86-87, and 84-85 for the March released fish; Fig. 5, A and B). After the annulus (black dots, Fig. 5), the spacing of circuli increased sharply to peak values of about Table 4 Correlations (r) between average circulus spacing (mmi, average estimated scale growth rate (mm/d), average estimated fish growth rate (mm/dl, and average esti- mated circulus formation rate (circuli/d) between ocean entry and late summer for nine year classes of juvenile coho salmon and during the entire ocean growth period for 17 groups of CWT maturing coho salmon (see Table 2). All correlations were significant (P<0.05), except were noted ("n.s"). Comparison Circulus spacing vs. scale growth rate Circulus spacing vs. fish growth rate Circulus spacing vs. circulus formation rate Scale growth rate vs. fish growth rate Scale growth rate vs. circulus formation rate Fish growth rate vs. circulus formation rate Juvenile Maturing fish fish r r 0.89 0.57 0.82 0.55 0.38, n.s. 0.05, n.s 0.97 0.91 0.76 0.85 0.81 0.75 0.055 mm-0.060 mm and remained high for a vari- able distance. Compared to the peak spacing, spacing of circuli at the scale margin was relatively high for maturing fish caught in late June or July 1982, 1984, 1985, 1986, 1987, 1991, and 2000, whereas, spacing at the scale margin was quite low compared to the peak spacing for fish caught in July 1983, 1989, 1990, and 1992 (Fig. 5, A, C, and D). Spacing at the scale margin was very low among unmarked maturing fish caught in late September 2001 (Fig. 5D). Compared to the large interseasonal variation in spacing of circuli in the pre- and postannulus zones, from about 0.03 mm in the annual ring to about 0.06 mm for the most widely spaced circuli, interannual variation the peak and minimum spacing of circuli was quite small. The peak spacing of circuli was similar among year classes, even when total growth differed greatly (e.g., the 82-83 vs. the 81-82 and 83-84 year classes, Fig 5A). The unusually small postannulus scale growth of fish caught during a strong El Nino in July 1983 (Fig. 5A) was characterized by a much narrower region of widely spaced circuli and more closely spaced circuli at the scale margin than in other years. In general, pre-annulus scale growth was greatest for the fish released in March (Fig. 5B), was slightly less for the fish released in late April or early May (Fig. 5A), and was smallest for the fish released in late May or early June (Fig. 5C). These data indicate that date of release may strongly affect the amount of growth at- Fisher and Pearcy: Seasonal changes in growth of Oncorhynchus kisutch off Oregon and Washington 41 Spacing = 0.839 ■ScaleCrouvthRate0 613 . n - 9. r= 0.88. I2 = 0.78 a) E 020 0) S> 017 CircRate= 2 03V ScaleGrowthRale0502 .n = 9, r=0.74, ^ = 0.55 0.007 0 008 0 009 0.010 0 011 Scale growth rate (mm/d) Figure 3 Estimated average scale growth rate versus (A) average spac- ing of ocean circuli and (B) estimated average rate of circulus formation for nine year classes (see Table 2) of juvenile coho salmon (O. kisutch) caught in the ocean in research nets in August (1981) or September (1982-84 and 1998-2002; black symbols, ±2 SE). Regressions are GM linear regressions of In-transformed variables (presented in their power function form). tained by juvenile coho salmon during their first sum- mer, fall, and winter in the ocean. Do the seasonal changes in circulus spacing in the ocean growth zones of scales coincide with similar sea- sonal changes in growth rates of juvenile and maturing coho salmon? In Figure 6 we plotted the average lengths of juvenile and maturing coho salmon from all research cruises 1981-2002 and the average apparent growth rates of coho salmon during different seasons (dashed lines). Apparent average growth rate of juvenile coho salmon between June and September was 1.30 mm/d, about twice the apparent growth rate of 0.64 mm/d between September and the following May. Apparent growth rates of maturing fish between late May and late June was very rapid (2.11 mm/d), about twice as great as the apparent growth rate of maturing fish later between June and September (1.01 mm/d). In a general sense, this pattern of changing apparent growth rate over time in the ocean corresponds well to the pattern of changing circulus spacing seen in Fig- ure 5, A-E. The rapid growth of juvenile coho salmon between June and September occurs during a period when the spacing of circuli generally is high (Fig. 5E). When maturing fish were caught in the ocean fisheries in late June and in July and August a zone of widely spaced circuli already was present on the scales (Fig. 5, 42 Fishery Bulletin 103(1) E E. 0050 Spacing =0 61 7'ScaleGrowthRate □ A o o Cowlilz.rel days 123-124 Fall Creek, rel. days 121-122 Fall Creek, rel days 74-76 Sandy, rel days 119-127 Sandy, rel days 151-157 0 0040 0 0045 0 0050 0 0055 0 0060 0 0065 0 0070 0.0075 0 0060 CircRate = 6 61 b' ScaleGrowthRate n=17. /-=087, i2 = 0.75 B 0.0040 0.0045 0.0050 0.0055 0.0060 0.0065 0 0070 0.0075 0 0080 Scale growth rate (mm/d) Figure 4 Estimated average scale growth rate versus (A) average spac- ing of ocean circuli and (B) estimated average rate of circu- lus formation for 17 groups (see Table 2) of maturing coho salmon (O. kisutch) caught in the Oregon ocean fisheries (±2 SE). Regressions are GM linear regressions of In-transformed variables (presented in their power function form). Data for Sandy Hatchery fish caught in 1983 and 1984 are labeled as examples of year when average growth rates were extremely different. A-C), indicating that these widely spaced circuli were produced earlier during the period of apparently rapid growth in the spring and early summer (Fig. 6). Circu- lus spacing at the scale margin was already declining in July among maturing fish in some years (Fig. 5A), and was clearly lower among maturing fish caught in August or September (Fig. 5, B and D) indicating that these more narrowly spaced circuli were produced some- time during the apparently slower growth of maturing fish between late June and September (Fig. 6). Finally, the low spacing of circuli in the annual ring occurs sometime between late September of the first year and mid-May of the second year, which was also the period of lowest apparent growth rate (Fig. 6). The pattern of changing circulus spacing at the scale margin is most clearly seen when average spacing of the outer two circulus pairs is plotted against the average Julian day of capture (Fig. 7, A and B). Among juvenile fish caught in research nets, the average spacing of the circuli at the scale margin was narrower in September than in June (Fig. 7A, see also Fig. 5E). We lack suf- ficient FL data from mid and late summer to deter- mine whether or not a decrease in the average growth rate of juvenile fish was associated with the observed Fisher and Pearcy: Seasonal changes in growth of Oncorhynchus kisutch off Oregon and Washington 43 Late April-Early May release — July recovery 0 060 - 0.055 - 0.050 0.045 H 0.040 0.035 0.030 0.5 2.0 0 060 0.055 0.050 0.045 0.040 - 0.035 - 0 030 84-85. n = 27 85-86, n = 54 0.0 0.5 — I — 1.0 2.0 2.5 0060 0.055 0.050 0.045 0.040 0.035 0.030 1 1 1 1 1 00 05 1.0 1.5 20 25 Mean scale radius (mm, OE = 0) Figure 5 Profiles of changing average circulus spacing (±2 SE) versus average scale radius at 50 intervals along the axis of mea- surement (see "Methods and Materials" section) for matur- ing coho salmon (O. kisutch) caught in the ocean fisheries and (A) released as smolts from hatcheries in late April or early May and caught in July, (B) released in March, (C) released in late May or early June and caught late June to late July, (D) unmarked maturing fish caught in research nets in June 2000 and September 2001, and (E) juvenile fish caught September 1999-2002. For clarity, error bars for the average scale radius at each interval are not shown. decrease in spacing of circuli at the scale margin in September. Among maturing fish, average spacing of the last two circulus pairs at the scale margin decreased greatly between the spring through early summer period and early fall (Fig. 7B). The decrease in circulus spacing at the scale margin during the summer occurred for both maturing fish of mixed stocks caught in research nets (gray and white symbols) and for CWT fish of known stocks caught in the ocean sport and troll fisheries (black symbols). The decrease also was very consis- tent among year classes; 11 of the 12 year-class groups (grouped by release period and pooled across hatcher- ies) of Table 2 showed significant negative correlations between spacing at the margin and date of capture (P<0.05, r=-0.40 to -0.59). In September the aver- age circulus spacing at the scale margin was about as low as the average circulus spacing in the annual ring (about 0.035 mm). The decrease in spacing of circuli at the scale margin over the summer mirrors a similar decrease over the summer in apparent growth rates in FL of maturing fish caught in research nets (Fig. 7C). The apparent growth rates of maturing coho salmon were usually 44 Fishery Bulletin 103(1) March release 0 065 - 0.060 - 0.055 - 0.050 - 0.045 - 0.040 0 035 ■ 83-84, n = 24. 7/9 • 8/8 recovery • 84-85. n = 15. 7/29 - 9/2 recovery 00 0.5 3.0 Late May-June release, June 19— July 19 recovery F fc 0.060 - 0.055 - CO 0.050 - in in 0 045 - -j 0.040 - C ) ~> 0.035 - c 0.030 - Surface trawl research: maturing fish 0 060 - 0.055 0.050 - 0.045 0.040 0.035 0 030 — 99-00. n= 78, 6/19-6/25 recovery Y\ 00-01 , n = 50. 9/21 - 9/29 recovery 05 1.0 1 5 20 Mean scale growth (mm) — i — 25 Surface trawl research: juvenile fish E b 0 060 en c 0.055 ra 0.050 W) 0.045 0.040 o 0.035 c 0030 1999 „ = 60,9/21- 0/1 recovery 2000 n = 75,9'19 9/24 recovery 2001 n = 67. 9/21 9;27 recovery 2002 n = 122,9/26 ■ 9/30 recovery 0.2 0.4 0.6 08 Mean scale radius (mm, OE = 0) Figure 5 (continued) —\ — 1.0 higher between the May and June research cruises (2-3 mm FL/d) than between cruises later in the sum- mer (0.5-1.5 mm FL/d)(Fig. 7C, see also Fig. 6). The concurrent decreases in spacing of circuli at the scale margin and in apparent growth rate of coho salmon in the ocean is consistent with the hypothesis that sea- sonal changes in scale circulus spacing reflect seasonal changes in fish growth rate. Additional evidence for decreasing growth rate of maturing coho salmon over the course of the summer comes from FLs of CWT fish in the hook-and-line fish- eries (sport and troll fisheries). Generally, apparent growth rates in FL of maturing coho salmon originat- ing from northern coastal Oregon streams and from both the Oregon and Washington sides of the Columbia river basin were highest from late May to mid-June and Fisher and Pearcy: Seasonal changes in growth of Oncorhynchus kisutch off Oregon and Washington 45 700 600 500 400 300 200 0 1981 □ 1982 A 1983 V 1984 O 1985 o 1998 □ 1999 A 2000 V 2001 o 2002 From Ishida etal. 1998 1 30 mm/d 5/30 7/19 9/7 1 1 1 10/27 12/16 2/04 Date -r T" 3/25 5/15 7/03 — I 1 — 8/22 10/11 Figure 6 Average lengths (±2 SE) of juvenile and maturing coho salmon (O. kisutch) caught during research cruises off Oregon and Washington in different months and years (gray and white symbols. The dashed lines are linear regressions and indicate apparent growth rates in FL between the differ- ent catch periods. The late April 2000 sample of maturing coho salmon was from a single trawl off the mouth of the Columbia River (Robert L Emmett, NMFS/NWFSC/HMSC, 2030 S Marine Science Drive, Newport, OR 97365. personal commun.). The small open circles are average lengths (±2 SE) of coho salmon from Ishida et al. (1998) (their Appendix Table 6) plotted against the 15th day of the months in which they were sampled. decreased greatly by mid-August (Fig. 8, A and B). For three periods, 20 May-29 June, 29 June-8 August, and 8 August-27 September, median apparent growth rates were 1.43 mm/d (ra=19), 0.64 mm/d (re=24), and 0.24 mm/d (n=27), respectively. Growth rates of fish from coastal Washington rivers also decreased over the summer, but the decrease was not as great as for the Oregon and Columbia River fish, and the apparent growth rates of the Washington fish were higher at comparable times during the summer (Fig. 9, A and B). The apparent growth rates of Gray Harbor basin fish were over 2 mm/d from late June to mid- July and remained comparatively high (about 1.0 mm/d) into late October (Fig. 9B). Washington fish generally were not caught in the fisheries until mid- or late June, about a month after the first catches of the Oregon and Columbia River fish. For three peri- ods 19 June-29 July, 29 July-7 September, and 7 Sep- tember-27 October, median apparent growth rates of the coastal Washington fish were 1.23 mm/d (n=13), 0.92 mm/d (n = \Q), and 1.06 mm/d (n=9), respectively. The growth data for CWT fish from the sport and troll fisheries, especially those for the coastal Oregon and Columbia River stocks, were consistent with the growth data from the mixed stock catches of coho salmon in research nets off Oregon and Washington in that both data sets indicated a substantial decrease in growth rate (FL) of maturing coho salmon between the May-June period and the August-September period. The decreases over the summer in circulus spacing at the scale margin (Fig. 7B) and in apparent growth rates of maturing CWT coho salmon of known origin (Fig. 8B) is further evidence that scale circulus spacing and fish growth rate are correlated seasonally. Discussion Our data indicate that the seasonal cycle of chang- ing ocean circulus spacing on scales of juvenile and adult coho salmon mirrors a similar seasonal cycle in the growth rate of these fish. We lack direct data for coho salmon collected between late September of the first calendar year of ocean residence and mid-May of the second calendar year, but growth rate during part of the fall and winter may be as low as 0.5mm/d 46 Fishery Bulletin 103(1) - 1 r- Apr 30 May 20 Jun 9 Jun 29 Jul 19 Aug 8 Aug 28 Sep 17 Oct 7 Oct 27 0065 0 060 0.045 & 0.040 - ro 0.035 - 0.030 B Maturing fish Apr 30 May 20 Jun 9 Jun 29 Jul! 9 Aug 8 Aug 28 Sep 17 Oct 7 Oct 27 □ A Maturing fish $ 2 o Research purse seines and surface trawls O 1981 mixed stocks D 1982 mixed stocks A 1983 mixed stocks V 1984 mixed stocks 0 1985 mixed stocks o 1998 mixed stocks □ 1999 mixed stocks A 2000 mixed stocks V 2001 mixed stocks O 2002 mixed stocks Oreg on ocean fisheries: ■ Cowlitz (all years of Table 2) • Sandy (all years of Table 2) A Fall Creek (all years of Table 2) Apr 30 May 20 Jun 9 Jun 29 Jul 19 Aug 8 Aug 28 Sep 17 Oct 7 Oct 27 Date Figure 7 Average spacing of the last two intercircular spaces at the scale margin versus average catch date for (Al juvenile coho salmon (O. kisutch) caught during research cruises and (B) maturing coho salmon caught during research cruises (gray and white symbols) and in the ocean fisheries (black symbols, averaged by 10-day periods, all years combined). Also shown for comparison with the temporal changes in circulus spacing are (C) the apparent growth rates of maturing coho salmon between research cruises (based on changes in mean FL; see Fig. 6) plotted against the mid-point of each growth period. based on data in Ishida et al. (1998). Therefore, the roughly twofold range in spacing of circuli in the ocean growth zone of scales from maturing fish that we found probably represents about a fourfold range in fish growth rate in the ocean (from about 0.5 mm/d in the winter to 2.1mm/d in the spring and early summer). Thus, changes in the spacing of scale circuli are relatively small when compared to the corresponding changes in fish growth Fisher and Pearcy: Seasonal changes in growth of Oncorhynchus kisutch off Oregon and Washington 47 LOCR Oregon 4/24 - 5/20 release LOCR Oregon 5/21 • 6/14 release LOCR Wash. 4/24 - 5/20 release LOCR Wash. 5/21 -6/14 release NOOR 3/01 - 3/31 release NOOR 4/10- 5/10 release Ape 30 May 20 Jun 9 Jun 29 Jul 19 Aug 8 Aug 28 Sep 17 Oct 7 Oct 27 Nov 16 B T3 E E 1 * v * S . 2 9 ' ■ ' r * - - ♦ ' ♦ to T n a □ □ O ■ D Apr 30 May 20 Jun 9 Jun 29 Jul 19 Aug 8 Aug 28 Sep 17 Oct 7 Oct 27 Nov 16 Date Figure 8 (A) Grand-mean FLs (±2 SE) by 10-day intervals over the years 1975-2002 of CWT lower Columbia River (LOCR) Oregon and Washington stocks and northern coastal Oregon stocks (NOOR) of maturing coho salmon (O. kisutch). Only intervals with five or more years of data are shown. (B) The corresponding aver- age apparent growth rates between each 10-day interval. Note the apparent decrease in growth rate between early and late summer. rate. However, the large seasonal changes in growth rate of coho salmon in the ocean are readily detectable from the changes in circulus spacing on the scale. In June 2001, 2002, and 2003 average spacing of the last two circulus pairs at the scale margin was positive- ly correlated (P<0.01) with plasma IGF-I (insulin-like growth factor-I) concentrations from juvenile fish caught in the ocean in research nets (n=119, 163, and 206 and r=0.52, 0.52, and 0.59 in 2001, 2002, and 2003, respec- tively) (Beckman2 and Fisher, unpubl. data). Because plasma IGF-I levels have been shown to be positively 2 Beckman, B. 2004. Unpubl. data. Integrative Fish Biol- ogy Program, Northwest Fisheries Science Center, National Marine Fisheries Service, 2725 Montlake Boulevard East, Seattle, Washington 98112. 48 Fishery Bulletin 103(1) correlated with instantaneous growth rates (in length) of juvenile coho salmon (Beckman et al., 2004), the finding that plasma IGF-I is also correlated with the spacing of circuli at the scale margin of juvenile coho salmon is further evidence that circulus spacing and growth rate are positively related for coho salmon. Our data suggest that growth rate in FL of matur- ing coho salmon is usually highest between early or mid-April and late June. This is a period of increasing photoperiod and often rising sea-surface temperature (SST) at 50°N in the northeastern Pacific Ocean, but is well before the maximum SST in late August (Fig. 10). Both increased day length and temperature stimulate growth in salmonids (Brett, 1979; Bjornsson, 1997). The 750 -i A I 700 - xf] p" 650 - s ^^^ length o o $r Fork o w —9- NWC 3/21 • 4/20 Release 500 ■ — O- NWC 4/21 - 5/25 Release —A— GRAY 4/20 - 5/30 Release 450 - —A— WILP 4/05 • 5/20 Release Apr 30 May 20 Jun 9 Jun 29 Jul 19 Aug 8 Aug28Sep17 Oct 7 Oct 27 Nov 16 B 3 ■ (mm/d) ro A A A A 3 2 1 '- CD C 0) a a. a. o - < • A A o 6 a A 8 . » • A A 8 ' ° • 0 A * A m O Apr30 May20 Jun 9 Jun 29 Jul 19 Aug 8 Aug 28 Sep17 Oct 7 Oct27 Nov 16 Date Figure 9 (A) Grand-mean FLs by 10-day intervals over the years 1975-2002 of CWT coastal Washington stocks of matur- ing coho salmon (O. kisuteh) from the Willapa Bay basin (WILP), Grays Harbor basin (GRAY), and coast north of Grays Harbor (NWC). (B) The corresponding average appar- ent growth rates between each 10-day interval. decreases in apparent growth rate in length of maturing coho salmon after the summer solstice could be associ- ated with a number of factors. One possibility is that there is a shift during the summer away from skeletal growth to growth in weight (with a resultant increase in condition) or to gonadal development. Data in Ishida et al. (1998) for coho salmon caught in research nets in the North Pacific tend to support this proposition (their Appendix Table 6). Their data indicate that the rate of growth in FL of maturing coho salmon decreased from 1.45 mm/d between April and May to 0.49 mm/d between July and August. (See also Fig. 6, present study). Over the same time period the condition index (weight (g)x(107/FL[mm]3)) of the fish they sampled increased from 113.3 to 143.8, an increase of 27%. Thus, skeletal growth slowed over the summer, but the condition of the fish increased. In contrast to growth rates of Columbia River co- ho salmon, which decreased greatly between early and late summer, and were quite low (s0.5 mm/d) by August and September, the growth rates of fish from the Grays Harbor basin, although also declin- ing during the summer, remained high well into September and early October (-0.7-1.4 mm/d), al- lowing the Grays Harbor fish to attain a significantly larger final average FL. Several factors may result in the differing growth patterns of maturing fish from these two groups. Many of the fish from the Columbia River are early spawners, and peak spawning occurs from late October to early November, whereas the Grays Harbor fish are mainly late spawners, and peak spawning occurs from mid-November to late- December (Weitkamp et al., 1995). Because of their later spawning the Grays Harbor fish may shift from somatic to gonadal growth later in the summer or fall than do the earlier spawners from the Columbia River. Maturing coho salmon from the Grays Harbor drainage also have a much more northerly distribu- tion than do maturing fish from the Columbia River (Weitkamp and Neely, 2002) and, therefore, the two groups encounter very different ocean conditions (e.g., temperature, salinity, prey fields, prey distributions, and potential competitors for food) while feeding in coastal waters. The different environmental condi- tions experienced by the Columbia River and Grays Harbor fish may also contribute to their differing temporal growth patterns. Because of the poor conditions for growth of fish associated with the 1983 El Nino, adult coho salmon in 1983 were exceptionally small off Oregon and were in poor condition (Pearcy et al., 1985; Johnson, 1988). Our scale analysis indicates that the small size of fish in 1983 was largely due to a failure of growth of maturing fish after formation of the winter annulus. Although the average scale radius between OE and the winter annulus was slightly smaller for the 1982-83 year class than for other year classes, the average scale radius between the winter annu- lus and the scale margin, representing the growth of maturing fish in spring and early summer, was Fisher and Pearcy: Seasonal changes in growth of Oncorhynchus kisutch off Oregon and Washington 49 16 - 14 - ~ 12 - O H co W 10- 8 - Jr\ Day length (h) CO (D "T CM O 4 SST ( °C ) ™^~ Day length (hours) Yearly and ol Point, Canad tions s gc.ca/c the pe Jan Mar May Jul Sep Nov Month (15th) Figure 10 cycle of day length (sunrise to sunset; black line) at 50°N sea surface temperature (SST) (°C; ±2 SE) at Amphitrite Vancouver Island, B.C. SST data from Fisheries and Oceans, a, Pacific Region, Science Branch, British Columbia lightsta- alinity and temperature data, URL: http://www-sci.pac. dfo-mpo. sap/data/lighthouse/amphitr.day. SST is the daily average for riod 22 August 1934-31 July 1999. exceptionally low for this year class (Fig. 5A). Circulus spacing revealed two notable trends. First, in 1983 the maximum spacing of circuli following the winter annulus was only very slightly lower than in other years, which indicates that spring growth in FL of maturing fish in 1983 was not unusually low. Perhaps maturing coho salmon continued to grow in length in spring 1983, when photoperiod was increasing rapidly, despite low food availability. Bjornsson (1997) found that changes in photoperiod may possibly control the level of pituitary growth hormone (GH), which strongly stimulates skeletal growth in salmonids and that in- creased levels of GH can induce growth in length even during starvation. Second, the spacing of circuli at the scale margin for fish caught in July 1983 was unusu- ally low. similar to the spacing at the scale margin from fish caught in August of most years. This find- ing indicates very slow growth rates for maturing fish by July 1983. Length data1 for maturing CWT coho salmon from the Oregon side of the Columbia River basin caught in the ocean sport and troll fisheries indicated that between June and September 1983 the average length of fish changed very little, which would indicate that somatic growth ceased during the summer. Our results confirm the utility of circulus spacing as an indicator of growth rate in FL of coho salmon in the ocean. Correlations between average circulus spacing and estimated average growth rates of groups of fish were significant and positive (Table 4), even when growth was measured over long intervals of time (four to five months for juveniles, and over a year for maturing coho salmon), and even when the estimates of growth rate were subject to error In addition, our data indicate large seasonal changes in growth rate in FL of coho salmon in the coastal ocean off Oregon and Washington, a result also suggested by data in Ishida et al. (1998) for coho salmon in the North Pacific (see Fig. 6), and these seasonal changes in growth rate ap- pear to be tracked by seasonal changes in spacing of scale circuli. Acknowledgments We thank all personnel from the Estuarine and Ocean Ecology Division of the National Marine Fisheries Ser- vice and from Oregon State University who participated either in the research cruises or in processing samples from those cruises. We also thank Lisa Borgerson of the Oregon Department of Fish and Wildlife for supplying scales from coho salmon caught in the ocean fisheries, and the captains and crews of the FV Sea Eagle, FV Ocean Harvester, FV Frosti and the RV Ricker for their expert assistance during the cruises. Ric Brodeur and Edmundo Casillas provided helpful comments on an earlier version of this paper. This study was funded by the Bonneville Power Administration through a grant to the National Marine Fisheries Service and from NMFS to Oregon State University. 50 Fishery Bulletin 103(1) Literature cited Bailey, K. M., and E. D. Houde. 1989. Predation on eggs and larvae of marine fishes and the problem of recruitment. Adv. Mar. Biol. 25:1-83. Bax, N. J. 1983. Early marine mortality of marked juvenile chum salmon (Oneorhynehus keta) released in Hood Canal, Puget Sound, Washington, in 1980. Can. J. Fish. Aquat. Sci. 40:426-435. Beamish, R. J., and D. R. Bouillon. 1993. Pacific salmon production trends in relation to climate. Can. J. Fish. Aquat. Sci. 50:1002-1016. Beckman, B. R., W. Fairgrieve, and K. A. Cooper, C. V. W. Mahnken, and R. J. Beamish. 2004. Evaluation of endocrine indices of growth in indi- vidual postsmolt coho salmon. Trans. Am. Fish. Soc. 133:1057-1067. Bhatia, D. 1932. Factors involved in the production of annual zones on the scales of the rainbow trout (Salmo irideus). II. J. Exp. Biol. 9:6-11. Bigelow, P. E., and R. G. White. 1996. Evaluation of growth interruption as a means of manipulating scale patterns for mass-marking hatchery trout. N. Am. J. Fish. Manag. 16:142-153. Bilton, H. T. 1975. Factors influencing the formation of scale char- acters. Int. North Pac. Fish Comm. Bull. 32:102- 108. Bilton, H. T„ and G. L. Robins. 1971. Effects of feeding level on circulus formation on scales of young sockeye salmon (Oneorhynehus nerka). J. Fish. Res. Board Can. 28:861-868. Bjbrnsson, B. T. 1997. The biology of salmon growth hormone: from day- light to dominance. Fish Physiol. Biochem. 17:9-24. Blackbourn, D. J. 1990. Comparison of release size and environmental data with the marine survival rates of some wild and enhanced stocks of pink and chum salmon in British Columbia and Washington state. In Proceedings 14th northeast Pacific pink and chum salmon workshop (P. A. Knudsen, ed.), p. 82-87. Washington Dept. Fish., Olympia, WA. Brett, J. R. 1979. Environmental factors and growth. In Fish physi- ology, vol. 8, Bioenergetics and growth (W. S. Hoar, D. J. Randall, and J. R. Brett, eds.), p. 599-675. Academic Press, New York, NY. Clutter, R. I., and L. E. Whitesel. 1956. Collection and interpretation of sockeye salmon scales. Int. Pac. Salmon Fish. Comm. Bull. 9, 159 p. Dawley, E. M., R. D. Ledgerwood, and A. Jensen. 1985a. Beach and purse seine sampling of juvenile sal- monids in the Columbia River estuary and ocean plume. 1977-1983. Volume I: Procedures, sampling effort, and catch data. NOAA Tech. Memo. NMFS F/NWC- 74. i-x, 397 p. 1985b. Beach and purse seine sampling of juvenile salmonids in the Columbia River estuary and ocean plume, 1977-1983. Volume II: Data on marked recoveries. NOAA Tech. Memo. NMFS F/NWC-75. i-vii, 260 p. Doyle, R. W., A. J. Talbot, and R. R. Nicholas. 1987. Statistical interrelation of length, growth, and scale circulus spacing: appraisal of a growth rate estimator for fish. Can. J. Fish. Aquat. Sci. 44:1520-1528. Emmett, R. L., and R. D. Brodeur. 2000. Recent changes in the pelagic nekton community off Oregon and Washington in relation to some physical oceanographic conditions. North Pacific Anadr. Fish Comm. Bull. 2:11-20. Fisher, J. P., and W. G. Pearcy. 1988. Growth of juvenile coho salmon (Oneorhynehus kisutch) in the ocean off Oregon and Washington, USA, in years of differing coastal upwelling. Can. J. Fish. Aquat. Sci. 45:1036-1044. 1990. Spacing of scale circuli versus growth rate in young coho salmon. Fish. Bull. 88:637-643. Friedland, K. D., and R. E. Haas. 1996. Marine post-smolt growth and age at maturity of Atlantic salmon. J. Fish Biol. 48:1-15. Friedland, K. D., L. P. Hansen, D. A. Dunkley, and J. C. MacLean. 2000. Linkage between ocean climate, post-smolt growth, and survival of Atlantic salmon (Salmo salar L.) in the North Sea area. ICES Journal of Marine Science. 57:419-429. Friedland, K. D„ and D. G. Reddin. 2000. Growth patterns of Labrador Sea Atlantic salmon postsmolts and the temporal scale of recruitment syn- chrony for North American salmon stocks. Can. J. Fish. Aquat. Sci. 57:1181-1189. Friedland, K. D., D. G Reddin, and J. F. Kocik. 1993. Marine survival of North American and Euro- pean Atlantic salmon: effects of growth and envi- ronment. ICES J. Mar. Sci. 50:481-492. Fukuwaka, M., and M. Kaeriyama. 1997. Scale analyses to estimate somatic growth in sock- eye salmon, Oneorhynehus nerka. Can. J. Fish. Aquat. Sci. 54:631-636. Glenn, C. L., and J. A. Mathias. 1985. Circuli development on body scales of young pond- reared walleye (Stizostedion bitreum). Can. J. Zool. 63:912-915. Hare, S. R., N. J. Mantua, and R. C. Francis. 1999. Inverse production regimes: Alaska and West Coast Pacific salmon. Fisheries 24:6-15. Healey, M. C. 1982. Timing and relative intensity of size-selective mortality of juvenile chum salmon {Oneorhynehus keta) during early sea life. Can. J. Fish. Aquat. Sci. 39:952-957. Hobday, A. J., and G. W. Boehlert. 2001. The role of coastal ocean variation in spatial and temporal patterns in survival and size of coho salmon (Oneorhynehus kisutch). Can. J. Fish. Aquat. Sci. 58:2021-2036. Holtby, L. B„ B. C. Andersen, and R. K. Kadawaki. 1990. Importance of smolt size and early ocean growth to interannual variability in marine survival of coho salmon (Oneorhynehus kisutch). Can. J. Fish. Aquat. Sci. 47:2181-2194. Ishida, Y., S. Ito, Y Ueno, and J. Sakai. 1998. Seasonal growth patterns of Pacific salmon (On- eorhynehus spp.) in offshore waters of the North Paci- fic Ocean. N. Pac. Anadr. Fish Comm. Bull. 1:66- 80. Jaenicke, H. W., M. J. Jaenicke, and G. T. Oliver. 1994. Predicting northern southeast Alaska pink salmon returns by early marine scale growth, p. 97-110. North- Fisher and Pearcy: Seasonal changes in growth of Oncorhynchus kisutch off Oregon and Washington 51 east Pacific pink and chum salmon workshop, Alaska Sea Grant Prog., Alaska Univ., Fairbanks, AK. Johnson, S. L. 1988. The effects of the 1983 El Nino on Oregon's coho {Oncorhynchus kisutch) and chinook (O. tshawytscha) salmon. Fish. Res. 6:105-123. Litvak, M. K., and W. C. Leggett. 1992. Age and size-selective predation on larval fishes: the bigger-is-better hypothesis revisited. Mar. Ecol. Prog. Ser. 81:13-24. Mathews, S. B., and Y. Ishida. 1989. Survival, ocean growth and ocean distribution of differentially timed releases of hatchery coho salmon (Oncorhynchus kisutch). Can. J. Fish. Aquat. Sci. 46:1216-1226. Matricia, T. A. J. Talbot and R. W. Doyle. 1989. Instantaneous growth rate of tilapia genotypes in undisturbed aquaculture systems. I, "Red" and "grey" morphs in Indonesia. Aquaculture 77:295-306. Mantua, N. J., S. R. Hare, Zhang, Y., J. M. Wallace and F. C. Francis. 1997. A Pacific interdecadal climate oscillation with impacts on salmon production. Bull. Am. Meteorol. Soc. 78:1069-1080. Miller, T. J., L. B. Crowder, J. A. Rice, and E. A. Marschall. 1988. Larval size and recruitment mechanisms in fishes: toward a conceptual framework. Can. J. Fish. Aquat. Sci. 45:1657-1670. Parker, R. R. 1971. Size selective predation among juvenile salmonid fishes in a British Columbia inlet. J. Fish. Res. Board Can. 28:1503-1510. Pearcy, W., J. Fisher, R. Brodeur, and S. Johnson. 1985. Effects of the 1983 El Nino on coastal nekton off Oregon and Washington. In El Nino North, Nino effects in the eastern subarctic Pacific Ocean, p. 188- 204. Washington Sea Grant Program, Univ. Washing- ton, Seattle, WA. Pearcy, W. G., and J. P. Fisher. 1988. Migrations of coho salmon, Oncorhynchus kisutch, during their first summer in the ocean. Fish. Bull. 86:173-195. 1990. Distribution and abundance of juvenile salmonids off Oregon and Washington, 1981-1985. NOAA Tech. Rep. NMFS 93, 83 p. Pyper, B. J., R. M. Peterman, M. F. Lapointe, and C.J . Walters. 1999. Patterns of covariation in length and age at matu- rity of British Columbia and Alaska sockeye salmon (On- corhynchus nerka) stocks. Can. J. Aquat. Sci. 56: 1046-1057. Ricker, W. E. 1973. Linear regressions in fishery research. J. Fish. Res. Board Can. 30:409-434. 1992. Back-calculation offish lengths based on propor- tionality between scale and length increments. Can. J. Fish. Aquat. Sci. 49:1018-1026. Scarnecchia, D. L. 1979. Variation of scale characteristics of coho salmon with sampling location on the body. Prog. Fish-Cult. 41:132-135. Sogard, S. M. 1997. Size-selective mortality in the juvenile stage of tele- ost fishes: a review. Bull. Mar. Sci. 60:1129-1157. Sokal, R. R., and F. J. Rohlf. 1995. Biometry. The principles and practice of statistics in biological research, 3rd ed., 887 p. W.H. Freeman and Company, New York, NY. Talbot, A. J., and R. W. Doyle. 1992. Statistical interrelation of length, growth, and scale circulus spacing: use of ossification to detect nongrowing fish. Can. J. Fish. Aquat. Sci. 49:701-707. Teel, D. J., D. M. Van Doornik, D. R. Kuligowski, and W. S. Grant. 2003. Genetic analysis of juvenile coho salmon (Oncorhyn- chus kisutch) off Oregon and Washington reveals few Columbia River wild fish. Fish. Bull. 101:640-652. Weitkamp, L., and K. Neely. 2002. Coho salmon (Oncorhynchus kisutch) ocean migra- tion patterns: insight from marine coded-wire tag recoveries. Can. J. Fish. Aquat. Sci. 59:1100-1115. Weitkamp, L. A., T. C. Wainwright, G. J. Bryant, G. B. Milner, D. J. Teel, R. G. Kope, and R. S. Waples. 1995. Status review of coho salmon from Washington, Oregon, and California. NOAA Tech. Memo. NMFS- NWFSC-24, 258 p. Willette, T M. 1996. Impacts of the Exxon Valdez oil spill on the migra- tion, growth, and survival of juvenile pink salmon in Prince William Sound. Am. Fish. Soc. Symp. 18:533-550. 2001. Foraging behaviour of juvenile pink salmon (Oncorhynchus gorbuscha) and size-dependent preda- tions risk. Fish. Oceanogr. 10(Suppl. 1):110-131. 52 Abstract — Metal-framed traps cov- ered with polyethylene mesh used in the fishery for the South African Cape rock lobster (Jasus lalandii) incidentally capture large numbers of undersize (<75 mm CD specimens. Air-exposure, handling, and release procedures affect captured rock lob- sters and reduce the productivity of the stock, which is heavily fished. Optimally, traps should retain legal- size rock lobsters and allow sublegal animals to escape before traps are hauled. Escapement, based on lobster morphometric measurements, through meshes of 62 mm, 75 mm, and 100 mm was investigated theoretically under controlled conditions in an aquarium, and during field trials. SELECT models were used to model escapement, wherever appropriate. Size-selectivity curves based on the logistic model fitted the aquarium and field data better than asymmetrical Richards curves. The lobster length at 50% retention (L50) on the escape- ment curve for 100-mm mesh in the aquarium (75.5 mm CL) approximated the minimum legal size (75 mm CL); however estimates of Z/50 increased to 77.4 mm in field trials where trap- entrances were sealed, and to 82.2 mm where trap-entrances were open. Therfore, rock lobsters that cannot escape through the mesh of sealed field traps do so through the trap entrance of open traps. By contrast, the wider selection range and lower L25 of field, compared to aquarium, trials ^SR = 8.2 mm vs. 2.6 mm; L.,5=73.4 mm vs. 74.1 mm), indicate that small lobsters that should be able to escape from 100-mm mesh traps do not always do so. Escape- ment from 62-mm mesh traps with open entrance funnels increased by 40-60% over sealed traps. The find- ings of this study with a known size distribution, are related to those of a recent indirect (comparative) study for the same species, and implications for trap surveys, commercial catch rates, and ghost fishing are discussed. Escapement of the Cape rock lobster (Jasus lalandii) through the mesh and entrance of commercial traps Johan C. Groeneveld Marine and Coastal Management 5lh floor Foretrust Building Martin Hamerschlacht Street, Foreshore Cape Town, South Africa E-mail address. Jgroenevdeat.gov.za Jimmy P. Khanyile National Research Foundation P.O. Box 2600 Pretoria 0001, South Africa David S. Schoeman Department of Zoology University of Port Elizabeth Port Elizabeth 6031, South Africa Manuscript submitted 20 March 2003 to the Scientific Editor. Manuscript approved for publication 1 July 2003 by the Scientific Editor. Fish Bull. 103:52-62 (2005). The traps used in lobster and crab fisheries are a versatile fishing gear that can be modified to target specific species and size ranges through choice of design and bait (Miller, 1990). Selec- tion by traps of only the desired size classes reduces sorting time and may increase the catch rates of legal-size animals (Fogarty and Borden, 1980; Everson et al., 1992; Rosa-Pacheco and Ramirez-Rodriguez, 1996). Cap- ture, sorting and release procedures have furthermore been implicated in accidental and stress-induced mor- talities (Brown and Caputi, 1983; 1985; Hunt et al., 1986), as well as in sublethal injuries, such as limb loss (legs or antennae), which may retard somatic growth (Davis, 1981; Brown and Caputi, 1985). Air expo- sure, even over short periods, can induce behavioral changes such as reduced responsiveness to threatening stimuli (Vermeer, 1987) and lead to higher predation risk among released animals (Brown and Caputi, 1983). Furthermore, displacement from home reefs disrupts feeding behav- ior and can affect growth increments (Brown and Caputi, 1985). Manag- ers of many crustacean trap fisher- ies have responded to these problems by introducing escape vents of vari- ous sizes and shapes (Krouse, 1989; Miller, 1990; Everson et al., 1992; Arana and Ziller, 1994; Rosa-Pacheco and Ramirez-Rodriguez, 1996; Treble et al., 1998; Schoeman et al., 2002a), because they successfully allow under- size specimens to escape (Arana and Ziller, 1994; Treble et al. 1998). In fisheries management, size selec- tivity curves are important for esti- mates of incidental mortality, recruit- ment in yield-per-recruit analysis, and age- and length-based popula- tion models (Millar and Fryer, 1999). Notably, size selectivity can be used to evaluate the minimum legal size (MLS) and the effects of changing escape vent or mesh size regulations on the future productivity of the re- source (Treble et al., 1998). Most selectivity studies on which mesh- or escape vent size are based are comparative (indirect), imply- ing that the size distribution of the population is unknown and that variants of the same gear type are fished simultaneously (Millar and Fryer, 1999). Results from indirect studies can, however, be influenced by trap soak times, trap saturation effects (Miller, 1990), seasonal size and sex-specific patterns in catchabil- ity (Pollock and Beyers, 1979), and by differences in morphometric ratios of subpopulations (Fogarty and Borden, Groeneveld et al.: Escapement of Jasus lalandn from traps 53 1980; Maynard et al., 1987). These disadvantages are offset by the convenience with which indirect studies measure selectivity under operational conditions. Far fewer direct studies, in which the size distribution of the fished population is known (Millar and Fryer, 1999), have been published, and those that have been pub- lished have included several laboratory studies where the escape of crustaceans from traps was monitored (Krouse and Thomas, 1975; Krouse, 1978; Everson et al., 1992). Direct studies do not recreate true commer- cial conditions, but rather provide a contact-selectivity curve (or retention curve) that quantifies the difference in length distribution between the catch and the popula- tion offish coming in contact with the gear (Millar and Fryer, 1999). This information is useful as a benchmark against which operational, seasonal, and spatial selec- tivity patterns can be measured. Commercial fishing for the South African Cape rock lobster (Jasus lalandii) originated in the late nine- teenth century and reached its pinnacle in the 1950s, when nearly 11,000 tons were landed annually (Pol- lock, 1986). However, since then catches have declined markedly, especially during the 1990s, when annual catch restrictions based on the assumption of decreased population strength, reduced the yield to 2000-3000 tons per year (Pollock et al., 2000). In response to these operational changes, several recent modifications have been made to the regulations governing gear used in the fishery (Schoeman et al., 2002b). The changes most pertinent to this study took place in 1984, when mesh size was increased from 62 to 100 mm (stretched) to reduce the relative catch of undersize J. lalandii (Schoeman et al., 2002b), and during the early 1990s, when the minimum size limit was reduced from its historic level of 89 mm carapace length (CL) to 75 mm CL (Cockcroft and Payne, 1999; Pollock et al., 2000). Despite these two measures, the proportion of the com- mercial catch <75 mm CL that has to be released re- mains around 35-40% (MCM1). At present, the biomass of the J. lalandii resource that is larger than the mini- mum legal size is estimated at about 6% of its pristine value, whereas the spawning biomass (of mature female rock lobsters) is estimated to be 21% (Johnston, 1998). Consequently, it is clear that the resource is heavily depleted and that there is little scope for wasted produc- tion through unnecessary damage to undersize lobsters (Schoeman et al., 2002a). Most studies on trap selectivity of J. lalandii (New- man and Pollock, 1969; Crous, 1976; Pollock and Bey- ers, 1979) predate the changes to mesh and minimum legal size described above and did not provide selectiv- ity curves. In the only recent study, Schoeman et al. (2002a) used the SELECT (Share Each LEngth class's Catch Total) method (Millar, 1992, Milllar and Walsh, 1992) to investigate the selectivity properties of vari- 1 MCM (Marine and Coastal Management). 2002. Unpubl. data. MCM, Martin Hamershclacht St., Cape Town, South Africa. ous modifications to commercial and research traps in comparison with the standard 100-mm stretched mesh trap design. This study was indirect, in that it simu- lated commercial fishing and compared catch rates in other traps to those made with a small-mesh (62 mm, stretched) trap, which acted as a control. Several processes are involved in the selectivity of traps: namely the attraction of rock lobsters by bait; their ability to enter traps through trap openings of various sizes, shapes, and localities within the trap; their behavior in and around traps; their escapement through the trap opening and their escapement through mesh openings or escape vents (Miller, 1990). The pres- ent study focuses on escapement of captured J. lalandii through the mesh of stretched mesh traps and through trap entrances. The aims are to investigate the relation- ships between CL and other morphometric measures for male rock lobsters in order to use these relationships to estimate theoretical escapement curves for any given mesh size; to compare these curves to observed escape- ment rates through selected meshes in the aquarium; and to extend these comparisons to field conditions. The overall aim is to determine the optimum mesh charac- teristics that maximize efficiency in targeting legal-size male J. lalandii. Material and methods Mesh size of lobster traps Mesh size is defined as the measurement from inside of knot to inside of knot when the net is stretched in the direction of the long diagonal of the meshes, i.e., lengthwise of the net. Netting is made of polyethylene. Commercial rock lobster traps (Fig. 1) are covered with 100-mm stretched mesh (or 50-mm bars, also measured from the insides of knots), which are stretched in such a manner over the metal frame that the openings are square. Morphometric variables measured Following manual trials that involved fitting lobster car- apaces of different sizes through an adjustable square hole, three carapace dimensions were identified as likely to play a role in regulating escapement. These were the following: 1) carapace width (CW), measured laterally, across the widest point of the carapace; 2) carapace depth (CD), measured dorsoventrally, extending from the highest point of the dorsal carapace surface to the lowest point on the ventral surface of the thoracic plate; and 3) carapace base (CB), measured ventrally, between the distal edges of the second segment of the last walking legs, with the legs folded flush against the carapace. Each of these dimensions was measured (±1 mm) for each of 169 male rock lobsters caught in research traps deployed off the Cape Peninsula between 1999 and 2002. Corresponding data regarding carapace length 54 Fishery Bulletin 103(1) Figure 1 (A) Standard metal-framed traps (0.8 m x 0.5 mxl.35 m high) covered with stretched polyethylene mesh used in the commercial fishery for J. lalandii and during field experiments (note the 100-mm mesh size covering on the com- mercial traps, and the 62-mm mesh size on the codend and entrance funnel). (B) Metal-framed escapement cages covered with 62-mm (shown), 75-mm, and 100-mm mesh used in the aquarium experiments (frames were 0.6 mx0.6 m, with a depth of 0.25 m). (CL), measured mid-dorsally from the posterior edge of the carapace to the anterior tip of the rostral spine, were also collected. This was done because CL is the dimension most frequently mentioned in legislation pertaining to this species (Schoeman et al., 2002b) and has therefore been the focus of most size-based studies (Newman and Pollock, 1969; Pollock and Beyers, 1979; Schoeman et al., 2002a). Relationships between the CL and each of CW, CD, and CB were explored by using simple least-squares regression analyses. Theoretical calculations of escapement In order to investigate morphological characteristics that physically limit escapement through meshes of various dimensions as a function of CL, digital photographs were taken of the posterior cross section of 46 male carapaces (tail removed) covering a range of sizes between 40 mm CL and 106 mm CL. Using standard graphics software, we superimposed a square on each image to represent a square of polyethylene mesh, similar to that used in a South African rock lobster trap. This simulated mesh was orientated so that its base was parallel with the carapace base of the lobster under consideration. It was then proportioned so that each of its sides was equal in length to the corresponding CB. Once this procedure had been completed, the simu- lated mesh square was rotated and resized so that we could determine the dimensions of the smallest square through which each lobster could pass. This measure was designated the "critical mesh size" for that image. Critical mesh size was related to CL by using simple linear regression analysis. In this way, the theoretically appropriate mesh aperture required to target all lobster larger than a given size could be predicted from the minimum CL of the target group (for convenience, this CL will be designated the "critical CL"). Aquarium trials Having addressed the matter of whether or not lobsters theoretically should be able to escape a mesh of given dimensions, the next question to be posed is whether or not they can do so under ideal (laboratory) conditions? For these purposes, three stretched mesh sizes were considered: 1) 62 mm, which coincides with the mesh size used in the commercial fishery prior to 1984 and also with the mesh currently used on traps deployed in the Fishery Independent Monitoring Survey (FIMS) (Schoeman et al., 2002a); 2) 100 mm, which corresponds with the mesh currently used on commercial traps for J. lalandii; and 3) 75 mm, which was used to provide Groeneveld et al.: Escapement of Jasus lalandu from traps 55 information on selectivity for meshes of intermediate aperture dimensions. Each of these experimental meshes was used to con- struct an escapement cage by stretching the mesh over a mild-steel frame in order to present square escape apertures of varying dimensions, as determined by the size of the mesh used (Fig. 1). These cages were deployed in an aquarium tank measuring 1.8 mxl.8 m and having a depth of 1.5 m. Fresh sea water was continuously supplied to this tank by a through-flow system that regulating water temperature between 12° and \&°C, well within the natural temperature range of J. lalandii (Heydorn, 1969). For each mesh size, male rock lobsters of various carapace lengths (373 lobsters measuring 34-91 mm CL for 62-mm mesh; 351 lobsters measuring 34-75 mm CL for 75-mm mesh; and 142 lobsters measuring 70-91 mm CL for 100-mm mesh) were collected live from the sea and transported to the experimental aquarium tank. Care was taken to ensure that approximately equal numbers of lobsters were available for each 2-mm size- class within the respective size ranges, although fewer lobsters tended to be available in size classes towards the ends of the frequency distributions. Once at the aquarium, lobsters were placed inside the experimental cages in groups of up to 20 and left for 30 minutes. Individuals that did not escape during this period were gently pushed towards the mesh open- ings, encouraging escapement, where this was possible. Subsequently, the CL frequency distributions were de- termined both for those lobsters that escaped the mesh as well as those that were retained. Several replicate escapement experiments were conducted for each mesh size, but because the experimental cages were too small to hold large numbers of lobsters, replicate selection curves could not be computed. Instead, all data were pooled for each mesh size for further analyses. Field trials The final question to be posed is whether or not lobsters do escape from traps when afforded the opportunity to do so under field conditions? To address this problem, field trials were undertaken off the Western Cape Pen- insula during monthly sampling sessions conducted by the research vessel Sardinops in July 2000 and from December 2001 to March 2002— a total of five distinct sampling surveys. Four categories of standard rock lobster traps (Fig. 1) were employed: 1) 62-mm stretched mesh, with en- trance funnels open; 2) 62-mm stretched mesh, with en- trance funnels blocked by a fine-mesh insert; 3) 100-mm stretched mesh, with entrance funnels open; and 4) 100- mm stretched mesh, with entrance funnels blocked. Duplicate bottom long-lines consisting of 10 traps each were prepared, of which six were normal commer- cial traps, and the remaining four were experimental traps, and these 10 traps were spread in haphazard order along the line, excluding the end traps. Into each trap was placed a sample of approximately 40 male rock lobsters, each of which had been measured (CL) and marked by cutting a notch in its uropod. In this way, it was possible to distinguish between lobsters that had been placed in the trap and those that had entered the trap of their own accord. Experimental traps were deployed without bait, in order to limit their ability to attract lobsters and also to remove one of the prime incentives that captive lob- sters might have to remain in a trap, even when it could escape. These trap lines were soaked overnight and on their retrieval, each remaining lobster was re- measured (CL) and inspected to identify specimens that had entered the traps voluntarily. Eight replicates were completed for each of the four categories of traps. Construction of selectivity curves The contact-selection curves (sensu Millar and Fryer, 1999) for the meshes used in the laboratory and field trials were modeled by using the SELECT method (Millar and Walsh, 1992) as applied to covered codend experiments (Millar and Fryer, 1999). We felt that this approach was warranted because we collected data with respect to lobsters in both a "codend" (those retained in the traps) and a "cover" (those that escaped, but for which data were available by inference). The logistic and Richards formulations of the general selectivity curve were fitted by using Excel (Microsoft, Redmond, WA) routines (Tokai2). These two selectivity functions were chosen because of their relative simplic- ity, their broad use over a range of different fisheries, and the availability of estimation routines for their parameters (Millar and Fryer, 1999). The Richards curve has the equation r(l) ( exp(a+b, (l + exp(a + bl)_ bl) where r(l) is the probability that an individual of length I attempting to pass through a mesh of given size will be retained by it (Millar and Fryer, 1999); and a, b, and 5 are constants. The logistic curve is the special case of this formulation, where 5=1. According to these models, the lobster length at 50% retention (L50) and the selection range {SR=L75-L25) are defined as follows: In 0.5" 1 - 0.5' simplifying to L50 = - — when 5 = 1, and b : Tokai, T. 2002. Personal commun. Department of Marine Science and Technology, Tokyo University of Fisheries, Konan Minatoku, Tokyo 108, Japan. 56 Fishery Bulletin 103(1) In SR = 0.75° 1-0.755 In 0.25') 1-0.2515 simplifying to SR = — - — , when <5 = 1. All calculations were made on the basis of 2-mm-CL size classes covering the entire size range for each fre- quency distribution. The 2-mm-CL size classes were used to ensure consistency across models, and also to balance data resolution against the number of size classes expected to have either zero catch or zero escape- ment (Millar and Fryer, 1999). Wherever necessary, hypothesis tests were conducted in accordance with the recommendations of Millar and Walsh (1992) and Millar and Fryer (1999). Results Morphometric relationships Least-squares regression analysis indicated highly sig- nificant linear relationships between CL and each of the other morphometric variables measured (Fig. 2). In each case, at least 97% of the variability in the predic- tor variable was explained by CL, indicating a high degree of correlation among predictors. Nevertheless, for any given CL, CB was consistently the largest variable measured, whereas CD was the smallest. Furthermore, CB increased more rapidly in response to increasing CL than either CW or CD (ANCOVA: F=115.165; df=2, 167; P<0.001). We therefore concluded that CB would likely be the morphometric variable that limits escapement through stretched square meshes. 100-1 o CW (mm) = 0-74 x CL (mm) - 4 45 rmn go- r2=0.98; n=169; P< 0.001 • CD (mm) = 0.61 x CL (mm)- 16.74 mm 's? 80- E tn 70- o ? 60 ■ CO Q 50- O 3 40- o r2= 0.97; n= 169; P< 0.001 m^ • CB (mm) = 0 80 x CL (mm) - 2.25 nrm ■ ■^■"o * r*= 0.98; n = 169; P< 0.001 miM^^ \^S^^ JV>' Jfty^T^ _»Sw/» 30 40 50 60 70 80 90 100 110 CL (mm) Figure 2 Individual linear relationships between carapace length (CL) and each of carapace width (CW), carapace depth (CD), and carapace base (CB) for J. lalandii. Theoretical calculation of escapement The mesh size that appeared (on the basis of visual inspec- tion) to limit escapement was expressed as a function of CL with a simple, linear, least-squares regression model (Fig. 3). This relationship was highly significant and explained 99% of the variability in critical mesh size. Using inverse prediction methods (Zar, 1999), we calculated the critical CL (mean ±95% confidence inter- val) from the regression model illustrated in Figure 3 for any mesh size. For 62-mm mesh, the critical CL is estimated at 43.8 (±4.12) mm; for the 75-mm mesh the estimate is 52.3 (±4.15) mm; whereas for the 100-mm mesh it is 68.7 (±4.12) mm. Given the implicit assump- tion that lobsters smaller than the critical CL can es- cape, but that larger lobsters are retained, the mean critical CL can be used as an estimate of L50. Aquarium trials No lobsters larger than 48 mm CL escaped the 62-mm mesh traps in the aquarium and none smaller than 44 mm CL were retained. This finding resulted in an extremely steep selection curve with a narrow SR (Fig. 4; Table 1). For the 75-mm mesh, no lobsters larger than 61 mm CL escaped and no lobsters smaller than 54 mm CL were retained. This finding resulted in a slightly more gentle selection curve, but with a reasonably tight SR (Fig. 4, Table 1). Similarly, for the 100-mm mesh, no lobsters larger than 79 mm CL escaped and no lobsters smaller than 74 mm CL were retained. This finding resulted in a selection curve that closely resembled that for the 75-mm mesh, except that the curve shifted a few size categories to the right (Fig. 4; Table 1). For all meshes, the symmetrical logistic model was se- lected in preference to the asymmetrical Richards model (Table 1), and in all cases the selected model fitted the data reasonably well (Fig. 4). It should, however, be noted that all hypothesis tests were conducted by using the deviance residu- als and their degrees of freedom for all size classes sampled. This was necessary because the very tight selection curves (especially for the 62-mm mesh) resulted in relatively small numbers of size classes in which retention probability was neither zero nor one. The above results indicate that L50-esti- mates for each mesh size are substantially larger than the corresponding estimates of critical CL from the theoretical escapement model. In fact, assuming that the asymptotic standard errors provided in Table 1 could be converted to 95% confidence intervals by a multiplication factor of two, only the confi- dence intervals for these statistics from the 62 mm mesh would overlap. By contrast, con- fidence intervals for the critical CL are well below those for the L50 for both the 75 mm mesh and the 100-mm mesh. This impression is confirmed by inspecting the probabilities of Groeneveld et al.: Escapement of Jasus lalandu from traps 57 Table 1 Statistics from SELECT analysis for the aquarium escapement trials. Values in parentheses are asymptotic standard errors sensu Millar (1993). The ^e standard errors are provided only for the best model fits for each of the various categories of data. Note that all hypothesis tests were cond ucted by using deviance residual 3 for the full model and their degrees of freedom (see text for explanation). 62-mm mesh 75 mm mesh 100-mm mesh Logistic Richards Logistic Richards Logistic Richards a -76.479 (23.159) -351.538 -58.217 (10.079) -400.075 -64.101 (11.655) -41.224 b (/mm) 1.649 (0.500) 7.463 0.991 (0.173) 6.589 0.849 (0.155) 0.615 o 6.567 13.173 0.010 L50 (mm) 46.389 (0.309) 46.493 58.717 (0.302) 59.333 75.459 (0.376) 75.144 SR (mm) 1.333 (0.404) 0.989 2.216 (0.386) 2.200 2.587 (0.471) 2.567 Selection factor 0.75 0.78 0.76 H0: data have binomial d stribution (i.e.. data are not overdispersed) Deviance 0.802 0.209 1.984 1.092 8.675 6.435 df 27 26 19 18 12 11 P-value 1 1 0.999 0.728 0.730 0.843 ff0:6=l Deviance 0.593 0.892 2.240 df 1 1 1 P-value 0.441 0.345 0.134 retention, r(l), by each mesh size of a lobster at its cor- responding mean critical CL. For the 62-mm mesh this probability is 0.014 (0.926 at the upper 95% confidence limit for the critical CL); for the 75-mm mesh it is 0.002 (0.096 at the upper 95% confidence limit for the critical CL); and for the 100-mm mesh it is 0.003 (0.099 at the upper 95% confidence limit for the critical CL). Field trials Escapement from traps with 62-mm mesh was highly variable both for the traps with entrance funnels left open, as well as for those with entrance funnels that were sealed, but was surprisingly high for the latter (Fig. 5). Furthermore, it is clear that the relationship between proportion of lobsters retained and CL was not logistic, as it was for the larger mesh sizes (Figs. 5 and 6). Instead, simple, least squares regression analysis indicated linear relationships between these variables both for traps with open entrance funnels as well as for those with entrance funnels closed (Fig. 5). There was no difference between the slopes (r=1.138; df=10; P=0.282; common slope = 0.795/mm), although their intercepts did differ significantly (r=14.079; df = 11;P«0.001). 170 - g 150 - Mesh size (mm) = 1 .53 x CL (mm) - 5.07 mn • r 2 = 0.99; n = 46; P=< 0.001 ^^ 8 130- *f^^ 01 • fcAt M 110 - _ 90 - to o ■■£ 70 - O i >_^* 40 SO 60 70 80 90 100 110 CL(mm) Figure 3 The relationship between carapace length of J. lalandil and mesh size below which escape should theoretically not be possible. No lobsters smaller than 62 mm CL were retained in the 100-mm mesh traps with open entrances, and no upper size limit was reached beyond which escape- ment was completely eliminated. By contrast, when the entrance funnels to the traps were sealed, no lobsters smaller than 64 mm CL were retained and no lobsters 58 Fishery Bulletin 103(1) 1.0 1 o 8 0.6 0.4 0.2- 0.0 30 90 100 T^ I.O-i 00 •rf.-*^ § 30 40 50 60 70 80 90 100 53.5 63.5 73.5 83.5 93.5 c o o §■ 1.0 1 0.8 C «*<- ao, */ 20 D 0.6 0.4 / 1"° /• 0.0 •7 -i.o ■ III 1 1 1 !■■ ■ i| I |"| | 0.2 y* -2 0 30 40 50 60 70 80 90 100 53.5 63.5 73.5 83.5 93.5 CL (mm) Figure 6 Fitted selectivity curves from the selected models (identified in Table 2) and their deviance residuals for escapement of -7. lalandii from commercial rock lobster traps covered with 100-mm stretched square meshes under field conditions. A and B are for traps with open entrance funnels; C and D are for traps with sealed entrance funnels. Discussion This study focuses on escapement of Cape rock lobster (J. lalandii) through mesh openings, and on escape- ment through the trap entrance of commercial traps. Three questions were initially posed, namely: through what mesh size, in theory, can a lobster of given CL escape; are lobsters physically able to escape through this theoretical mesh size, or are there other factors such as orientation and mobility of lobster appendages that prevent escapement; and what proportion of sublegal and legal size lobsters escape through the mesh and trap entrance of commercial traps? In brief, the results showed a weak leak between theoretical values and the ability of the lobsters to escape. Carapace base (CB) was isolated as the dimension most likely to limit escapement through stretched square meshes. This dimension superceded carapace width and depth, which have been more widely assumed to be the limiting factors to escapement of lobsters (Treble et al., 1998), mainly because our measurement included the width of the last pair of walking legs, folded flush against the carapace. Experimenting with lobster carapaces and an adjustable square hole showed that the joints of these appendages protrude ventrolater- al^ from the carapace, and the orientation and limited mobility of these appendages would prevent the lobster from escaping. Nevertheless, our theoretical escapement model included all three measurements in the underly- ing computer simulations to determine the appropriate mesh aperture required to target all lobsters larger than a given size. The theoretical escapement model produced surpris- ingly small values of "critical CL" for all three mesh sizes in comparison with the corresponding selectivity curves from the aquarium experiment. This result im- plies that many rock lobsters that should theoretically not have been able to escape, did so in the aquarium trials. We therefore concluded that the theoretical model was weak and that the mechanics of escapement ap- pear to be more complex than can be shown by simple measurements of the carapace dimensions and may rely also on the orientation of lobsters during escapement (Stasko, 1975). Selectivity curves developed from aquarium data in- dicated that an 85-mm-CL lobster should not have been able to escape a 100-mm mesh trap. However, field data indicated that escapement from 100-mm mesh traps with sealed trap openings exceeded 10%. Thus, rock lobsters that should not have been able to escape, ac- cording to aquarium experiments, did escape under field conditions. This result was expected, because the mesh of traps used in the commercial fishery (and field experiments) is often unevenly stretched across the met- al trap-frame, and therefore some openings lose their square dimensions. This unevenness in the stretch of the mesh was clearly illustrated by a random sample of 40 knot-to-knot aperture measurements from four 100- mm mesh commercial traps, which had diagonal dimen- sions significantly larger than the 70.71 mm predicted 60 Fishery Bulletin 103(1) by Pythagoras's theorem (r=4.470; df =39; P«0.001). In addition, repairs to torn meshes often leave openings that are somewhat larger than 100 mm and that are not square (Groeneveld, personal, observ. ). The wider SR of the selectivity curve for field data compared to the tight SR of the aquarium curves supports this "unevenly stretched mesh" hypothesis. Paradoxically, a 70-mm-CL lobster, which has a 1% chance of being retained by a 100 mm mesh in the laboratory has an 11% probability of being retained by a trap with the same mesh in the field (even when its entrances are sealed). Thus, some rock lobsters that should, and could have escaped through the 100 mm mesh of the field traps did not. Schoeman et al. (2002a) suggested that small rock lobsters that can escape do not always do so because they use the trap as a refuge against predators. Alternatively, overnight soak times (as used in the field trials) may be too short for all the small rock lobsters to escape. The probability of Table 2 Statistics from SELECT analysis for the field escapement trials. Values in parentheses are asymptotic standard errors sensu Millar (1993). These standard errors are provided only for the best model fits for each of the vari- ous categories of data. Note that all hypothesis tests were conducted by using deviance residuals for the full model and their degrees of freedom (see text for explanation). 100-mm mesh 100-mm mesh Trap-entrance open Trap-entrance sealed Logistic Richards Logistic Richards a -17.856 (2.460) -14.299 -20.801 (2.437) -51.967 b (/mm) 0.217 (0.031) 0.181 0.2686 (0.031) 0.626 6 0.641 4.238 L50 (mm) 82.274 (0.379) 82.222 77.444 (0.379) 78.458 Si? (mm) 10.124 (0.498) 10.809 8.181 (0.498) 7.997 Selection factor 0.82 0.77 H0: data have binomial distribution (i.e., data are not overdispersed) Deviance 21.593 21.257 18.698 15.807 df 19 18 17 16 P-value 0.305 0.267 0.346 0.467 ff0:S=l Deviance 0.336 2.891 df 1 1 P-value 0.562 0.090 escape is much reduced during hauling because captive specimens are then pressed into a tight mass within the fine-mesh (62-mm) codend of the trap. No escapement from sealed 62-mm mesh traps was expected during field trials, based on the aquarium L50 of 46.4 (±0.3) mm and the size range of lobsters used in the field (60 mm-95 mm CL). Nevertheless, small losses (0-18%, depending on lobster size; see Fig. 5) did occur. Only two explanations are possi- ble, namely: 1) that lobsters still managed to escaped through the mesh of the sealed 62-mm traps, despite precautions taken to ensure that the meshes of these traps were undamaged and that trap openings were properly sealed; and 2) that some individuals sus- tained injuries during exposure and handling, and subsequently were cannibalized by the healthy rock lobsters in the traps. This second conclusion is sup- ported by the presence of shell fragments observed in traps after their retrieval. Because these regres- sions had the same positive slopes, it seems likely that smaller rock lobsters would be more susceptible to injury and cannibalism than larger animals, and their susceptibility holds irrespective of whether the trap entrance is sealed or not. Escapement from 62-mm mesh traps with open en- trance funnels increased by 40-60% compared to es- capement from traps with sealed traps (Fig. 5). This finding has significant implications for the FIMS, which relies on catches made by 62-mm mesh traps and is con- ducted annually as a measure of the relative abundance of the J. lalandii resource. During a survey, it is as- sumed that all the Cape rock lobsters that are captured are retained and that trap-selection is uniform across all the size classes of these lobsters (Johnston, 1998). It appears that neither of these two assumptions can be met: significant escapement does occur through the trap entrance and there is a greater retention of larger specimens than smaller specimens . When the trap entrance was left open in the 100- mm mesh field trials, L50 increased to 82.3 mm (from 77.4 mm in sealed traps), thus indicating that captive Cape rock lobsters can and do use the trap entrance of commercial traps to escape. The open traps also have a wider SR of 10.1 mm (compared to 8.2 mm in sealed traps), and therefore animals with a CL of >87 mm (L75=87.3 mm), which are very unlikely to be able to get through the mesh apertures, will still be able to use the trap entrance to exit. The presence of this escape vent implies that there is little danger of ghost-fishing when using this trap type and that Cape rock lobsters of all sizes should be able to vacate the trap once the bait has been consumed. From a commercial viewpoint, however, the problem of leaving traps in the water for too long is that legal-size specimens, which cannot fit through the mesh, will escape through the entrance, thus decreasing catch rates considerably. The aquarium result (L50=75.1 mm) is considered the most accurate direct measurement of the selectiv- ity of 100-mm square mesh for J. lalandii, because care was taken to ensure that the mesh was stretched Groeneveld et al.: Escapement of Jasus lalandu from traps 61 evenly with square openings across the metal trap- frame and because we made sure that all lobsters that could escape, did, resulting in a tight SR of 2.6 mm. This L50 is remarkably close to the present MLS of 75 mm CL for the commercial fishery, especially considering that 100-mm mesh was first used when the MLS was 89 mm CL, and that the commercial mesh size remained at 100 mm despite the 14 mm CL reduction in MLS during the early 1990s (Schoeman et al., 2002b). The L50 obtained from the field trials with sealed trap openings (77.4 mm) was also close to the present MLS. In a recent indirect study (i.e., where the size com- position of a population was unknown) Schoeman et al. (2002a) found L-0 to be 79.2 mm (SJR=11.1 mm) under commercial operational conditions. The increase in L50 (above the 75.1 mm and 77.4 mm found in the direct aquarium and field studies, respectively) is the result of the trap entrances of commercial traps remaining open, so that rock lobsters that are too large to fit through the mesh can still escape through the entrance. In the present direct study, this factor increased the L50 from 77.5 mm (sealed entrance) to 82.2 mm (open entrance) for 100-mm mesh. Thus, one conclusion of the indirect study, namely that the South African fishery for J. lalandii is unusual in that standard commercial traps are covered with mesh having an aperture considerably wider (L50=79.2 mm CL) than that required to retain Cape rock lobsters of the current MLS (Schoeman et al., 2002a), must now be seen in a different light. The selectivity of the 100-mm stretched mesh itself now appears not to be wider than that which is currently required (based on the direct results). Rather, the indi- rectly determined L50 appears to have been inflated by the numbers of larger lobsters that were able to escape through the trap entrance. Direct studies of the escapement of crustaceans from pots (Krouse and Thomas, 1975; Krouse, 1978; Everson et al., 1992) have often been criticized because these studies themselves may affect the behavior of the ani- mals and do not include the dynamics of the processes of entry and escapement (Xu and Millar, 1993; Treble et al., 1998). We recognize these weaknesses, but felt that direct studies remain useful because they can be used to set a theoretical benchmark against which the results of indirect studies can be tested, especially if the trap selectivity of the latter depends on area and season. Various insights were gained from the pres- ent study, particularly because it closely followed an indirect study of trap selectivity for J. lalandii (Schoe- man et al., 2002a). In conclusion, this study of escape- ment of J. lalandii through square meshes showed 1) that 100-mm mesh size is, theoretically, near optimal for the fishery; 2) that many Cape rock lobsters that are able to escape through the mesh do not do so; 3) that the rock lobsters that are shown theoretically to be unable to escape through the mesh of commercial traps, often can do so; and 4) that specimens too large to escape through the mesh can escape through the trap entrance. Acknowledgments This study would not have been possible without the funding and infrastructure provided by Marine and Coastal Management (Department of Environmental Affairs and Tourism, South Africa). In particular, we would like to thank our colleagues, Steven McCue, Neil van den Heever, and Danie van Zyl for technical sup- port. We are also grateful to the skipper and crew of the research vessel Sardinops, which was used to conduct the field trials. J.P.K. received financial assistance from the Fridtjof Nansen and NORAD, and would like to thank his supervisors, Anders Ferno and Geir Blom, at the University of Bergen in Norway, for their assistance with an earlier draft of this manuscript. D.S.S. thanks the University of Port Elizabeth for support in terms of finance and infrastructure. Finally, the constructive comments of three anonymous referees are acknowl- edged; these aided substantially in clarifying certain parts of the original manuscript. Literature cited Arana, P. E„ and S. V. Ziller. 1994. Modelling the selectivity of traps in the capture of spiny lobster {Jasus frontalis), in the Juan Fernan- dez archipelago (Chile). Investigation pesq., Santiago 38:1-21. Brown, R. S., and N. Caputi. 1983. Factors affecting the recapture of undersize west- ern rock lobster Panulirus cygnus George returned by fishermen to the sea. Fish. Res. 2:103-128. 1985. Factors affecting the growth of undersize west- ern rock lobster, Panulirus cygnus George, returned by fishermen to the sea. Fish. Bull. 83:567-574. Cockcroft, A. C, and A. I. L. Payne. 1999. A cautious fisheries management policy in South Africa: the fisheries for rock lobster. Mar. Policy 23(6):587-600. Crous, H. B. 1976. A comparison of the efficiency of escape gaps and deck grid sorters for the selection of legal-sized rock lobsters Jasus lalandii. Fish. Bull. S. Afr. 8:5-12. Davis, G. E. 1981. Effects of injuries on spiny lobster, Panulirus argus, and implications for fishery management. Fish. Bull. 78:979-984. Everson, A. R., R. A. Skillman, and J. J. Polovina. 1992. Evaluation of rectangular and circular escape vents in the northwestern Hawaiian Islands lobster fishery. N. Am. J. Fish. Manag. 12:161-171. Fogarty, M. J., and V. D. Borden. 1980. Effects of trap-venting on gear selectivity in the inshore Rhode Island American lobster, Homarus ameri- canus, fishery. Fish. Bull. 77:925-933. Heydorn, A. E. F. 1969. The rock lobster of the South African west coast Jasus lalandii (H. Milne-Edwards): 2. Population stud- ies, behaviour, reproduction, moulting, growth and migration. Invest. Rep. Div. Sea Fish. S. Afr. 71, 52 p. Hunt, J. H., W. G. Lyons, and F. S. Kennedy. 1986. Effects of exposure and confinement on spiny 62 Fishery Bulletin 103(1) lobsters, Panulirus argus, used as attractants in the Florida trap fishery. Fish. Bull. 84:69-76. Johnston, S. J. 1998. The development of an operational management procedure for the South African west coast rock lobster fishery. Ph.D. diss., 370 p. Univ. Cape Town, Cape Town, South Africa. Krouse, J. S. 1978. Effectiveness of escape vent shape in traps for catching legal-sized lobster, Homarus americanus, and harvestable-sized crabs, Cancer borealis and Cancer irroratus. Fish. Bull. 76:425-432. 1989. Performance and selectivity of trap fisheries for crustaceans. In Marine invertebrate fisheries: their assessment and management (J. F. Caddy, ed.), p. 307-325. Wiley, New York, NY. Krouse, J. S., and J. C. Thomas. 1975. Effects of trap-selectivity and some lobster popu- lation parameters on size composition of the American lobster, Homarus americanus catch along the Maine coast. Fish. Bull. 73:862-871. Maynard, D. R., N. Branch, Y. Chiasson, and G. Y. Conan. 1987. Comparison of three lobster (Homarus americanus) trap escape mechanisms and application of a theoretical retention curve for these devices in the southern Gulf of St. Lawrence lobster fishery. Canadian Atlantic Fisheries Scientific Advisory Committee, Research Document, 87/87, 34 p. Millar, R. B. 1992. Estimating the size-selectivity of fishing gear by conditioning on the total catch. J. Am. Stat. Assoc. 87: 962-968. 1993. Analysis of trawl selectivity studies (addendum): implementation in SAS. Fish. Res. 17:373-377. Millar, R. B., and R. J. Fryer. 1999. Estimating the size-selection curves of towed gears, traps, nets and hooks. Revs. Fish Biol. Fish. 9:89-116. Millar, R. B., and S. J. Walsh. 1992. Analysis of trawl selectivity studies with an appli- cation to trouser trawls. Fish. Res. 13:205-220. Miller, R. J. 1990. Effectiveness of crab and lobster traps. Can. J. Fish. Aquat. Sci. 47:1228-1251. Newman, G. G., and D. E. Pollock. 1969. The efficiency of rock lobster fishing gear. S. Afr. Shipp. News Fish. Ind. Rev. 24(6):79-81. Pollock, D. E. 1986. Review of the fishery for and biology of the Cape rock lobster Jasus lalandii with notes on larval recruitment. Can. J. Fish. Aquat. Sci. 43(111:2107- 2117. Pollock, D. E., and C. J. de B. Beyers. 1979. Trap selectivity and seasonal catchability of rock lobster Jasus lalandii at Robben Island sanctuary, near Cape Town. Fish. Bull. S. Afr. 12:75-77. Pollock, D. E., A. C. Cockcroft, J. C. Groeneveld, and D. S. Schoeman. 2000. The commercial fisheries for Jasus and Palinurus species in the south-east Atlantic and south-west Indian oceans. In Spiny lobsters: fisheries and culture (B. F. Phillips and J. Kittaka, eds.), p. 105-120. Blackwell Science, UK. Rosa-Pacheco, R. D. L., and M. Ramirez-Rodriguez. 1996. Escape vents in traps for the fishery of the Cal- ifornia spiny lobster, Panulirus interruptus, in Baja California Sur, Mexico. Cienc. Mar., Baja Calif, Mex. 22:235-243. Schoeman, D. S„ A. C. Cockcroft, D. L. Van Zyl, and P. C. Goosen. 2002a. Trap selectivity and the effects of altering gear design in the South African rock lobster Jasus lalandii commercial fishery. S. Afr. J. Mar. Sci. 24:37-48. 2002b. Changes to regulations and the gear used in the South African commercial fishery for Jasus lalandii. S. Afr. J. Mar. Sci. 24:365-370. Stasko, A. B. 1975. Modified lobster traps for catching crabs and keeping lobsters out. J. Fish. Res. Board Can. 32(12): 2515-2520. Treble, R. J., R. B. Millar, and T I. Walker. 1998. Size-selectivity of lobster pots with escape-gaps: application of the SELECT method to the southern rock lobster (Jasus edwardsii) fishery in Victoria, Australia. Fish. Res. 34:289-305. Vermeer, G. K. 1987. Effects of air exposure on dessication rate, hemo- lymph chemistry, and escape behaviour of the spiny lobster, Panulirus argus. Fish. Bull. 85:45-51. Xu, X., and R. B. Millar. 1993. Estimation of trap selectivity for male snow crab (Chionoecetes opilio) using the SELECT model- ing approach with unequal sampling effort. Can. J. Fish. Aquat. Sci. 50:2485-2490. Zar, J. H. 1999. Biostatistical analysis, 663 p. Prentice-Hall, Inc., Englewood Cliffs, NJ. 63 Abstract — The recent development of the pop-up satellite archival tag (PSAT) has allowed the collection of information on a tagged animal, such as geolocation, pressure (depth), and ambient water temperature. The suc- cess of early studies, where PSATs were used on pelagic fishes, has spurred increasing interest in the use of these tags on a large variety of species and age groups. However, some species and age groups may not be suitable candidates for carrying a PSAT because of the relatively large size of the tag and the consequent energy cost to the study animal. We examined potential energetic costs to carrying a tag for the cownose ray iRhinoptera bonasus). Two forces act on an animal tagged with a PSAT: lift from the PSATs buoyancy and drag as the tag is moved through the water column. In a freshwater flume, a spring scale measured the total force exerted by a PSAT at flume velocities from 0.00 to 0.60 m/s. By measuring the angle of deflection of the PSAT at each velocity, we separated total force into its constituent forces — lift and drag. The power required to carry a PSAT horizontally through the water was then calculated from the drag force and velocity. Using published metabolic rates, we calculated the power for a ray of a given size to swim at a specified velocity (i.e., its swimming power). For each velocity, the power required to carry a PSAT was compared to the swimming power expressed as a percentage, Power (W) Drag as OTAX Total force (N) Total power (W) Total force as %TAX Wildlife Computers 0.00 0.000 0.000 0.00 0.064 0.000 0.00 0.15 0.017 0.003 0.34 0.074 0.011 1.44 0.30 0.076 0.023 3.01 0.103 0.031 4.05 0.45 0.113 0.051 2.55 0.147 0.066 3.33 0.60 0.159 0.095 4.80 0.186 0.112 5.63 Microwave Telemetry 0.00 0.000 0.000 0.00 0.115 0.000 0.00 0.15 0.030 0.004 0.59 0.120 0.018 2.36 0.30 0.063 0.019 2.46 0.132 0.040 5.20 0.45 0.116 0.052 2.62 0.157 0.071 3.55 0.60 0.159 0.095 4.77 0.211 0.126 6.37 mal that has adequate food resources in nature; higher loads are felt to be energetically significant. In this ex- ample using a 15.5-kg cownose ray, the Drag as %TAX is within acceptable parameters; however, at 0.60 m/s the Total force as %TAX begins to exceed these guide- lines. At this point, a researcher would have to consider whether diving behavior at this speed would be a sig- nificant factor in the animal's survival. Another application of this information would be to determine the minimum reasonable size for a study ani- mal of a particular species. Blaylock (1990) attempted to address this issue for cownose rays by considering the transmitter-to-ray mass ratio using dry weights. The advantage of using metabolic rates is that it identifies subtler but significant increases in energy requirement to carry a PSAT. In his study, Blaylock examined two age groups, a 0+ age group that had an average weight of 1.8 kg and a 1+ age group that ranged in size be- tween 4.3 kg and 7.8 kg. He concluded that the 0+ age group was negatively impacted by the sonic tag but that the 1+ age group was not effected. A PSAT is physically smaller than the sonic tags used in his experiment; in addition, it is attached to the animal at the nose-end of the tag so that it is carried with the long axis of the tag parallel to the long axis of the animal (Blaylock's sonic tags were attached so that the long axis of the tag was carried perpendicular to the long axis of the animal). Both these factors — smaller physical size and nose-end orientation in space — decrease the projected surface area of the tag. As an example, consider the metabolic cost of carrying a Wildlife Computers PAT to each of these sizes (1.8, 4.3, and 7.8 kg) of cownose ray (Table 3). For the 1.8-kg ray, only the exertion of carrying the PSAT at 0.15 m/s horizontally was associ- ated with a %TAX of <5%; higher swimming speeds or downward diving markedly increased the %TAX. It is obvious why short-term effects of carrying a sonic tag were evident. For the 4.3-kg ray, all swimming speeds greater than 0.15 m/s, whether horizontal or diving, required increased energy expenditures of >5%. For the 7.8-kg ray, %TAX was acceptable at 0.15 m/s, marginal to slightly elevated for mid-range speeds, and was clear- ly excessive at high speed. According to this analysis, rays of these size classes would not be good candidates for carrying a PSAT. As determined in this study, the smallest cownose ray that ought to be considered for PSAT tracking would be 14.8 kg. Drag as %TAX is s5% for all speeds and only slightly >57c for Total force as %TAX at 0.60 m/s. Because prolonged high speed div- ing behavior is not likely a factor in this ray's ability to survive, the minor elevation of %TAX for diving at 0.60 m/s can be disregarded. When applying this type of analysis to other species that predominantly swim at speeds greater than 0.60 m/s, several caveats make unwise the extrapolation of these data to higher velocities. Referring back to the equations describing drag and power. Equation 1 and Equation 2, respectively, drag is proportional to veloc- ity squared and power is proportional to velocity cubed provided that all other factors are constant. However, in examining Figure 3, as velocity increases from 0.00 m/s to 0.60 m/s, all other factors are not constant. Spe- cifically, the angle of deflection, 9, decreases from 90° at 0.00 m/s to as low as 31.5° at 0.60 m/s. First, the projected surface area, S, over which water flows de- creases as velocity increases. Second, the orientation (effective shape) of the object also effectively changes as velocity increases. Hence the drag co-efficient, CD Grusha and Patterson: Quantification of the drag and lift of pop-up satellite archival tags 69 Table 3 Metabolic costs to various sizes of cownose ray to increase in energy expenditure, normalized by thi ray to carry the PSAT while swimming in the hor applies when the ray is diving. carry a Wildlife Computers routine or active metabolic zontal plane. Total force as PSAT expressed as %TAX. Drag as %TAX is the rate (speed dependent — see text), required by the %TAX accounts for the buoyancy of the PSAT and Weight of ray (kg) Drag as %TAX Total force as %TAX Swimming velocity (m/s) Swimming velocity (m/s) 0.15 0.30 0.45 0.60 0.15 0.30 0.45 0.60 1.8 2.39 21.32 18.53 34.84 10.25 28.69 24.19 40.86 4.3 1.05 9.32 8.06 15.16 4.48 12.58 10.53 17.78 7.8 14.8 0.62 0.35 5.50 3.15 4.70 2.66 8.84 5.00 2.65 1.51 7.41 4.24 6.41 3.48 10.37 5.87 also changes. At some velocity greater than 0.60 m/s, 9 will approach 0°, and at that point S and CD would remain constant for higher velocities. After that veloc- ity is reached, then for higher velocities, drag would increase proportionately to the square of velocity and power would increase proportionately to the cube of ve- locity. In other words, between 0.00 m/s and 0.60 m/s, the changes in S and CD mask the parabolic relation- ship of drag with velocity. Because the velocity at which S and CD become constant is not known, extrapolations far beyond the maximum velocity for which drag was measured would be risky. The effect of the changing values of S and CD is evi- dent in this data set. For example in Table 1, as velocity doubles from 0.30 m/s to 0.60 m/s, drag increases by only 2.09 and 2.52 for the Wildlife Computers PAT and the Microwave Telemetry PTT-100, respectively, rather than by a factor of four. Similarly, power increases by 4.13 and 5.00 for the two PSATs and not by a factor of eight. For both these tags, d decreases with increasing velocity resulting in a smaller value for S and a differ- ent value for CD. By examining the forces exerted by a PSAT at various velocities, insights regarding the impact of these forces on a study animal can be gained. The combined forces of lift and drag act chronically on the anchor site of the PSAT. Although this study does not specifically address attachment methods, the forces of lift and drag exerted by a PSAT are not negligible and cannot be ignored when evaluating an attachment technique. A PSAT also imposes an energetic cost to the study animal. If that energy cost compromises the animal's behavior or survival, the information gained from the tag is not rep- resentative of an untagged animal. By estimating the energetic cost to an intended study animal, a researcher can make a more informed decision regarding the suit- ability of the animal for this type of tagging. Although direct extrapolation to higher swimming speeds is not possible with our data, the principles outlined in this study can be applied to faster swimming species such as tunas and billfishes that are frequently tagged. Acknowledgments We would like to thank T. Nelson, S. Wilson, W. Reisner and R. Gammisch for their assistance in running and setting up the flume, T. Mathes for his enthusiastic sup- port, and R. Brill, D. Kersetter, and J. Hoenig for helpful discussions. Financial support was provided by NOAA Office of Sea Grant. Literature cited Arnold, G., and H. Dewar 2001. Electronic tags in marine fisheries research: a 30-year perspective. In Electronic tagging and track- ing in marine fisheries (J. R. Sibert and J. L. Nielsen, eds.), p. 7-64. Kluwer Academic Publishers, Dordrecht, The Netherlands. Blaylock, R. A. 1990. Effects of external biotelemetry transmitters on behavior of the cownose ray Rhinoptera bonasus (Mitchill 1815). J. Exp. Mar. Biol. Ecol. 141:213-220. 1992. Distribution, abundance, and behavior of the cow- nose ray, Rhinoptera bonasus (Mitchill 1815), in lower Chesapeake Bay. Ph.D. diss., 129 p. College of Wil- liam and Mary, Williamsburg, VA. Block, B. A., H. Dewar, C. Farwell, and E. D. Prince. 1998. A new satellite technology for tracking the move- ment of Atlantic bluefin tuna. Proc. Natl. Acad. Sci. 95:9384-9389. Dagorn, L., E. Josse, and P. Bach. 2001. Association of yellowfin tuna (Thunnus albacares) with tracking vessels during ultrasonic telemetry experiments. Fish. Bull. 99:40-48. DuPreez, H. H., A. McLachlan, and J. F. K. Marais. 1988. Oxygen consumption of two nearshore marine elasmobranchs, Rhinobatos annulatus (Muller & Henle, 1841) and Myliobatus aquila (Linnaeus, 1758). Comp. Biochem. Physiol. 89A:283-294. Graves, J. E., B. E. Luckhurst, and E. D. Prince. 2002. An evaluation of pop-up satellite tags for estimat- ing post-release survival of blue marlin. Fish. Bull. 100:134-142. 70 Fishery Bulletin 103(1) Gunn, J., and B. Block. 2001. Advances in acoustic, archival, and satellite tagging of tunas. In Tuna: physiology, ecology, and evolution (B. A. Block and E. D. Stevens, eds.l, p. 167-224. Aca- demic Press, San Diego, CA. Hill, R. D., and M. J. Braun. 2001. Geolocation by light level — the next step: latitude. In Electronic tagging and tracking in marine fisheries (J. R. Sibert and J. L. Nielsen, eds.), p. 315-330. Kluwer Academic Publishers, Dordrecht, The Netherlands. Kerstetter, D. W. 2002. Use of pop-up satellite tag technology to estimate survival of blue marlin (Makaira nigricans) released from pelagic longline gear. M.S. thesis, 109 p. College of William and Mary, Williamsburg, VA. Lutcavage, M. E.. R. W. Brill, G. B. Skomal. B. Chase, and P. Howey. 1999. Results of pop-up satellite tagging of spawning size class fish in the Gulf of Maine: Do North Atlantic bluefin tuna spawn in the mid-Atlantic? Can. J. Fish. Aquat. Sci. 56:173-177. Sibert, J. R., M. K. Musyl, and R. W. Brill. 2003. Horizontal movements of bigeye tuna (Thunnus obesus) near Hawaii determined by Kalman filter analysis of archival tagging data. Fish. Oceanogr. 12: 141-151. Smith, J. W. 1980. The life history of the cownose ray, Rhinoptera bonasus (Mitchill 1815), in lower Chesapeake Bay, with notes on the management of the species. M.A. thesis, 151 p. College of William and Mary, Williamsburg, VA. 71 Abstract — Fish bioenergetics models estimate relationships between energy budgets and environmental and physi- ological variables. This study presents a generic rockfish (Sebastes) bioen- ergetics model and estimates energy consumption by northern California blue rockfish (S. mystinus) under average (baseline I and El Nino con- ditions. Compared to males, female S. mystinus required more energy because they were larger and had greater reproductive costs. When El Nino conditions I warmer tempera- tures; lower growth, condition, and fecundity) were experienced every 3-7 years, energy consumption decreased on an individual and a per-recruit basis in relation to baseline conditions, but the decrease was minor (<4% at the individual scale, <7% at the per- recruit scale) compared to decreases in female egg production (12-19% at the individual scale. 15-23% at the per-recruit scale). When mortality in per-recruit models was increased by adding fishing, energy consumption in El Nino models grew more similar to that seen in the baseline model. However, egg production decreased significantly — an effect exacerbated by the frequency of El Nino events. Sensitivity analyses showed that energy consumption estimates were most sensitive to respiration param- eters, energy density, and female fecundity, and that estimated con- sumption increased as parameter uncertainty increased. This model provides a means of understand- ing rockfish trophic ecology in the context of community structure and environmental change by synthe- sizing metabolic, demographic, and environmental information. Future research should focus on acquiring such information so that models like the bioenergetics model can be used to estimate the effect of climate change, community shifts, and different har- vesting strategies on rockfish energy demands. Effects of El Nino events on energy demand and egg production of rockfish (Scorpaenidae: Sebastes): a bioenergetics approach Chris J. Harvey Northwest Fisheries Science Center National Marine Fisheries Service 2725 Montlake Blvd. E Seattle, Washington 98112 E-mail address Chris. Harveyignoaa gov Manuscript submitted 20 October 2003 to the Scientific Editor's Office. Manuscript approved for publication 2 August 2004 by the Scientific Editor. Fish. Bull. 103:71-83 (2005). Over 90 species of rockfish (Sebastes spp.) are found in kelp beds, rocky reefs, pelagic habitats, and continental shelf and slope zones of the temperate and subarctic North Pacific; these spe- cies feed on a range of organisms, from zooplankton to fish (Love et al., 20021. Although they are a key component of groundfish fisheries on the U.S. Pacific Coast, many rockfish have declined considerably in recent decades, owing to overfishing and climate-induced downturns in production (Parker et al., 2000). Conservation efforts, rang- ing from coast-wide fishery closures to establishment of marine reserves, have been enacted in order to rehabilitate rockfish stocks. The efficacy of such actions depends in part on the dynam- ics of the communities in which rock- fish exist. Key among these dynamics are trophic interactions, as influenced by abiotic factors and rockfish popula- tion structure. Although rockfish are widely dis- tributed and important to the ecolo- gy, fisheries, and conservation efforts of the Pacific Coast, little is known about their trophic dynamics. For ex- ample, of the 65 rockfish species that live along the North American West Coast, quantitative diet data are available for only 15 species (Murie. 1995). Better information on the food habits and energetics of both juvenile and adult rockfish would facilitate a greater understanding of the role they play in their communities, and how their role is affected by external forces. This is particularly true given observations that environmental vari- ation can have strong effects on rock- fish growth and condition (Lenarz et al., 1995; Woodbury, 1999). Fish bioenergetics models relate the energy consumption, growth, and energy allocation patterns of fishes to environmental and physiological variables such as temperature, food quality, body size, and reproductive status (Kitchell et al., 1977). These models, founded in thermodynamic laws of mass and energy balance, can successfully predict patterns of energy demands by fish (Madenjian et al., 2000). At the scale of the indi- vidual fish, bioenergetics models can estimate effects of a fish on its com- munity (in terms of the amount of prey it consumes) and effects of the environment on the fish, such as how changes in temperature or food avail- ability influence energy consumption and growth (Rice et al., 1983). When coupled to population models, bio- energetics models can predict prey- predator supply-demand relationships (Negus, 1995) and determine how different fishery management poli- cies will affect prey resources in the community from which the targeted fish is extracted (Kitchell et al., 1997; Essington et al., 2002; Schindler et al., 2002). Thus, these models may facilitate a more community- or eco- system-level approach to rockfish management. In this study, I develop a generic Sebastes bioenergetics model. My first objective is to detail the parameters and the sensitivity analysis of the model, thereby offering a synthesis of what is known about Sebastes en- ergetic physiology and identifying pa- rameters for which greater informa- tion is desirable. The second goal is to present a simple application of the model: an estimation of the effects of 72 Fishery Bulletin 103(1) El Nino related environmental changes on the energy demands of blue rockfish (S. mystinus) under unfished and fished conditions. Two relevant characteristics of El Nino events in U.S. West Coast waters are elevated temperatures and reductions in growth rates and re- productive condition of Sebastes (Lenarz et al., 1995; VenTresca et al., 1995; Woodbury, 1999). The bioener- getics approach can incorporate these changes and can therefore help to characterize the role of rockfish as consumers in a dynamic environment. Methods Model structure I followed the basic structure of bioenergetics models established for other fishes (e.g., Kitchell et al., 1977; Hewett and Johnson, 1992), in which energy intake (consumption) equals all energy outputs (respiration, wastes, growth, and reproduction). The basic model equation is C = (i? + A + S) + (F + U) + (AB + G) (1) where C = consumption, R = respiration, A = active metabolism, S = specific dynamic action (digestive costs), F = egestion, U = excretion, AB = somatic growth, and G = gonad production. The respiration and active metabo- lism portions of Equation 1 take the form R = RA x WRB x f(T) x ACT, (2) where RA and RB are constants that describe the allo- metric respiration function, W is wet biomass, f(T) is a temperature dependence function, and ACT is an activ- ity multiplier (Kitchell et al., 1977). The function f(T) (Kitchell et al., 1977) is a hump-shaped function that requires estimates of optimal (RTO) and maximum (RTM) temperatures for respiration, and a Q10 (RQ). The terms S, U, and F all scale to total consumption (Kitchell et al., 1977). One can thus think of them as a general energy loss term Loss = (S + U) x (C -F) + F. (3) Model parameters Although parameters are derived from studies of many rockfish species, I developed the present model to describe energetic dynamics of S. mystinus, for which a consider- able literature exists regarding diet and responses to climate variability (e.g., Hallacher and Roberts, 1985; Bodkin et al., 1987; Hobson and Chess, 1988; Lenarz et al., 1995; VenTresca et al., 1995). Respiration parameter estimates came from studies of other Sebastes species or related scorpaenid fishes (Table 1). For RTM, I used published estimates for S. thompsoni and S. schlegeli (Ouchi, 1977; Tsuchida and Setoguma, 1997), and assumed that RTO would be 5°C Table 1 Parameter model. values for the generic Sebastes bioenergetics Parameter Description Value RA Intercept of the allometric respiration function 0.0143 RB Slope for allometric respiration function -0.2485 RQ Slope for temperature dependence of respiration ( 2 ACT Multiplier for active metabolism 1 RTO Optimum temperature for respiration 23°C RTM Maximum temperature for respiration 28°C SDA Specific dynamic action coefficient 0.163 FA Egestion coefficient 0.104 UA Excretion coefficient 0.068 ED Energy density (somatic tissue) of wet mass 6,120 J/g GED Energy density (female gonadal tissue) of wet mass 8,627 J/g GA Coefficient of the female length-fecundity relationship 1.559 GB Exponent of the female length-fecundity relationship 3.179 GSI„lax Maximum male gonadosomatic index 0.008 cooler. The resulting RTO was similar to upper tem- peratures at which juvenile S. diploproa experienced zero growth while feeding (Boehlert, 1981). RQ was based on low-temperature Q10 values in several scorpaenid respi- ration studies (Boehlert et al. 1991; Yang et al., 1992; Kita et al, 1996; Vetter and Lynn, 1997). RA, the oxygen consumption rate for a 1-g fish at RTO, was derived from data for nongestating S. schlegeli (Boehlert et al., 1991). RB, which describes the allometric scaling of respiration, was also derived from data for nongestating S. schlegeli spanning a range of roughly 0.7 to 1.9 kg body mass (Boehlert et al., 1991). Respiration terms were converted to energy units by an oxycalorific correction (13.56 J/mg 09), and then to biomass by assuming that rockfish en- ergy density (ED) = 6,120 J/g wet mass (Perez, 1994). The ACT multiplier was assumed to equal 1. This assumption is best justified in cases where routine res- piration rates were used to determine parameters for the model. Boehlert et al. (1991) stated that S. schlegeli in their analysis were generally inactive, which implies that rates derived from their data represent resting Harvey: Effects of El Nino events on consumption and egg production of Sebastes spp. 73 metabolism. I chose to keep ACT at 1, however, because I could find no data describing a reasonable activity multiplier. Thus. Sebastes model outputs may underes- timate energy consumption under conditions in which individuals are especially active. I obtained growth (AB in Eq. 1) terms using von Bertalanffy length-at-age curves and data for length- to-mass conversions for S. mystinus as summarized by Love et al. (2002). Because female S. mystinus are larger at age than males, growth was modeled with sex-specific von Bertalanffy curves with the difference equation method of Gulland (1983). Digestion and waste terms S, F, and U were derived from previous teleost models (Hewett and Johnson, 1992). I estimated gonad production (G) with gonadosomatic indexes (GSI) and size-fecundity relationships (females only), assuming that female and male S. mystinus ma- ture gradually over the range of lengths observed by Wyllie-Echeverria (1987), and reproduce once annually. For males, I assumed that gonads have the same ED as somatic tissue; for females, I assumed that gonadal en- ergy density (GED) = 8,627 J/g, which was the average of gonadal energy density at the onset of embryogenesis for S. flavidus and S. jordani (MacFarlane and Norton, 1999). Estimated maximum female GSI was based on a fecundity-length relationship: fecundity = GA x TLGB, (4) where GA and GB were taken from a generic rockfish length-fecundity relationship (Love et al., 2002) and TL is total length in cm. Fecundity was converted to bio- mass units by assuming that each egg weighed 0.0003 g, which I derived from Love et al. (1990) by dividing the mean maximum female gonad weight by the estimated fecundity of modal mature females for several species. For mature males, I assumed a constant maximum GSI based on data for other species (Love et al., 1990). Post- spawning GSI was assumed to be 10% of the maximum for each sex, as with other rockfish (Love et al., 1990). The G terms were the difference between the maximum and minimum GSIs for each sex, expressed as mass (and, in females, adjusted by multiplying by GED/ED). Rockfish are viviparous, and developing larvae may receive energy from both yolk and maternal sources (Love et al., 2002). During gestation in a laboratory, female S. schlegeli consumed 35% to 117% more oxygen than nongestating fish of similar size (Boehlert et al., 1991). To account for the possibility that blue rockfish may also be matrotrophically viviparous, I increased female respiration by 50% during the gestation period (assumed to be 45 days per year based on gestation times of other species [Boehlert et al., 1991]). Model application: effects of El Nino on blue rockfish energy consumption To examine the effects of El Nino on S. mystinus energy consumption, I created two model conditions: a baseline model and an El Nino model that estimated S. mystinus Table 2 Changes in the S. mystinus bioenergetics model that were implemented in El N ino scenarios in relation to the base- line model. Variable Change Temperature Increased 1.5°C in El Nino years7 Growth (length increment) Decreased 17.5% in El Nino years' Female condition Decreased 10% in El Nino years; factor decreased 5% the year following an El Nino2 Male condition Decreased 7.5% in El Nino years; factor decreased 5% the year following an El Nino- Fecundity Decreased 67% in El Nino years2 1 Source: Lenarz et al.. 1995. - Source: VenTresca et al.. 1995. energy demands, in megajoules (MJ), required for neces- sary growth, reproduction, and related metabolic costs. I used MJ rather than prey biomass as the currency because quantitative, seasonal diet data for S. mystinus in northern California were available for average years (Hobson and Chess 1988) but not for El Nino years. During the 1982-83 El Nino, Lea et al. (1999) found that central Californian S. mystinus consumed large numbers of the pelagic crab Pleuroneodes planipes, which is typically found south of Point Conception during average years. During the same time period, S. fnystinus ate few tuni- cates or scyphozoans (Lea et al., 1999), which were the predominate prey of S. mystinus in average years (Hobson and Chess, 1988). These findings suggest a major shift in S. mystinus prey composition during El Nino events. The baseline model simulates energy consumption of northern California S. mystinus from age 0 to age 30, based on quarterly growth estimates from sex-specific von Bertalanffy curves (Love et al., 2002) and seasonal temperature data from Hobson and Chess (1988). Mature females released larvae in the fourth quarter of each year, and mature males released gametes in the third quarter (Wyllie-Echeverria, 1987). Energy consumption for both sexes from ages 0 to 30 was expressed at two scales: for the 30-year life span of an individual; and on a per-recruit basis (under the assumption that there was no fishing mortality and that the natural mortality rate [M] was 0.2, applied in quarterly time steps). The El Nino model was similar to the baseline model, except an El Nino occurred every three to seven years. During these years there were changes in temperature, growth, condition, and fecundity (Table 2). Temperature increases in El Nino years were similar to temperature anomalies in northern California waters during major El Nino events from 1957 to 1993 (Lenarz et al., 1995). Changes in growth (in terms of length increment), con- 74 Fishery Bulletin 103(1) dition (the ratio of actual to expected weight, based on length-weight relationships), and fecundity were based on empirical measures of S. mystinus during El Nino years (Lenarz et al., 1995; VenTresca et al., 1995). As in the baseline model, 1 expressed energy consumption by both sexes at individual and per-recruit scales. Finally, I ran simulations at the per-recruit scale in which the total mortality rate (Z) was increased by add- ing a fishing-induced mortality rate (F) in increments of 0.05 to M; fishing mortality was imposed on fish greater than 20 cm, the size at which S. mystinus enters fisher- ies in California waters (Laidig et al., 2003). The range of Z examined was 0.2 (natural mortality only) to 1.0 (a heavily overfished condition). These simulations were run under baseline conditions and El Nino conditions to determine if there was any interaction between El Nino effects and Z. Sensitivity analysis To measure sensitivity of the Sebastes bioenergetics model to different parameters, I used a Monte Carlo error analysis method (Bartell et al., 1986). In this method, parameters are drawn randomly from normal distributions with means equal to parameter estimates (Table 1) and with a coefficient of variation (CV) of either 2%, 10%, or 20%. Cases where randomly drawn RTO was greater than RTM were discarded. Female and male models were run 1000 times for each of the three CVs. Individual simulations ran to age 30 at 0.25-year increments; seasonal temperatures were those used in the baseline model. Parameter influence on 30-year cumulative consumption estimates was judged accord- ing to the parameters' relative partial sums of squares (RPSS), which quantify the influence of a parameter after all other parameters have been accounted for. RPSS for all parameters were calculated with SYSTAT (version 10.2, SYSTAT Software Inc., Richmond, CA). Additionally, means and standard deviations of con- sumption estimates from RPSS analyses were calculated to capture the range of energy consumption possible over the lifetime of female and male S. mystinus. Results Northern California S. mystinus baseline energy demands Baseline energetic demands of northern California S. mystinus were a function of size, sex, and the scale of calculation (i.e., individual versus per recruit). As size increased, more energy was allocated to respiration, elimination of wastes, and reproduction, and steadily less energy was allocated to growth (Fig. 1). At the individual scale, females consumed more than males at all ages. The sexes diverged markedly as fish matured (beginning at age 3 for females, age 4 for males), and continued to diverge as fish approached asymptotic sizes (Fig. 2A). The disparity was related to sex-based differences in growth rate, maximum size, GSI, and the increased respiration of gestating females. Cumu- lative consumption through age 30 was 285.0 MJ for individual females, and 174.6 MJ for individual males. Assuming a prey energy density of 1500 J/g (given S. mystinus diets [Hobson and Chess, 1988] and prey- density measurements of the same or related prey spe- cies [Paine and Vadas, 1969; Thayer et al., 1973; Foy and Norcross, 1999]), this energy density equates to a long-term average energy consumption rate of 2.7% body mass per day for females and 2.8% body mass per day for males. Females also had greater requirements than males at the per-recruit scale, although mortality gradually lessened the contribution of older age classes (Fig. 2B), nullifying some of the disparity between the sexes at the individual scale. Cumulative female and male per- recruit energy consumption was 20.7 MJ and 14.8 MJ, respectively. Per-recruit energy consumption, the prod- uct of age-specific consumption rate and relative fish abundance, peaked at ages 4-6, indicating that those age groups have the greatest potential to affect their prey species. Effects of El Nino on S. mystinus energetics El Nino events changed S. mystinus energy consumption compared to that in the baseline model, but the direction and magnitude of change were dependent on sex, age, scale of calculation (individual vs. per recruit), and the number and frequency of El Nino events experienced by a given cohort. To demonstrate this change, I modeled growth of two cohorts that experienced El Nino regimes of moderate or high intensity. The first cohort ("cohort A") experienced five El Nino events by age 30, whereas the second cohort ("cohort B") experienced eight El Nino events (Figs. 3 and 4). At the scale of individual fish, cohorts A and B experi- enced lower energy consumption in El Nino events, par- ticularly among females. During El Nino years, which first occurred at age 3 for cohort A and at age 1 for co- hort B, consumption by females was always lower than the baseline value (Fig. 3A). In immature females, the disparity was 7-10% lower than the baseline value and was 12-13% lower for mature females. These reductions in consumption were a function of lower growth rates, poor condition factor, and reduced fecundity during El Nino years. In contrast, consumption by males during El Nino years was 4-9% lower than the baseline value among immature individuals, but was roughly equal to the baseline value for mature individuals (Fig. 3B), in part because males did not experience drastic changes in reproductive condition during El Nino years. Both sexes experienced years when energy consumption was greater than the baseline value, particularly two years after an El Nino event when the somatic condition fac- tor returned to normal and greater-than-average growth for that age occurred. By age 30, sizes of fish in both El Nino models were close to the asymptotic maxima and were therefore similar to baseline sizes (Table 3). Cumulative 30-year energy consumption values were Harvey: bffects of El Nino events on consumption and egg production of Sebastes spp 75 12 - £ 0 10 15 20 25 30 Figure 1 Estimated allocation of energy consumption by northern California S. mystinus from ages 0 to 30 under baseline model conditions. Consumption (C) is allocated as respiration (R), waste, and digestive costs (F+U+SDA), growth (4B), and reproduction (G). (A) Females. (B) Males. also similar in all models and in both sexes, despite the declines experienced by females. Repeated exposure to El Nino also affected reproduc- tion by S. mystinus. Both sexes experienced delays in maturation as a result of slowed growth rates during El Nino events, and the delay was related to the num- ber of El Nino years experienced at young ages. In the baseline model, 50% maturity was reached at age 6 for both sexes. In cohort A, 50% maturity was reached at age 6 by females, but at age 7 by males. Under the more arduous conditions of cohort B, both sexes reached 50% maturity at age 7. The effect of delayed maturation in terms of energy consumption should be greatest in females because of their greater investments in repro- duction, although this was not especially noticeable at the scale of cumulative consumption per individual (Table 3). A further effect of El Nino events occurred in female egg production. The dramatic reduction in fecundity during El Nino years over the course of an individual female's life caused cumulative egg produc- tion in cohort A to be only 87.9%> of the baseline level, and cohort B female egg production was only 81.3% of the baseline level (Fig. 3C). More pronounced El Nino effects occurred at the per- recruit scale. El Nino conditions reduced per-recruit en- ergy consumption in both sexes in contrast to baseline conditions (Fig. 4, A and B). Incorporating mortality lowered the contribution of older age groups, where individual consumption was highest (Fig. 3, A and B), thereby magnifying the El Nino effects on young fish. The negative effects on young age classes were exac- erbated in females by slowed maturation and reduced 7b Fishery Bulletin 103(1) 0 5 10 15 20 Age (y) Figure 2 Estimated energy consumption by S. mystinus under baseline model con- ditions. (A) Females and males at the per-individual scale. (B) Females and males at the per-recruit scale, assuming a mortality rate (Z) of 0.2 (i.e., no fishing mortality). Table 3 Final weights and cumulative energy consumptions for female and male S. mystinus from bioenergetics models run under baseline and El Nino conditions. All values are taken from the end of the 30th year. Cohort-A and cohort-B individuals experienced five and eight El Nino events, respectively (see Figs. 3 and 4). Final weight (g) Total consumption (MJ) Model Females Males Females Males Baseline Cohort A Cohort B 1,134.3 1,129.4 1,126.8 617.2 616.5 616.1 285.0 278.1 273.6 174.6 173.3 172.1 fecundity (due to slower growth), resulting in lower per-recruit consumption to meet reproductive costs. Thirty-year cumulative per-recruit energy consumption was 20.0 MJ for cohort-A females (3.2% lower than the baseline value), and 19.4 MJ for cohort-B females (6.3% less than the baseline value). Cumulative per-re- cruit consumption by cohort-A males was 14.5 MJ (1.9% lower than baseline), whereas cohort-B males consumed 14.2 MJ (4.4%. less than the baseline level). The reduc- tion of cumulative egg production was also more drastic at the per-recruit scale: cohort-A females produced 15% fewer eggs than the baseline level, whereas cohort-B fe- males produced 23% fewer eggs at the per-recruit scale (Fig. 4C). These reductions in egg production were re- lated to smaller size, lower fecundity in El Nino years, delayed maturation, and accumulative mortality, all of which allowed fewer females to reach maturity. Harvey: Effects of El Nino events on consumption and egg production of Sebastes spp. 77 m 12 " O o O 180 120 60 " D 10 15 20 10 15 20 10 15 20 10 15 Age (y) 20 25 25 25 30 30 30 B B B B B B B B A A r A — i A 1 r A 1 30 Figure 3 Estimated energy consumption and egg production by S. mystinus at the per-individual scale, under baseline conditions and for two cohorts (A and B) in which El Nino events occurred every three to seven years. (A) Female energy consumption. (B) Male energy consumption. (C) Egg production. (D) Timing of El Nino events for cohorts A and B. Effects of El Nino on fished cohorts Adding fishing mortality to the total mortality rate applied in the per-recruit simulations caused changes in the total energy consumption and egg production of S. mystinus experiencing repeated El Nino events, in con- strast to the baseline state. Under both El Nino regimes, per-recruit consumption by both sexes increased slowly as Z increased until it was nearly identical to the base- line level for cohort A (Fig. 5A) or exceeded the baseline for cohort B (Fig. 5B). The reason for this is that the slower growth experienced during El Nino years meant that fish reached 200 mm (the size of recruitment into the fishery) later and therefore were not as rapidly sub- jected to fishing mortality as baseline fish. This extra period of feeding prior to reaching 200 mm was sufficient to equal or exceed the per-recruit energy consumption level in the baseline model. In contrast, increased Z caused strong declines in egg production, and that effect was exacerbated by the frequency of El Nino years, as demonstrated by the steeper decline in cohort B (Fig. 5B). Delayed matura- tion caused by El Nino meant that many females were removed by fishing before they were able to reproduce; 78 Fishery Bulletin 103(1) Baseline --■ — Cohort A —a— Cohort B 10 15 D B B B B B B B B A A A A A 10 15 Age (y) 20 25 30 Figure 4 Estimated energy consumption and egg production by S. mystinus at the per-recruit scale, under baseline conditions and for two cohorts (A and B) in which El Nino events occurred every three to seven years. (Al Female enrgy consumption. (B) Male energy consumption. (C) Egg production. (Dl Timing of El Nino events for cohorts A and B. furthermore, those that escaped fishing had lower fe- cundities because of their smaller size and reduced egg production because of the number of El Nino years experienced. Sensitivity analysis Based on the RPSS analysis, sensitivity of rockfish bioenergetics models to parameter variation was a func- tion of sex, size, and the CV of the parameter set. When CV = 2%, the model was most sensitive to respiration parameters in Equation 2 (particularly RB, RQ, and RTO) and to ED, although the rank order varied slightly by sex (Fig. 6, A and B). The sum of the RPSScv=2fJ for all parameters was >0.99 for both the male and female models. This result implies that energy consump- tion responded linearly to parameter variation because summed RPSS values scale from 0 to 1, with 1 implying a linear response to parameter perturbation (Bartell et al., 1986). When CV increased to 10%, the rank order of parameter sensitivity changed slightly, although res- piration parameters and ED remained most important Harvey: Effects of El Nino events on consumption and egg production of Sebastes spp. 79 1.0 - 0.9 - A 0.8 - 0.7 - Male consumption Female consumption 3 > 0.6 - — o — Egg production a) o 0.5 - 1 1 1 1 0.2 0.4 0.6 0.8 1.0 o c o 1.0 - ^__^^ Q O a. 0.9 - 0.8 - 0.7 - 0.6 - 0.5 - 1 1 1 ^" 0.2 0.4 0.6 0.! Total mortaility rate (Z) 1.0 Figure 5 Effects of mortality (Z, increased due to fishingl on S. mystinus responses to El Nino events, in relation to a baseline model with identical Z. (A) Cohort A, which experienced 5 El Nino years (see Figs. 3 and 4). (B) Cohort B, which experienced 8 El Nino years. (Fig. 6, C and D). RPSScv=10(-r values declined to 0.84 and 0.94 for females and males, respectively, indicating a greater degree of nonlinearity in response to parameter variation. Finally, when CV increased to 20%, there were major changes in parameter rank order and RPSS, especially for females (Fig. 6E). All female parameters essentially had equal weight, and RPSScv=2(r; dropped dramatically to 0.14, indicating a nonlinear response to parameter variation. Males experienced slight changes in parameter rank order at CV = 20% (Fig. 6F) and increasingly nonlinear behavior related to parameter variation (RPSScv=20r; = 0.81). Because the major differ- ence in the models for the two sexes is the reproductive terms (i.e., Eq. 4 for females vs. the simple GSI calcula- tion for males), the GA or GB terms (or both) appear to be the cause of poor female model performance at high parameter uncertainty. Also, because GA and GB should only affect female energy budgets as the females mature, model sensitivity to those parameters is likely size dependent. Energy consumption estimates generated in RPSS analyses were consistently greater than estimates gen- erated by the baseline deterministic model, which used the parameter values from Table 1. Mean consump- tion estimates and standard deviations increased as the parameter CV increased (Table 4). This effect was more pronounced in females than in males, especially when parameter CV=20%. At that level of parameter uncertainty, male and especially female consumption estimates had very large standard deviations. 80 Fishery Bulletin 103(1) in C/3 D- 0.30 - A 0.25 0.20 0.15 0.10 H 0.05 0.00 n n, 0.30 -| Q 0.25- 0.20 - 0.15 - 0.10 - 0.05 - 1 0 00 -±K- nnnnnnnn fi 0.30 0.25 0.20 H 0.15 0.10 - 0.05 - 0.00 E vims edu Manuscript submitted 20 January 2004 to the Scientific Editor's Office. Manuscript approved for publication 2 August 2004 by the Scientific Editor. Fish. Bull. 103:84-96 (2005). Atlantic white marlin (Tetrapturus albidus Poey, 1860) are targeted by a directed recreational fishery and occur as incidental bycatch in commercial fisheries throughout the warm pelagic waters of the Atlantic Ocean. Total reported recreational and commercial landings of white marlin peaked at 4911 metric tons (t) in the mid-1960s, declined steadily during the next 15 years, and have since fluctuated with- out trend between 1000 and 2000 t despite substantial increases in fish- ing effort (ICCAT, 2003). Recent popu- lation assessments conducted by the Standing Committee for Research and Statistics (SCRS) of the International Commission for the Conservation of Atlantic Tunas (ICCAT) indicate that the Atlantic-wide white marlin stock is currently at historically low levels and has been severely overexploited for over three decades (ICCAT, 2003). In the 2002 white marlin assessment, the 2001 biomass was estimated to be less than 12% of that required for maximum sustainable yield (MSY) under the continuity case (ICCAT, 2003). Current harvest is estimated to be more than eight times the replace- ment yield (ICCAT, 2003). In response to the overfished status of white marlin, ICCAT has adopted binding international recommendations to decrease overall Atlantic landings of this species by 67% from 1996 or 1999 levels (whichever is greater) through the release of all live white marlin from commercial pelagic longline and purse-seine gears (ICCAT, 2001). How- ever, even these dramatic reductions may be ineffective in rebuilding the white marlin stock. Goodyear (2000) estimated that a 60% decrease from 1999 fishing mortality levels would be required to halt the reduction of At- lantic blue marlin (Makaira nigricans). Because white marlin experience high- er levels of fishing-induced mortality, it is expected that the reduction in mortality required to stabilize this stock will be even greater. Management measures within the United States, established by the At- lantic Billfish Fishery Management Plan (FMP) (NMFS, 1988) and sub- sequent Amendment 1 (NMFS, 1999), have also been implemented to reduce white marlin fishing mortality. U.S. commercial fishermen have been pro- hibited from landing or possessing all Atlantic istiophorids since 1988. Dead discards of white marlin from the U.S. commercial pelagic longline fishery peaked at 107 t in 1989, and have decreased to 40-60 t over the last several years (White Marlin Status Review Team1). Management ♦Contribution 2610 from the Virginia Institute of Marine Science, College of William and Mary, Gloucester Point, VA 23062. 1 White Marlin Status Review Team. 2002. Atlantic white marlin status review document, 49 p. Report to the National Marine Fisheries Ser- vice. Southeast Regional Office, September 3, 2002. www.nmfs.gov/ prot_res/readingrm/Candidate_Plus/ wh it e„m a rl in/ whm_status_review.pdf Horodysky and Graves: Estimation of survival of Tetrapturus albidus caught and released in the North Atlantic recreational fishery 85 measures for U.S. recreational anglers include a mini- mum size of 66 inches lower jaw fork length (NMFS, 1999) and mandatory reporting of landed billfishes (NMFS, 2003). White marlin landings by U.S. recre- ational anglers ranged between 40 and 110 t from 1960 to the mid-1980s (Goodyear and Prince, 2003) and have decreased to about 2 t in recent years. At present, over 99% of the 4000-8000 white marlin estimated to be caught annually by U.S. recreational fishermen are released (Goodyear and Prince, 2003). The benefit of current management measures that rely on the release of white marlin cannot be evaluated because levels of postrelease survival are not known for this species. Recapture rates of billfishes tagged with conventional tags are very low (0.4-1.83%; Prince et al., 2003; Ortiz et al., 2003), which may result from high postrelease mortality, tag shedding, or a failure to report recaptures (Bayley and Prince, 1994; Jones and Prince, 1998). Little acoustic tracking has been conducted on white marlin (Skomal and Chase, 2002; n=2 tracks), but similar work on other istiophorid spe- cies indicates relatively high postrelease survival for periods ranging from a few hours to a few days for fish released from recreational fisheries (e.g., sailfish: Jolley and Irby, 1979; blue marlin: Holland et al, 1990; Block et al., 1992; black marlin: Pepperell and Davis, 1999). However, data from acoustic tracking studies bear limi- tations and biases that preclude their use in estimating billfish postrelease survival (Pepperell and Davis, 1999; Graves et al., 2002). In the absence of better data, all recreationally released billfishes have been assumed to survive (Peel, 1995), and estimates of white marlin postrelease mortality are currently not incorporated into ICCAT landing statistics or assessments (White Marlin Status Review Team, 2002). Developments in pop-up satellite archival tag (PSAT) technology have greatly improved scientific under- standing of the behavior, movements and postrelease survival of highly migratory marine fishes, including bluefin tuna (Block et al., 2001), swordfish (Sedber- ry and Loefer, 2001), white sharks (Boustany et al., 2002), blue marlin (Graves et al., 2002; Kerstetter et al., 2003), black marlin (Gunn et al., 2003), and striped marlin (Domeier et. al, 2003). To estimate the postrelease survival of billfishes, researchers have used PSAT deployment durations ranging from five days to seven months (Graves et al., 2002; Domeier et al., 2003; Kerstetter et al., 2003). Goodyear (2002) cautioned that longer duration deployments increase the potential for tag shedding, tag malfunction, and data corruption, and may bias postrelease survival estimates by including additional sources of mortality other than the capture event. Graves et al. (2002) con- sidered five days to be an appropriate window to detect mortality in blue marlin released from recreational gear in offshore waters of Bermuda, citing recaptures of blue marlin tagged with conventional tags within five days of the initial tagging event as evidence that some istiophorids may recover sufficiently to resume feeding shortly after capture. Survival estimates for other istiophorid species re- leased from recreational fishing gear may not be ap- plicable to white marlin. One reason may involve body size: recreationally caught blue marlin and striped marlin are generally larger than white marlin. Inter- and intra-specific differences in body size may affect feeding behavior, fight time, handling time, as well as postrelease recovery (Kieffer. 2000). Another reason may involve the different angling techniques used to catch certain istiophorid species. Blue marlin often hook themselves in the mouth and head while aggressively pursuing high speed trolled lures (Graves et al., 2002). In contrast, as white and striped marlin approach a specific baitfish in the trolling spread, many anglers free-spool (i.e., "drop-back") rigged natural baits to feeding marlin to imitate stunned baitfish (Mather et al., 1975). This process increases the probability that straight-shank ("J") hooks rigged with natural baits will damage vital internal areas such as the gills, esophagus, and stomach (Prince et al., 2002a). Recently, several studies have documented a reduction in hook- induced trauma associated with the use of circle hooks in fisheries targeting estuarine and pelagic fishes (Lucy and Studholme, 2002). However, there is little research specifically comparing levels of postrelease survival of pelagic fishes caught on circle and straight-shank ("J") hooks. Prince et al. (2002a) and Skomal et al. (2002) examined hooking locations and injuries in sailfish and bluefin tuna caught on both hook types but lacked postrelease survival data from study animals. Domeier et al. (2003) did not detect a significant difference be- tween striped marlin caught on circle and straight- shank ("J") hooks, although the authors did observe significantly decreased rates of deep-hooking and tissue trauma with circle hooks compared to straight-shank ("J") hooks. We used data recovered from PSATs to estimate the survival of 41 white marlin caught on circle and straight-shank ("J") hooks in the recreational fishery and released in the western North Atlantic Ocean dur- ing 2002-2003. In addition, differences in hooking locations and hook-induced trauma for white marlin caught on circle and straight-shank ("J") hooks were assessed. Methods Tags The Microwave Telemetry, Inc. (Columbia, MD) PTT-100 HR model PSAT tag was used in our study. This tag is slightly buoyant, measures 35 cm by 4 cm, and weighs <70 grams. The body of the tag contains a lithium com- posite battery, a microprocessor, a pressure sensor, a temperature gauge, and a transmitter, all housed within a black resin-filled carbon fiber tube. Flotation is provided by a spherical resin bulb embedded with buoyant glass beads. This tag model is programmed to record and archive a continuous series of temperature, 86 Fishery Bulletin 103(1) Figure 1 White marlin {Tetrapturus albidus) tagged with a Microwave Telemetry PTT-100 HR pop-up satellite tag lA) and conventional streamer tag (B). light, and pressure (depth) measurements, and can withstand pressure equivalent to a depth of 3000 m. Tags programmed to disengage after five days (n = 5) recorded measurements approximately every two min- utes, whereas tags programmed to disengage after ten days («=35) recorded measurements about every four minutes. Additionally, both 5-day and 10-day tag models transmitted archived and real-time surface temperature, pressure, and light level readings to orbiting satellites of the Argos system for 7-10 days following release from the study animals. PSATs were attached to white marlin by an assembly composed of 16 cm of 400-pound test Momoi® brand (Momoi Fishing Co., Ako City, Japan) monofilament fishing line attached to a large hydroscopic, surgical- grade nylon intramuscular tag anchor according to the method of Graves et al. (2002). Anchors were implanted with 10-cm stainless steel applicators attached to 0.3-m, 1-m, or 2-m tagging poles (the length of the tagging pole varied depending on the distance from a boat's gun- whales to the water) and were inserted approximately 9 cm deep into an area about 10 cm posterior to the origin of the dorsal fin and 5 cm ventral to the base of the dorsal fin (Fig. 1). In this region, the nylon anchor has an opportunity to pass through and potentially interlock with pterygiophores supporting the dorsal fin well above the coelomic cavity (Prince et al., 2002b; Graves et al., 2002). When possible, a conventional tag was also implanted posterior to the PSAT. Deployment White marlin were tagged in the offshore waters of the U.S. Mid-Atlantic Bight, the Dominican Repub- lic, Mexico, and Venezuela (Table 1). These locations were chosen for vessel availability and seasonal concen- trations of white marlin. All tagging operations were conducted on private or charter recreational fishing vessels targeting billfishes and tunas. White marlin were caught on 20-40 lb class sportfishing tackle and fought in a manner consistent with typical recreational fishing practice (G. Harvey, personal commun.2). The first 41 white marlin caught and successfully positioned boatside were tagged. Fish were not brought to the boat until they were sufficiently quiet to facilitate optimal tag placement. When possible, crew members positioned white marlin for tagging by holding them by the bill and dorsal fin in the water alongside the boat, a technique often used when controlling a billfish to remove hooks. On boats with high gunwhales that prohibited holding the captured fish by the bill, the marlin were "leadered" to the boat's side and moved into position for tagging when calm. Six hooked white marlin escaped prior to tagging because frayed leaders broke or hooks slipped during this process. Hooks were removed when feasible; 2 Harvey, G. 2002. Personal commun. Guy Harvey Enter- prises. 4350 Oakes Rd. Suite 518. Davie, FL 33314. Horodysky and Graves: Estimation of survival of Tetrapturus albidus caught and released in the North Atlantic recreational fishery 87 Table 1 Summary of white marlin {Tetrapturus albidus) tagging locations during 2002- 2003. Location Dates of tagging Tag deployment duration (in days) Number of tags deployed Mid-Atlantic Coast 2002: 18-22 Aug, 5-21 Sep 10 11 2003: 22 Aug 10 1 Punta Cana, Dominican Republic 2002: 15-19 May 5 5 Isla Mujeres, Mexico 2003: 10-12 June 10 3 La Guaira, Venezuela 2002 : 23-25 Nov 10 6 2003: 12-13 Sep 1 Oct 10 15 otherwise, they were left in the fish and the leader was cut as close to the animal as possible prior to release. Both practices are common in the recreational billfish fishery. After capture and positioning alongside tagging vessels, six white marlin were observed to have lost color, and were lethargic and unable to maintain vertical posi- tion in the water. These fish were resuscitated alongside the moving boat for 1-5 minutes prior to release — also a common practice in the recreational fishery. Gear type, fight time, handling time, fight behav- ior, hooking location, overall fish condition, estimated weight, and GPS coordinates of the release location were recorded for each tagged white marlin. Fight time was defined as the interval from the time the fish was hooked to the time it was "leadered" alongside the boat prior to tagging. Handling time included tagging and resuscitation, if applicable. In accordance with Prince et al. (2002a), straight-shank ("J") hooks were defined as those with a point parallel to the main hook shaft, whereas circle hooks were defined as having a point perpendicular to the main hook shaft. All circle and straight-shank ("J") hooks were rigged with dead bal- lyhoo (Hemiramphus brasiliensis) bait. Size 7/0 Mustad straight-shank ("J") hooks (models 9175 and 7731) were rigged with the hook exiting the ventral surface of the ballyhoo. Two models of circle hooks were employed in this study: Mustad Demon Fine Wire (model C39952BL, size 7/0; 5° offset, «=9) and Eagle Claw Circle Sea (L2004EL, sizes 7/0-9/0; non-offset, n = ll). All circle hooks were rigged so that they pointed upwards from the head of the ballyhoo (see Prince et al., 2002a). The rigging designations and fishing techniques unique to each hook type were maintained in our study to reflect the usual application of circle and straight-shank ("J") hooks in the white marlin recreational fishery. Other than these differences, all handling, tagging, and re- cording methods were the same for both treatments. Hooking locations were pooled into two categories: jaw, externally visible (including all lip-hooked, foul- hooked, and bill-entangled white marlin) and deep, not externally visible (including all white marlin hooked in the palate, gills, esophagus, and everted stomachs). Bleeding was recorded as present or absent, and the general location of bleeding was recorded when it was possible to identify the source. Data analysis Survival of released white marlin was determined from two distinct lines of evidence provided by the satellite tags: net movement, and water temperature and depth profiles. Time series of water temperature and depth measurements taken about every 2 minutes (5-day tags) or 4 minutes (10-day tags) were used to discriminate surviving from moribund animals. Net movement was determined as a minimum straight line distance trav- eled between the coordinates of the initial tagging event and the coordinates of the first reliable satellite contact with the detached tag (inferred to be the location of tag pop-up) derived from Argos location codes 1, 2, or 3 for the first or second day of transmission. In cases where tags did not report more precise location codes, an average of all location code 0 readings for the first day of transmission was used as a proxy for the loca- tion of the tag pop-up. To determine the directions (and magnitudes) of observed surface currents in areas where fish were tagged, GPS coordinates (Argos location codes of 1, 2, or 3, or a daily mean of location code 0, for tags lacking these) were plotted for the 7-10 days that the tags were floating at the surface and transmitting data to satellites. Maps, tracks, and distances were generated by using MATLAB (version 6.5, release 13.1, Mathworks Inc, Natick, MA). Cochran-Mantel-Haenszel (CMH) tests were used to address the effect of circle and straight-shank ("J") hooks on survival, hooking location, and the degree of hook-induced trauma. A Yates correction for small sample size was applied when expected cell values were less than 5 (Agresti, 1990). The effects of fight time and total handling time on survival were assessed with Wilcoxon-Mann-Whitney exact tests, with the null hy- pothesis that there was no difference between surviving and moribund white marlin. All statistical analyses were conducted by using SAS (version 8, SAS Institute, Cary, NC). The lone nonreporting tag observed in our study was excluded from all subsequent analyses. 88 Fishery Bulletin 103(1) We conducted bootstrapping simulations to examine the effect of sample size on the 95% confidence intervals of the release mor- tality estimates using software developed by Goodyear (2002). Distributions of estimates were based on 10,000 simulations with an underlying release mortality equivalent to that observed for straight-shank ("J") hooks for experiments containing 10-200 tags and no sources of error (e.g., no premature re- lease of tags, no tagging-induced mortality, and no natural mortality). Results Forty-one white marlin were tagged in four geographic locations during 2002-2003 (Table 1). Information for each fish is summa- rized in Table 2. Fight times were fairly typi- cal for this fishery (mean: 15.8 min, range: 3-83 min), although two animals required more than 30 minutes before they were suf- ficiently calm at boatside for tag placement. Overall, forty tags (97.6%) transmitted data to the satellites of the Argos system and of these, thirty-seven tags remained attached to study animals for the full five- or ten- day duration. One five-day tag was released prematurely from a surviving white marlin after 2.5 days, presumably because it had not been attached securely. This individual showed behavior similar to other surviving white marlin while the tag was attached and was presumed to have survived for the purposes of our study. Additionally, two 10- day tags attached to moribund white marlin disengaged from the carcasses prior to the expected date after an extended amount of time at a constant depth and temperature on the seafloor. Approximately 61% of data (range: 19-95%) were successfully transmit- ted from reporting tags. Overall, 33 of 40 tags (82.5%) returned data that indicated the survival of tagged animals throughout the duration of tag de- ployment. Surviving white marlin exhib- ited daily variations in water temperature and depth data while carrying PSATs (Fig. 2A). The net movement of surviving ani- mals could not be explained by the speed or direction of current patterns alone over the course of the tag deployment (Table 2, Fig 3A). In contrast, moribund white mar- lin (Fig. 2B) sank to the seafloor (237-1307 m) and to constant water temperatures (3.7-12.5°C), where they remained until the tags disengaged and floated to the surface not far from the initial tag- ging location (Fig 3B). Five of the seven moribund white marlin died within the first six hours of release; Surviving white marlin 30 25 15 10 rr S. "60 80 120 08/22 08/23 08/24 08/25 08/26 08/27 08/28 08/29 08/30 08/31 09/01 09/02 Moribund white marlin 08/18 08/19 08/20 08/21 08/22 08/23 08/24 08/25 08/26 08/27 09/28 09/29 Figure 2 Depth and temperature tracks for a surviving (A) (MA12) and moribund (B) (MA01) white marlin (Tetrapturus albidus). Filled symbols correspond to measurements taken while tags were attached to animals, hollow symbols refer to measurements taken after pop- up while tags were transmitting data to Argos satellites. Gray bars denote periods of local night. four of these five animals died within the first hour (Table 2). The two white marlin that experienced the longest fight times (46 and 83 min) died more than 24 hours following their release. White marlin VZ03-11 had a Horodysky and Graves: Estimation of survival of Tetrapturus albidus caught and released in the North Atlantic recreational fishery 89 Table 2 Summary information for tagged white marlin (Tetrapturus albidus) released from recreational fishing gear in the western North Atlantic Ocean. Total fight time is defined as the interval between the time that the fish was hooked and the time that it was brought to the side of the boat prior to tagging. Handling time included tagging and resuscitation, where applicable. "D/N" refers to deep, not externally visible hooking locations, "foul" refers to a white marlin hooked in the dorsal musculature. Tail- wrapped fish are denoted with the symbol "T", resuscitated marlin are denoted with the symbol "R". Estimated Fight Handling Location Fate Movement weight time time Hook of hook Bleeding (living or (nmi/km Tag number cbl umcesedu Kristen Munk Alaska Department of Fish and Game Division of Commercial Fisheries 1255 W. 8th Street Juneau, Alaska 99801 Kenneth H. Coale Moss Landing Marine Laboratories California State University 8272 Moss Landing Road Moss Landing, California 95039 Brian R. Frantz Center for Accelerator Mass Spectrometry Lawrence Livermore National Laboratory 7000 East Avenue Livermore, California 94551 Gregor M. Cailliet Moss Landing Marine Laboratories California State University 8272 Moss Landing Road Moss Landing, California 95039 Thomas A. Brown Center for Accelerator Mass Spectometry Lawrence Livermore National Laboratory 7000 East Avenue Livermore, California 94551 Manuscript submitted 11 April 2003 to the Scientific Editor's Office. Manuscript approved for publication 24 August 2004 by the Scientific Editor. Fish. Bull. 103:97-107 (2005). Rockfishes {Sebastes spp.) comprise one of the most commercially impor- tant fisheries in the northeast Pacific Ocean. Some rockfish species possess life history characteristics, such as long life, slow growth, late age at maturity, low natural mortality, and variable juvenile recruitment success, all of which make them particularly vulnerable to overfishing (Adams, 1980; Archibald et al., 1981; Leaman and Beamish, 1984; Cailliet et al., 2001.1. Rockfish population biomass and size composition have declined to very low levels today in part because of continued high exploita- tion rates (Love et al., 2002). Preven- tion of further population declines is a management imperative. Sustain- able management of marine fisheries requires accurate life history infor- mation, of which validated age and growth characteristics can be one of the most important aspects. Underestimated age can lead to inflated estimates of total allow- able catch for a fishery that is un- sustainable at that level of exploita- tion (Beamish and McFarlane, 1983; Campana, 2001). For example, un- derestimated longevity and improper management allowed overfishing that accelerated the decline of the Pacific ocean perch (Sebastes alutus) of the northeastern Pacific Ocean (Beamish, 1979; Archibald et al., 1983). Reliable estimates of age are also essential for understanding life history traits, 98 Fishery Bulletin 103(1) such as age at maturity, rate of growth, longevity, and reproduction frequency (Beamish and McFarlane, 1983). For production (large-scale) aging purposes, age vali- dation is especially important because it provides a standardized basis for ongoing aging efforts to identify strong and weak cohorts (Campana, 2001). The most common method of age estimation of bony fishes is counting growth zones in their calcified in- ner ear bones, or otoliths (Chilton and Beamish, 1982; Beamish and McFarlane, 1987). A pair of translucent and opaque growth zones is often assumed to represent one year of growth (Williams and Bedford, 1974). By counting growth zones an estimate of fish age is possi- ble; however, growth patterns are not easily discernible for all species. Age interpretations in long-lived species can be particularly difficult and subjective because of the compression of growth zones within the otolith (Munk, 2001). Therefore, it is necessary to validate the periodicity of growth zones in otoliths with an indepen- dent and objective method. Despite the importance of accurate age estimates for understanding and manag- ing fish populations, validated age and growth charac- teristics are often not available (Beamish and McFar- lane, 1983; Campana, 2001). Traditional age validation techniques, such as captive rearing, mark-recapture, and tag-recapture, can be difficult or impractical for long-lived and deep-dwelling fishes. An alternative technique to traditional age valida- tion methods uses radiocarbon (14C) produced by the atmospheric testing of thermonuclear devices in the 1950s and 1960s as a time-specific marker (Kalish, 1993). This established method of validating otolith- derived age estimates of fishes involves relating the discrete temporal variation of 14C recorded in otoliths to an established 14C chronology. Otoliths are closed systems, accreting calcium carbonate throughout the life of the fish and this calcium carbonate is conserved through time. The measurement of bomb-produced 14C in otoliths of fishes is considered one of the best objec- tive means to validate otolith-based age estimates in long-lived fishes (Campana, 2001). This technique is most reliable for fishes that inhabit the surface mixed layer of the ocean, at least during a portion of their life history. Uncertainty regarding mix- ing rate at depth and limited data on the 14C signal in deeper waters make it difficult to use this technique for organisms that live below the mixed layer throughout their lives (Kalish, 1995, 2001). Studies indicate that the main source of carbon (70-90%) for otoliths is from dissolved inorganic carbon (DIC) in seawater and that the remainder (10-30%) is dietary (Kalish, 1991; Far- rell and Campana, 1996; Schwarcz et al., 1998). An understanding of the life history of the fish (in par- ticular diet, movement, and habitat) and the regional oceanography of the area are integral for interpreting otolith 14C data. One caveat of this technique is that it must use otoliths with birth dates, including the period of initial increase in 14C (mid-1950s to mid-1960s; Ka- lish, 1995). Consequently, this technique is well suited for age validation of long-lived species or species for which there is an archived otolith collection with birth years that span this period. The application of bomb 14C for age validation of long-lived species is advantageous, in that it provides a minimum longevity and verifies the periodicity of growth zones in otoliths with only a small amount of material and with a high degree of precision (Kalish, 1993; Campana, 2001). However, the high cost of 14C analysis (~$400-$500 per sample) has been a limiting factor in the widespread application of this technique. The quillback rockfish (Sebastes maliger) is a com- mercially important rockfish that represents a portion (~8%) of the demersal shelf rockfish assemblage land- ings in the Gulf of Alaska (O'Connell et al.1). Species within the demersal shelf group are considered long- lived, late maturing, and sedentary as adults, making them highly susceptible to fishing pressure (O'Connell et al.1). Estimated exploitation rates are low; once ex- ploited beyond a sustainable level, recovery is slow (Lea- man and Beamish, 1984; Francis, 1985; O'Connell et al.1). Longevity estimates for the quillback rockfish are wide ranging, from 15 to 90 years (38 years. Barker, 1979; 55 years, Richards and Cass, 1986; 15 years, Reilly et al.2; 76 years, Yamanaka and Kronland, 1997; >32 years, Casillas et al., 1998; 90 years, Munk, 2001), and no age validation has been performed for this spe- cies to date. Quillback rockfish are found associated with rocky substrate in relatively shallow continental shelf waters (9 to 146 m) — their abundance decreasing with increas- ing depth below 73 m (Kramer and O'Connell, 1995). As juveniles, quillback rockfish inhabit nearshore benthic habitat. Tagging studies confirmed that they do not demonstrate migratory behavior and are residents in their shallow-water habitat (Matthews, 1990). Because 1) most longevity estimates indicate that some present- day adult quillback rockfish were born in the prebomb era, 2) quillback rockfish in the juvenile stage are found in the ocean mixed layer, and 3) a suitable 14C time series exists for the waters off southeast Alaska (previ- ously determined from yelloweye rockfish [S. ruberri- mus] otoliths [Kerr et al., 2004]), the quillback rockfish is an ideal candidate for 14C age validation. The objectives of our study were 1) to develop a meth- od for determining the minimum number of samples required for bomb 14C age validation to minimize cost, 2) to validate both age and age estimation methods of the quillback rockfish by measuring 14C in aged otoliths and to compare the timing of the initial rise in 14C with 1 O'Connell, V. M., D. W. Carlile, and C. Brylinsky. 2002. De- mersal shelf rockfish assessment for 2002. Stock assessment and fishery evaluation report for the groundfish resources of the Gulf of Alaska, 36 p. North Pacific Fishery Management Council (NPFMCl, P.O. Box 103136, Anchorage AK 99510. 2 Reilly, P. N., D. Wilson-Vandenberg, R. N. Lea, C. Wilson, and M. Sullivan. 1994. Recreational angler's guide to the common nearshore fishes of Northern and Central California. California Department of Fish and Game, Marine Resources Leaflet, 57 p. Calif. Dep. Fish and Game, 20 Lower Ragsdale Drive, Suite 100, Monterey, CA 93940. Kerr et al.: Age validation for Sebastes maliger with bomb radiocarbon 99 the chronology determined for the waters of southeast Alaska (i.e., yelloweye rockfish), and 3) to analyze 14C in aged quillback rockfish otoliths, spanning the pre- to postbomb era, in order to examine the complete 14C time series and demonstrate the effectiveness of using the timing of the initial rise in UC as an age valida- tion method. Materials and methods Sample size assessment Because of the considerable cost of accelerator mass spectrometry (AMS) analyses, the minimum number of 14C samples required to validate the aging method of the quillback rockfish was mathematically determined from a previously determined yelloweye rockfish otolith 14C time series for the waters of southeast Alaska (Kerr et al., 2004). To assess minimum sample size, estimated years of initial rise in 14C levels (and associated errors) were determined for different numbers of data points (n = 3, 5, 7, 9, and 11) subsampled from the bomb-rise region of the yelloweye rockfish data set. The estimated years of initial rise from each subsample set were then compared to the initial year of rise and error deter- mined from all bomb-rise yelloweye rockfish 14C samples (n=23). Because the error associated with the yelloweye rockfish bomb-rise data set is limited by the uncertainty associated with age estimates for yelloweye rockfish, a maximum error of ±2 years for fish with birth years during the bomb rise (1956 to 1971; Kerr et al., 2004) was our precision criterion for defining the minimum number of quillback rockfish otolith samples. Twenty-three yelloweye rockfish otolith 14C values with birth years from 1956 to 1971 were divided into data sets of 3, 5, 7, 9, and 11 data points. A stratified sampling approach was applied by creating bomb 14C linear regressions from repeated selection of 3, 5, 7, 9, and 11 data points at uniform intervals from 1956 to 1971. Random selection of data points was not practi- cal because it is established that the careful choice of sample year during the rapid rise in 14C is required for this technique (Baker and Wilson, 2001). The year of initial rise in 14C, and associated error, was determined from the bomb 14C regressions. The year of initial 14C rise was calculated with the following formula: x = (y - b)lm, where x = year of initial rise in 14C values; y = average prebomb 14C value; m = slope of the line; and b = y-intercept. The error associated with the year of initial rise in 14C values (ox) was calculated by using the delta method (treating 6 as a scaler; Wang et al., 1975): °y/°m> where av = error associated with average prebomb 14C value am = error associated with the slope of the line. Radiocarbon analysis Sagittal otoliths of quillback rockfish were collected from a random subsampling of catches from commercial long- line fishing vessels in the coastal waters off southeast Alaska by the Alaska Department of Fish and Game (ADFG), Juneau, AK in 2000 (Fig. 1). A single otolith from a pair taken from each fish was aged by using the break-and-burn method developed by researchers at the Mark, Tag, and Age Laboratory, ADFG in Juneau, AK, and the corresponding intact otolith was analyzed for 14C. Whole and broken-and-burned otoliths were stored dry in paper envelopes. Year of capture, estimated final age, assigned year class, readability code, and reader identification information were archived and provided by ADFG for each sample. Fifteen quillback rockfish otoliths, with estimated birth years ranging from the prebomb 1950s to the postbomb mid-1980s were selected for 14C analysis. The core of each otolith, which constitutes the first year of growth, was analyzed for 14C. From life history infor- mation, it is known that the core was formed while the fish inhabited the ocean mixed layer during its early growth stage (Yoklavich et al., 1996). To determine the average length and width, and minimum depth of the core, whole and broken-and-burnt otoliths from adult quillback rockfish were examined under a Leica- dis- secting microscope with an attached Spot RT® video camera and were measured using Image Pro Plus® im- age analysis software (version 4.1 for Windows, Media Cybernetics, Silver Spring, MD). Cores were extracted with a milling machine with a 1.6-mm (1/16") diam- eter end mill. To minimize the extraction of material deposited after the first year of growth, length, width, and depth parameters of the otolith core were used to guide coring. Because the first year of growth in quill- back rockfish otoliths is clearly visible from the distal surface of the otolith we were able to visually correct for individual variability in otolith core size. The core (first year of growth in the otolith) was reduced to powder, collected, and weighed to the nearest 0.1 mg. For 14C analysis, otolith calcium carbonate (CaC03) was converted to pure carbon in the form of graphite (Vogel et al., 1984, 1987) and measured for 14C content by using AMS at the Center for Accelerator Mass Spec- trometry, Lawrence Livermore National Laboratory. The 14C values were reported as 414C (Stuvier and Polach, 1977). The 14C values measured in quillback rockfish otolith cores were plotted with respect to corresponding birth years assessed from break-and-burn age estimates, taking into consideration the potential variation of the age estimate (coefficient of variation=2.6%, rounded to the nearest whole number; Chang, 1982). The 14C time series for the waters of southeast Alaska established from the otoliths of the age-validated yelloweye rockfish 100 Fishery Bulletin 103(1) 137° 136° 135° 134° 133° 132° 131° 130° Figure 1 Map of southeast Alaska with regions where quillback rockfish (Sebastes maliger) used for otolith radiocarbon analyses were captured. Quillback rockfish were collected from random subsam- pling of catches from commercial longline fishing vessels in the coastal waters off southeast Alaska (CSEO: Central Southeast Offshore (outside), SSEI: Southern Southeast Inshore, SSEO: Southern Southeast Offshore, and NSEI Northern Southeast Inshore (inside)) by the Alaska Department of Fish and Game, Juneau, AK, in 2000. Note that the specific geographic location for individual fish during the first year of life is unknown; how- ever, life history information indicates that quillback rockfish are not migratory and exhibit residential behavior in shallow- water habitat. Hence, the general location of the fish collected and used in this study may be useful in a broad context. (rc=43) was used for temporal calibration of the quill- back rockfish record (Andrews et al., 2002; Kerr et al., 2004). The level of concordance between the years of initial rise in 14C in the two time series was the basis for validating the otolith-based age estimates of the quillback rockfish. The degree of agreement between the 14C time series, spanning the pre- to postbomb era, for the quillback and yelloweye rockfishes was examined to demonstrate the effectiveness of determining the year of initial rise in 14C as an age validation method, and whether the entire time series for the quillback provided any further information relevant to age validation. To do this, the yelloweye rockfish 14C time series was di- vided into three intervals (prebomb, bomb rise, and postbomb) and fitted with confidence intervals. The pre- bomb era 14C values (1950-57) were fitted with an aver- age (±2 SD); the bomb rise (1959-71) and postbomb era values (1966-85) were fitted with a linear regression and corresponding 95% prediction intervals. A qualita- tive comparison of the quillback rockfish 14C record was made with other existing marine records: two Hawaiian Islands coral records — Oahu (Toggweiler et al., 1991) Kerr et al.: Age validation for Sebastes maliger with bomb radiocarbon 101 Table 1 Range of the year of calculated initial rise in radiocarbon and the associated range of error calculated for bomb radiocarbon regressions. Each regression comprised varying numbers of yelloweye rockfish radiocarbon data points (n = 3, 5, 7, 9, and 11) and was compared to the year of initial 14C rise and error was determined from all bomb-rise yelloweye rockfish 14C samples (n=23, last row) to determine the minimum number of quillback rockfish otolith samples sufficient to achieve the desired degree of precision (±2 years). Number of data points in regression 3 5 7 9 11 23 Number of regressions Range of the year of calculated initial rise in 14C Error range (± years) 1954.1-1960.3 1956.0-1959.4 1956.5-1957.9 1957.1-1957.3 1957.0-1957.8 1957.3 0.8-6.8 1.3-2.9 1.0-2.5 0.9-1.8 1.2-1.5 n/a and Kona (Druffel et al., 2001)— and two otolith-based northern hemisphere 14C records — for northwest At- lantic haddock (Campana, 1997) and the Barents Sea Arcto-Norwegian cod (Kalish et al., 2001). Results Sample size assessment The estimated years of initial rise in 14C calculated for the bomb-14C regressions, composed of 3, 5, 7, 9, and 11 yelloweye rockfish data points spanning 1956 to 1971, converged towards the calculated year for all 23 data points as the number of samples comprising the regressions increased (Table 1). In parallel, the errors associated with the estimated years of initial rise in 14C decreased as the number of 14C samples increased (Table 1). The degree of precision within the quillback rock- fish record was limited by the uncertainty associated with age estimates for yelloweye rockfish (a maximum error of ±2 years based on growth zone counts for fish with birth years from 1956 to 1971; Kerr et al., 2004). Examination of the error (years) associated with the year of initial rise in 14C for the number of data points comprising each regression in relation to our ±2 year criterion indicated that a sample size of nine data points resulted in error values that ranged below 2 years (Table 1). Therefore, it was concluded that nine 14C samples spanning 1956-71 would be sufficient to provide a suit- able degree of precision in the quillback rockfish record. In addition, a limited number of samples, in this case 4, were required to establish an average prebomb level for the intercept year. Radiocarbon analysis The 14C measured in 15 previously aged quillback rock- fish otoliths with presumed birth years from 1950 to 1985 varied considerably over time (Table 2). Otoliths Table 2 Summary offish and otolith data from qui llback rockfish collected off the coast of southeast Alaska . Resolved age is the final age estimate given by Alaska Department of Fish and Game. Birth year is the collection year (2000) minus the resolved age Age error is the uncertainty asso- ciated with the age estimate (CV=2.6%; year rounded to the nearest whole number). Radiocarbon values in the otolith cores of yelloweye rockfish are expressed as 414C with the AMS analytical uncertainty. Resolved age Birth year 414C (years) (± age error) (%c) 50 1950 ±1 -76.9 ±3.3 46 1954 ±1 -104.8+3.2 45 1955 ±1 -89.0 ±4.0 43 1957 ±1 -92.2 ±3.8 41 1959 ±1 -66.9 ±3.3 40 1960 ±1 -54.7 ±4.2 39 1961 ±1 -57.8 ±3.7 37 1963 ± 1 -49.1 ±3.3 35 1965 ±1 28.2 ±3.7 33 1967 ±1 105.4 ±4.2 31 1969 ±1 19.4 ±4.0 30 1970 ±1 47.5 ±3.6 25 1975 ±1 43.9 ±4.0 20 1980 ±1 76.3 ±5.5 15 1985 ±0 15.4 ±3.7 from quillback rockfish with birth years 1950-57 con- tained prebomb 14C levels. Although there was more variation in these prebomb values than expected from 14C uncertainties, the level was relatively consistent over time, averaging -90.7 (±11.5)%c (mean ±SD). A sharp rise in otolith 14C values was evident in 1959 (±1 year); 102 Fishery Bulletin 103(1) 200 150 100 ~ 50 O < 0 -50 -100 -150 ° Quillback rockfish (n=15) ^^— Exponential Rise - - - Mean prebomb value (-907 %o) Two sigma value (-67.7 %„) * ^ 1930 1940 1950 1 960 1 970 Birth year 1980 1990 2000 Figure 2 Radiocarbon (414C) values for quillback rockfish iSebastes maliger) otolith cores (n = 15) in relation to estimated birth year. Horizontal error bars represent the age estimate uncertainty from growth zone counts (CV=2.6%, year rounded to the nearest whole number) and vertical error bars represent the 1-aAMS (accelerator mass spectometry) analytical uncertainty. The solid line represents the exponential curve fitted to the data that was used to determine the year of initial rise in 14C levels from prebomb levels (the fitted function had the form Y=A+B exp(CX) with Y =14C, X=birth year, and A, B, and C as fitted param- eters). The dashed line represents the +2 SD level (-67.7%r) associated with the average prebomb 14C value (-90.7 ±11.5r/rr ; dotted line); the intersection of the +2 SD line and the curve was used to define the year-of-initial-rise in 14C values. this sample was the first to have a 14C value (-66.9 [±3.3]%e) that was above prebomb radiocarbon levels with a +2 SD criteria (upper limit of-67.7%r). This first indication of a rise in 14C related to the rise of the bomb was in agreement with the exponential fit of the quill- back rockfish 14C times series (Fig. 2). The 14C record for quillback rockfish otoliths peaked in 1967 with a maximum 14C concentration of +105.4 (±4)%e. This peak was followed by a generally declining, but inconsistent, trend in 14C values to 1985 (last birth year sampled). The 14C values measured in quillback rockfish oto- liths plotted against estimated birth years produced a characteristic increasing and decreasing curve rep- resentative of bomb-generated 14C changes over time (Fig. 3). The quillback rockfish 14C record was syn- chronous with a 14C time series for southeast Alaskan waters determined from yelloweye rockfish otoliths (Kerr et al., 2004); the average prebomb 14C values for the quillback rockfish were in close agreement with the average yelloweye rockfish prebomb levels (-102.2 [±9.3]%c [mean ±SD]). The year of initial rise in the quillback and yelloweye rockfish records (1959 [±1 year] cf. 1958 [±2 years]) and peak in 14C values (1967 cf. 1966) for these two species coincided within one year, a period encompassed within the uncertainty associated with break-and-burn age estimates. Furthermore, the postbomb decline in quillback rockfish 14C values was similar to that of the yelloweye rockfish. In addition, thirteen of the fifteen quillback rockfish 14C values fell within the confidence intervals of the yelloweye rockfish 14C curve (Fig. 3). The comparison of the quillback rockfish 14C record with that for Hawaiian Islands corals (Toggweiler et al., 1991; Druffel et al., 2001) and two otolith-based north- ern hemisphere 14C chronologies (northwest Atlantic haddock [Campana, 1997] and Barents Sea Arcto-Nor- wegian cod [Kalish et al., 2001]) revealed similarities in the year of initial rise and rate of rise of 14C values, and differences in the pre- and postbomb eras that can be explained by regional oceanographic effects (Fig. 4). Discussion Sample size assessment Although the 14C technique has great potential for vali- dating the age of many long-lived fishes, one of the main disadvantages has been the high cost of AMS 14C analyses. By providing a means of defining the Kerr et al.: Age validation for Sebastes maliger with bomb radiocarbon 103 200- 150- 100- _ 50 o 0 * -50 -100 -150 -200 • Yelloweye rocktish (n=43) □ Quillback rockfish (n=15) 1930 1940 1950 1960 1970 Birth year 1980 1990 2000 Figure 3 Radiocarbon (414C) values from quillback rockfish [Sebastes maliger) otoliths and the yelloweye rockfish (S. ruberrimus) 14C time series for the waters of southeast Alaska. The yelloweye rockfish 14C data were divided into three intervals (prebomb, bomb rise, and postbomb) and fitted with confidence intervals. The prebomb era 14C values (1950-571 were fitted with an average (±2 SD), and the bomb rise (1959-71) and postbomb era values (1966-85) were fitted with a linear regression and corresponding 95% prediction intervals. minimum number of samples required to achieve the desired degree of precision, the present study takes a step toward reducing the number of prescribed samples, (i.e., 20-30 otoliths; Campana, 2001), effectively making age validation more affordable. To determine the number of samples necessary for an age validation study, an assessment of the degree of precision is required. The degree of precision may be defined by the level of variation in the chronology or the uncertainty associated with age estimates. It can also be dependent on the estimated longevity of the fish and the resolution of age that is sufficient for the purposes of the study. For example, a resolution of ±5 years may be sufficient for a species estimated to live 100 years, but would not be satisfactory for a species estimated to live 20 years. The higher degree of preci- sion, the greater the cost will be for a study. However, a maximum precision can be attained at a minimum cost by taking into consideration the precision of the 14C time series, the error associated with age estimates, and the age resolution necessary to accomplish the goals of the study. In our study, the ±2-year variation of the yelloweye rockfish 14C time series limited the precision to which the age of the quillback rockfish could be determined through comparison. Stratified sampling of nine quill- back rockfish 14C values between 1956 and 1971 re- vealed an average year of initial rise in 14C of 1959 (±1 year) that was in close agreement with the year of initial rise determined for the yelloweye rockfish time series. Thus, given the unique circumstances for this species, we have quantitatively reduced the number of samples required for age validation to 9 (given that some sampling or additional information is used to es- tablish prebomb levels). It can also be envisioned that 14C analysis of a single fish otolith could establish a minimum longevity for a species if the 14C levels mea- sured in the otolith core of an adult fish with a known capture year were consistent with established prebomb 14C levels for the regional waters in which that fish spent its first year. This exercise illustrates the neces- sity of defining precision on a species-by-species basis prior to beginning a 14C study. Despite the high cost of AMS analyses, the overall project cost may be lower and of shorter duration than traditional age validation studies because of the relatively short time required to prepare and process the minimum number of otoliths. Currently, the 14C technique is considered one of the most effective methods for age validation of long-lived fishes (Campana, 2001) and as costs are minimized, future application of the bomb 14C age-validation tech- nique of marine fishes should increase. Radiocarbon analysis To interpret radiocarbon values recorded in marine or- ganisms it is essential to put them in the context of the regional oceanography. The Alaska coastal current, driven by wind stress and enriched with freshwater runoff, is the driving force behind the coastal dynamics off southeast Alaska (Royer, 1982). The coastal environ- ment off southeast Alaska is characterized by significant 104 Fishery Bulletin 103(1) 200 i □ Quillback rockfish •* • *> 150 ■ O Hawaiian Island corals * "" V A Arcto-Norwegian cod fii * * 100 - • North Atlantic haddock » , > * \ ' 1 50 - • :r;- o V** - J ° n < 0 - •> ■ t -50 - -J *'**.. K—ff' \ ten.10 1 * . • *ta -100 - nn -150 J 1 1 1 ' 1 1900 1920 1940 1960 1980 2000 Year Figure 4 Radiocarbon data (zl14C) from otolith cores of quillback rockfish tSebastes maliger), Hawaiian hermatypic corals (Toggweiler et al., 1991; Druffel et al., 2001), and two age-validated fishes, the northwest Atlantic haddock iMelanogrammus aeglefinus; Campana, 1997) and the Arcto-Norwegian cod (Gadus morhua; Kalish et al., 2001). Note the strong agreement in the timing of the year of initial rise in 14C values. downwelling, high wind stress, eddies, and storm activ- ity, resulting in a high degree of mixing. The rapid rise and early peak recorded in quillback rockfish otoliths, followed by a postbomb decline, indicated rapid ocean- atmosphere gas exchange in the shelf waters off south- east Alaska. Shallow continental shelf waters, such as the environment inhabited by juvenile quillback rock- fish, have a thin mixed layer and relatively long surface residence time, resulting in a relatively fast response and build up of bomb-14C from the atmospheric signal. In addition, low prebomb 14C values in the quillback rockfish record may indicate the influence of upwelled 14C-depleted waters on southeast Alaskan coastal sur- face waters. This is expected because surface waters sampled off the Alaskan Peninsula (GEOSECS; Ostlund and Stuvier, 1980) in 1973 had low 14C values (+62%c) in relation to the subtropical Pacific (Oahu coral, +174. 5%r in 1973, Toggweiler et al., 1991), indicating the influence of upwelled 14C-depleted waters. A comparison of the 14C time series determined from quillback rockfish otoliths to the established 14C time se- ries exhibited synchronicity with the global rise in radio- carbon. The quillback rockfish record and high latitude northern hemisphere records from Arcto-Norwegian cod and haddock exhibited nearly identical years of initial rise and rates of 14C increase. Note that there are differ- ences among these records in the prebomb and peak 14C levels attained and the behavior of bomb-14C after the peak, but it is irrelevant to the utility of the technique as an age validation tool. The quillback rockfish record was also temporally similar to a Hawaiian Island corals record (Oahu, Toggweiler et al., 1991; Hawaii, Druffel et al., 2001), both increasing rapidly from the late 1950s. However, as expected, the corals had higher prebomb levels (-50%o cf. -90%o), a later peak (1971 cf. 1967) at a higher value (YlA%c cf. 105%c), and the indication of a more rapid decline in the postbomb years. These dif- ferences are indicative of the different oceanographic influences on the subtropical waters (e.g., lesser relative influence of upwelled, 14C-depleted, deep water). Possible sources of error in the quillback rockfish 14C record are the specific location of each fish during its first year of growth, possible inaccuracies in the method of extracting the core, age estimate uncertainty, and variable oceanographic conditions during the year of otolith formation. The unknown geographic location of individual fish during the first year of life is a potential source of 14C variation. Although juvenile quillback rockfish occupy relatively limited regions, factors such as local bathymetry, coastal upwelling, and freshwater input are likely to impact the 14C content of the lo- cal waters. Two of the quillback otolith samples (birth years 1967 and 1980) had considerably higher (~50%o) 14C values when compared to the highest yelloweye rockfish value for that same year. These elevated 14C values may indicate that the individuals resided in different water masses. The variability of otolith 14C values from regional effects is evident in the observed ±11.5%o (1 SD) associated with prebomb values, a higher variability than expected from the analytical uncertain- Kerr et al.: Age validation for Sebastes maliger with bomb radiocarbon 105 ties of the AMS 14C measurements (~±3-4%o). Elevated 14C levels have also been recorded in otoliths of the black drum {Pogonias cromis), known to reside in es- tuaries during the juvenile stage (Campana and Jones, 1998); these elevated values are attributed to the rapid exchange of atmospheric 14C in the well-mixed estuarine environment and the influence of river input. Quillback rockfish are known to inhabit more nearshore waters than those inhabited by yelloweye rockfish (Love et al., 2002), which could explain the elevated 14C levels. The core-extraction method was designed to limit the inclusion of more recently formed material (older than age 1); however, the inclusion of some of this material may have inadvertently occurred, perhaps introducing error to the quillback rockfish 14C record. This kind of error could alter the 14C value from the actual core year value depending on the time of otolith formation in relation to the bomb 14C signal. A small addition of material with 14C content different from the core ma- terial, however, may not produce a significant change in the timing of the initial rise and the shape of the rise. We feel that in most cases this would lead to an underaging of the fish and provide us with a minimum age estimate. Perhaps the most significant potential source of er- ror is the uncertainty associated with age estimation methods (coefficient of variation=2.6%). Growth zone counting error could have contributed to variation in the quillback rockfish record; however, the otoliths used in our study were chosen specifically to provide clearly definable growth zones and the highest rank in age-es- timate confidence. The samples chosen were best-case examples of precise age determinations. Short-term regional-scale changes in oceanographic conditions, such as upwelling events, may have affected 14C levels at the time of otolith formation. The variation in postbomb measurements exemplifies this factor. Considering the discussions above and the similar biology, ecology, and distribution of the two rockfish species, we believe that the use of the yelloweye rockfish 14C time-series (Kerr et al., 2004) as a means of tempo- ral calibration for the quillback rockfish record is well supported. The year of initial rise in 14C for quillback rockfish otoliths (1959 [±1 year]) is in agreement with the yelloweye rockfish record (1958 [±2 years]); this finding validates the age estimates of the quillback rockfish and the accuracy of the break-and-burn age estimation method. In addition, the concordance of the quillback time series (1950 to 1985) provides further support for the age validation. Note that the 14C levels, timing of the peak, and the subsequent decline were similar between species. In addition, all but two of the quillback rockfish 14C values (sample years 1967 and 1980) fell within the confidence intervals for the yelloweye rockfish 14C curve, further supporting the concordance of the two rockfish records. If there had been consistent underaging or overaging of quillback rockfish otoliths, this discrepancy would have resulted in a chronology that was not in phase with the yellow- eye rockfish time series (Campana et al., 2002). This application of the bomb-14C technique has con- firmed the longevity of quillback rockfish to a minimum of 43 (±1) years. This minimum age estimate is based on the last individual fish sample (estimated birth year of 1957 from growth zone counting) to have prebomb levels, immediately preceding the significant rise in 14C levels observed in 1959 (±1) year. These findings effectively refute previous longevity estimates less than 43 years (Barker, 1979; Reilly et al.2). In addition, it is reasonable to assume that the annual growth pattern continues throughout life; hence, these findings strongly support longevity estimates exceeding 43 years and ranging up to 90 years (Richards and Cass, 1986; Yamanaka and Kronland, 1997; Casillas et al., 1998; Munk, 2001). Conclusions It is our intention to not only validate the age and age estimation method for the quillback rockfish, but to determine the most effective number of samples for age validation with bomb radiocarbon. From our results, it appears that the concordance of the full 14C time series is not entirely necessary for validating the age of fish, and perhaps of any other organism. Because the evolu- tion and magnitude of the bomb-14C rise from the pre- bomb to postbomb era is subject to variations due to the specific oceanography of the region, the 14C time series are in fact regional and are not universally applicable to all validation studies. The agreement of the entire 14C time series does not provide additional information relevant to age validation. Hence, we propose that the year-of-initial-rise method be considered an effective 14C age validation approach. This method both reduces the number of samples required for age validation and effectively precludes the perceived need to establish a pre- to postbomb 14C reference time series for every region of the world's oceans. Because the year of initial rise in 14C levels in surface waters is well defined (1958 [±2 years]), it should be treated as a time-specific marker for organisms that inhabit the mixed layer of the oceans for some or all of their life cycle. Acknowledgments We thank the Alaska Department of Fish and Game for providing aged otolith samples. This article was supported in part by the National Sea Grant College Program of the U.S. Department of Commerce's National Oceanic and Atmospheric Administration under NOAA Grant no. NA06RG0142, project number R/F-190, through the California Sea Grant College Program, and in part by the California State Resources Agency. This work was performed, in part, under the auspices of the U.S. Department of Energy by University of Cali- fornia, Lawrence Livermore National Laboratory under contract no. W-7405-Eng-48. This research was also funded in part by the Pacific States Marine Fisheries Commission, Earl H. and Ethel M. Myers Oceanographic 106 Fishery Bulletin 103(1) and Marine Biological Trust, Packard Foundation, and Project AWARE. Literature cited Adams, P. B. 1980. Life history patterns in marine fishes and their consequences for fisheries management. Fish. Bull. 78:1-12. Andrews, A. H., G. M. Cailliet, K. H. Coale, K. M. Munk, M. M. Mahoney, and V. M. O'Connell. 2002. Radiometric age validation of the yelloweye rockfish iSebastes ruberrimus) from southeastern Alaska. Mar. Freshw. Res. 53:139-146. Archibald, C. P., D. Fournier, and B. M. Leaman. 1983. Reconstruction of stock history and development of rehabilitation strategies for Pacific ocean perch in Queen Charlotte Sound. Canada. N. Am. J. Fish. Manag. 3:283-294. Archibald. C. P., W. Shaw, and B. M. Leaman. 1981. Growth and mortality estimates of rockfish ( Scor- paenidae) from B.C. coastal waters. 1977-1979. Can. Tec. Rep. Fish Aquat. Sci. 1048, 57 p. Baker, M. S., and C. A. Wilson. 2000. Use of bomb radiocarbon to validate otolith sec- tion ages of red snapper Lutjanus campechanus from the northern Gulf of Mexico. Limnol. Oceanogr. 46(71:1819-1824. Barker, M. W. 1979. Population and fishery dynamics of recreation- ally exploited marine bottomfishes of northern Puget Sound. Ph.D. diss., 135 p. Univ. Wash., Seattle, WA. Beamish, R. J. 1979. New information on the longevity of Pacific ocean perch (Sebastes alutus). J. Fish. Res. Board Can. 36:1395-1400. 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. 1987. Current trends in age determination metho- dology. In The age and growth of fish (R. C. Sum- merfelt and G. E. Hall, eds.), p. 15-42. The Iowa State Univ. Press, Ames, IA. Cailliet, G. M., A. H. Andrews, E. J. Burton, D. L. Watters, D. E. Kline, and L. A. Ferry-Graham. 2001. Age determination and validation studies of marine fishes: do deep-dwellers live longer? Exp. Geron. 36:739-764. Campana, S. E. 1997. Use of radiocarbon from nuclear fallout as a dated marker in the otoliths of haddock Melanogrammus aegle- finus. Mar. Ecol. Prog. Ser. 150:49-56. 2001. Accuracy, precision and quality control in age deter- mination, including a review of the use and abuse of age validation methods. J. Fish Biol. 59:197-242. Campana, S. E., and C. M. Jones. 1998. Radiocarbon from nuclear testing applied to age validation of black drum, Pogonias cromis. Fish. Bull. 96:185-192. Campana, S. E., L. J. Natanson, and S. Myklevol. 2002. Bomb dating and age determination of large pelagic sharks. Can. J. Fish. Aquat. Sci. 59:450-455. Casillas, E., L. Crockett, Y. DeReynier, J. Glock, M. Helvey, B. Meyer, C. Schmitt, M. Yoklavich, A. Bailey, B. Chao, B. Johnson, and T. Pepperell. 1998. Essential fish habitat West Coast groundfish appen- dix, 778 p. NMFS, Montlake, Seattle, WA. Chang, W. Y. B. 1982. A statistical method for evaluating reproducibil- ity of age determination. Can. J. Fish. Aquat. Sci. 39:1208-1210. Chilton, D. E., and R. J. Beamish. 1982. Age determination methods for fishes studied by the groundfish program at the Pacific biological station. Can. Spec. Publ. Fish. Aquat. Sci. 60. 102 p. Druffel, E. R. M., S. Griffin, T. P. Guilderson, M. Kashgarian, J. Southon, and D. P. Schrag. 2001. Changes in subtropical North Pacific radiocarbon and correlation with climate variability. Radiocarbon 43(11:15-25 Farrell, J., and S. E. Campana. 1996. Regulation of calcium and strontium deposi- tion on the otoliths of juvenile tilapia, Oreochromis niloticus. Comp. Biochem. Physiol. 115A:103-109. Francis, R. C. 1985. Fisheries research and its application to west coast groundfish management, In Proceedings of the conference on fisheries management: issues and options (T. Frady ed.), p. 285-304. Alaska Sea Grant Report. 85-2, Univ. Alaska, Fairbanks, AK. Kalish.J. M. 1991. 13C and lsO isotopic disequilibria in fish otoliths: metabolic and kinetic effects. Mar. Ecol. Prog. Ser. 75:191-203. 1993. Pre- and post-bomb radiocarbon in fish otoliths. Earth Planet. Sci. Lett. 114:549-554. 1995. Radiocarbon and fish biology. In Recent develop- ments in fish otolith research (D. H. Secor, J. M. Dean, and S. E. Campana, eds.), p. 637-653. Univ. South Carolina Press, Columbia, SC. 2001. Use of the bomb radiocarbon chronometer to vali- date fish age. Final Report FRDC Project 93/109, 384 p. Fisheries Research and Development Corporation, Canberra, Australia. Kalish, J. M., R. Nydal, K. H. Nedreaas, G. S. Burr, and G. L. Eine. 2001. A time history of pre- and post-bomb radiocarbon in the Barents Sea derived from Arcto-Norwegian cod otoliths. Radiocarbon 43(2B):843-855. Kerr, L. A., A. H. Andrews, B. R. Frantz, K. H. Coale, T. A. Brown, and G. M. Cailliet. 2004. Radiocarbon in otoliths of yelloweye rockfish iSebastes ruberrimus): a reference time series for the waters of southeast Alaska. Can. J. Fish. Aquat. Sci. 61:443-451. Kramer, D. E., and V. M. O'Connell. 1995. Guide to northeast Pacific rockfishes: genera Sebastes and Sebastolobus. Alaska Sea Grant Advisory Bulletin 25, 78 p. Univ. Alaska, Fairbanks, AK. Leaman, B. M., and R. J. Beamish. 1984. Ecological and management implications of longev- ity in some northeast Pacific groundfishes. Int. North Pac. Fish. Comm. Bull. 42:85-97. Love, M. S., M. Yoklavich, and L. Thorsteinson. 2002. The rockfishes of the Northeast Pacific, 404 p. Univ. California Press, Berkeley, CA. Matthews, K. R. 1990. An experimental study of the habitat prefer- Kerr et al.: Age validation for Sebastes maliger with bomb radiocarbon 107 ences and movement patterns of copper, quillback, and brown rockfishes iSebastes spp.l. Environ. Biol. Fishes 29:161-178. Munk, K. M. 2001. Maximum ages of groundfishes in waters off Alaska and British Columbia and considerations of age determination. Alaska Fish. Res. Bull. 8(11:12- 21. Ostlund, H. G., and M. Stuiver. 1980. GEOSECS Pacific Radiocarbon. Radiocarbon 22(11:25-53. Richards. L. J., and A. J. Cass. 1986. The British Columbia inshore rockfish fishery: Stock assessment and fleet dynamics of an unrestricted fishery, p. 299-308. In Proceedings of the international rockfish symposium, Lowell Wakefield fisheries sympo- sium, Anchorage, Alaska. October 20-22, 1986. Alaska Sea Grant Report 87-2: Royer, T. C. 1982. Coastal fresh water discharge in the northeast Pacific. J. Geophys. Res. 87(C3):2017-2021 Schwarcz, H. P., Y. Gao, S. Campana, D. Browne, M. Knyf, and U. Brand. 1998. Stable carbon isotope variations in otoliths of Atlantic cod [Gadus morhua). Can. J. Fish. Aquat. Sci. 55:1798-1806. Stuvier, M., and H. A. Polach. 1977. Discussion: reporting of 14C data. Radiocarbon 19(3):355-363. Toggweiler, J. R., K. Dixon, and W. S. Broecker. 1991. The Peru upwelling and the ventilation of the South Pacific thermocline. J. Geophys. Res. 96:20467- 20497. Vogel, J. S., D. E. Nelson, and J. R. Southon. 1987. 14C background levels in an accelerator mass spec- trometry system. Radiocarbon 29(31:323-333. Vogel, J. S., J. R. Southon, D. E. Nelson, and T. A. Brown. 1984. Performance of catalytically condensed carbon for use in accelerator mass spectrometry. Nucl. Instr. Methods B5:289-293. Wang, C. H.. D .L. Willis, and W. D. Loveland. 1975. Radiotracer methodology in the biological, envi- ronmental, and physical sciences, 480 p. Prentice Hall, Englewood Cliffs, NJ. Williams, T, and B. C. Bedford. 1974. The use of otoliths for age determination. In The ageing offish. Proceedings of an international sympo- sium (T. B. Bagenal, ed.l, p. 114-123. Unwin Brothers Limited, Surrey, England. Yamanaka, K. L., and A. R. Kronlund. 1997. Inshore rockfish stock assessment for the west coast of Canada in 1996 and recommended yields for 1997. Can. Tech. Rep. Fish. Aquat. Sci. 2175, 80 p. Yoklavich, M. M., V. J. Loeb, M. Nishimoto, and B. Daly. 1996. Nearshore assemblages of larval rockfishes and their physical environment off central California during an extended El Nino event, 1991-1993. Fish. Bull. 94:766-782. 108 Abstract — Seasonal and cross-shelf patterns were investigated in larval fish assemblages on the continental shelf off the coast of Georgia. The influence of environmental factors on larval distributions also was exam- ined, and larval transport processes on the shelf were considered. Ichthyo- plankton and environmental data were collected approximately every other month from spring 2000 to winter 2002. Ten stations were repeatedly sampled along a 110-km cross-shelf transect, including four stations in the vicinity of Gray's Reef National Marine Sanctuary. Correspondence analysis (CA) on untransformed com- munity data identified two seasonal (warm weather [spring, summer, and fall] and winter) and three cross-shelf larval assemblages (inner-, mid-, and outer-shelf). Five environmental factors (temperature, salinity, den- sity, depth of the water column, and stratification) were related to larval cross-shelf distribution. Specifically, increased water column stratification was associated with the outer-shelf assemblage in spring, summer, and fall. The inner shelf assemblage was associated with generally lower tem- peratures and lower salinities in the spring and summer and higher salini- ties in the winter. The three cross- shelf regions indicated by the three assemblages coincided with the loca- tion of three primary water masses on the shelf. However, taxa occurring together within an assemblage were transported to different parts of the shelf; thus, transport across the con- tinental shelf off the coast of Georgia cannot be explained solely by two- dimensional physical factors. Cross-shelf and seasonal variation in larval fish assemblages on the southeast United States continental shelf off the coast of Georgia Katrin E. Marancik Department of Biology East Carolina University East Fifth Street Greenville. North Carolina 27858 Present address: Center for Coastal Fisheries and Habitat Research NOAA Beaufort Laboratory 101 Pivers Island Road Beaufort, North Carolina 28516 E mail address: Katey Marancikiffinoaa.gov Lisa M. Clough Department of Biology East Carolina University East Fifth Street Greenville, North Carolina 27858 Jonathan A. Hare Center for Coastal Fisheries and Habitat Research NOAA Beaufort Laboratory 101 Pivers Island Road Beaufort, North Carolina 28516 Manuscript submitted 20 December 2003 to the Scientific Editor's Office. Manuscript approved for publication June 25 2004 by the Scientific Editor. Fish. Bull. 103:108-129(2005). The study of larval fish assemblages provides information on community structure, spawning, and larval transport. Larval fish assemblages are groups of larvae with similar temporal and spatial distributions (Cowen et al., 1993). Larval distribu- tion patterns are initially determined by spawning time and location; larvae of species with similar spawning pat- terns are initially in the same larval assemblage (Rakocinski et al., 1996). Physical forcing and larval behavior then modify the structure of larval assemblages and ultimately deter- mine the outcome of larval transport (Cowen et al., 1993; Smith et al., 1999; Hare et al., 2001). Marine protected areas (MPAs) are portions of the marine environment designated to "provide lasting protec- tion for part or all of the natural and cultural resources therein" (Federal Register, 2000). A number of specific conservation objectives are encom- passed by this definition, such as protecting small areas with histori- cal significance or aesthetic quality, or protecting much larger areas to enhance fisheries through increases in spawning stock biomass and the supply of recruits to surrounding ar- eas (Crowder et al., 2000). However, whether an MPA provides recruits to other areas is difficult to quantify and involves determining the fate of larvae and juveniles spawned in a protected area (Stephenson, 1999; Warner et al., 2000). MPAs are under consideration as a fisheries management tool on the southeast United States continental shelf (Plan Development Team, 1990), and larval assemblage studies would provide useful information regard- ing spawning and larval transport. Although substantial larval fish re- search has been conducted on the southeast U.S. continental shelf, no studies have examined the dynamics Marancik et al .: Fish assemblages on the southeast United States continental shelf 109 of larval fish assemblages in this area. For example, during the RV Dolphin cruises, the Marine Resources Monitoring, Assessment, and Prediction (MARMAP) cruises, and the Southeast Area Monitoring and As- sessment Program (SEAMAP) cruises, ichthyoplankton surveys were conducted on the southeast United States continental shelf. From these surveys, spawning time was denned for a large group of species (Fahay, 1975), and the temporal and spatial distribution of larvae were described for a few select species (Kendall and Walford, 1979; Collins and Stender, 1987; 1989; Smith et al., 1994) and for multiple taxa, but mostly at the family level (Powles and Stender, 1976). Similarly, other programs (e.g., the South Atlantic Bight Recruitment Experiment) examined spawning and larval transport of "estuarine-dependent" species such as Atlantic men- haden (e.g., Judy and Lewis, 1983; Hoss et al., 1997; Hare et al., 1999; Checkley et al., 1999), but results for the entire suite of species sampled were not reported. For studies where the broader community of larval fish on the southeast U.S. shelf was addressed, the structure and dynamics of larval assemblages were not defined (Powell and Robbins, 1994, 1998; Govoni and Spach, 1999; Powell et al., 2000). The purpose of this study was to examine larval fish assemblages on the continental shelf off the coast of Georgia, USA. This region of the continental shelf was targeted because of 1) the nature of the broad shallow shelf, 2) the location of Gray's Reef National Marine Sanctuary 20 km from shore, and 3) the location of sev- eral proposed deepwater MPAs (70-200 m water depth) in the region. Temporal and spatial patterns in larval distributions were described to explain spawning and larval transport processes on the continental shelf off the coast of Georgia, and the implications for MPAs in the region were addressed. Materials and methods Study site The southeast United States continental shelf extends from West Palm Beach, Florida, to Cape Hatteras, North Carolina. Moving north from West Palm Beach (15 km), the shelf widens to Georgia (200 km) and then narrows to Cape Hatteras (35 km). Physical forcing by the Gulf Stream, which is part of the North Atlantic Western Boundary Current system, varies along the shelf. As the Gulf Stream flows northward along the shelf edge, it meanders, and cyclonic frontal eddies form in meander troughs (Lee et al., 1991). Meanders and frontal eddies grow in dimension from just north of the Straits of Florida (27°N latitude) to St. Augustine, Florida (30°N latitude), and then decrease from St. Augustine to just south of Charleston, South Carolina (32°N latitude). Meanders and frontal eddies grow in dimension again downstream of the Charleston Bump (32-33°N latitude), and then decrease again from Cape Fear, North Carolina (33°N latitude), to Cape Hatteras, North Carolina (36°N latitude). Table 1 Year, month, and season of ichthyoplan kton sampling and number of stations sampled in the Georgia Bight region of the southeast United States continental shelf. Year Month Season Number of stations 2000 April spring 4 2000 August summer 8 2000 October fall 7 2001 January winter 8 2001 March winter 8 2001 May spring 7 2001 June summer 7 2001 August summer 10 2001 October fall 8 2002 February winter 10 In addition to along-shelf variation in geophysical structure and Gulf Stream forcing, the southeast Unit- ed States continental shelf can be divided into three cross-shelf zones based on physical circulation dynamics (Boicourt et al., 1998). Circulation on the inner-shelf (0-20 m water depth) is influenced by tidal currents, river inflow, and wind (Atkinson and Menzel, 1985; Pi- etrafesa et al., 1985a). Wind-driven flow predominates on the mid-shelf (20-40 m water depth) and there is only minor Gulf Stream and tidal influence (Atkin- son and Menzel, 1985). Flow on the outer-shelf (40-75 m water depth) is dominated by the passage of Gulf Stream frontal eddies and upwelling at the shelf break (Pietrafesa et al„ 1985b). Inner and mid-shelf physical processes are relatively more important off the coast of Georgia compared to other segments of the southeast United States conti- nental shelf (Boicourt et al., 1998). The continental shelf off the coast of Georgia is the area of diminish- ing meanders and eddies from St. Augustine, Florida, to Charleston, South Carolina. Tidal range and fresh- water inflow is greatest in the Georgia portion of the southeast shelf (Atkinson and Menzel, 1985). Further, because the shelf is widest off the coast of Georgia (ap- proximately 200 km), the Gulf Stream is less influential on mid- and inner-shelf dynamics compared to the rest of the southeast United States continental shelf (Lee et al., 1991). Collection of larval fish and CTD data Ichthyoplankton sampling was conducted approximately every other month from April 2000 through February 2002 (Table 1). A maximum of ten stations, approxi- mately 18.5 km apart, were sampled during each cruise. Stations were missed on some cruises owing to weather and equipment failure. The transect was 110 km long and spanned 10 to 50 m water depth (Fig. 1). Four sta- 110 Fishery Bulletin 103(1) 32°0'N - 31 ON 31°0'W I 80°0'W pBrtglarWV^SUdv ares South Carolina | Gray's Reef National Marine Sanctuary Ichthyoplankton station Depth contour (m) 25 50 Kilometers - 32°0'N Georgia - - 20 m 30 m 40 m [Brunswick 2-+ 23/ • 50 m 200m • 5 7.' • 1 81°0'W 1 80°0'W - 31°0N Figure 1 Map of the study area and the cross-shelf transect used for sampling larval abundance and environmental data bimonthly from April 2000 to February 2002 (see Table 1). Four stations (stations 2.1-2.4) were located around Gray's Reef National Marine Sanctuary. tions were placed immediately adjacent to the four sides of Gray's Reef National Marine Sanctuary. At each station, temperature, salinity, density, and water depth were measured from the water's surface to one meter above the bottom with a Seabird conductivity-tempera- ture-depth (CTD probe (SBE19, Seabird Electronics, Inc., Bellevue, WA). Ichthyoplankton was collected at each sta- tion with a five-minute single oblique net tow to within one meter of the bottom. For all but one cruise (August 2000), a 61-cm paired bongo frame fitted with 333-fim or 505-/xm mesh nets was used. During the remain- ing cruise, a 1-m ichthyoplankton sled with 333-|um mesh net was used because of the smaller size of the research vessel. A flow meter (General Oceanica) was used to measure the volume of water filtered. A gear comparison study, conducted during October 2000, showed that ichthyoplankton samples collected with the two gear types (61-cm bongo versus 1-m2 ich- thyoplankton sled) were similar. An analysis of variance (ANOVA) on the mean larval concentration revealed no significant differences between the two gear types (one- way ANOVA: F=0.489; df=l; P>0.5). Also, an analysis of similarities (ANOSIM, Clarke and Warwick, 2001) determined that the community structure varied more within than between gear types (ANOSIM: i? = -0.11; S=77.57). Similarly, preliminary analysis of the effect of gear selectivity due to mesh size indicated that the larval communities collected by 333-f = l/h j (p-p)gzdz, where /) = water column depth; 7? = average water column density; p = water density; g = acceleration due to gravity; and z = depth. The stratification parameter, (jowles/m3), is a measure of the resistance of water to mixing; higher numbers signify higher resistance to mixing. Temperature and salinity data were further used to define water masses on the continental shelf off the coast of Georgia. Pietrafesa et al. (1994) defined four wa- ter masses on the southeast U.S. continental shelf: Geor- gia Bight Water, Carolina Capes Water, Virginia Coastal Water, and Gulf Stream Water. However, temperature data collected on the continental shelf off the coast of Georgia exhibited greater seasonal variability (10-29°C) than reported by Pietrafesa et al. (1994; 14-29°C). As a result, water mass definitions for our study, although based largely on the definitions of Pietrafesa et al. (1994), reflect the greater range of temperature and reflect the natural breaks in temperature, salinity, and stratification data. Specifically, two water masses (inner- shelf water and mid-shelf water) and two mixes (inner- shelf-mid-shelf mixed water and mid-shelf-Gulf Stream mixed water) were defined (Fig. 2). Inner-shelf water was characterized by salinities <35 ppt and seasonally vari- able temperatures. This water mass was found during winter and spring and was distributed inside the 20-m isobath (Fig. 3). Mid-shelf water, with salinities >36 (Fig. 2), was typically well mixed vertically (Simpson's stratification parameter value <10). Mid-shelf water was found year round over large sections of the shelf, particularly in the fall (Fig. 3). A mixture between in- ner-shelf and mid-shelf water was defined with salinities between 35 and 36 (Fig. 2). A mixture was also defined 114 Fishery Bulletin 103(1) 30.0 Georgia Bight Wat: Watermass / Inner-shelf water O Inner-shelf- mid-shelf mixed water + Mid-shelf water A Mid-shelf-Gulf Stream mixed water 33 35 Average salinity Figure 2 The average temperature and salinity for each station; symbols used represent the water mass designation for each station. The black polygons represent the temperature and salinity boundaries (data for all seasons bounded by one polygon) of three water masses defined by Pietrafesa et al. (1994; Georges Bight water; Carolina Capes water, and Gulf Stream water). Four water masses were defined in our study (inner-shelf water, inner-shelf-mid-shelf water, mid-shelf water, and mid-shelf-Gulf Stream mixed water). as mid-shelf water and Gulf Stream water (Fig. 2). Gulf Stream water was not encountered, but its temperature and salinity properties are well documented (Churchill et al., 1993; Pietrafesa et al., 1994). Mid-shelf-Gulf Stream mixed water was highly stratified (Simpson's stratification parameter value >10), with warm highly saline water intruding on the surface during fall, win- ter, and spring and cool highly saline water intruding at depth during summer. Mid-shelf-Gulf Stream mixed water was encountered on most cruises and was found farthest offshore (Fig. 3). Cruises were assigned to one of four seasons (Ta- ble 1) based on wind and temperature regimes. Al- though Blanton et al. (1985) identified five seasons for the southeast United States based on wind regimes (Spring [March-May], summer [June- July], transition [August], autumn [September-October], and winter [November-February]), the temperature data collected in our study supported classifying both August cruises as summer and the March cruise as winter. Data analyses Multivariate analyses were used to define larval assem- blages and to explore the factors that influence distri- bution of larval assemblages on the continental shelf off the coast of Georgia. Multivariate analyses arrange sites and species along environmental gradients creating a low dimensional map (an ordination). Analyses can be conducted for samples where the distance between points in the ordination represents the similarity of species abundance between samples. Analyses also can be conducted for species where the distance between points in the ordination represents the similarity in the sample distribution between species. Ordinations, then, can be analyzed in two ways: with regard to proximity and dimensionality. Points that occur in close proximity can be considered similar based on similar composition. Points that occur on the same dimension define gradients in the data. The effects of data transformation (untransformed, square root transformed, and fourth root transformed) and species inclusions (1% and 10% data sets) on the ordination of community and environmental data by two multivariate ordination techniques, multidimen- sional scaling and correspondence analysis (CA), were compared to determine which method was more effec- tive at analyzing the larval fish data collected on the continental shelf off the coast of Georgia (Marancik, 2003). Overall, the two analytical methods produced similar ordinations and were robust to the inclusion of rare species and to the type of data transformation. Correspondence analysis on untransformed larval fish concentration data was used to define larval as- semblages in relation to season and the entire two-year data set. One of the strengths of CA is that it allows one to plot analyses of species and station data simul- taneously on one ordination, thereby, allowing immedi- ate comparisons between those stations that occur in close proximity in ordination space and those taxa that influence that proximity. Eigenvalues are a measure of the importance of each CA dimension (ter Braak and Smilauer, 2002). Thus, the dimensions needed to describe patterns in the data can be determined by an abrupt drop in the magnitude of eigenvalues from one dimension to the next. Marancik et al.: Fish assemblages on the southeast United States continental shelf 115 B . Summer :iM--J. X »m ■ • • c c °o o o o o ^^^o, . c . Fall :> •'M, "JOO, D Winter 20- *" ■ • • 0 o o u o ° ^ ***>■** Water masses Salinity Stratification # Inner-shelf water <35 # Inner-shelf-mid-shelf mixed water 35-36 O Mid-shelf water >36 O Mid-shelf-Gulf Stream mixed water r~l No data collected <10 >10 Figure 3 Water mass designations for each station for each cruise. Cruises within a season were put together in one map with transects offset from center: (Al spring, (B) summer, (C) fall, and (D) winter. Inner-shelf water was the least saline and found farthest inshore. Mid-shelf-Gulf Stream mixed water was a highly stratified mix of Gulf Stream water and mid-shelf water and was found farthest offshore. Canonical correspondence analysis (CCA), which in- corporates environmental variables by aligning species and station data along environmental gradients, was used to explore the relationship between larval assem- blages and the environment. The species-environment correlation is a measure of the strength of the rela- tion between the species data and the environmental data for each CCA dimension (ter Braak and Smilauer, 2002). The product of the species-environment correla- tion and the eigenvalue can be used to describe the variance in the data. CA and CCA were performed by using the statistical package CANOCO (Ter Braak, 1988). Multivariate analyses were used to determine which fish species spawn on the continental shelf off the coast of Georgia, to examine what environmental factors in- fluence larval distribution, and to explore the physical factors affecting the transport of larvae spawned on the shelf. Specifically, six objectives were addressed: 1) cross-shelf patterns in the larval fish community; 2) larval assemblages associated with cross-shelf patterns in the larval fish community; 3) the relation among cross-shelf patterns in the larval fish community, larval assemblages, and environmental variables; 4) the rela- tion between water mass and larval assemblages; 5) seasonal patterns in the larval fish community and lar- val assemblages; and 6) the relation between seasonal larval assemblages and environmental variables. In addition to addressing the six specific objectives, the implications for larval transport were considered. By comparing the distributions of specific taxa to the patterns discerned by addressing the objectives above, some insights were gained into larval transport pro- cesses. The distribution of taxa representative of each larval assemblage was examined for patterns through space and time. Mechanisms driving larval transport were then explored by linking these patterns to water mass and other environmental variables. 116 Fishery Bulletin 103(1 Results Two dimensions were sufficient to explain the majority of the variance in the larval concentration data (Table 4). The winter data eigenvalues indicated the relevance of a third dimension; yet, inspection of three dimensions did not define any patterns not indicated by the first two dimensions. Thus, two dimensions were analyzed for each season in both the CA and CCA analyses. Cross-shelf patterns in the larval fish community A cross-shelf pattern in the larval community was observed. In spring, summer, and fall, the inshore sta- tions (stations 1-3) were in close proximity, forming an inner-shelf station group in the ordination resulting from the CA (Fig. 4). Along the same dimension (axis) as the inner-shelf group was a mid-shelf station group of stations 3-6 (stations 2.1-2.4 were also included in this group in spring, summer, and winter). An outer-shelf group composed of offshore stations (stations 5-7) was distributed along a nearly perpendicular dimension, and the mid-shelf group was at the intersection of the two dimensions (Fig. 4). Analysis of the one-percent species data set revealed an identical pattern for each season (not shown). The winter station ordination resulted in a less dis- tinct cross-shelf pattern (Fig. 4D). In January 2001, stations 1, 2, 3, and 6 were in the inner-shelf group; whereas, stations 4 and 7 from the same cruise were in the mid-shelf group, and station 5 was in the outer-shelf group. Some of this blurring of the cross-shelf pattern in the ordination may be explained by a lower total catch, giving the taxa found across the shelf {Brevoor- tia tyrannus and Leiostomus xanthurus) more influence over the data. In addition, most of the variance was explained by the first dimension (Table 4), meaning that the separation of the outer-shelf group (stations 5 and 6) from the mid- and inner-shelf groups is based on a weak relationship among the stations. Larval assemblages associated with cross-shelf patterns in the larval fish community Three larval assemblages were defined that corre- sponded to the three station groups (Fig. 5). The inner- shelf assemblage was composed of species that spawn in coastal and estuarine habitats. Larvae in this assem- blage were distributed within the 20-m isobath and con- fined largely to stations classified as inner-shelf (Fig. 6). The inner-shelf assemblage was primarily represented by Menticirrhus americanus during spring, summer, and fall, and by Micropogonius undulatus and Lagodon rhomboides during winter (Table 5). Taxa included in the mid-shelf assemblage were generally found between the 20- and 40-m isobaths. Some mid-shelf taxa, how- ever, were found across the shelf (stations 1-7) and a large percentage of the larvae occurring in each region were mid-shelf taxa (Fig. 6). The outer-shelf assemblage comprised offshore or deepwater spawned taxa and was CA1 Figure 4 Correspondence analysis ordinations (portraying the first and second dimension scores) of the larval fish community data showing station groups in each season (A) spring, (B) summer, (C) fall, and (D) winter. Three cross-shelf sta- tion groups were identified within each season. Solid lines enclose the boundary of each station group with three or more stations. Station groups comprising one or two stations are not enclosed by a solid line. Each station group is labeled and portrayed with a different symbol. The dashed lines intersect at the origin of the plot. Analyses were conducted with larval concentration data only. Data from each cruise within a season are shown together. Marancik et al.: Fish assemblages on the southeast United States continental shelf 117 Table 4 Eigenvalues and species-environment correlations (r2l for each axis analyzed (correspondence analysis [CA] and canonical cor- respondence analysis [CAA]) by season and the entire year. A sharp drop in the eigenvalue marks the axes that explain most of the data. Species and environment correlations represent the strength of the relation between the species data and the envi- ronmental data for each axis within each season. Values of zero denote no relation; values of one denote a perfect relation. The product of the species-environment correlation and the eigenvalue explains the variance in the data for CCA. Eigenvalues alone explain the variance in the data for CA. Season CA axis CCA axis Spring Eigenvalue r2 Summer Eigenvalue r2 Fall Eigenvalue r2 Winter Eigenvalue r2 Year Eigenvalue 0.932 0.674 0.348 0.792 0.621 0.738 0.544 0.537 0.273 0.526 0.937 0.287 0.197 0.788 0.607 0.107 0.292 0.106 0.165 0.54 0.89 0.631 0.329 0.068 0.98 0.969 0.969 0.796 0.703 0.564 0.409 0.159 0.959 0.959 0.889 0.799 0.707 0.443 0.228 0.053 0.983 0.909 0.935 0.946 0.42 0.104 0.059 0.041 0.894 0.665 0.645 0.496 0.773 0.61 0.319 0.276 0.923 0.899 0.8 0.735 Table 5 Three cross-shelf larval assemblages (inner-shelf, mid-shelf, and outer-shelf) were persistent in the Georgia Bight with sea- sonal changes in membership. Shown are the assemblages from the ten-percent data set. "Bothus ocellatus 1 robinsi" means B. ocellatus and B. robinsi or one of either of them. Season Inner Mid Outer Spring Menticirrhus americanus Diplogrammus pauciradiatus Auxis rochei Otophidium omostigmum Opisthonema oglinum Bothus ocellatus 1 robinsi Xyrichthys spp. Micropogonias undulatus Etropus crossotus Anchoa hepsetus Summer M. americanus D. pauciradiatus A. rochei O. oglinum O. omostigmum Ophidion marginatum Xyrichthys spp. E. crossotus M. undulatus A. hepsetus B. ocellatus 1 robinsi Fall M. americanus D. pauciradiatus Xyrichthys spp. A. hepsetus M. undulatus B. ocellatus 1 robinsi O. marginatum E. crossotus Leiostomus xan hurus O. omostigmum Winter M. undulatus L. rhomboides B. tyrannus M. punctatus C. spilopterus D. pauciradiatus O. omostigmum L. xanthurus B. ocellatus 1 robinsi 118 Fishery Bulletin 103(1) Oomo DPau Outjer Oog\ Aroc' . 4hep~ , Mame Inner "D 3 CO Oomi Omi' Mid /Wura B 10 1 \ // 0 i i i i o i i i i Inner Mid Outer Inner Mid Outer Station group Figure 7 The number of taxa collected in each station group during each season for the (A) ten-percent and (B) one-percent data sets. pattern in the larval community was revealed. Physical data delineated four water masses (Fig. 3). Larval fish assemblages differentiated only three of these water masses. Stations associated with inner-shelf water (the inshoremost water mass) and mid-shelf-Gulf Stream mixed water (the offshoremost water mass) formed dis- tinct groups in the ordination of larval community data (Fig. 9). Stations associated with mid-shelf water also 120 Fishery Bulletin 103(1) STRAT DEP„-_SALGRA£ . AYGQER . . . AVGSAL DENGRA_ AVGTEM o Outer "O 3 CO AVGTEM c 3 3 YOuter c STOAT te^grad A inner !/ M / AVGDE.N AVGJE^^ ^0 CCA 1 Figure 8 Canonical correspondence analysis (CCA) ordinations (portray- ing the first and second dimension scores) of the larval fish community data showing the correlations between environ- mental variables, species, and station groups: (A) spring. (B) summer, (C) fall, and (D) winter. The solid triangles mark the location of taxa (as in Fig. 5), and the polygons surround the three cross-shelf station groups (as in Fig. 4). The arrows depict the gradient of each environmental variable. The dashed lines intersect at the origin of the plot. Analyses were conducted with both larval and environmental data. Refer to Table 3 for definitions of environmental variable codes. Table 6 The P values from a Monte Carlo permutation test on the environmental variables for each season. Significant values (P<0.05) are shown in bold font. See Table 3 for definitions of variable codes. Variable code Season Spring Summer Fall Winter AVGDEN 0.002 0.01 0.34 0.494 AVGSAL 0.002 0.022 0.016 0.004 AVGTEM 0.152 0.1 0.04 0.016 DENGRAD 0.836 0.076 0.466 0.958 SALGRAD 0.456 0.086 0.78 0.634 TEMGRAD 0.074 0.076 0.38 0.574 DEP 0.468 0.002 0.002 0.68 STRAT 0.036 0.014 0.012 0.504 formed distinct groups. The fourth water mass, inner- shelf-mid-shelf mixed water overlapped with either inner-shelf or mid-shelf water depending on season. In summary, the cross-shelf distribution and assemblages of water masses coincided with the three cross-shelf regions described: inner-shelf, mid-shelf, and outer-shelf characterized by inner-shelf water, mid-shelf water, and mid-shelf-Gulf Stream mixed water, respectively. Seasonal patterns in the cross-shelf distributions of the larval fish community The ten percent data set revealed two distinct seasonal station groups (Fig. 10). The winter stations occurred in close proximity and were separate from stations sampled during the rest of the seasons (Fig. 10A). However, inner- shelf stations sampled during fall overlapped with the winter stations because of the presence of winter and fall spawning species (L. xanthurus and M. undulatus). There was also overlap of the winter and the warm weather outer-shelf stations (Fig. 10, A and B). Similarly, the ten percent data set revealed two seasonal assemblages in the larval community data (Fig. 10, C and D). The warm weather assemblage com- prised taxa associated with the warm weather station group and were collected during spring, summer, and fall. The winter assemblage was associated with the winter station group and comprised taxa collected dur- ing winter. Taxa from the warm weather inner- and mid-shelf assemblages were different from those rep- resenting the winter inner- and mid-shelf assemblages (Table 5). The outer-shelf assemblage, however, was less seasonally distinct, represented by Bothus ocellatus/rob- insi in summer, fall, and winter and by Auxis rochei in spring, summer, and fall (Table 5). Marancik et al .: Fish assemblages on the southeast United States continental shelf 121 Relation between seasonal larval assemblages and environmental variables The seasonal pattern in the larval concentration data described above was maintained when constrained by environmental variables in the CCA. The community data clearly showed a seasonal influence on the first dimension in ordination space; winter taxa were sepa- rate from taxa collected during the rest of the seasons. This seasonal pattern was also reflected in the environ- mental data (Fig. 11). Salinity, density, temperature, depth, and stratification of the water column were again the most significant environmental variables for explain- ing variance in the species data (P<0.05, Monte Carlo permutation test, Table 6). The warm weather stations and taxa coincided with higher water temperature, lower density, and a lower density gradient. In addition, the cross-shelf pattern evident in the second and third dimensions of the full larval concentration data (Fig. 10, A and B) appeared to correlate with depth of the water column, the degree of stratification in the water column, and salinity (Fig. 11). Implications for larval transport The structure of larval assemblages was linked to water mass distributions and the cross-shelf zonation of physi- cal circulation processes. Three cross-shelf zones of physical dynamics have been defined previously (Atkin- son and Menzel, 1985; Pietrafesa et al., 1985a, 1985b; Lee et al., 1991; Boicourt et al., 1998). Three analogous cross-shelf zones were delineated in the larval com- munity data. The cross-shelf larval assemblages were linked to three water masses with cross-shelf structure, and to the physical-chemical characteristics of the region (temperature, salinity, density, and stratification of the water column). The three cross-shelf zones identified pre- viously in terms of physical dynamics coincided with the station groups and larval assemblages identified in our study. Thus, larval distribution and physical properties of the ocean are linked and indicate a strong influence of physical properties and processes on the distribution of larval fish on the southeast United States continental shelf. Retention on the inner-shelf was a clear larval trans- port pattern identified in the analyses. Menticirrhus americanus represents the inner-shelf group (Table 5) and were always found inshore of the 20-m isobath in inner-shelf water, in inner-shelf-mid-shelf mixed water, or in mid-shelf water, (Fig. 12). Spawning likely occurs on the inner-shelf (Cowan and Shaw, 1988), and larvae are retained in the inner-shelf region. The analyses also demonstrated that transport from offshore onto the shelf is limited on the continental shelf off the coast of Georgia. Ceratoscopelus maderensis and Auxis rochei were found only at offshore stations (Fig. 13), representing the outer-shelf group (Table 5) and the mid-shelf-Gulf Stream mixed water mass. The presence of C. maderensis identified transport of a me- sopelagic fish to waters inshore of the shelf break; how- _M3GS MSGS ISMS D ISMS '■1 -■■'•.- CA1 Figure 9 Correspondence analysis (CA) ordinations (portraying the first and second dimension scores) of the larval fish community data showing the full ten-percent data set: (A) spring, iBi summer, (C) fall, and (D) winter. The points represent stations classified by water mass. Solid lines enclose the boundary of each station group with three or more stations. Station groups comprising one or two stations are not enclosed by a solid line. Each station group is labeled and portrayed with a different symbol. Stations with inner-shelf water are labeled with IS (inner-shelf), inner-shelf-mid-shelf mixed water with ISMS, mid-shelf water with MS, and mid-shelf-Gulf Stream mixed water with MSGS. The dashed lines intersect at the origin of the plot. Analyses were conducted using larval data only. 122 Fishery Bulletin 103(1) CM < O Warm • Winter < O Inner □ Mid ir Outer I) St -Mpvn^Lxari Lrtia Winter CA 1 CA 1 Figure 10 Correspondence analysis (CA) ordinations of the larval fish community data showing (A) the first and second dimension scores and (B) the first and third dimension scores of the station groups (inner, mid, and outer) defined within each season when the 10% data set was used. Open symbols denote stations sampled during the warm weather season and filled symbols denote stations sampled during the winter season. (C) The first and second dimensions and (D) the first and third dimensions of the station and species groups in the full data set are shown without the incorporation of the environmental data. The dashed lines intersect at the origin of the plot. ever, the rarity of this species on the continental shelf off the coast of Georgia provides evidence for relatively limited onshore transport from off the shelf. Powell and Robins (1994, 1998) and Govoni and Spach (1999) also collected tropical and deepwater taxa inshore of the shelf break. The presence of these taxa was likely due to frequent but variable exchange of larvae across the Gulf Stream front (Govoni and Spach, 1999). Less is known about spawning of A. rochei but the species' lar- val distribution represents restriction to offshore waters (always collected offshore of the 40-m isobath). During winter, when B. tyrannus was found across the shelf (Fig. 14), Bothus ocellatus /robinsi was col- lected only on the outer part of the shelf (Fig. 14). Both B. tyrannus and B. ocellatus /robinsi likely spawn on the outer shelf. However, unlike B. tyrannus, Bothus ocel- latus /robinsi was never collected inshore of station 3 (the boundary between the inner- and mid-shelf zones), indicating that the two taxa may experience different transport pathways or different seasonal spawning pat- terns (see "Discussion" section). Discussion Three cross-shelf regions were defined on the continental shelf off the coast of Georgia based on the distribution and abundance of larval fish: inner-shelf, mid-shelf, and outer-shelf. Each region was dominated by a distinct group of species (i.e., larval assemblage). The inner-shelf Marancik et al.: Fish assemblages on the southeast United States continental shelf 123 A l Summer B DEP \ AVG&AL ^V\ 'STRAT DENGBA^ALGRAlf \ * AV A/ V \a \ /^ \\l\i TEMGRAD \WinteK FairV Spring i Figure 11 The correlation between environmental variables and station groups portrayed by canoni- cal correspondence analysis (Fig. 10). (A) The proximity of seasonal station groups (black polygons) and taxa (black triangles) when environmental and larval concentration data were analyzed. (B) The relationship between the environmental variables (black arrows) and the seasonal station groups (gray polygons). The direction of the arrows depicts the gradient of each environmental variable. The dashed lines intersect at the origin of the plot. region was defined inshore of the 20-m isobath (Figs. 4, 5, 12). The inner-shelf larval assemblage was the least diverse taxonomically (Table 2, Fig. 7B), and most taxa in the assemblage were nearshore or estuarine spawning species (e.g., Cynoscion regalis, Menticirrhus americanus. Table 2). Gradients in salinity and density were associ- ated with the separation of the inner-shelf region but the direction of the gradient varied among seasons; in the spring and summer the inner-shelf region was char- acterized by lower salinity and density, whereas in the fall and winter, the inner-shelf region was characterized by higher salinities and densities (Fig. 8). The restricted inshore distribution of the assemblage indicated mecha- nisms of larval retention in the inner-shelf zone. The mid-shelf region was defined between the 20- and 40-m isobaths (Figs. 4, 5, 12). The mid-shelf larval as- semblage was distributed over the widest area (Figs. 4, 5, 12) and species in the assemblage were found in all three regions defined (Fig. 6). The mid-shelf region and larval assemblage were related to the average environ- mental parameters encountered on the shelf (Fig. 8), which varied seasonally. The broad distribution of the assemblage indicated either broad spawning distribu- tions of member species or mechanism of larval trans- port to both the inner- and outer-shelf regions. The outer-shelf region was defined as the area off- shore from the 40-m isobath (Figs. 4, 5, 12). The outer- shelf region was related to increased stratification of the water column, which was likely a result of Gulf Stream waters mixing onshore. These periodic intru- sions would help explain the higher species richness of rare taxa found on the outer-shelf during fall and win- ter (Fig. 7B). Taxa in the outer-shelf assemblage were either spawned on the outer-shelf (e.g., Hemanthias vivanus), spawned offshore of the shelf break and trans- ported onto the shelf (e.g., Ceratoscopelus maderensis), or spawned south of the study area and transported onto the shelf (e.g., Abudefduf sp.). Most outer-shelf taxa, however, were restricted to outer-shelf stations indicating limited onshore exchange between the outer- and mid-shelf regions. Larval assemblages on the continental shelf off the coast of Georgia are derived from a combination of spawning distributions and larval transport; Brevoor- tia tyrannus and Bothus ocellatus I robinsi provide an example. Brevoortia tyrannus spawn in water tempera- tures between 16° and 23°C during winter (Checkley et al. 1999); these temperatures were experienced in the mid- and outer-shelf regions during winter. Bothus ocel- latus/robinsi adults also occur on the mid- and outer- shelf of the continental shelf off the coast of Georgia (Gutherz, 1967). Thus, during winter the spawning distribution of these two species are likely similar. The larval distributions, however, are different: B. tyrannus larvae were collected in all three regions of the shelf during winter, whereas B. ocellatus /robinsi were col- lected on the mid- and outer-shelf (Fig. 14). The verti- cal distributions of the two species also are different. B. tyrannus larvae occur higher in the water column than do B. ocellatus /robsini (Hare and Govoni1). The observed differences in horizontal distribution could result from the differences in vertical distributions. Alternatively, the distributional differences could result from physiological differences that allow B. tyrannus larvae to survive cooler inshore waters or could result from seasonal cross-shelf spawning patterns that result 1 Hare, J. A., and J. J. Govoni. 2004. In review. Vertical distribution and the outcome of larval fish transport along the southeast US continental shelf during winter. 124 Fishery Bulletin 103(1) B Summer o +**< two ■ o> v«> - c ■ • Fall oo •e ' ""--•.■,.; a ^^ " -=- D Winter "z», '«*.* .<-. SJO; Water mass • Inner-shelf water • Inner-shelf-mid-shelf mixed water o Mid-shelf water o Mid-shelf-Gulf Stream mixed water □ No water mass data Fish abundance (larvae/100 m3) 0 • 0.001-1 £ 1001-10 ft 10.001-100 100.001-1000 Figure 12 Distribution of Menticirrhus americanus in (A) spring, (B) summer, (C) fall, and (D) winter. Transects for each cruise within a season are offset from one another. The size of the circle for each station varies with larval fish concentra- tion (larvae/100 m3). The fill color for each circle varies with water mass. in B. tyrannus spawning inshore during the fall. This example demonstrates that there are multiple mecha- nisms or pathways that affect the transport of larval fish, and that each species may be subject to different transport regimes. Therefore, to understand larval transport, many factors, including physical forcing mechanisms, the horizontal and vertical distributions of larvae, seasonal patterns, and the physiology of a species, need to be considered. Temporal larval assemblages were defined in addi- tion to the spatial assemblages. Larvae clearly sepa- rated into two seasonal spawning groups: winter and warm seasons (Fig. 10). The winter assemblage was associated with cool, denser water, whereas the warm water assemblage was associated with warmer, less dense water (Fig. 11). The cross-shelf structure in lar- val assemblages was still evident in the two seasonal assemblages, but there was overlap in the winter and warm-weather outer-shelf assemblages (Fig. 10). This overlap occurred in waters with the least seasonal vari- ability in temperature and salinity and likely results from year-round spawning by species in the outer-shelf assemblage or year-round supply of larvae to the outer- shelf region by the Gulf Stream. Marancik et al.: Fish assemblages on the southeast United States continental shelf 125 Auxis rochei KJ Spring i - 3B« . 0 *»** on. .,, " B Summer □ •o □ **-, ■ o Jooo 'Vl 3*)j Ceratoscopelus maderensis u? Water mass • Inner-shelf water • Inner-shelf-mid-shelf mixed water o Mid-shelf water o Mid-shelf-Gulf Stream mixed water □ No water mass data Fish abundance (larvae/100 m3) 0 • 0.001-1 £ 1.001-10 ft 10.001-100 100.001-1000 Figure 13 Distribution of Auxis rochei in (A) spring, (B) summer, and distribution of Cera- toscopelus maderensis in (C) spring ID) winter, across the shelf and across water masses. Transects for each cruise within a season are offset from one another. The size of the circle for each station varies with fish concentration (larvae/100 m3). The fill color for each circle varies with water mass. Winter-spawning species that use estuaries are fre- quently grouped together as "estuarine-dependent" taxa (sensu Warlen and Burke, 1990). However, Hare and Govoni1 found that vertical distributions of these winter taxa are different. In addition, our study demonstrated that the horizontal distributions of these species are distinct: Lagadon rhomboides and Micropogonias un- dulatus were members of the inner-shelf assemblage and Leiostomus xanthurus, Myrophis punctatus, and Brevoortia tyrannus were members of the mid-shelf assemblage. These findings imply that often grouped "estuarine-dependent" species have different spawning locations or experience different larval transport pro- cesses (or both) and may not reflect a single group. The definition of three regions based on larval fish distributions is consistent with the division of the shelf into three cross-shelf zones based on physical dynamics. The inner-shelf (0-20 m) is dominated by freshwater discharge, tides, and winds; the mid-shelf (20-40 m) is influenced by wind and tides; and the outer-shelf (40-75 m) is affected by the Gulf Stream and wind (At- kinson and Menzel, 1985; Pietrafesa et al., 1985a, 1985b; Lee et al., 1991; Boicourt et al., 1998). Thus, the physical dynamics of the shelf appear to be closely linked to spa- 126 Fishery Bulletin 103(1) Bothus ocellatus/robinsi B Summer o o o °00 'o. *»«. •">« '■Wo, ***«, 2P07 Brevoortia tyrannus D Winter Vn »■ '■"■"•ad • • O ( ) ° e ^a* Water mass • Inner-shelf water • Inner-shelf-mid-shelf mixed water o Mid-shelf water o Mid-shelf-Gulf Stream mixed water D No water mass data Fish abundance (larvae/100 m^) 0 • 0.001-1 £ 1.001-10 ft 10.001-100 100.001-1000 Figure 14 Distribution of Bothus ocellatus/robinsi in (A) spring, (B) summer, (C) fall, and ID) winter, and Brevoortia tyrannus (E) in winter, across the shelf and across water masses. Transects for each cruise within a season are offset from one another. The size of the circle for each station varies with fish concentration (larvae/100 m3). The shading for each circle varies with water mass. tial patterns in the distribution of larval fish. Further physiochemical characteristics of the environment (e.g., temperature, salinity, water masses) are highly associ- ated with the structure of larval assemblages (Tables 4, 6, Fig. 9), again indicating a strong link between physi- cal dynamics and larval distribution. However, patterns in spawning and behaviorally modified vertical distribu- tions also have an influence on larval distributions and thus a simple two-dimensional passive model will not adequately explain the distribution of larval fish on the continental shelf off the coast of Georgia. The three regions defined in our study have impor- tant implications for the consideration of MPAs on the southeast United States shelf. The described cross-shelf zones (inner-, mid-, or outer-shelf) provide information needed to protect spawning habitat of specific species (e.g., Rhomboplites aurorubens spawns on the outer- shelf; Table 2). Conversely, the species included in an area under consideration for protection can also be derived (e.g., Gray's Reef National Marine Sanctuary potentially protects species spawning at the interface between the inner- and mid-shelf. Table 2). Further, spawning location information can be derived for sev- eral species protected under the South Atlantic Fish- eries Management Council's coastal migratory pelag- ics management plan (e.g., Rachycentron canadum, Scomberomorus cavalla, Scomberomorus maculatus, or Coryphaena hippurus. Table 2), but individuals of Marancik et al.: Fish assemblages on the southeast United States continental shelf 127 these species range so widely (Sutter et al., 1991), only very large MPAs would afford protection from fishing (Parrish 1999, Beck and Odaya 2001). Unfortunately, many species in the snapper-grouper complex, a more sedentary group of species of particular importance in the southeast United States, were not collected. Either these taxa do not spawn on the continental shelf off the coast of Georgia and their larvae are rarely transported into the area, or snapper-grouper spawning on the con- tinental shelf off the coast of Georgia is at a very low level and larvae are quite rare. Another aspect of MPAs designed for fisheries man- agement is production of individuals in the MPA and their supply to surrounding areas; larval transport is a major mechanism of supply. On the continental shelf off the coast of Georgia, larval assemblages suggest that the supply of larvae from the south (by the Gulf Stream) and even between cross-shelf zones is limited. Members of the outer-shelf assemblage rarely occurred on the mid- and inner-shelf, and members of the inner- shelf assemblage rarely occurred on the mid- and outer- shelf. Thus, larvae spawned on the inner-shelf and to a lesser degree on the mid-shelf likely remain on the continental shelf off the coast of Georgia and appear to be subject to local retention. MPAs in the region, there- fore, could provide a local benefit by supplying recruits to nonprotected areas on the continental shelf off the coast of Georgia. Acknowledgments We would like to thank all who helped with sample collections, sorting, and analyses: G. Bohne, R. Bohne, C. Bonn, J. Burke, M. Burton, B. Degan, M. Duncan, J. Govoni, M. Greene, E. Jugovich, S. Lem, J. Loefer, R. Mays, R. McNatt, A. Powell, R. Rogers, S. Shoffler, S. Varnam, H. Walsh, and T. Zimanski. We appreciate the hard work and dedication of the officers and crew of the NOAA Ship Ferrel, NOAA Ship Jane Yarn, NOAA Ship Oregon II, and RV Cape Fear. Frank Hernandez provided invaluable help with the CTD processing and stratification calculations. We would also like to thank J. Johnson, S. Norton, A. Powell, F. Hernandez, E. Wil- liams, P. Marraro, W. Richards, and an anonymous reviewer for their comments on previous drafts. Most of all, we thank Gray's Reef National Marine Sanctuary and the National Marine Sanctuary Office for funding the project. Literature cited Atkinson, L. P., and D. W. Menzel. 1995. Introduction: Oceanography of the southeast United States continental shelf. In Oceanography of the south- eastern U.S. continental shelf (L. P. Atkinson, D. W. Menzel, and K. A. Bush, eds.), p 1-9. Am Geophysical Union, Washington, D. C. Beck, M. W., and M. Odaya. 2001. Ecoregional planning in marine environments: identifying priority sites for conservation in the northern Gulf of Mexico. Aquat. Conserv.: 11:235-242 Blanton, J. O., F. B. Schwing, A. H. Weber, L. J. Pietrafesa, and D. W. Hayes. 1985. Wind stress climatology in the south Atlantic Bight. In Oceanography of the Southeastern U.S. con- tinental shelf (L. P. Atkinson, D. W. Menzel, and K. A. Bush, eds.). p. 10-22. American Geophysical Union, Washington, D.C. Boicourt, W. C, W. J. Wiseman Jr., A. Valle-Levinson, and L. P. Atkinson. 1998. Chapter 6. Continental shelf of the southeastern United States and the Gulf of Mexico: in the shadow of the western boundary current. In The sea, vol. 11 (A. R. Robinson and K. H. Brink, eds.), p. 135-182. John Wiley and Sons, Inc., New York, NY. Checkley. Jr.. D. M., P. B. Ortner, F. E. Werner, L. R. Settle, and S. R. Gumming. 1999. Spawning habitat of the Atlantic menhaden in On- slow Bay, North Carolina. Fish. Oceanogr. 8:22-36. Churchill, J. H., E. R. Levine, D. N. Connors, and P. C. Cornillon. 1993. Mixing of shelf, slope and Gulf Stream water over the continental slope of the Middle Atlantic Bight. Deep- Sea Res. 40:1063-1085. Clarke, K. R., and R. M. Warwick. 2001. Change in marine communities: an approach to statistical analysis and interpretation. PRIMER-E, 2nd ed., 144 p. Plymouth Marine Laboratory, Plymouth, Cornwall, UK. Collins, M. R., and B. W. Stender. 1987. Larval king mackerel iScomberomorus cavalla), Spanish mackerel (S. maculates), and bluefish (Poma- tomus saltatrix) off the southeast coast of the United States, 1973-1980. Bull. Mar. Sci. 41:822-834. 1987. Larval striped mullet {Mugil eephalus) and white mullet (Mugil curema) off the southeastern United States. Bull. Mar. Sci. 45:580-589. Cowan, J. H., and R. F Shaw. 1988. The distribution, abundance, and transport of larval sciaenids collected during winter and early spring from the continental shelf waters off west Louisiana. Fish. Bull. 86:129-142. Cowen, R. K. , J. A. Hare, and M. P. Fahay. 1993. Beyond hydrography: can physical processes explain larval fish assemblages within the Middle Atlantic Bight? Bull. Mar. Sci. 53:567-587. Crowder, L. B., S. J. Lyman, W. F Figueira, and J. Priddy. 2000. Source-sink population dynamics and the problem of siting marine reserves. Bull. Mar. Sci. 66:799-820. Fahay, M. P. 1975. An annotated list of larval and juvenile fishes captured with surface-towed meter net in the South Atlantic Bight during four RV Dolphin cruises between May 1967 and February 1968. NOAA/NMFS Technical Report SSRF 685, 39 p 1983. Guide to the early stages of marine fishes occur- ring in the western North Atlantic Ocean, Cape Hat- teras to the southern Scotian Shelf. Northwest Atl. Fish. Sci. 4:1-423. Federal Register. 2000. Presidential documents. Executive Order 13158 of 128 Fishery Bulletin 103(1) May 26, 2000. Volume 65, number 105, May 31 2000. U.S. Government Printing Office, Washington, D.C. Govoni, J. J., and H. L. Spach. 1999. Exchange and flux of larval fishes across the west- ern Gulf Stream front south of Cape Hatteras, USA, in winter. Fish. Oceanogr. 8(suppl. 2):77-92. Gutherz, E. J. 1967. Field guide to the flatfishes of the family Bothidae in the Western North Atlantic, 47 p. United States Department of the Interior, Washington, D.C. Hare, J. A., J. A. Quinlan, F. E. Werner, B. O. Blanton, J. J. Govoni, R. B. Forward, L. R. Settle, and D. E. Hoss. 1999. Larval transport during winter in the SABRE study area: results of a coupled vertical larval behavior- three-dimensional circulation model. Fish. Oceanogr. 8:57-76. Hare, J. A., M. P. Fahay, and R. K. Cowen. 2001. Springtime ichthyoplankton of the Slope Sea: larval assemblages, relation to hydrography and implications for larval transport. Fish. Oceanogr. 10:164-192. Hoss, D. E., H. L. Spach, L. R. Settle, J. A. Hare, and E. H. Laban. 1997. The growth and behaviour of two species of clupeid larvae and how it affects their oceanic transport. In Ichthyoplankton ecology (A. J. Geffen, J. M. Fives, J. E. Thorpe, eds.), 22 p. Fisheries Society of the British Isles, Galway, Ireland. Johnson, G. D., and P., Keener. 1984. Aid to the identification of American grouper larvae. Bull. Mar. Sci. 34:106-134. Judy, M. H., and R. M. Lewis. 1983. Distribution of eggs and larvae of Atlantic menha- den, Brevoortia tyrannus along the Atlantic coast of the United States. U. S. National Marine Fisheries Service, Special Scientific Report— Fisheries 774, 23 p. Kendall, A. W., Jr., and A. C. Materese. 1994. Status of early life history descriptions of marine teleosts. Fish. Bull. 92:725-36. Kendall, A. W., Jr. and L. A. Walford. 1979. Sources and distribution of bluefish, Pomatomus saltatrix, larvae and juveniles off the east coast of the United States. Fish. Bull. 77:213-27. Lee, T. N., J. A. Yoder, and L. P. Atkinson. 1991. Gulf Stream frontal eddy influence on productiv- ity of the southeast U.S. continental shelf. J. Geophy. Res. 96:22191-2205. Leis, J. M. 1989. Larval biology of butterflyfishes (Pisces, Chae- todontidae): what do we really know? Environ. Biol. Fishes 25:87-00. Marancik, K. E. 2003. Larval fish assemblages of the Georgia Bight. M.S. thesis, 149 p. East Carolina Univ., Greenville. NC. Parrish, R. 1999. Marine reserves for fisheries management: why not. CalCOFI Rep. 40:77-6 Pietrafesa, L. J., J. O. Blanton J. D. Wang, V. Kourafalou, T. L. Lee, and K. A. Bush. 1985a. The tidal regime in the South Atlantic Bight. In Oceanography of the southeastern U.S. continental shelf (L. P. Atkinson, D. W. Menzel, K. A. Bush, eds.), p. 63-6. American Geophysical Union, Washington, D.C. Pietrafesa, L. J., G. S. Janowitx, and P. A. Wittman. 1985b. Physical oceanographic processes in the Caro- lina Capes. In Oceanography of the southeastern U.S. continental shelf (L. P. Atkinson, D. W. Menzel, K. A. Bush, eds.), p. 23-32. American Geophysical Union, Washington, D.C. Pietrafesa, L. J., J. M. Morrison, M. P. McCann, J. Churchill, E. Bohm, and R. W. Houghton. 1994. Water mass linkages between the Middle and South Atlantic Bights. Deep-Sea Res. II 41:365-89. Plan Development Team. 1990. The potential of marine fishery reserves for reef fish management in the U.S. southern Atlantic. NOAA Tech. Memo. NMFS-SEFC-261, 40 p. Powell, A. B., D. G. Lindquist, and J. A. Hare. 2000. Larval and pelagic juvenile fishes collected with three types of gear in Gulf Stream and shelf waters in Onslow Bay, North Carolina, and comments on ichthyo- plankton distribution and hydrography. Fish. Bull. 98:427-38. Powell, A. B., and R. E. Robbins 1994. Abundance and distribution of ichthyoplankton along an inshore-offshore transect in Onslow Bay, North Carolina. NOAA Tech. Report NMFS 120, 28 p. 1998. Ichthyoplankton adjacent to live-bottom habitats in Onslow Bay, North Carolina. NOAA Tech. Report NMFS 133, 32 p. Powles, H., and B. W. Stender. 1976. Observations on composition, seasonality and distri- bution of ichthyoplankton from MARMAP cruises in the South Atlantic Bight in 1973. South Carolina Marine Resources Center, Technical Report Series Number 11, Charleston, SC. Rakocinski, C. F, J. Lyczkowski-Shultz, and S. L. Richardson. 1996. Ichthyoplankton assemblage structure in Missis- sippi Sound as revealed by canonical correspondence analysis. Estuar. Coast. Shelf Sci. 43:237-57. Richards, W.J. led.). 2001. Preliminary guide to the identification of the early life history stages of fishes of the Western Central Atlantic, (http://www4.cookman.edu/noaa). [Accessed 12 May 2004.] Simpson, J. H., and I. D. James. 1986. Coastal and estuarine fronts. In Baroclinic pro- cesses on continental shelves (C. N. K. Mooers, ed.), p. 63-93. American Geophysical Union, Washington, D.C. Smith, W., P. Berrien, and T Potthoff. 1994. Spawning patterns of bluefish, Pomatomus salta- trix in the northeast continental shelf ecosystem. Bull. Mar. Sci. 54:8-6. Smith, K. A., M. T Gibbs, J. H. Middleton, and I. M. Suthers. 1999. Short term variability in larval fish assem- blages of the Sydney shelf: tracers of hydrographic variability. Mar. Ecol. Prog. Ser. 178:1-5. Stephenson, R. L. 1999. Stock complexity in fisheries management: A perspective of emerging issues related to population sub-units. Fish. Res. 43:247-249 Sutter, F. C, III, R. I. Williams, and M. F. Godcharles. 1991. Movement patterns and stock affinities of king mackerel in the southeast United States. Fish. Bull. 89:315-324 ter Braak, C. F. J. 1988. CANOCO— a FORTRAN program for canonical com- munity ordination by correspondence analysis, principal components analysis and redundancy analysis. Tech- nical Report: LWA-88-02. Groep Landbouw-wiskunde, Wageningen, The Netherlands. Marancik et a\ :. Fish assemblages on the southeast United States continental shelf 129 ter Braak, C. F. J., and P. Smilauer. 2002. CANOCO Reference manual and CanoDraw for Windows User's guide: software for canonical community ordination (vers. 4.5), 500 p. Microcomputer Power Ithaca, New York, NY. Warlen, S. M., and J. S. Burke. 1990. Immigration of larvae of fall/winter spawning marine fishes into a North Carolina estuary. Estuaries 13:453-61. Warner, R. R., S. E. Swearer, and J. E. Caselle. 2000. Larval accumulation and retention: Implications for the design of marine reserves and essential fish habitat. Bull. Mar. Sci. 66:821-830. 130 Abstract — Inter and intra-annual var- iation in year-class strength was ana- lyzed for San Francisco Bay Pacific herring (Clupea pallasi) by using oto- liths of juveniles. Juvenile herring were collected from March through June in 1999 and 2000 and otoliths from subsamples of these collections were aged by daily otolith increment analysis. The composition of the year classes in 1999 and 2000 were deter- mined by back-calculating the birth date distribution for surviving juve- nile herring. In 2000, 729% more juveniles were captured than in 1999, even though an estimated 12% fewer eggs were spawned in 2000. Spawn- ing-date distributions show that survival for the 2000 year class was exceptionally good for a short (approx- imately 1 month) period of spawn- ing, resulting in a large abundance of juvenile recruits. Analysis of age at size shows that growth rate increased significantly as the spawning season progressed both in 1999 and 2000. However, only in 2000 were the bulk of surviving juveniles a product of the fast growth period. In the two years examined, year-class strength was not predicted by the estimated number of eggs spawned, but rather appeared to depend on survival of eggs or larvae (or both) through the juvenile stage. Fast growth through the larval stage may have little effect on year-class strength if mortality during the egg stage is high and few larvae are available. Year-class formation in Pacific herring (Clupea pallasi) estimated from spawning-date distributions of juveniles in San Francisco Bay, California Michael R. O Fan ell Ralph J. Larson Department of Biology San Francisco State University 1600 Holloway Avenue San Francisco, CA 94132 Present address (for M. R. O'Farrell, contact author): Department of Wildlife. Fish and Conservation Biology University of California, Davis One Shields Avenue Davis, California 95616 E-mail address (for M R O'Farrell)' mrofarrellffi'ucdavis edu Manuscript submitted 27 February 2003 to the Scientific Editor's Office. Manuscript approved for publication 2 August 2004 by the Scientific Editor. Fish. Bull. 103:130-141 (2005). Both biological and physical sources of mortality have been suggested as important in determining year-class strength in fish populations. Food lim- itation at first feeding (Hjort, 1914; Cushing, 1975; Lasker, 1975; Cushing, 1996), larval retention (lies and Sin- clair, 1982; Sinclair and lies, 1985), a juvenile critical period (Bollens et al„ 1992; Thorisson, 1994), as well as predation and environmental condi- tions may ultimately affect recruit- ment. Egg development time and larval growth rate have the capacity to adjust the relative impacts of these mortality sources on individual prop- agules by modifying stage duration (Houde, 1989; Yoklavich and Bailey, 1990). Juvenile fishes can be used to as- sess both inter- and intra-annual variation in egg and larval survival. Interannual variation in year-class strength is often inferred from mea- sures of juvenile abundance (e.g., Baxter et al., 1999). In addition, when the total number of eggs spawned is known, juvenile abundance can be used to assess overall variation in egg and larval survival. Intra-annual variation in egg and larval survival can be estimated from the birth-date distribution of surviving juveniles, as determined from otolith daily in- crement analysis. Particularly when data on actual spawning-date distri- butions are available, the birth date distribution of survivors can be used to identify periods of spawning that contributed differentially to juvenile recruitment (Methot, 1983; Rice et al., 1987; Yoklavich and Bailey, 1990; Moksness and Fossum, 1992; Fox, 1997; Takahashi et al., 1999). Recruitment of juvenile Pacific her- ring (Clupea pallasi) varies interan- nually by over an order of magnitude in San Francisco Bay (Baxter et al., 1999) and is the culmination of sever- al processes. Schools of adult herring enter San Francisco Bay in discrete batches during the fall and winter. These schools shoal and deposit eggs and milt during spawning events that often correspond to the quarter moon phase. Spawning events can vary in duration from approximately one day to one week, and simultaneous events may occur at different spawning sites throughout the bay. Herring lay adhe- sive eggs intertidally and subtidally on rocks, algae, aquatic plants, pier pilings, and other substrates (Alderd- ice and Velsen, 1971; Hay, 1985). Eggs can experience extremely high mortal- ity due to predation (McGurk, 1986; Bishop and Green, 2001), suboptimal temperature and salinity conditions (Alderice and Velsen, 1971; Griffin et al., 1998), as well as reduced hatch- ing and developmental abnormalities associated with certain substrate se- O'Farrell and Larson: Year-class formation in Clupea palllasi 131 lection (Vines et al., 2000). Larvae hatch from eggs after an incubation period, and the San Francisco Bay estuary can serve as a larval nursery area until after metamorphosis into the juvenile stage (Hay, 1985). Our objectives were 1) to identify periods in the spawning season that lead to successful (or unsuc- cessful) juvenile recruitment and 2) to evaluate larval and juvenile growth variation for two herring year classes. We used otoliths of juvenile herring from the 1999 and 2000 year classes to back-calculate spawn- ing-date distributions and determine spawning times that lead to successful recruitment. Distributions of spawning were obtained from management surveys. Growth was then evaluated to determine its role in year-class formation. Methods Surveys All information on adult herring spawning events and juvenile herring specimens were obtained from ongoing monitoring and management surveys conducted by the California Department of Fish and Game (CDFG). Data on timing, location, and magnitude of her- ring spawning events for the 1998-99 and 1999-2000 spawning seasons were obtained from the herring spawn survey conducted by the California Department of Fish and Game (CDFG). The survey is conducted from November through March throughout central San Francisco Bay, the area of most herring spawning (Wat- ters et al., 2004). The central bay region is searched for herring spawning on a daily basis from a small boat, and the entire spawning region is covered at least once per week. Eggs are located visually at low tide and by rake in shallow subtidal areas. When a spawning area is located, the number of eggs per square meter is measured from a subsample of the spawning area and is expanded to an estimate of total eggs spawned (for spawning survey method details, see Spratt, 1981; Watters et al., 2004). At the end of the 1998-99 and 1999-2000 spawning seasons, information on date, location, spawning area, average eggs/m2, total eggs, and the spawning biomass estimate was provided for the purpose of this study (Watters1). Juvenile (age-0) herring were sampled monthly from 30 stations in San Francisco Bay aboard the RV Long- fin as part of CDFG's Bay/Delta Division's Bay study (Fig. 1). Each station was visited once a month and juvenile herring were retained from catches during the months of April- June 1999 and March- June 2000. Stations were sampled by mid-water trawl with a 3.7-m2 mouth and 1.3-cm mesh codend, towed against the cur- rent, for 12 minutes. Volume of water filtered was cal- culated by using a flowmeter and was used to calculate -122°30'W 38'00'N 37°30'N 1 Watters, D. 2000. Personal commun. Calif. Dep. Fish and Game, 411 Burgess Dr., Menlo Park, CA 94025. Figure 1 Midwater trawl sampling stations in San Francisco Bay. catch per unit of effort (CPUE) for each station. Juve- nile herring were measured onboard, sorted from the catch, kept on ice, and transported to the laboratory, where they were frozen. Relative recruitment in each year was calculated by summing the CPUE at each sta- tion for the months of March-June in 1999 and 2000. Otolith preparation and analysis Frozen juvenile herring, separated by date and station, were thawed in batches and all fish were re-measured for standard length to the nearest mm. If the catch was small at a particular station (less than approximately 10 individuals), all specimens from that station were reserved for otolith analysis. If the catch was large, a subsample of the measured catch was reserved for otolith analysis. Subsampling consisted of randomly selecting at least two specimens from each 1-mm length bin in the catch. Both sagittal otoliths were extracted from each fish, cleaned with fresh water, and transferred to a micro- scope slide where they were allowed to dry. When com- pletely dry, both otoliths were mounted on the slide, convex side up, with clear nail polish. Otoliths were read with a compound microscope. Be- cause otoliths were too thick to allow sufficient light transmission for increment reading, all otoliths were 132 Fishery Bulletin 103(1) ground with 2000 grit sandpaper. Otoliths were al- ternately ground and examined under the microscope at 100 x to ensure that the section was thin enough to allow sufficient light transmission, yet not over-ground so that the edges of the otolith were lost. Daily increment deposition in herring begins at yolksac absorption, corresponding with the first heavy ring near the nucleus (Geffen, 1982; McGurk, 1984a; McGurk, 1987; Moksness and Wespestad, 1989). This heavy ring was located in all herring examined and increment counts were initiated there. Increment counts were made at 1000 x (with an oil immersion objective) and 400x (without oil immersion) magnification along the axis of maximum resolution. All increments were counted from the first heavy ring until the last ring on the edge of the otolith. Several days after the first reading, the same reader performed a reading on the second otolith. If the two increment counts differed by more than a value of 7, a third reading was conducted at a later date on the highest quality otolith. If the three increment counts differed from each other by more than a value of 7, otolith data from that fish were not used in further analyses. Where two readings differed by 7 or fewer in- crements, the final increment number for each fish was determined by averaging the two increment counts. Daily otolith increment deposition has been demon- strated in Pacific herring larvae reared in captivity (McGurk, 1984a; Moksness and Wespestad, 1989) and in the field (McGurk, 1987). In our study, otolith in- crements were assumed to be deposited daily and the validity of this assumption is treated in the "Results" and "Discussion" sections. Precision of otolith incre- ment counts was determined by computing the average percent error for each otolith examined (Beamish and Fournier, 1981). Spawning-date distributions Spawning-date distributions were constructed from specimens retained for otolith analysis in 1999 and 2000. Distributions were calculated 1) by adding a con- stant of 14 days to the otolith increment count and 2) by subtracting that value (otolith increments+14) from the Julian date of capture. Because Pacific herring begin daily increment deposition at yolksac absorption, the constant of 14 days was added to the increment value to account for egg incubation and the yolksac larval period. Taylor (1971) reported a 9-day egg incubation period for a British Columbia Pacific herring stock between 13.4°C and 13.8°C. For San Francisco Bay spawned herring, Griffin et al. (1998) found developmental rate to be influenced by salinity; the greatest hatching rate occurred 10 days after fertilization at a salinity of 14 ppt. Yolksac absorption occurs in Pacific herring 4-7 days after hatching (McGurk, 1987; Griffin et al., 2004, and references therein). The final value of 14 days for egg incubation and yolksac absorption used in our study was determined 1) from laboratory-derived values reported for British Columbia (Taylor, 1971; McGurk, 1987) and San Francisco Bay (Griffin et al., 1998) herring popu- lations and 2) by visually matching back-calculated spawning-date distributions with the observed spawn- ing-date distribution from the CDFG spawn-deposition survey. The back-calculated spawning-date distributions determined from specimens used for otolith analy- sis were extrapolated to include as many herring as possible caught in the juvenile surveys of 1999 and 2000. Length-frequency distributions were converted to spawning-date distributions by using age-length keys. Separate age-length keys were constructed for each survey in both 1999 and 2000. In some cases, the monthly survey was split into two legs separated by several days. When the monthly survey was split into legs, separate age-length keys were constructed for each leg. It was not possible to fit all herring caught between the months of March and June into age-length keys because some samples were inadvertently discarded after measurement in the field. If the range of lengths in the discarded samples extended beyond the sizes of samples aged, a complete age-length key could not be constructed. To avoid ascribing a possibly inaccurate age to a fish outside the size range of the age-length key, those fish were not included in the spawning-date distribution. Table 1 displays the number of herring caught in each leg, the number of otoliths used to con- struct the age-length key for that survey leg, and the total number and proportion of juveniles caught that are represented in the spawning-date distribution. The number of juveniles caught was greater than the num- ber of juveniles in the spawning-date distribution for all but one survey leg. This discrepancy was due to discarded fish (in the field) with lengths not within the range of the age-length key constructed from the subsampled individuals. Mortality estimate corrections are often superimposed upon spawning-date or hatching-date distributions to account for different size juveniles captured (Methot, 1983). Presumably a larger juvenile is older, and thus has been exposed to mortality factors for a longer pe- riod of time than has a smaller juvenile. The lack of a correction for juvenile mortality can lead to an under- representation of larger juveniles in the distribution. Because of the noncontinuous mid-water trawl sampling schedule, mortality rates could not be estimated from the data used in our study. As a result, mortality cor- rections were calculated by using an instantaneous mortality rate value of 0.016/d, corresponding to the greater of two mortality rates calculated from juve- nile Pacific herring in Prince William Sound, Alaska (Stokesbury et al., 2002). Spawning-date distributions were corrected for mor- tality by calculating abundance at age 100 days (N100). For fishes aged at less than 100 days: M - M e-0.0161100 -al JV100 _ JVne ' (1) where a is the age of the fish in days. O'Farrell and Larson: Year-class formation in Clupea palllasi 133 Table 1 Summary of the catch. number of Clupea pallaai otoliths examined fi om the catch. number and percent available for use in the spawning-date dist ributions. and catch per unit of effort ( CPUE ) for the midwater trawl survey in 1999 and 2000. iUPUE repre- sents summed CPUE for all stations in each survey leg. Juveniles were not used in analysis if they were inadvertently discarded in the field and if a complete age-length key could not be constructed. Juveniles Otoliths Used in Percent Survey dates Area surveyed caught examined analysis used ZCPUE 1999 Mar 99 entire bay 0 0 0 0 0 21 Apr 99 central and north 41 0 0 0% 1653 26-28 Apr 99 south and north 66 53 60 91% 2360 18-19 May 99 north 19 4 2 11% 771 24-27 May 99 north, central, and south 280 251 273 98% 12,856 9-10 Jun 99 north and central 91 25 45 49% 3457 15 Jun 99 south 61 0 0 0% 2551 Total 558 333 380 68% 23,648 2000 8-9 Mar 00 north and central 11 0 0 0% 637 13-14 Mar 00 south 7 7 6 86% 294 4-5 Apr 00 north 25 25 25 100% 1053 10-11 Apr 00 central and south 302 115 284 94% 14,712 10 May 00 north 898 77 740 82% 38,270 22-24 May 00 central and south 2244 77 2237 100% 102,516 6-7 Jun 00 central and south 569 74 569 100% 25,352 13 Jun 00 north 13 0 0 0% 539 Total 4069 375 3861 95% 183,373 For fishes aged greater than 100 days: N = N- ■"100 „-0.016la-100l (2) Combining the results of Equations 1 and 2 produced the mortality-corrected spawning-date distributions. Growth To evaluate correlates of both inter- and intra-annual variation in survival to the juvenile stage, we wanted to compare growth rates of herring up to the juvenile stage. However, because it was apparent that growth rates may have differed for specimens spawned at different times of the year, either a linear or nonlinear growth curve fitted to size-at-age data would be erroneous (O'Farrell, 2001). Larger (older) and smaller (younger) individuals would have experienced different growth histories; therefore a plot of size versus age for any sample of fish would not reflect the growth history of any one cohort. Further- more, consecutive samples rarely contained individuals from any given cohort because older juveniles appeared to leave San Francisco Bay. Finally, we did not have data on size at age of larvae; therefore growth curves would be incomplete. Instead, we used age at size to compare growth with- in and between years. To do this, we computed the num- Table 2 Summary statistics and distribution of juvenile Clupea harengus lengths within the 40-50 mm size bin for sam- pling events where size-at -age data were used. Other sampling events were not ncluded in growth analyses because they did not contain juvenile herring bel ween the sizes of 40 mm and 50 mm. Survey leg n Mean (mm) SD(mm) 26-27 Apr 99 15 43.80 2.54 24-27 May 99 162 45.02 2.63 9 Jun 99 10 42.20 2.82 5 Apr 00 16 46.25 3.00 10-11 Apr 00 23 46.43 2.94 10 May 00 9 43.67 3.04 22-24 May 00 36 44.81 3.19 6-7 Jun 00 36 46.56 2.82 ber of otolith increments (days after yolksac absorption) present in fish between 40 mm and 50 mm standard length. This size group was chosen to analyze growth because it was well represented in both in the 1998-99 and 1999-2000 spawning seasons. The mean and stan- 134 Fishery Bulletin 103(1 dard deviation of the length distribution within the 40-50 mm bin for each sampling event is provided in Table 2. Thus, the amount of time (measured by otolith increments) needed for fish to grow to the 40 mm-50 mm size group was used to compare growth. Differences in age at length were evaluated and compared with observed variation in juvenile abundance. distributed throughout the bay (Fig. 3). Peak abun- dances occurred in May for both 1999 and 2000, and juveniles were caught throughout the study area. By June, abundances decreased and herring became more concentrated in the central Bay region, presumably ag- gregating in this area prior to exiting San Francisco Bay for the coastal ocean (Fig. 3). Results Egg and juvenile abundance Both the magnitude and timing of estimated egg deposi- tion differed little between the 1998-99 and 1999-2000 spawning seasons (Fig. 2). Total egg deposition was esti- mated to be 9.66 x 1011 eggs for 1998-1999 and 8.59 x 1011 eggs for 1999-2000 (Watters2). Peak egg deposition in both spawning years occurred in January (Fig. 2). Abundance of juvenile herring resulting from these two spawning seasons differed greatly. The cumula- tive estimated relative recruitment (ICPUE) of juve- nile herring was 7.75 times greater in 2000 than 1999 (Table 1). General patterns of juvenile herring distribution were similar in 1999 and 2000. Juvenile herring recruited to the sampling gear in March and April and were widely Watters, D. 2000. Unpubl. data. Calif. Dep of Fish and Game, 411 Burgess Dr., Menlo Park, CA 94025. 7x10" 6x10" ■a 5x10 CD cd 4x10 "D cfl lu 3x10 2x10" 1x10" 1998-1999 1999-2000 Nov Dec Jan Feb Mar Spawning month Figure 2 Total egg deposition by Pacific herring [Clupea pal- lasi), summed by spawning month for the 1998-99 and 1999-2000 spawning seasons. Data provided by the Cali- fornia Department of Fish and Game, Menlo Park. Spawning-date distributions The temporal distribution of successful spawning-dates differed between the 1999 and 2000 year classes (Fig. 4, A and B). In 1999, the earliest spawning-date that resulted in juvenile recruitment was 30 November 1998. The greatest numbers of juvenile recruits were a product of the middle of the spawning season, from approxi- mately early January 1999 though early February 1999, and the highest recruitment occurred from spawnings between 10 January and 14 January 1999 (Fig. 4A). An additional spike of recruitment was observed from spawning events at the end of the season (early March). The period of highest recruitment came at the same time as the highest spawning intensity. Spawning events early in the spawning season (November-December) appeared to produce few juveniles (Fig. 4A). In 2000, juveniles recruited from much earlier spawn- ing events. Back-calculated spawning dates indicated that spawning may have occurred as early as 13 Octo- ber 1999 (Fig. 4B). Both the March 2000 and April 2000 juvenile surveys contained herring with back-calculated spawning dates that ranged from mid to late October, indicating that a spawning event occurred extremely early in the spawning season and was undetected by the spawn-deposition survey (which commences in No- vember). Although early spawnings appeared to produce some recruitment success, a near lack of success was noted for many of the mid-season spawnings that oc- curred from mid-November through mid-January 2000 (Fig. 4B). This period of poor survival was then followed by the period of highest recruitment; spawning dates ranged from mid-January to early March and peak re- cruitment resulted from February spawning (Fig. 4B). Juvenile mortality corrections superimposed upon the spawning-date distributions had little effect on the general results. An instantaneous juvenile mortal- ity rate of 0.016/d produced minor adjustments on the percent recruitment resulting from particular spawning periods in both years (Fig. 4, A and B). This mortality correction did not alter the general spawning periods that resulted in juvenile recruitment. Increasing the in- stantaneous juvenile mortality rate to 0.05/d (O'Farrell. unpubl. data) also had negligible effects on the general results of the spawning-date distributions. Data for both 1999 and 2000 are not totally complete. The spawning-date distribution for 1999 was based on a total of 380 herring, whereas 558 herring were caught between the months of March and June. Similarly, the 2000 spawning-date distribution was based on a total of 3861 herring, whereas 4069 herring were caught dur- ing the same months (Table 1). Fish were omitted from O'Farrell and Larson: Year-class formation in Clupea palllasi 135 >NNC N^f NoC0 O o ONCO Apr 99 NCNO. NC«N(NC NC %. : nc o NC°. D NC NC o NO?* NCN© o Mar 00 NC oo*nc °o9-# o o CNC oNS° OqNC B May 99 NC NC NC, NC E NC*N(NC O NC N(Nfic Apr 00 Nc .NCNC ^JC N(«- NC NC NC^C NC Jun 99 NC >OoO >€>' NC NC» NC NCN' ^C May Qfl CPUE f~) > 1 0.000 O 5001-10,000 O 1001-5000 o 1-1000 NC No Catch rfeNS G A • NC NCN^C NC Jun 00 NC) Figure 3 Juvenile herring {Clupea pallasi) CPUE distribution by station and month for 1999 and 2000. April 1999 (A) dark bubbles represent the 21 April survey leg and light bubbles represent the 28-28 April survey leg. May 1999 (B) dark bubbles represent the 18-19 May survey leg and light bubbles represent the 24-27 May survey leg. June 1999 (C) dark bubbles represent the 9-10 June survey leg and light bubbles represent the 15 June survey leg. March 2000 (D) dark bubbles represent the 8-9 March survey leg and light bubbles represent the 13-14 March survey leg. April 2000 (E) dark bubbles represent the 4-5 April survey leg and light bubbles represent the 10-11 April survey leg. May 2000 (F) dark bubbles represent the 22-24 May survey leg and light bubbles represent the 10 May survey leg. June 2000 (G) dark bubbles represent the 6-7 June survey leg and light bubbles represent 13 June survey leg. the spawning-date distribution because some samples were discarded and otoliths were unavailable. Because of evidence for intrayear growth-rate variation, other age-at-length data were not used to infer spawning dates for these fish. The standard length data for the fish not included in this analysis were used for all other analyses in our study. Precision of multiple otolith readings was calcu- lated for all otoliths examined. Average percent error (Beamish and Fournier, 1981) was 3.60% in 1999 and 1.64% in 2000, indicating that aging precision was less than 4 days for 100-day old herring in both years. Growth Different patterns of age at length (40-50 mm) were observed in 1999 and 2000. In 1999, specimens between 40 mm and 50 mm were captured in three survey legs. A significant decrease in the number of otolith incre- ments for juveniles 40 mm-50 mm standard length was detected in 1999 (Fig. 5A; Kruskal-Wallis test; r7=27.93, P<0.0001). Nonparametric multiple compari- sons indicated that there was a nonsignificant difference in otolith increment counts for herring caught in the April 1999 and the May 1999 surveys, but herring from these surveys had significantly higher median otolith increment counts than those from the June 1999 survey. In this later survey, juvenile herring were caught that were a product of spawning events occurring late in the spawning season. Figure 5C displays the median and range of spawning dates of the specimens aged for Figure 5A. Juvenile herring that were a product of spawning between 27 February 1999 and 7 March 1999 reached a 40-50 mm size range significantly faster than 136 Fishery Bulletin 103(1) 20 15 10 5 - with mortality correction without mortality correction eggs 0 10/1/98 6x10' 5x10" 4x10' 3x101 2x10' 1x10" 12/1/98 2/1/99 4/1/99 25 20 15 - 10 - 5 - with mortality correction without mortality correction eggs B o 10/1/99 ^l MJ 6x10' _ 5x10' 4x10' 3x10' 2x10' 1x10' 12/1/99 2/1/00 Spawning date 4/1/00 Figure 4 Spawning-date distributions for juvenile herring (Clupea pallasi) caught in (A) 1999 and (B) 2000. Vertical bars represent dates and magnitude of observed spawning (eggs deposited), heavy lines represent the spawning-date distribution of juveniles without the mortality correction, and light lines represent the spawning-date distribution corrected for juvenile mortality at an instantaneous rate of 0.016/d. Distributions are smoothed with a cubic spline interpolation. Data on observed spawning were provided by D. Watters, CDFG (see Footnote 2 in the text). O'Farrell and Larson: Year-class formation in Clupea palllasi 137 160 140 - S 120 - E o 5 15 162 160 140 - 120 - 100 80 60 40 3/1/99 4/1/99 5/1/99 6/1/99 7/1/99 B \ 36 Jb 40 3/1/00 4/1/00 5/1/00 6/1/00 7/1/00 Collection date - 12 140 - - 10 - 8 120 - - 6 100 - - 4 80 - - 2 60 - 40 D i — ■ 1 NL 9/1/98 11/1/98 1/1/99 3/1/99 5/1/99 9/1/99 11/1/99 1/1/00 3/1/00 5/1/00 Spawning date Figure 5 The upper panels display otolith increments present in 40 mm-50 mm juvenile herring {Clupea pallasi) arranged by capture date for (A) 1999 and (B) 2000. Boxes represent median number of otolith increments, bars indicate ±1 SD, and the number above each point is the sample size. The lower panel displays growth histories for juvenile herring originating from various periods within the spawning season for (C) 1999 and (D) 2000. Boxes represent median spawn- ing-dates and bars represent range of spawning dates at that growth rate. The spawning-date distribution (uncorrected for juvenile mortality) is superimposed upon C and D to ascertain how changes in growth are reflected in survival to the juvenile stage. specimens recruiting from earlier spawning periods. The period of greatest recruitment occurred during the slower growth period in 1999 (Fig. 5C). In 2000, 40 mm-50 mm juvenile herring were caught in five survey legs conducted during three months (April, May, and June). The data are displayed by survey leg; pooling the data by month, however, does not change the result. Median increment counts differed significantly for the 2000 surveys (Fig. 5B; #=76.39, P<0.0001). Oto- lith increment counts for 40 mm-50 mm specimens did not differ for the 5 April 2000 and 10-11 April 2000 surveys. However, the age at length for these surveys was significantly greater than for the three later survey legs (10 May 2000, 22-24 May 2000, and 6 June 2000), which did not significantly differ from each other. Her- ring caught in the three later surveys grew significantly faster than herring caught in the two earlier surveys. The significant decrease in age at length indicates that juvenile herring that were a product of spawning be- tween 15 January 2000 and 18 March 2000 grew faster than specimens recruiting from earlier spawning events. The majority of juvenile recruits in 2000 were a product of the fast growth period (Fig. 5D). Accuracy of growth-rate estimates determined from growth increments on otoliths The above analyses depended upon the assumption that increments were deposited daily in the otoliths exam- ined. Two lines of evidence point to the validity of this assumption. First, back-calculated spawning-dates gen- erally agreed with the known spawning season of San Francisco Bay herring, and several peaks in back-cal- culated spawning dates match known spawning events quite closely (Fig. 4, A and B). Second, juvenile growth rates appear to be high enough for daily growth (McGurk, 1984b). Clear length- frequency modes were visible for three sampling events 138 Fishery Bulletin 103(1) in 2000. Assuming linear growth between these time periods, the advancement of these length-frequency modes resulted in growth rates of 0.75 mm/d (Fig. 6, arrow in A), 0.83 mm/d (arrow in B), and 0.64 mm/d (arrow in C). McGurk (1984b) demonstrated daily in- crement deposition in herring if the larval growth rate exceeded 0.36 mm/d. Our data did not allow us to es- timate growth rates of larvae; however, the estimated juvenile growth rates presented above are much greater than necessary for daily increment deposition. Discussion Catches of juvenile herring were much greater in 2000 than in 1999. Between the months of March and June 2000, cumulative CPUE was more that seven times greater than during the same period in 1999, yet an estimated 12% more eggs were deposited during the 10 May 2000 20 30 40 50 60 Standard length (mm) Figure 6 Length frequencies for juvenile herring (Clupea pallasi) captured on 10 May 2000, 22 May 2000, and 6 June 2000. Arrows represent the estimated propagation of length modes through time. Linear growth rates, calculated from each trajectory, are as follows: A=0.75 mm/d; trajectory B = 0.83 mm/d); and C = 0.64 mm/d. 1988-99 spawning season. Because observed differences in recruitment between 1999 and 2000 far exceeded dif- ferences in the total eggs spawned, differential survivor- ship during the egg or larval stages (or both) must be responsible for disparate year-class strengths. The spawning-date distributions presented for 1999 and 2000 did not contain all herring caught by the mid- water trawl survey between the months of March and June. Because they could not be accurately assigned ages with an age-length key (Table 1), 178 herring were omitted from the distribution in 1999. Most specimens omitted from this distribution were caught in the early April 1999 and late June 1999 survey legs. As a result, the spawning-date distribution likely underestimated the recruitment from very early and very late season spawnings. In 2000, 208 specimens, from a variety of survey legs, were omitted from the spawning-date dis- tribution (Table 1). Because a large number of herring were caught in 2000, it is unlikely that these omissions would significantly change the shape of the spawn- ing-date distribution. The loss of data in this case does not change the overall result of large year-class- strength variation. The noncontinuous sampling schedule for juve- niles may have resulted in either an underestima- tion or overestimation of CPUE and thus year-class strength. In several months, the mid-water trawl sur- vey was conducted over two legs separated by several days (Table 1, Fig. 3). This noncontinuous sampling could have produced error in our estimates because aggregations of juveniles, through movement between areas, could conceivably have escaped detection by trawls (resulting in CPUE underestimation) or have been sampled twice in the same month (resulting in CPUE overestimation). However, O'Farrell (2001) showed that dispersal of herring from a successful spawning event could occur through much of San Francisco Bay. Therefore, we do not believe that ag- gregations of juveniles were completely missed by the mid-water trawl survey. The degree to which ag- gregations of juveniles were sampled more than once in a sampling month is not known. Variation in age estimates undoubtedly produced back-calculated spawning-dates that did not match exactly with true spawning dates. Yet, for some spawning events, very good matches between back- calculated and reported spawning events indicate that the age estimations were accurate for many of the cohorts examined (O'Farrell, 2001). Other cohorts that did not match as well with reported spawnings may be the result of 1) a spawning event undetected by the spawn-deposition study, 2) a small, "spot" spawning that did not qualify as a true spawning event for the spawn-deposition study, or 3) very slow or fast growth through a portion of the larval life history that interrupted daily increment deposition (McGurk, 1984b, 1987). Increased survival did not occur throughout the entire 2000 spawning season. Instead, periods of good survival and poor survival were present, yet the O'Farrell and Larson: Year-class formation in Clupea palllasi 139 periods of good survival in 2000 led to a much stronger year class than that of 1999. Detecting a "match" of favorable conditions that led to recruitment success was not possible in our study because of the myriad factors that can determine recruitment success. Rather than attempting to explain the observed survival differences with specific mechanisms, we suggest what may pos- sibly contribute to the observed patterns. Larval survival The degree to which larval survival depends upon biotic or abiotic factors is difficult to estimate. Fox (2001) presented data showing that year-class strength in the Blackwater stock of Atlantic herring {Clupea harengus L.) was determined by survival after the egg stage. How- ever, it is not clear whether variation in survival was due to density-dependent or environmental factors. A recent study has shown that salinity can affect larval survival after hatching in San Francisco Bay herring (Griffin et al., 20041. Here, the salinity during embryonic develop- ment was a factor in yolksac larval survival in different salinity treatments. Regardless of the form of mortality operating on larvae, small changes in larval growth rate can lead to large changes in levels of recruitment (Houde, 1987). Faster larval growth results in shorter larval stage duration and thus decreased exposure to the characteristically high mortality of the larval stage. Age at size for herring in this study decreased signifi- cantly as the spawning season progressed both in 1999 and 2000. From this finding, we infer that positive changes in growth rate occurred during the spring and summer. Seasonal positive shifts in growth have also been observed in Pacific herring populations in Prince William Sound, Alaska, between the months of June and October (Stokesbury et al., 1999). In 1999, the greatest number of recruits came from mid to late-season spawning events. The late February to early March spike in recruitment (Fig. 4 A) may be partially explained by within-year growth variation. This group of survivors appeared to be derived from a rela- tively small number of eggs. Recruits from that spawn- ing period grew significantly faster than recruits from earlier spawning events. The largest spawning events of the 1998-99 spawning season produced recruits that grew slower than the recruits spawned in early March and thus may have experienced lower relative survival. Within-year growth rate variation also partially ex- plains the 2000 year class. The 2000 year class was dominated by late season recruitment, primarily from spawning in February 2000. Herring from spawning events occurring between late October 1999 and mid- January 2000 had a significantly higher median age at length than herring produced from subsequent spawn- ing times. This slow growth may in part explain the near lack of recruitment from the two highest magni- tude spawns occurring from 1 to 3 Jan 2000 and from 19 to 24 Jan 2000. However, age at length decreased (and thus growth rate increased) for spawning events occurring from late January 2000 to early March 2000. The timing of the growth rate switch (from slow to fast) coincided closely with the spawning period producing the greatest amount of recruitment. The general trend of high levels of recruitment from late season spawning events indicates that increased growth rate played a role in the good survival during this period. However, recruitment from very early spawning events and the small number of recruits resulting from late March 2000 spawning was not explained solely by this within- year growth variation. Egg mortality Variation in mortality during the egg stage may also affect recruitment in San Francisco Bay herring. Fertil- ization, embryonic development, and hatching success of Pacific herring are strongly tied to environmental condi- tions (Alderdice and Velsen, 1971, Griffin et al., 1998). The optimal range for fertilization and development of the San Francisco Bay population is between 12 ppt and 24 ppt, and both percent fertilization and percent hatching is maximized at 16 ppt (Griffin et al., 1998). The herring spawning season in San Francisco Bay is a time of rapidly changing salinities. High salinities generally persist through the fall months. In winter, rapid decreases in salinity due to freshwater from the San Joaquin-Sacramento Delta, storm drain runoff and local creek purges (Oda3) are common, yet the magnitude varies between years (Conomos et al., 1985). In the two years examined, salinity during the winter spawning season varied both above and below the optimum range determined by Griffin et al. (1998). These salinity fluc- tuations could have a large effect on the supply of larvae into the San Francisco Bay system. Mortality during the egg stage can be exceedingly high in Pacific herring due to predation and other biotic interactions (Alderdice and Velsen, 1971; McGurk, 1986; Rooper et al., 1999, Bishop and Green, 2001). As a re- sult, egg incubation time may have a significant effect upon eventual recruitment. The length of times of egg incubation and the yolksac larval stage were combined in our study and the combined period was given a con- stant value of 14 days. In actuality, egg incubation time (Taylor, 1971; McGurk, 1987) and embryonic develop- ment (Alderdice and Velsen, 1971; Griffin et al., 1998) are strongly linked to environmental factors and likely have a significant effect upon recruitment before growth rates can determine survival. Analysis of egg incubation and yolksac larval duration for separate cohorts was not performed in our study. It may, however, play a large role in larval abundance. Conclusion The 1999 and 2000 spawning-date distributions indicate that year classes can be shaped by periods of good and 3 Oda, K. 2000. Personal commun. Calif. Dep. Fish and Game, 411 Burgess Dr., Menlo Park, CA 94025. 140 Fishery Bulletin 103(1) poor survival lasting shorter than the duration of the spawning season, yet longer than the duration of an individual spawning event. The distributions indicated that variation in survivorship was not only a function of individual spawn success. Rather, periods of good and poor survivorship in 1999 and 2000 were of longer duration than one spawning event. The period of excep- tionally good survival that led to the majority of the strong 2000 year class was approximately one month in duration and incorporated several spawning events. Yet this window of good survival was much shorter than the entire 2000 spawning season. Variation in survivorship between individual spawnings may be less important in shaping the year class than survivorship variation on a longer time scale. Visual examination of the spawning-date distribu- tion superimposed upon juvenile age at length indicate that faster growth had a positive effect on recruitment in 2000, and a negligible effect in 1999. For larval growth to affect recruitment, larvae must be available from hatching eggs. Year-class strength variation in Pacific herring could depend upon both egg and larval survival. The timing of peak herring spawning in San Fran- cisco Bay may be a tradeoff between maximizing larval growth rates and spawning when hydrographic condi- tions are optimal for embryonic development. In the two years examined, growth rate increased with the progres- sion of the spawning season. It follows that the herring population could maximize recruitment by spawning later so that larvae grow faster. However, because delta outflow is generally high in February and March on account of winter storms, late season spawning may ex- pose eggs to low salinities and thus decreased hatching rates. Peak spawning may occur in January as a trade off between growth-rate and egg-hatching success. Acknowledgments This research would not have been possible without the extensive cooperation of the California Department of Fish and Game Belmont and Stockton offices. In par- ticular, we would like to thank Diana Watters, Ken Oda, Sara Peterson, Kathy Hieb, Kevin Fleming, Tom Greiner, Suzanne Deleon, and the entire crew of the RV Longfin. Stephen Bollens, Steven Obrebski, Ken Oda, and three anonymous reviewers provided very helpful comments on various drafts of this manuscript. Literature cited Alderdice, D. F., and F. P. J. Velsen. 1971. Some effects of salinity and temperature on early development of Pacific herring (Clupea harengus pallasi). J. Fish. Res. Board Can. 28:1545-1562. Baxter, R„ K. Hieb, S. DeLeon, K. Fleming, and J. Orsi. 1999. Report on the 1980-1995 fish, shrimp and crab sampling in the San Francisco estuary. Technical Report 63, California Department of Fish and Game, Stockton, CA. Beamish, R. J., and D. A. Fournier. 1981. A method for comparing the precision of a set of age determinations. Can. J. Fish. Aquat. Sci. 38:982-983. Bishop, M. A., and P. G. Green. 2001. Predation on Pacific herring (Clupea pallasi) spawn by birds in Prince William Sound, Alaska. Fish. Ocean- ogr. 10(suppl. 11:149-158. Bollens, S. M., B. W. Frost, H. R. Schwaninger, C. S. Davis, K. J. Way, and M. C. Landsteiner. 1992. Seasonal plankton cycles in a temperate fjord and comments on the match-mismatch hypothesis. J. Plankton Res. 14:1279-1305. Conomos, T. J., R. E. Smith, and J. W. Gartner. 1985. Environmental setting of San Francisco Bay. Hy- drobiologia 129:1-12. Cushing, D. H. 1975. Marine ecology and fisheries. Cambridge Univ. Press, Cambridge, UK. 1996. Toward a science of recruitment in fish pop- ulations. Excell. Ecol. 7, 175 p. Fogarty, M. J., M. P. Sissenwine, and E. B. Cohen. 1991. Recruitment variability and the dynamics of ex- ploited marine populations. Trends Ecol. Evol. 6: 241-246 Fox, C. J. 2001. Recent trends in stock-recruitment of Blackwa- ter herring {Clupea harengus L.) in relation to larval production. ICES J. Mar. Sci. 58:750-762. Fox, D. A. 1997. Otolith increment analysis and the application toward understanding recruitment variation in Pacific hake (Merluceius produetus) within Dabob Bay, WA. M.S. thesis, 73 p. Univ. Washington, Seattle, WA. Geffen, A. J. 1982. Otolith ring deposition in relation to growth rate in herring (Clupea harengus) and Turbot (Scophthalmus maximus) larvae. Mar. Biol. 71:317-326. Griffin, F. J., M. R. Brenner, H. M. Brown, E. H. Smith, C. A. Vines, and G. N. Cherr. 2004. Survival of Pacific herring larvae is a function of external salinity. In Early life history of fishes in the San Francisco Estuary and watershed (F. Feyrer, L. R. Brown, R. L. Brown, and J. J. Orsi, eds.), p. 37-46. Am. Fish. Soc. Symp. 39, Bethesda, MD. Griffin, F. J., M. C. Pillai, C. A. Vines, J. Kaaria, T. Hibbard-Robbins, R. Yanagimachi, and G. N. Cherr. 1998. Effects of salinity on sperm motility, fertiliza- tion, and development in the Pacific herring, Clupea pallasi. Biol. Bull. 194:25-35. Hay, D. E. 1985. Reproductive biology of Pacific herring (Clupea harengus pallasi). Can. J. Fish. Aquat. Sci. 42(supp. l.):lll-126. Hjort, J. 1914. Fluctuations in the great fisheries of northern Europe viewed in light of biological research. Rapp. P.V. Reun. Cons. Int. Explor. Mer 20:1-228. Houde, E. D. 1987. Fish early life dynamics and recruitment varia- bility. Am. Fish. Soc. Symp. 2:17-29. 1989. Subtleties and episodes in the early life of fishes. J. Fish Biol. 35(suppl. A):29-38. O'Farrell and Larson: Year-class formation in Clupea palllasi 141 lies, T. D. and M. Sinclair. 1982. Atlantic herring: stock discreteness and abun- dance. Science 215(5):627-633. Lasker, R. 1975. Field criteria for survival of anchovy larvae: the relation between inshore chlorophyll maximum layers and successful first feeding. Fish. Bull. 73:453-462. McGurk, M. D. 1984a. Ring deposition in the otoliths of larval Pacific herring, Clupea harengus pallasi. Fish. Bull. 82: 113-120. 1984b. Effects of delayed feeding and temperature on the age of irreversible starvation and on the rates of growth and mortality of Pacific herring larvae. Mar. Biol. 84:13-26. 1986. Natural mortality rates of marine pelagic fish eggs and larvae: role of spatial patchiness. Mar. Ecol. Prog. Ser. 34:227-242. 1987. Age and growth of Pacific herring based on length- frequency analysis and otolith ring number. Env. Biol. Fish. 20(11:33-47. Methot, R. D„ Jr. 1983. Seasonal variation in survival of larval northern anchovy, Engraulis mordax, estimated from the age distribution of juveniles. Fish. Bull. 81:741-750. Moksness, E., and P. Fossum. 1992. Daily growth rate and hatching-date distribu- tion of Norwegian spring-spawning herring (Clupea harengus L.l. ICES J. Mar. Sci. 49:217-221. Moksness, E., and V. Wespestad. 1989. Ageing and back-calculating growth rates of Pacific herring, Clupea pallasii, larvae by reading daily otolith increments. Fish. Bull. 87(31:509-513. O'Farrell, M. R. 2001. Year class formation of Pacific herring in San Francisco Bay. M.A. thesis, 73 p. San Francisco State Univ., San Francisco, CA. Rice, J. A., L. B. Crowder, and M.E. Holey. 1987. Exploration of mechanisms regulating larval sur- vival in Lake Michigan bloater: a recruitment analysis based on characteristics of individual larvae. Trans. Am. Fish. Soc. 116:703-718. Rooper, C. N., L. J. Haldorson, and T. J. Quinn II. 1999. Habitat factors controlling Pacific herring {Clupea pallasi) egg loss in Prince William Sound, Alaska. Can. J. Fish. Aquat. Sci. 56:1133-1142. Sinclair, M., and T. D. lies. 1985. Atlantic herring (Clupea harengus) distribution in the Gulf of Maine-Scotian Shelf area in relation to oceanographic features. Can. J. Fish. Aquat. Sci. 42:880-887. Spratt, J. D. 1981. Status of the Pacific herring, Clupea harengus pallasi, resource in California 1972 to 1980. Cal. Dep. Fish and Game, Fish. Bull. 171, 107 p. Stokesbury, K. D. E., R. L. Foy, and B. L. Norcross. 1999. Spatial and temporal variability in juvenile Pacific herring, Clupea pallasi, growth in Prince William Sound, Alaska. Environ. Biol. Fish. 56:409-418. Stokesbury K. D. E., J. Kirsch E. V. Patrick, and B. L. Norcross. 2002. Natural mortality estimates of juvenile Pacific herring (Clupea pallasi) in Prince William Sound, Alaska. Can. J. Fish. Aquat. Sci. 59:416-423. Takahashi, I., K. Azuma, H. Hiroyuki, and F. Shinji. 1999. Different mortality in larval stage of Ayu Pleeoglas- sus altivelis by birth dates in the Shimanto estuary and adjacent coastal waters. Fish. Sci. 65(2):206-210. Taylor, F. H. C. 1971. Variation in hatching success in Pacific herring (Clupea pallasi) eggs with water depth, temperature, salinity, and egg mass thickness. Rapp. P.-V. Reun. Cons. Perm. Int. Explor. Mer 160:128-136. Thorisson, K. 1994. Is metamorphosis a critical interval in the early-life of marine fishes? Environ. Biol. Fish. 40:23-36. Vines, C. A., T. Robbins, F. J. Griffin, and G. N. Cherr. 2000. The effects of diffusible creosote-derived com- pounds on development in Pacific herring (Clupea pallasi). Aquat. Toxicol. (NY) 51:225-239. Watters, D. L., H. M. Brown, F. J. Griffin, E. J. Larson, and G. N. Cherr. 2004. Pacific herring Clupea pallasii spawning grounds in San Francisco Bay: 1973 to 2000. In Early life history of fishes in the San Francisco Estuary and watershed (F. Feyrer, L. R. Brown, R. L. Brown, and J. J. Orsi, eds.), p. 3-14. Am. Fish. Soc. Symp. 39, Bethesda, MD. Yoklavich, M. M., and K. M. Bailey. 1990. Hatching period, growth and survival of young wall- eye pollock Theragra chalcogramma as determined from otolith analysis. Mar. Ecol. Prog. Ser. 64:13-23. 142 Abstract — Diet analysis of 52 log- gerhead sea turtles (Caretta caretta) collected as bycatch from 1990 to 1992 in the high-seas driftnet fishery oper- ating between lat. 29.5°N and 43°N and between long. 150°E and 154°W demonstrated that these turtles fed predominately at the surface; few deeper water prey items were pres- ent in their stomachs. The turtles ranged in size from 13.5 to 74.0 cm curved carapace length. Whole tur- tles (n = 10) and excised stomachs (n = 42) were frozen and transported to a laboratory for analysis of major faunal components. Neustonic species accounted for four of the five most common prey taxa. The most common prey items were Janthina spp. (Gas- tropoda); Carinaria cithara Benson 1835 (Heteropoda); a chondrophore, Velella velella (Hydrodia); Lepas spp. (Cirripedia), Planes spp. (Decapoda: Grapsidae), and pyrosomas (Pyrosoma spp.). Diet of oceanic loggerhead sea turtles (Caretta caretta) in the central North Pacific Denise M. Parker Joint Institute for Marine and Atmospheric Research 8604 La Jolla Shores Drive La Jolla, California 92037 Present address: Northwest Fisheries Science Center National Marine Fisheries Service, NOAA Newport, Oregon 97365-5275 E-mail address Denise Parkers noaa gov William J. Cooke AECOS, Inc. 970 N. Kalaheo Avenue, Suite C311 Kailua. Hawaii 96734 George H. Balazs Pacific Islands Fisheries Science Center, Honolulu Laboratory National Marine Fisheries Service 2570 Dole Street Honolulu, Hawaii 96822-2396 Manuscript submitted 15 July 2003 to the Scientific Editor's Office. Manuscript approved for publication 8 July 2004 by the Scientific Editor. Fish. Bull. 103:142-152 12005). Loggerhead sea turtles are circum- global, inhabiting temperate, sub- tropical, and tropical waters of the Atlantic, Pacific, and Indian Oceans. In the Pacific, loggerhead sea turtles have been found in nearshore waters of China, Taiwan, Japan, Australia, and New Zealand and are seen in off- shore waters of Washington, Califor- nia, and northwestern Mexico (Dodd, 1988; Pitman, 1990). Nesting in the North Pacific Ocean occurs in Japan; there is no known nesting in the east- ern North Pacific (Marquez and Vil- lanueva, 1982; Frazier, 1985; Bartlett, 1989). Trans-Pacific migrations of juveniles have been documented from mitochondrial DNA analyses of indi- viduals found feeding off Baja Cali- fornia. Bowen et al. (1995) identified these Baja sea turtles as originating from Japanese rookeries, although a a small percentage come from Aus- tralia. Recent research indicates that all loggerhead sea turtles found in the oceanic realm of the central North Pacific Ocean are of Japanese stock (Dutton et al., 1998). Tagging studies in Japan and the Eastern Pacific also demonstrate transpacific migrations of loggerhead sea turtles between the east and west Pacific (Balazs, 1989; Resendiz et al., 1998; Uchida and Teruya1). Recent oceanic satellite tracking studies of loggerhead sea turtles in- dicate that they are active in their oceanic movements. These turtles follow subtropical fronts as they travel toward Japan from east to west across the Pacific Ocean, often swimming against weak geostrophic currents (Polovina et al., 2000; Po- lovina et al., 2004). One hypothesis discussed in Polovina et al. (2000; 2004) suggests that this species ob- tains prey items from the subtropi- cal fronts along which they travel. A sharp gradient in surface chlorophyll is observed along the main frontal area where these turtles are com- monly encountered. This frontal area, the transition zone chlorophyll front Uchida, S., and H. Teruya. 1991. A) Transpacific migration of a tagged log- gerhead, Caretta caretta. B) Tag-return result of loggerhead released from Oki- nawa Islands, Japan. In International symposium on sea turtles '88 in Japan (I. Uchida, ed.), p. 169-182. Himeji City Aquarium, Tegarayama 440 Nishinobu- sue, Himeji-shi, Hyoyo 670, Japan. Parker et al.: Diet of Caretta caretta in the central North Pacific 143 60 N 40CN 20° N 0°N 20°N 40'N O less than 50 cm CCL • 50 cm CCL or greater 120°E 160°E 160W 120°W 80°W Figure 1 Distribution of loggerhead sea turtles {Caretta caretta) incidentally captured in the international high seas driftnet fishery in the central North Pacific Ocean. Turtles smaller than 50 cm curved carapace length (CCL) are shown as open diamonds and those larger than 50 cm CCL are shown as black circles. (TZCF), is an area of concentrated phytoplankton that also collects and attracts a variety of neustonic and oce- anic organisms — many of which may be potential prey times, as well as predators, of oceanic-stage loggerhead sea turtles in the Pacific. Polovina et al. (2000, 2004) have suggested that the turtles are foraging along the TZCF. The duration of the juvenile oceanic stage for logger- head sea turtles in the Pacific is currently unknown. In the Atlantic, juvenile turtles inhabit the oceanic zone for approximately 10 years (Bjorndal et al., 2000). Based on growth analyses (Zug et al., 1995; Chaloupka, 1998), it is probable that this sea turtle from the Pacific can have a similar extended oceanic stage, which in some cases may last until sexual maturity (30+ years). Understanding the diets of sea turtles is important for their conservation. Foraging studies have been done with oceanic-stage turtles in the Atlantic (Van Nierop and den Hartog, 1984). However, there is a paucity of information regarding the foraging ecology of oceanic- stage loggerhead sea turtles in the Pacific. Such infor- mation can help identify important food resources and foraging areas necessary for guiding decisions regarding the management of endangered sea turtle populations (Bjorndal, 1999). The objective of the present study is to determine the diet composition of loggerhead sea turtles from the central North Pacific Ocean and to discuss the possibility of interactions between these turtles and commercial fisheries that may occur as a result of the foraging behavior of these sea turtles. Method National Marine Fisheries Service (NMFS) observers between 1990 and 1992 obtained 52 dead loggerhead sea turtles. These specimens were taken as bycatch in the international high-seas driftnet fishery, which targeted squid and albacore (Wetherall et al., 1993). NMFS observers recorded capture position and sea sur- face temperature aboard commercial driftnet vessels. Samples were collected between latitude 29.5°N and 43°N and longitude 150°E and 154°W (Fig. 1). A total of 10 whole specimens and 42 excised stomachs were frozen and transported to a Honolulu laboratory for analysis. Stomachs were removed from whole specimens and all stomachs were examined from anterior to posterior. Gross observations of stomach contents were made and the contents were sorted to the lowest identifiable taxo- nomic level by using a dissecting microscope. Major fauna were identified, quantified by volume, and the percent contribution (to stomach contents) of each major organ- ism was calculated (Forbes, 1999). Presence of jellyfish or other jellies were identified by presence of tentacles, nematocysts, and whole or partial individuals. Planes spp. were identified from descriptions of Spivak and Bas (1999). Frequency of occurrence of major components was calculated by dividing the number of stomachs in which the prey item occurred by the total number of turtle stomachs examined. Percent sample volume was calcu- lated for all prey items by summing the total volume of each prey item and dividing it by the total volume of all 144 Fishery Bulletin 103(1) prey collected. Summing the total volume of each prey item and dividing it by the total stomach volume for those samples, where the prey item was present, yielded the mean percent volume. Regression analysis was done to determine if any correlation existed between sea surface temperature, sample volume, and size of turtle. Results Loggerhead sea turtles collected in our study were found widely distributed over the central North Pacific Ocean and there was no apparent difference in distribution 16 14 12 10-19 cm 20-29 cm 30-39 cm 40-49 cm 50-59 cm Curved carapace length (cm) 60-69 cm 70-79 cm Figure 2 Size distribution for the 52 loggerhead sea turtles [Caretta caretta) obtained as samples in the high-seas driftnet fishery. Sizes were grouped into 10-cm size classes. 21 0 20.0 19.0 180 17.0 16.0 15.0 • • • I • •* * • • • * '• • • • ym ••• 0.0 10.0 20.0 30.0 40.0 50 0 Curved carapace length (cm) 60.0 700 Figure 3 Relationship between curved carapace length (CCL, cm) of loggerhead sea turtles [Caretta caretta) and sea surface temperature iSST, n = 52). between size classes (Fig. 1). The turtle specimens ranged from 13.5 cm to 74.0 cm curved carapace length (CCL, Fig. 2); the mean was 44.8 [±14.5] cm CCL. Figure 2 shows the distribution of turtles in each 10-cm size class. Sea surface temperatures in the area of cap- ture ranged from 16° to 20°C. There was no correlation between size of turtle and sea surface temperature in the area of capture (F=0.58, r2=0.01, Fig. 3). All 52 stomachs examined contained prey items; the level of fill varied from 6 mL to 1262 mL. Items found in the anterior portion of the stomach were the most identifiable and contents varied between turtles. Un- identifiable remains were located mainly in the poste- rior end of the stomach or the intes- tines if a whole gastrointestinal tract was analyzed. Only one of the samples analyzed included an entire gastroin- testinal tract. A taxonomic listing of diet items identified for the loggerhead sea turtles of the central North Pacific is shown in Table 1 along with frequency of occur- rence and mean percent sample volume of each prey item. The six most com- mon (frequent) prey items were iden- tified. These included Janthina spp., which occurred in 75% of samples, and Planes spp., which occurred in 56% of samples. Lepas spp. occurred in 52% of the samples, and Carinaria cithara was found in 50% of samples. Velella velella, was found in 25% of the sam- ples, and pyrosomas were found in 21% of samples (Table 1). Other common food items found in stomachs were fish eggs (25% of stomachs), salps, amphi- pods (46% of stomachs), small fish, and plastic items (35% of stomachs. Table 1). Some plastic items included small plastic beads, thin plastic sheets, poly- propylene line, and even a small plastic fish, which had been an individual soy sauce container. Although Velella, py- rosomas, and salps were represented as prey items in our samples, other types of jellies may not have been well represented because their soft bodies may dissolve more quickly in stomach acids. It is also possible that unidenti- fied jellies may comprise the unidenti- fied remains, which occurred in 71% of stomachs and comprised 13.8% of total sample volume; however, a por- tion of the unidentified remains were likely masticated portions of identified prey items. Table 2 shows the mean percent prey item volumes for the six most common prey items. The six most common prey items can be ranked from largest to smallest mean volumes in 80.0 Parker et al .: Diet of Caretta caretta in the central North Pacific 145 Table 1 Percent occurrence and percentage of total sample volume (volume of prey for all stomachs combined) for prey items (listed to lowest taxonomic order) found in loggerhead sea turtles {Caretta caretta, n = 52 turtles). Occurrence Percent volume Prey group (%) (%) Carinaria eithara Benson 1835 50.0 43.8 Janthina spp. (includes J.janthina and J. prolongata = J. globosa) 75.0 14.4 Lepas spp. (includes L.anserifera Linnaeus 1767 and L.anatifera anatifera Linnaeus 1758) 51.9 6.7 Velella velella Linneaus 1758 (by-the-wind-sailor) 25.0 10.6 Planes spp. Dana 1852 55.8 1.2 Pyrosoma spp. 21.0 3.4 Fish eggs iHirundicthys speculiger and unidentified spp.) 25.0 1.9 Cephalopoda (squid and octopus fragments and paralarvae) 21.2 0.5 Debris (plastic, styrofoam, paper, rubber, polypropylene, etc.) 34.6 0.3 Debris (wood, bird feathers) 11.5 <0.1 Salpidae 13.5 0.5 Family Sternoptychidae (hatchetfish) 7.7 0.1 Electrona sp. — Myctophidae 1.9 0.1 Gammaridea and Hyperiidea amphipods 46.2 <0.1 Thecosomate pteropods 13.5 <0.1 Cavolinia globulosa (Gray 1850) 11.5 <0.1 POLYCHAETA (polychaete worms)— Alciopidae 5.8 <0.1 ISOPODA 3.8 <0.1 MYSIDACEA— mysid 3.8 <0.1 Creseis sp. 1.9 <0.1 PHAEOPHYTA (brown algae )—Cystoseira sp. 1.9 <0.1 EUPHAUSIACEA— euphausiid 1.9 <0.1 Unidentified tunicate spp. 13.5 1.0 Unidentified jellies 13.5 0.5 Unidentified crustaceans 5.8 0.5 Unidentified remains 71.2 13.8 the following order: 1) Carinaria eithara, 2) Pyrosoma spp., 3) Janthina spp., 4) Velella velella, 5) Lepas spp., and 6) Planes spp. Mean sample volume was 370.2 [±319.4] mL. Size of loggerhead sea turtles did not influence the volume of prey items for turtle sizes 35-70+ cm (F=0.11, r2=0.05). However, the smaller turtles did have smaller volumes of prey items present in their stomachs, because all turtles 13-34 cm had less than 80 mL total stomach volume (Fig. 4). The size of the turtle did not appear to be a factor in the type of prey ingested. The one excep- tion may be Velella velella. Turtles smaller than 30 cm CCL in our sample did not ingest this prey item, albeit sample size for less than 30-cm turtles was relatively small compared to the number of 40- and 50-cm size class turtles (Fig. 2); therefore, this apparent trend may not be the case for the general population. Of the six most common prey items, Carinaria ei- thara had the highest percent sample volume, 43.8% of total sample volume. In general, percent volumes of C. eithara were high; 20 of the 27 turtle stomachs Table 2 Mean percent volume and percent volume ranges for the six most frequently observed prey items found in driftnet captured loggerhead sea turtles (Caretta caretta) Mean Standard percent deviation Prey item volume (±%) Range Janthina spp. 30.7% 34.8% 1-97% Carinaria eithara 52.8% 33.1% 1-98% Lepas spp. 19.1% 24.7% 1-99% Velella velella 22.7% 29.4% 1-84% Planes spp. 5.6% 10.1% 1-38% Pyrosoma spp. 44.7% 33.7% 1-88% had percent volumes greater than 30% with this prey item and a number of stomachs had percent volumes greater than 90%. Janthina spp. had the next highest 146 Fishery Bulletin 103(1) percent sample volume at 14.4%. The percent volume of Janthina was generally high; 15 of 37 turtle stomachs had greater than 30% volume of this species. Only 4 of the 13 stomachs with Velella velella had greater than 30% sample volume; yet Velella made up almost 11% of total sample volume, and one of the stomach samples was almost entirely filled (84% volume) with Velella prey. In the samples that contained pyrosomas, this prey item often comprised a high percent of the total gut content — up to 88% stomach volume — and 7 out of 11 stomachs had greater than 30% stomach volume of pyrosomas. Planes spp. comprised more than 30% of stomach volume in only 2 of the 29 stomachs contain- ing this species. Lepas spp. often occurred in very high percent volumes (up to 99% of total gut content in one sample), although only 6 of 21 stomachs had percent volumes greater than 30% for Lepas. Discussion Prey items Loggerhead sea turtles in North Pacific oceanic habi- tats are opportunistic feeders that ingest items floating at or near the surface. Availability of prey in the oce- anic realm is generally characterized as patchy. This means that the majority of the ocean contains little to no forage, but in some areas high densities of prey can be found. This unpredictability of prey availability likely contributes to the opportunistic feeding behavior of the loggerhead sea turtle. The TZCF, an area of convergence created within the subtropical frontal zone by cooler denser water masses converging and sinking below warmer lighter water masses (Roden, 1991), may serve to help concentrate different prey items. Prey items such as Velella can often concentrate in large numbers in such areas (Evans, 1986). All size classes of this sea turtle 1400-, f 1200 • sz • S 1000- • E • o to 800 ■ g 600- Q. • . • • • ° 400- • • • E id o 200- ;• : • . • • • • • • 0 -I 1 w ■ 1 1 1 1 1 1 1 00 10.0 200 30.0 40.0 50.0 60.0 70.0 80.0 Curved carapace length (cm) Figure 4 Relationship between curved carapace length (CCL, cm) of loggerheads (Caretta caretta) and stomach volume IraL, n = 52) collected in our study were found between 16° and 21°C (Fig. 3), which typically are the temperatures that define the subtropical frontal zone and TZCF (Roden, 1991). Eighty-three percent of prey items that were recorded were found floating on the surface or were found on floating objects and would also likely be concentrated at convergent fronts such as the TZCF, driven there by the currents and winds (Polovina, et al., 2000; Polovina et al., 2004). It is suggested that this concentration of prey, along the convergent fronts, may be aggregating the loggerhead sea turtles traveling along this area, which are likely foraging on the increased densities of prey (Polovina et al., 2003a). Turtles in our study smaller than 30-cm CCL had very low volumes of prey in their stomachs. It is unknown whether the paucity of prey items in these turtle stomachs was related to the individual's size, e.g. they were physically not able to capture or ingest certain types of prey items, or perhaps to a lack of experience in foraging due to youth, given that turtles in this size range were determined to be between 1 and 4 years of age by Zug et al. (1995), or to other mitigating factors. Another indication that loggerhead sea turtles are opportunistic feeders is the presence of oceanic, me- sopelagic fish as prey items. The total number of fish (lanternfish and hatchetfish) in the samples was low (only 0.1 % of total stomach volume). These species of fish tend to stay below the photic zone usually at depths greater than 300 m during the day and migrate up near the surface at night. Lanternfish make diel vertical mi- grations where they reach maximum densities at 100 m at night. During nightly movements some species can also come directly to the surface (Hulley, 1990). Some species of hatchetfish also make diel vertical migrations, which would bring them to within 100 m of the surface at night (Weitzman, 1986; Froese and Pauly, 2003). Because of the low numbers, it is likely that loggerhead sea turtles ingest only dead or debilitated fish rather than actively hunt and chase such spe- cies. The presence of these species also indicates that the turtles may be feeding at night when they would be more likely to encounter the fish during their diel movement. Another prey item exhibiting diel vertical migration is the pyrosomas. Pyrosomas, which are a part of Pacific leatherback sea turtle diets (Davenport and Balazs, 1991), were also present in loggerhead sea turtle stomach samples. Pyrosomas are colonial tunicates com- prising individual zooids embedded in the walls of a gelatinous tube. These colonies can become quite large (some greater that 4 meters in length) and tend to drift with ocean currents and accumulate along frontal zones which make them accessible to the sea turtle that forages opportunistically. At least one species (P. atlanticum) has been re- corded to stay below 300 m during the Parker et al : Diet of Caretta caretta in the central North Pacific 147 day and move up near the surface at night (Andersen and Sardou, 1994); this activity again may indicate ac- tive night foraging by the loggerhead sea turtle. Loggerhead sea turtles may feed by swallowing float- ing prey whole and also by biting whole prey (or por- tions off a whole prey) found on large floating objects. A commonly ingested prey item, Velella velella, known as "by-the-wind-sailor" (Eldredge and Devaney, 1977), typi- cally was found intact. Janthina spp.. predatory gastro- pods whose main prey item is Velella velella, were also frequently found whole in stomachs. Small Janthina spp. have been observed directly on Velella, and it has been hypothesized that Janthina use Velella to settle on and use the Velella as floatation until they become too large for the host (Bayer, 1963). This behavior may be a reason why whole Janthina and Velella were often found together in stomach samples. Janthina spp. had been previously noted as a prey item of loggerhead sea turtles in the Azores and South Africa (Dodd, 1988) but was first identified as a prey item in the Pacific Ocean in a preliminary unpublished report by Cooke in 19922 — data that are included in the present study. The high frequency of occurrence of Velella velella and whole Janthina spp. support the hypothesis that loggerhead sea turtles will feed on the surface, swallowing their prey whole. Distribution of Velella velella is patchy; den- sities range from <1/1000 m3 to 1000/1000 m3 and den- sities of Janthina spp. are considerably less than those of Velella. When optimum combinations of prevailing winds and currents converge, densities of Velella velella have been observed to be in concentrations upward of 10,000/1000 m'-, forming patches so large and dense they have been likened to oil tanker sludge by mariners (Evans, 1986; Parker, personal observ.). It is possible that the one turtle that had a stomach volume of 84% Velella found one of these patches on which to feed. Velella velella was the one common prey item that was not found in stomachs of turtles less than 30-cm CCL. Because Velella were commonly swallowed whole, it is possible that an average size Velella, which range from 5 to 10 cm (Evans, 1986), might have been too large for a 13-29 cm CCL turtle to swallow whole. The epibiotic oceanic crabs and the gooseneck bar- nacles (Lepas spp.) usually occur on floating objects; Planes sometimes even rides on Velella (Chace, 1951). Planes spp. also have been observed and collected from the tail area of loggerhead sea turtle themselves (Dav- enport, 1994; NMFS observers3). Although approximate- ly 80% of stomach samples with Planes spp. contained whole crabs, which were identified as P. cyaneus, there 2 Cooke, W. J. 1992. A taxonomic analysis of stomach contents from loggerhead turtles (Caretta caretta ). AECOS report no. 697, 12 p. Prepared for NOAA, NMFS, Honolulu Laboratory, 2570 Dole Street, Honolulu, Hawaii 96822. (Available from AECOS, Inc., 45-939 Kamehameha Hwy., Rm. 104, Kaneohe, Hawaii 96744.] :) NMFS (National Marine Fisheries Service) observers. 1997- 2000. Personal commun. Pacific Islands Fisheries Science Center. 2570 Dole Street, Honolulu, HI 96822-2396. were also numerous masticated crabs and pieces of crabs. These pieces could have been P. marinus because whole specimens are necessary to identify Planes spp. (Spivak and Bas, 1999); therefore the lowest taxonomic identification for this study was limited to Planes spp. Densities of Planes spp. and Lepas spp. are not well documented but are likely limited by the amount of substrate on which they can settle or on the amount of floating objects available. Natural drifting objects such as tree logs or pumice from volcanic eruptions have been documented since the nineteenth century (Kew, 1893, cited in Jokiel, 1990). The "floating islands," as they have been called, continue to be important for transporting organisms, from corals to reef fish across the oceans (Jokiel, 1990). Man-made objects also sup- ply substrate and habitat on which different organisms can settle. Buoys and logs that wash ashore often have Lepas spp. attached to them, some with Lepas spp. cov- ering 100% of the area that was underwater (Parker, personal observ.). Although the frequency of occurrence of Planes spp. in stomach samples was high, the percent sample volume of Planes was relatively low (1.2% total volume) and the mean volume of Planes found was also low (5.6%, Table 2), indicating that this prey was either taken opportunistically or accidentally. It is not known whether the Planes were ingested along with other prey items or were actually grazed from larger floating objects. In contrast, Lepas spp. often occurred in very high percent volumes, indicating that the turtles were actively grazing these prey. The constant presence of Lepas spp. in samples strongly supports the hypothesis that loggerhead sea turtles feed not only by swallowing prey whole, but also by biting prey off larger floating objects. Small chunks of Styrofoam were still attached to the bases of some Lepas specimens indicating that the turtle had bitten off some of the floating object itself while grazing on prey found on the floating debris. Among other floating items that often occurred in the turtles' stomachs, one common element was fish eggs. Some of these fish eggs were identified as Hirundicthys speculiger or flying fish eggs. Amphipods were another common item but comprised a very small fraction of total gut content (<1%), indicating that they were not a targeted prey item. Amphipods were possibly ingested incidentally as epiphytes on other items or as part of the gut contents of other prey items. The proportion of man-made drift debris in our sample was low in contrast to prior studies (Balazs, 1985; Allen, 1992; Bjorndal et al., 1994; Kamezaki, 1994; Tomas et al., 2002). Plastics and other man-made debris were com- monly found, occurring in about 35% of stomachs, but they comprised a very small fraction of the total gut content (<1%). Loggerhead sea turtles also actively forage at deeper depths if high densities of prey items are present. An initial study of pelagic dive behavior of this species (Polovina et al., 2003) indicates that they regularly dive down to depths of 100 m and may also forage at those depths, which may account for the high fre- quency of occurrence and high total percent volume of 148 Fishery Bulletin 103(1) the heteropod Carinaria cithara. Okutani (1961) first recorded sea turtles consuming Carinaria (including Carinaria cithara, Benson 1835), in the western North Pacific. Heteropods are found in the upper photic zone (within 100 m of the surface) but are not typically a neustonic or floating species. Recorded heteropod densities in the Pacific are variable (<1/1000 m3 to 150/1000 m3, Seapy, 1974, cited in Lalli and Gilmer, 1989). Although these densities seem very low, it is clear that in this area of the central North Pacific heteropods are numerous enough within diving depths of loggerhead sea turtles to make this an attractive prey item for the turtles. Conclusion — Interactions with fisheries The bycatch of nontargeted species in different fisher- ies has been an issue for many years (Wetherall et al., 1993; Wetherall, 1996; Gardner and Nichols, 2001; Suganuma4). Bycatch of sea turtles has also been an issue for the conservation management of most sea turtle species. Sea turtle mortalities have occurred in nearly all fisheries (gillnet, driftnet, trawl, and long- line). During their transpacific migrations loggerhead sea turtles move through areas of multinational long- line fishing (Lewison et al., 2004). Mortalities of sea turtles after longline fishery interactions have been estimated between 28% and 50% by both U.S. and Japa- nese researchers (Nishemura and Nakahigashi, 1990; Kleiber,5 McCracken'M and loggerhead sea turtles com- prise a large percentage of the sea turtle interactions in longline fisheries, as high as 59% of sea turtles cap- tured in the Hawaii-based longline fleet. The longline fishery as well as various other fisheries in the Pacific (Gardner and Nichols, 2001) have been implicated as part of the reason for recent declines in the loggerhead sea turtle populations both in Japan (Kamezaki and Matsui, 1997; Sato et al., 1997; Suganuma4) and also in Australia, and southern nesting areas (Limpus and Couper, 1994; Limpus and Reimer7). Research on feed- ing behavior may help with the mitigation of fisheries interactions. 4 Suganuma, H. 2002. Population trends and mortality of Japanese loggerhead turtles, Caretta caretta, in Japan. In Proc. Western Pacific Sea Turtle Coop. Res. and Mgmt. Workshop (I. Kinan, ed.). p. 74-77. Western Pacific Regional Fishery Management Council, 1164 Bishop Street, Suite 1400, Honolulu, HI 96813. 5 Kleiber, P. 1998. Estimating annual takes and kills of sea turtles by the Hawaiian longline fishery, 1991-1997, from observer program and logbook data. Administrative report H-98-08, 21 p. Southwest Fisheries Science Center, Nat. Mar. Fish. Serv., NOAA, 2570 Dole St., Honolulu, HI 96822. 6 McCracken, M. L. 2000. Estimation of sea turtle take and mortality in the Hawaiian longline fisheries. Administrative report H-00-06, 29 p. Southwest Fisheries Science Center, Nat. Mar. Fish. Serv., NOAA, 2570 Dole St., Honolulu. HI 96822. Learning more about the life history of loggerhead sea turtles and understanding more about the move- ments, foraging behavior, and prey of these turtles are important for making well-informed management decisions because foraging behavior may change as seasons change and as these turtles move through dif- ferent habitats (Bjorndal, 1997). Although our study indicates that these turtles forage mainly on floating or near-surface prey in the open ocean, studies in dif- ferent areas show different feeding habits. The oceanic, near-surface feeding behavior of loggerhead sea turtles is likely one reason for the numerous longline fishery interactions in the central North Pacific. The recorded dive data for these turtles indicate that they spend a large percentage of their time near the surface — as much as 78% of their time is spent within 10 m of the surface (Polovina et al., 2003b). Juvenile loggerhead sea turtles are rarely found in the waters adjacent to Japan (Uchida, 1973); the juvenile turtles are thought to use the Kuroshiro Current to move out into the Pacific and the southern edge of the Subartic Gyre during their eastward movement toward foraging grounds in the Eastern Pacific (Bowen et al., 1995). In the Atlantic, however, small neonate loggerhead sea turtles have been found associated with drifts of floating material, especially Sargassum rafts (Witherington, 2002), and although large, regular drifts of floating material are rare in the Pacific, small loggerhead sea turtles may also be associated with floatsam (Pitman, 1990). Studies have indicated that foraging changes through- out the lifecycle of loggerhead sea turtles (van Nierop and den Hartog, 1984; Plotkin et al., 1993; Godley et al., 1997; Tomas et al., 2001). In the Pacific, oceanic immature turtles (present study) forage on different prey from that foraged by subadults in the pelagic and neritic areas off Baja California (Nichols et al., 2000; Peckham and Nichols, 2003; Seminoff et al., 2004), and adults in benthic neritic habitats, in turn, forage on different prey near Japan and China (Hitase et al., 2002). Japanese loggerhead sea turtles foraging in the Eastern Pacific target Pleuroncodes planipes, the pelagic red crab, which occurs year round off Baja California. These turtles interact with the artisanal fisheries in the area which are both pelagic and benthic fisheries (Gomez-Gutierrez and Sanchez-Ortiz, 1997; Bartlett, 1998; Gomez-Gutierrez et al., 2000; Peckham and Nich- ols, 2003). Loggerhead sea turtles have also been found on the Gulf of California side of Baja California, likely foraging on the large abundance of invertebrate fauna found there (Brusca, 1980), and these turtles face fish- ing pressure from the artisanal gillnet fishery in this area (Seminoff et al., 2004). 7 Limpus, C. J., and D. Reimer. 1994. The loggerhead turtle, Caretta caretta, in Queensland: a population in decline. In Proceedings of the Australian marine turtle conservation workshop iR. James, compiler), p. 39-59. Queensland Dep. Environ and Heritage and Aust. Nat. Conserv. Agency, GPO Box 787, Canberra ACT 2601, Australia. Parker et al .: Diet of Caretto caretta in the central North Pacific 149 Converting CCL to straight carapace length (SCL; using the conversion equation: CCL=1.388+(1.053) SCL, in Bjorndal et al., 2000), size classes found in our study ranged from 11.5 cm to 68.9 cm SCL with a mean of 41.2 [±12.4] cm SCL. The East Pacific recruits were slightly larger with means of 46.9-61.9 cm SCL (Semi- noff et al., 2004). Most of these turtles were immature to subadult turtles, and only a few were adult-size tur- tles. According to Zug et al. (1995), the loggerhead sea turtles recruiting to the nearshore and neritic habitats of Baja California are likely 10 years of age or older, indicating that these turtles might spend as many as 10 years before arriving at their East Pacific foraging habitat. After returning to the West Pacific, satellite telemetry has found that adult loggerhead sea turtles also reside in both neritic and pelagic habitats (Baba et al., 1992, 1993; Kamezaki et al., 1997; Sakamoto et al., 1997) putting them at risk of interaction with nearshore gillnet fisheries as well as pelagic longline fisheries. Hitase et al. (2002) found a size difference between adults in neritic and oceanic habitats — the postnesting females that chose oceanic habitats were smaller (mainly <80.0 cm) than those that used neritic habitats for postnesting foraging — and also suggested that some adult turtles may not recruit to neritic areas near Japan and China. This may be evidence that some loggerhead sea turtles remain in the oceanic habitat their whole life cycle, returning nearshore only to mate or nest. In the Atlantic, juveniles as well as adults of this species can be found in neritic foraging habitats of the Gulf of Mexico, and these turtles can have interac- tions with coastal trawl and other coastal fisheries in the area (Plotkin et al., 1993). Juvenile turtles have also been observed and captured in areas along the eastern coast of the United States where they have been found feeding on benthic invertebrates (Burke et al., 1990; Epperly et al., 1990). Very small, neonate loggerhead sea turtles have been found associating with and foraging in Sargassum drifts while they are transported by the Gulf Stream into the mid-Atlantic (Witherington, 2002); therefore, the harvest of Sargas- sum or trawling through this area would affect these juveniles. There is some evidence that juvenile Atlantic loggerhead sea turtles may move between coastal and pelagic forage habitats, which would expose them to both coastal and pelagic fisheries (Witzell, 2002). In the Mediterranean, both juvenile and adult loggerhead sea turtles also have variety of foraging behaviors. In the eastern Mediterranean, postpelagic juveniles and adults forage mainly in neritic habitats on benthic prey items where they would interact with coastal trawl and other artesianal fisheries (Godley et al., 1997). In the western Mediterranean, juvenile turtles of this species forage in both pelagic as well as neritic habitats, where they are at risk of fishery interactions in many differ- ent fisheries including longline, trawling, and coastal fisheries (Tomas et al., 2001). Postpelagic juveniles in the Mediterranean may be recruits from the Atlantic Ocean or may come from the endemic Mediterranean population. Adult loggerhead sea turtles have been noted to also move between the eastern and western basins of the Mediterranean in response to seasonal temperature changes (Bentivegna, 2002). During this migration between two benthic feeding areas, some of the turtles would spend extensive amounts of time in the pelagic habitat likely foraging on pelagic prey items. This intra-Mediterranean movement puts these turtles at risk of interactions with a multinational fishery contingent of pelagic as well as coastal fisheries (Bentivegna, 2002). One possible way to mitigate increased fisheries in- teractions in the Pacific and other areas might be to identify specific loggerhead foraging areas for protec- tion, such as the area around Baja California, Mexico. In the central North Pacific, our study (Fig. 1), as well as recent satellite tracking studies of juvenile and adult loggerhead sea turtles (Hitase et al., 2002; Parker et al., 2003; Polovina et al., 2004), has indicated that the area west of and around the Emperor seamounts, between 160° and 180°E might also be an important foraging habitat. Most of the turtles in our study were collected from this area (Fig. 1) and one juvenile spent 10 months west of the Emperor Seamounts, between 160° and 170°E, before its satellite transmitter stopped transmitting data (Parker et al., 2003). In this area, the southern edge of the Kuroshiro Extension Current forms numerous eddies that are semipermanent fea- tures throughout the year. Reduction of fishing effort or other fishery mitigation techniques in this area may greatly decrease the number of fisheries interactions that Pacific loggerhead sea turtles experience. Interna- tional cooperation is needed in order to manage these foraging habitats. More studies also need to be done on the ecology of these turtles so that fishery interac- tions at all life stages can be addressed and so that a total picture of the life history of this species can be obtained. Acknowledgments We would like to acknowledge the hard work of all the NMFS fishery observers for obtaining the samples, Russ Miya and Bryan Winton for their help in the initial sorting, and Mike Seki and Kevin Landgraf for their help in identifying prey items. We would also like to acknowledge the review and comments of Alan Bolten, Jeffrey Seminoff, George Antonelis, Jerry Wetherall, Colin Limpus, Judith Kendig, Francine Fiust, Shawn Murakawa, and two anonymous reviewers. Literature cited Allen, W. 1992. Loggerhead dies after ingesting marine debris. Mar. Turtle Newsl. 58:10. Andersen, V., and J. Sardou. 1994. Pyrosoma atlanticum (Tunicata, Thaliacea): diel migration and vertical distribution as a function of colony size. J. Plankton Res. 16(4):337-349. 150 Fishery Bulletin 103(1) Baba, N, M. Kiyota, H. Suganuma, and H. Tachikawa. 1992. Research on migratory routes of loggerhead turtles and green turtles by the Argos system. Report on com- missioned project for data analysis by scientific observ- ers aboard fishing vessels in 1991. Fish. Agency Jpn., Tokyo, p. 89-99. [In Japanese.] 1993. Research on migratory routes of loggerhead turtles and green turtles by the Argos system. Report on com- missioned project for data analysis by scientific observers aboard fishing vessels in 1992. Fish. Agency of Jpn, Tokyo, p. 86-99. [In Japanese.] Balazs, G. 1989. New initiatives to study sea turtles in the eastern Pacific. Mar. Turtle Newsl. 47:19-21. 1985. Impact of ocean debris on marine turtles: entangle- ment and ingestion. In R.S. Shomura and H.O. Yoshida (eds.), Proceedings of the workshop on the fate and impact of marine debris, 26-29 November 1984, Hono- lulu, Hawaii, p. 387-429. NOAA Tech. Memo. NMFS, NOAA-TM-SWFC-54. Bartlett, G. 1989. Juvenile Caretta off Pacific coast of Baja Califor- nia. Not. Caguamas 2:2-10. Bayer, F. M. 1963. Observations on pelagic mollusks associated with siphonophores Velella and Physalia. B. Mar. Sci. 13(3):454-466. Bentivegna, F. 2002. Intra-Mediterranean migrations of loggerhead sea turtles (Caretta caretta) monitored by satellite telemetry. Mar. Biol. 141: 795-800. Bjorndal, K. A. 1997. Foraging ecology and nutrition of sea turtles. In The biology of sea turtles (P. Lutz and J. Musick, eds.), p. 199-32. CRC Press, Boca Raton, FL. 1999. Priorities for research in foraging habitats. In Research and management techniques for the conserva- tion of sea turtles (K. L. Eckert, K. A. Bjorndal, F. A. Abreu-Grobois, and M. Donnely, eds.), p. 12-18. IUCN (International Union for the Conservation of Nature and Natural Resources)/SSC (Species Survival Com- mision), Mar. Turtle Spec. Group Publ. No. 4, Wash- ington, DC. Bjorndal, K. A., A. B. Bolten, and C. J. Lagueux. 1994. Ingestion of marine debris by juvenile sea tur- tles in coastal Florida habitats. Mar. Pollut. Bull. 28: 154-158. Bjorndal, K. A., A. B. Bolten, and H. R. Martins. 2000. Somatic growth model of juvenile loggerhead sea turtles Caretta caretta: duration of pelagic stage. Mar. Ecol. Prog. Ser. 202: 265-272. Bowen, B. W., F A. Abreu-Grobois, G. H. Balazs, N. Kamezaki, C. J. Limpus and R. J. Ferl. 1995. Trans-Pacific migrations of the loggerhead turtle (Caretta caretta) demonstrated with mitochondrial DNA markers. Proc. Natl. Acad. Sci. 92: 3731-3734. Brusca, R. C. 1980. Intertidal invertebrates of the Gulf of California, 513 p. Univ. Arizona Press, Tucson, AZ. Burke, V. J., S. J. Morreale, and E. A. Standora. 1990. Comparisons of diet and growth of Kemp's Ridley and loggerhead turtles from the Northeastern U.S. /;; Proceedings of the tenth annual workshop on sea turtle biology and conservation (T. H. Richardson. J. I. Rich- ardson, and M. Donnelly (comps.), p. 135. NOAA Tech. Memo NMFS-SEFC-278. Chace, F. A. 1951. The oceanic crabs of the genera Planes and Pachy- grapus. Proc. U.S. Natl. Mus. 101:65-103. Chaloupka, M. 1998. Polyphasic growth in pelagic loggerhead sea turtles. Copeia 1998(21:516-518. Davenport, J. 1994. A cleaning association between the oceanic crab (Planes minutus) and the loggerhead sea turtle (Caretta caretta). J. Mar. Biol. Assoc. UK 74(3):735-737. Davenport, J., and G. H. Balazs. 1991. Fiery Bodies: Are pyrosomas an important com- ponent of the diet of leatherback turtles. Brit. Herpt. Soc. 37:33-38. 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(14), 110 p. Dutton, P. H., G H. Balazs, and A. E. Dizon. 1998. Genetic stock identification of sea turtles caught in the Hawaii-based pelagic longline fishery. In Proceed- ings of the seventeenth annual sea turtle symposium (S. P. Epperly and J. Braun, compilers), p. 43. NOAA Tech. Memo. NMFS-SEFSC-415. Eldredge, L. G, and D. M. Devaney. 1977. Other hydrozoans. In Reef and shore fauna of Hawaii, Seel: Protozoa through Ctenophora. B. P. Bishop Mus. Spec. Publ. 64(11:105-107. Epperly, S. P., J. Braun, and A. Veishlow. 1990. Distribution and species composition of sea turtles in North Carolina. In Proceedings of the tenth annual workshop on sea turtle biology and conservation (T. H. Richardson. J. I. Richardson, and M. Donnelly (comps.), p. 95-96. NOAA Tech. Memo NMFS-SEFC-278. Evans, F. 1986. Velella velella (L.), the 'by-the-wind-sailor,' in the North Pacific Ocean in 1985. Mar. Obs. 56(7): 196- 200. Forbes, G A. 1999. Diet sampling and diet component analysis. In Research and management techniques for the conserva- tion of sea turtles (K. L. Eckert, K. A. Bjorndal, F A. Abreu- Grobois, and M. Donnelly, eds.), p. 144-148. IUCN/SSC Mar. Turtle Spec. Group Publ. no. 4., Washington, D.C. Frazier, J. 1985. Misidentification of sea turtles in the East Pacific: Caretta caretta and Lepidnchelys olivacea. J. Herpetol. 1:1-11. Froese, R., and D. Pauly, eds. 2003. FishBase. World Wide Web electronic publication. www.fishbase.org. [Accessed 6 March 2004.] Gardner, S. C, and W. J. Nichols. 2001. Assessment of sea turtle mortality rates in the Bahia Magdalena region, Baja, California Sur, Mexico. Chel. Cons. Biol. 4:197-199. Godley, B. J., S. M. Smith, P. F. Clark, and J. D. Taylor. 1997. Molluscan and crustacean items in the diet of the loggerhead turtle, Caretta caretta (Linnaeus, 1758) ITes- tudines: Chelonidael in the eastern Mediterranean. J. Molluscan Stud. 63:474-476. Gomez-Gutierrez, J., E. Dominguez-Hernandez, C. J. Robinson, and V. Arenas. 2000. Hydroacoustical evidence of autumn inshore resi- dence of the pelagic red crab Pleuroncodes planipes at Punta Eugenia, Baja California, Mexico. Mar. Ecol. Prog. Ser. 208:283-291. Parker et al : Diet of Caretta caretta in the central North Pacific 151 Gomez-Gutierrez, J., and C. A. Sanchez-Ortiz. 1997. Larval drift and population structure of the pelagic phase of Pleuroncodes planipes (Stimpson) (Crustacea: Galatheidaei off the southwest coast of Baja California. Mexico. B. Mar. Sci. 61(21:305-325. Hitase, H., N. Takai, Y. Matsuzawa, W. Sakamoto, K. Omuta, K. Goto, N. Arai, and T. Fujiwara. 2002. Size-related differences in feeding habitat use of adult female loggerhead turtles Caretta caretta around Japan determined by stable isotope analyses and satellite telemetry. Mar. Ecol. Prog. Ser. 233: 273-281. Hulley, P. A. 1990. Myctophidae. In Fishes of the Southern Ocean (O. Gon and P. C. Heemstra, eds.), p. 146-178. J. L. B. Smith Institute of Ichthyology, Grahamstown, South Africa. [ref. 5182 from website FishBase: www.fishbase.org.] Jokiel, P. L. 1990. Long-distance dispersal by rafting: reemer- gence of an old hypothesis. Endeavour, New Series. 14(21:66-73. Kamezaki, N. 1994. Review of quantitative data on plastic debris found in the intestine (sic) of the sea turtle. LTmigame Newsl. 22: 9-14. [In Japanese.] Kamezaki, N., and M. Matsui. 1997. A review of biological studies on sea turtles in Japan. Jpn. J. Herpetol. 17(11:16-32. Kamezaki, N„ I. Miyawaki, H. Suganuma, K. Omuta, Y. Nakajima, K. Goto, K. Sato, Y. Matsuzawa, M. Samejima, M. Ishii, and T. Iwamoto. 1997. Post-nesting migration of Japanese logger- head turtle, Caretta caretta. Wildl. Conserv. Jpn. 3:29-39. fin Japanese with English abstract.] Lalli, C. M„ and R. W. Gilmer. 1989. Pelagic snails: the holoplanktonic gastropod mol- lusks, 259 p. Stanford LIniv. Press, Stanford, CA. Lewison, R. L., S. A. Freeman, and L. D. Crowder. 2004. Quantifying the effects of fisheries on threatened species: the impact of pelagic longlines on loggerhead and leatherback sea turtles. Ecol. Lett. 7(31:221-231. Limpus, C. J., and P. Couper. 1994. Loggerheads: a species in decline. Wildl. Aust. 30:11-13. Marquez, M. R., and A. O. Villanueva. 1982. Situacion actual y recomendaciones para el manejo de las tortugas marinas de la costa occidental de Mexico, en especial la tortuga golfina Lepidochelys olivacea. Cienc. Pesquera INP 2:83-91. Nichols, W. J., A. Resendiz, and C. Mayoral-Russeau. 2000. Biology and conservation of loggerhead turtles in Baja California, Mexico. In Proceedings of the nine- teenth annual symposium on sea turtle biology and conservation (H. J. Kalb and T. Wibbels, comps), p. 169-171. NOAA Tech. Memo. NMFS-SEFSC-443. Nishemura W., and S. Nakahigashi. 1990. Incidental capture of sea turtles by Japanese research and training vessels: results of a question- naire. Mar. Turtle Newsl. 51:1-4. Okutam, T. 1961. Notes on the genus Carinaria (Heteropoda) from Japanese and adjacent waters. Pubis. Seto Mar. Biol. Lab. 9:333-353. Parker, D. M., J. Polovina, G. H. Balazs, and E. Howell. 2003. The lost years: long-term movement of a maturing loggerhead turtle in the northern Pacific Ocean. In Pro- ceedings of the twenty-second annual symposium on sea turtle biology and conservation (J. A. Seminoff, comp.l, p. 294-296. NOAA Tech. Memo. NMFS-SEFSC-503. Peckham, H., and W. J. Nichols. 2003. Why did the turtle cross the ocean? Pelagic red crabs and loggerhead turtles along the Baja California coast, /n Proceedings of the twenty-second annual symposium on sea turtle biology and conservation (J. A. Seminoff, comp.l, p. 47. NOAA Tech. Memo. NMFS- SEFSC-503. Pitman, R. L. 1990. Pelagic distribution and biology of sea turtles in the Eastern Tropical Pacific. In Proceedings of the tenth annual workshop on sea turtle biology and con- servation (T. H. Richardson, J. I. Richardson, and M. Donnelly, comps. 1, p. 143-148. 1. NOAA Tech. Memo. NMFS-SEFC-278. Plotkin, P. T, M. K. Wicksten, and A. E. Amos. 1993. Feeding ecology of the loggerhead sea turtle Caretta caretta in the Northwestern Gulf of Mexico. Mar. Biol. 115:1-15. Polovina, J. J., G. H. Balazs, E. A. Howell, D. M. Parker, M. P. Seki, and P. H. Dutton. 2004. Forage and migration habitat of loggerhead (Caretta caretta) and olive ridley (Lepiodchelys olivacea) sea turtles in the central North Pacific Ocean. Fish. Oceanogr. 13:36-51. Polovina, J. J., E.A. Howell, D. M. Parker, and G. H. Balazs. 2003. Dive depth distribution of loggerhead (Caretta caretta) and olive ridley (Lepiodchelys olivacea) turtles in the central North Pacific Ocean: Might deep longline sets catch fewer turtles. Fish. Bull. 101: 189-193. Polovina, J. J., D. R. Kobayashi, D. M. Parker, M. P. Seki and G. H. Balazs. 2000. Turtles on the edge: movement of loggerhead tur- tles (Caretta caretta) along oceanic fronts spanning longline fishing grounds in the Central North Pacific, 1997-1998. Fish. Oceanogr. 9:71-82. Resendiz, A., B. Resendiz, W. J. Nichols, J. A. Seminoff, and N. Kamezaki. 1998. First confirmed east-west transpacific movement of a loggerhead sea turtle, Caretta caretta, released in Baja California, Mexico. Pac. Sci. 52(2):151-153. Roden, G. I. 1991. Subartic-subtropical transition zone of the North Pacific: large-scale aspects and mesoscale structure. In Biology, oceanography, and fisheries of the North Pacific Transition Zone and Subartic Frontal Zone (J. A. Wether- all, ed.), p. 1-38. NOAA Tech Rep. NMFS, SWFSC 105. Sakamoto, W., T. Bando, N. Arai, and N. Baba. 1997. Migration paths of adult female and male log- gerhead turtles, Caretta caretta, determined through satellite telemetry. Fisheries Sci. 63:547-552. Sato, K., T. Bando, Y. Matsuzawa, H. Tanaka, W. Sakamoto, S. Minamikawa, and K. Goto. 1997. Decline of the loggerhead turtle, Caretta caretta, nesting on Senri beach in Minabe, Wakayama, Japan. Chel. Conserv. Biol. 2:600-603. Seminoff, J. A., A. Resendiz, B. Resendiz, and W. J. Nichols. 2004. Occurrence of loggerhead sea turtles (Caretta caretta) in the Gulf of California, Mexico: evidence of life-history variation in the Pacific Ocean. Herpetol. Rev. 35(11:24-27. Spivak, E. D„ and C. C. Bas. 1999. First finding of the pelagic crab Planes marinus (Decapod: Grapsidae) in the Southwestern Atlantic. J. Crustacean Biol. 19(11:72-26. 152 Fishery Bulletin 103(1) Tomas, J., F. J. Aznar, and J. A. Raga. 2001. Feeding ecology of the loggerhead turtle Caretta caretta in the western Mediterranean. J. Zool. Lond. 255:525-532. Tomas, J., R. Guitart, R. Mateo, and J. A. Raga. 2002. Marine debris ingestion in loggerhead sea turtles, Caretta caretta, from the western Mediterranean. Mar. Poll. Bull. 44:221-216. Uchida, I. 1973. Pacific loggerhead turtle — pursuing its mysterious oceanic life. Anima 1(3):5-17. [In Japanese.] Weitzman, S. H. 1986. Sternoptychidae. In Smith's sea fishes (M. M. Smith and P. C. Heemstra, eds.), p. 253-259. Springer- Verlag, Berlin. Wetherall, J. A. 1996. Assessing impacts of Hawaiian longline fishing on Japanese loggerheads and Malaysian leatherbacks: some exploratory studies using TURTSIM. In Status of marine turtles in the Pacific Ocean relevant to inci- dental take in the Hawaii-based pelagic longline fishery (A. Bolten, J. A. Wetherall, G. H. Balazs, and S. G. Pooley, comps.), p. 57-75. NOAA Tech. Memo. NMFS- SWFSC-230. Wetherall, J. A., G. H. Balazs, R. A. Tokunaga, and M. Y. Y. Yong. 1993. Bycatch of marine turtles in North Pacific high- seas driftnet fishery and impacts on stock. In INPFC symposium on biology, distribution, and stock assess- ment of species caught in the high seas driftnet fishery in the North Pacific Ocean, 53(111) (J. Ito et al., eds.), p. 519-538. Int. N. Pac. Fish. Comm., Vancouver, Canada. Witherington, B. E. 2002. Ecology of neonate loggerhead turtles inhabiting lines of downwelling near a Gulf Stream front. Mar. Biol. 140:843-853. Witzell, W. N. 2002. Immature Atlantic loggerhead turtles (Caretta caretta): suggested changes to the life history model. Herpetol. Rev. 33:266-269. Van Nierop, M. M., and J. C. den Hartog. 1984. A study of the gut contents of five juvenile log- gerhead turtles, Caretta caretta (Linnaeus) (Reptilia Cheloniidae), from the south-Eastern part of the North Atlantic Ocean, with emphasis on coelenterate identification. Zool. Meded. Leiden 59:35-54. Zug, G. R., G. H. Balazs, and J. A. Wetherall. 1995. Growth in juvenile loggerhead sea turtles [Caretta caretta) in the North Pacific pelagic habitat. Copeia 2:484-487. 153 Abstract — Two examples of indirect validation are described for age-read- ing methods of Pacific cod [Gadus macrocephalus). Aging criteria that exclude faint translucent zones (checks) in counts of annuli and cri- teria that include faint zones were both tested. Otoliths from marked and recaptured fish were used to back-calculate the length of each fish at the time of its release by using measurements of the area of annuli. Estimated fish size at time of release and actual observed fish size were similar, supporting the assumption that translucent zones are laid down on an annual basis. A second method for validating read- ing criteria used otolith age and von Bertalanffy parameters, estimated from the tagging data, to predict how much each fish grew in length after tagging. We found that otolith aging criteria applied to otoliths from tagged and recovered Pacific cod pre- dicted quite accurately the growth increments that we observed in these specimens. These results provide fur- ther evidence that the current aging criteria are not underestimating the age of the fish and support our cur- rent interpretation of checks (i.e., as subannual marks). We expect these indirect validations to advance age determination for Pacific cod, which in turn would enhance development of stock assessment methods based on age structure for this species in the eastern Bering Sea. Indirect validation of the age-reading method for Pacific cod (Gadus macrocephalus) using otoliths from marked and recaptured fish Nancy E. Roberson Daniel K. Kimura National Marine Fisheries Service, NOAA Alaska Fisheries Science Center 7600 Sand Point Way, NE Seattle, Washington 98115 E-mail address (for N E Roberson): Nancy Roberson (a1 noaa gov Donald R. Gunderson University ol Washington School of Aquatic and Fishery Sciences 1122 NE. Boat Street Seattle, Washington 98105 Allen M. Shimada National Marine Fisheries Service, NOAA Office of Research 1315 East- West Hwy Silver Spring, Maryland 20910 3282 Manuscript submitted 7 November 2002 to the Scientific Editor's Office. Manuscript approved for publication 20 September 2004 by the Scientific Editor. Fish. Bull. 103:153-160 (2005). Pacific cod (Gadus macrocephalus Tilesius, 1810) is an important spe- cies in eastern Bering Sea commercial fisheries and is second only to walleye pollock (Theragra chalcogramma) in landings (Thompson and Dorn1). It is also considered to be one of the most difficult of all commercially impor- tant Alaska groundfish species to age. Scientists from both Canada and the United States have experienced simi- lar difficulties in finding an appro- priate aging structure, which can be consistently interpreted to track large year classes of cod through time. Historically, scales and otoliths have been the two most common struc- tures used for determining the ages of fish species. Unfortunately, age- readers employing these structures have experienced limited success in the case of Pacific cod (Kimura and Lyons, 1990). The Pacific Biological Station in Canada stopped aging Pacific cod in 1978, after age estimates derived from scale readings began yielding year classes that were inconsistent with length-frequency time series from field surveys (Westrheim and Shaw, 1982). The Alaska Fisheries Science Center's (AFSC) Resource Ecology and Fisheries Management (REFM) Division is responsible for stock assessment of Pacific cod in the Gulf of Alaska and eastern Bering Sea. The REFM Division's Age and Growth Program used scales for de- termining the age of Pacific cod from 1976 to the early 1980s. Thereafter, the program used the break-and-burn method with otoliths to age Pacific Thompson, G. G., and M. W. Dorn. 1999. Pacific cod. In Stock assessment and fishery evaluation report for the groundfish resources for the Bering Sea/ Aleutian Islands regions (plan team for groundfish fisheries of the Bering Sea/ Aleutian Islands), p. 151-205. North Pacific Fishery Management Council, 605 W. 4th Avenue Suite 306, Anchor- age, AK 99501. 154 Fishery Bulletin 103(1) cod (Thompson and Methot2). In the otoliths of young Pacific cod (under 6 years), there is a tendency for sub- annual marks (also known as "checks") to be very dark and evenly spaced, making them difficult to distinguish from annuli. This confusion makes it difficult to age the species to an exact age. From 1990 through 1992, the AFSC noticed that the average length at a specific age was smaller than it had been in previous years. The decrease was noticed in ages 1-6 but was especially dramatic in 1-, 2-, and 3-year-olds. It is generally theorized that the shift was the result of one of two scenarios: either the fish popu- lation experienced an actual decrease in length-at-age or the age readers were over-aging fish by counting marks other than annuli. Unable to pinpoint the reason for the shift and given the inherent difficulty of aging cod, production (large-scale) aging of Pacific cod was indefinitely suspended at the AFSC. Pacific cod stock assessments in Alaska have since depended largely on length-frequency data alone to model population age structure because of the difficul- ties in obtaining age estimates (Thompson and Dorn1). However, the use of length-frequency data as proxies for age data can be problematic. If external factors such as ocean conditions affect somatic growth to such a degree that length-at-age within the population is highly vari- able, such as appears to be the case for Pacific cod, then the population becomes difficult to model. Otoliths, on the other hand, are permanent records of growth that are more independent of external factors. Consequently, the Age and Growth Program initiated a new study in 1998 to re-examine the otolith aging structure for Pacific cod. This study used otoliths from tagged Alaska Pacific cod to validate aging criteria for otoliths. Methods Otoliths and length data were collected during a tag- ging study conducted by the AFSC. Between 1982 and 1990, 12,396 Pacific cod were tagged and released in the eastern Bering Sea during summer bottom-trawl surveys (See Shimada and Kimura, 1994). Fish were measured to the nearest 0.5 cm fork length, tagged with uniquely marked spaghetti tags, and set free. Over a period of 13 years, commercial fishing vessels recaptured 375 (3%) of the tagged fish and returned otoliths from 112 fish (106 of which were usable) (Table 1). More details on the tagging methods can be found in Shimada and Kimura (1994). 2 Thompson, G. G., and R. D. Methot. 1993. Pacific cod. In Stock assessment and fishery evaluation report for the groundfish resources for the Bering Sea/Aleutian Islands regions as projected for 1994 (plan team for groundfish fish- eries of the Bering Sea/Aleutian Islands), p. 2-28. North Pacific Fishery Management Council, 605 W. 4lh Avenue Suite 306, Anchorage, AK 99501. Otolith preparation One sagittal otolith from each recaptured fish was selected for our study. We did not discriminate between left and right otoliths based on the results of Sakurai and Hukuda (1984) who were unable to detect any con- sistent differences between the weight and length of right and left Pacific cod otoliths. Each otolith was cleaned and preserved in 95% etha- nol. After having been preserved for approximately one month, a line was penciled across the otolith center from the dorsal apex to the ventral apex to ensure that the otoliths would later be sectioned at the core. The otolith was then placed in a polyester mold and set in black resin (Technovit 3040, Energy Beam Sci- ences, Agawam, MA), forming a block of resin. A slow- speed saw was used to cut the blocks in half. This pro- duced two smaller blocks, each with an exposed view of the otolith in the transverse plane and cut through the center. One of the two blocks was selected and glued (otolith side down) to a glass slide. The glass slide was mounted to a Hillquist thin section machine (Hillquist Inc., Fall City, WA) and the section was ground down to a thickness of 0.25 mm. A coverslip was permanently glued on the top of the section with marine-grade epoxy. Sections were placed on a piece of black velvet (which added contrast) on the stage of a 50x dissecting micro- scope, and reflected light was used for illumination. The sections were viewed on a computer monitor by using a Cohu 6500 monochrome video camera, Integral Flashpoint 128 frame grabber and Optimas 6.5 imaging software (Media Cybernetics, Silver Spring, MD). Age-reading criteria Traditional qualitative aging criteria were used to distin- guish annuli from checks. The criterion for identification as an annulus was a continuous translucent band that could be seen along the entire structure or as a ridge or groove on the structure (Secor et al., 1995). Checks (i.e., subannual marks) are translucent zones that appear very similar to annuli. They were determined primarily by the incompleteness of the zone around the entire sec- tion, by zone darkness, and by spacing between zones. When translucent zones could be classified as either annuli or additional subannular marks, they were clas- sified as checks. Annuli, checks, and edges (the space between the last annulus and the edge of the otolith) were traced by using the Optimas 6.5 software package and measurements of their areas and major axis lengths were collected (Fig. 1). All otoliths were read blind; that is, information about fish length and date of capture (Table 1) was withheld from the reader to prevent bias. When all the otoliths had been aged and measured, the age reader returned to each otolith section to es- timate the area and length of the otolith when the fish was tagged. This was accomplished by following a two-step process. The first step was to approximate the location of the otolith cross-section that corresponded to Roberson et al .: Indirect validation of the age-reading method for Gadus macrocephalus 155 Table 1 Mark and recapture data for spaghetti-tagged Pacific cod [Gadus macrocephalus). L, = fork length at tagging (mm), L., = fork length at recapture (mm), )], where Lx = length at tagging; L2 = length at recapture; Linf = maximum size; K = growth rate; a2 = estimated age at recovery determined from our otolith ages; d = time at liberty; and r0 = age at length 0 mm. Given von Bertalanffy parameters and age at recovery, a (fish) length increment for time after tagging can be predicted. Using published Lln{ and K estimates from tagging data, Linf= 1043 mm, K = 0.222 (Kimura et al., 1993), we estimated L2-Lv One weakness in these estimates is that the growth parameters estimated by Kimura et al. (1993) were based on only positive growth increments (there were some instances where recaptured fish were smaller than they were at tagging, demonstrat- ing negative growth increments). A value for t0 was estimated iteratively in the von Bertalanffy equation by using the subroutine Solver (Frontline Systems Inc., Incline Village, NE) from the Excel software package with the following parameter values: K = 0.222, t = 1 year old, Llnf= 1043 mm (Kimu- ra et al., 1993), /, = length at age one = 180 mm (from the 1977 year class [Foucher et al., 1984]). Because these von Bertalanffy parameters are not based on age determination, they provide an indirect method for validating aging criteria. In addition, ages determined by readers were scaled smaller (by 0.75) and larger (by 1.25) in order to simulate the results of younger and older aging criteria. Plots of observed and predicted growth increments should agree if the aging criteria for a2 reflect the true age of fish. E c .2 2.5 3 Ln Otolith area (mm2) Figure 2 Relationship between Ln fish length and Ln otolith area based on tagged and recaptured Pacific cod (r2 = 0.735, y=4.6+0.66X, n = 96). Results Predicting fish length from tagged fish and back-calculations The parameter v used in all back-calculations in our study was estimated by using otolith area and again by using the major axis of the otolith (Fig. 1). Based on the slopes from the regression of Ln fish length on Ln otolith size (Table 2, [Fig. 2]), the coefficients should be v=1.01 for otolith length and v=0.66 for otolith areas. Back-calculations were performed by using the otolith area and were repeated by using the major axis. Scatter plots of estimated and observed fish lengths were used to visually inspect how well back-calculation determines fish length (Fig. 3). Assuming that the residuals of the back-calculated length at tagging have independent chi-square distributions, an F-test indicates that back- calculations derived from otolith areas are significantly more accurate than back-calculations derived from the major axis (P<0.05). However, because we used the two 158 Fishery Bulletin 103(1) different methods on the same otoliths, the residuals were correlated and thus this result can be considered only approximate. Predicting growth increments of fish length from tagged fish Using the von Bertalanffy curve fitted with the data from the tagged fish sample, we estimated the value of t0 to be 0.147. 1000 -i / jf* 800- V* * ' *&* • iff* # • ♦ 'it* 600- **>Jtr .J***- .y*m. 400- *S ♦ +X X ♦ £ 200- s E jf a> / c yf 2£ / cc o . E _ 0 200 400 600 800 1000 ra 3, Back-calculated lengths at tagging (mm) c CD j; 1000 -i B y TD / CD > / + <5 __X ♦ £ 800 - x * -Q •/ o *•&/ * is* » ♦ ♦ S*T* ♦ n^ •• 600- +r — ♦ . *?{**<•*• 400- y4f • jf ♦ 200- / Q / 0 200 400 600 800 1000 Back-calculated lengths at tagging (mm) Figure 3 Fish length at tagging was back-calculated from estimated otolith size at tagging and time at liberty. Two separate back-calculations were performed, each with a different measure of the otolith size at tagging: one using (A) the area; the other using (B) the major axis. The amount that each tagged fish grew after tagging was calculated three times by using fish age at recovery and the von Bertalanffy equation (Fig. 4). The calcu- lated sums of squared deviations for the three sets of predicted values are as follows: 433,955 when fish age is scaled by 0.75, 419,477 when fish age is scaled by 1.0, and 761,545 when fish age is scaled by 1.25. The lowest sum of squared deviations accompanied ages that were scaled by 0.86. Assuming that the residuals of the esti- mated growth increments have independent chi-square distributions, an F-test indicates that residu- als were significantly larger (P<0.05) when ages were scaled 25% older and there was no significant difference (P<0.05) between reader-determined ages and ages scaled 25% younger. The three sets of residuals came from the same otoliths and would be corre- lated; therefore, this result can be considered only approximate. Another test of our reading criteria was performed through a more direct compari- son: simply "aging" the tagged fish from esti- mated age at tagging (based on length), plus the time after tagging (Table 1). Out of 106 samples, 75% of these fish were within one year of the age that we had determined from otolith readings, and 94% were within two years. The average percent error (Beamish and Fournier, 1981) was 8.70, and the aver- age deviation from tagged-based age was -0.075. Results of a Z-test indicated that the average deviation was not significantly different from zero (P= 0.724) and indicated no bias in the age estimates. Discussion Beamish and McFarlane (1983) noted that "validating a method of age determination is as important in fishery biology as standard- izing solutions or calibrating instruments are in other sciences." Age determination must reflect the actual age of each fish in order to be a useful tool for use in stock assessments. Although much effort has been devoted in the past to finding an appropriate aging struc- ture for Pacific cod, particularly with dorsal fin rays (Beamish, 1981; Lai et al., 1987; Kimura and Lyons, 1990), scales and otoliths (Lai et al., 1987; Kimura and Lyons, 1990), a directly validated method of age determina- tion has yet to be found ( Westrheim, 1996). The otolith seems to be the most promising structure for production (large-scale) age reading of Pacific cod (Kimura and Lyons, 1990); however it is not without weaknesses (i.e., the faint patterns of some translucent zones can lead to low precision between read- ers and are a constant concern in regard to Roberson et al Indirect validation of the age-reading method for Gadus macrocephalus 159 under- or overestimated ages). The key difficulty of the cod otolith patterns is differentiating the translucent zones that are annual from the translucent zones that are checks, particularly in young fish. It is necessary to have validated criteria in order to confidently eliminate checks without under-aging the fish. In our study, we have given two examples of indirect validation for Pacific cod age determination by using otoliths from marked and recaptured fish. In the first example, we used back-calculations to test our reading criteria, which exclude counting lighter translucent zones. Early in the study, we found a strong relationship between otolith size and fish length, which supported using back-calculations as a vehicle to test accuracy. Overall, using strong translucent zones to back-calculate fish length at tagging gave fairly accu- rate results. This finding supports the assumption that translucent zones are laid down on an annual basis. An ancillary finding was that otolith area measure- ments provided more accurate estimates of fish length than otolith lengths. Although back-calculations are typically performed by using radial or diametral mea- surement, the more accurate estimates of fish length from otolith area measurements are not surprising in that otolith area is a more comprehensive measure of otolith three-dimensional growth. A second indirect validation of reading criteria was possible by estimating how much each tagged fish grew (millimeters) between tagging and recapture by its estimated age at recovery and von Bertalanffy growth parameters (derived only from tagging data). When compared to the observed growth increments, we found that the results support our proposed aging criteria (which exclude lighter translucent zones) because these criteria give the best fit to growth increments based on the mark-recapture growth increments. Aging the fish older (by counting light translucent zones) or younger (counting less annuli by banding translucent zones to- gether) increases the residual fit to the mark-recapture growth increments. Large growth increments of fish length were difficult to estimate (Fig. 4). A possible explanation is that the longer a fish remains at liberty, the more likely that the growth becomes asymptotic, making the relationship between the growth increment and time at liberty less exact. The final test for reading criteria was performed through a more direct comparison: simply "aging" the tagged fish by its length-at-release plus its time at liberty after tagging and comparing that age to the otolith-based age at recovery. Dwyer et al. (2003) also used this method in their study of yellowtail flounder (Limanda ferruginea). Average deviation from tag-based age was -0.075; 75% of these fish were found to be within one year of our age according to otolith readings, and 94% were within two years. These results provide further evidence that the current criteria do not result in the underestimation of the age of the fish and sup- port the practice of not counting checks. We found that growth information residing in oto- liths from tagged and recovered Pacific cod provided 200 400 600 600 -, 400 ™ 200- 200 400 600 600 400 200 0 200 400 600 Observed growth Increments (mm) Figure 4 Three plots comparing predicted and observed fish-length growth increments by using recap- tured fish from tagging experiments (n = 97). Estimated ages at recovery (B) were scaled 25% smaller (A) and 25% larger (C). The lines indi- cate theoretical 1:1 line of perfect agreement. significant information applicable to indirectly validat- ing otolith aging criteria. Therefore, it seems that oto- liths from other species that were tagged and recovered might be useful for indirect age validation as well. The 160 Fishery Bulletin 103(1) information provided in our study indicates that our aging criteria for Pacific cod are roughly correct and that errors are probably within plus or minus 1 or 2 years. However, the problem of the shift in length at age alluded to in the introduction is more difficult to elucidate. Analysis made during this study seems to indicate that both environmental growth factors and changes in otolith aging criteria could have played a role in this apparent shift. Acknowledgments This work was performed in partial fulfillment of the requirements for an M.S. degree at the School of Aquatic and Fishery Sciences at the University of Washington and was supported by the National Marine Fisheries Service. We would like to express our appreciation to those individuals who assisted in the tagging and recap- ture process and to the two anonymous reviewers for their helpful comments on the manuscript. Literature cited Beamish, R. J. 1981. Use of sections of fin-rays to age walleye pol- lock, Pacific cod, albacore, and the importance of this method. Trans. Am. Fish. Soc. 110:287-299. Beamish, R. J., and D. A. Fournier. 1981. A method for comparing the precision of a set of age determinations. Can. J. Fish. Aquat. Sci. 38:982- 983. Beamish, R. J., and G. A. McFarlane. 1983. The forgotten requirement for age validation in fisheries biology. Trans. Am. Fish. Soc. 112:735 Dwyer, K. S., S. J. Walsh, and S. E. Campana. 2003. Age determination, validation and growth of Grand Bank yellowtail flounder (Limanda ferruginea). ICES J. Mar. Sci. 60:1123-1138. Foucher, R. P., R. G. Bakkala, and D. Fournier. 1984. Comparison of age frequency derived by length- frequency analysis and scale reading for Pacific cod in the North Pacific Ocean. Int. North Pac. Fish. Comm. Bull. 42:232-242. Kimura, D. K., and J. J. Lyons. 1990. Choosing a structure for the production aging of Pacific cod (Gadus macrocephalus). Int. N. Pac. Fish. Comm. Bull. 50:9-23. Kimura, D. K., A. M. Shimada, and S. A. Lowe. 1990. Estimating von Bertalanffy growth parameter of sablefish (Anoplopoma fimbria) and Pacific cod (Gadus macrocephalus) using tag-recapture data. Fish. Bull. 91:271-280. Lai, H. L., D. R. Gunderson, and L. L. Loh. 1987. Age determination of Pacific cod, Gadus mac- rocephalus, using five aging methods. Fish. Bull. 85:713-723. Ricker, W. E. 1975. Computation and interpretation of biological sta- tistics offish populations. Bull. Fish. Res. Board Can. 191, p. 382. Sakurai, Y., and S. Hukuda. 1984. The age and growth of the spawning Pacific cod in Mutsu Bay. Sci. Rep. Aquat. Cen., Aomori Pref. 3:9- 14. Secor, D. H., J. M. Dean, and S. E. Campana (eds.). 1995. Recent developments in fish otolith research. Belle W. Baruch Institute for Marine Biology and Coastal Research., Univ. South Carolina Press, Columbia, SC. Shimada, A. M., and D. K. Kimura. 1994. Seasonal movements of Pacific cod, Gadus mac- rocephalus, in the eastern Bering Sea and adjacent waters based on tag-recapture data. Fish. Bull. 92:800- 816. Smedstad, O. M., and J. C. Holm. 1996. Validation of back-calculation formulae for cod otoliths. J. Fish Biol. 49:973-985. Westrheim, S. J. 1996. On the Pacific cod (Gadus macrocephalus) in Brit- ish Columbia waters, and a comparison with Pacific cod elsewhere, and Atlantic cod (G. morhua). Can. Tech. Rep. Fish. Aquat. Sci. 2092, 390 p. Westrheim, S. J., and W. Shaw. 1982. Progress report on validating age determination methods for Pacific cod (Gadus macrocephalus). Can. Manuscr. Rep. Fish. Aquat. Sci. 1670, 41 p. 161 Abstract — The northwest Atlantic population of thorny skates (Ambly- raja radiata) inhabits an area that ranges from Greenland and Hudson Bay, Canada, to South Carolina. Despite such a wide range, very little is known about most aspects of the biology of this species. Recent stock assessment studies in the northeast United States indicate that the bio- mass of the thorny skate is below the threshold levels mandated by the Sustainable Fisheries Act. In order to gain insight into the life history of this skate, we estimated age and growth for thorny skates, using verte- bral band counts from 224 individuals ranging in size from 29 to 105 cm total length (TL). Age bias plots and the coefficient of variation indicated that our aging method represents a nonbiased and precise approach for the age assessment of A. radiata. Mar- ginal increments were significantly different between months (Kruskal- Wallis P<0.001); a distinct trend of increasing monthly increment growth began in August. Age-at-length data were used to determine the von Ber- talanffy growth parameters for this population: Lx = 127 cm (TL) and 6 = 0.11 for males; Lr = 120 cm (TL) and 6 = 0.13 for females. The oldest age estimates obtained for the thorny skate were 16 years for both males and females, which corresponded to total lengths of 103 cm and 105 cm, respectively. Age and growth estimates of the thorny skate (Amblyraja radiata) in the western Gulf of Maine James A. Sulikowski Jeff Kneebone Scott Elzey Zoology Department, Spaulding Hall 46 College Road University of New Hampshire Durham, New Hampshire 03824 E mail address !for J A Sulikowski, senior author): isulikowigihotmail.com Joe Jurek Yankee Fisherman's Cooperative P.O. Box 2240 Seabrook, New Hampshire 03874 Patrick D. Danley Department of Biology University of Maryland College Park, Maryland 20724 W. Huntting Howell Zoology Department, Spaulding Hall 46 College Road University of New Hampshire Durham, New Hampshire 03824 Paul C.W. Tsang Department of Animal and Nutritional Sciences Kendall Hall 129 Main St. University of New Hampshire Durham, New Hampshire 03824. Manuscript submitted 21 August 2003 to the Scientific Editor's Office. Manuscript approved for publication 8 July 2004 by the Scientific Editor. Fish. Bull. 103:161-168(2005). The northeast skate complex consists of seven species endemic to the waters off the New England coast of the United States (New England Fisheries Management Council (NEFMC1-2). In the past, these skates were generally discarded because of their low commer- cial value (NEFMC1-2). More recently, the rapidly expanding markets for human consumption of skate wing and for use as lobster bait have made three of these species (winter skate [Leucoraja ocellata], little skate [L. erinacea], and thorny skate [Amblyraja radiata]) commercially more viable (Sosebee, 2000; NEFMC1-2). Despite an increasing commercial importance, harvests for skate in the U.S. portion of the western north Atlantic remain unregulated and have the potential to over-exploit the stocks. Moreover, life history information is almost nonex- istent for most of these elasmobranch fishes (Frisk, 2000 NEFMC1-2). The available information from the few skates that have been studied cat- egorizes them as equilibrium strate- gists (K selected) because they reach sexual maturity at a late age, have a low fecundity, and are relatively long- lived (Holden 1977; Winemiller and Rose, 1992; Zeiner and Wolfe, 1993; Francis et al., 2001; Frisk et el., 2001, Sulikowski et al., 2003). These characteristics, coupled with the prac- tice of selective removal of large in- dividuals, make these animals more likely to be over-exploited by commer- cial fisheries (Brander 1981; Hoenig and Gruber, 1990; Casey and Myers 1998; Dulvey et al., 2000; Frisk et al., 2001). The thorny skate (Amblyraja radi- ata) is a cosmopolitan species found on both sides of the Atlantic ocean from Greenland and Iceland to the English Channel in the eastern At- lantic (Compagno et al., 1989) and from Greenland and Hudson Bay, Canada, to South Carolina, United States, in the western Atlantic (Rob- ins and Ray, 1986; Collette and Klein, 2002). Along with this broad geo- graphic range, marked differences in size exist for specimens captured in different regions of the Atlantic. For 1 NEFMC (New England Fishery Man- agement Council. 2001. 2000 stock assessment and fishery evaluation (SAFE) report for the northeast skate complex. NEFMC, 50 Water Street, Mill 2 Newburyport, MA 01950. 2 NEFMC (New England Fishery Man- agement Council). 2003. Skate fish- eries management plan. NEFMC, 50 Water Street, Mill 2 Newburyport, MA. 01950. 162 Fishery Bulletin 103(1) example, the thorny skate reaches total lengths of over 100 cm in the Gulf of Maine (Collette and Klein, 2002), whereas specimens captured off the Labrador coast do not reach total lengths >72 cm (Templeman, 1987). Although no directed fisheries for this species exists in the Gulf of Maine, this skate meets the minimum 1*4 pound-cut pectoral-fin size sought after by wing proces- sors (Sosebee, 2000; NEFMC1-2). Unfortunately, because landings are not reported by species, the proportion of thorny skates to the total wing market is unknown. Re- cent assessment studies in the northeast United States (NEFSC3) indicate that the biomass of thorny skates is declining, and is below threshold levels mandated by the Sustainable Fisheries Act (SFA; NMFS4). Thus, owing to the recent commercial interest in this species and the concomitant decline in population size, obtain- ing life history information for this species has become more important. In order to provide insight into the biology of this species and to determine the stock status (Simpfendorfer, 1993; Frisk et al., 2001), our objectives were to estimate age and growth rates of A. radiata based on banding patterns in vertebral centra from specimens collected in the western Gulf of Maine. Marine Laboratory (CML). There, individual fish were euthanized (0.3g/L bath of MS222). Total length (TL in cm) was measured as a straight line distance from the tip of the rostrum to the end of the tail, and disc width (DW in cm) as a straight line distance between the tips of the widest portion of pectoral fins. Total wet weight (kg) was also recorded. Preparation of vertebral samples Vertebral samples, taken from above the abdominal cavity, were removed from 320 thorny skates ( 154 females and 166 males), labeled, and stored frozen. After having been thawed, three centra from each specimen were removed from the vertebral column, stripped of excess tissue and air dried. Large centra were cut sagittally with a Dremel™ tool fitted with a mini saw attachment while held with a vice-like device. Smaller centra were sanded with a Dremel™ tool to replicate a sagittal cut. Processed vertebrae were mounted horizontally on glass microscope slides and ground with successively finer-grit (no. 180, no. 400, no. 600) wet-dry sandpaper. Each ver- tebra was then remounted and the other side ground to produce a thin (300-micrometer) hourglass section. Materials and methods Sampling Thorny skates were captured by otter trawl in an approx- imate 900 square mile area centered at 42°50'N and 70°15'W in the Gulf of Maine between June 2001 and May 2002. These locations varied from 30 to 40 km off the coast of New Hampshire. Approximate depths at this location ranged between 100 and 120 m. This area was chosen for two reasons: 1) these waters support the vast majority of commercial fishing in New Hamp- shire and can be easily accessed during normal fishing operations; and 2) because of rolling closures within the Gulf of Maine, an experimental fishing permit was granted to us by the National Marine Fisheries Service (NMFS) to collect thorny skates in this location during the months of April, May, and June, when these waters are closed to commercial fishing. Although our sam- pling was conducted in a small portion of the species' range, the sizes of thorny rays collected corresponded to those collected during the NEFSC bottom trawl surveys conducted throughout the Gulf of Maine and Georges Bank (NEFMC1 ; NEFSC3). From this information, it is unlikely that differences in other biological parameters exist. Skates were maintained alive on board the vessel un- til arrival at the LTniversity of New Hampshire's Coastal 3 NEFSC (Northeast Fisheries Science Center). 1999. 30th Northeast regional stock assessment workshop. NEFSC, 166 Water Street, Woods Hole, MA 02543-1026. 4 NMFS (National Marine Fisheries Service). 2002. Annual report to Congress on the status of U.S. fisheries 2001, 142 p. NMFS, NOAA, Silver Spring, MD 20910. Band counts Vertebral sections were digitally photographed with a Canon Powershot S40 attached to a Leica S8PO dis- secting microscope and reflected light. Magnification depended on the size of the section and varied from 4x to 12x (Fig. 1). A growth ring (band count) was defined as one opaque and translucent band pair that traversed the intermedialia and that clearly extended into the corpus calcareum (Casey et al., 1985; Brown and Gruber, 1988; Sulikowski et al., 2003). The birth mark (age zero) was defined as the first distinct mark distal to the focus that coincided with a change in the angle of the corpus calcareum (Casey et al., 1985; Wintner and Cliff, 1996; Sulikowski et al, 2003). Precision and bias Nonconsecutive band counts were made independently by two readers for each specimen used in the study with- out prior knowledge of the skate's length or of previous counts. A Tukey test was used to test for differences between ages. Age determination bias between read- ers was assessed through the use of an age-bias plot. This type of graph displays band counts of one reader against a second reader in reference to an equivalence line. Specifically, reader 2 is represented as mean age and 95% confidence intervals corresponding to each of the age classes estimated by reader 1 (Campana et al., 1995). Divergence from the equivalence line, where reader 1 = reader 2, would indicate a systematic dif- ference between readers. Precision estimates of each reader were calculated by using the coefficient of varia- tion (CV) as described by Chang (1982) and Campana et al. (1995). Sulikowski et al.: Age and growth of Amblyro/a rodiata 163 Marginal increment analyses The annual periodicity of band pair formation was investigated using marginal increment analyses (MIA). Because the annuli in older adult specimens were compressed, marginal increments were calculated from randomly selected juvenile specimens (Simpfendorfer, 1993; Sulikowski et al., 2003). For MIA. the distance of the final opaque band and the penultimate opaque band, from the centrum edge, was measured with a compound micro- scope and optical micrometer. The marginal increment was calculated as the ratio of the distance between the final and penultimate bands (Branstetter and Musick, 1984; Cail- liet, 1990; Simpfendorfer, 1993; Simpfendorfer et al., 2000; Sulikowski et al., 2003). Average increments were plotted by month of capture to identify trends in band formation, and a Kruskal-Wallis one-way analysis of variance on ranks was used to test for differences in marginal increment by month (Simpfendorfer et al., 2000; Sulikowski et al., 2003). Growth estimates A von Bertalanffy growth function (VBGF) was fitted to the data with the following equa- tion (von Bertalanffy, 1938): Lt=LJl -e -kit - t„\ V), where /, = total length at time t (age in years); L = theoretical asymptotic length; k = Brody growth constant; and t0 = theoretical age at zero length. Figure 1 Longitudinal cross-section of a vertebral centrum from a 97-cm-TL female Amblyraja radiata estimated to be 12 years. BM = birth mark; Black dots represent age in years. The VBGF was calculated by using FISH- PARM, a computer program for parameter estimation of nonlinear models with Marquardt's (1963) algorithm for least-square estimation of nonlinear parameters (Prager et al., 1987). Results Morphological measurements Out of the 320 specimens collected, a total of 224 were used for our study (Table 1). Males (rc=103) ranged between 29 and 103 cm TL, 18-75 mm DW, and 0.125- 10.5 kg body weight (data not shown), whereas females (n=121) ranged between 31 and 105 cm TL, 18-74 cm DW, and 0.170-11.4 kg body weight (data not shown). Total length, disk width, and body weight were strongly correlated in males, females, and when data from the sexes were combined (all coefficient of determination [r2] values were greater than 0.87). Vertebral analyses Comparison of counts between two readers indicated no appreciable bias in the counting process (Fig. 2) and the coefficient of variation for all sampled vertebrae was 2.8% This level of precision is considered acceptable (Campana, 2001) and counts generated by both readers were combined (averaged) for the analyses (Skomal and Natanson, 2003). The relationship between TL and centrum diameter was linear (r2=0.93; P<0.05) and there were no signifi- cant differences in this relationship (ANCOVA, P<0.05) between males and females. Because no significant difference existed between TL and centrum diameter between the sexes, these data were combined (Fig. 3). A total of 120 skates (10 per month) were used for marginal increment analyses. Marginal increments were significantly different between months (Kruskal- Wallis P<0.001); there was a distinct trend of increasing 164 Fishery Bulletin 103(1) Number of bands (age) of reader one Figure 2 Age bias graph for pair-wise comparison of 224 thorny skate {Amblyraja radiata) vertebral counts by two independent readers. Each error bar represents the 95% confidence interval for the mean age assigned by reader 2 to all fish assigned a given age by reader 1. The diagonal line represents the one-to-one equivalence line. Sample sizes are given above each corresponding age. Table 1 Average total length (TL) and d isc width (DW) at age for male and female thorny skates (A. radiata). Sizes are presented as mean ±1 SEM sample sizes are given in parentheses. Age (yes rs) Male TL (cm) Female TL (cm) S exes combined Male DW (cm) Female DW (cm) Sexes combined 2 33 (5) ±1 33 (5) ±1 23±1 2±1 3 37 (3) ±1 37 (7) ±1 37(10)±1 26 ±1 27 ±0 27 ±1 4 43 (5) ±1 42 (2) ±2 42 (7) ±1 29 ±0 29 ±1 29 ±1 5 48(2)±2 49 ( 7) ±2 48 (9) ±1 33 ±0 35 ±0 34 ±1 6 64(1)±1 54(17)±1 54(18)±1 44 ±0 39 ±2 39 ±2 7 69 (5) ±1 62 (5) ± 3 64(10)±1 50 ±1 44 ±2 47 ±1 8 71 (6) ±1 73 (11) ±2 72 (17) ±2 52 ±1 53 ±2 53 ±1 9 78 (9) ±1 78 (8) ±2 78(17)±1 57 ±2 57 ± 1 57 ±1 10 86 (14) +1 82(15)±1 84(29)±1 63 ±2 60+1 61 ±1 11 88(17)±1 89<17)±1 89(34)±1 65 ±1 65 ±1 65 ±1 12 93(18)±1 92(19)±0 92(37)±1 68 ±1 66 ±1 67 ±1 13 99(10)±1 95 (8) ±1 97(18)±1 73 ±1 68 ±1 70+1 14 97 (3) ±3 98(1)±0 96 (4) ±2 70 ±2 70 ±0 70 ±1 15 102 (2) ±1 101 (3) ±0 101 (5) ±1 70 ±1 74 ±2 74 ±1 16 101 (3) ±2 105(1)±0 102 (4) ±2 75 ±1 70 ±1 75 ±1 monthly increment growth that peaked in July, followed by a large decline in August (Fig. 4). Based on this information, the increment analyses support the likeli- hood that a single opaque band may be formed annually on vertebral centra during August or September. The marginal analysis was only conclusive for juvenile-size animals (skates s80 cm TL). As thorny skates matured, their growth slowed dramatically and the band counts in older specimens became compressed, making it dif- ficult to discern monthly changes in margin width. Suhkowski et al.: Age and growth of Amblyra/a radiata 165 Age and growth estimates The von Bertalanffy growth curves (VBGC) fitted to total length-at-age data (Fig. 5) pro- vided a reasonable fit with a low standard error for males, females, and both sexes combined (Table 2). Furthermore, the VBGC parameters for males, females, and the sexes combined were similar, and because no differences in age-at- size existed between males and females (P>0.05 ANOVA), those data were combined (Fig. 5). Discussion Precision estimates, the relationship between TL and centrum diameter, and marginal incre- ment analysis support the use of vertebral centra for age analyses of thorny skates cap- tured in the Gulf of Maine (Conrath et al., 2002; Sulikowski et al., 2003). Furthermore, the 2.8% coefficient of variation indicates that our aging method represents a precise approach for the age assessment of A. radiata (Cam- pana, 2001). Minimal width of the marginal increment for thorny skates captured in August and September (Fig. 4) supports the hypoth- esis of annual band formation in this species. Moreover, these results compare favorably to cycles in marginal increments (Sulikowski et al., 2003) and to annual vertebral band pat- terns in other skate species (Holden and Vince, 1973; Waring, 1984; Natanson, 1993; Zeiner and Wolfe, 1993; Walmsley-Hart et al., 1999; Francis et al., 2001). However, because the band counts of the largest and oldest animals in the population were compressed (too small for us to discern marginal increments from their widths), the marginal increment analysis included only juvenile skates that were s80 cm total length and the annular nature of the growth bands was verified for only those length groups. Nevertheless, we assumed that as the skates grew larger and older, the annual nature of growth ring deposition continued throughout their lifetime (Conrath et al, 2002). During 42 sampling trips from June 2001 through May 2002 (approximately three trips per month), individuals <30 cm TL were rarely captured. The lack of specimens in this size class and smaller size classes was most like- ly due to the mesh size of the commercial trawl nets. Trawl nets used by commercial fishermen in the Gulf of Maine are restricted to a 6V2-inch diamond mesh- size opening, which facilitates the release of most fish below 30 cm TL. Our estimates of Lr exceed the largest specimens in our field collections for both females and males. Growth parameters estimated from the Gompertz and logistics equations also produced over-estimations of La for the 140 -i r2 = 0.93 120 - P<0.05 100 - •jtfS^ ? u — 80 - ^ 60- ro 40 - S* 20 - ^ 0 2 4 6 8 10 12 Centrum diameter (mm) Figure 3 The relationship of total length (cm) to centrum diameter (mm) for combined sexes of thorny skate (Amblyraja radiata). 0 9 -| 0.8 - nal increment o o | 0.5 - Relative o 0 3 - Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Month Figure 4 Mean monthly marginal increments of opaque bands for Ambly- raja radiata from the Gulf of Maine. Marginal increments were calculated each month from 10 specimens whose centra contained less than 10 annuli. Error bars represent 1 +SEM. thorny skates in our study. Because the von Bertalanffy growth curve (VBGC) is most widely used and accepted in elasmobranch age and growth studies, we chose to use this function to fit our data. Over estimation of L^ with the VBGC has been documented in most skate species studied to date (Table 3). Campana (2001) sug- gested that the smallest and largest specimens are the most influential in the estimation of growth. Moreover, both Walmsley-Hart et al. (1999) and Sulikowski et 166 Fishery Bulletin 103(1) al. (2003) suggested that rareness of large in- dividuals was most likely responsible for their over estimation of Lx. Similarly, in a study of the blue shark (Prionace glauca) in the northwest Atlantic, Skomal and Natanson (2003) suggested that earlier studies on the same species contained artificially inflated Lx estimates and lower growth rates because of the lack of maximum-size fish. The rareness of large specimens in our study may have been due to these larger individuals being able to avoid the fishing gear or may indicate that mortality, natural or fishing induced, prevents them from attaining these lengths. Exploratory manipulation of our data indicated that inclusion of maximum observed sizes (i.e., thorny skates over 103 cm TL) produced divergent results with regard to von Bertalanffy parameters. For ex- ample, the addition of maximum-size fish, using 20 years as the maximum age (Templeman, 1984) and 105 cm as the maximum total length (from the present study), reduced the combined sex Lx from 124 cm TL to 116 cm TL. However, the same effect was not documented when adding hatching size (age zero) fish (note: because no documented size for thorny skates exists within the Gulf of Maine, the authors used known hatching sizes for a similar congener species, the winter skate [Leucoraja ocellata]). Be that as it may, the reasonable fit of the thorny skate data (Table 2) to the VBGC (Fig. 5) for A. radiata, along with a comparison with other batoid studies (Table 3), indicates that this is an appropriate model for this skate species. Growth rates were similar for both sexes of thorny skate (£ = 0.13 for females and 6 = 0.11 for males) and commensurate with other skate species of a similar size. The oldest age obtained for male and female thorny skates was 16 years (Table 1). These data are in agreement with the assumption that larger batoids, such as A. radiata, R. pullopunctata (Walmsley-Hart et al„ 1999), and L. ocellata (Sulikowski et al., 2003) are longer lived and slower growing than smaller species. For instance, R. erinacea, which reaches a total length of 52 cm, has been aged to 8 years and found to have 110 -i 100 - 90 - 80 - .lit 'Jr^^ ? 70- Jf\ • Females £ 60 - J5 50 - 2 40- 30 - */• * Males 20 - / 10 - / 0 2 4 6 8 10 12 14 16 Band count (age in years) Figure 5 Von Bertalanffy growth curve generated from combined ver- tebral data for female and male thorny skates iAmblvraja radiata) from the western Gulf of Maine. Individual VBGC parameters are given in Table 2. Table 2 Calculated von and combined Bertalar sexes of A ffy parameters radiata. for male, female. Parameter Male Female Combined sexes L.(cmTL) 127.00 120.00 124.00 K (/year) 0.11 0.13 0.12 t0 (year) -0.37 -0.4 -0.35 r2 0.96 0.92 0.94 SE 0.01 0.01 0.001 n 103.00 121.00 224.00 a corresponding k value of 0.352 (Johnson, 1979; War- ing, 1984). The reduction in biomass of the thorny skate below threshold levels mandated by the SFA necessitates the development of management measures to rebuild these stocks in accordance with the Magnuson-Stevens Fish- ery Conservation and Management Act. However, the development and implementation of a successful fisher- ies management plan for the species require in-depth analyses of appropriate biological information. More- over, accurate stock assessment data for skates is dif- ficult to collect in the northeast U.S. because individual species are rarely differentiated in landings information (NEFMC1 2). As a result, fluctuations in stock size will continue to be difficult to detect and successful imple- mentation of fisheries management plans will remain problematic. This article is the first in a series aimed at providing life history data for the management of thorny skates indigenous to the Gulf of Maine. The basic age and growth parameters for the thorny skate provided in the present study support the hypothesis that A. radiata, like other elasmobranchs, require con- servative management because of their slow growth rate and susceptibility to over-exploitation (Brander, 1981; Kusher et al., 1992; Zeiner and Wolf, 1993; Frisk et al., 2001; Sulikowski et al., 2003). Acknowledgments Collection of skates was conducted on the FV Mystique Lady. We thank Noel Carlson for maintenance of the fish at the U.N.H. Coastal Marine Laboratory and Matt Ayer for his help in digitizing the vertebrae samples. This Sulikowski et al.: Age and growth of Amblyra/a radiata 167 Table 3 Comparison of von Bertal inffy derived Lx to the observed total lengths (L) for a number of skate species, i, (mm) = disk width. Scientific name Sex L, tmml L observed (mm) Max. age (yr) Source Raja microocellata 9 a 1370 (TL) 875' 9 Ryland and Ajayi, 1984 Raja montagui 9 a* 978 (TL) 710' 7 Ryland and Ajayi, 1984 Raja clavata 9 1047 (TL) 1392 7 Ryland and Ajayi, 1984 Raja erinacea 9 o" 527 (TL) 520 8 Waring, 1984 Raja rhina O* 967 (TL) 1322 13 Zeiner and Wolf, 1993 Raja rhina 9 1067 (TL) 1047 12 Zeiner and Wolf. 1993 Raja wallacei o* 405 (DW) 512 15 Walmsley-Hart et al., 1999 Raja wallacei 9 435 (DW) 571 15 Walmsley-Hart et al., 1999 Raja pullopunctata o* 771 (DW) 696 18 Walmsley-Hart et al., 1999 Raja pullopunctata 9 1327 (DW) 747 14 Walmsley-Hart et al., 1999 Raja pullopunctata 9 1327 (DW) 747 14 Walmsley-Hart et al., 1999 Dipturus nasutus 9 o* 700 (TL) 913 9 Francis et al., 2001 Dipt unit: innominatus 9 o* 1330 (TL) 1505 24 Francis et al., 2001 Leucoraja ocellata 9 o* 1314 (TL) 936' 19 Sulikowski et al., 2003 Amblyraja radiata 9 o* 1240 (TL) 1020' 16 Present study ' Average of male and female observed TL. project was supported by a Northeast Consortium grant (no. NA16FL1324) to PCWT, JAS, and PDD. Literature cited Brander, K. 1981. Disappearance of common skate flora batis from Irish Sea. Nature 290 (58011:48-49. Branstetter, S., and J. A. Musick. 1984. Age and growth estimates for the sand tiger in the northwestern Atlantic ocean. Trans. Am. Fish. Soc. 123:242-254. Brown, C. B., and S. H. Gruber. 1988. Age assessment of the lemon shark, Negaprion brevirostris, using tetracycline validated vertebral centra. Copeia 3:747-753. Casey, J. G., H. L. Pratt, and C. E. Stillwell. 1985. Age and growth of the sandbar shark (Carcharhinus plumbeus ) from the western North Atlantic. Can. J. Fish. Aquat. Sci. 42(51:963-975. Casey, J. M., and R. A. Myers. 1998. Near extinction of a large widely distributed fish. Science 28:690-692. Cailliet, G. M. 1990. Elasmobranch age determination and verifica- tion: an updated review. In Elasmobranchs as living resources: advances in the biology, ecology, systemat- ics and the status of the fisheries (L. Pratt Jr, S. H. Gruber, and T. Tanuichi (eds.), p. 157-165. NOAA Tech. Report NMFS 90. Campana, S. E. 2001. Accuracy, precision and quality control in age determination including a review of the use and abuse of age validation methods. J. Fish. Biol. 59:197-242. Campana, S. E., M. C. Annand, and J. I. Mcmillan. 1995. Graphical and statistical methods for determining the consistency of age determinations. Trans. Am. Fish. Soc. 124:131-138. Chang, W. Y. B. 1982. A statistical method for evaluating the reproduc- ibility of age determination. Can. J. Fish. Aquat. Sci. 39:1208-1210. Collette, B., and G. Klein-MacPhee, eds. 2002. Bigelow and Schroeder's fishes of the Gulf of Maine, 3rd ed., p. 62-66. Smithsonian Institution Press, Washington, D. C. Compagno, L. J. V., D. A. Ebert, and M. J. Smale. 1989. Guide to the sharks and rays of southern Africa, 158 p. New Holland (Publ.) Ltd., London. Conrath, C. L., J. Gelsleichter, and J. A. Musick. 2002. Age and growth of the smooth dogfish, Mustelus canis, in the northwest Atlantic. Fish. Bull. 100:674- 682. Dulvey, N. K., J. D. MetCalfe, J. Glanville, M. G. Pawson, and J. D. Reynolds. 2000. Fishery stability, local extinctions, and shifts in community structure in skates. Cons. Biol. 14 (1): 283-293. Francis, M., C. O. Maolagain, and D. Stevens. 2001. Age, growth, and sexual maturity of two New Zealand endemic skates, Dipturus nasutus and D. inno- minatus. N. Z. J. Mar. Freshw. Res. 35:831-842. Frisk, M. G. 2000. Estimation and analysis of biological parameters in elasmobranch fishes and the population dynamics of the little skate Raja erinacea, thorny skate R. radiata and Barndoor skate R. laeuis. M.S. thesis, 112 p. Center for Environmental Science, Univ. Maryland, Cambridge, MD. Frisk, M. G„ T. J. Miller, and M. J. Fogarty. 2001. Estimation and analysis of biological parameters in elasmobranch fishes: a comparative life history study. Can. J. Fish. Aquat. Sci. 58(5 ):969-981. 168 Fishery Bulletin 103(1) Hoenig, J., and S. H. Gruber. 1990. Life history patterns in the elasmobranchs: impli- cations for fisheries management. In Elasmobranchs as living resources: advances in the biology, ecology, systematics and the status of the fisheries (H. L. Pratt Jr, S. H. Gruber, and T. Tanuichi, eds. ), p. 1-16. NOAA Tech. Report, NMFS 90. Holden, M. J. 1977. Elasmobranchs. In Fish population dynamics (J. A. Gulland, ed.l, p. 187-214. J. Wiley and Sons, London, England. Holden, M. J., and M. R. Vince. 1973. Age validation studies on the centra of Raja cla- vata using tetracycline. J. Cons. Int. Explor. Mer 35:13-17. Johnson, G. F. 1979. The biology of the little skate, Raja erinacea, in Block Island Sound, Rhode Island. Unpubl. M.A. thesis, 119 p. Univ. Rhode Island. Kingston, RI. Kusher, D. I.. S. E. Smith, and G. M. Cailliet 1992. Validated age and growth of the leopard shark. Tri- akis semifasciata, with comments on reproduction. En- viron. Biol. Fish. 35:187-203. Magnuson Fishery Conservation and Management Act. 1994. 16 U.S.C. 1801-1882 (1994), amended by Sustain- able Fisheries Act, Pub. L. No. 104-297, 110 Stat. 3559 (1996). Marquardt, D. W. 1963. An algorithm for the least squares estimation of nonlinear parameters. J. Soc. Ind. Appl. Math. 2: 431-441. Natanson, L. J. 1993. Effect of temperature on band deposition in the little skate, Raja erinacea. Copeia 1993:199-206. Prager, M. H., S. B. Saila, and C. W. Recksiek. 1987. Fishparm: a microcomputer program for parameter estimation of nonlinear models in fishery science. Old Dominion University, Norfolk Virginia, Tech. Rep. (87- 10):l-37. Robins, C. R., and G. C. Ray. 1986. A field guide to Atlantic coast fishes of North America, 354 p. Houghton Mifflin Co., Boston, MA. Ryland, J. S., and T. O Ajayi. 1984. Growth and population dynamics of three Raja species (Batoidei) in Carmarthen Bay, British Isles. J. Cons. Int. Explor. Mer 41:111-120. Simpfendorfer, C.A. 1993. Age and growth of the Australian sharpnose shark, Rhizoprionodon taylori, from north Queensland, Australia. Environ. Biol. Fish. 36:1 3 1:233-241. 2000. Age and growth of the whiskery shark, Furgaleus macki, from south-western Australia. Environ. Biol. Fish. 58:335-343. Skomal, G. B, and L. J. Natanson. 2003. Age and growth of the blue shark iPrionace glauca) in the North Atlantic Ocean. Fish. Bull. 101:627-639. Sosebee, K. 2000. Skates. Status of fishery resources off the north- eastern United States. NOAA Tech. Mem. NMFS-NE. 115:114-115. Sulikowski, J. A., M. D. Morin, S. H. Suk, and W. H. Howell. 2003. Age and growth of the winter skate, Leucoraja ocellata, in the Gulf of Maine. Fish. Bull. 101:405- 413. Templeman, W. 1984. Migrations of thorny skate, Raja radiata, tagged in the Newfoundland area. J. Northw. Atl. Fish. Sci. 5(l):55-63. 1987. Differences in sexual maturity and related char- acteristics between populations of thorny skate Raja- radiata in the northwest atlantic. J. Northw. Atl. Fish. Sci. 44 (1):155-168. Von Bertalanffy, L. 1938. A quantitative theory of organic growth (inquiries of growth laws II). Human Biology 10:181-183. Walmsley-Hart, S. A., W. H. H. Sauer, and C. D. Buxton. 1999. The biology of the skates Raja wallacei and R. pullopunctata (Batoidea:Rajidae) on the Agulhas Bank, South Africa. S. Afr. J. Mar. Sci. 21:165-179. Waring, G. T 1984. Age, growth and mortality of the little skate off the northeast coast of the United States. Trans. Am. Fish. Soc. 113:314-321. Winemiller, K. O., and K. A. Rose. 1992. Patterns of life history diversification in North American fishes: implication for population regu- lation. Can. J. Fish. Aquat. Sci. 49:2196-2218. Wintner, S. P., and G. Cliff. 1996. Age and growth determination of the blacktip shark, Carcharhinus limbatus, from the east coast of South Africa. Fish. Bull. 94 (11:135-144. Zeiner, S. J., and P. G. Wolf. 1993. Growth characteristics and estimates of age at maturity of two species of skates (Raja binoculata and Raja rhina) from Monterey Bay, California. In Conser- vation biology of elasmobranches (S. Branstetter, ed.), p. 87-90. NOAA Tech. Report NMFS 115. 169 Abstract — Age estimates for striped trumpeter (Latris lineata) from Tas- manian waters were produced by counting annuli on the transverse section of sagittal otoliths and were validated by comparison of growth with known-age individuals and modal progression of a strong recruit- ment pulse. Estimated ages ranged from one to 43 years; fast growth rates were observed for the first five years. Minimal sexual dimorphism was shown to exist between length, weight, and growth characteristics of striped trumpeter. Seasonal growth variability was strong in individuals up to at least age four, and growth rates peaked approximately one month after the observed peak in sea surface temperature. A modified two-phase von Bertalanffy growth function was fitted to the length-at-age data, and the transition between growth phases was linked to apparent changes in physiological and life history traits, including offshore movement as fish approach maturity. The two-phase curve was found to represent the mean length at age in the data better than the standard von Bertalanffy growth function. Total mortality was estimated by using catch curve analysis based on the standard and two-phase von Bertalanffy growth functions, and estimates of natural mortality were calculated by using two empirical models, one based on longevity and the other based on the parameters L, and k from both growth functions. The interactions between an inshore gillnet fishery targeting predominately juveniles and an off- shore hook fishery targeting predomi- nately adults highlight the need to use a precautionary approach when developing harvest strategies. Age validation, growth modeling, and mortality estimates for striped trumpeter (Latris lineata) from southeastern Australia: making the most of patchy data Sean R. Tracey Jeremy M. Lyle Marine Research Laboratories Tasmanian Aquaculture and Fisheries Institute Private Bag 49 Hobart 7001, Tasmania, Australia E-mail address (for S R Tracey): straceyigutas edu au Manuscript submitted 20 December 2003 to the Scientific Editor's Office. Manuscript approved for publication 7 September 2004 by the Scientific Editor. Fish. Bull. 103:169-182 (2005). Striped trumpeter (Latris lineata) are widely distributed around the tem- perate latitudes of southern Austra- lia, New Zealand (Last et al., 1983), the Gough and Tristan Da Cunha Island groups in the southern Atlan- tic Ocean (Andrew et al, 1995), and the Amsterdam and St. Paul Island groups in the southern Indian Ocean (Duhamel, 1989). They are opportu- nistic carnivores associated with epi- benthic communities over rocky reefs at moderate depths from 5 to 300 m along the continental shelf. The spe- cies can grow to a relatively large size, 1200 mm in total length and 25 kg in weight (Gomon et al., 1994). Spawning apparently occurs offshore, and females are highly fecund mul- tiple-spawners (Furlani and Ruwald, 1999). Although there have been a number of ichthyoplankton surveys in Tasmanian waters, only a few striped trumpeter larvae have been collected, caught during the late austral winter through early spring months at near- shore (30-50 m) and shelf edge sites (-200 m) (Furlani and Ruwald, 1999). Larval rearing trials have shown that the presettlement phase is complex and extended; individuals can remain in this neritic-pelagic phase for up to 9 months after hatching before metamorphosis into the juvenile stage takes place (Morehead1). As juveniles striped trumpeter settle on shallow rocky reefs. In Tasmania striped trumpeter are taken commercially over inshore reefs (5 to 50 m), generally as a by- catch of gillnetting, and are targeted with hook methods (handline, drop- line, longline, and trotline) on deeper reefs (80 to 300 m). Small, subadult individuals dominate inshore catches, whereas larger individuals are taken from offshore reefs. In recent years the combined annual commercial catch has been typically less than 100 metric tons (Lyle2). Striped trumpeter also attract significant interest from recreational fishermen, who use both hooks and gill nets. Furthermore, the aquaculture potential for this species is currently being assessed in Tasmania (Furlani and Ruwald, 1999; Cobcroft et al., 2001). Despite wide interest in this species, there is a general paucity of information on age, growth, and stock structure of wild populations. Assessing the growth of a species is a fundamental part of fisheries popu- lation dynamics. Ever since Beverton and Holt (1957) introduced the von Bertalanffy growth model to fisheries research it has become ubiquitous in descriptions of the increase in size 1 Morehead. D. 2003. Personal commun. Tasmanian Aquaculture and Fisheries Institute, Univ. Tasmania. GPO Private Bag 49, Hobart, Tasmania, Australia 7001. 2 Lyle. J. M. 2003. Tasmanian scale- fish fishery — 2002. Fishery assessment report of the Marine Research Laborato- ries, Tasmanian Aquaculture and Fish- eries Institute, Tasmania. [Available from TAFI GPO Private Bag 49, Hobart, Tasmania, Australia 7001.] 170 Fishery Bulletin 103(1) of a species as a function of age. The parameters com- mon to the von Bertalanffy growth function (VBGF) are used in stock assessment models such as empiri- cal derivatives of natural mortality (Pauly, 1980) and assessments of yield per recruit and spawning stock biomass (Beverton and Holt, 1957). Despite the von Bertalanffy growth parameters being well established as cornerstones of many stock assessment models, sev- eral authors have highlighted limitations of the original derivation of the growth function to adequately repre- sent growth of a population (Knight, 1968; Sainsbury, 1979; Roff, 1980; Schnute, 1981; Bayliff, et al, 1991; Hearn and Polacheck, 2003). This limitation becomes especially evident with limited or patchy data. The limitations of the von Bertalanffy growth function have created three scenarios: 1) use the VBGF and retain the use of the parameters to derive per-recruit estimates at the possible expense of physiological integrity; 2) derive or employ a model that is not based on the von Bertalanffy parameters (such as a linear or logistic model) or another polynomial function (for instance, the Gompertz equation [Schnute, 1981]) and in doing so the expediency of the von Bertalanffy parameters in stock assessments is compromised; 3), use or develop an extension of the von Bertalanffy equation with the caveat that, by introducing additional parameters, the problem of reduced parsimony by over parameterisation would need to be considered. While investigating the life history characteristics of striped trumpeter, we became aware that the original description of the VBGF would not adequately represent growth of this species, in part because of the patchy data available for analysis. This study aims to describe the age and growth of striped trumpeter from Tasmania. Seasonal growth oscillations are considered for the first four years by using actual length-at-age data from a strong cohort. We then employ and evaluate an extension of the VB- GF that offers a better fit to the sample population of aged individuals and allows the flexibility of assigning representative growth and mortality parameters to different life phases of the population. Growth param- eters derived from both the standard von Bertalanffy and extended von Bertalanffy models are used in our catch curve analyses, and the empirical models of Pauly (1980) and Hoenig (1983) are used to allow comparison of mortality estimates. Materials and methods Striped trumpeter were collected opportunistically from various sites off the east and southeast coasts of Tasma- nia from a variety of fisheries dependent and indepen- dent sources spanning the period 1990-2002 (Table 1). Inshore catches were predominately taken with gill nets ranging in mesh sizes from 64 to 150 mm. Offshore catches were taken by hook-and-line methods. Samples ranged from intact specimens, for which the full range of biological information was collected, to processed frames Table 1 Composition of Tasmanian sampling data from 1990 through 2002 showing data from inshore gill net and off- shore hook fisheries. Numbers in parentheses represent the number of individuals aged from each particular sam- pling regime. Year Gill net Hook Total 1990 45 45 1991 — 332 332 1992 — 126 126 1994 3 8 11 1995 228 12 240 1996 529 55 585 1997 193 2 195 1998 7 171 178 1999 205 902 1107 2000 — 91 91 2001 — 60 60 2002 — 97 99 Total 1165 1901 3069 (268) (508) (776) from which length and, depending on condition of the body, sex and gonad weight were recorded. All specimens were measured for fork length (±1 mm) and, where pos- sible, total weight was recorded (±1 gram). Otoliths were collected when possible. This ad hoc sampling approach created a temporally irregular data set. Kolmogorov-Smirnov tests (o=0.05) were used to de- termine whether significant differences existed between male and female length-frequency distributions or be- tween length-frequency distributions by depth strata. Analysis of residual sums of squares (Chen, 1992) was used to determine whether a significant difference existed between the sex-specific length-weight rela- tionships that were fitted by minimizing the sum of square residuals and that are described by the power function W=aLb, (1) where W = whole weight (g); L = fork length (mm); and a and b = constants. Sex ratios were compared for significant deviation from 1:1 by chi-square tests. Aging technique Sagittal otoliths were removed from 873 individuals and a subsample of 295 otoliths were individually weighed to the nearest milligram. One randomly selected oto- lith from each fish was embedded in clear polyester casting resin. A transverse section was taken through Tracey and Lyle: Age validation, growth modeling, and mortality estimates for Latris lineata 171 the primordial region (width approximately 300 ^m) and mounted on a microscope slide. A stereo dissector microscope at 25 x magnification was used to aid the interpretation of increments in the mounted sections. Increment measurements were made by using Leica IM " image digitization and analysis software (Leica Micro- systems, Wetzlar, Germany). All counts and increment measurements were made without knowledge offish size, sex, or date at capture to avoid reader bias. Position of the first annual increment was determined by testing the close correspondence of otolith micro- structure and body size between known-age individuals reared from eggs in aquaria and wild-caught specimens. To ensure that growth in cultured individuals also re- flected growth in wild specimens, a hypothesis of com- parable growth was tested by fitting traditional VBGFs to the length-at-age data of 288 cultured individuals (maximum known age: 4 years) and 268 wild specimens (maximum otolith-derived age: 4 years). A likelihood ratio test (Kimura, 1980) was then used to test for sig- nificance. The VBGF model was in the form sampled from the strong 1993 cohort over the period 1995 through 1997, where the model was described as L, = LAI - ts 3 Data sourced from the NOAA-CIRES Climate Diagnostics Center, Boulder, CO 80305. http://www.cdc.noaa.gov/. [Accessed 15 Sep. 2002) 172 Fishery Bulletin 103(1) where Lxl, kv t01 VBGF parameters applied to the first growth phase; Ll2, k2, t02 = VBGF parameters applied to the second growth phase; Ls = length of transference from one growth phase to the next; and ts = age of transference from one growth phase to the next; calculated as. * =v In A L, (5) Having fitted Equation 4, we smoothed the discon- tinuity from the first growth stanza to the second, as- suming normal distribution around the age at transfer- ence by integrating a normal probability cumulative distribution function (PDF) where the mean is equal to the age of transference (4.4 years) and where the standard deviation is arbitrarily set at 1.0. This model is referred to as the two-phase von Bertalanffy growth function (VBGFTP) and is now represented as ♦''' i -■<-(„, f)\ ,1, Os/2k (L_1(l-e-*'"-'»,)+£)+ , (6) (*•. " 1 ', o-n/2/t (L6+(L,2-L')a-e-k'"'"') + e) where tmax = maximum age present in the sample; and a2 = standard deviation of cumulative density function with mean t6. The model that best represented the data was judged on a combination of parsimony as determined by the Akaike information criterion (AIC) (Akaike, 1974), qual- ity of fit by minimization of the negative log-likelihood value derived from each model, visual inspection of the residuals, and as an index of fit, the percent deviation of Lx for each model from the maximum observed length The hypothesis of sexual dimorphism in growth was tested by using likelihood ratio tests (Kimura, 1980) for both the VBGFS and VBGFTP models fitted to the length-at-age data of all individuals whose sex had been determined. Mortality estimation Mortality estimates were calculated by using the param- eters of both the VBGFS and VBGFTP functions. An esti- mate of instantaneous rate of total mortality (Z) for the offshore hook fishery was calculated for 1998 by applying a length converted catch curve analysis (LCCCA sensu Pauly, 1983) to the length-frequency data. Estimates of instantaneous rate of natural mortality (M) were calculated by using two empirical equations. The first equation, derived by Pauly (1980), is described log10M = -0.0066 - 0.279 log^L^y + 0.6543 log10 ky + 0.4634 log10 T, (7) where L„ and ky = parameters derived from the VBGFS or from the second growth phase of the VBGFTP; and T = average annual sea surface tempera- ture (°C) at the area of capture. The mean annual sea surface temperature on the east coast of Tasmania in 1998 was estimated as 14°C (NOAA-CIRES3). The second equation used was the regression equation of Hoenig (1983): In Z = 1.46 - 1.01 In tmax; M~Z assuming F~0, (8) where tmax = the maximum age for the species in years. Estimates of fishing mortality (F) were calculated by subtracting natural mortality from total mortality. Results Males ranged in length from 203 mm to 815 mm (n = 504) and females ranged from 269 mm to 950 mm (n = 565). Length-frequency distributions did not differ signifi- cantly between sexes (Kolmogorov-Smirnov; Z = 0.91 P=0.38). Pooling the length-frequency data of all individuals produced a bimodal frequency distribution. However, when grouped by depth (Fig. 1), the data revealed a significant depth-based stratification between the shal- low (<50 m stratum) and the deeper strata (Kolmogorov- Smirnov; Z=13.8 P<0.001), occurring at around 450 mm in length. Analysis of residual sums of squares indicated no significant difference between the sex-specific length- weight relationships (F=0.02 df=2 P=0.10); consequently a power regression was applied to the length-weight data of all individuals combined (Table 2). The sex ratio of males to females (1.0:1.3) from the inshore net fishery showed a low level of significant difference from 1:1 (x" = 3.88 P=0.049 n=232), whereas, the ratio of males to females (1.0:1.1) caught from the offshore hook fishery did not show significant difference from 1:1 (x~ = 0.933 P=0.334 n = 840). Age estimates Age was successfully estimated for 776 (89%) individu- als. Transverse otolith sections showed typical distinct alternate light and dark zone formations within the Tracey and Lyle: Age validation, growth modeling, and mortality estimates for Latns lineata 173 30 25 20 15 10 5 0 30 25 20 15 10 5 0 30 25 20 15 10 5 0 30 25 20 15 10 5 0 Length (Latris Depth: 0 - 50 m n = 1039 K Depth: 51 - 100 m n = 244 J\^n\ P — — i I i rrr-| cC. Depth: 101 - 150 m n=246 Tr-fl^TTr^ n Gill net □ Hook and line Depth: 151 -200 m n= 147 Qh j~|hi-^ t n n — i 1 1 1 — 0 100 200 300 400 500 600 700 800 900 1000 Length class (20-mm bin category) Figure 1 -frequency distribution by 50-m depth strata for striped trumpeter lineata) samples collected from 1990 through 2002. Table 2 Predictive equations used to compare weight and length, otolith weight and age, and reader variability across age classes, for striped trumpeter (Latris lineata). Dependent variable Independent variable n Equation r2 Weight ( W I Fork length IL) 491 ff = 2x 10"5 x L3-00 0.99 Otolith weight (OW) Age it) 295 OW = 7.32 + (1.70 xn 0.89 Primary reader, count 2 (P9) Primary reader, count 1 (Pj) 339 P2 = 0.05 + (0.99 xP,) 0.99 Secondary reader, count 1 (Sj) Primary reader, count 1 (P,) 46 S, =0.27 + 10.97 xP,) 0.97 174 Fishery Bulletin 103(1) B 0 Figure 2 Photomicrograph of transverse otolith sections of striped trumpeter (Latris lineata) from (A) a 5-year-old male (515 mm, FL), and (B) a 15-year-old female (724 mm, FL), using transmitted light. Scale bar = 1 mm. otolith matrix. Viewed under transmitted light the zones showed as dark (opaque) and light (translucent) (Fig. 2). A robust linear relationship existed between otolith mass and individual age (Table 2). The core area of each section consisted of an opaque region. Immediately adjacent to this was a faint thin translucent zone followed by the first broad opaque an- nual increment. In some cases the transition from core to the first expected increment could not be discerned because of a continuation of the opaque region (the expected thin translucent zone was too faint to see). In such cases, increment measurements were required to ensure that the annulus was not overlooked. Mean increment radius (±SD) from the primordia to the first annulus was 491 ±63 f ■14 W -13 B 200 - t V v - 12 Jan 95 Jul 95 Jan 96 Jul 96 Jan 97 Jul 97 Jan 98 Date Figure 5 Length-at-age data (O) of the 1993 striped trumpeter (Latris lineata) cohort fitted with a modified von Bertalanffy growth function to rep- resent seasonal growth (black line), plotted against a 7-day average SST at the time of sampling (gray line). 19-1 r 0 15 c o g 18- /' i ZJ /i/\ /A /\ - 0.10 > CD J / \ / \ / V \ 3 C 17 - 1 w \ / \ / \ -g_ (/) 1 ;i \ / \ Is \ .C \J\ \ A \ W/A \ c - 0.05 9- i 16- y\\ /A \ // \ \ o "O / \ \ / / \ \ // \ \ (1) / l\ \ /// A \ // A \ tfl — 15 - / i\ \ /l / M \ // Sv \ / i \ / i / V<\ \ /l / ^t \ - 000 ^ o \A \ /J/ u \ /(/ \ \ (Q \ \ U\ \ \ hi \ \ o H 14- w \ is 1 A \ /i/ \\ £ 01 X \ ' a \ / '/ v\ - -0.05 =r w \w V\// Wi c 13 - o CD \w \#/ \>y E --0.10 § » 12- \'\| ra j TJ K 11 Jan 95 Jul 95 Jan 96 Jul 96 Jan 97 Jul 97 Jan 98 Date Figure 6 Seven-day mean SST (broken gray line) fitted with a sine wave (gray line) plotted against the sine function (black line) extracted from the seasonal von Bertalanffy growth function fitted in Figure 5. supports the hypothesis that a more complex growth model was required for striped trumpeter. The VBGFTP was sensitive to the value age at trans- ference. A profile of negative log likelihood for a range of age-at-transference values (Fig. 8) assisted in determin- ing the correct absolute minima. The negative log-like- lihood profile revealed a low minima range across age at transference from 3.5 to 4.6 years, which, however, converged to a lowest value at age 4.4 years. Fitting the PDF to the growth curve substantially smoothed the point of transition, producing a curve that represented the data well. Setting an arbitrary standard devia- tion of 1.0 around the age at transference provided a normally distributed two-tailed range at transference (90 percent confidence adjusted for bias) from 1.3 to 7.8 years. 178 Fishery Bulletin 103(1) A likelihood ratio test (LRT) identified a slight sig- nificant difference between male and female VBGFS growth curves (/2=13.20 df=3 P=0.04), but there was no significant difference when the VBGFTP was tested (X2=10.83 df=6P=0.09). Mortality estimation Ages 9-23 and 7-25 were included in the LCCCA regres- sions of the VBGFS and the VBGFTP, respectively, to n = 776 100 10 15 20 25 Age (yrs) 30 35 40 Figure 7 Pooled length-at-age data for striped trumpeter (Latris lineata). The black line represents the optimal two-phase von Bertalanffy growth function (VBGFTP), with a mean age at transference of 4.4 years and a standard deviation equal to 1; the gray line represents the optimal standard von Bertalanffy growth function (VBGFS). estimate Z (Fig. 9). Individuals below these ranges were assumed, by their respective model, not to have fully recruited to the offshore fishery, and individuals over the age of 25 were excluded due to poor sample size. These age ranges effectively excluded the strong 1993 recruitment pulse from the regression, thereby avoiding the complication of including a known strong year class in the analysis. Application of the VBGFTP model resulted in lower estimates of Z and M (based on the Pauly equation), compared with those calculated by using the VBGFS parameters (Table 4). The estimate of Z based on the Hoenig (1983) equation was assumed to be close to M because F is low for this species. The Hoenig M was very similar to the Pauly estimate when VBGFTP parameters were used. In this case M was just below 0.1, indicating an annual natural mortality rate of about 9%. The VBGFTP estimates indicate that F was slightly higher than M in the offshore fishery. By contrast, the standard VBFGS parameters produced a substantially higher estimate of M (0.15) based on the Pauly equation than predicted by the Hoenig approximation, indicating an annual natural mortality rate of about 14%. Derived estimates of F with the VBGFTP were slightly higher than M, whereas F in relation to M was variable for the VBGFS, depending on the equation used to derive M. 45 2 3 4 5 6 7 8 9 10 1 1 12 13 14 15 16 17 18 19 Age at transference Figure 8 Negative log-likelihood profile plot of increasing age-at-transference values for striped trumpeter iLatris lineata). Discussion The present study represents the first report of age and growth of striped trumpeter. Despite having available a patchy data set, we were able to validate age and overcome the limitations of the von Bertalanffy equa- tion to represent these data by the use of a robust growth model. Striped trumpeter are long lived, have a maximum age in excess of 40 years, and growth is particularly rapid up to age five, after which it slows dramatically. The species has a complex early life his- tory involving a long planktonic larval phase of around nine months (Morehead1), an in- shore juvenile phase, and then movement offshore into deepwater. Gear selectivity (gill nets in the shallow and hook catches in the deeper waters) may have influenced the fish-size structure of our samples, especially when grouped by depth, although it is highly unlikely that the size differences could be completely attributed to gear type alone. For instance, small indi- viduals (<400 mm) were occasionally taken by hooks in the deeper strata and individu- als over 500 mm were taken by gill nets in less than 50 m. The commercial hook fishery Tracey and Lyle: Age validation, growth modeling, and mortality estimates for Latns lineata 179 I 6- CD O) CO 61 ° • y = -0.253* + 6.881 7 • r2 = 0.853 5 . o°» y = -0.192x + 5.977 • r2 = 0.860 O) c CO o 4 _ 0 ^W •N. 4 •^. irithm (frequency [F • >v 3 o N^ o • »^ 0 ^s. • ^v • ^"S. a, 1 - o \ 1 - * 2 A B m D - z 0 5 10 15 20 25 Ane fvearsi Figure 9 Length-converted catch curve analysis for striped trumpeter iLatris lineata) length and age data from 1998. (A) Age composition was based on the standard von Bertalanffy growth function (VBGFS), (B) age composition was based on the second stanza of the two-phase von Bertalanffy growth function (VBGFTp). Solid points were included in the respective linear regressions. for striped trumpeter is largely restricted to depths of greater than 50 m, and despite considerable hook-fish- ing effort at shallower depths targeting other demersal reef species, notably the wrasses Notolabrus fucicola and N. tetricus, minimal catches of striped trumpeter are taken and those that are caught tend to be small in size (Lyle2). Rather, size structuring by depth is believed to reflect the movement of striped trumpeter offshore into deeper water as they grow and mature. Seasonal growth was dramatic in young striped trum- peter (Fig. 5 1. This phenomenon is common in temper- ate species (Haddon, 2001; Jordan, 2001; McGarvey and Fowler, 2002), and has been linked to fluctuations in environmental factors, such as water temperature and oceanographic conditions, as well as biotic factors, such as seasonality in primary productivity (Harris et al., 1991; Jordan, 2001). Our study supports a correla- tion between water temperature and seasonal growth (Fig. 6); maximum growth was observed to take place consistently over a three-year period, approximately one month after the peak sea-surface temperatures. Knowledge of growth and growth variability is es- sential to the understanding of a stock's population dynamics. To achieve an accurate assessment of these characteristics, several issues need to be addressed. Foremost, is a rigorous approach to the validation and precision testing of age estimates (Campana, 2001). In this study, a combination of age validation protocols outlined by Fowler and Doherty (1992) and Campana (2001) were subscribed to: 1) otoliths must display an internal structure of increments, (Fig. 3); 2) otoliths must grow throughout the lives of fish at a perceptible rate, which was confirmed by the otolith weight-at-age Table 3 Parameter estimates derived from the ;wo growth functions (standard von Berta anffy growth function, [VBGFS] and the two-phase von Bertala nffy growth function tVBGFTP|) applied to the length-at -age data of striped trumpeter tLatris lineata) in Tasmania. Growth parameters are defined in the text, NOP = = number of parameters in the model, AIC = Akaike information cri- terion, and Lmax = the maximum length of all individuals included in th* growth models. VBGFS VBGFTP Growth i-l 773.27 532.77 parameters *i 0.15 0.43 'oi -1.46 0.03 L" — 450.11 i-2 — 871.59 *2 — 0.08 ?02 — 3.49 r" — 4.4 a2 oft" — 1.0 Diagnostics NOP 3.0 9.0 -log likelihood 3759.98 3700.13 AIC 5335.12 5211.30 % deviation of -13.7 -2.7 Lm2bomLmal regression (Table 2); 3) the age of first increment forma- tion must be determined; and 4) increment periodicity across the entire age range of interest must be veri- 180 Fishery Bulletin 103(1) Table 4 Estimates of instantaneous rates of total (Z), natural (M), and fishing IF) mortality for striped trumpeter (Latris lineata) deter- mined with age-based catch curve analysis and the empirical equations of Hoenig ( 1983 ) and Pauly ( 1980 ). VBGFTP = estimates derived from the parameters of the two-phase von Bertalanffy growth function, VBGFS = estimates derived from the parameters of the standard von Bertalanffy growth function and LCCCA = length converted catch curve analysis. Z M F Method VBGFS VBGFTP VBGFS VBGFTP VBGFS VBGFTP LCCCA 0.253 0.192 — Hoenig 0.096 Pauly 0.151 0.096 0.157 0.092 0.102 0.096 0.100 fled. We used cultured individuals to determine which opaque or translucent zone represented the first growth increment, although the accuracy of age validation with cultured individuals has been questioned by Campana (2001). In our study, the close correspondence between the growth of cultured and wild fish over a period of several years gives us confidence in using this approach to validate first increment position. The slightly slower growth rate observed in cultured striped trumpeter can be attributed to jaw malformation — a phenomenon that has been shown to affect feeding ability (Cobcroft et al., 2001). Modal progression of the 1993 cohort through time provided indirect validation for annual periodicity in in- crement formation up until age seven. Validation across all age classes was not possible in our study, although validation after the age of five years was significant. That is, validation was achieved past the average age at which fish moved offshore into deeper water, and past the age at which there was a significant reduction in growth rate. The second consideration to address when studying animal growth is model selection. Akaike's information criterion is a standard method for model selection that provides an implementation of Occam's razor, in which parsimony or simplicity is balanced against goodness-of- fit (Forster, 2000). However, model selection should not rely on statistical fit alone; it should also provide a bio- logically sensible interpretation across the entire range of ages in the sampled population (Haddon, 2001). In the case of striped trumpeter, the standard von Berta- lanffy function provided a poor representation of growth in older individuals, resulting in an unrealistically low L r. This problem was largely overcome by the applica- tion of a two-phase growth function. Similar to that used on large pelagics, such as Thunnus maccoyii (Bay- liff et al., 1991; Hearn and Polacheck, 2003). In their application of the model, Hearn and Polacheck (2003) considered biological traits when discussing the justi- fication for age at transference, namely the reduction in growth rate, and inshore to offshore migration. In the present study we have considered analogous traits to seed the age of transference for striped trumpeter. In this species there is a marked transition in size structure between shallow and deeper reefs that occurs at around 450 mm or between 4 and 5 years (Fig. 8). In addition, a visual assessment of the length-at-age data highlighted a marked decrease in growth rate at a similar age. Solving for the age at transference produced a point estimate that results in a sharp discontinuity in the growth curve; an observation that Hearn and Polacheck (2003) highlighted as biologically unrealistic. The range of low negative log likelihood values described by the age at transference profile is due to the patchiness of data around these ages, creating uncertainty in the model. We have assumed in this case that the converged value of 4.4 years is accurate and that the variability around this point is normally distributed with a stan- dard deviation equal to one. By including the normal probability distribution function we have effectively created a smooth transition between growth phases. This function implies that age at transference has some level of inherent variability, which is likely to be more biologically plausible than knife-edge transition. A further extension of the two-phase model was test- ed by applying the seasonal growth version of the VBGF (described in Eq. 3) to the first phase and a standard VBGF to the second phase, but was disregarded because of the effect of over parameterization on parsimony. However, this approach did highlight the flexibility of the two-phase model to allow for a more dynamic rep- resentation of population growth characteristics. This study supports the assertion by Hearn and Po- lacheck (2003) that discontinuity in growth rate may be a more common phenomenon in fish than implied by growth models reported in the literature. Such a two-phase growth model, where age at transference coincides with the transition phase from one fishery to another, has proven useful. It allows separate growth parameters to be tracked to each fishery, and as such, provides a precursor to developing a more biologically robust production model with dynamic parameters at age and for fishing method. The predictive regression developed by Pauly (1980) that estimates natural mortality is based on the direct Tracey and Lyle: Age validation, growth modeling, and mortality estimates for Latns lineata 181 relationship between longevity (.tmax) and the magnitude of the physiological growth parameters k and Lr. As such, it would be reasonable to assume that if a good fit exists between length at age that the growth param- eters, when employed in such an empirical model, would yield a natural mortality estimate approximately equal to that determined by a regression model that is based on tmax (Hoenig, 1983). The two-phase growth function also provided a more conservative estimate of M than the standard von Bertalanffy model. Overestimates of M can lead to unrealistically high estimates of produc- tivity and a potential yield that may in turn lead to overexploitation of a stock. Protracted longevity, slow growth in later life, large body size, recruitment variability, and relatively low natural mortality once individuals reach adulthood are all characteristics typical of a K-selected species (where equilibrium is the biological strategy). Such species are often regarded as being susceptible to growth over-fish- ing and stock depletion (Booth and Buxton, 1997). For instance, increased fishing effort on the inshore fishery, as has been observed with the recruitment of strong cohorts, will affect subsequent recruitment to the off- shore fishery and spawning stock. The current analysis indicates that fishing mortality is slightly higher than natural mortality and, in the absence of further strong recruitment, a decline in the stock size is likely if fish- ing pressure is not reduced. Acknowledgments The authors gratefully acknowledge Ray Murphy and Alan Jordan who collected many of the earlier samples and undertook preliminary examination of the otoliths. The assistance of the captain and crew of FRV Challenger in collecting samples is also thankfully acknowledged. Philippe Ziegler, Dirk Welsford, and Malcolm Haddon provided constructive criticism and ideas in terms of the analyses and reviewed the manuscript; Sarah Irvine and an anonymous reviewer provided constructive feedback on final versions of this manuscript. Literature cited Akaike, H. 1974. A new look at the statistical model identification. Institute of Electrical and Electronic Engineers Transac- tions on Automatic Control, AC-19, p. 716-723. IEEE Control Systems Society, New York, NY. Andrew, T. G., T. Hecht, P. C. Heemstra, and J. R. E. Lutjeharms. 1995. Fishes of the Tristan Da Cunha group and Gough Island, South Atlantic Ocean. JLB Smith Institute of Ichthyology. Ichthyol. Bull. 63:1-41. Bayliff, W. H.. I. Ishizuka, and R. B. Deriso. 1991. Growth, movement, and attrition of northern blue- fin tuna, Thunnus thynnus, in the Pacific Ocean, as determined by tagging. Inter-Am. Trop. Tuna Comm. Bull. 20(11:1-94. Beamish, R. J., and D. A. Fournier. 1981. A method for comparing the precision of a set of age determinations. Can. J. Fish. Aquat. Sci. 38: 982-983. Beverton, R. J. H , and S. J. Holt. 1957. On the dynamics of exploited fish populations. U.K. Ministry of Agriculture and Fisheries, Fisheries Inves- tigations (series 2), 19:1-533. Booth, A. J. and C. D. Buxton. 1997. Management of the panga Pterogymnus laniarius (Pisces: Sparidae), on the Agulhas Bank, South Africa using per-recruit models. Fish. Res. 32:1-11. Campana, S. E. 2001. Accuracy, precision and quality control in age deter- mination, including a review of the use and abuse of age validation methods. J. Fish. Biol. 59:197-242. Campana, S. E„ M. C. Annand, and J. I. Mcmillan. 1994. Graphical and statistical methods for determining the consistency of age determinations. Trans. Am. Fish. Soc. 124(11:131-138. Chen, Y, D. A, Jackson, and H. H. Harvey. 1992. A comparison of von Bertalanffy and polynomial functions in modelling fish growth data. Can. J. Fish. Aquat. Sci. 49:1228-1235. Cobcroft, J. M., P. M. Pankhurst, J. Sadler, and P. R. Hart. 2001. Jaw development and malformation in cultured striped trumpeter Latris lineata. Aquaculture 199(3-4): 267-282. Duhamel, G. 1989. Ichtyofaune des iles Saint-Paul et Amsterdam (Ocean Indien sud). Mesogee. 49:21-47. Forster, M. R. 2000. Key concepts on model selection: performance and generalizability. J. Math. Psych. 44:205-231. Fowler, A. J., and P. J. Doherty. 1992. Validation of annual growth increments in the otoliths of two species of damsel fish from the southern Great Barrier Reef. Aust. J. Mar. Freshw. Res. 43: 1057-1068. Furlani, D. M., and F. P. Ruwald 1999. Egg and larval development of laboratory-reared striped trumpeter Latris lineata (Forster in Bloch and Schneider 1801) (Percoidei: Latridiidae) from Tasmanian waters. N.Z. J. Mar. Freshw. Res. 33:16-83. Gomon, M. F„ J. C. M. Glover, and R. H. Kuiter. 1994. The fishes of Australia's south coast, 992 p. The Flora and Fauna of South Australia Handbooks Committee. State Printers, Adelaide, South Australia, Australia. Haddon, M. 2001. Modelling and quantitative methods in fisheries. Chapman and Hall, Boca Raton, FL. Harris, G. P., F. B. Griffiths, L. A. Clementson, V. Lyne, and H. Van der Doe. 1991. Seasonal and interannual variability in physical processes, nutrient cycling and the structure of the food chain in Tasmanian shelf waters. J. Plankton Res. 13 (Suppl. 1:109-131. Hearn, W. S., and T. Polacheck. 2003. Estimating long-term growth-rate changes of south- ern bluefin tuna {Thunnus maccoyii) from two periods of tag-return data. Fish. Bull. 101:58-74. Hoenig, J. M. 1983. Empirical use of longevity data to estimate mor- tality rates. Fish. Bull. 82:898-902. Jordan, A. R. 2001. Age, growth and spatial and interannual trends in age composition of jackass morwong, Nemadactylus 182 Fishery Bulletin 103(1) macropterus, in Tasmania. Aust. J. Mar. Freshw. Res. 52:651-660. Kimura, D. K. 1980. Likelihood methods for the von Bertalanffy growth curve. Fish. Bull. 77:765-776. Knight, W. 1968. Asymptotic growth: an example of nonsense disguised as mathematics. J. Fish. Res. Board Can. 25:1303-1307. Last. P. R.. E. O. G. Scott, and F. H. Talbot. 1983. Fishes of Tasmania, 563 p. Tasmanian Fish-eries Development Authority, Hobart, Tasmania, Australia. McGarvey, R., and A. J. Fowler 2002. Seasonal growth of King George whiting {Sillagi- nodes punctata) estimated from length-at-age samples of the legal-size harvest. Fish. Bull. 100:545-558. Pauly, D. 1980. On the interrelationships between natural mor- tality, growth parameters, and mean environmental temperature in 175 fish stocks. J. Cons. Int. Explor. Mer 39(21:175-192. 1983. Length-converted catch curves: a powerful tool for fisheries research in the tropics (part 1). Fishbyte 1(2):9-13. Pitcher, T J., and P. D. M. MacDonald. 1973. Two models for seasonal growth in fishes. J. Appl. Ecol. 10:559-606. Reynolds, R. W., N. A. Rayner, T. M. Smith, D. C. Stokes, and W. Wang. 2002. An improved in situ and satellite SST analysis for climate. J. Climate 15:1609-1625. Roff, D. A. 1980. A motion for the retirement of the von Bertalanffy function. Can. J. Fish. Aquat. Sci. 37:127-129. Sainsbury, K. J. 1979. Effect of individual variability on the von Berta- lanffy growth equation. Can. J. Fish. Aquat. Sci. 37: 241-247. Schnute, J. 1981. A versatile growth model with statistically stable parameters. Can. J. Fish. Aquat. Sci. 38:1128-1140. 183 Abstract — Morphological develop- ment of the larvae and small juve- niles of estuary perch {Macquaria colonorum) 1 17 specimens, 4.8-13.5 mm body length) and Australian bass (M. novemaculeata) (38 specimens, 3.3-14.1 mm) (Family Percichthyidae) is described from channel-net and beach-seine collections of both species, and from reared larvae of M. novemac- uleata. The larvae of both are charac- terized by having 24-25 myomeres, a large triangular gut (54-67% of BL) in postflexion larvae, small spines on the preopercle and interopercle, a smooth supraocular ridge, a small to moderate gap between the anus and the origin of the anal fin, and distinctive pigment patterns. The two species can be distinguished most easily by the different distribution of their melanophores. The adults spawn in estuaries and larvae are presumed to remain in estuaries before migrating to adult freshwa- ter habitat. However, larvae of both species were collected as they entered a central New South Wales estuary from the ocean on flood tides; such transport may have consequences for the dispersal of larvae among estuar- ies. Larval morphology and published genetic evidence supports a reconsid- eration of the generic arrangement of the four species currently placed in the genus Macquaria. Larval development of estuary perch (Macquaria colonorum) and Australian bass (M novemaculeata) (Perciformes: Percichthyidae), and comments on their life history Thomas Trnski Amanda C Hay Ichthyology. Australian Museum 6 College Street Sydney, New South Wales 2010, Australia E-mail address (for T Trnski, senior author): tomt@austmus gov au D. Stewart Fielder New South Wales Fisheries Port Stephens Fisheries Centre Private Bag 1 Nelson Bay, New South Wales 2315, Australia Manuscript submitted 20 November 2003 to the Scientific Editor's Office. Manuscript approved for publication 15 June 2004 by the Scientific Editor. Fish. Bull. 103:183-194 (2005). The Percichthyidae is a family of freshwater fishes restricted to Aus- tralia (8 genera, 17 species) and South America (2 genera, 7 species) (John- son, 1984; Nelson, 1994; Allen et al., 2002; Paxton et al., in press). There is continuing debate regarding the mono- phyly of the family; several genera are variously allocated to separate fami- lies: Gadopsis is allocated to Gadop- sidae (Allen et al., 2002; see Johnson, 1984 for a history of the systematic placement of the genus) and Edelia, Nannatherina, and Nannoperca are allocated to Nannopercidae (Allen et al., 2002). Other Australian genera of Percichthyidae include Bostockia, Guyu, Maccullochella , and Macquaria (Pusey and Kennard, 2001; Allen et al., 2002; Paxton et al., in press). The genera Percolates and Plectroplites were synonymized with Macquaria, based on morphological and biochemi- cal characters (MacDonald, 1978), and although this arrangement was accepted by Paxton and Hanley (1989), Paxton et al. (in press), Eschmeyer (1998), Johnson (1984), and Nelson (1994) recognized both Percolates and Plectroplites as valid genera. There are four described species in the genus Macquaria, all confined to southeastern Australia. Macquaria ambigua occurs naturally in fresh- waters of the Murray-Darling river system and has been translocated outside of its natural range (Kai- lola et al., 1993; Allen et al., 2002). There is genetic evidence for an ad- ditional undescribed freshwater spe- cies closely related to M. ambigua from central Australian drainages (Musyl and Keenan, 1992). Mac- quaria australasica is also confined to freshwater of the Murray-Darling river system, and an isolated popu- lation exists from the Shoalhaven and Hawkesbury Rivers, New South Wales (Allen et al„ 2002) that may be a separate species (Dufty, 1986). The other two species (M. colonorum and M. novemaculeata) are catadromous and occur in coastal southeastern Australian drainages between south- ern Queensland and eastern South Australia (Paxton et al., in press). They are sympatric from northern New South Wales (NSW) to eastern Victoria. Adults of M. novemaculeata occur in freshwater, whereas M. colo- norum prefers brackish water of estu- aries (Williams, 1970). Both species migrate to estuarine areas to breed in winter (Allen et al., 2002). Both species are protected from commer- cial fishing but are highly prized by recreational fishermen (Harris and Rowland, 1996; Allen et al., 2002) and M. novemaculeata is an impor- tant aquaculture species. 184 Fishery Bulletin 103(1) Of the 17 Australian percichthyids, larvae of only Maccullochella macquariensis, M. peelii peelii, and Macquaria ambigua have been described (Dakin and Kesteven, 1938; Lake, 1967; Brown and Neira, 1998). Larval and early juvenile development of the estuary perch (Macquaria colonorum) and the Australian bass (Macquaria novemaculeata) is described from specimens collected from the central and southern coast of NSW, and from reared larvae of the latter species obtained from brood stock from central NSW. This is the first description of the morphological development of the early life history of these two species. Materials and methods Morphological definitions, measurements, and abbrevia- tions follow Neira et al. (1998) and Leis and Carson- Ewart (2000). Larvae and juveniles were examined and measured under a dissecting microscope at magnifica- tions from 6 to 50x. Precision of the measurements varied with magnification but ranged from 0.02 to 0.16 mm. Where morphometric values are given as a percent- age, they are as a proportion of body length (BL) unless otherwise indicated. All pigment described is external unless otherwise specified. The juveniles collected are in transition from larvae to juveniles because they retain some of their larval characters and squamation is incomplete; these are called "transitional juveniles" ( Vigliola and Harmelin-Vivien, 2001). Illustrations were prepared with a Zeiss SR with an adjustable drawing tube. Field-caught larvae were collected in a fixed 2-m2 channel net with about 1-mm mesh in Swansea Chan- nel, Lake Macquarie, central NSW. The net filtered surface waters to 1 m depth during night flood tides (Trnski, 2002). Small juveniles were collected in a 30-m beach seine dragged over sand, mud, and Zostera sea- grass in the Clyde River, southern NSW. Reared larvae of M. novemaculeata were obtained from rearing tanks at the Port Stephens Fisheries Centre, an aquaculture research facility of NSW Fisheries. Brood stock came from the Williams River, central NSW. All specimens were initially fixed in 10% formalin and subsequently transferred to 70% ethanol. Field-caught larvae were restricted to a narrow size range: 4.8-7.1 mm body length (BL) for M. colono- rum (n=12), and 4.6-7.6 mm BL for M. novemaculeata (n=15). Juveniles of both species ranged from 10.3 to 13.5 (n = 5) and from 10.1 to 14.1 mm BL (n = 5), respec- tively. Reared larvae of AT. novemaculeata were available to confirm the identification of the larvae and to extend the developmental series for this species to 3.3-10.2 mm BL (ra=18). All material examined is registered in the fish collec- tion at the Australian Museum. Registration numbers of M. colonorum larvae are AMS 1.20052-010, 1.41690- 005 to -008, 1.41691-002, 1.41692-001, 1.41693-001; M. novemaculeata are AMS 1.20052-012, 1.27051-013, 1.41561-001 to -008, 1.41590-001, 1.41641-001, 1.41661- 001 and -002, 1.41662-001, 1.41668-001, 1.41690-001 to -0004, 1.41691-001, 1.41694-001. Identification Field-caught larvae and juveniles were identified as per- cichthyids by using the characters in Brown and Neira (1998), particularly the combination of a relatively large gut, the small to moderate gap between the anus and origin of the anal fin prior to complete formation of the anal-fin, continuous dorsal fin, fin-ray, and vertebral counts, and head spination including small preopercular spines, a small interopercular spine, and a smooth supra- ocular ridge. The larvae and juveniles described here were confirmed as being Macquaria colonorum and M. novemaculeata because of their coastal distribution and meristics; all other species in the family are restricted to freshwater. The overlap in meristics between M. colo- norum and M. novemaculeata made separation of the species difficult. The availability of reared M. novemacu- leata from positively identified adults determined the species allocations. Results Development of Macquaria colonorum Adult meristic data Dorsal (D) IX-X,8-11; Anal (A) 111,7-9; Pectoral (Pj) 12-16; Pelvic (P2) 1,5; Vertebrae 25 17 specimens: 4.8-7.1 and 10.3-13.5 mm BL General morphology (Tables 1 and 2, Fig. 1) Larvae and transitional juveniles are moderately deep bodied (body depth, BD 30-35%). The body and head are lat- erally compressed. There are 24-25 myomeres (12-14 preanal and 11-13 postanal). The large, triangular gut is fully coiled in the smallest larva examined. The pre- anal length ranges from 60% to 67%. The conspicuous gas bladder located over the midgut is small to moder- ate in size but difficult to distinguish in transitional juveniles. The round to slightly elongate head is large (head length, HL 32-41%). The snout is slightly concave to straight. The snout is approximately the same length as the eye diameter but becomes shorter from 7 mm. The eye is round and moderate in size (27-32% of HL) in larvae but becomes moderate to large in transitional juveniles (32-36% of HL). The large mouth reaches to the middle of the pupil. Small canine teeth are present in both jaws in all larvae examined. The nasal pit closes shortly after settlement, by 12.5 mm. Head spination is weak. Three short spines are pres- ent on the posterior preopercular border in the small- est larva examined; a fourth spine is present in some postflexion larvae from 6.3 mm and in all transitional juveniles. The spine at the angle of the preopercle is longest but remains shorter than the pupil diameter. A minute interopercular spine is present from 6.0 mm and persists in all transitional juveniles. A low, smooth Trnski et al : Larval development of Macquana colonorum and M. novemaculeata 185 Table 1 Morphometric data for Macquaria colonorum la •vae from channel-net samples and juveniles from beach-seine samples. Measurements are in mm. VAFL = : vent to anal-fin length. Preanal Predorsal Body Head Snout Eye Body length length length depth length length diameter VAFL Flexion 4.80 3.40 2.48 1.49 1.96 0.58 0.58 0.04 5.10 3.40 3.00 1.60 1.88 0.60 0.56 0 5.40 3.40 2.80 1.60 1.72 0.50 0.50 0 5.48 3.32 2.91 1.74 1.80 0.56 0.56 0 Postflexion 5.73 3.49 2.80 1.99 2.08 0.56 0.60 0 5.98 3.68 3.24 1.99 2.04 0.60 0.60 0 6.00 3.72 2.60 1.92 2.00 0.50 0.60 0 6.31 3.98 2.57 2.16 2.20 0.60 0.68 0 6.60 4.00 3.00 2.20 2.40 0.64 0.64 0 6.81 4.15 3.07 2.16 2.32 0.66 0.66 0 7.00 4.32 3.32 2.08 2.24 0.60 0.72 0 7.10 4.36 2.91 2.32 2.28 0.60 0.72 0 Settled 10.29 6.81 4.81 3.15 3.74 0.91 1.25 0 10.62 6.81 4.98 3.24 3.90 0.95 1.25 0 11.29 7.47 5.56 3.74 4.48 1.00 1.58 0 12.45 7.97 5.64 4.15 4.57 1.00 1.66 0 13.45 8.70 6.64 4.48 5.23 1.41 1.83 0 Table 2 Meristic data for Macquaria colonorum arvae and juveniles. ( ) indicates only fin bases present, [ ] incipient rays or spines, 1 1 ray transforming to a spine d = damaged. Body length Dorsal Anal Pectoral Pelvic Caudal Myomeres Flexion 4.80 (V), 9 (D,8[l] 9+711] 14+11=25 5.10 d, (10) (I),9 [1]8+7[1] 13+11=24 5.40 d, (9) (II), 9 [2] 9+8 13+12=25 5.48 (III), 9 (II), 8 3 9+8 12+12=25 Postflexion 5.73 (IV), 11 (II), 8 5 9+8 13+12=25 5.98 (V), 10 (II),8 2 9+8 13+12=25 6.00 (IV), 10 (II), 8 3 9+8 13+12=25 6.31 (IV)I, 11 [II], 9 9 buds 9+8 13+12=25 6.60 IV, 11 11,9 5 buds 9+8 13+12=25 6.81 VII, 11 II, 10 9 buds 9+8 13+12=25 7.00 VII, 11 II, 10 5(d) buds 9+8 12+13=25 7.10 VIII, 10 11111,9 inn buds 9+8 14+11=25 Settled 10.29 VIII II), 10 11111,8 15 1,5 7+9+8+6 13+12=25 10.62 VIII III, 10 Hill, 8 13 1,5 7+9+8+4 12+13=25 11.29 IX, 10 111,8 15 1,5 7+9+8+8 12+13=25 12.45 IX, 10 111,8 14 1,5 12+9+8+7 12+13=25 13.45 IX, 10 111,9 14 1,5 9+9+8+8 12+13=25 186 Fishery Bulletin 103(1) A 4.8 mm B 7.1 mm C 10.3 D 12.5 mm Figure 1 Larvae of Macquaria colonorum. (A and B) postflexion larvae from Swansea Chan- nel, central New South Wales (NSW) (C and D) recently settled juveniles from the Clyde River, southern NSW. supraocular and supracleithral ridge form by the time notochord flexion is complete. A weak posttemporal ridge is present from 7 mm, and a small spine develops in transitional juveniles from 11.3 mm. A small spine develops on the supracleithrum from 10.6 mm. An oper- cular spine is present in transitional juveniles. Dorsal-fin soft rays are ossified by the completion of notochord flexion, the posteriormost rays being the last to ossify. The pterygiophores of the spinous rays of the dorsal fin develop from posterior to anterior and begin to form during notochord flexion. Spines begin to ossify in postflexion larvae by 6.3 mm, and the full comple- ment of dorsal-fin elements is present by 7.1 mm. All soft rays of the anal fin are ossified by the completion of notochord flexion, by which time 1-2 pterygiophores of the spinous rays are present. The first two anal-fin spines are ossified by 6.6 mm. The last spinous soft ray of the dorsal and the third spinous ray of the anal fin transforms from a soft ray after settlement and they are fully transformed by 11.3 mm. Incipient rays begin to form in the pectoral fin during notochord flexion, and the rays ossify from dorsal to ventral in postflex- ion larvae. A few pectoral-fin rays remain unossified at 7.1 mm and are fully ossified prior to settlement. Trnski et al.: Larval development of Macquana colonorum and M. novemaculeata 187 Pelvic-fin buds appear in postflexion larvae from 6.3 mm, but no elements have formed in the largest speci- men; they are all ossified in the transitional juveniles. All primary caudal-fin rays are ossified by the end of notochord flexion. Procurrent caudal rays are present in the transitional juveniles. Notochord flexion commences before 4.8 mm, and is complete by 5.7 mm. Scales have not begun to develop in the largest transitional juvenile examined (13.5 mm). Pigment (Fig. 1, A-D) Larvae are moderately to heav- ily pigmented; melanophores are concentrated on the dorsal and ventral midlines, and midlateral surface of the trunk and tail. Small expanded melanophores are present at the tips of the upper and lower jaws, and there are one or two melanophores ventral to the nasal pit. Additional internal melanophores are present along the roof of the mouth, and posterior to the eye below the mid- and hindbrain. External melanophores may be present on the operculum in line with the eye. One or two melanophores are present on the ventral midline of the lower jaw, and there is one at the angle of the lower jaw. Four to seven large, expanded melanophores are pres- ent along the dorsal midline of the trunk and tail, from the nape to just posterior to the dorsal-fin base. There are one or two melanophores on the nape and four or five along the dorsal-fin base. A series of large, expand- ed melanophores is present along the lateral midline of the trunk and tail, commencing at the gas bladder and extending to the posterior end of the dorsal and anal fins. In postflexion larvae, this series extends onto the anterior third of the caudal peduncle. Internal melano- phores are present over the gas bladder, the mid- and hindgut, and may be present along the notochord. The external and internal pigment series thus give the im- pression of a line of heavy pigment from the tip of the snout, across the head and trunk, to the tail. Small melanophores are present along the ventral midline of the gut; one melanophore on the isthmus immediately anterior to the cleithral symphysis, usually three (range: 2-4) melanophores between the cleithral symphysis and pelvic-fin base, and usually three (range: 1-4) melanophores between the pelvic-fin base and the anus. Expanded melanophores are present along the ventral midline of the tail, from above the anus to the posterior end of the anal-fin base. Between one and three melanophores occur along the anal-fin base. A small melanophore is occasionally present in early post- flexion larvae at the base of ventral primary caudal-fin rays 1-2. In transitional juveniles, the expanded melanophores are relatively smaller, and are most prominent midlat- erally along the trunk and tail. The expanded melano- phores along the dorsal and ventral midlines become small to absent during the juvenile stage. Additional ex- panded melanophores develop laterally on the head and body, and the dorsal and anal fins become pigmented. Small melanophores cover the head and body — coverage lightest ventrally on the head and gut. Three broad vertical bands become apparent dorsally on the nape, below the center of the spinous dorsal fin and below the center of the soft dorsal fin by 13.5 mm. Development of Macquaria novemaculeata larvae Adult meristic data D VIII-X,8-11; A 111,7-9; Pj 12-16; P2 1,5; Vertebrae 25; 38 specimens: 3.3-14.1 mm BL Eggs and hatching Eggs are approximately 900 pm in diameter and have multiple oil globules. Larvae are 3.3 mm SL at time of hatching. General morphology (Tables 3 and 4, Fig. 2) Yolksac and early preflexion larvae are elongate (BD 15-18%), but in late preflexion and flexion larvae, body depth becomes moderate (BD 26-34%). Body depth of field- caught postflexion larvae ranges from 29%. to 35%, and in transitional juveniles from 33% to 34%. Reared postflexion larvae and transitional juveniles are deeper than wild larvae, ranging from 32% to 44%, which is an artifact of the extremely full guts in the reared larvae. Body depth decreases abruptly posterior to the anus, although this becomes less marked with development. The head and body are laterally compressed. There are 25 myomeres (10-13 preanal+12-15 postanal). In general, there are 10-12 preanal myomeres in preflex- ion and flexion larvae, and 12-13 preanal myomeres in postflexion larvae and transitional juveniles. The gut is initially straight in yolksac larvae but is coiled by 3.9 mm. The gut is oval to triangular in shape; preanal length reaches 44-56% of BL in yolksac and preflexion larvae, 54-60% in flexion stage larvae, and 54-66% in postflexion larvae and transitional juveniles. The gut mass is large, particularly in reared postflexion larvae and transitional juveniles. The conspicuous gas blad- der, which is located over the midgut, is moderate to large in size, except in the yolksac larvae where it is small and inconspicuous. The head is round and small in yolksac larvae (HL 15-16%), moderate in preflexion larvae (HL 22-31%), and becomes moderate to large in flexion (29-35%) and postflexion larvae and transitional juveniles (32-38%). The snout is always shorter than the eye diameter and is initially concave, but becomes convex to straight in postflexion larvae. The eye is moderate to large (27-36% of HL) but is relatively larger in yolksac larvae (42-45% of HL). The eye is initially unpigmented, but is fully pigmented by 3.6-3.8 mm, prior to the com- plete absorption of the yolk. The moderate mouth reaches to the middle of the pupil. Small canine teeth appear in both jaws in late preflexion larvae by 4.4 mm. The number of teeth increases with development. The nasal pit begins to close by 8.6 mm, and both nostrils are developed by 10.3 mm. Head spination is weak. A small spine appears at the preopercular angle by the end of the preflexion stage. By the time notochord flexion is complete, there are three spines on the posterior preopercular border, and the spine at the angle is the longest. All spines are shorter than the pupil diameter. Additional spines 188 Fishery Bulletin 103(1) Table 3 Morphometric data for Macquaria by "R"), and juveniles from beach- novemaculeata larvae from channel net samples seine samples. Measurements are in mm. VAFL and reared in aquaria (body length preceded = vent to anal-fin length. Body length Preanal length Predorsal length Body depth Head length Snout length Eye diameter VAFL Yolksac R3.32 1.48 0.52 0.53 0.16 0.24 R3.60 1.60 0.58 0.53 0.16 0.22 Preflexion R3.60 2.00 0.92 0.96 0.24 0.34 R3.80 2.00 1.00 1.04 0.30 0.38 R3.90 1.76 0.64 0.84 0.18 0.30 R4.20 2.00 0.64 0.93 0.20 0.33 R4.40 2.00 0.78 1.06 0.26 0.34 4.57 2.36 2.16 1.20 1.40 0.28 0.40 0.22 Flexion 5.00 2.72 2.40 1.52 1.48 0.32 0.48 0.12 5.14 2.90 2.32 1.48 1.76 0.44 0.48 0.10 R5.31 2.84 2.60 1.40 1.60 0.48 0.56 0.20 5.39 2.74 2.66 1.58 1.60 0.40 0.48 0.12 R5.39 2.90 2.66 1.36 1.56 0.40 0.56 0.20 5.47 2.80 2.74 1.60 1.72 0.44 0.48 0.12 R5.47 3.00 2.60 1.56 1.76 0.52 0.60 0.12 5.70 3.40 2.92 1.96 2.00 0.52 0.56 0.06 5.90 3.32 2.90 1.80 1.88 0.52 0.56 0.04 Postflexion 5.64 3.07 2.41 1.66 2.00 0.52 0.60 0.08 5.89 3.52 2.64 2.00 2.00 0.52 0.60 0.06 6.06 3.32 2.81 1.99 2.00 0.52 0.56 0.10 6.30 3.40 2.57 1.91 1.99 0.50 0.60 0.20 6.60 3.73 2.91 2.24 2.08 0.50 0.66 0 6.72 3.74 2.82 2.24 2.16 0.60 0.64 0.08 R6.72 3.74 2.60 2.16 2.16 0.52 0.76 0.08 R7.20 3.98 3.00 2.32 2.28 0.52 0.68 0.08 7.40 4.15 3.02 2.49 2.32 0.66 0.72 0 R7.47 4.15 3.04 2.49 2.48 0.64 0.72 0.08 7.55 4.30 3.15 2.66 2.57 0.66 0.72 0 R8.18 5.31 3.49 3.07 2.91 0.75 0.91 0 R8.60 5.56 4.15 3.24 3.24 0.83 1.08 0 R9.20 5.56 3.98 3.75 3.50 0.75 1.21 0 Settled 10.13 6.64 4.81 3.49 3.65 0.83 1.33 0 R 10.20 6.64 4.57 3.74 3.74 0.83 1.33 0 R 10.30 6.64 4.57 3.99 3.82 1.05 1.33 0 11.62 7.55 5.56 3.82 4.23 1.00 1.49 0 11.62 7.55 5.47 3.98 4.39 1.07 1.49 0 13.28 8.30 5.98 4.56 4.98 1.41 1.66 0 14.10 8.63 6.47 4.65 5.15 1.41 1.74 0 Trnski et al.: Larval development of Macquana colonorum and M novemaculeata 189 Table 4 Meristic data of Macquaria novemaculeata larvae and juveniles. Body length preceded by ium. ( ) indicates only fin bases present. [ 1 incipient rays or spines, 1 1 ray transforming to a 'R" indicates larvae spine. reared in aquar- Body length Dorsal Anal Pectoral Pelvic Caudal Myomeres Yolksac R3.32 10+15=25 R3.60 11+14=25 Preflexion R3.60 12+13=25 R3.80 11+14=25 R3.90 11+14=25 R4.20 10+15=25 R4.40 10+15=25 4.57 (9) (8) [2+3] 10+15=25 Flexion 5.00 (VI), (6) (8) 7+6 10+15=25 5.14 (VI), (8) (9) 8+7 10+15=25 R5.31 (III), (9) (8) [7+6] 12+13=25 5.39 (IV), [9] (I), 19] 8+7 11+14=25 R5.39 (III), (9) (9) 6+6 12+13=25 5.47 (V), [8] [II, [7] [1]7+7[1] 11+14=25 R 5.47 (VI), [101 (I), |8](1) [1)8+7[1] 12+13=25 5.70 VI, 10 (I),8 6 9+8 13+13=26 5.90 [I]V, 10 (I), 8 6 9+8 12+13=25 Postflexion 5.64 VI, 10 (I),9 5 9+8 10+15=25 5.89 [VI], 10 (I), 9 5 9+8 12+13=25 6.06 VI, 10 (I), 9 6 9+8 12+13=25 6.30 VI, 11 1,9 6 buds 9+8 11+14=25 6.60 VI, 11 1,9 8 9+8 12+13=25 6.72 VI, 11 1,9 8 buds 9+8 12+13=25 R6.72 VI, 10 1,9 12 buds 9+8 12+13=25 R7.20 VII, 11 11,9 10 buds 9+8 12+13=25 7.40 VII, 11 11.9 12 buds 9+8 13+12=25 R7.47 VII, 11 11,9 13 buds 9+8 12+13=25 7.55 VII, 11 11,9 12 buds 9+8 12+13=25 R8.18 VIII. 11 11,9 13 1,5 9+8 13+12=25 R8.60 VIII, 11 11,8 13 1,5 9+8 13+12=25 R9.20 IX, 10 11111,7 15 1,5 9+8 13+12=25 Settled 10.13 IX, 10 11111,8 14 1,5 7+9+8+7 12+13=25 R 10.20 IX, 10 11111,8 14 1,5 9+8 13+12=25 R 10.30 IX, 10 11111,8 14 1,5 9+8 13+12=25 11.62 VIII 111,10 11111,8 14 1,5 9+9+8+8 13+12=25 11.62 IX, 9 11111,8 14 1,5 8+9+8+7 13+12=25 13.28 IX, 10 111,8 14 1,5 7+9+8+9 12+13=25 14.1 IX, 10 111,8 14 1,5 9+9+8+7 12+13=25 190 Fishery Bulletin 103(1) A 4.4 B 46 C 5.4 D 67 E 10.3 mm 13.3 Figure 2 Larvae of Macquaria novemaculeata. (A) yolksac larva, 10 days after hatching, note remnant of yolk below pectoral-fin base; (B) preflexion larva; (C) flexion stage larva; (D) postflexion larva; (E) postflexion larva, 57 days after hatching; (F) recently settled juvenile. Specimens A and E were reared at Port Stevens Fisheries Centre, New South Wales (NSW); B-D from Swansea Channel, central NSW; specimen F is a recently settled juvenile from the Clyde River, southern NSW. Trnski et al.: Larval development of Macquana colonorum and M. novemaculeata 191 form as larvae develop; four or five spines are present in larvae and transitional juveniles from 7.5-8.2 mm. A minute spine (rarely two) develops on the anterior preopercular border from 9 mm; a third spine devel- ops in transitional juveniles from 13.3 mm. A small interopercular spine develops by the time notochord flexion is complete. Low posttemporal and supraocular ridges, but no spines, develop during notochord flexion; they both become inconspicuous in postflexion larvae from 8.2 and 8.6 mm, respectively. An opercular spine is present from 8.6 mm. A small supracleithral spine is present in transitional juveniles from 10.1 mm. The pterygiophores of all the soft rays and up to six of the pterygiophores of the first dorsal fin form during notochord flexion. Soft rays of the dorsal fin are ossi- fied by the time notochord flexion is complete, whereas spinous rays ossify from posterior to anterior in late flexion and early postflexion larvae by 5.7-6.1 mm. The full complement of spines is present by 8.2 mm. Anal- fin pterygiophores form during notochord flexion, and all soft rays are ossified by the time notochord flexion is complete. Spinous rays of the anal fin begin to ossify in postflexion larvae by 6.3 mm, and all anal-fin ele- ments are present by 7.2 mm. The last spinous ray of the dorsal fin and the third spinous ray of the anal fin transform from a soft ray between 7.6 and 9.2 mm. Pec- toral-fin elements begin to ossify by the time notochord flexion is complete, and all rays are present in postflex- ion larvae by 7.5 mm. Pelvic-fin buds form in postflexion larvae by 6.7 mm, and all elements are ossified by 8.2 mm. Caudal-fin rays first appear in preflexion larvae from 4.6 mm, and all principal rays are ossified by the time notochord flexion is complete. Procurrent caudal rays are present in field-caught transitional juveniles. Notochord flexion commences between 4.6 and 5.0 mm, and is complete by 5.6-6.1 mm. There is a prominent gap between the anus and anal fin while the anal fin forms (vent to anal-fin length [VAFL] up to 5% of BL). The gap reduces in size as the anal fin develops, and it is absent by 7.6 mm. Scales have not developed in the largest specimen examined. Pigment (Fig. 2, A-F) Larvae are moderately to heav- ily pigmented. An expanded melanophore is present on the tip of the snout and a small melanophore develops under the tip of the lower jaw in preflexion larvae from 3.6 mm. A second melanophore on the snout develops posterior to the first by the time notochord flexion is complete. A single melanophore is present at the angle of the lower jaw. A few small melanophores develop ventrally along the lower jaw in postflexion larvae from 7.2 mm. A series of internal melanophores underlie the mid- and hindbrain. There are two very large expanded melanophores on the dorsal midline of the tail; the first is on the trunk centered over the hindgut, and the second is mid way along the tail. Once the dorsal fin forms they are centred under the middle of the spinous portion of the dorsal fin and under the posterior end of the soft dorsal fin, respectively. An additional smaller expanded melanophore is present from 7.2 to 7.5 mm on the dorsal midline of the nape above the pectoral-fin base. Two very large expanded melanophores occur ven- trally, opposite the two large dorsal melanophores. The anteriormost of these melanophores reduces in promi- nence as larvae develop and is inconspicuous to absent by metamorphosis. Internal expanded melanophores over the gas bladder may have filaments that emerge externally, particularly in preflexion and flexion lar- vae. Internal melanophores along the notochord may be apparent on the caudal peduncle in postflexion lar- vae from 7 mm. There is an expanded melanophore on the midline of the isthmus, immediately anterior to the cleithral symphysis. A series of three to six small, expanded melanophores is present along the ventral midline of the gut. In postflexion larvae there is a bi- laterally paired melanophore anterior to the pelvic-fin base, and two to four melanophores along the midline of the gut between the pelvic-fin base and the anus. A small contracted melanophore ventrally on the posterior margin of the caudal-fin base develops between 5.0 and 6.1 mm, and is located between ventral rays 1-5. This melanophore expands from 6.7 to 7.6 mm and spreads across up to four ray bases. Pigment distribution spreads rapidly over most of the head from 7.2 to 7.5 mm, and laterally on the trunk, gut and tail from 8.2 mm. The expanded melanophores on the dorsal and ventral midlines of the trunk and tail remain large as the larvae develop; the posterior- most of these increases in intensity in reared larvae. The expanded melanophores on the dorsal and ventral midlines of the body become relatively smaller after settlement. By settlement, small melanophores develop on the membranes of the pectoral, pelvic, anal, and cau- dal fins, and the membrane of the spinous portion of the dorsal fin becomes heavily pigmented. After settlement, small melanophores cover most of the head and body, but the heaviest cover is seen dorsally. Three broad vertical bands become apparent dorsally on the nape, below the center of the spinous dorsal fin, and below the center of the soft dorsal fin in the largest specimen examined (14.1 mm). Discussion Adults of M. colonorum and M. novemaculeata, which have only minor morphological differences, such as the relative length of the snout, the profile of the head dor- sally, postorbital head length, and gill-raker counts, are difficult to distinguish (Williams, 1970). None of these characters are useful for distinguishing larvae. The larvae of these two species could be positively identified only by comparison with reared larvae derived from positively identified brood stock. Melanophore distribution is the most distinguishing character between the larvae of M. colonorum and M. novemaculeata. Macquaria colonorum has between four and seven expanded melanophores along the dorsal midline of the trunk and tail between 4.8 and 7.1 mm. 192 Fishery Bulletin 103(1) Macquaria novemaculeata has only two melanophores, and these are much larger; a third expanded melano- phore develops on the nape from 7.2 mm. In addition, M. novemaculeata lacks a midlateral series of melano- phores along the tail until settlement, and it is never as well developed as that in M. colonorum. On the other hand, M. colonorum has a prominent midlateral series until after settlement. One other morphological char- acter that distinguishes the larvae is a snout length which is about equal to eye diameter in M. colonorum larvae until 7 mm, but snout length is always smaller than the eye diameter in M. novemaculeata. Within the genus Macquaria, larval development of only M. ambigua has been described (Lake, 1967; Brown and Neira, 1998). There are several differences in the life history and development of the larvae of M. ambig- ua compared with M. colonorum and M. novemaculeata. Macquaria ambigua is restricted to freshwater, the eggs are large (3.3-4.2 mm in diameter, compared with 0.9 mm in reared M. novemaculeata) and the yolk sac in M. ambigua is large in small larvae and is not resorbed until the flexion stage (Brown and Neira, 1998). Com- pared with the larvae described in the present study, larvae of M. ambigua have more myomeres (24-28, but typically 26-27), and these larvae are relatively large by the time they complete notochord flexion (7.3 mm). They also lack an interopercular spine and supraocular ridge, and lack dorsal and lateral pigment on the tail until the postflexion stage. Larvae of several other generalized percoid families are morphologically similar to Macquaria, including Latidae (Trnski et al., 2000), Microcanthidae (Walker et al., 2000a), Kyphosidae (Walker et al., 2000b), and some Apogonidae (Leis and Rennis, 2000). The latid genus Lates is morphologically most similar to the Macquaria larvae described in the present study but is tropical and does not have an overlapping distribution with Macquaria. Lates can be distinguished by the small size at notochord flexion (3.0-3.8 mm), dorsal and pectoral fin-ray counts when complete, and heavier melanophore distribution at a given size. Microcanthid and kyphosid larvae can be distinguished from coastal percichthyid larvae by the higher number of fin elements in the dorsal and anal fins, and the presence of supracleithral spines that are absent in larval percichthyids until the juvenile stage. Some deep-bodied apogonids resemble Macquaria larvae but can be distinguished by having separate spinous and soft dorsal fins and a large, con- spicuous gas bladder. Larvae of M. colonorum and M. novemaculeata were collected in Swansea Channel from July to August. This collection period coincides with adults of M. novemacu- leata spawning from June to September in central New South Wales (Harris, 1986). Macquaria colonorum prob- ably spawns at a similar time (McCarraher and McK- enzie, 1986), and eggs have been collected from June to November in western Victoria (Newton, 1996). Adults of both species are thought to spawn in the middle reaches of estuaries at salinities above 8-10 g/kg (Harris, 1986; McCarraher, 1986), but M. novemaculeata will spawn in waters up to 35 g/kg in culture (Battaglene and Selosse, 1996). The optimal conditions for incubation and hatch- ing of M. novemaculeata eggs are 18 [±1]°C and salinity at 25 to 35%r (van der Wal, 1985). Eggs are buoyant within this salinity range and hatch in 42 h at 18 C. The presence of field-caught larvae of both species on incoming tides in Swansea Channel indicates that the larvae have spent some time in the ocean and that the eggs were potentially spawned in the ocean rather than in an estuary if they were not carried out to sea by outgoing tides. Macquaria novemaculeata adults move downstream into estuaries to spawn in water of suitable salinity. In low rainfall years, the spawning location is further upstream than in wet years, when spawning can occur in shallow coastal waters adjacent to estuaries (Searle1). Mature M. novemaculeata adults can be found outside of estuaries in wet years (Williams 1970). This is verified by the collection of mature adults by trawl in July 1995 in 11-17 m of water off Newcastle, NSW (AMS 1.37358-001). Macquaria colonorum adults have also been collected on the continental shelf (Mc- Carraher and McKenzie, 1986). In addition, larvae can tolerate waters of marine salinity in culture, and late in their larval phase wild larvae can tolerate marine sa- linity as shown from our field collections. The presence of larvae and adults in continental shelf waters may provide two modes of dispersal among estuaries. Thus, these two species of Macquaria may not be confined to freshwater and estuarine conditions as often assumed (Harris and Rowland, 1996; Allen et al., 2002). The smallest juveniles of M. colonorum and M. novemaculeata collected in the wild are from the Clyde River estuary, southern NSW. These range in size from 10 to 14 mm SL, and were collected among Zostera sea- grass. They are morphologically similar to the largest pelagic larvae collected in the channel net in Swansea Channel. Based on the largest larvae and smallest juveniles, settlement occurs between 7.1 and 10.3 mm SL in M. colonorum and between 9.2 and 10.1 mm in M. novemaculeata. Transition to the juvenile stage is gradual, because scales are not present and juvenile pigmentation is still forming at about 15 mm. Juveniles of both species have been collected in estuarine waters until at least 100 mm SL (AMS fish collection). Juve- niles of M. novemaculeata would be expected to migrate to freshwater because this is the nominal adult habitat (Williams, 1970), but the size at which this migration occurs is unclear. The two species described in the present study were the only members of the genus Percolates, until this genus (along with Plectroplites) was synonymized with Macquaria by MacDonald (1978). Analyzing morpho- logical and biochemical similarities of the three genera, MacDonald (1978) listed eight morphological differences that distinguished Percolates from Macquaria and Plec- troplites. Protein electrophoresis similarities were stron- 1 Searle, G. 2002. Personal commun. Searle Aquaculture, 255 School Rd, Palmers Island NSW 2463. Trnski et al.: Larval development of Macquana colonorum and M novemaculeata 193 ger between Pe. (currently Macquaria) colonorum and Pe. (Macquaria) novemaculeata (similarity coefficient 0.95), and M. australasica and PL {Macquaria) am- bigua (0.71) than between the Percolates and Macquaria + Plectroplites (0.63) (MacDonald, 1978). The species of Percolates are euryhaline, whereas Macquaria and Plectroplites are strictly freshwater. This fact, combined with the difference in larval morphological features be- tween Macquaria ambigua (Brown and Neira, 1998) and M. colonorum and M. novemaculeata, provides evidence that the genus Macquaria as defined by MacDonald may be polyphyletic. Recent phylogenetic analysis of the Percichthyidae with the use of molecular data in- dicates that M. colonorum and M. novemaculeata are more closely related to Maceullochella species than to Macquaria (sensu stricto) (Jerry et al., 2001). Molecular and larval evidence indicates the two catadromous spe- cies (M. colonorum and M. novemaculeata) belong in a genus separate from the freshwater species (M. ambigua and M. australasica). Acknowledgments Comments by Dave Johnson, Jeff Leis, and Tony Miskie- wicz improved the manuscript. Sue Bullock illustrated the larvae from camera lucida sketches by TT Glen Searle provided information on spawning habits that aided interpretation of larval distributions. Mark McGrouther provided access to specimens held in the Fish Collection at the Australian Museum (AMS). Larval collections in the field were supported by funds from Lake Macquarie City Council. Preparation of this paper was supported by a NSW Government Biodiversity Enhancement Grant to AMS, and by AMS. Literature cited Allen, G. R., S. H. Midgley, and M. Allen. 2002. Field guide to the freshwater fishes of Australia, 394 p. Western Australian Museum, Perth, Western Australia, Australia. Battaglene, S. C and P. M. Selosse. 1996. Hormone-induced ovulation and spawning of cap- tive and wild broodfish of the catadromous Austra- lian bass, Macquaria novemaculeata (Steindachner), (Percichthyidae). Aquacult. Res. 27:191-204. Brown. P., and F. J. Neira. 1998. Percichthyidae: basses, perches, cods. In Larvae of temperate Australian fishes: laboratory guide for larval fish identification (F. J. Neira, A. G. Miskiewicz and T. Trnski, eds.), p. 259-265. Univ. Western Australia Press, Perth, Western Australia, Australia. Dakin, J. W., and G. L. Kesteven. 1938. The Murray cod {Maceullochella macquariensis (Cuv. And Val.)). Bull. New South Wales State Fish. 1:1-18. Dufty, S. 1986. Genetic and morphological divergence between populations of macquarie perch (Macquaria australasica) east and west of the Great Dividing Range. Honours thesis, 43 p. Univ. New South Wales, Sydney, New South Wales, Australia. Eschmeyer, W. N. (ed.) 1998. Catalog of fishes, 2905 p. Special publication 1, Center for Biodiversity Research and Information, California Academy of Sciences, San Francisco, CA. Harris. J. H. 1986. Reproduction of the Australian bass, Macquaria novemaculeata ( Perciformes: Percichthyidae) in the Sydney Basin. Aust. J. Mar. Freshw. Res. 37:209- 235. Harris, J. H., and S. J. Rowland. 1996. Family Percichthyidae: Australian freshwater cods and basses. In Freshwater fishes of south-eastern Aus- tralia (R. M. McDowall, ed.), p 150-163. Reed Books, Chatswood, New South Wales., Australia. Jerry, D. R., M. S. Elphinstone, and P. B. Baverstock. 2001. Phylogenetic relationships of Australian members of the family Percichthyidae inferred from mitochon- drial 12s rRNA sequence data. Molec. Phylogenetics Evol. 18:335-347. Johnson, G. D. 1984. Percoidei: development and relationships. //; Ontogeny and systematics of fishes (H. G. Moser, W. J. Richards, D. M. Cohen, M. P. Fahay, A. W. Kendall Jr., and S. L. Richardson, eds.), p 464-498. Am. Soc. Ichthyol. Herpetol., Spec. Publ. 1. Kailola, P. J., M. J. Williams, P. C. Stewart, R. E. Reichelt, A. McNee, and C. Grieve. 1993. Australian fisheries resources, 422 p. Bureau of Resource Sciences, and Fisheries Research and Develop- ment Corporation, Canberra, New South Wales, Australia. Lake, J. S. 1967. Rearing experiments with five species of Aus- tralian Freshwater fishes. II. Morphogenesis and ontogeny. Aust. J. Mar. Freshw. Res. 18:155-173. Leis, J. M., and B. M. Carson-Ewart. 2000. The larvae of Indo-Pacific coastal fishes: an identi- fication guide to marine fish larvae. Fauna Malesiana Handbooks, 2850 p. Brill, Leiden, The Netherlands. Leis, J. M., and D. S. Rennis. 2000. Apogonidae (cardinalfishes). In The larvae of Indo-Pacific coastal fishes: an identification guide to marine fish larvae (J. M. Leis and B. M. Carson-Ewart, eds.), p. 273-279. Fauna Malesiana Handbooks 2, Brill, Leiden, The Netherlands. McCarraher, D. B. 1986. Observations on the distribution, spawning, growth and diet of Australian bass (Macquaria novemaculeata) in Victorian waters. Arthur Rylah Inst. Environ. Res. Tech. Rep. Ser. 47. McCarraher, D. B., and J. A. McKenzie. 1986. Observations on the distribution, growth, spawn- ing and diet of estuary perch (Macquaria colonorum) in Victorian waters. Arthur Rylah Inst. Environ. Res. Tech. Rep. Ser. 42. MacDonald, C. M. 1978. Morphological and biological systematics of Austra- lian freshwater and estuarine percichthyid fishes. Aust. J. Mar. Freshw. Res. 29:667-698. Musyl, M. K., and C. P. Keenan. 1992. Population genetics and zoogeography of Australian freshwater golden perch, Macquaria ambigua (Richard- son 1845) (Teleostei: Percichthyidae), and electropho- retic identification of a new species from the Lake Eyre Basin. Aust. J. Mar. Freshw. Res. 43:1585-1601. 194 Fishery Bulletin 103(1) Neira, F. J., A. G. Miskiewicz, and T. Trnski. 1998. Larvae of temperate Australian fishes: labora- tory guide for larval fish identification, 474 p. Univ. Western Australia Press, Perth, Western Australia, Australia. Nelson, J. S. 1994. Fishes of the world, 3rd ed., 600 p. John Wiley and Sons, New York, NY. Newton, G. M. 1996. Estuarine ichthyoplankton ecology in relation to hydrology and zooplankton dynamics in a salt-wedge estuary. Mar. Freshw. Res. 47:99-111. Paxton, J. R., and J. E. Hanley. 1989. Percichthyidae. In Zoological catalogue of Aus- tralia, vol. 7: Petromyzontidae to Carangidae (J. R. Paxton, D. F. Hoese, G. R. Allen, and J. E. Hanley (eds.), p. 509-515. Australian Government Publishing Service, Canberra, New South Wales, Australia. Paxton, J. R., J. E. Hanley, D. J. Bray, and D. F Hoese. In press. Percichthyidae. In Pisces (part 2), Mugilidae to Molidae. Zoological catalogue of Australia, vol. 7 (part 2) (D. F. Hoese, D. J. Bray, D. J., G. R. Allen, C. J. Allen, N. J. Cross, and J. R. Paxton, eds.). Australian Government Publishing Service, Canberra, New South Wales, Australia. Pusey, B. J., and M. J. Kennard. 2001. Guyu wujalwujalensis, a new genus and species (Pisces: Percichthyidae) from north-eastern Queensland, Australia. Ichthyol. Explor. Freshw. 12:17-28. Trnski, T. 2002. Behaviour of settlement-stage larvae of fishes with an estuarine juvenile phase: in situ observations in a warm-temperate estuary. Mar. Ecol. Progr. Ser. 242:205-214. Trnski, T., D. J. Russell, and J. M. Leis. 2000. Latidae (barramundi, sea basses). In The larvae of Indo-Pacific coastal fishes: an identification guide to marine fish larvae (J. M. Leis and B. M.Carson-Ewart, eds.), p. 313-316. Fauna Malesiana Handbooks 2, Brill, Leiden, The Netherlands. VanderWal, E. 1985. Effects of temperature and salinity on the hatch rate and survival of Australian bass iMacquaria novemacu- leata ) eggs and yolk-sac larvae. Aquaculture 47:239- 244. Vigliola, L., and M. Harmelin-Vivien. 2001. Post-settlement ontogeny in three Mediterranean reef fish species of the genus Diplodus. Bull. Mar. Sci. 68:271-286. Walker Jr, H. J., A. G. Miskiewicz, and F. J. Neira. 2000a. Microcanthidae (stripey). In The larvae of Indo- Pacific coastal fishes: an identification guide to marine fish larvae (J. M. Leis and B. M.Carson-Ewart, eds.), p. 470-473. Fauna Malesiana Handbooks 2, Brill, Leiden, The Netherlands. Walker Jr, H. J., F. J. Neira, A. G. Miskiewicz, and B. M. Carson-Ewart. 2000b. Kyphosidae (rudderfishes, sea chubs). In The larvae of Indo-Pacific coastal fishes: an identification guide to marine fish larvae (J. M. Leis and B. M.Carson- Ewart, eds.), p. 466-469. Fauna Malesiana Handbooks 2, Brill, Leiden, The Netherlands. Williams, N. J. 1970. A comparison of the two species of the genus Perca- lates Ramsay and Ogilby (Percomorphi: Macquariidae), and their taxonomy. State Fisheries Research Bulletin 11, 60 p. Chief Secretary's Department, Sydney, New South Wales, Australia. 195 Abstract — The Argentine sandperch Pseudopercis semifasciata (Pinguipe- didae) sustains an important commer- cial and recreational fishery in the northern Patagonian gulfs of Argen- tina. We describe the morphological features of larvae and posttransition juveniles of P. semifasciata and ana- lyze the abundance and distribution of early life-history stages obtained from 19 research cruises conducted on the Argentine shelf between 1978 and 2001. Pseudopercis semifasciata larvae were distinguished from other larvae by the modal number of myomeres (between 36 and 38), their elongated body, the size of their gut, and by osteological features of the neuro- and branchiocranium. Pseudopercis semi- fasciata and Pinguipes brasilianus (the other sympatric species of pin- guipedid fishes) posttransition juve- niles were distinguished by their head shape, pigmentation pattern, and by the number of spines of the dorsal fin (five in P. semifasciata and seven in P. brasilianus). The abundance and distribution of P. semifasciata at early stages indicate the existence of at least three offshore reproductive grounds between 42-43°S, 43-44°S, and 44-45°S, and a delayed spawning pulse in the southern stocks. Early life history of the Argentine sandperch Pseudopercis semifasciata (Pinguipedidae) off northern Patagonia Leonardo A. Venerus Centra Nacional Patagonico-Conseio Nacional de Investigaciones Cientificas y Tecnicas Boulevard Brown s/n, (U9120ACV) Puerto Madryn, Chubut, Argentina E-mail address leoigicenpat edu ar Laura Machinandiarena Martin D. Ehrlich Institute Nacional de Investigacion y Desarrollo Pesquero PO Box 175, (B7602HSA) Mar del Plata, Buenos Aires, Argentina Ana M. Parma Centra Nacional Patagonico-Conseio Nacional de Investigaciones Cientificas y Tecnicas Boulevard Brown s/n, (U9120ACV) Puerto Madryn, Chubut, Argentina Manuscript submitted 20 November 2003 to the Scientific Editor's Office. Manuscript approved for publication 16 September 2004 by the Scientific Editor. Fish. Bull. 103:195-206 (2005). The family Pinguipedidae (Osteich- thyes, Perciformes) includes six genera and about 50 marine species and one freshwater species (Froese and Pauly, 2004). On the Argentine continental shelf this family is represented by two species, Pseudopercis setJiifasciata (Cuvier, 1829) and Pinguipes brasilia- nus Cuvier, 1829. The Argentine sandperch P. semi- fasciata is an important incidental catch in the bottom trawl and long- line commercial fisheries that target hake (Merluccius hubbsi) in the north- ern Patagonian coast off Argentina (Otero et al., 1982; Elias and Bur- gos, 1988; Gonzalez, 1998). In recent years, the reported annual landings have oscillated between 1900 and 3780 metric tons (official statistics, SAGPyA-DNPyA1). In northern Pata- gonia, P. semifasciata is also targeted by sport anglers and spear fishermen and represents a tourist attraction for recreational divers. It inhabits rocky and sandy bottoms, from 23°S in Brazil to 47°S in Argentina (Cous- seau and Perrotta, 2000), mainly in coastal waters, although it has been found in depths of up to 100 m (Mene- zes and Figueiredo, 1985). Very little is known about the ecol- ogy and behavior of P. setnifasciata, and most of what is known is based on limited observations during un- derwater visual censuses on shal- low reefs where adults concentrate (Gonzalez, 1998). Previous studies have focused on morphological fea- tures (Herrera and Cousseau, 1996; Rosa and Rosa, 1997; Gosztonyi and Kuba2), age and growth (Elias and Burgos, 1988; Fulco, 1996; Gonzalez, 1998), diet (Elias and Rajoy, 1992; Gonzalez, 2002), and reproductive traits, including reproductive sea- son, spawning modality, and age at first maturity (Macchi et al., 1995; SAGPyA-DNPyA. 2003. Capturas maritimas totales 1992-2002. Manu- script, 71 p. [Available from Sec- retaria de Agricultura, Ganaderia y Pesca de la Nacion, Direccibn de Pesca y Acuicultura, Paseo Colon 982 P.B. Of. 59 - (C1063ACW) Buenos Aires, Argentina.] http://www.sagpya.mecon. gov.ar (Accessed July 2004. J Gosztonyi, A. E„ and L. Kuba. 1996. At- las de huesos craneales y de la cintura escapular de peces costeros patagonicos. Inf. Tec. FPN 4, 29 p. [Available from CENPAT, Blvd. Brown s/n (U9120ACV), Puerto Madryn, Chubut, Argentina.) 196 Fishery Bulletin 103(1) Fulco, 1996; Gonzalez, 1998). Pseudopercis semifas- ciata is a multiple spawner with low batch fecundity and an extended reproductive season (Macchi et al., 1995; Gonzalez, 1998). There is little information on the early life history of the species because only specimens >20-25 cm are found on reefs and the habitat of juve- niles has not been described. In general, information about the early stages of pinguipedid fishes from the southwest Atlantic Ocean is scarce. De Cabo3 reported pinguipedid larvae from the Argentine shelf but did not identify the specimens to species level. In the present study, we describe development of P. semifasciata from larvae to the posttransition juvenile stage {sensu Vigliola and Harmelin-Vivien, 2001) and analyze data on distribution and abundance on the northern Patagonian shelf. This information is needed to locate main reproductive and nursery grounds for the species. 3 De Cabo, L. 1988. Descripcidn de tres larvas de peces teleos- teos del Mar Argentino: Mugiloididae, Ophidiidae (Genypterus blacodes) y Tripterygidae (Tripterygion eunninghami). Un- publ. manuscript, 58 p. Facultad de Ciencias Exactas y Natu- rales, Universidad de Buenos Aires-INIDEP. lAvailable from INIDEP: P.O. Box 175 (B7602HSA) Buenos Aires, Argentina.] Materials and methods Fish larvae and posttransition juveniles were collected during 19 research cruises conducted by INIDEP (Insti- tuto Nacional de Investigacion y Desarrollo Pesquero) between 1978 and 2001. A total of 592 ichthyoplankton samples and 277 juvenile trawl samples were analyzed (Table 1). Larvae Ichthyoplankton was sampled by using Bongo, Nack- thai, and PairoVET nets. The Bongo net was fitted with 300-/01' 63°51'-66c42' 2000 OB-05/00 01 Jun-20 Jun 112 43°45'-47°02' 61°53'-67°25' 2001 OB-02/01 12 Feb-25 Feb 23 42°54'-45°25' 62o30'-66°12' Venerus et al .: Early life history of Pseudoperas semifasciata 197 A total of 68 preserved larvae, ranging in body length (BL) from 3.3 to 11.7 mm, were used to describe larval development. Terminology for morphometries followed Neira et al. (1998). Additionally, head depth (HD) was defined as the maximum depth of the head. Preserved larvae were measured to the nearest 0.1 mm with an ocular micrometer fitted to a dissecting microscope, and their pigmentation pattern was recorded. Possible shrinkage was not considered in the measurements. Whenever possible, the number of vertebrae and num- bers of dorsal, anal, caudal, pectoral, and pelvic fin rays were recorded. In addition, 14 larvae from 3.4 to 11.7 mm BL were cleared and stained following the methods of Potthoff (1984) and Taylor and Van Dyke (1985), and then examined for meristics and osteological features. Myomere and fin-ray counts and morphometric measure- ments were made on the left side of the body. Larval abundance was expressed as the number of larvae/ 10 m2 of sea surface as recommended by Smith and Richardson (1977). Posttransition juveniles Posttransition juveniles were collected with a small bottom trawl called "Piloto," with the following features: 6 m total length, 6-m headrope and groundrope, 25-mm wing mesh size, 10-mm codend mesh size, 0.25-m2 otter board surface and 12 kg weight, 10-m bridles and 0.80- m vertical opening. In Argentina, commercial fishing vessels use this gear for locating shrimp concentra- tions. Additionally, an epibenthic sampler (Rothlisberg and Pearcy, 1976) fitted with 1-mm mesh was used on one cruise (EH-02/92). We believe that individuals up to 12 cm total length were well represented in samples obtained with this gear. A total of 27 posttransition juveniles, ranging from 22 to 83 mm body length (BL), were used to describe Argentine sandperch developmental stages. Samples were either frozen or fixed in 5% formalin to seawater solution. Measurements and degree of pigmentation were recorded after preservation. Total length (TL), body length ( = standard length), head length (HL), predorsal length (PDL), and preanal length (PAL) were measured to the nearest 1 mm. Head depth (HD), body depth (BD), and eye diameter (ED) were measured to the nearest 0.2 mm. Three juveniles between 22 and 33 mm BL were cleared and stained (Potthoff, 1984; Taylor and Van Dyke, 1985) and exam- ined for meristics. The density of posttransition juveniles, expressed as individuals/square nautical mile (nmi2), was estimated from swept area. The family Pinguipedidae includes two species (morphologically very similar as juveniles) that overlap in the Argentine Sea. Unfortunately, not all individuals caught during the cruises were examined by us; therefore, to avoid biases caused by identifica- tion errors, the posttransition juveniles of both species were considered as a group. Distributional centroids and ellipses were calculated by following the method of Kendall and Picquelle (1989), that is by weighting 30 - A 8 25- -3 20 - CQ I 15 - Aa AA £. A A* fak A AA 10 n =55 ■ 0 5 10 15 30 - 8 25- 1 15- if^A^ 10 ■ n =53 1 0 5 10 15 BL (mm) Figure 1 Relative head length (HL/BLxlOO) and relative head depth (HD/BLx 100) against body length (BL) in Pseudoper- cis semifasciata larvae, regardless of the flexion stage of the notochord. Solid line represents a linear trend (rc = 53; r2 = 0.2576; P<0.001). each station by the density of juveniles caught. For this purpose each density value was standardized with re- spect to the maximum density observed for each survey season over all years. Results Description of larvae General morphological features The larval body was elongate and relative BD was <25% in all stages of development (Table 2). The smallest larva collected (yolksac larva) was 3.3 mm BL. Its yolk sac was small and the single oil globule was located on the anterior part of the yolk mass. Notochord flexion began at 6 mm and was complete by 7-8 mm BL. As development proceeded, larvae became slightly deeper and laterally compressed. The head was small, with a rounded snout and no spines. The oblique mouth was open by the end of the yolksac larval stage. By 10 mm BL, premaxilla and dentary bones were covered with caniniform teeth. The premaxilla was an elongated bone with three processes on its dorsal margin — the first one perpendicular to the premaxilla. Relative head length remained constant, whereas relative head depth diminished during develop- ment (Fig. 1). The eyes were pigmented and their relative diameter decreased during the preflexion stage, and 198 Fishery Bulletin 103(1) Table 2 Body proport ions of Pseudopercis semifasciata larvae, according to the flexion stage of the notochord. Mean (±SE), range and number of observations are shown in the table. BD =body depth; BL=body length; ED=eye diameter ; HD=head depth; HL=head length; PAL = preanal length. BD/BLxlOO HD/BLxlOO Preflexion 3.3-7.1 mm BL;o = 36: 16.4 ±3.1 (12.4-25.5)71=27 18.5 ±2.3 (14.8-23.1) o=25 Flexion 6.2-8.7 mm BL;o = 8: 14.5 ±1.4 (12.2-16.1) n=5 17.1 ±1.2(15.9-19.4)0 = 8 Postflexion 7.3-11.7 mm BL;o=20: 16.8 ±1.3 (14.3-19.5) o=20 PAL/BLxlOO 17.4 ±1.4 (15.2-19.8) o=20 ED/BLxlOO Preflexion 53.6 ±4.1 (45.0-62.5)n = 29 7.8 ±1.1 (5.9-10.9) o=29 Flexion 52.0 ±2.2 (49.4-56.5) o = 8 6.1 ±0.3(5.6-6.5)0 = 8 Postflexion 52.5 ±2.7 (47.3-57.5)n=20 HL/BLxlOO 6.1 ±0.7(4.9-7.6)0 = 20 Preflexion 21.2 ±2.4 (17.0-27.7) o = 27 Flexion 20.5 ±1.9 (18.2-24.2) ra=8 Postflexion 23.4 ±1.5 (19.3-26.0)»=20 then remained constant (Fig. 2, A and B). The gut was initially straight but began to constrict at 4 mm BL and was loosely constricted throughout development (Fig. 3, A-C). It was moderate to long and extended to near the midpoint of the body, resulting in a relative preanal length of 0.45 to 0.62 BL. 15 - A 10 - Wa 5 - 2pfeeA n =29 o U - o x ( _l m ) 5 10 15 Q w 15 - B 10 - 5 - A A 0 - n =28 ■ . i 0 5 10 15 BL (mm) Figure 2 Relative eye diameter (ED/BLxlOO) against body length (BL) in Pseudoper- cis semifasciata larvae. (A) Preflexion larvae. Solid line represents a linear trend (rc=29; r2 = 0.6228; P<0.001). (B) Flexion and postflexion larvae. Body pigmentation Argentine sandperch larvae were lightly pigmented during all stages of development (Fig. 3; A-C). The pigmentation on the ventral body surface, between the isthmus and the anus, consisted of small stellate melanophores. Several small melanophores were scattered on the lateral surface of the anterior part of the gut. A double row of minute melanophores along the ventral surface ended in a single melanophore at the constriction of the gut. Pigmentation along the lateral midline of the tail consisted of four to seven stellate melanophores. In preflexion larvae (Fig. 3A), small spots were evi- dent along the lower jaw and the ventral part of the head. Several small stellate melanophores were present on the dorsal surface of the gut. A few melanophores were scattered at the base of the pectoral fin bud. Preflexion and flexion larvae (Fig. 3, A and B) showed a distinct pattern of 12 to 23 small postanal melano- phores serially arranged, about one per myomere, along the ventral midline. A total of 11 to 18 melanophores, about one melanophore per anal fin pterygiophore, was observed in postflexion larvae (Fig. 3C). As flexion pro- gressed (Fig. 3, B and C), the number of melanophores on the ventral part of the head and over the gut diminished. Fins and meristic features Modes of preanal and post- anal myomeres were 14 and 23, respectively. All speci- mens examined had 33-40 total myomeres (mode:36-38 myomeres). Vertebral column ossification started anteri- orly. A total of 38-39 vertebrae were recorded in 10-12 mm BL postflexion larvae (n = 2). In yolksac larvae, finfold and pectoral buds were the first fin development distinguished. In preflexion and flexion larvae, the finfold was present and it was gradu- ally lost as the true fins developed. The sequence of fin-ray formation, characterized by initial development of fin elements, was caudal (7-8 mm BL), then pectoral Venerus et a\ .: Early life history of Pseudopercis semifasciata 199 Figure 3 Larvae and posttransition juvenile of Pseudopercis semifasciata. (A) Preflexion (4.3 mm BL). (B) Flexion (8.7 mm BL). (C) Postflexion (10.7 mm BL). (Dl transi- tion juvenile (22 mm BL). (9-10 mm BL), anal (9-10 mm BL), dorsal (9-10 mm BL), and pelvic (10-11 mm BL). Elements of the cau- dal fin began forming at flexion stage, and remaining fins at the postflexion stage. By 9-10 mm BL, dorsal (V+26-27) and anal (11+20-22) fin elements reached their full complement. Description of posttransition juveniles The posttransition juvenile stage was characterized by the acquisition of complete fin-ray complements and by morphological similarities with the adults (Table 3, Fig. 4). The transition from pelagic to benthic habitat in this species, i.e. settlement, probably occurred at about 20 mm BL because the smallest benthic juvenile of Pseu- dopercis semifasciata reported was 22 mm BL. Table 3 Body proportions (mean [±SE| and range) of Pseudoper- cis semifasciata posttransition juveniles. BD=body depth; BL=body length; ED = eye diameter; HD=head depth; HL=head length; PAL = preanal length; PDL=predorsal length. BD/BLxlOO 15.1 ±1.4(12.7-19.3) PAL/BLxlOO 41.7 ±2.2 (37.7 -48.0) HL/BLxlOO 23.0 ±2.4 (17.6-31.6) HD/BLxlOO 13.8 ±1.5 (11. 9-19.31 ED/BLxlOO 8.4 ±1.1 (6.4-11.9) PDL/BLxlOO 27.8 ±1.3 (25.7-30.2) 200 Fishery Bulletin 103(1) Individuals became more thick bodied as they devel- oped. The body was elongate and relative body depth remained fairly constant throughout development. The snout was longer and rounded, and relative head length was moderate. The mouth was terminal, reaching to the middle of the eye, and had fleshy lips. Both jaws presented only caniniform teeth. Two opercular spines were also present in all specimens studied. Relative head depth decreased slightly during development, but not relative eye diameter. Gut length was moderate (PAL/BL 0.38-0.48), and the anus was situated near the midpoint of the body (Fig. 3D). Relative predorsal length (0.26-0.30) diminished during development. The scales were ctenoid. Smaller posttransition juve- niles (BL s33 mm) retained some of the larval pigmen- tation pattern. Larger juveniles showed several dark vertical bars, not completely defined at this stage of de- velopment, and three horizontal stripes along the body (Fig. 3D). Vertical bars developed progressively from the caudal peduncle to the head. Two lateral stripes formed continuous bands along each side of the body, almost entirely above the midline. The upper stripe developed from the tip of the snout and the lower one began below the eye, both extending to the anterior caudal peduncle. Another stripe developed from the dorsal region of the head between the eyes and extended along the dorsal fin, joining the upper lateral stripe at the posterior third part of the body. In large posttransition juveniles (a47 mm BL), the membrane of the dorsal fin was pig- mented more densely between the spines than between the rays; there were also dark blotches on the mem- brane between the rays. Anal-fin membranes were more pigmented than those of the dorsal fin. The membranes of the pectoral, pelvic, and caudal fins, and the external border of the membranes of the dorsal fin, were yellow in frozen individuals. By 22 mm BL, the conspicuous 50 -i ~ 40 30 ■ 20 10 ■ t r BD/TLx100 HL/TLx100 PAL/TLx100 PDL/TLx100 Body proportions Figure 4 Comparisons between body proportions in posttransition juveniles (white bars) and adults (gray bars) of Pseudopercis semifasciata. Relative measures were taken with respect to total length (TL). Body depth (BD), head length (HLi, preanal length (PAL), predorsal length (PDL). Proportions for adults were estimated from 99 individuals between <30 cm and 90 cm TL (Gonzalez, 1998). dark blotch observed in adult P. semifasciata on the base of the caudal fin upper lobe (Herrera and Cous- seau, 1996) was already present (Fig. 3D). The pelvic fin was large and slightly shorter than the pectoral fin, whose margin was rounded. Abundance and distribution Larvae Larvae of Argentine sandperch occurred between 36°42'S and 46°30'S, mainly in coastal waters, in the vicinity of the 50-m isobath (Fig. 5). The southernmost limit where larvae were collected was within San Jorge Gulf, which was surveyed in late March (fall). Larvae were present in only 3.55% of the stations in densities that varied between two and 74 larvae/10 m2 of sea surface (Table 4). Greater densities (>20 larvae/10 m2 of sea surface) were obtained in December 1986, 1996, and 1999, off the coast between Engano Bay and Isla Escondida. Positive stations formed scattered clumps along the whole distributional area of the species. Min- imum and maximum depths sampled were 20 and 71 m, respectively. Water temperature at 10 m depth at posi- tive stations varied between 12.3°C (March 1985) and 18.7°C (December 1999) (mean temperature [±SE]: 15.2°C [±2.1°C]). Posttransition juveniles Posttransition pinguipedid juveniles were found between 42°27'S and 43°37'S in February and March, and between 43°17'S and 44°58'S from April to June, primarily in the vicinity of the 50-m isobath (Fig. 6, A and B). The percentages of positive stations were 5.9% and 7.7% in summer and fall surveys, respectively. Maximum juvenile densities were 4410 individuals/nmi- in summer and 27,027 individuals/nmi2 in fall (Table 5). The grid of stations used during the summer and fall cruises overlapped (Fig. 6, A and B), cover- ing the main area of concentration of P. semifas- ciata (Otero et al., 1982). Minimum and maximum depths were 54 and 74 m in summer surveys (mean depth [±SE]: 64.5 [±10.0] m), and 34 and 79 m in fall surveys (mean depth [±SEJ: 60.4 [+13.7] m). The distributional ellipses calculated for summer and fall from the positive stations were small and widely separated. Maximum summer densities of posttransition pinguipedids were found southeast of Peninsula Valdes, whereas greatest fall densities were detected northeast of Camarones Bay (Fig. 6, A and B). Discussion Literature describing the early stages of species belonging to the family Pinguipedidae (formerly Mugiloididae) is scarce. The few available studies refer to the larval development of Parapercis spp. (Leis and Rennis, 1983; Watson et al., 1984; Houde et al., 1986; Neira, 1998; Leis and Rennis, 2000) Venerus et al.: Early life history of Pseudoperas semifasciata 201 and Prolatilus jugularis (Velez et al., 2003). Larval abundance and distribution have been studied for a few species of Parapercis (Houde et al., 1986; Gaughan et al., 1990; Neira et al., 1992) and, more recently, for Prolatilus jugularis (Velez et al., 2003); no information is available for posttransi- tion pinguipedid juveniles. Larvae of P. semifasciata resembled the larvae of other pinguipedids in their gut size, meristics, and general pattern of pigmentation. They differed from Parapercis spp. and P. jugularis larvae in some relevant features: • The head had no spines and was less rotund, rather moderate instead of large (HL ranged from 0.17 to 0.30 BL; mean HL/BL = 0.22 [±0.02]); • The body was rather elongate instead of mod- erate (BD ranged from 0.12 to 0.26 BL; mean BD/BL = 0.16 [±0.03]); • The notochord flexion occurred between 6.2 and 8.7 mm BL, at a relatively large size range compared to that for Parapercis spp. (3.7-4.8 mm BL) and to P. jugularis (5.7-6.9 mm BL). Pseudopercis semifasciata is a larger and more rotund species; • The finfold was still present in preflexion and flexion larvae. De Cabo3 described some osteological, meristic, and morphological characteristics of Argentine Sea pinguipedid larvae. Like De Cabo3 we found that the first cranial bones that appeared during larval development in P. semifasciata were the premax- illa, the dentary and the cleithrum. These struc- tures were already ossified in 3.4 mm BL preflexion larvae. From the adult osteological descriptions by Herrera and Cousseau (1996) and Gosztonyi and Kuba,2 we determined that the larvae studied were P. semifasciata. The only other sympatric species of Pinguipedidae in the Argentine shelf is the Brazilian sandperch (Pinguipes brasilianus), which shares several similarities in meristic counts with P. semifasciata (Rosa and Rosa, 1987; Herrera and Cousseau, 1996). However, some osteological features from the neuro- and branchiocranium are of great value for identification of larval stages of P. semifasciata. The two species could be dis- tinguished by the placement of the first process of the premaxilla, which is perpendicular to the premaxilla in the Argentine sandperch, and back- inclined in the Brazilian sandperch, drawing an acute angle with the premaxilla (Herrera and Cousseau, 1996). The dentary in P. semifasciata has a quadrangulate anterior end and a margin almost straight, whereas the margin of the dentary in P. brasilianus is oblique (Herrera and Cousseau, 1996). In addition, the head and the teeth patch of the vomer are quadrangulate in Pinguipes and triangular in Pseudopercis (Herrera and Cous- seau, 1996). 35°S 70°W 35°S 70°W Figure 5 Distribution of ichthyoplankton stations (upper) and Pseu- dopercis semifasciata larvae (lower) in the Argentine Sea in the period 1978-2001. Dot diameter, classified into four cat- egories, is proportional to larval abundance at each station (expressed as larvae/10 m2 of sea surface). 202 Fishery Bulletin 103(1) Table 4 Positive stations for Pseudopereis semifasciata larvae in the Argentine Sea, during 1978-2001. ses indicate that only surface temperature was registered. W/d=missing data. Temperature values in parenthe- Abundance Water Cruise Date Sampler Lat. S Long. W (larvae/10 m2 of sea surface) temperature (at 10 m depth) Depth (m) SM-IX 28 Dec 1978 Bongo 42°27' 63°08' 7.36 w/d 70 EH-05/82 22 Nov 1982 Bongo 4039' 60=40' 5.85 (12.8) 53 EH-01/83 21 Jan 1983 Bongo 43°44' 65°00' 4.82 16.7 52 OB-02/85 30 Mar 1985 Bongo 46c30' 67°18' Presence 12.3 56 OB-07/86 20 Dec 1986 Nackthai 43°25' 64'45' 73.91 14.2 34 OB-07/86 20 Dec 1986 Nackthai 43°50' 64" 17' 19.78 14.0 47 OB-01/86 22 Jan 1986 Nackthai 41°33' 62 = 15' 13.76 18.7 45 OB-01/86 22 Jan 1986 Nackthai 41c35' 63=40' 8.64 17.6 51 OB-07/91 02 Nov 1991 Nackthai 36=42' 56°21' 15.33 w/d 20 OB-14/95 12 Dec 1995 Pairovet 43°04' 63°59' Presence 13.0 65 EH-17/96 15 Dec 1996 Nackthai 43°30' 65=05' 23.92 14.3 24 OB-10/98 10 Dec 1998 Nackthai 42 = 21' 62=40' Presence w/d 66 OB-09/99 12 Dec 1999 Nackthai 43°21' 64°52' 17.22 (14.6) 20 OB-09/99 12 Dec 1999 Nackthai 43 = 30' 64 = 29' 41.00 (21.0) 49 OB-14/00 11 Dec 2000 Bongo 43°19' 64°35' 1.81 (12.8) 37 OB -14/00 11 Dec 2000 Bongo 43°30' 64=24' 8.44 13.8 52 EH-01/01 26 Jan 2001 Bongo 43 = 29' 64°35' 2.51 15.9 47 OB-02/01 16 Feb 2001 Bongo 43°18' 64=08' 5.20 15.8 59 OB-13/01 10 Nov 2001 Bongo 42°30' 62°30' 9.30 w/d 71 OB-13/01 11 Nov 2001 Bongo 42°50' 62°55' 9.36 w/d 71 OB-13/01 13 Nov 2001 Bongo 43°25' 64°49' 7.99 w/d 38 The modal number of myomeres (36-38; n = 47) in P. semifasciata larvae matched the number of vertebrae reported for adults (36-37; ?? = 50) by Gonzalez (1998). The dorsal and anal fin elements reached their full complement by 9-10 mm BL, whereas the caudal-, pel- vic-, and pectoral-fin elements were still incomplete in the size range analyzed in this study (3.3 to 11.7 mm BL). Pseudopereis semifasciata and P. brasilianus post- transition juveniles differ in their head shape, pigmen- tation pattern, and in the number of spines of the dorsal fin. The snout is larger in the Brazilian sandperch and the dorsal profile of the head is less convexly shaped than in P. semifasciata. These head shape differences increased with size. In P. brasilianus, the lateral stripes were less conspicuous than in P. semifasciata, and the vertical bars appeared earlier in the development (seven vertical bars were present in ca. 50 mm BL individu- als). Furthermore, vertical bars in P. semifasciata were more defined at the base of the dorsal fin, whereas they extended below the midline in P. brasilianus. Pseu- dopereis semifasciata had five dorsal-fin spines, and P. brasilianus had seven spines, both in the range reported by Herrera and Cousseau (1996). Both the epibenthic sampler and the "Piloto" trawl used to collect juveniles sample the fauna from the bot- tom to approximately one meter above the bottom. The fact that juveniles were caught in the lowest strata of the water column indicates that juveniles had settled to benthic habitat, even though the P. semifasciata post- transition juveniles still conserved some larval pigmen- tation, had not completely developed adult pigmentation pattern, and had already acquired morphological pro- portions similar to adults. Even though the abundance and distribution data used in our study came from cruises that targeted other species, they provide satisfactory spatiotemporal cover- age. This was particularly true for the ichthyoplancton surveys, which covered a great portion of the distribu- tional area of P. semifasciata in the northern Patago- nian shelf, mainly during the peak of the reproductive season (November-December). Among the Piloto posi- tive stations (« = 20), P. brasilianus was found by itself only at three stations. Also, P. brasilianus was far less abundant than P. semifasciata posttransition juveniles in the trawl samples. As a consequence, we consider that the abundance and distribution patterns of post- transition pinguipedid juveniles adequately reflect the abundance and distribution of P. semifasciata posttran- sition juveniles in the Argentine shelf. The abundance and distribution of P. semifasciata larvae and posttransition juveniles indicate the pres- ence of at least three main reproductive grounds, one Venerus et al .: Early life history of Pseudopercis semifasciata 203 Posttransition Pinguipedidae summer surveys 70°W 41 °S 43: 41 °S 70°W B Posttransition Pinguipedidae fall surveys Camarones Bay 41 S 70°W 1-2702 Juvemles/nmr' O 2703-6757 Juveniles/nmi2 O 6758-13.514 Juveniles/nmP Q 13,515-27.027 Juveniles'nmpr J Camarones Bay 70°W 41 °S Figure 6 Distribution of "Piloto" or epibenthic sampler stations (left) and Pinguipedidae posttransition juveniles (right) in the Argentine Sea by season. (A) Summer surveys. (B) Fall surveys. Dot diameter, classified into four categories, is pro- portional to posttransition juvenile abundance at each station (expressed as no. of juveniles/nmi2). located off Peninsula Valdes (42-43°S, 63°W), another off the coast between Engano Bay and Isla Escondida (43-44°S, 64°W to the coast), and the third off north- eastern Camarones Bay (44-45°S, 65°W to the coast). These areas are linked to a frontal zone, the Northern Patagonia frontal system, which is highly productive during the spring and summer and could offer reten- tion mechanisms for larvae (Bogazzi et al., in press). In December 1978, Argentine sandperches of both sexes were observed running near Isla Escondida (Ehrlich, personal observ.). In addition, Elias and Burgos (1988) reported great concentrations of Argentine sandperches off Peninsula Valdes (42-44°S) between October and De- cember, based on commercial fishery data for the period 1981-88. These reproductive grounds are consistent with the principal areas of summer concentration described by Otero et al. (1982). Furthermore, Elias and Burgos (1988) attributed the decline in yields and average size observed in January and February to the dispersal of postspawning individuals. However, initial results from an ongoing tag-recapture program in San Jose Gulf indi- cate that this species may have a high site fidelity and a limited dispersal (Venerus et al., 2003). In this case, the declines in yield and average size as the fishing season progresses could be a consequence of the fishing effort itself. Macchi et al. (1995) detected a decrease in the proportion of females in January, which also may imply an emigration from the reproductive sites. 204 Fishery Bulletin 103(1) Table 5 Positive stations for posttransition pinguipedids in the Argentine Sea, during 1992-2001. The "Species" column show the catego- ries assigned in the survey reports. Underlined items in the "Abundance" column indicate that some or all of the specimens were preserved and at least one individual was correctly identified as Pseudopercis semifasciata. EBS= epibenthic sampler. Cruise Date Season Sampler Lat. S Long. W Species Abundance individuals/nmi') Depth (m) EH-02/92 18 Mar 1992 Summer EBS 42=27' 62°45' Pseudopercis Presence 71 OB-02/01 14 Feb 2001 Summer P loto trawl 43°08' 63°32' Both 4409.5 74 OB-02/01 16 Feb 2001 Summer P loto trawl 43 16' 6407' Pinguipedidae 2572.2 59 OB-02/01 17 Feb 2001 Summer P loto trawl 43°37' 64° 28' Pinguipedidae 1286.1 54 EH-04/98 07 Apr 1998 Fall P loto trawl 44°40' 65°13' Pseudopercis 1492.6 74 EH-04/98 07 Apr 1998 Fall P loto trawl 44°43' 65°00' Pseudopercis 10,204.1 79 EH-04/98 07 Apr 1998 Fall P loto trawl 44°38' 6501' Pseudopercis 4761.9 78 EH-04/98 07 Apr 1998 Fall P loto trawl 44°34' 65°20' Pseudopercis 3448.3 52 EH-04/98 07 Apr 1998 Fall P loto trawl 44°28' 65°14' Pseudopercis 1587.3 61 EH-04/99 28 May 1999 Fall P loto trawl 44°12' 65°14' Pseudopercis 1449.3 34 EH-04/99 28 May 1999 Fall P loto trawl 43°50' 64°44' Pinguipes 1315.8 64 EH-04/99 29 Mayl999 Fall P loto trawl 43°54' 64c30' Pinguipes 1265.8 65 EH-04/99 29 May 1999 Fall P loto trawl 4317' 63°51' Pseudopercis 1250.0 73 OB-05/00 HJun 2000 Fall P loto trawl 44°27' 65'' 13' Pinguipes 2631.6 64 OB-05/00 HJun 2000 Fall P loto trawl 44°34' 65:19' Pseudopercis 1250.0 58 OB-05/00 HJun 2000 Fall P loto trawl 44°41' 65°31' Pseudopercis 1351.4 43 OB-05/00 11 Jun 2000 Fall P loto trawl 44°43' 65°37' Both 27.027.0 38 OB-05/00 15 Jun 2000 Fall P loto trawl 44°15' 6459' Pinguipes 1449.3 72 OB-05/00 18 Jun 2000 Fall P loto trawl 43°50' 64°44' Both 5194.8 59 OB-05/00 18 Jun 2000 Fall Piloto trawl 43°46' 65°01' Pseudopercis 1388.9 52 The low number of positive stations in spite of the intense sampling conducted within the area of distribu- tion of P. semifasciata suggests a reduced spawning site. Both the area off Peninsula Valdes and the one near Isla Escondida have rocky bottoms, which complicates trawling operations. A few experienced captains were able to target P. se?nifasciata by trawling along sandy corridors between rocky outcrops off Peninsula Valdes during the reproductive season (Elias4). Likewise, where running Argentine sandperches were observed near Isla Escondida, trawling is possible only in one orientation (Ehrlich, personal observ. ). This could indicate that spawning grounds are associated with rocky outcrops. Spawning associated with rocky reefs and the existence of chromatic sexual dimorphism is compatible with Mac- chi et al.'s (1995) and Gonzalez's (1998) suggestions of a complex mating system involving sexual courtship. Spawning activity of P. semifasciata in northern Patagonia (42-44°S) peaks in November and Decem- ber (Elias and Burgos, 1988; Macchi et al., 1995), and in October within San Matias Gulf (Gonzalez, 1998). Maximum densities of larvae (>20 larvae/10 m2 of sea surface) were found in December 1986, 1996, and 1999. The temperature at 10 m depth at positive ichthyo- plankton stations varied between 12.3°C and 18.7°C. Such a wide range of temperature reflects the wide latitudinal range in the distribution of P. se?nifasciata and the extended time period (November-March) in which larvae were collected. Posttransition pinguipedid juveniles were mainly col- lected at depths between 60 and 65 m, in both sea- sons sampled (summer and fall). A total of seven P. semifasciata juveniles ranging in total length from 66 to 82 mm were collected in fall (June), near the northern coast of San Matias Gulf (40°58'S-41°00'S; 64°18'W-64024'W), at 29-54 m depth, associated with rib mussel beds (Aulacomya ater) (Gonzalez5). Our dis- tributional data indicate that settlement and nursery grounds could be located near shore. The absence of posttransition juveniles off northeast of Camarones Bay during summer and their presence in the fall could be a consequence of a delayed spawning pulse in the southern stocks. Some independent observations sup- port this hypothesis: 1) back-calculations of hatching date based on daily growth increments from 19 post- 4 Elias, I. 2004. Personal commun. Centra Nacional Pata- gonico, Puerto Madryn, Chubut, Argentina. 5 Gonzalez, R. A. C. 2004. Personal commun. Instituto de Biologia Marina y Pesquera "Alte. Storni," San Antonio Oeste, Rio Negro, Argentina. Venerus et al .: Early life history of Pseudopercis semifasaata 205 transition juveniles collected in northeast Camarones Bay, between 43°50'S and 44°43'S, indicated birth dates between February and March (Venerus and Brown, 2003); 2) the collection of one P. se/nifasciata larva in San Jorge Gulf (46°30'S 67°18'W) on 30 March 1985; and 3) macroscopic observations of the ovaries from 24 mature females angled near Islas Blancas. Camarones Bay (ca. 44°46'S 65°38'W) on 26 and 27 January 2002, most of which (58.3%) were in the late developing stage (rc = 4) or in the gravid and running stage (;;=10) (mac- roscopic maturation stages sensu Gonzalez, 1998). This delayed spawning pulse in the southern stocks appar- ently follows the annual cycle of seawater warming on the Argentine shelf (Ciancio6). Similar delays have been reported for the Argentine hake (Merluccius hubbsi) (Pajaro and Macchi7; Machinandiarena et al.8). Further investigations focused on the seasonal distri- bution of spawners are needed to confirm the existence of spawning aggregations indicated by the presence of larvae and posttransition juveniles. Mark-recapture and telemetry studies could be used to investigate the spa- tial dynamics of reproductive activity of this species in the Argentine Sea. Given the relative sedentary habits of adult Argentine sandperches, the use of reproductive refuges appear a priori to provide a suitable approach to protect this species. Acknowledgments We thank the crew and scientific staff on board for col- lecting the material. We also thank Atila Gosztonyi, Raul Gonzalez, and two anonymous reviewers for pro- viding useful comments on the manuscript. L.A.V. was supported by a fellowship from Consejo Nacional de Investigaciones Cientificas y Tecnicas (CONICET). Literature cited Bogazzi, E., A. Baldoni, A. Rivas, P. Martos. R. Reta, J. M. Orensanz, M. Lasta, and P. Dell Arciprete. In press. Spatial correspondence between areas of concen- tration of Patagonian scallop (Zygochlamys patagonica) and frontal systems in the Southwestern Atlantic. Fish. Oceanograph. 6 Ciancio, J. 2004. Unpubl. data. Centro Nacional Pata- gonieo, Puerto Madryn, Chubut, Argentina. ' Pajaro, M., and G. J. Macchi. 2001. Distribucion espacial y estimaeion de la talla de primera maduracion del stock patagonico de merluza (Merluccius hubbsi) en el periodo de puesta diciembre 2000-abril 2001. INIDEP Inf. Tec. Int. 100, 14 p. I Available from INIDEP: P.O. Box 175 (B7602HSA) Buenos. Aires, Argentina.] 8 Machinandiarena, L., M. D. Ehrlich, D., Brown, M. Pajaro, and E. Leonarduzzi. 2004. Distribucion y abundancia de huevos y larvas de merluza (Merluccius hubbsi) en el litoral norpatagonico. Periodo diciembre 2000 a marzo-abril 2001. Inf. Tec. Int. DNI-INIDEP 29, 16 p. [Available from INIDEP: P.O. Box 175 (B7602HSA) Beunos Aires, Argentina.] Cousseau, M. B.. and R. G. Perrotta. 2000. Peces marinos de Argentina: Biologia, distribu- cion, pesca, 163 p. INIDEP, Mar del Plata, Buenos Aires, Argentina. Elias, I., and G. Burgos. 1988. Edad y crecimicnto del "salmon de mar," Pseu- dopercis semifasciata (Cuvier, 1829) (Osteichthyes. Pin- guipedidae) en aguas norpatagonicas argentinas. Inv. Pesq. 52:533-548. Elias, I., and C. R. Rajoy. 1992. Habitos alimentarios del "salmon de mar" Pseu- dopercis semifasciala (Cuvier, 1829): Pinguipedidae en aguas norpatagonicas argentinas. Rev. Biol. Mar. 27:133-146. Froese, R., and D. Pauly (eds.) 2004. Fishbase. World wide web electronic publication. www.fishbase.org. [Accessed April 2004|. Fulco, V. K. 1996. Aspectos ecologicos y pesqueros del "salmon de mar" Pseudopercis semifasciata (Cuvier, 1829) (Osteichthyes, Pinguipedidae) en aguas norpatagonicas. Licenciatura diss., 21 p. Facultad de Ciencias Naturales, Universi- dad Nacional de la Patagonia San Juan Bosco, Puerto Madryn, Chubut, Argentina. Gaughan, D. J., F. J. Neira, L. E. Beckley, and I. C. Potter. 1990. Composition, seasonality and distribution of ich- thyoplankton in the lower Swan Estuary, south-western Australia. Aust. J. Mar. Freshw. Res. 41:529-543. Gonzalez, R. A. C. 1998. Biologia y explotacion pesquera del salmon de mar Pseudopercis semifasciata (Cuvier, 1829) (Pinguipedidae) en el Golfo San Matias, Patagonia, Argentina. Ph.D. diss., 135 p. Universidad Nacional del Sur, Bahia Blanca, Buenos Aires, Argentina. 2002. Alimentacion del salmon de mar Pseudopercis semi- fasciata (Cuvier, 1829) en el golfo San Matias. IBMP- Serie Publicaciones 1:14-21. Herrera, M., and M. B. Cousseau. 1996. Comparacion del esqueleto oseo de dos especies de peces de la familia Pinguipedidae. Nat. Patagon. (Cienc. Biol.) 4:95-110. Houde, E. D., S. Almatar, J. C. Leak, and C. E. Dowd. 1986. Ichthyoplankton abundance and diversity in the Western Arabian Gulf. Kuwait Bull. Mar. Sci. 8:107-393. Kendall Jr., A. W., and S. J. Picquelle. 1989. Egg and larval distributions of walleye pollock Theragra chalcogramma in the Shelikof Strait, Gulf of Alaska. Fish. Bull. 88:133-154. Leis, J. M., and D. S. Rennis. 1983. The larvae of Indo-Pacific coral reef fishes, 269 p. New South Wales University Press, Sidney, Australia, and Univ. Hawaii Press, Honolulu, HI. 2000. Pinguipedidae — grubfishes, sandperches. In The larvae of Indo-Pacific coastal fishes: an identification guide to marine fish larvae (Fauna Malesiana Hand- books 2) (J. M. Leis and B. M. Carson-Ewart, eds.), p. 565-658. E. J. Brill, Leiden, The Netherlands. Macchi, G. J., I. Elias, and G. E. Burgos. 1995. Histological observations on the reproductive cycle of the Argentinean sandperch, Pseudopercis semifas- ciata (Osteichthyes, Pinguipedidae). Sci. Mar. 59: 119-127. Menezes, N. A., and J. L. Figuereido. 1985. Manual de peixes marinhos do sudeste do Brasil. 206 Fishery Bulletin 103(1) V. Teleostei (4), 105 p. Mus. Zool. Univ. Sao Paulo, Brazil. Neira, F. J. 1998. Pinguipedidae: sandperches, grubfishes In Larvae of temperate Australian fishes: Laboratory guide for larval fish identification I F. J. Neira, A. G. Miskiewicz, and T. Trnski, eds.), p. 362-365. Univ. Western Aus- tralia Press, Nedlands, Western Australia, Australia. Neira, F. J., A. G. Miskiewicz, and T. Trnski. (eds.) 1998. Larvae of temperate Australian fishes. Laboratory guide for larval fish identification, 474 p. Univ. Western Australia Press, Nedlands, Western Australia. Neira, F. J., I. C. Potter, and J. S. Bradley. 1992. Seasonal and spatial changes in the larval fish fauna within a large temperate Australian estuary. Mar. Biol. 112:1-16. Nellen, W., and G. Hempel. 1969. Versuche zur Fangigkeit des "Hai" und des modi- fizierten Gulf-V-plankton-samplers "Nackthai. Meeres- forsch. 20:141-154. Otero, H., S. I. Bezzi, M. Renzi, and G. Verazay. 1982. Atlas de los recursos pesqueros demersales del Mar Argentine 248 p. Contrib. INIDEP 423, Mar del Plata, Buenos Aires. Potthoff, T. 1984. Clearing and staining techniques. //; Ontogeny and systematics of fishes (G. Moser, W. J. Richards, D. M. Cohen, M. P. Fahay, A. W. Kendall, and S. L. Richardson, eds.), p. 35-37. Am. Soc. Ichthy. Herp., Spec. Publ. 1. Rosa, I. L., and R. S. Rosa. 1997. Systematic revision of the South American spe- cies of Pinguipedidae (Teleostei, Trachinoidei). Revta. Bras. Zool. 14:845-865. Rothlisberg, P. C, and W. G. Pearcy. 1976. An epibenthic sampler used to study the ontogeny of vertical migration of Pandalus jordani (Decapoda, Caridea). Fish. Bull. 74:994-998. Smith, P. E., and S. L. Richardson. 1977. Standard techniques for pelagic fish egg and larva surveys. FAO Fish. Tech. Paper 175, 100 p. Smith, P. E., W. C. Flerx, and R. P. Hewitt. 1985. The CalCOFI vertical egg tow (CalVET) net. In An egg production method for estimating spawning bio- mass of pelagic fish: application to the northern anchovy (Engraulis mordax) (R. Lasker, ed.), p. 27-33. NOAA Tech. Rep. NMFS 36. Taylor, W. K. and G. C. Van Dyke. 1985. Revised procedures of staining and clearing small fishes and other vertebrates for bone and cartilage study. Cybium 9:107-119. Velez, J., W. Watson, and E. M. Sandknop. 2003. Larval development of the Pacific sandperch iPro- latilus jugularis) (Pisces: Pinguipedidae) from the Inde- pendencia Bight, Pisco, Peru. J. Mar. Biol. Assoc. UK 83:1137-1142. Venerus, L. A., and D. Brown. 2003. Analisis de la microestructura de otolitos lapilli en juveniles de edad 0 del salmon de mar, Pseudopereis semifasciata. Abstracts of the V Jornadas Nacionales de Ciencias del Mar. Instituto Nacional de Investigacion y Desarrollo Pesquero and LTniversidad Nacional de Mar del Plata, 8-12 Dec. 2003, p. 181. Mar del Plata, Buenos Aires, Argentina. Venerus, L. A., A. M. Parma, and D. E. Galvan. 2003. Dinamica espacial del salmon de mar, Pseudoper- eis semifasciata, en los arrecifes rocosos del golfo San Jose, Argentina. Resultados iniciales de un programa de marcacion. Abstracts of the XXIII Congreso de Ciencias del Mar, Sociedad Chilena de Ciencias del Mar, 5-8 May 2003, p. 152. Sociedad Chilena de Ciencias del Mar and Universidad de Magallanes, Punta Arenas, Chile. Vigliola, L., and M. Harmelin-Vivien. 2001. Post-settlement ontogeny in three Mediterra- nean reef fishes of the genus Diplodus. Bull. Mar. Sci. 68:271-286. Watson, W., A. C. Matarese, and E. G. Stevens. 1984. Trachinoidea: development and relationships. In Ontogeny and systematics of fishes (G. Moser, W. J. Richards, D. M. Cohen, M. P. Fahay, A. W. Kendall, and S. L. Richardson, eds.), p. 554-561. Am. Soc. Ichthyol. Herpet. Spec. Publ. 1. 207 Abstract — Nurseries play an impor- tant part in the production of marine fishes. Determining the relative importance of different nurseries in maintaining the parental population, however, can be difficult. In the west- ern Gulf of Alaska, the Kodiak Island vicinity may be particularly well suited as a pollock nursery because of a prey-rich nearshore environment. Our objectives were 1) to examine age-0 pollock body condition, growth, and diet for evidence of a nearshore-shelf effect, and 2) to determine if variation in the potential prey field of zooplank- ton was associated with this effect. This was a pilot study that occurred in three bays and over the adjacent shelf off east Kodiak Island during 5-18 September 1993. Sampling occurred only during night at loca- tions where echo sign indicated the presence of age-0 pollock. Echo sign was targeted to increase the chance of collecting fish given the limited vessel time. Fish condition was indicated by length-specific body weight. Growth rate indices were estimated for three different periods by using fish length- age data and daily otolith increment widths: 1) from hatching date to cap- ture, 2) 1-5 d before capture, and 3) 6-10 d before capture. Fish diet was determined from gut content analysis. Considerable variation among areas was evident in zooplankton composi- tion, and fish condition, growth, and diet. However, relatively high prey densities, as well as fish condition and growth rates indicated that Chin- iak Bay was particularly well suited as a pollock nursery. Hatching-date distributions indicated that most of the age-0 walleye pollock from bays were spawned earlier than were those from the shelf. The benefit of being reared in nearshore areas is therefore realized more by individuals that were spawned early than by individuals spawned relatively late. Geographic variation among age-0 walleye pollock (Theragra chalcogramma)'. evidence of mesoscale variation in nursery quality?* Matthew T. Wilson Annette L. Brown Kathryn L. Mier Alaska Fisheries Science Center National Marine Fisheries Service, NOAA 7600 Sand Point Way. NE Seattle, Washington 98115 E-mail address (for M T Wilson) matt wilson(a>noaa gov Manuscript submitted 20 November 2003 to the Scientific Editor's Office. Manuscript approved for publication 16 September 2004 by the Scientific Editor. Fish. Bull. 103:207-218 (2005). The location of suitable fish nurseries has long been of interest to fishery scientists (Kendall and Duker, 1998). Such areas are a link in the chain of resources that sustain the produc- tivity of a population and shape its evolution. Although the presence of juvenile fish in an area may indicate a nursery, relative importance among nursery areas ultimately depends on the number and reproductive fit- ness of reared individuals that con- tribute to the parental population. These qualities, however, are usually not measurable. Instead, we focus on measuring the size of juveniles, their body condition, diet, growth, and other characteristics that are acces- sible and relevant to fish survival. However, because these indices are not free of measurement error, it is advisable to consider more than one index (Suthers, 1998). In the North Pacific Ocean, wall- eye pollock {Theragra chalcogramma) have adapted to the heterogeneity and productivity of coastal areas; they now support one of the world's most productive fisheries. Walleye pollock are a semidemersal gadid. Spawning typically occurs in mid-water during the spring at locations near, or over, the continental shelf (Kendall and Picquelle, 1989; Bailey et al., 1997). Fertilization is external. The eggs and larvae are pelagic, remaining in the plankton for ca. 4 months while they are dispersed over large areas. At 25-40 mm standard length (SL), larvae transform to juveniles (Brown et al., 2001) and become increasingly nektonic. Juveniles are referred to as "age-0" when they are between tran- sition and 12-months old (40-130 mm SL, Brodeur and Wilson, 1996a). They are zooplanktivorous, feeding mostly on copepods and euphausiids, but other taxa sometimes dominate their diet (Brodeur and Wilson, 1996a). Age-0 juveniles commonly occur in various habitats from nearshore to the outer continental shelf (Nakatani and Maeda, 1987; Sobolevskiy et al., 1992; Carlson, 1995; Natsume and Sasaki, 1995; Brodeur and Wilson, 1996a; Wilson, 2000). Occasionally, they are found farther offshore (Tang et al., 1995), but probably in small numbers (Brodeur et al., 1999; Shida et al., 1999). The early life stages of walleye pol- lock have been extensively studied in the Gulf of Alaska (GOA) (Kendall et al., 1996). In the Gulf, young pollock are most abundant in the western region (Brodeur and Wilson, 1996a). This region is naturally divided into two areas by the Shelikof Sea Val- ley, which cuts through the shelf at ca.l56°N longitude (Fig. 1). To the east, the Kodiak vicinity includes the continental shelf around the Kodiak Island Archipelago. To the west, the lower Alaska Peninsula vicinity ex- tends to Unimak Pass at the Penin- sula's southwestern terminus. During the 1980s, age-0 abundance in the Contribution FOCI-0417 to NOAA's Fisheries- Oceanography Coordinated Investigations, 7600 Sand Point Way NE, Seattle, WA 98115. 208 Fishery Bulletin 103(1) 165'OOTM 160°0'OW 155'0'OW ISO'O'CTW 145J0'0W Alaska /' _■ ,. V /" 57.8 57.6 - 57.4 - 57 2 NE shelf Sampling gear + CTD o plankton u trawl Vkv> 152.8 152 4 152.0 Longitude (°W) 151.6 151 2 Figure 1 Location of sampling operations (CTD, plankton, and trawl) conducted during 5-18 September 1993, Kodiak Island, Alaska, to examine geo- graphic variation among age-0 walleye pollock I Theragra chalcogramma I. The ocean currents, shown as arrows on upper map, are adapted from Reed and Schumacher (1986). Kodiak vicinity was related to the recruitment of pollock to the GOA fishery (Wilson, 20001. Furthermore, age-0 juveniles in this vicinity were large in comparison to those collected elsewhere (Wilson, 2000). The large size of the "Kodiak" juveniles may reflect faster growth (Bai- ley et al., 1996) due to a rich diet of euphausiids (Me- rati and Brodeur, 1996). In contrast, the diet of age-0 pollock along the Lower Peninsula was dominated by larvaceans (Merati and Brodeur, 1996). Interestingly, high densities of age-0 pollock were closer to shore in the Kodiak vicinity than along the Lower Peninsula where the shelf is relatively broad. The apparent richness of the Kodiak Island vicinity may reflect its relative upstream position in the Alaska Coastal Current (ACC) (Fig. 1). Stabeno et al. (2004) integrated much research on the ACC to provide a com- prehensive view of its importance in circulation over the GOA shelf. The ACC is wind driven and structured by seasonal influxes of fresh water. Flow is generally southwestward over the shelf but there is considerable topographic influence. For example, landmasses at the northern entrance to Shelikof Strait (Kennedy-Steven- son Entrance) allow only about 70% of the ACC water to enter the Strait. The remaining 30% of the water flows south around the northeastern end of the Kodiak Archipelago. This bifurcation of flow occurs in an area of vigorous tidal mixing and localized upwelling, both of which contribute to increased biological productiv- ity. Off the northeastern Archipelago, Stabeno et al. (2004) have shown that the ACC follows bathymetric contours into and out of sea valleys, thus, providing some across-shelf movement of water. Advection of wa- ter was found by Coyle et al. (1990) to be important in the enhancement of zooplankton in Auke Bay, which is in the eastern GOA. Less is known about the exchange of water and zooplankton between the bays and fjords of the western GOA and the adjacent shelf. Thus, the ACC probably helps enrich the waters off northeastern Wilson et al.: Geographic variation among age-0 Theragra chakogramma 209 Kodiak Island, but we do not yet understand how this actually affects walleye pollock in nearshore nurseries. In this article, we present information from a pilot study to better understand the environmental basis for the apparent richness of the Kodiak Island vicinity as a pollock nursery. Our objectives were 1) to examine age-0 pollock size, body condition, growth, and diet for evidence of geographic effect (nearshore versus shelf), and 2) to determine if their potential prey field (i.e., zooplanktonl was associated with this effect. Materials and methods This study was conducted as an ancillary project during a research cruise off east Kodiak Island, 5-18 Septem- ber 1993 (Fig. 1). In this area, the shelf is about 50 nmi wide and has an offshore bank (Albatross Bank) crossed by deep gullies (Barnabas and Chiniak gullies) extend- ing from the slope to the coast. Bays form the upper reaches of these troughs and receive seasonal influxes of freshwater (Rogers et al.1). Over the shelf, net transport is southwestward (ca. 5 em's) (Stabeno et al., 1995). A boundary current, the Alaska Stream, exists farther offshore and flows rapidly to the southwest (Reed and Schumacher, 1986). Sampling was conducted from the NOAA ship Miller Freeman (Fig. 1). Sampling occurred only at night to avoid complications of diel fish movement (Brodeur and Wilson, 1996b) and feeding patterns (Merati and Bro- deur, 1996). A 38-kHz, Simrad-EK500 echo-sounder system was used to help guide our sampling to locations where age-0 pollock were likely present. The targeting of echo signs resulted in an irregular sample-location pattern and biased estimation of fish abundance; how- ever, it focused our sampling at locations where age-0 pollock were likely present and thereby contributed to successful fish collections. Sampling was accomplished in four areas: Chiniak Bay, Ugak Bay, Kiliuda Bay, and over the adjacent shelf. All data analyses included these four areas as geographic strata; finer divisions (e.g., in- ner and outer Kiliuda Bay, and NE and Albatross Bank) were not possible given the available data and chosen analytical methods. Age-0 pollock were obtained from the four areas with a bottom trawl and a midwater trawl (Wilson et al., 1996). The codend of each trawl was lined with a 3-mm mesh net. Towing speed averaged 4.5 k/h. Previous comparisons between these trawls indicated no sig- nificant difference with regard to estimation of age-0 pollock size or abundance (Brodeur and Wilson, 1996a; Wilson et al., 1996). Differences in the sampling effort Rogers, D. E., D. J. Rabin, B. J. Rogers, K. J. Garrison, and M. E. Wangerin. 1979. Seasonal composition and food web relationships of marine organisms in the nearshore zone of Kodiak Island — including ichthyoplankton, meroplankton (shellfish), zooplankton, and fish. Annual rep. OCSEAP RU553, FRI-UW-7925. 291 p. Fish. Res. Inst., Univ. Wash- ington, Seattle, WA. used to collect each sample were corrected by dividing the age-0 catch by the volume filtered. Volume filtered was estimated by multiplying the distance fished (me- ters traveled while at depth) by the mouth opening of the trawl (m2) (Wilson, 2000). Thus, age-0 catches are reported as number of fish per m3. Size composition of walleye pollock for each area was estimated by measuring the standard length (SL) of fresh age-0 pollock to the nearest millimeter. For large catches, a random subsample of about 300 individuals was used to represent the entire catch; otherwise, SL on every individual was measured. Length frequencies were expanded to the standardized catch estimates. Age-0 juveniles were clearly distinguishable from older pollock (<130 mm versus >150 mm SL) as indicated by Brodeur and Wilson (1996a). Random subsamples of age-0 pollock were also frozen at sea for subsequent de- termination of body condition, age, growth, and diet. In the laboratory, length-specific weights of 776 age-0 pollock were used to examine area differences in body condition (Table 1). The fish were thawed within four months of collection. Excess water was blotted from each individual, and each specimen was measured to the nearest millimeter SL and weighed whole to the nearest 0.01 gram. Afterwards, each carcass was stored in 95% ethanol for eventual gut content analysis. Lengths and somatic weights, obtained from the subset of fish used in the gut analysis, were also analyzed to verify that geographic differences in condition were not dependent on whole versus somatic weight. Growth rate was estimated for 128 individuals by using fish length and age data. Age, in days, was esti- mated as the number of daily increments visible in the microstructure of sagittal otoliths following Brown and Bailey (1992). Length-age relationships were examined for evidence of an area effect on growth rates integrated over the period from hatching to capture. We used these relationships to convert the length composition for each sample to a hatching-date distribution, and by summing across samples we then obtained area-specific hatching- date distributions. To estimate growth rate realized near the point of capture we measured the width of recent daily otolith increments. Following Bailey (1989), we measured the width of the two outermost, nonoverlapping 5-increment bands on each of 97 sagittal otoliths. These widths were assumed to relate directly to body growth during the first (1-5 days) and second (6-10 days) 5-d periods before capture, and that the increments were deposited while individuals were near the point of capture. Thus, growth rate indices were obtained for three different periods: 1) hatching date to capture date, 2) 1-5 days before capture, and 3) 6-10 days before capture. Gut content analysis was conducted on 300 individu- als according to the method of Merati and Brodeur (1996) to determine feeding intensity and taxonomic composition of age-0 prey. No more than 15 fish per sample were examined. Each fish was measured (SL), blotted dry, and weighed immediately prior to dissec- tion. Stomachs were excised between the esophagus and 210 Fishery Bulletin 103(1) Table 1 Number of age-0 walleye pollock (Theragra chalcogramma) collected near Kodiak Island, Alaska, September 1993, measured for standard length, and examined in the laboratory to estimate condition, growth, and the weight and taxonomic composition of stomach contents. Sample is the number of trawl hauls. At-sea collections Laboratory examinations (no. offish) Growth (no offish) Condition Band width Evaluated for gut content Sample Measured whole' somatic- weight and Location (n) Caught forSL wt. wt. Age 1-53 6-10' composition Chiniak Bay 7 1858 709 223 75 23 17 17 75 Ugak Bay 4 2506 773 218 91 28 12 12 91 Kiliuda Bay 7 562 279 165 66 41 33 33 66 Shelf 14 358 358 170 68 36 35 35 65 All combined 32 5284 2119 776 300 128 97 97 297 ; Whole wet weights from thawed fish. - Somatic wet weights from fish preserved in 95f4 ethanol after freezing at sea. 3 Collective width of daily otolith increments 1-5; numbering begins with the most peripheral increment. 4 Collective width of daily otolith increments 6-10. pylorus. Gut contents were dissected from the speci- mens and weighed to the nearest 0.001 gram. Somatic weight represented whole wet weight minus the gut content weight. Three fish were omitted from further consideration because of apparent regurgitation. Taxo- nomic composition of age-0 diets was determined by counting the organisms in the gut after sorting them into broad taxonomic groups. Zooplankton was collected by using a 1-m Tucker net (333-/mi mesh) to sample where age-0 pollock had been collected. The net was fished through acoustic echo lay- ers believed to be age-0 pollock in order to characterize their immediate prey field. Potential prey items were sorted into broad taxonomic groups and enumerated at the Polish Plankton and Identification Center, Szezcin, Poland. Temperature and salinity profiles (near surface to 10 m off bottom) were obtained by using a Seabird SBE- 911+ CTD system. Profile data were collected during deployment at a descent rate of ca. 0.5 m/s. Statistically significant differences in age-0 condition, growth, and feeding intensity among geographic areas were detected with split-plot analysis of covariance (AN- COVA) and post hoc multiple comparison tests (Proc Mixed, SAS software, Littell et al., 1996). The covari- ates were fish length or age (days since hatching). Fol- lowing Milliken and Johnson (2002), we first tested for covariate significance (H0: all slopes = 0) and homogene- ity of slopes (H(): equal slopes) to ensure appropriateness of the following reduced, common-slope model: Y = a + (5x:j + Area ( + Sample, I Area 1 1 + eljk , where Y = dependent variable; a = intercept parameter; P = slope parameter; x = covariate for sample i and area,/'; and e k = replicate error for sample /', area./', and fish k. A split-plot design was necessary to account for the nesting of samples (trawl catches) within area, and individuals within sample. To avoid pseudoreplication, trawl catch was the sampling unit instead of individual fish. Area was a fixed effect; sample was a random effect. For body condition, lengths and weights were log(,-transformed according to the method of Patterson (1992); two points were omitted because of suspiciously low length-specific, whole-body weight. For feeding in- tensity, gut content weights (GCW) were fourth-root transformed (GCW025) to linearize the GCW-length relationship and remove heteroscedasticity (Clarke and Warwick. 2001). Significance of post hoc pairwise dif- ferences was based on a Bonferroni-corrected, 0.05-level of significance. The standardized catch data were not incorporated into these tests; therefore the conclusions pertain to the samples not weighted by catch. Nonmetric multidimensional scaling (NMS, PC-Ord, McCune and Mefford, 1999) was used to ordinate the diet and plankton samples according to taxonomic composition. Each diet sample represented the aver- age numerical composition of the diet of all fish in the sample. This value was calculated by dividing the sum of all items within each taxonomic category by the num- ber of fish in the sample. The ordinations, one for diet and another for plankton, were based on Bray-Curtis similarity coefficients of fourth root-transformed data. Differences among the four areas were statistically tested by using a two-way nested analysis of similarity Wilson et al.: Geographic variation among age-0 Theragra chalcogramma 211 Salinity (psu) 30 31 32 33 34 30 31 32 33 34 30 31 32 33 34 30 31 32 33 34 0 - if 25 - psu I J «c / I50: u Depth 00 25 - Ugak Bay I i 8 10 12 Chiniak Bay 8 10 12 4 6 8 10 12 4 6 8 10 12 Temperature (°C) Figure 2 Water salinity and temperature profiles obtained in 10 casts at locations where age-0 walleye pollock {Theragra chalcogramma) were collected near Kodiak Island, Alaska, during 5-18 September 1993. (ANOSIM, PRIMER, Clarke and Warwick, 2001) ap- plied to the Bray-Curtis similarity matrices. Results Overall, salinity ranged from 30.3 to 33.0 ppt, and water temperature ranged from 4.4 to 11.3°C (Fig. 2). Shal- low surface layers of relatively fresh water were evident from low near-surface salinities in Ugak Bay and in the inner part of Kiliuda Bay. This part of Kiliuda Bay was also well stratified thermally. Unfortunately, it was not possible to include inner Kiliuda Bay as a fifth area in subsequent statistical analyses because of insufficient sampling. Thermal stratification was also evident at shelf sampling locations. A total of 5284 age-0 pollock were collected in 25 of the 32 successful trawl hauls (Table 1). These fish were absent only at the four most-offshore locations over Albatross Bank and Chiniak Gully (Fig. 3 1. In ad- dition, no age-0 pollock were caught in shallow (<35-m depth) tows at locations on Albatross Bank; a dense and expansive school of capelin (Mallotus villosus) may have displaced them downward. Median age-0 density was 0.0006 fish/m3; the maximum (0.095 fish/m3) was found in Ugak Bay. Standard lengths of 2119 age-0 pollock ranged from 25 to 121 mm SL (Table 1, Fig. 4). The fish in Chiniak Bay (91 mm SL), Ugak Bay (90 mm SL), and Kiliuda Bay (89 mm SL) all had a median SL that were larger than the median length of fish collected over the shelf (71 mm SL). A surprising number of individuals <50 mm SL were collected in Ugak Bay and inner Kiliuda Bay. Body condition, based on the reduced, common-slope ANCOVA model, varied among the four areas (Table 2). Because of this effect, area-specific equations were used to describe the length-weight relationship (Table 3, Fig. 5A). After accounting for differences in length, we found that fish from the shelf weighed less than the individuals collected in Chiniak Bay and Ugak Bay. Individuals from Kiliuda Bay were intermediate in weight, differing only from the Ugak Bay fish (Table 4). Similar conclusions from the somatic-weight data of fish used in the diet examinations indicated that gut- content weight was not responsible for the relatively low length-specific weights of fish from Kiliuda Bay and the shelf (Tables 2 and 4). The fish age-length relationship also varied by area. The relationship was described by using a reduced, common-slope model (Table 2). The common slope was 0.78 mm/d (Table 3, Fig. 5B). Differences in line eleva- tion, or age-specific length, indicated that fish from the shelf grew more slowly during the hatch-to-capture period than did the fish from Chiniak or Kiliuda bays (Table 4). Applying these equations to the length data resulted in hatching-date distributions that ranged from mid March to mid July (Fig.6). The fish collected in Chiniak Bay (17 April), Kiliuda Bay (20 April), and Ugak Bay (25 April) all had earlier median hatching 212 Fishery Bulletin 103(1) 578 576 57.4 - 57.2 152 8 152 4 152 0 Longitude (°W) 151.6 151.2 Figure 3 Geographic distribution of standardized catches (no. of individuals/m3) of age-0 walleye pollock iTheragra chalcogramma) collected in trawl hauls conducted near Kodiak Island during 5-18 September 1993. dates in comparison to fish from the shelf (8 May). In- terestingly, the hatching dates of the cohort of small in- dividuals from Ugak Bay and inner Kiliuda Bay ranged from June to July. Mean otolith increment width varied with area. It was not necessary to include fish length as a covariate (Table 2). For the 1-5 d precatch period, the large mean increment width associated with fish from Chiniak Bay (0.036-mm band width) was different from the means of Chiniak Bay Ugak Bay Kiliuda Bay Shelf 30 40 50 60 70 80 90 100 110 120 Standard length (mm) Figure 4 Size composition (mm SL) of age-0 walleye pollock (Thcr- agra chalcogramma) by area from samples collected near Kodiak Island, 5-18 September 1993. each other area (Table 4). The only other difference was between the Kiliuda Bay (0.026 mm) and shelf (0.030 mm) areas. The only difference for the 6-10 d precatch period was again between the Kiliuda Bay (0.029 mm) and shelf (0.036 mm) areas. No area effect on gut content weight (GCW) was de- tected (Table 2). There was, however, a significant fish length effect (Fig. 5C), and this was incorporated in the final model (Table 3). After adjusting for length, area- specific mean GCW agreed in rank with area-specific fish weight (Table 4). Differences in taxonomic composition of age-0 pollock diets resulted in a good separation of samples by area (Fig. 7A, ANOSIM, K = 0.533, P=0.001). Each pair-wise comparison of areas resulted in a significant difference (P<0.05) (the one sample of small fish from Kiliuda Bay, and two samples from the shelf of fish with empty stomachs were omitted from the ANOSIM). The diet of fish from Ugak Bay and Kiliuda Bay were mostly crab larvae or copepods, depending on fish size (Table 5A). Over the shelf, fish diets comprised mostly euphausiids (74%). In contrast, fish from Chiniak Bay had a much more varied diet; no single prey category exceeded 40% of the items per stomach. Note the correspondence be- tween the number of prey per fish (Table 5A) and mean gut-content weight (Table 4); both were lowest for fish from the shelf. Differences in taxonomic composition also resulted in separation of the plankton samples by area (Fig. 7B, ANOSIM, i? = 0.886, P=0.001). Pair-wise comparisons indicated a difference between Chiniak Bay and the shelf (fl = 0.813, P=0.029). Ugak Bay was not included in the comparisons because only one sample was avail- Wilson et al.: Geographic variation among age-0 Theragra chalcogramma 213 Table 2 Summary results of six ANCOVA tests of an area effect on six pollock: body condition (whole or somatic weight), three indices and denominator degrees of freedom for the F test, respectively dependent of growth, variables obtained from laboratory analysis of age-0 ind gut content weight. NDF and DDF are numerator Dependent variable H0: all slopes = 0 Ho: equal slopes Reduced model Source NDF DDF Type III F P>F Condition whole weight P=0.0001 P=0.3340 Area 3 14.2 6.81 0.0045 ln(SL) 1 769 105824 0.0001 somatic weight P=0.0001 P=0.3115 Area 3 17.8 10.01 0.0004 ln(SL) 1 294 31000 0.0001 Growth age-specific length P=0.0001 P=0.4140 Area 3 4.3 14.43 0.0106 Age 1 117 475.39 0.0001 1-5 d band width P=0.2645 Area 3 93 8.05 0.0001 6-10 d band width P=0.2267 Area 3 5.76 3.89 0.0768 Gut content weight P=0.0001 P=0.7208 Area SL 3 1 16.2 285 0.36 201.99 0.7850 0.0001 able. Copepods dominated the catches in Chiniak Bay and over the shelf, whereas larval crabs were most prevalent in the Ugak and Kiliuda samples (Table 5B). In terms of overall abundance, mean prey densities were lowest among samples collected from the shelf and highest for the Chiniak Bay samples. Discussion The presence of age-0 pollock in bays and over the inner shelf, but not over the outer shelf, indicates that the principal pollock nurs- ery off east Kodiak Island during autumn is relatively close to shore. Earlier studies of age-0 pollock in the western GOA focused on near- shore areas (Smith et al., 1984; Wilson, 2000) and did not docu- ment the absence of age-0 pollock over the outer shelf. Our results point to prey resource as a likely explanation for the observed distribution of and differences among age-0 walleye pollock. Seasonal declines in zooplankton density underscore the importance of nearshore areas as pollock nurseries. Rogers et al.1 and Kendall et al.2 observed an order-of- magnitude autumnal decline in prey3 density off Kodiak Island during 1977-79 (Fig. 8). This decline was accom- panied by a shoreward shift in the region of highest eu- Table 3 Least-squares linear relationships used to describe the condition, growth, and feeding intensity of age-0 walleye pollock collected September 1993, Kodiak Island, Alaska. GCW = gut content weight. Relationship Location Equation Condition whole weight Chiniak Bay ln(£) = 3.228(ln SLmm) - 12.646 Ugak Bay ln(£) = 3.228(ln SLmm) - 12.609 Kiliuda Bay \n(g) = 3.2281 In SLmm) - 12.659 Shelf ln(£) = 3.228(ln SLmm) - 12.698 somatic weight Chiniak Bay ln(g) = 3.127* In SLmm) - 12.708 Ugak Bay ln(g) = 3.127(ln SLmm) - 12.702 Kiliuda By ln(£) = 3.127(ln SLmm) - 12.774 Shelf ln(g) = 3.127(ln SLmm) - 12.816 Growth length-at-age Chiniak Bay age(d) = 0.782(SLmm) - 22.294 Ugak Bay age(d) = 0.782(SLmm) - 26.928 Kiliuda Bay age(rf) = 0.782(SLmm) - 21.346 Shelf age(rf) = 0.782(SLmm) - 31.011 Gut content weight All combined GCW(£025) = 0.007(SLmm) - 0.192 2 Kendall, A. W., Jr., J. R. Dunn, R. J. Wolotira Jr., J. H. Bowerman Jr., D. B. Dey, A. C. Matarese, and J. E. Munk. 1980. Zooplankton, including ichthyoplankton and deca- pod larvae, of the Kodiak shelf. NOAA NWAFC proc. rep. 80-8, 393 p. Alaska Fishery Science Center, Seattle, WA. 3 All invertebrate zooplankters are considered potential age-0 pollock prey except cnidarians, ctenophores, siphonophores, and larval shrimps and crabs. Shrimp and crab were omitted from Figure 8 because density estimates were not available separately for the shelf and slope regions. 214 Fishery Bulletin 103(1) phausiid density. Similar to our findings, the estimates of larval crab densities from Rogers et al.'s and Kendall et al.'s studies were always highest in bays. In autumn. 18 16 14 12 10 all locations Chiniak Bay Ugak Bay Kiliuda Bay Shelf 50 75 100 Standard length (mm) 125 B I^U - 100 ■ ojljj 80 - "J^Be* t 60 - t ' Chiniak y^^ a Ugak 40 ■ o Kiliuda • Shelf 80 100 120 140 Age (d) c 160 180 0.5 • V Chiniak Bay Ugak Bay CO 04 • G Kiliuda Bay • Shelf D) 0.3 ■ all locations 1 c c o o 02 ■ 0 3 0.1 ■ 00 ■ Jgjfc ^ k.' * •*, 20 40 60 80 Standard length (mm) 100 Figure 5 Least-squares regressions of age-0 walleye pollock tTheragra chalcogramma) length on weight (Al, length on age (B), and gut content weight on length (C) for individuals collected from four areas off east Kodiak Island, 5-18 September 1993. the larger zooplankters are of principal importance to age-0 pollock because of size-related changes in diet (Table 5A, Merati and Brodeur, 1996). By all accounts, age-0 pollock collected from Chiniak Bay fared as well or better than individuals in each of the other areas sampled. Wilson (2000) found that the density of age-0 pollock in the Chiniak Bay vicinity predicted Gulf-wide recruitment. However, these fish represent a minuscule part of the Gulf-wide population of age-0 pollock. Even if two cohorts, from spring- and summer-spawnings, were produced, it seems unreason- able to expect that local production would dramati- cally affect gulf-wide recruitment. Alternatively, the abundance and condition of age-0 pollock in this vicin- ity might reflect larger-scale processes that relate to gulf-wide recruitment. Identifying large-scale processes based on small-scale sampling, however, is complicated by variation at high spatial and temporal frequencies. For example, the relatively high density of pea crab (Fabia subquadrata) megalopae in combination with influxes of freshwater (Epifanio, 1988) indicate that local dynamics are important in sustaining prey popu- lations in Ugak and Kiliuda bays. In contrast, Chiniak Bay might be more affected by influxes of oceanic prey. Such influxes could be facilitated by cross-shelf sea val- leys, which extend into all the fjords that we sampled. Indeed, Kendall et al.,2 Lagerloef (1983), and Stabeno et al. (2004) have all shown that the local sea valleys in- duce cross-shelf flow in the ACC. Furthermore, Inzce et al. (1997) found that zooplankton density was elevated in the Shelikof Sea Valley above the density found at adjacent shelf areas; a similar phenomenon, however, was not observed off northeastern Kodiak Island (Ken- dall et al.2). Compared to the other bays, Chiniak Bay might be best positioned to receive enriched ACC water that flows south from where it bifurcates at the en- trance to Shelikof Strait. Such enriched water may also be an important transport mechanism for immigrating larval and juvenile pollock (Wilson, 2000). Because of the inconsistency among our various indi- ces (i.e., weight-at-length, length-at-age, otolith incre- ment width), it is difficult to conclude that fish over the shelf and in Ugak and Kiliuda bays were prey limited. Over the shelf, recent growth rates were not low despite relatively small individual size and low prey density. For example, the low prey densities and small fish sizes over the shelf contrasted with recent fish growth that was not low. Age-0 pollock are capable of social foraging behavior to compensate for food scar- city (Ryer and Olla, 1992), but it is unclear that the associated energetic cost (Ryer and Olla, 1997) would depress body weight before slowing otolith growth. In contrast, fish in Kiliuda Bay had relatively slow recent growth and low body weight, but age-specific length was large. The observed differences in age-specific length are somewhat discounted by the fact that such differences may have arisen any time after hatching and are not necessarily indicative of recent differences in growth. Another complication was our inability to reconstruct the spatial history of the sampled fish; Wilson et al.: Geographic variation among age-0 Theragra chalcogramma 215 Table 4 Least-squares ad lected September comparison tests. usted means of indices of body condition, growth, and gut content weight (GCW) 1993, Kodiak Island, Alaska. Means sharing the same superscript letter are not P>0.05). of age-0 walleye pollock col- different (post hoc multiple Location Condition Growth GCW („0.25) whole wt. (lng) somatic wt. (\ng) age-specific SL (mm) 1-5 d band width (mm) 6-10 d band width (mm) Chiniak Bay 1.47"6 0.82" 82.3° 0.036" 0.033°* 0.40 LTgak Bay 1.50* 0.83° 77.6°* 0.027*'' 0.032°* 0.43 Kiliuda Bay 1.45ac 0.76* 83.2" 0.026* 0.029° 0.39 Shelf 1.42c 0.72* 73.6* 0.030' 0.036* 0.36 Chiniak Bay Ugak Bay Kiliuda Bay Shelf n~~ 1 1 r^ r~ 21 Mar 10 Apr 30 Apr 20 May 9 Jun 29 Jun Hatching date during 1993 Figure 6 Hatching-date composition of age-0 walleye pollock (Theragra chalcogramma) by area from samples collected near Kodiak Island, 5-18 September 1993. in other words, we did not know where they had been prior to capture. As evidenced by geographic variation in hatching-date distributions, cohort-specific differences persisted well into the juvenile stage and had important implications for inter-cohort differences in survival. The median hatching dates of fish in bays were similar to those es- timated for north Kodiak Island by Brown and Bailey (1992). In contrast, fish over the shelf had substantially later hatching dates. There is little evidence of pollock spawning within our study area; therefore it seems likely that the differences in hatching dates reflect suc- cessive immigration of sequential cohorts. However, the presence of the youngest cohort, fish hatched during A ▲ ▲ ▲ 2.0 1.0 s ■ _!_ V Chiniak B A Ugak B □ Kiliuda B O Shelf stress=17.84 r2=0.81 ■ 0 0 - O ov V A A □ D D 10 - o o O □ D -2 0 -1 5 -1.0 -0 5 0 0 0.5 10 1.5 E CO B 15 1 0 05 1 0.0 -0.5 -1.0 -I -1 5 stress=9.79 r2=0.87 V O O -15 -10 -0.5 0 0 0 5 NMS dimension 1 1 0 Figure 7 Nonmetric multidimensional scaling (NMS) of samples based on age-0 walleye pollock (Theragra chalcogramma) diet composition (A), or zooplankton composition (B). Symbols indicate four different sampling areas. For diet, the small (<66 mm SL) fish in bays are represented by filled symbols. 216 Fishery Bulletin 103(1) Bays (Rogers el al 1) Nearshore (Kendall et a! *) 1000 A - ^ T Middle shelf (Kendall etal ) .A B Slope (Kendall et al.3) <| 800 - / s / \ « ' * \ CO / .••••. \ 3 / •• ■■ \ ? 600 - / •••' '■•■ \ / •■' '•■ v c / • \ „_ /.■' ■ \ d 400 / A \\ c >> / / \ \ 'to / \ V S 200 •i— -^ ^~^\ \^~ J£— ■- -*" ^""--^.^^ D — ■*** -^ T**Tt Spring Summer Fall Winter Figure 8 Seasonal variability in densities (no. of indiv iduals/m3) of inverte- brate zooplankton near Kodiak Island based on samples collected with 60-cm bongo nets with 0.333-mm mesh ( modified from Rogers et al.1, Kendall et al.2). June and July, only in the innermost parts of Kiliuda Bay and Ugak Bay, may indicate an alternative mechanism such as local spawn- ing and geographic differences in retention. Regardless, the relationship between fish size and hatching date indicates that large individuals were spawned early; thus, early spawned individuals might experience higher overwinter survival, which often increases with fish size (Sogard, 1997). We chose to track echo sign in our study to reduce the sampling effort expended in areas devoid of age-0 pollock. This meth- od maximized our chance of collecting the samples needed to study differences among age-0 pollock given the limited vessel time. Unfortunately, this method also introduced a bias, thereby reducing the utility of den- sity estimates to indicate habitat suitability (Brown et al., 2000; Stoner et al., 2001) and to extrapolate from samples to at-sea popu- lations. Our focus, however, was on other measures that might eventually provide a Table 5 Numerical composition of age-0 pollock diet samples (B) concurrently collected during 5- by location and predator SL (A), and composition of the plankton in 1-m2 -18 September 1993, Kodiak Island, Alaska, "t" signifies trace (<0.05). Tucker A Prey (number of individuals/fish )' No. of Area Sample (ra) no. of fish amphipod chaetognath copepod crab larvae cumacean euphausiid mysid shrimp larvae prey/ fish Chiniak Bay 5 75 2.8 t 4.1 0.1 0.0 3.7 0.4 t 11.0 Ugak Bay <66 mm 3 31 0.1 0.6 11.2 0.6 0.0 0.1 t 0.0 12.6 >66 mm 4 60 0.1 t 5.9 15.9 t 3.3 0.2 t 25.6 Kiliuda Bay <66 mm 1 5 0.0 0.2 2.2 1.0 0.0 0.2 0.0 0.0 3.6 >66 mm 6 61 t t t 6.4 t 2.7 t t 9.2 Shelf 6 65 0.3 0.0 0.1 0.1 t 2.8 0.4 0.1 3.8 ' 070 cm ML. come from the four squid that grew from 47-53 to 71-74 cm ML in 207-224 days, and these measurements yield a mean DGR of 1.05 [±0.05] mm/day (Fig. 5 and V in Fig. 6). Solid and dashed curves in Figure 6 represent DGR independently determined for both sexes through analysis of statolith increments (Markaida et al., 2004) for squid of a comparable size range. These growth rates are about twice those determined in the present study by direct ML measurements. Determination of daily growth rate (DGR) Dorsal ML was measured from forty-four squid tagged off Guaymas after recapture at 4 to 224 days. ML values ranged between 46 and 80.7 cm. Variability in DGR determination, as indicated by the standard devia- tion (SD) of binned data from 20-day intervals, clearly decreased as the time to recapture increased. Thus, a significant negative correlation exists between the SD of DGR and recovery time (r2=0.88, P<0.01, n = 6) (Fig. 5). Six measurements of squid caught before 40 days yielded negative growth rates. This finding indicates that large discrepancies in DGR calculations exist in measure- ments on squid with short recapture times, because any errors in ML measurement are generally much larger. Growth rate estimates from squid captured after 40 days yielded values of 1.0-1.5 mm/day (SD of 0.05-0.6). We regard these as the only reliable data. Further analysis of DGR was limited to squid cap- tured after 40 days. Figure 6 illustrates DGR versus "mean" ML (average of ML at times of tagging and re- capture) for selected squid of different sexes and stages of maturity. Probably the most reliable DGR estimates Tagged off Guaymas n = 995 Recaptured off Guaymas n = 58 Recaptured off Sta Rosalia n=18 Recapture rate 0.20 - 0.15 0.10 0.05 000 Discussion Tag return rates High recovery rates obtained in our study clearly demon- strate that D. gigas in the Gulf of California is suitable for tagging studies. This large species is relatively easily tagged with conventional plastic tags, and the tagging operation produced no obviously deleterious effects on the squid. These features make jumbo squid an attrac- tive species for application of archival electronic tags or telemetry devices. Despite extensive tagging efforts and intense com- mercial fisheries, recapture rates for other species of ommastrephid squid have generally been much lower. In the extreme case, no recaptures whatsoever were ob- tained for the northern shortfin squid (lllex illecebrosus) tagged in offshore waters of Newfoundland (Hurley and Dawe, 1981). In other studies recaptures ranged from 0.03-0.1% for the Argentine shortfin squid (/. argenti- nus) in the Southwest Atlantic (Brunetti et al., 2000), to 1.0-6.2% for the European flying squid (Todarodes sag- ittatus) off Norway (Wiborg et al.2). The neon flying squid iOmmastrephes bartramii) from the North Pacific also yielded low rates (0.1-0.5%; Murata and Nakamura, 1998; see also Nagasa- wa et al., 1993). In 62 years of tagging studies of Japanese flying squid (Todarodes pacificus), only a few experiments carried out in the Sea of Japan and Tsugaru Strait yielded return rates that match those of the present study (up to 16.4%; see Nagasawa et al., 1993). The highest tag recovery rate (19-32%) was found for the northern shortfin squid in Newfound- land inshore areas (Hurley and Dawe, 1981). Recapture rates of up to 12.7% have also been reported for large, neritic loliginid squid (Naga- sawa et al., 1993; Sauer et al., 2000). In the present study, recapture rate was found to be directly proportional to mantle length, ranging from <3.5% for squid <50 cm ML (cm) at tagging Figure 4 Mantle length (ML) distribution for all squid tagged off Guay- mas (white bars) and for those recaptured off Guaymas (gray bars) and Santa Rosalia (black bars). Black circles represent recapture rate. 2 Wiborg, K. F., J. Gjosaeter, I. M. Beck, and P. Fossum. 1982. The squid Todarodes sagittatus (Lamarck). Distribution and biology in Northern waters, August 1981-April 1982. Council Meet. Int. Coun. Explor. Sea (K:30):l-17. ICES, Palaegade 2-4, DK-1261, Copenhagen K, Denmark. info@ices.dk. NOTE Markaida et al .: Tagging studies on Dosidicus gigas 223 ML to 20% for squid close to 80 cm ML (Fig. 4). Reasons for this strong size-dependence are not clear. Smaller squid may either suffer a higher natural mortality rate or migrate southward out of the Guaymas basin more readily than the larger squid. We do not be- lieve that the tagging process itself leads to such a difference in mortality rate, but this possibility cannot be ruled out. Several factors are relevant to evaluating differences in recapture rates for jumbo squid and other ommastrephids. First, squid of the other species are not as large as jumbo squid. We are not aware of any other published data on size-dependence of recapture rates, but this phenomenon may be relevant. Second, the localized nature of the fisheries surround- ing the Guaymas basin equates with high concentrations of squid in relatively small ar- eas that are intensively fished. Most recent tagging studies of other ommastrephids have taken place in oceanic waters in the Sea of Japan and North Pacific, where the fishing zone is extremely large and far from any lo- calized coastal fishing areas (Nagasawa et al, 1993). The extreme disparity in return rates for nearshore versus offshore studies in Newfoundland supports this idea. Third, an ambitious advertising campaign (posters) and the substantial reward offered for tag returns undoubtedly stimulated a high degree of cooperation in the largely artisanal Mexi- can fishery that is highly concentrated in Sta. Rosalia and Guaymas. A strong dependence of tag-return rate on rewards and advertis- ing has been previously noted (see Nagasawa et al., 1993). Seasonal migration Results from this study directly demonstrate that jumbo squid in the Guaymas basin migrate across the Gulf on a seasonal basis. Squid appear to migrate from Sta. Rosalia to Guaymas during the second half of November and early December and to make the reverse trip in late May and early June. Thus, large squid (40-80 cm ML) remain available to fish- eries surrounding the Guaymas basin through- out the year. These data support the idea that these fishing areas are feeding grounds (Markaida and Sosa-Nishizaki, 2001). What fraction of squid, if any, migrate southward out of the Guaymas basin and potentially into the Pacific cannot be ascertained from our data. Transit time across the Gulf for the migrat- ing squid appears to be fairly brief — proba- bly less than 16 days based on the overlap of recaptures in both fishing areas. Assuming a straight-line distance of 130 km between CO 1 XI ' E E DC O a • . • *T • 6 * □ • • 0 -- -2 -3 -I 0 10 20 30 40 50 60 70 80 90 100 210 220 Days elapsed between tagging and recapturing Figure 5 Daily growth rate (DGR) in mantle length (ML) determined for squid recaptured at different times after tagging. Black circles represent measurements from individual animals. Gray squares represent means ±1 SD for squid grouped in 20-day bins. Note that a gap exists between 100 and 200 days. v Recaptured after 200 days o Immature Female • Mature Female Maturing Male Mature Male Mean ±SD for each 5-cm ML bin Females (Markaida et al.. 2003) Males ra 2.0 E E rr a 60 65 70 Mean ML (cm) 80 Figure 6 Relationship of daily growth rate (DGR) and mean mantle length (ML) (average of measured ML at time of tagging and time of recapture). Small symbols represent measurements from selected individual animals as follows: squid recaptured after >200 days (V), immature female (Ol, mature female (•), maturing male (A), mature male (A). Larger squares (■) indicate means ±1 SD for all data pooled into 5-cm bins. Analysis was limited to squid recaptured after 40 days and of identified sex. Curves represent DGR vs. ML relationship as determined by counting statolith increments for females (solid) and males (dashed) and are adapted from Markaida et al. (2004). 224 Fishery Bulletin 103(1) these areas, the average maintained speed during the migration would be about 8 km/day. A comparable fig- ure can be derived from one of our first squid to be recaptured. This animal was tagged at Pt. Prieta (see Fig. 2) and recaptured 20 km away off Sta. Rosalia (Fig. 2) after three days. This estimated velocity for a trans-Gulf migration is well within the range of rates observed in other studies of ommastrephids (O'Dor, 1988). Jumbo squid tracked with acoustic telemetry off Peru covered 3-5 miles in 8-14 hours, or about 14 km/day (Yatsu et al., 1999). Neon flying squid tracked in the same way covered up to 22 km per day (Nakamura, 1993). Migration rates obtained from tagging studies yielded even higher es- timates. Maximum speed for migrating short-finned squid has been estimated at 20-30 km/day (Dawe et al., 1981; Hurley and Dawe, 1981), and high rates have also been reported for the Japanese squid (see Nagasawa et al., 1993). Large loliginid squid have been reported to migrate at rates of 3 to 17 km/day (see Sauer et al., 2000). Daily growth rates Variance in DGR estimates from ML measurements decreased dramatically after 30 days after tagging, and became fairly consistent by 50 days. Clearly, estimates of DGR in our study are only reliable for these later times, and a DGR of 1-1.5 mm/day in ML is evident for squid in the 50-70 cm range of ML (Fig. 6). These absolute rates would correspond to relative rates of 0.15-0.22% increase in ML per day. There are few comparable estimates of growth rates for other ommastrephid squids based on tag-recapture studies. However, the neon flying squid grows 0.5-2.7 mm/day in the 18-48 cm ML range (Araya, 1983), and good agreement exists between growth rates obtained from tag-recapture studies and those from statolith ag- ing studies (Yatsu et al., 1997). When converted to rela- tive growth rate, this species would thus appear to grow substantially faster than the jumbo squid. The common Japanese squid grows 0.45 mm/day (Nagasawa et al., 1993), but for this species, mantle lengths were not given; therefore relative rates cannot be estimated. More importantly, absolute growth rates determined by direct ML measurements in the present study dis- agreed with those derived from statolith aging methods (Markaida et al., 2004), and this discrepancy merits re-evaluation of previous longevity estimates. Squid of 50 cm ML are thought to be about 260 days old based on statolith ring counts, and our tag-recapture study revealed that it can take another 200 days to grow to 70 cm ML. The estimated age at this size would therefore be 460 days, about 100 days more than that estimated by statolith aging for squid of 70 cm ML (Markaida et al., 2004). Thus, the largest squid found in the Gulf of California (about 90 cm ML) might be up to 2 years old. Reasons for the apparent underestimates in longevity with statolith aging are unclear. Difficulty in resolving discrete rings late in life of a specimen is one possibil- ity. Another is that the assumed daily ring deposition may not occur throughout the lifetime of a jumbo squid. No successful validation studies have been reported for this species, either in the laboratory or in the wild. Squid distributions in the Gulf in relation to commercial landings Although large-scale migrations of jumbo squid within the Guaymas basin are apparently responsible for the seasonal pattern in the commercial landings (Fig. 1; see also Markaida and Sosa-Nishizaki, 2001), the biological and oceanographic reasons for these migrations are not well established. The reciprocal pattern in squid distri- bution between the eastern and western central Gulf is correlated with the wind-driven upwelling seasonality in this area (Roden and Groves, 1959) and is probably highly influenced by this oceanographic feature. A simi- lar situation exists in the life cycle of another important pelagic resource, the Pacific sardine (Sardinops caeru- leus) (Hammann et al., 1988). However, other biological factors are also probably important. Summer upwelling in the western Gulf is actually less intense than off the eastern coast in win- ter (Hammann et al., 1988; Santamaria-del-Angel et al., 1999), yet 80% of squid landings were made at Sta. Rosalia between 1995 and 1997 (Markaida and Sosa- Nishizaki, 2001). We propose that concentrations of spawning myctophids (lanternfishes) off Baja California in the summer (Moser et al., 1974) may be largely re- sponsible for this disparity because these fish are a ma- jor prey item for squid in the Guaymas basin (Markaida and Sosa-Nishizaki, 2003). Data in the present study also indicate that jumbo squid may be available to commercial fishing efforts off each coast for a longer period than previously thought. Our data indicate that squid were recovered in the waters off Guaymas throughout the year; therefore it is likely that some squid do not undergo the westward spring migration (Fig. 3B). However, it is not certain that the final returns from Guaymas after 7 months were of this resident stock, because they would have had time to migrate to Sta. Rosalia and back again. It is also unclear whether a resident stock of squid exists in the Sta. Rosalia area year-round. Strong northern winds in this area lead to a cessation of commercial fishing efforts during the winter months, and the lack of tag returns during winter may simply reflect this fact. Long-distance migrations into and out of the Gulf of California Although data in this paper have demonstrated seasonal migrations of jumbo squid within the Guaymas basin, migration patterns into this region from the southern Gulf and open Pacific (and back out) remain unknown. The much lower level of commercial fishing effort in these latter areas will greatly constrain efforts to elu- NOTE Markaida et al : Tagging studies on Dosidicus gigas 225 cidate migrations over these longer distances using conventional tag-and-recapture approaches. Presumably, as the largest eunektonic squid, jumbo squid should be able to perform large-scale migrations covering its whole geographic range as do other om- mastrephids (O'Dor, 1988). The high tag return rates achieved in the present study, in conjunction with the large size of the squid, make application of a variety of archival electronic tagging devices an attractive pos- sibility. Such devices could reveal long-distance migra- tions across the large range of jumbo squid in a fishery- independent manner. Acknowledgments We acknowledge funding for this project by the Tagging of Pacific Pelagics (TOPP) program and the Census of Marine Life (COML). We thank Oscar Sosa-Nishizaki (CICESE, Ensenada) for administering this project in Mexico and providing laboratory space and facili- ties. Volunteer field workers in Sta. Rosalia included A. Novakovic, J. Schulz, S. Sethi (Stanford Univ.), and L. Roberson (California State University, Northridge). We are also indebted to personnel of Centro Regional de Investigacion Pesquera, especially Manuel O. Nevarez, Paco Mendez, and Araceli Ramos, for their support during tagging and tag recovering at Guaymas, and to Sandra Patricia Garaizar and Vicente Monreal for recovering tags in Sta. Rosalia. We extend our sincere gratitude to all fishermen and squid factory personnel for their cooperation. Literature cited Araya, H. 1983. Fishery, biology and stock assessment of Ommas- trephes bartrami in the North Pacific Ocean. Mem. Nat. Mus. Victoria 44:269-283. Bischoff, J. L., and J. W. Niemitz. 1980. Bathymetric maps of the Gulf of California: bathym- etry of the Central Gulf Province, Map 1-244. Miscel- laneous Investigation Series, U. S. Geological Survey, Denver, CO. Sheet 3 of 4 sheets. Brunetti, N. E., M. L. Ivanovic, G. R. Rossi, B. Elena, H. Benavides, R. Guerrero, G. Blanco, C. Marchetti, and R. Pinero. 2000. JAMARC-INIDEP joint research cruise on Argen- tine short-finned squid [Illex argentinus). January- March 1997. Argentine Final Report. INIDEP Inf. Tec. 34:1-36. Dawe, E. G., P. C. Beck, H. S. Drew, and G. H. Winter. 1981. Long distance migration of a short-finned squid Illex illecebrosus. J. Northwest Atlantic Fish. Sci. 2:75-76. Ehrhardt, N. M., P. S. Jacquemin, F. Garcia B., G. Gonzalez D., J. M. Lopez B., J. Ortiz C. and A. Soli's N. 1983. On the fishery and biology of the giant squid Dosidi- cus gigas in the Gulf of California, Mexico. In Advances in assessment of world cephalopod resources (J. F. Caddy, ed.), p. 306-339. FAO Fish. Tech. Pap. 231. Hammann, M.G., T. R. Baumgartner, and A. Badan-Dangon. 1988. Coupling of the Pacific sardine iSardinops sagax caeruleus) life cycle with the Gulf of California pelagic environment. CalCOFI Rep. 29:102-109. Hurley, G. V., and E. G. Dawe. 1981. Tagging studies on squid {Illex illecebrosus) in the Newfoundland area. North Atlantic Fisheries Organi- zation (NAFO) SCR Doc, No. 80/11/33, serial no. 072, lip. NAFO, Dartmouth, Nova Scotia, Canada. Ichii, T., K. Mahapatra, T. Watanabe, A. Yatsu, D. Inagake, and Y. Okada. 2002. Occurrence of jumbo flying squid Dosidicus gigas aggregations associated with the countercurrent ridge off the Costa Rica Dome during 1997 El Nino and 1999 La Nina. Mar. Ecol. Prog. Ser. 231:151-166. Klett, A. 1982. Jumbo squid fishery in the Gulf of California, Mexico. In Proceedings of the International Squid Symposium. August 9-12, 1981, Boston, Massachusetts (prepared by the New-England Fisheries Development Found., Inc.), p. 81-100. UNIPUB. New York, NY. Lipinski, M. R., and L. G. Underhill. 1995. Sexual maturation in squid: Quantum or continuum? S. Afr. J. Mar. Sci. 15:207-223. Mangold, K. 1976. La migration chez les Cephalopodes. Oceanis 2(8):381-389. [In French.! Markaida, U., and O. Sosa-Nishizaki. 2001. Reproductive biology of jumbo squid Dosidicus gigas in the Gulf of California, 1995-1997. Fish. Res. 54(l):63-82. 2003. Food and feeding habits of jumbo squid Dosidi- cus gigas (Cephalopoda: Ommastrephidae) from the Gulf of California, Mexico. J. Mar. Biol. Assoc. U.K. 83:507-522. Markaida, U., C. Quinonez-Velazquez, and O. Sosa-Nishizaki. 2004. Age, growth and maturation of jumbo squid Dosi- dicus gigas (Cephalopoda: Ommastrephidae) from the Gulf of California, Mexico. Fish. Res. 66(11:31-47 Morales-Bojorquez, E., M. A. Cisneros-Mata, M. O. Nevarez-Martinez and A. Hernandez-Herrera. 2001. Review of the stock assessment and fishery biology of Dosidicus gigas in the Gulf of California, Mexico. Fish. Res. 54:83-94. Moser, H. G., E. H. Ahlstrom, D. Kramer, and E. G. Stevens. 1974. Distribution and abundance offish eggs and larvae in the Gulf of California. CalCOFI Rep. 17:112-128. Murata, M., and Y Nakamura. 1998. Seasonal migration and diel vertical migration of the neon flying squid, Ommastrephes bartramii, in the North Pacific. In Contributed papers to the interna- tional symposium on large pelagic squids, Tokyo, July 18-19, 1996 (T. Okutani, ed.), p. 13-30. Japan Marine Fishery Resources Research Center (JAMARC). Nakamura, Y 1993. Vertical and horizontal movements of mature females of Ommastrephes bartramii observed by ultra- sonic telemetry. In Recent advances in cephalopod fisheries biology: contributed papers to 1991 CIAC inter- national symposium and proceedings of the workshop on age, growth and population structure (T. Okutani, R. K. O'Dor, and T. Kubodera, eds.), p. 331-336. Tokay Univ. Press, Tokyo, Japan. Nagasawa, K, S. Takayanagi, and T. Takami. 1993. Cephalopod tagging and marking in Japan: a review. In Recent advances in cephalopod fisheries 226 Fishery Bulletin 103(1) biology: contributed papers to the 1991 Cephalopod International Advisory Council (CIAC) international symposium and proceedings of the workshop on age, growth, and population structure (T. Okutani, R. K. O'Dor, and T. Kubodera, eds.), p. 313-329. Tokay Univ. Press, Tokyo, Japan. Nesis, K. N. 1983. Dosidicus gigas. In Cephalopod life cycles, vol. I, species accounts (P. R. Boyle, ed), p. 215-231. Academic Press, London, UK. Nigmatullin, Ch. M., K. N. Nesis, and A. I. Arkhipkin. 2001. A review of the biology of the jumbo squid Dosi- dicus gigas (Cephalopoda: Ommastrephidae). Fish. Res. 54:9-19. O'Dor, R. K. 1988. The energetic limits on squid distributions. Mal- acologia 29(1):113-119. Roden, G. I., and G. W. Groves. 1959. Recent oceanographic investigations in the Gulf of California. J. Mar. Res. 18:10-35. Roper, F. E., M. J. Sweeney, and C. E. Nauen. 1984. FAO species catalogue: vol. 3, Cephalopods of the world, 272 p. FAO, Rome. SAGARPA (Secretaria de agricultura, ganaderia, desarrollo rural, pesca y alimentacion). 2001. Anuario estadistico de pesca 2000. SAGARPA- CONAPESCA (Comision Nacional de Acuacultura y Pesca), Mexico, 268 p. [In Spanish.] Santamaria-del-Angel, E., S. Alvarez-Borrego, R. Millan-Niinez, and F. E. Miiller-Karger. 1999. On the weak effect of summer upwelling on the phytoplankton biomass of the Gulf of California. Rev. Soc. Mex. Hist. Nat. 49:207-212. [In Spanish, English abstract.] Sauer, W. H. H.. M. R. Lipinski, and C. J. Augustyn. 2000. Tag recapture studies of the chokka squid Loligo vulgaris reynaudii d'Orbigny, 1845 on inshore spawning grounds on the south-east coast of South Africa. Fish. Res. 45:283-289. SEMARNAP (Secretaria del medio ambiente, recursos naturales y pesca). Distributed by the Mexican Secretary on Natural Resources. 1996. Anuario estadistico de pesca 1995, 235 p. SEMAR- NAP, Mexico. [In Spanish.] 1997. Anuario estadistico de pesca 1996, 232 p. SEMAR- NAP, Mexico. [In Spanish.] 1998. Anuario estadistico de pesca 1997, 241 p. SEMAR- NAP, Mexico. [In Spanish.] 1999. Anuario estadistico de pesca 1998, 244 p. SEMAR- NAP, Mexico. [In Spanish.] 2000. Anuario estadistico de pesca 1999, 271 p. SEMAR- NAP, Mexico. [In Spanish.] Taipe, A., C. Yamashiro, L. Mariategui, P. Rojas, and C. Roque. 2001. Distribution and concentrations of jumbo flying squid iDosidicus gigas) off the Peruvian coast between 1991 and 1999. Fish. Res. 54:21-32. Yatsu, A., S. Midorikawa, T. Shimada, and Y Uozumi. 1997. Age and growth of the neon flying squid, Ommas- trephes bartrami, in the North Pacific Ocean. Fish. Res. 29:257-270. Yatsu, A., K. Yamanaka, and C. Yamashiro. 1999. Tracking experiments of the jumbo squid, Dosidi- cus gigas, with an ultrasonic telemetry system in the Eastern Pacific Ocean. Bull. Nat. Res. Inst. Far Seas Fish. 36:55-60. Fishery Bulletin 103(1) 227 Superintendent of Documents Publications Order Form *5178 I I YrLo, 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) MasterCard ( Purchase Order No. ) 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! U.S. Department of Commerce Volume 103 Number 2 April 2005 U.S. Department of Commerce Carlos M. Gutierrez Secretary National Oceanic and Atmospheric Administration Vice Admiral Conrad C. Lautenbacher Jr., USN (ret.) Under Secretary for Oceans and Atmosphere National Marine Fisheries Service William T. Hogarth Assistant Administrator for Fisheries ^T0Fc%, S?ATES 0* The Fishery Bulletin (ISSN 0090-0656) is published quarterly by the Scientific Publications Office, National Marine Fish- eries Service, NOAA, 7600 Sand Point Way NE, BIN C15700, Seattle, WA98115-0070. Periodicals postage is paid at Seattle, WA, and at additional mailing offices. POST- MASTER: Send address changes for sub- scriptions to Fishery Bulletin, Superin- tendent of Documents, Attn.: Chief, Mail List Branch, Mail Stop SSOM, Washing- ton, DC 20402-9373. Although the contents of this publica- tion have not been copyrighted and may be reprinted entirely, reference to source is appreciated. The Secretary of Commerce has deter- mined that the publication of this peri- odical is necessary according to law for the transaction of public business of this Department. Use of funds for printing of this periodical has been approved by the Director of the Office of Management and Budget. For sale by the Superintendent of Documents, U.S. Government Printing Office, Washington, DC 20402. Subscrip- tion price per year: $55.00 domestic and $68.75 foreign. Cost per single issue: $28.00 domestic and $35.00 foreign. See back for order form. Scientific Editor Norman Bartoo, Ph.D. Associate Editor Sarah Shoffler National Marine Fisheries Service, NOAA 8604 La Jolla Shores Drive La Jolla, California 92037 Managing Editor Sharyn Matriotti National Marine Fisheries Service Scientific Publications Office 7600 Sand Point Way NE, BIN C15700 Seattle, Washington 981 15-0070 Editorial Committee Harlyn O. Halvorson, Ph.D. Ronald W. Hardy, Ph.D. Richard D. Methot, Ph.D. Theodore W. Pietsch, Ph.D. Joseph E. Powers, Ph.D. Harald Rosenthal, Ph.D. Fredric M. Serchuk, Ph.D. George Watters, Ph.D. University of Massachusetts, Boston University of Idaho. Hagerman National Marine Fisheries Service University of Washington, Seattle National Marine Fisheries Service Universitat Kiel, Germany National Marine Fisheries Service National Marine Fisheries Service Fishery Bulletin web site: wwAV.fishbull.noaa.gov The Fishery Bulletin carries original research reports and technical notes on investigations in fishery science, engineering, and economics. It began as the Bulletin of the United States Fish Commission in 1881; it became the Bulletin of the Bureau of Fisheries in 1904 and the Fishery Bulletin of the Fish and Wildlife Service in 1941. Separates were issued as documents through volume 46; the last document was No. 1103. Beginning with volume 47 in 1931 and continuing through volume 62 in 1963, each separate appeared as a numbered bulletin. A new system began in 1963 with volume 63 in which papers are bound together in a single issue of the bulletin. Beginning with volume 70, number 1, January 1972, the Fishery Bulletin became a periodical, issued quarterly. In this form, it is available by subscription from the Superintendent of Documents, U.S. Government Printing Office, Washington, DC 20402. It is also available free in limited numbers to libraries, research institutions, State and Federal agencies, and in exchange for other scientific publications. U.S. Department of Commerce Seattle, Washington Volume 103 Number 2 April 2005 Fishery Bulletin Contents MAY 1 0 2005 The conclusions and opinions expressed in Fishery Bulletin are solely those of the authors and do not represent the official position of the National Manne Fisher- ies Service (NOAA) or any other agency or institution. The National Manne 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 229—245 Alonzo, Suzanne H., and Marc Mangel Sex-change rules, stock dynamics, and the performance of spawmng-per-recruit measures in protogynous stocks 246-257 Brandon, Elisif A. A., Donald G. Calkins, Thomas R. Loughlin, and Randall W. Davis Neonatal growth of Steller sea lion (Eumetopias jubatus) pups in Alaska 258-269 Brouwer, Stephen L, and Marc H. Griffiths Reproductive biology of carpenter seabream (Argyrozona argyrozona) (Pisces: Sparidae) in a marine protected area 270-279 Burn, Douglas M., and Angela M. Doroff Decline in sea otter (Enhydra lutris) populations along the Alaska Peninsula, 1986-2001 280-291 Carlson, John K„ and Ivy E. Baremore Growth dynamics of the spinner shark (Carcharhinus brevipinna) off the United States southeast and Gulf of Mexico coasts: a comparison of methods 292-306 Domeier, Michael L, Dale Kiefer, Nicole Nasby-Lucas, Adam Wagschal, and Frank O'Brien Tracking Pacific bluefin tuna (Thunnus thynnus orientalis) in the northeastern Pacific with an automated algorithm that estimates latitude by matching sea-surface-temperature data from satellites with temperature data from tags on fish 307-319 Fischer, Andrew J., M. Scott Baker Jr., Charles A Wilson, and David L. Nieland Age, growth, mortality, and radiometric age validation of gray snapper (Lut/anus gnseus) from Louisiana 320-330 Grabowski, Robert C, Thomas Windholz, and Yong Chen Estimating exploitable stock biomass for the Maine green sea urchin {Strongylocentrotus droebachiensis) fishery using a spatial statistics approach Fishery Bulletin 103(2) 331—343 Lowry, Mark S., and Karin A. Forney Abundance and distribution of California sea lions (Zalophus caltfornianus) in central and northern California during 1998 and summer 1999 344—354 Mackie, Michael C, Paul D. Lewis, Daniel J. Gaughan, and Stephen J. Newman Variability in spawning frequency and reproductive development of the narrow-barred Spanish mackerel (Scomberomorus commerson) along the west coast of Australia 355—370 Ruggerone, Gregory T., Ed Farley, Jennifer Nielson, and Peter Hagen Seasonal marine growth of Bristol Bay sockeye salmon (Oncorhynchus nerka) in relation to competition with Asian pink salmon (O. gorbuscha) and the 1977 ocean regime shift 371—379 Shoji, Jun, and Masaru Tanaka Distribution, feeding condition, and growth of Japanese Spanish mackerel (Scomberomorus niphonius) larvae in the Seto Inland Sea 380-391 Wang, You-Gan, and Nick Ellis Maximum likelihood estimation of mortality and growth with individual variability from multiple length-frequency data 392-403 Williams, Erik H., and Kyle W. Shertzer Effects of fishing on growth traits: a simulation analysis Notes 404-406 Burton, Michael L, Kenneth J. Brennan, Roldan C. Muhoz, and Richard O. Parker Jr. Preliminary evidence of increased spawning aggregations of mutton snapper (Lut/anus ana/is) at Riley's Hump two years after establishment of the Tortugas South Ecological Reserve 411—416 Carpentieri, Paolo, Francesco Colloca, Massimiliano Cardinale, Andrea Belluscio, and Giandomenico D. Ardizzone Feeding habits of European hake (Mer/ucaus mer/uccius) in the central Mediterranean Sea 417—425 Gobert, Bertrand, Alain Guillou, Peter Murray, Patrick Berthou, Maria D. Oqueli Turcios, Ester Lopez, Pascal Lorance, Jerome Huet, Nicolas Diaz, and Paul Gervain Biology of the queen snapper (Etelis oculatus: Lutjanidae) in the Caribbean 426—432 Graham, Rachel T., and Daniel W. Castellanos Courtship and spawning behaviors of carangid species in Belize 433—437 Hewitt, David A., and John M. Hoenig Comparison of two approaches for estimating natural mortality based on longevity 438-444 Lindquist, David C, and Richard F. Shaw Effects of current speed and turbidity on stationary light-trap catches of larval and juvenile fishes 445—452 Macchi, Gustavo J., Marcelo Pajaro, and Adrian Madirolas Can a change in spawning pattern of Argentine hake (Merlucaus hubbsi) affect its recruitment? 453—460 Raymundo-Huizar, Alma R., Horacio Perez-Espana, Maite Mascaro, and Xavier Chiappa-Carrara Feeding habits of the dwarf weakfish (Cynosaon nannus) off the coasts of Jalisco and Colima, Mexico 461-466 Wood, Anthony D. Using bone measurements to estimate the original sizes of bluefish (.Pomatomus saltatnx) from digested remains 467 Subscription form 229 Abstract— Predicting and under- standing the dynamics of a popula- tion requires knowledge of vital rates such as survival, growth, and repro- duction. However, these variables are influenced by individual behavior, and when managing exploited popu- lations, it is now generally realized that knowledge of a species' behav- ior and life history strategies is required. However, predicting and understanding a response to novel conditions — such as increased fish- ing-induced mortality, changes in environmental conditions, or specific management strategies — also require knowing the endogenous or exogenous cues that induce phenotypic changes and knowing whether these behaviors and life history patterns are plastic. Although a wide variety of patterns of sex change have been observed in the wild, it is not known how the specific sex-change rule and cues that induce sex change affect stock dynamics. Using an individual based model, we examined the effect of the sex-change rule on the predicted stock dynamics, the effect of mating group size, and the performance of traditional spawn- ing-per-recruit (SPR) measures in a protogynous stock. We considered four different patterns of sex change in which the probability of sex change is determined by 1) the absolute size of the individual, 2) the relative length of individuals at the mating site, 3) the frequency of smaller individuals at the mating site, and 4) expected reproductive success. All four pat- terns of sex change have distinct stock dynamics. Although each sex- change rule leads to the prediction that the stock will be sensitive to the size-selective fishing pattern and may crash if too many reproductive size classes are fished, the performance of traditional spawning-per-recruit mea- sures, the fishing pattern that leads to the greatest yield, and the effect of mating group size all differ distinctly for the four sex-change rules. These results indicate that the management of individual species requires knowl- edge of whether sex change occurs, as well as an understanding of the endogenous or exogenous cues that induce sex change. Manuscript submitted 22 September 2003 to the Scientific Editor's Office. Manuscript approved for publication 20 December 2004 by the Scientific Editor. Fish. Bull. 103:229-245 (2005). Sex-change rules, stock dynamics, and the performance of spawning-per-recruit measures in protogynous stocks Suzanne H. Alonzo Institute ol Marine Sciences and the Center for Stock Assessment Research (CSTAR) University of California Santa Cruz 1156 High Street Santa Cruz. California 95064 Present address: Department of Ecology and Evolutionary Biology Yale University 165 Prospect St., OML 427 New Haven, Connecticut 06511 Email address Suzanne Alonzoia'yaleedu Marc Mangel Department of Applied Mathematics and Statistics Jack Baskm School of Engineering and the Center for Stock Assessment Research (CSTAR) University of California Santa Cruz 1156 High Street Santa Cruz California 95064 Growth, survival, and reproduction all affect the dynamics of a population and its response to fishing and man- agement (Quinn and Deriso, 1999; Haddon, 2001). However, these three key variables are influenced by many aspects of a species' biology, environ- ment, and evolutionary history. There is an increasing realization that the management of populations requires an understanding of their behavior, life history strategies, and repro- ductive patterns (Sutherland, 1990; Huntsman and Schaaf, 1994; Col- lins et al„ 1996; Greene et al„ 1998; Sutherland, 1998; Beets and Fried- lander, 1999; Coleman et al., 1999; Fulton et al., 1999; Kruuk et al., 1999; Constable et al., 2000; Cowen et al., 2000; Koeller et al„ 2000; Fu et al., 2001; Apostolaki et al., 2002; Levin and Grimes, 2002). Although it is important to document the normal patterns of behavior and reproduction within a population, predicting and understanding a stock's response to novel conditions also requires knowl- edge of the degree of plasticity in behaviors that affect growth, survival, and reproduction, and the cues that induce phenotypic changes. Numer- ous examples exist of context- and condition-dependent behavior in fish (e.g., Metcalfe et al., 1989; Snyder and Dingle, 1990; Schultz and Warner, 1991; Wainwright et al., 1991; Mit- telbach et al., 1992; Nishibori and Kawata, 1993; Ridgeway and Shuter, 1994; Breden et al., 1995), and this kind of plasticity has the potential to affect the dynamics of a stock. For example, many commercially impor- tant species of fish change sex from female to male. Researchers have argued that this life history pat- tern will lead to different population dynamics and responses to fishing and management strategies than will the life history pattern of dioecious (sep- arate-sex) species (e.g., Snyder and Dingle, 1990; Schultz and Warner, 1991; Wainwright et al., 1991; Nishi- bori and Kawata, 1993; Ridgeway and Shuter, 1994; Alonzo and Mangel, 2004). However, it is important to consider not only whether sex change occurs, but also how it occurs; whether plasticity in sex change exists and what cues determine sex change in an individual species. A variety of patterns of sex change have been observed in the wild (War- ner and Lejeune, 1985; Charnov, 1986; Shapiro, 1987; Charnov and Bull, 1989; Iwasa, 1990; Warner and Swearer, 1991; Lutnesky, 1994, 1996; 230 Fishery Bulletin 103(2) Kuwamura and Nakashima, 1998; Koeller et al., 2000; Nakashima et al., 2000). At one extreme, sex change may occur at a fixed size or age threshold. However, sex change is known in many species to be mediated by local factors such as population density, reproduc- tive skew, sex ratio, and size distribution (Warner and Lejeune. 1985; Warner and Swearer, 1991; Lutnesky, 1994, 1996; Kuwamura and Nakashima, 1998; Koeller et al., 2000; Nakashima et al., 2000). In many sex- changing species, overlap exists between the sexes in size and age and this overlap indicates that sex change may also depend on individual experience and local conditions (Munoz and Warner, 2003). The pattern of sex change may have important implications for a spe- cies' response to fishing. For example, if the size at sex change is fixed, then the population sex ratio may be affected by size-selective fishing of males, resulting in sperm limitation and decreased larval production (Alonzo and Mangel, 2004). In contrast, if sex change is mediated at the level of the spawning group in single- male harems and mating group size remains the same, sex ratios are maintained if the largest female always changes sex. In such a case, larval production will be reduced only because of the decreased size distribution of the population due to fishing. However, if sex change is controlled by the reproductive skew in the group (e.g., the expected potential for reproduction as a male versus present fecundity as a female), then the largest individ- ual might not change sex and the spawning group could be without a male (Munoz and Warner, 2003). This re- sult would clearly lead to a much greater effect on the productivity of the stock. A detailed understanding of the factors determining sex change and the cascading effects on sperm production, fecundity, and sex ratio can be critical to predicting stock dynamics. Furthermore, most animals have "rules-of-thumb" which determine their behavior and reproduction. Although these rules will have evolved under normal conditions, in the pres- ence of fishing or other human-induced disturbances, animals are likely to continue to use these behavioral rules on ecological time scales even if they no longer function to maximize reproduction. Although previous fisheries models have examined sex change, a consensus does not exist regarding how sex change is predicted to affect stock dynamics. Some research has suggested that sex-changing stocks will be more sensitive to fishing and cannot be managed as if they were identical to separate-sex stocks (Bannerot et al., 1987; Punt et al., 1993; Huntsman and Schaaf, 1994; Coleman et al., 1996; Beets and Friedlander, 1999; Brule et al., 1999; Coleman et al., 1999; Arm- sworth, 2001; Fu et al., 2001). However, it has also been argued that, in the absence of sperm limitation, protogynous stocks should be less sensitive to size-selec- tive fishing because female biomass and thus population fecundity should not decrease as much as in a dioecious population, making traditional management and theory conservative when applied to these species. In general, protogynous stocks have been predicted to be at risk of population crashes because of their potential for nonlin- ear population dynamics in the presence of exploitation, yet there is no consensus regarding the importance of the exact pattern of sex change. For example, Arms- worth (2001) examined protogynous stock dynamics when the probability of sex change was a fixed func- tion of individual age and when the probability of sex change depended on the mean age of individuals in the population. He found that these two patterns of sex change had similar general dynamics and argued that management of a protogynous stock might not require knowledge of the precise pattern of sex change. In con- trast. Huntsman and Schaaf (1994) and Coleman et al (1999) have argued that a consideration of the pattern of sex change can be important to managing stocks. But, past theory has generally focused on comparing fixed patterns of sex change with fully compensating reproductive patterns that maintain a fixed sex ratio or ratio of female to male biomass. However, a variety of patterns of sex change exist and there is no reason to believe that all species have evolved to exhibit full compensation under natural conditions, let alone un- der new situations. Thus, it is important to consider how specific sex change rules will affect the dynamics and management of protogynous stocks and whether knowledge of the cues that determine sex change will be important. We (Alonzo and Mangel, 2004) developed a general modeling approach for examining the impact of repro- ductive behavior and life history pattern on stock dy- namics. Using this approach, we then compared the dynamics of a protogynous population with fixed size at sex change and an otherwise identical dioecious species (Alonzo and Mangel, 2004). These analyses showed that although dioecious and protogynous stocks clearly have distinct dynamics, simple statements arguing that one life history pattern is more or less sensitive to fishing cannot be made. Protogynous stocks with fixed patterns of sex change were predicted to experience sperm limi- tation and lowered larval recruitment at high fishing pressure, whereas the dioecious stock was predicted to show a large drop in mean population size even at low fishing mortality, but was not predicted to experience lowered fertilization rates due to size-selective fishing. Both stocks were predicted to be sensitive to fishing pattern, but a fixed pattern of sex change was predicted to put a population at risk of crashing if all male size classes were fished even at relatively low fishing mor- tality. Finally, classic spawning-per-recruit (SPR) mea- sures were not predicted to be good indicators of chang- es in the mean population size of protogynous stocks because they cannot indicate whether a population is experiencing sperm limitation and whether this limita- tion may lead to decreased population size or cause the stock to crash with small changes in fishing mortality. Although we found that whether or not a stock changes sex was important, that knowledge alone was not suf- ficient to understand and predict the response of the stock to fishing or management. We also found that sperm production and mating system were important variables affecting the probability that a population Alonzo and Mangel: Sex-change rules, stock dynamics, and the performance of spawning-per-recruit measures in protogynous stocks 231 would experience sperm limitation and would affect the performance of traditional spawning-per-recruit mea- sures. However, we did not consider the possibility that size at sex change may be plastic and depend on local social conditions or relative rather than absolute size. Plastic sex change may allow a protogynous species to compensate for any effect of size-selective fishing on the sex ratio of the population, rendering its dynamics iden- tical to the dynamics of a dioecious species. However, as described above, a wide variety of patterns of sex change have been observed in the wild and have been proposed to occur. Therefore, the exact pattern of sex change and cue driving phenotypic changes may lead to unique stock dynamics. In this study we apply the same general method we used previously (Alonzo and Mangel, 2004) to examine the effect of four different patterns of sex change (one fixed and three plastic) on the stock dynamics of a protogynous species. Methods We applied the same general method and individual- based population dynamic model as our previous study (Alonzo and Mangel, 2004). However, we now included the effect of four different patterns of sex change on the stock dynamics and performance of spawning-per-recruit measures in a protogynous species. Individuals vary in age, size, sex, and location (i.e. mating site). We assumed annual time periods and determined individual survival, size, and reproduction as described below. We simulated 100 years prior to examining the impact of fishing on stock dynamics and then simulated 100 more years in the presence of fishing with a constant mean fishing- induced mortality. This allowed the population to reach a stable age, sex, and size distribution prior to fishing which is independent of initial conditions. Because a number of elements of the model are stochastic, we examined 20 simulations for each scenario and set of parameter values, which was more than sufficient in all cases to lead to low variability in the key measures of interest. Fishing and adult survival We assume age and size do not affect natural adult mortality, i.iA and that adult mortality is density-inde- pendent. The fishery is size selective; if L represents fish size, F annual fishing mortality, Lf the size at which there is 50% chance an individual of that size will be taken, and r the steepness of the selectivity pattern, the fishing selectivity per size class s(L) is given by 11) SiL) = - — r l+eiq)y-r(L-Lf)\ and adult annual survival is a(L) = exp(-fiA - Fs(D). Population dynamics The number of larvae that enter the population is deter- mined by larval survival and the total production of fer- tilized eggs Pit), which is determined by total fecundity and fertility within each mating site as described below. Larval survival is assumed to have both density-inde- pendent and density-dependent components (e.g., Cowen et al., 2000; Sale, 2002), and we use a Beverton-Holt recruitment function (Quinn and Deriso, 1999; Jennings et al., 2001) to calculate larval survival . The number of larvae surviving to recruit in any year t, N0(t), is given by NQ(t) = (aPit))/(l + pP(t)) if(aP(t))/(l + l3P(t))+'£Nn{t)Nn where a gives density-independent survival, ft deter- mines the strength of the density-dependence in the larval phase, and Nmax represents the maximum popula- tion size. We assume that the population is open between mating sites, a single larval pool exists, larval recruit- ment is random among mating sites, and there is no emi- gration to or immigration from outside populations. Growth dynamics Larvae that survive to recruit begin at size L0 and growth is assumed to be deterministic and indepen- dent of sex or reproductive status. We calculate growth between age classes using a discrete time version of the von Bertalanffy growth equation (Beverton, 1987, 1992) where Llnf represents the asymptotic size and k is the growth rate. Then an individual of length Lit) at time t will grow in the next time period to size LU+1): LU + 1) = Linf (1 - exp(-&)) + LU)exp(-£). (4) Mating system (2) As in our previous model, we assume that reproduction occurs at the level of the mating group, and we examine the effect of varying mating group size and the number of mating sites. Juveniles and adults are assumed to exhibit site fidelity and larvae settle randomly among mating sites. The carrying capacity of the population is split equally among the mating sites and the total capacity of all mating sites exceeds the maximum popu- lation size in the absence of fishing as determined by 232 Fishery Bulletin 103(2) adult mortality and the recruitment function. As before (Alonzo and Mangel, 2004), we examine the following three cases: 1) the entire population mates at one site (1 mating site with up to 1000 individuals); 2) a few large mating groups exist (10 sites with a maximum of 100 individuals per site); and 3) many small mating aggregations exist (20 mating sites with a maximum of 50 individuals per site). We assume that within a mating site, individuals mate in proportion to their fertility and fecundity and that males that are large enough to change sex have a chance of reproducing that is proportional to their fertility and thus a large male reproductive advantage exists. This is equivalent to assuming that females exhibit a mate choice threshold (Janetos, 1980) that has evolved with the size at pat- tern of sex change and that male fertilization success is proportional to fertility. Reproduction We assume female fecundity E(L) and male sperm pro- duction SiL) can be represented by the allometric rela- tionships EiL)=aLh and SiL)=cLb respectively where a, b and c are constants. We assume that at any body length males produce 1000 times more sperm than females produce eggs. This leads to the realistic pattern that (in the absence of fishing) fertilization rates are high and that multiple males are needed to fertilize all the eggs produced by females. We calculate the average expected fertilization rate per mating site based on the total production of sperm and eggs at the site, where S represents the number of sperm released (in millions) and E the number of eggs released at each mating site. The proportion of eggs fertilized per mating site pF is given by PF = 1 + IkE + x)S (5) examples represent four plausible patterns that differ in the cues or mechanisms that induce sex change, the degree of compensation or plasticity assumed, and encompass the diversity that has been observed and hypothesized for a variety of sex-changing fish popula- tions (Helfman, 1997). Rule 1 : Fixed For the first sex-change rule, we assume that the probability of sex change pc(L) is determined by the absolute length of the individual and is pc(L)-- 1 l + exp(-p(L-Lc)) (6) where Lc represents the size at which 50% of mature females change sex and p is a constant that determines the steepness of the probability function. With this sex change rule, we also assume that the probability an individual matures p^L) is determined by absolute size. Once an individual matures, she remains female until sex change. LM represents the length at which 50% of juveniles are expected to mature. PM&) l + exp(-LM. Rule 2: Relative size For the second sex change rule, the mean size of all individuals in the mating group deter- mines the probability of sex change for an individual. First, we find the mean size of all individuals at each mating site. We let Lt represent the mean size in the mating site i. Then the probability of sex change for an individual of length L is where k and % are constants fitted to data. The pro- portion of eggs fertilized (pF) depends on both total sperm production (S) and egg production (E). If sperm production is very high in relation to egg production, fertilization rates will be at or near 100%. However, if total sperm production (S) decreases and egg production remains the same, fertilization rates will decrease. Simi- larly, as egg production (E) increases in relation to total sperm production (S) fertilization rates will decrease (see Fig. 2, Alonzo and Mangel, 2004). The number of eggs fertilized per group is pFE and the total production of fertilized eggs Pit) is the sum of the number of eggs fertilized in all mating groups. For more details on the fertilization function and individual sperm production see Alonzo and Mangel ( 2004). Patterns of sex change We examine four possible patterns of sex change, deter- mined by absolute or relative size of the individual. Although a variety of other possibilities exist, these Pc(L) = l + exp(-p(L-(L, +ALr)) (8) where ALC represents the difference from the mean at which the probability of sex change is 0.5. For these analyses, we also assumed that the probability an individual matures also depends on the mean size of individuals at the mating site. Then the probability of maturity is Pm Larval recruitment function parameter (see text) Fishing r 1(0.1) Steepness of selectivity curve h 30(25,35) Length at which 50% chance a fish will be removed F 0-3 Fishing mortality Reproduction a 7.04 Constant in the fecundity relationship (Warner, 1975) b 2.95 Exponent in the allometric relationship (Warner, 1975) c 10"3a Constant in the sperm production function (measured in millions of sperm) K 0.000003 Slope of fertilization function parameter X 0.09 Intercept of fertilization function parameter (based on Peterson et al., 2001) see text for details Rule 1 L. 30 cm Length at which 50% offish change sex P 1 Shape parameter in the sex-change function K 20 cm Length at which 50% of fish mature q 1 Shape parameter in the maturity function Rule 2 ALC 10 cm Difference from the mean size at which p(,(L)= 0.5 P 1 Shape parameter in the sex change function ALm 0 cm Difference from the mean size at which pM(L)= 0.5 Q 1 Shape parameter in the maturity function Rule 3 Fc 0.67 Frequency of smaller mature individuals wherepl.(L) = 0.5 P 50 Shape parameter in the sex-change function Fm 0.50 Frequency of smaller individuals at whichpw(L) = 0.5 Q 50 Shape parameter in the maturity function Rule 4 No additional parameters required dynamics rather than on exploring all possible param- eter combinations. However, it would certainly be use- ful in the future to examine the same question using parameter estimates based on other commercially ex- ploited species that change sex. Results We present the average across simulations of the mean population measures of the last 50 years for each simu- lation. The variation around the mean in all measures considered is hundredths of a percent of the mean or less. For the spawning-per-recruit (SPR) measures we give the mean value across the first 50 years of fishing to ensure that the entire cohort under consideration had died before the end of the simulation. Parameter values used are given in Table 1 General dynamics In all cases, size-selective fishing is predicted to decrease population size and decrease the mean length offish in the population. Although all scenarios are predicted to lead to the same change in average fish length, the effect of fishing on predicted population size and the mechanisms leading to changes in population size differ between the four sex-change rules (Figs. 1 and 2, Table 2). The largest differences occur between the fixed rule and the three plastic patterns of sex change. How- Alonzo and Mangel; Sex-change rules, stock dynamics, and the performance of spawnmg-per-recruit measures in protogynous stocks 235 80.000 -, A Rule 1: Fixed 60.000 - 40.000 - 20,000 - 0 0.5 1 15 2 2.5 3 80.000 -, B Rule 2: Relative size in 60.000 - ® 40,000 - tu |= 20,000 - ^ 0 0 0.5 1 1.5 2 2.5 3 o u 3 1. 80.000 -, C Rule 3: Relative frequency | 60.000 - < 40,000 - 20,000 - 0 0.5 1 1.5 2 2.5 3 80.000 n D Rule 4: Reproductive success 60,000 - 40,000 - 20.000 - U I 1 I 1 I 1 0 0.5 1 1.5 2 2.5 3 Fishing mortality (F) Figure 1 The predicted effect of fishing mortality on the production of fertilized eggs. Results are shown for the case where one mating group exists and the fishing selectivity is characterized by Lf— 30 and r=l. The same basic pattern is predicted for multiple mating sites as well. ever, the exact pattern of sex change has an important and qualitative effect on the predicted stock dynamics (Table 2). All three plastic patterns of sex change are predicted to show lower sperm limitation and higher fer- tilization rates in the presence of fishing than the fixed pattern of sex change (Table 2). However, associated with plastic sex change is also a greater predicted drop in egg production (total and fertilized) and mean popula- tion size than when the effect of size on the probability of sex change is fixed (Fig. 1, Table 2). This drop in egg production and mean population size occurs because female biomass is predicted to decrease as a result of the combination of fishing on larger individuals and smaller sizes at sex change (Fig. 2). The basic patterns are the same for the case with multiple mating sites. Most of the significant reductions in stock size are predicted at high fishing mortality. However, it is important to remember that we have assumed that the stock is very resilient (Table 1), and our focus is on the differences among sex-change rules and fishing patterns rather than on absolute fishing mortality. The effect of mating group size Although mating group size is predicted to have an effect in most cases on the stock dynamics of the population. 236 Fishery Bulletin 103(2) E o 20,000 1 A Rule 1: Fixed 16,000 ' \ \ 12.000 ; \ 8,000 4,000 ■ V^ 20,000 16.000 12,000 8,000 4.000 0 5 1 1.5 2.5 B Rule 2: Relative size \ 05 1.5 2.5 20,000 -X 16,000 ' 12,000 8,000 4,000 H C Rule 3: Relative frequency 0.5 1.5 2.5 20,000 16,000 12,000 ■ 8,000 • 4.000 0 l) Rule 4: Reproductive success 0 5 1 1.5 2 Fishing mortality (F) 25 Figure 2 The predicted effect of fishing mortality on the spawning stock biomass per male recruit (dashed lines) and per female recruit (solid lines) for all four patterns of sex change. Results are shown for the case of one mating group and the fishing selectivity is characterized by L^=30 and r=l. The same basic pattern is predicted for multiple mating sites as well. the strongest effect is predicted when size at sex change is fixed or determined by the frequency of small fish in the population (Fig. 3, A and C). When the size at sex change is fixed, populations are predicted to crash when mating sites are very small (Fig. 3A). In the case where size at sex change is determined by expected reproduc- tive success, group size is predicted to have no effect on the relative production of eggs and mean population size (Fig. 3D). However, for all the other rules of sex change considered, smaller mating sites are predicted to experi- ence sperm limitation in the presence of fishing, lead- ing to a decrease in the relative production of fertilized eggs and a decrease in mean population size (Fig. 3). However, unlike in the case of fixed size at sex change, the smaller mating groups (20 mating sites with up to 50 individuals per site) are stable both in the presence and absence of fishing and are not predicted to collapse for most fishing patterns. Sensitivity to fishing pattern Rule 1 The size-selective pattern of the fishery has a large effect on the predicted stock dynamics when the size at sex change is fixed. When the selectivity of the Alonzo and Mangel: Sex-change rules, stock dynamics, and the performance of spawning-per-recruit measures in protogynous stocks 237 Table 2 A comparison of stock dynamics for four sex-change rules. Results are reported for the situation where the fishing selectivity pattern and the probability of sex change are both centered at the same size (L^30). These results assume a near knife-edge selectivity (r=l) and that one mating site exists. Numbers given are for the predicted relative change as a result of fishing (when F=3 compared to F=0\. SSBR = spawning stock biomass per recruit. Rule 1: Fixed Rule 2: Relative size Rule 3: Relative frequency Rule 4: Reproductive success Mean population size 90% 90% 73% 72% Total SSBR 40% 45% 44% 39% Male SSBR 11% 22% 39% 397r Female SSBR 98% 92% 58% 39% Sex ratio 0.67 - 0.92 0.67^0.84 No change 0.8 - 0.66 Mean size 88% 88% 88% 88% Sperm production 11% 23% 40% 40% Egg production 98% 93% 59% 41% Fertilized egg production 88% 86% 59% 41% fishery is centered below the mean size at sex change (L^=25, r=l), the stock was predicted to crash at high fishing mortality (F>1, Fig. 4A). Furthermore, when the selectivity pattern was not steep (L,= 30, r=0.1), the population was always predicted to crash even at low fishing mortality (and thus this case is not shown in Figs. 4A-6A). When the steepness of the fishery's selec- tivity changes, the size range over which fish are targeted also changes. Thus, smaller and younger fish are removed by the fishery when r=0.1 and hence a greater number of age classes are affected by fishing. At an extreme, fishing mortality could be high enough that all of the individuals in any size classes targeted by the fishery are removed. As a result, although the steepness of the selectivity function only affects the spread of the function mathematically, it has the biological effect of decreasing the size at which fish experience fishing mortality and can have a large effect on the size and age distribution of the population. In contrast, when the fishery's selectiv- ity is steep (r=l) and only fish at or above the mean size at sex change (L^&30) are targeted, the effect of fishing on the population is predicted to be much less (Fig. 4A). Independent of the selectivity pattern, the population sex ratio is predicted to be more female-biased in the presence of fishing than in the absence of fishing. The lower the mean size removed by the fishery, the greater the predicted change in population sex ratio as a result of fishing (Fig. 5A). For situations in which the stock is not predicted to crash (i.e., L^30 and r=l), yield is pre- dicted to increase with diminishing returns with fishing mortality (Fig. 6A), catch is not predicted to decline with increased fishing mortality (at least up to F=3), and steep size-selective fishing patterns with lower size thresholds are predicted to lead to more yield (Fig. 6A). Rule 2 When sex change is determined by the mean size of individuals in the mating site and the size-selec- tivity is weak (r=0.1), the population is predicted to crash when F^l.67 (Fig. 4B). This crash occurs because individuals do not escape fishing mortality even at small sizes. However, unlike when sex change is fixed (Fig. 4A), the population is predicted not to crash when the size selected by the fishery is less than the mean size at sex change in the absence of fishing (L,=25, Fig. 4B). The larger the mean size selected by the fishery, the smaller the predicted effect of fishing on the mean population size and the population sex ratio (Figs. 4B and 5B). Although catch is predicted to increase with diminishing returns as fishing mortality increases from zero to three, the difference between the size-selectivity patterns is predicted to decrease and yield will be greater annually if larger fish are targeted (Fig. 6B). Rule 3 As above, when the probability of sex change depends on the relative frequency of smaller mature individuals, the population is predicted to crash when- ever size-selectivity is weak because fish do not escape fishing even when small (r=0.1, Fig. 4C). Although the population is predicted not to crash when the size tar- geted by the fishery is less than the mean size at sex change in the absence of fishing (L/=25, Fig. 4C), this fishing pattern is predicted to lead to a large decrease in mean population size and a marked decrease in popu- lation sex ratio (Figs. 4C and 5C). In contrast fishing selectivity that is centered at or above the mean size of sex change in the absence of fishing (L,— 30 and L^35) is predicted to lead to a weaker effect on mean popula- tion size and to almost no effect on the population sex ratio (Figs. 4C and 5C). However, in contrast to the two scenarios described above this pattern of sex change leads to the prediction that targeting fish at or larger than the normal mean size of sex change (Zy=30 and r=l) will lead to the greatest annual yield over time for most fishing mortalities (Fig. 6C). 238 Fishery Bulletin 103(2) A Rule 1: Fixed o> o a. 1.0 ft 0.8 15 Rule 2: Relative size Egg production Fert. egg production Mean population per recruit per recruit size 1.0 08 06 0.4 02- C Rule 3: Relative frequency D Rule 4: Reproductive success en £ 2 Egg production Fert. egg production Mean population per recruit per recruit size Figure 3 Effects of mating group size on the response of egg production per recruit, fertilized egg production per recruit, and mean population size to fishing pressure. Large (one large mating aggregation), medium 1 10 medium-sized mating aggrega- tions) and small (20 small mating aggregations) situations are compared. Percent change in the presence of fishing (from F =0 to F=l) is given. Total population fecundity and mean body size are lower for smaller mating aggregations as well. Results are shown for Z^— 30 and r=l. No bars are shown for small mating groups with fixed size at sex change because these populations are predicted to crash. Rule 4 As with all of the other patterns of sex change, populations with sex change based on expected reproduc- tive success are predicted to crash whenever small fish experience fishing mortality (r=0.1, Fig. 4D). Further- more, as with the other two plastic sex change rules, populations are predicted not to crash when fish below the normal mean size at sex change are included in the fishery because the population can compensate with smaller sizes at sex change in the presence of fishing (Fig. 4D). Although only small differences among fish- ing patterns are predicted in the mean population sex ratio, the effect on the population size is predicted to be greatest when many size classes are fished, and large differences are predicted between the fishing patterns in mean population size (Fig. 5D). Finally, in the scenario of sex change based on expected reproductive success. the fishing pattern predicted to lead to the greatest catch is to target only fish above the normal mean size at sex change (1^=35, Fig. 6D). In summary, fishing is always predicted to decrease total production of fertilized eggs and mean population size. However, the strength of the effect depends both on fishing selectivity and the pattern of sex change (see above and Figs. 4-6). Although populations with fixed patterns of sex change are predicted to crash in the pres- ence of fishing below the mean size at sex change, plastic patterns of sex change are predicted to lead to more resilience since these populations can compensate for the removal of large males more effectively. However, all scenarios are predicted to crash in the presence of fishing across a broad range of size classes (when r=0.1) even in completely compensatory patterns of sex change. Alonzo and Mangel: Sex-change rules, stock dynamics, and the performance of spawning-per-recruit measures in protogynous stocks 239 1 A Rule 1: Fixed L,=35 r=1 800 600 ' L,=30r=1 400 ' 200 ' \ L,=25 r=1 0.5 1.5 2.5 800 600 400 200 ' 0 B Rule 2: Relative size r L,=35 r=1 I --- ; ,- 1 L,=25 r=1 . L,=30 r=1 1 *i 1- i 0 0.5 1 15 C Rule 3: Relative frequency 2.5 S 800 L,=30r=1 0.5 1.5 2.5 D Rule 4: Reproductive success 2.5 Fishing mortality Lj=35 r=1 "77=30 r=1 L,=25 r=1 L,=35 r=1 Figure 4 The effect of size-selective fishing on the predicted mean population size for all four patterns of sex change. We present results for a sex- changing stock with one mating site. Means across 20 simulations are given. For details see text. The same basic patterns are predicted with multiple mating sites. A line is not shown in panel A (when sex change is fixed) where L^— 30 and r=0.1 because the population is predicted to crash at any fishing mortality in this scenario. Yet, the exact response depends greatly on the specific pattern of sex change. For example, the population sex ratio is not predicted to change much in the presence of fishing when sex change is based on expected reproduc- tive success and fishing pattern has little effect on the sex ratio (Fig. 5). However, when sex change is based on expected reproductive success, the annual yield is greater for fishing patterns with larger size thresholds (Fig. 6). In contrast, when sex change is determined by the mean size of individuals at the mating site, sex ratio is predicted to increase with fishing and increase more when smaller size classes are fished. However, for this pattern of sex change, the smallest size threshold is also predicted to lead to the largest yield of the fishery, although as fishing mortality increases the difference between fishing pat- terns with differing size thresholds decreases. Therefore, the fishing pattern that will produce optimal yield will depend on the exact pattern of sex change (Fig. 6). 240 Fishery Bulletin 103(2) 1 o i A Rule 1 : Fixed 0 L,=30r=1 1 0 08 0.6 0.4 ' 02 0 L» Huie •£ : Heia ive size L,=25 r=1 ^r=~- -z=~- :-t==-=""" \ /_,=35r=1 \ L,=30r=0.1 \ X 0 0.5 1 1.5 1 C Rule 3: Relative frequency 0.8- 0.6- . ..„■■.... 04 0.2 0 1.0 2.5 Lr=35 r=1 L,=25 r=1 \ L,=30r=0.1 0.5 1.5 2.5 0.5 1 1.5 2 Fishing mortality 2.5 L,=30r=1 L,=30r=1 0.8. L,=35r=1 0.6- 0.4. 0.2. 0 -""^ — \ \ L,=30 r=0.1 \ 1 1 1 > 1 r L,=25r=1 1 r L,=30r=1 Figure 5 The effect of size-selective fishing on the predicted population sex ratio for all four patterns of sex change. We present results for a sex-changing stock with one mating site. Means across 20 simulations are given. For details see text. The same basic patterns are predicted with multiple mating sites. A line is not shown in panel A (when sex change is fixed I where L,= 30 and r=0.1 because the population is predicted to crash at any fishing mortality in this scenario. Spawning-per-recruit (SPR) measures and a comparison of protogynous and dioecious stocks Our previous results (Alonzo and Mangel. 2004) have shown that whether species change sex or are dioecious is predicted to have dramatic effects on both the stock dynamics and performance of classic SPR measures. However, our results show that the exact pattern of sex change, and not just whether the pattern is plastic or fixed, can have a strong effect on these measures as well (Fig. 7). Because of the population dynamics of the model, all the scenarios represented in the present study show a great resiliency to fishing. Hence, the predicted changes in stock size are all above the common threshold of allowing a reduction of spawning per recruit mea- sures to 40% of their values in the unfished condition. However, our aim is not determine if this population is overfished. Instead, it is to determine whether classic Alonzo and Mangel: Sex-change rules, stock dynamics, and the performance of spawning-per-recruit measures in protogynous stocks 241 OJ A Rule 1: Fixed L,30r=1 L,=35r=1 L,=25 r=1 0 0 5 1 1.5 B Rule 2 Relative size 25 Z.,=25r=1 L,=30 r=1 L,=35 r=1 L,=30r=0.1 0 0.5 1 15 (_> Rule 3: Relative frequency — i — 2.5 L,=30r=1 L,=35r=1 L,=25 r=1 D Rule 4: Reproductive success 35 r=1 1.5 2 Fishing mortality Figure 6 The effect of size-selective fishing on the predicted annual yield for all four patterns of sex change. We present results for a sex-changing stock with one mating site. Means across 20 simulations are given. For details see text. The same basic patterns are predicted with multiple mating sites. A line is not shown in panel A (when sex change is fixed) where Zy=30 and r = 0.1 because the population is predicted to crash at any fishing mortality in this scenario. spawning per recruit measures based on egg produc- tion or fecundity could accurately assess the status of sex-changing stocks. Although the fixed pattern of sex change is predicted to show the greatest difference between egg production per recruit and fertilized eggs produced per recruit, each population shows deviations between egg production and the production of fertilized eggs. Thus egg production alone cannot tell us how the population is being affected by fishing and classic SPR measures based on population fecundity may be misleading for sex-changing stocks in cases where the sex-change rule is not completely compensatory (rules 1-3). It is also interesting to ask whether consistent differences exist (as has been suggested) in the resil- iency of sex-changing stocks, compared to stocks with separate sexes. Our results indicate that sex change based on expected reproductive success is predicted to have very similar dynamics to the dioecious population, 242 Fishery Bulletin 103(2) Rule 1; Fixed 0.9 Q. 0.7 o 06 05 0.4 650 700 750 Mean population size 900 950 Figure 7 Spawning-per-recruit (SPR) measures for all four patterns of sex change and an otherwise identical dioecious stock: mean egg production per recruit (filled) and mean fertilized eggs per recruit (open) are shown for a population with one large mating group when Zy=30 and r=l. The same basic patterns are predicted for multiple mating sites. Each line represents the same range of fishing mortalities, and each point represents fishing mortality increasing from 0 to 3 in increments of 1/3 moving from the right to the left. For the fourth rule (expected reproductive success), the two lines (eggs produced per recruit and eggs fertilized per recruit) overlap. whereas sex change based on relative size or the relative frequency of individuals in a mating site is predicted to have similar dynamics to those for the fixed pat- tern of sex change. Thus, it is not possible to say that sex-changing stocks tend to be more or less resilient to fishing than are dioecious populations. However, the sex change rule clearly affects the predicted relation- ship between fishing mortality and the response of the stock to fishing. Discussion We apply a general approach using individual-based simulation models to determine the predicted effect of the pattern of sex change on the stock dynamics of a protogynous species. Although the model structure and parameter values considered will not apply to all com- mercially important protogynous species, it is important to realize that all the scenarios considered are identical except for the pattern of sex change. As a result, any predicted differences that arise between these situa- tions are a result of the sex-change rule and indicate that knowing simply that a species exhibits sex change but not what the behavioral rule of sex change is will lead to an incomplete ability to understand and predict the dynamics of the stock and its response to fishing or management strategies. Independent of the sex-change rule, the protogy- nous stocks are always predicted to be sensitive to the size-selective fishing pattern. Mean population size is always predicted to decrease as fishing mortality increases, despite the fact that we have assumed that recruitment is strongly density dependent and that the species is very productive. Stocks are predicted to crash even at low fishing mortality when the size- selective fishing pattern targets all reproductive size classes and for the fixed sex change rule whenever all male sizes sizes are targeted by the fishery. It will be necessary but not sufficient to avoid overfishing at spawning aggregations. Our results indicate that it will also be important to allow smaller and nonreproductive individuals to escape fishing as well. These results indicate that independent of the exact pattern of sex change, management strategies for all protogynous stocks need to be sensitive to the size-selectivity of Alonzo and Mangel: Sex-change rules, stock dynamics, and the performance of spawning-per-recruit measures in protogynous stocks 243 the fishing pattern in relation to size at maturity and size at sex change observed in the population, and a failure to do so can lead to a sudden and unexpected collapse of the fishery — a collapse from which it may be difficult to recover. We assume in all cases that the same cues determine both the probability of maturity and the probability of sex change within a species. For example, when sex change was affected by the relative size of individuals at the mating site, we assume that this same cue af- fected the probability of maturing. This assumption has a large effect on the predicted dynamics of the stock. Alternatives exist. For example, the size at which fish mature could be determined by endogenous rather than exogenous factors even in a population where the prob- ability of sex change is affected by external cues. If this were the case, the population can easily be fished into a situation where it cannot compensate for size-selective fishing and is predicted to crash for any fishing pat- tern that targets reproductive individuals. For example, when L,= 30 and r=l, populations with plastic size at maturity and sex change were not predicted to crash independent of fishing mortality. In contrast, simula- tions where populations were assumed to have fixed size at maturity rules (Lm=20) but plastic patterns of sex change crashed at most fishing mortalities with L^=30 and r=l. Hence, knowledge of the cues determining both maturity and sex change will be important in predict- ing and understanding larval production and the effect of fishing on a population. It is possible to argue that a protogynous species with fixed patterns of sex change may have very different dynamics than dioecious stocks, but the compensatory patterns of sex change will be less sensitive to fishing and exhibit dynamics very similar to their dioecious counterparts. However, our results indicate that even stocks with plastic patterns of sex change are predicted to have dynamics distinctly different from otherwise identical dioecious populations. For example, sperm limitation is predicted to occur for all sex change rules, except for the pattern where sex change is determined by expected reproductive success (rule 4). However, even a stock exhibiting the reproductive sucess rule has dynamics that are distinctly different from those of a dioecious species because a change in the size dis- tribution of the population due to size-selective fishing is predicted to have a large effect on the productivity and sex ratio of the protogynous population. Similarly, mating group size is predicted to affect the stock dy- namics in all cases except for the reproductive success rule. Therefore, although knowing the pattern of sex change is predicted to be important in understanding stock dynamics, it is also clear that the pattern of sex change must be considered in the context of the mat- ing system of the stock, as well as in the context of the basic biology of the stock. Protogynous stocks are thus predicted to be sensitive to the fishing pattern, and nonlinear stock dynamics are possible when fishing operations target a wide range of fish sizes. However, each stock is also predicted to have a unique response to the same fishing pattern (Figs. 4-6) and to have different relationships between traditional spawning-per-recruit measures and changes in mean population size with fishing mortality (Figs. 1, 2, and 7). As a result, monitoring changes in spawn- ing stock biomass per recruit or egg production per recruit alone will not make it possible to determine the relationship between these measures and mean popula- tion size or to know whether the population is at risk for large and sudden declines in population size. Our results indicate that although it is important to know whether sex change occurs when managing a stock, it will also be important to know what endogenous or exogenous cues induce sex change and how behavioral patterns and life history strategies affect the demo- graphic rates of the stock. Plasticity is not predicted to yield populations that have stock dynamics that are identical to those of di- oecious species, and the performance of spawning-per- recruit measures and the relationship between egg pro- duction and population size differed greatly between all four patterns of sex change, despite the fact that the basic patterns of growth, survival, and fecundity where identical between all the scenarios considered. Because sperm limitation is more common with the fixed and relative size rules of sex change, these situations are predicted to have the greatest difference between clas- sic SPR measures and the production of fertilized eggs. Clearly it is not just whether a population changes sex or not, but also how sex change is induced, that deter- mines the population's predicted response to fishing and the performance of spawning-per-recruit measures in predicting and indicating the effect of fishing on the population. Although it is important to know what life history strategy and behavioral patterns are observed in a species, these alone will not always be sufficient to predict expected changes in population size and pro- ductivity under new conditions. Instead, knowledge of the plasticity of behavioral and life history patterns, as well as information about the internal and exter- nal cues that induce phenotypic changes, may also be necessary. Phenotypic plasticity is often expressed as a threshold response (such as sex change) to a continuous endogenous or exogenous cue. Therefore, as predicted by our model, plasticity can generate nonlinear changes in important demographic characters. An understanding of the natural variation in behavior and life history combined with knowledge of fish vital rates and envi- ronmental conditions will lead to a better understand- ing of and ability to predict the response of a stock to fishing mortality, environmental changes, and specific management strategies. Acknowledgments This research was supported by National Science Foun- dation grant IBN-0110506 to Suzanne Alonzo and the Center for Stock Assessment Research (CSTAR). 244 Fishery Bulletin 103(2) Literature cited Alonzo, S. H., and M. Mangel. 2004. Size-selective fisheries, sperm limitation and spawning per recruit in sex changing fish. Fish. Bull. 102:1-13. Apostolaki, P., E. J. Milner-Gulland, M. K. McAllister, and G. P. Kirkwood. 2002. Modelling the effects of establishing a marine reserve for mobile fish species. Can. J. Fish. Aquat. Sci. 59:405-415. Armsworth, P. R. 2001. Effects of fishing on a protogynous hermaphro- dite. Can. J. Fish. Aquat. Sci. 58:568-578. Bannerot, S., W. F. Fox, and J. E. Powers. 1987. Reproductive strategies and the management of snappers and groupers in the gulf of Mexico and Caribbean. In Tropical snappers and groupers: biology and fisheries management, t J. J. Polovina and S. Ralston, eds.), p. 561-603. Westview Press, Boulder, CO. Beets, J., and A. Friedlander. 1999. Evaluation of a conservation strategy: a spawn- ing aggregation closure for red hind, Epinephelus gut- tatus, in the U.S. Virgin Islands. Environ. Biol. Fish. 55:91-98. Beverton, R. J. H. 1987. Longevity in fish: some ecological and evolution- ary considerations. In Evolution of longevity in ani- mals. (A. D. Woodhead and K.H. Thompson, eds.) p. 161-185. Plenum Press, New York, NY. Beverton, R. J. H. 1992. Patterns of reproductive strategy parameters in some marine teleost fishes. J. Fish Biol. 41 (Suppl. B) 41:137-160. Breden, F., D. Novinger, and A. Schubert. 1995. The effect of experience on mate choice in the Trinidad guppy, Poecilia reticulata. Environ. Biol. Fish. 42:323-328. Brule, T. C. Deniel, T. Colas-Marrufo, and M. Sanchez-Crespo. 1999. Red grouper reproduction in the southern Gulf of Mexico. Trans. Am. Fish. Soc. 128:385-402. Charnov, E. L. 1982. The theory of sex allocation, 355 p. Princeton Univ. Press, Princeton, NJ. 1986. Size advantage may not always favor sex change. J. Theor. Biol. 119:283-285. Charnov, E. L., and J. J. Bull. 1989. Non-Fisherian sex ratios with sex change and envi- ronmental sex determination. Nature 338:148-150. Coleman. F. C, A. Eklund. and C. B. Grimes. 1999. Management and conservation of temperate reef fishes in the grouper-snapper complex of the southeast- ern United States. Am. Fish. Soc. Sym. 23:233-242. Coleman, F. C, C. C. Koenig, and L. A. Collins. 1996. Reproductive styles of shallow-water groupers (Pisces: Serranidae) in the eastern Gulf of Mexico and the consequences of fishing spawning aggregations. En- viron. Biol. Fish. 47:129-141. Collins, L. A., A. G. Johnson, and C. P. Keim. 1996. Spawning and annual fecundity of the red snap- per iLutjanus campechanus) from the northeastern Gulf of Mexico. In Biology, fisheries, and culture of tropical groupers and snappers (F. Arreguin-San- chez, J. L. Munro, M. C. Balgos, and D. Pauly, eds.) p. 174-188. ICLARM (International Center for Living Aquatic Resources Management! conference proceedings 48, ICLARM, Makati City, Philippines. Constable, A. J., W. K. de la Mare, D. J. Agnew, I. Everson, and D. Miller. 2000. Managing fisheries to conserve the Antarctic marine ecosystem: practical implementation of the con- vention on the conservation of Antarctic marine living resources (CCAMLR). ICES J. Mar. Sci. 57:778-791. Cowen, R. K. 1985. Large-scale pattern of recruitment by the labrid, Semicossyphus pulcher: causes and implications. J. Mar. Res. 43:719-742. 1990. Sex change and life history patterns of the labrid, Semicossyphus pulcher, across an environmental gradient. Copeia 1990:787-795. Cowen, R. K., K. M. M. Lwiza, S. Sponaugle, C. B. Paris, and D. B. Olson. 2000. Connectivity of marine populations: open or closed? Science 287:857-859. Fu, C, T. J. Quinn, II. and T. C. Shirley. 2001. The role of sex change, growth and mortality in Panda/us population dynamics and management. ICES J. Mar. Sci. 58:607-6*21. Fulton, E., D. Kault, B. Mapstone, and M. Sheaves. 1999. Spawning season influences on commercial catch rates: computer simulations and Plectropomus leopardus. a case in point. Can. J. Fish. Aquat. Sci. 56:1096-1108. Greene. C. J. LTmbanhowar, M. Mangel, and T. Caro. 1998. Animal breeding systems, hunter selectivity, and consumptive use in wildlife conservation. In Behav- ioral ecology and conservation biology (T. Caro, ed.) p. 271-305. Oxford Univ. Press, New York, NY. Gunderson, D. R. 1997. Trade-off between reproductive effort and adult survival in oviparous and viviparous fishes. Can. J. Fish. Aquat. Sci. 54:990-998. Haddon, M. 2001. Modelling and quantitative methods in fisheries, 406 p. Chapman and Hall, Boca Raton, FL. Helfman, G. S. 1997. The diversity of fishes, 528 p. Blackwell Science, Maiden, MA. Huntsman, G. R., and W. E. Schaaf. 1994. Simulation of the impact of fishing on reproduc- tion of a protogynous grouper, the graysby. No. Am. J. Fish. Manag. 14:41-52. Iwasa, Y. 1990. Sex change evolution and cost of reproduction. Be- hav. Ecol. 2:56-68. Janetos, A. C. 1980. Strategies of female mate choice: a theoretical analysis. Behav. Ecol. Sociobiol. 7:107-112. Jennings, S., M. J. Kaiser, and J. D. Reynolds. 2001. Marine fisheries ecology, 417 p. Blackwell Sci- ence, Oxford, UK. Koeller, P., R. Mohn, and M. Etter. 2000. Density dependant sex change in northern shrimp, Pandalus borealis, on the Scotian Shelf. J. No. Atl. Fish. Sci. 27:107-118. Kruuk, L. E. B., T. H. Clutton-Brock, S. D. Albon, J. M. Pember- ton, and F. E. Guinness. 1999. Population density affects sex ratio variation in red deer. Nature 399:459-461. Alonzo and Mangel: Sex-change rules, stock dynamics, and the performance of spawnmg-per-recruit measures in protogynous stocks 245 Kuwamura, T., and Y. Nakashima. 1998. New aspects of sex change among reef fishes: Recent studies in Japan. Environ. Biol. Fish. 52:125-135. Levin, P. F. and C. B. Grimes. 2002. Reef fish ecology and grouper conservation and management. In Coral reef fishes: dynamics and diversity in a complex ecosystem (P. F. Sale, ed.l p. 377-390. Academic Press, Amsterdam, The Netherlands. Lutnesky, M. M. F. 1994. Density-dependent protogynous sex change in ter- ritorial-haremic fishes: models and evidence. Behav. Ecol. 5:375-383. 1996. Size-dependent rate of protogynous sex change in the pomacanthid angelfish, Centropyge potteri. Copeia 1996:209-212. Metcalfe, N. B., F. A. Huntingford, W. D. Graham, and J. E. Thorpe. 1989. Early social status and the development of life- history strategies in Atlantic salmon. Proc. Roy. Soc. Lond.B 236:7-19. Mittelbach, G. G., C. W. Osenberg, and P. C. Wainwright. 1992. Variation in resource abundance affects diet and feeding morphology in the pumpkinseed sunfish iLepmis gibbosus). Oecologia 90:8-13. Munoz, R. C, and R. R. Warner. 2003. A new version of the size-advantage hypothesis for sex change: Incorporating sperm competition and size-fecundity skew. Am. Naturalist 161:749-761. Nakashima, Y., Y. Sakai, K. Karino, and T. Kuwamura. 2000. Female-female spawning and sex change in a haremic coral-reef fish, Labroides dimidiatus. Zool. Sci. 17:967-970. Nishibori, M., and M. Kawata. 1993. The effect of visual density on the fecundity of the guppy, Poecilia reticulata. Environ. Biol. Fish. 37:213-217. Petersen, C. W.. R. R. Warner, D. Y. Shapiro, and A. Marconato. 2001. Components of fertilization success in the blue- head wrasse. Thalassemia bifaseiatum. Behav. Ecol. 12:237-245. Punt, A. E., P. A. Garratt, and A. Govender. 1993. On an approach for applying per-recruit mea- sures to a protogynous hermaphrodite, with an illus- tration for the slinger Chrysoblephus puniceus (Pisces: Sparidae). S. Afr. J. Mar. Sci. 13:109-119. Quinn, T. J., and R. B. Deriso. 1999. Quantitative fish dynamics, 542 p. Oxford Univ. Press, New York, NY. Ridgeway, M. S., and B. J. Shuter. 1994. The effects of supplemental food on reproduction in parental male smallmouth bass. Environ. Biol. Fish. 39:201-207. Sale, P. F. 2002. The science we need to develop more effective man- agement. In Coral reef fishes: dynamics and diversity in a complex ecosystem (P. F. Sale, ed.l, p. 361-376. Aca- demic Press, Amsterdam, The Netherlands. Schultz, E. T., and R. R. Warner. 1991. Phenotypic plasticity in life-history traits of female Thalassoma bifaseiatum (Pisces: Labridael: 2. Corre- lation of fecundity and growth rate in comparative studies. Environ. Biol. Fish. 30:333-344. Shapiro, D. Y. 1987. Reproduction in groupers. In Tropical snappers and groupers: biology and fisheries management (J. J. Polvina and S. Ralston, eds.l, p. 295-328. Westview Press, Boulder, CO. Snyder, R. J., and H. Dingle. 1990. Effects of freshwater and marine overwintering environments on life histories of threespine stickle- backs: evidence for adaptive variation between anadro- mous and resident freshwater populations. Oecologia 84:386-390. Sutherland, W. J. 1990. Evolution and fisheries. Nature 344:814-815. 1998. The importance of behavioural studies in conser- vation biology. Anim. Behav. 56:801-809. Wainwright, P. C, C. W. Osenberg, and G. G. Mittelbach. 1991. Trophic polymorphism in the pumpkinseed sunfish {Lepomis gibbosus Linnaeus): effects of environment on ontogeny. Funct. Ecol. 5:40-55. Warner, R. R. 1975. The reproductive biology of the protogynous hermaphrodite Pimelometopon pulchrum (Pisces: Labridae). Fish. Bull. 73:262-283. Warner, R. R., and P. Lejeune. 1985. Sex change limited by paternal care: A test using four Mediterranean labrid fishes, genus Symphodus. Mar. Biol. 87:89-100. Warner, R. R., and S. E. Swearer. 1991. Social control of sex change in the bluehead wrasse, Thalassoma bifaseiatum (Pisces: Labridae). Biol. Bull. 181:199-204. 246 Abstract— The growth rate of Steller sea lion (Eumetopiasjubatus) pups was studied in southeast Alaska, the Gulf of Alaska, and the Aleutian Islands during the first six weeks after birth. The Steller sea lion population is cur- rently stable in southeast Alaska but is declining in the Aleutian Islands and parts of the Gulf of Alaska. Male pups (22.6 kg [±2.21 SD]) were sig- nificantly heavier than female pups (19.6 kg [±1.80 SD]) at 1-5 days of age, but there were no significant dif- ferences among rookeries. Male and female pups grew (in mass, standard length, and axillary girth) at the same rate. Body mass and standard length increased at a faster rate for pups in the Aleutian Islands and the western Gulf of Alaska (0.45-0.48 kg/day and 0.47-0.53 cm/day, respectively) than in southeast Alaska (0.23 kg/day and 0.20 cm/day). Additionally, axillary girth increased at a faster rate for pups in the Aleutian Islands (0.59 cm/ day) than for pups in southeast Alaska v(0.25 cm/day). Our results indicate a greater maternal investment in male pups during gestation, but not during early lactation. Although differences in pup growth rate occurred among rookeries, there was no evidence that female sea lions and their pups were nutritionally stressed in the area of population decline. Neonatal growth of Steller sea lion (Eumetopias jubatus) pups in Alaska Elisif A. A. Brandon Department of Marine Biology Texas A&M University at Galveston 5007 Avenue U Galveston, Texas 77551 Present address: 97A Lowell Ave Newton, Massachusetts 02460 Donald G. Calkins Alaska SeaLife Center P.O. Box 1329 Seward. Alaska 99664 Thomas R. Loughlin National Marine Mammal Laboratory Alaska Fisheries Science Center, NMFS 7600 Sand Point Way, NE Seattle, Washington 98115 Randall W. Davis Department of Marine Biology Texas A&M University at Galveston 5007 Avenue U Galveston, Texas 77551 E-mail address (for R. W Davis, contact author): davisngitamug edu Manuscript submitted 26 April 2004 to the Scientific Editor's Office. Manuscript approved for publication 2 December 2004 by the Scientific Editor. Fish. Bull. 103:246-257 (2005). Sea lion (order Carnivora, family Otariidae) pups depend entirely on milk for neonatal growth (Bonner, 1984). Studies of sea lions and fur seals have shown that if a pup does not obtain enough milk from its mother, it will exhibit poor body condi- tion (i.e., reduced lean mass and total lipid mass for a given age or standard length) and a reduced growth rate (Trillmich and Limberger. 1985; Ono et al., 1987). Poor body condition and reduced growth rate, in turn, may have lifelong consequences because neonatal growth is an important factor in determining adult size and survival (Bryden, 1968; Innes et al., 1981; Calambokidis and Gentry, 1985; Albon et al., 1992; Baker and Fowler, 1992; Gaillard et al., 1997; Boltnev et al., 1998; Tveraa et al„ 1998; Burns, 1999). Because of their large size, aggressive behavior, sensitivity to disturbance, and the remote location of their rookeries, less is known about the early growth of Steller sea lions (SSL) than of most other pinniped (seals, sea lions, and walrus) species. Higgins et al. (1988) measured body mass of SSL pups on Ano Nuevo Island in California but only reweighed five pups to measure growth rates. Mer- rick et al. (1995) weighed SSL pups at a number of locations throughout the Gulf of Alaska and the Aleutian Islands but did not reweigh them to assess individual growth rates. Genetic studies show that there are distinct eastern and western popula- tions of SSL (Bickham et al., 1996, 1998) (Fig. 1). The eastern population comprises animals in California, Ore- gon, British Columbia, and southeast Alaska. The western population com- prises animals in the Gulf of Alaska, the Aleutian Islands, the Bering Sea, the Commander Islands, Kamchatka, and the Kuril Islands. A severe popu- Brandon et al.: Neonatal growth of Eumetopias /ubatus 247 ~65°N - 60° -55° 180° 1 Bering Sea Sequam Island Yunaska >j^ Kl, Hid ■to.*.* \ ^teutian Islands Chinkof Island Lowrie - Island 250 miles 170° / 160° I 150°W I 250 kilometers 1 Figure 1 Study sites for Steller sea lions iEumetopias jubatus) in Alaska. The Lowrie Island rookery in south- east Alaska has a stable population but rookeries at Fish, Marmot and Chirikof Islands in the Gulf of Alaska and Yunaska and Seguam Islands in the Aleutian Islands are areas where the population of Steller sea lions has declined. lation decline (>80'7f) occurred in the western popula- tion between the 1970s and the 1990s. In 1997, these population changes led to the reclassification of the western population from "threatened" to "endangered" and a classification of the eastern population as "threat- ened" under the Endangered Species Act (U.S. Federal Register 62:24345-24355). One hypothesis for the decline in population of SSLs is a decrease in food availability or quality in the Gulf of Alaska and the Aleutian Islands (Pascual and Adki- son, 1994; York, 1994; Calkins et al., 1999; NMFS1-2). If females are unsuccessful in obtaining sufficient food, pups will develop more slowly or die because of a de- crease in milk supply. To examine the potential effects NMFS (National Marine Fisheries Service). 1992. Re- covery plan for the Steller sea lion tEumetopias jubatus), 92 p. Prepared by the Steller Sea Lion Recovery Team for the National Marine Fisheries Service, Silver Spring, MD. [Available from the National Marine Mammal Laboratory, 7600 Sandpoint Way. NE, Seattle, Washington 98115.) ; NMFS (National Marine Fisheries Service). 1995. Sta- tus review of the United States Steller sea lion (Eumeto- pias jubatus) population. 61 p. Prepared by the National Marine Mammal Laboratory, Alaska Fisheries Science Center. [Available from the National Marine Mammal Labo- ratory, 7600 Sandpoint Way, NE, Seattle, Washington 98115.] of food availability on pup development, we measured growth rates of male and female pups from stable and declining populations of SSL in Alaska from 1990 to 1997. Our null hypothesis was that there was no differ- ence in pup growth rates among rookeries in southeast Alaska, the Gulf of Alaska, and the Aleutian Islands. The alternative hypothesis was that pups grew at a faster rate in southeast Alaska, the area of stable popu- lation. However, our results showed that pups grew faster in the area of declining population during the first six weeks after birth. In addition, females invested more energy in male pups at all locations during gesta- tion, but not during early lactation. Materials and methods Animals and study sites From 1990 to 1997, SSL pups were studied at loca- tions in southeast Alaska, the Gulf of Alaska, and the Aleutian Islands (Fig. 1 and Table 1). At Lowrie Island <54°51'N, 133°32'W) in southeast Alaska, measure- ments were made in 1993, 1994, and 1997. The rookery at Lowrie Island is in the area of the stable population (Calkins et al., 1999). In the Gulf of Alaska, measure- 248 Fishery Bulletin 103(2) Table 1 Locations dates, and the number of Steller sea lior lEumetopias jubatus) pups captured (/i). Location Dates n Stable population Lowrie Island (1993) 26 May-5 June 15-19 June 3 July 25 5 1 Lowrie Island (1994) 15-22 June 24-30 June 13-14 July 28 9 3 Lowrie Island (1997) 5-12 June 16-29 June 25 11 Declining population Fish Island (1995) 9-10 June 24-26 June 13-14 July 20 13 12 Marmot Island (1990) 27 June 8 Marmot Island (1991) 30 June 11 Marmot Island (1994) 27 June 15 July 21' 11-' Chirikof Island (1993) 11-17 June 27-28 June 7 July 18 July 20 14 11 4 Yunaska and Seguam Islands (1997) 8-16 June 22-24 June 4 July 16 12 5 ; Nine known-age pups. 2 Six known-age pups. merits were made in 1990, 1991, and 1994 on Marmot Island (58°12'N, 151°50'W), in 1993 on Chirikof Island (55°10'N, 155°8'W) and in 1995 on Fish Island (59°53'N, 147°20'W). On the Aleutian Islands of Seguam (52°30'N, 172°30'W) and Yunaska (52°45'N, 170°45"W), pups were studied in 1997. Data from Seguam and Yunaska Islands were combined because the islands are geographically close and can be considered part of one rookery complex. Rookeries in the Gulf of Alaska and the Aleutian Islands are in the area of declining population, although the rookery on Fish Island has not shown as precipitous a decline. Samples could not be obtained from all rooker- ies in all years because of logistical constraints and the need to minimize disturbance to rookeries. However, concurrent data were obtained from the declining and stable populations in 1993, 1994, and 1997. Only pups that had an attached umbilical cord or an unhealed umbilicus were selected for study. The fresh- ness of the umbilical cord was used as a rough estimate of age between 1 and 5 days (Davis and Brandon3). 3 Davis, R. W, and A. A. Brandon. Unpubl. data. I Data are on file at Texas A&M University, 5007 Avenue U, Galveston, Texas 77551.1 Choosing only pups with fresh umbilical cords mini- mized the age bias (Trites, 1993) that occurs when pups are captured at different times and rookeries (Table 1). Although pups were not selected by sex, sex was noted and used as a factor in analyses. Body mass (BM), standard length (SL), axillary girth (AG) (Am. Soc. Mammalogists, 1967) and body composition were measured for each pup. BM was measured to the near- est kilogram with a mechanical spring scale (Chatillon 160, Ametek, FL) on Marmot Island in 1990 and 1991 and on Lowrie Island in 1993. Body mass of pups at all other sites and years was measured to the nearest tenth of a kilogram by using an electronic scale (Rice Lake Weighing Systems, Rice Lake, WI; Ohaus I-20W, Ohaus, Pine Brook, NJ). Standard length was measured as a straight line from tip-of-nose to tip-of-tail, ventral sur- face down. Pups were restrained by hand and marked for later identification with hair bleach (Lady Clairol Maxi Blond, Clairol, Inc.) and with flipper tags attached in the axillary area of the fore-flippers. Body composition was measured by using the labeled water method (Nagy 1975; Nagy and Costa, 1980; Cos- ta, 1987; Bowen and Iverson, 1998). In this study, water labeled with a stable isotope of hydrogen (deuterium) Brandon et al.: Neonatal growth ot Eumetopias /ubatus 249 was used to estimate total body water (TBW in kg and %TBW as a percentage of BM). Background concentra- tion of deuterium was determined from blood samples taken from pups that were subsequently injected intra- muscularly with 10 mL deuterium oxide (D._,0) (99% en- riched, Cambridge Isotope Laboratories, Andover, MA). After a two-hour equilibration period (Costa, 1987), blood samples were taken to determine the dilution of injected deuterium in total body water. Pups were recaptured at approximately two-week intervals over periods ranging in length from 18 to 38 days (average measurement period was 29.6 days) (Table 1) and were weighed, measured, and a blood sample was taken from each pup. Similar protocols were used at all rookeries, except Marmot Island in 1990 and 1991, when only BM and SL were measured, and the age of pups was not estimated. Therefore, no growth rates were obtained from these data. Labeled water sample analysis Blood samples were centrifuged in the field in serum separator tubes, and the serum was transferred to cryo- vials that were frozen at -20°C until analysis. Isotope- ratio mass spectrometry was used to determine the ratio of deuterium (2H) to hydrogen (H) (Laboratory of Biochemical and Environmental Studies at University of California, Los Angeles, CA). The hydrogen-isotope dilution space was calculated from this ratio by using Equation 3 in Schoeller et al. (1980). However, the hydro- gen-isotope dilution space has been shown to underesti- mate TBW in a number of pinniped species (Reilly and Fedak, 1990; Arnould et al., 1996b), leading Bowen and Iverson (1998) to develop a single predictive equation to estimate '/'< TBW from hydrogen-isotope dilution space in pinnipeds for which data on the accuracy of the hydro- gen-isotope method are lacking. The equation 9cTBW = 0.003 + 0.968 H-dilution space (1) was used in the present study to correct the overesti- mated %TBW by 3.3% (Bowen and Iverson, 1998, Eq. 5). Percent total body lipid (%TBL, as a percentage of BM) was calculated by using predictive equations derived from the relationship between %TBW and 7cTBL for Antarctic fur seals (Arnould et al., 1996b): %TBL = 66.562 - 0.845 %TBW. (2) 9cTBL was then compared between male and female pups and among rookeries. Statistical analyses Statistics were performed by using Systat (version 11, SPSS, Inc, Chicago, ID, and by first treating each study site and year as a separate "location," then combining data for multiple years at a location (e.g., Marmot Island and Lowrie Island) when no significant interannual differences were found. Significance was determined at PsO.05. Data were examined for heteroscedasticity (unequal variances) before analysis (Zar, 1984). A\\ post hoc pairwise comparisons were made with the Tukey multiple comparison test. Data from the first capture (1-5 days of age) were analyzed for comparison by loca- tion and sex by using two-way ANOVA. Pup growth rate was estimated by performing a linear regression for each pup and extrapolating to t = 0 to estimate birth mass. Differences among means of pup growth rate and birth mass were then analyzed by using two-way ANOVA to determine differences by location and sex. Results Neonatal size There were no significant differences by rookery in pup mass at 1-5 days of age (Table 2) and no signifi- cant interaction between rookery and sex. The only significant difference in SL of 1-5 day old pups was that both genders were significantly longer on Seguam and Yunaska Islands than on Fish Island (P=0.0395). Pups on Chirikof Island had significantly smaller AG than pups on Lowrie, Fish, and Seguam and Yunaska Islands (P<0.02). Male and female pups were significantly differ- ent for all three morphometric measurements. Overall, male pups averaged 22.6 kg (±2.21 SD, ?i=71) and female pups averaged 19.6 kg (±1.80 SD, ;?=74) at first capture ( 1-5 days of age). There was no significant difference by rookery or sex and no significant interaction between rookery and sex in %TBW or %TBL of pups at first capture. When all pups at all rookeries were combined ( « =116), %TBW was 72.1% of BM (±3.17 SD) and %TBL was 5.6% of BM (±2.68 SD). Male pups had a significantly greater absolute TBW than female pups (P<0.0001), as would be expected because of the difference in BM at birth. There was a significant correlation between TBW and BM (Pearson r=0.945, P<0.001, ra=116; TBW (kg) = 0.6895 xBM + 0.6618). Neonatal growth Growth rates were treated as linear over the period monitored; there were not enough data to determine if growth was nonlinear. Male and female pups on the same rookery grew at the same rate (in BM, SL, and AG) during the first six weeks after birth (Fig. 2). When compared by rookery, BM increased at a faster rate for pups on Chirikof Island (P=0.0005) and on Seguam and Yunaska Islands (P= 0.0002) than on Lowrie Island (Fig. 3 and Table 3). The increase in BM for pups on Fish Island did not differ significantly from that at other rookeries. Marmot Island pups grew significantly more slowly than pups on Seguam and Yunaska Islands (P= 0.0382) but did not differ significantly from growth of pups at other rookeries. Standard length increased at a faster rate for pups on Chirikof Island (P=0.0068) and Seguam and Yu- 250 Fishery Bulletin 103(2) naska Islands (P= 0.0050) than it did for pups on Low- rie Island (Table 3). Growth in SL was also faster on Chirikof (P=0. 0383) and Seguam and Yunaska Islands (P=0.0230) than on Fish Island, whereas the increase in SL on Marmot Island did not differ significantly from the other rookeries. The increase in AG was sig- nificantly greater on Seguam and Yunaska Islands (P=0.0021) and Marmot Island (P=0.0364) than on Lowrie Island. There was no significant interaction between rookery and sex in the growth rate of BM, SL, and AG. Body mass at birth extrapolated to t = 0 from growth rates did not differ by rookery. There was no significant interaction between rookery and sex, but extrapolated birth mass did differ by sex (P<0.0001). Male pups at all rookeries averaged 22.4 kg (±2.36 SD, rc = 39), whereas female pups averaged 18.7 kg (±2.08 SD, n = 35). These extrapolated birth masses were similar to the average BM measured on the rookery for male (22.6 kg) and female (19.6) pups 1-5 days old. There was no correla- tion between extrapolated birth mass and growth rate (Pearson r=-0.09, P=0.45). Table 2 Body mass (BM), standard length (SL), and ixillary girth (AG) of neonatal li- 5 day old) Steller sea lion (Eumetopias jubatus) pups in the stable (Lowrie Island I a rid declini ng (Fish s.. Marmot I s., Chirikof Is., Seguam Is. Yunaska Is.) populations (mean ±SD). An asterisk (*) ndicates sign ificant differences Tom all other sites, and t indicates a significant difference between two sites. Standard length from Fish Is was sign ificantly different from SL on Seg jam and Yunaska Is. Ax llary girth on Chirikof Is. was significantly different from AG at all other sites In all cases. males were significantly larger than females. There were no significant interannua 1 differences; therefore data from all years at Lowrie Is. were combined. Location n BM kg) SL(cm) AG (cm) male female male female male female Lowrie Is. (1993-97) 39M 22.1 19.5 98.3 94.1 64.9 64.3 41F ±2.20 ±1.67 ±4.56 ±3.96 ±3.33 ±5.01 Fish Is. (1995) 11M 22.6 19.2 96. 2t 93. 3t 68.5 64.0 9F ±1.69 ±2.39 ±26.76 ±6.39 ±2.96 ±4.00 Marmot Is. (1994) 3M 21.7 20.2 101.7 97.4 65.5 61.8 6F ±1.80 ±2.42 ±1.53 ±2.67 ±2.78 ±5.38 Chirikof Is. (1993) 11M 23.21 19.02 99.1 94.9 62.7* 60.1* 9F ±2.59 ±1.05 ±5.24 ±2.40 ±3.52 ±2.15 Aleutian Is. ( Seguam and Yunaska Is. ) ( 1997 ) 7M 24.2 20.5 101.4+ 96.3t 67.7 63.9 9F ±1.97 ±1.88 ±4.29 ±2.55 ±3.50 ±3.66 Table 3 Steller sea lion (Eumetopias jubatus) pup growth from 0 to 40 days of age (mean between male and female pups. BM=body mass; SL = standard length; AG = axillar> no significant differences within an underlined grouping (e.g., for body mass grow and A was significantly different from M and L). ±SD). girth. :h rate There were no significant differences Underlining indicates that there were , C was significantly different from L, Location n BM growth rate (kg/day) SL growth rate (cm/day) AG growth rate (cm/day) Lowrie Is. (L) 26 Fish Is. (F) 13 Marmot Is. (M) 6 Chirikof Is. (C) 17 Aleutian Is. (A) (Seguam and Yunaska Is.) 12 0.23 ±0.176 0.35 ±0.171 0.28 ±0.141 0.45 ±0.126 0.48 ±0.168 0.20 ±0.322 0.22 ±0.183 0.22 ±0.287 0.47 ±0.171 0.53 ±0.163 0.25 ±0.244 0.41 ±0.235 0.59 ±0.510 0.47 ±0.187 0.59 ±0.257 ANOVA results LMFCA LFMCA LFCMA Brandon et al.: Neonatal growth of Eumetopias /ubatus 251 A D male ■ female 10 15 20 25 30 35 40 E >, 13 c □ male ■ female I ' ' ' ' I ' ' ' ' I i | i i i i | i i i i | i i i i | 5 10 15 20 25 30 35 40 B 1 ' i ' > ■ ' i ■ > > > i > > ' ■ i ' ' > > i ' > ■ > i > > ' > i > > ' ' i 0 5 10 15 20 25 30 35 40 Age (days) Figure 2 Change in body mass of individual Steller sea lion (Eumetopias jubatus) pups captured on I A) Lowrie Island in 1993, 1994, and 1997, (B) Fish Island in 1995, (C) Marmot Island in 1994, (D) Chirikof Island in 1993, and (E) Yunaska and Seguam Islands in 1997. 252 Fishery Bulletin 103(2) Discussion Compared to other species of sea lions and fur seals, SSL pups are large, although this species produces smaller pups in relation to adult size than do smaller otariids (Kovacs and Lavigne, 1992; McLaren, 1993). In the present study, male pups averaged 22.6 kg and female pups averaged 19.6 kg at 1-5 days of age, which is in the range of birth masses reported in the literature. Two studies conducted before the recent population decline reported 17 kg for male pups at birth (Scheffer, 1945) and a range of 9.1-21.8 kg for male and female pups (Mathisen et al., 1962). Late in the population decline, studies reported a range of 16-23 kg for pups at birth in Alaska (Calkins and Pitcher, 1982) and an extrapolated birth mass of 17.9 kg for five pups for which growth rates were measured in California (Higgins et al., 1988). This is the first, large-scale (in terms of sample size and geographic area) longitudinal study of growth in Steller sea lion pups. Growth rates reported in our study are the highest absolute growth rates reported for any sea lion or fur seal. This is to be expected because adult SSLs are the largest otariids (Kovacs and Lavi- gne, 1992). The growth rate of 0.38 kg/day measured for five SSL pups at Afio Nuevo Island in California (Higgins et al., 1988) falls within the range of average growth rates measured in the present study (0.23-0.48 kg/day). The only other measurement of pup growth in SSLs was conducted on captive pups that were already Chmkof Island 1993 Lowne Island 1993-94, 1997 Marmot Island 1994 15 20 25 Age (days) 35 Figure 3 Summary of Steller sea lion lEumetopias jubatus) pup growth (body mass) during the first six weeks after birth for all five rookeries. The length of each line indicates the length of the study period at that location. Pups from Seguam, Yunaska, and Chirikof Islands, in the declining population, grew significantly faster than pups from Lowrie Island, in the stable population. Pups from Seguam and Yunaska Islands also grew significantly faster than pups from Marmot Island. several months old. In terms of growth rate in relation to size at birth, SSL pups gained 1-2.3% of their birth weight per day (Lowrie Island and Seguam and Yu- naska Islands, respectively, based on an average birth mass of 21.1 kg), which was faster than the relative growth rates reported for other otariid species (Kovacs and Lavigne, 1992, calculated from Table 1), except for northern fur seals. In contrast, seals (order Carnivora, family Phocidae) exhibit faster growth rates (1.3-5.6 kg/day or 8-26% birth weight per day) (Stewart and Lavigne, 1980; Bowen et al., 1985; Kovacs and Lavigne, 1985; Bowen et al., 1987; Bowen et al., 1992; Campagna et al., 1992). Although adult SSLs are larger than many species of phocid seals, phocids have much shorter lacta- tion periods and their pups grow at a more accelerated rate than do otariids. Male-female differences Male pups weighed 159c more than females at birth, indi- cating a difference in maternal investment during gesta- tion, which has been found in other otariids including Antarctic fur seals (Doidge et al., 1984; Lunn and Boyd, 1993; Goldsworthy, 1995; Boyd, 1996), South American fur seals (Arctocephalus australis) (Lima and Paez. 1995), California sea lions (Ono and Boness, 1996), and southern sea lions (Otaria byronia) (Cappozzo et al., 1991). These results are consistent with the predictions of Maynard- Smith's (1980) theory on sexual investment. Steller sea lion adults are extremely sexually dimorphic; females weigh 263 kg on average (maximum of approximately 350 kg); males weigh more than twice as much (average of 566 kg, maxi- mum of approximately 1120 kg) (Calkins and Pitcher, 1982). In view of this dimorphism and the fact that size is more important to male fitness than to female fitness in a polyg- ynous species (McCann, 1981) such as the SSL, theory predicts that males would be heavier than females at birth. Northern fur seal females with male fetuses are in poorer condition than mothers with female fetuses (Trites, 1992), and male fetuses grow at a faster rate than female fetuses (Trites, 1991), indicating that mothers invest more in male offspring during gestation. However, there were no male-female dif- ferences in neonatal growth (BM, SL, and AG) rate in SSL during the first six weeks after birth. In a species as sexually dimor- phic as SSL, one would expect males to grow at a faster rate than females during devel- opment. However, this difference may not occur until the animals are older. There is some evidence that male otariids undergo a sharp increase in growth rate near sexual maturity (McLaren, 1993; Bester and Van Jaarsveld, 1994), after females have already reached sexual maturity and their growth has slowed. "H 40 Brandon et al.: Neonatal growth of Eumetopias jubatus 253 Conflicting results have been reported in other growth studies of otariids. Several studies reported that male pups grew faster than female pups (Antarctic fur seals: Payne, 1979; Doidge et al.. 1984; Antarctic and Subantarctic fur seals: Kerley, 1985; New Zealand fur seals: Mattlin 1981). However, cross-sectional data on growth rate were used in these studies. Conversely, longitudinal data, considered to be more accurate, dem- onstrate no differences in neonatal growth rate between male and female Antarctic fur seal pups (Doidge and Croxall, 1989; Lunn et al., 1993; Lunn and Arnould, 1997); Goldsworthy (1995), however, is the exception. Ono and Boness (1996) collected longitudinal growth data on California sea lion pups and found that males grew faster than females, but they found no other evi- dence of differential maternal investment. In phocids, most studies have found no difference in neonatal male and female growth rates, regardless of whether the data were longitudinal or cross sectional (Stewart and Lavigne, 1980; Innes et al., 1981; Bowen et al., 1992). This is true for species with extreme sexual dimor- phism such as elephant seals (McCann et al., 1989; Campagna et al., 1992). The only other study where growth rates for SSL pups were measured did not have a large enough sample size for a comparison between males and females (Higgins et al., 1988). No differences between male and female pups were found for suckling behavior or maternal attendance behavior (Higgins et al., 1988). Total body lipid Average %TBL of neonatal pups was low (5.6% BM). Steller sea pups are born with small energy stores and normally fast for short periods (about one day) while their mothers make foraging trips to sea. There have been few measurements of lipid content in otariid neo- nates. Jonker and Trites (2000) found a blubber content of 9.7% BM in five SSL pups in the first month after birth. However, this measurement does not correspond directly to body fat content because they measured blub- ber content by weighing the sculp (skin plus blubber) and then calculating the fraction of sculp that was blubber by measuring skin and blubber thicknesses. Using the same labeled water method as in the present study, Arnould et al. (1996b) found a %TBL of 9.4% BM in four Antarctic fur seal pups in the first month after birth. In a similar study of one-day-old Antarctic fur seal pups, Arnould et al. (1996a) found a %TBL of 7.0% BM for female pups and 4.9% BM for male pups. Also using labeled water, Oftedal et al. (1987a) found an average %TBL of 5% BM for neonatal California sea lion pups. Arnould et al. (1996b) suggested two explanations for the higher lipid content that they found in Antarc- tic fur seal pups in comparison to California sea lion pups (Oftedal et al. 1987b). First, in colder habitats, a larger subcutaneous lipid store may be necessary for thermoregulation. The data here do not support that explanation. SSL live in a colder habitat than Cali- fornia sea lions, but have a similar %TBL. The more likely explanation is that larger lipid stores are found in species in which pups normally fast longer while their mothers are foraging. Steller sea lion pups have the smallest lipid stores and shortest fasting periods (Brandon, 2000) of the three species. Differences in pup size among rookeries Although male and female pups differed significantly in size, there were no significant differences in pup size at birth among the rookeries studied. Rookery location should have less influence on pup size at birth than on neonatal growth because maternal foraging range is much greater during gestation than during lactation (Merrick and Loughlin, 1997). This greater maternal foraging range during gestation reduces, among rook- eries, variation in maternal size and feeding conditions (quantity and quality of prey available) during gestation, both of which have been shown to influence pup birth mass in pinnipeds (Calambokidis and Gentry, 1985; Kovacs and Lavigne, 1986; Trites, 1991; Trites 1992). The lack of a difference in pup BM at birth among rook- eries could also be explained by the fact that females that are "successful" (i.e.. carry their fetuses to term) have a significantly better body condition than females that do not carry their fetuses to term (Pitcher et al., 1998). As a consequence of our study design, only those females that were successful were used, and therefore our sample was biased toward females in the population with better body condition. In addition, gestation is less energetically expensive than early lactation; therefore differences in food availability would have less of an effect during gestation (Robbins and Robbins, 1979; Albon et al., 1983; Oftedal, 1984). Although most pup morphometries at first capture did not differ among rookeries, growth parameters differed significantly (Table 3). Growth rates of pups on Seguam and Yunaska Islands (0.48 kg/day) and on Lowrie Is- land (0.23 kg/day) represented the extremes, whereas growth rates of pups on Chirikof, Marmot, and Fish Islands fell between these two extremes. In general, faster growth rates occurred in the west and slower growth rates in the east. In terms of mass, Seguam and Yunaska Islands and Chirikof Island pups grew twice as fast as Lowrie Island pups. A concurrent study of the attendance patterns of lactating females (Brandon, 2000) showed that foraging trip duration decreased from east (25.6 hours on Lowrie Island) to west (an average of9.4 hours on Chirikof and Seguam Islands). Therefore, it is possible that the higher growth rates in SSL pups in the western Gulf of Alaska and Aleutian Islands resulted from shorter periods of fasting while females were foraging at sea (Arnould et al., 1996a; Goldsworthy, 1995). Is food limiting growth in Steller sea lion pups in the area of population decline? If the cause of the population decline were decreased food availability, which is one of the leading hypotheses 254 Fishery Bulletin 103(2) (Pascual and Adkison, 1994; York, 1994; NMFS2!, one might expect the animals in the declining population to show signs of nutritional stress compared to those in the stable population. The results for pup size and growth give no indication of food stress during early lactation. In fact, pups from the declining population on Seguam, Yunaska, and Chirikof Islands grew faster than pups from the stable population on Lowrie Island during the first six weeks. Similar results were also found in a study of pup BM (Merrick et al., 1995), in which pups were weighed on rookeries from Oregon to the Aleutian Islands in late June and early July from 1987 to 1994. Although the pups' ages were unknown, weighing date was used as a covariate in the analysis. Merrick et al. (1995) found a continuous increase in pup BM from Oregon to southeast Alaska and to the Aleutian Islands. These investigators also concluded that pup BM was on average greater in the declining population. In most other studies of declining populations or dif- ferences among rookeries, such contradictory results have not been seen. A study of California sea lion pups during an ENSO (El Nino Southern Oscilliation) event revealed lower pup growth during the period of food stress (Boness et al., 1991). Trillmich and Limberger (1985) have also seen clear effects of low food avail- ability during an ENSO in Galapagos fur seals and sea lions. Antarctic fur seals are affected in predictable ways (increased pup mortality and increased female for- aging time) during times of decreased food availability (Costa et al., 1989). Hood and Ono (1997) found that in the declining California population of SSLs, pups spent less time suckling when adult females made longer for- aging trips in 1992 than in 1973 when the population was larger. The longer foraging trips suggested less abundant food resources. Considering the results for SSL pup growth in light of the population decline, we suggest three alternative hypotheses: 1) food availability was never a factor in the population decline; 2) food availability caused the overall decline, but lactating females and their pups were not affected during early lactation; or 3) our study was conducted when pups and lactating females were no longer experiencing decreased food availability. Faster rates of pup growth may be normal for the Aleutian Islands and western Gulf of Alaska despite the population decline. The declining and stable popula- tions are genetically distinct (Bickham et al., 1996), and perhaps the differences seen in our study are normal differences between the two populations. It is impos- sible to determine if growth and foraging behavior have changed over time because historical data on maternal investment are sparse. Juveniles rather than neonates may be the affected age class in the declining popula- tion (Merrick et al., 1988), whereas lactating females are feeding on either different prey or age classes and not experiencing decreased food availability. York (1994) constructed a population model for SSLs in Alaska and concluded that the current population decline could be accounted for by increased juvenile mortality. Alternatively, because our study was performed late in the decline, the higher growth rates could be the result of lower population density and less competition for food in the declining population. Trites and Bigg (1992) re- ported larger body sizes in northern fur seal populations during a period of decline. The northern fur seal popula- tion in the Pribilof Islands in the Bering Sea increased from the early 1900s to the 1950s. During this period, adult body size decreased. From 1950 to the 1970s the population declined and there was a concurrent increase in individual body size (Trites and Bigg, 1992). Scheffer (1955) hypothesized that increased body size was due to decreased competition for food, which in turn would be due to the lower population density. It is possible that the same density-dependent effects are occurring in the declining SSL population because our study was performed late in the decline, after the original cause may have abated. More information will be needed to determine the cause of the SSL decline and whether it is related to availability of food, especially for different age classes, and to different times of the year. Acknowledgments We thank T. Adams, R. Andrews , D. Bradley, J. Burns, M. Castellini, J. K. Chumbley, W. and S. Cunningham, J. Davis, F. Gulland, D. Gummeson, B. Heath, D. John- son, S. Kanatous, D. Lidgard, R. Lindeman, R. Merrick, D. McAllister, L. Milette, K. Ono, L. Polasek, T. Porter, D. Rosen, J. Sease, T. Spraker, U. Swain, W. Taylor, A. Trites, D. van den Bosch, T Williams, and the captain and crew of the RV Mecleia for assistance in the field. We thank K. Andrews for the map and D. Brandon for assistance in data collection and analysis. G. Worthy, A. Trites, T. Lacher, D. Owens, and M. Reynolds reviewed an early version of this manuscript. Funding and logis- tical support in the field were provided by the Alaska Department of Fish and Game, the National Marine Fisheries Service/National Marine Mammal Labora- tory, Texas A&M University, and the Texas Institute of Oceanography. This research was conducted under Marine Mammal permit no. 846 and 963. Literature cited Albon, S. D., T. H. Clutton-Brock, and R. Langvatn. 1992. Cohort variation in reproduction and survival: implications for population demography. In The biology of deer! R. D. Brown, ed.), p. 15-21. Springer-Verlag. New York, NY. Albon, S. D., F. E. Guinness, and T. H. Clutton-Brock. 1983. The influence of climatic variation on the birth weights of red deer iCervus elaphus). J. Zool. (Lond.) 203:295-298. American Society of Mammalogists. 1967. Standard measurements of seals. J. Mammal. 48:459-462. Arnould, J. R Y, I. L. Boyd, and D. G. Socha. 1996a. Milk consumption and growth efficiency in Ant- Brandon et af: Neonatal growth of Eumetopias jubatus 255 arctic fur seal (Arctocephalus gazella) pups. Can. J. Zool. 74:254-266. Arnould, J. P. Y.. L. L. Boyd, and J. R. Speakman. 1996b. Measuring the body composition of Antarctic fur seals [Arctocephalus gazella): validation of hydrogen isotope dilution. Physiol. Zool. 69:93-116. Baker, J. D„ and C. W. Fowler. 1992. Pup weight and survival of northern fur seals Callorhinus ursinus. J. Zool. (Lond.) 227:231-238. Bester, M. N., and A. S. Van Jaarsveld. 1994. Sex-specific and latitudinal variance in postna- tal growth of the Subantarctic fur seal (Arctocephalus tropicalis). Can. J. Zool. 72:1126-1133. Bickham, J. W„ T. R. Loughlin, J. R. Wickliffe. and V. N. Burkanov. 1998. Geographic variation in the mitochondrial DNA of Steller sea lions: haplotype diversity and endemism in the Kuril Islands. Biosphere Conservation, 1: 107-117 Bickham, J. W., J. C. Patton, and T. R. Loughlin. 1996. High variability for control-region sequences in a marine mammal: implications for conservation and bio- geography of Steller sea lions (Eumetopias jubatus). J. Mammal. 77:95-108. Boltnev, A. I., A. E. York, and G. A. Antonelis. 1998. Northern fur seal young: interrelationships among birth size, growth, and survival. Can. J. Zool. 76: 843-854. Boness, D. J., O. T. Oftedal, and K. A. Ono. 1991. The effect of El Nino on pup development in the California sea lion (Zalophus californianus) I. Early postnatal growth. In Pinnipeds and El Nino: responses to environmental stress (F. Trillmich and K. Ono, eds.l. p. 173-184. Springer-Verlag, Berlin. Bonner, W. N. 1984. Lactation strategies in Pinnipeds: problems for a marine mammalian group. Symp. Zool. Soc. Lond. 51:253-272. Bowen, W. D., and S. J. Iverson. 1998. Estimation of total body water in pinnipeds using hydrogen-isotope dilution. Physiol. Zool. 71:329-332. Bowen, D., D. J. Boness, and O. T. Oftedal. 1985. Birth to weaning in 4 days: remarkable growth in the hooded seal, Cystophora cristata. Can. J. Zool. 63:2841-2846. 1987. Mass transfer from mother to pup and subsequent mass loss by the weaned pup in the hooded seal, Cys- tophora cristata. Can. J. Zool. 65:1-8. Bowen, W.D., W. T. Stobo, and S. J. Smith. 1992. Mass changes of grey seal Halichoerus grypus pups on Sable Island: differential maternal investment reconsidered. J. Zool. (Lond.) 227:607-622. Boyd, I. L. 1996. Individual variation in the duration of pregnancy and birth date in Antarctic fur seals: the role of environ- ment, age, and sex of fetus. J. Mammal. 77:124-133. Brandon, E. A. A. 2000. Maternal investment in Steller sea lions in Alaska. Ph.D. diss., 137 p. Texas A&M University, Galveston, TX. Bryden, M. M. 1968. Control of growth in two populations of elephant seals. Nature 217:1106-1108. Burns, J. M. 1999. The development of diving behavior in juvenile Weddell seals: pushing physiological limits in order survive. Can. J. Zool. 77:737-747. Calambokidis. J., and R. Gentry. 1985. Mortality of northern fur seal pups in relation to growth and birth weights. J. Wildl. Dis. 21:327-330. Calkins, D. G., D. C. McAllister, K. W. Pitcher, and G. W. Pendleton. 1999. Steller sea lion status and trend in Southeast Alaska: 1979-1997. Mar. Mammal Sci. 15:462-477. Calkins. D. G., and K. W. Pitcher. 1982. Population assessment, ecology and trophic rela- tionships of Steller sea lions in the Gulf of Alaska, 140 p. Final report to Outer Continental Shelf Environment Assessment Program, contract 03-5-022-69. Alaska Department of Fish and Game, Anchorage, AK. Campagna, C, B. J. Le Boeuf, M. Lewis, and C. Bisioli. 1992. Equal investment in male and female off- spring in southern elephant seals. J. Zool. (Lond.) 226:551-561. Cappozzo, H. L., C. Campagna, and J. Monserrat. 1991. Sexual dimorphism in newborn Southern sea lions. Mar. Mamm. Sci. 7:385-394. Costa, D. P. 1987. Isotopic methods for quantifying material and energy intake of free-ranging marine mammals. In Marine mammal energetics (A. C. Huntley, D. P. Costa, G. A.J. Worthy, and M. A. Castellini, eds.), p. 43-66. Soc. Mar. Mammal. Special Publ. 1. Allen Press. Lawrence. KS. Costa, D. P., J. P. Croxall, and C. D. Duck. 1989. Foraging energetics of Antarctic fur seals in relation to changes in prey availability. Ecology 70:596-606. Doidge. D. W.. and J. P. Croxall. 1989. Factors affecting weaning weight in Antarctic fur seals Arctocephalus gazella at South Georgia. Polar Biol. 9:155-160. Doidge, D. W., J. P. Croxall, and C. Ricketts. 1984. Growth rates of Antarctic fur seal Arctocepha- lus gazella pups at South Georgia. J. Zool. (Lond.) 203:87-93. Gaillard, J.-M., J.-M. Boutin. D. Delorme, G. Van Laere, P. Dun- can, and J.-D. Lebreton. 1997. Early survival in roe deer: causes and consequences of cohort variation in two contrasted populations. Oeco- logia 112:502-513. Goldsworthy, S. D. 1995. Differential expenditure of maternal resources in Antarctic fur seals, Arctocephalus gazella. and Heard Island, southern Indian Ocean. Behav. Ecol. 6:218-228. Higgins, L.V, C. P. Costa, A. C. Huntley, and B. J. Le Boeuf. 1988. Behavioral and physiological measurements of maternal investment in the Steller sea lion, Eumetopias jubatus. Mar. Mammal Sci. 4:44-58. Hood, W. R., and K. A. Ono. 1997. Variation in maternal attendance patterns and pup behaviour in a declining population of Steller sea lions (Eumetopias jubatus). Can. J. Zool. 75:1241-1246. Innes, S., R. E. A. Stewart, and D. M. Lavigne. 1981. Growth in Northwest Atlantic harp seals Phoca groenlandica. J. Zool. (Lond.) 194:11-24. Jonker, R. A. H., and A. W. Trites. 2000. The reliability of skinfold-calipers for measuring blubber thickness of Steller sea lion pups ( Eumetopias jubatus). Mar. Mamm. Sci. 16:757-766. Kerley, G. I. H. 1985. Pup growth in the fur seals Arctocephalus tropi- calis and A. gazella on Marion Island. J. Zool. (Lond.) 205:315-324. 256 Fishery Bulletin 103(2) Kovacs, K. M., and D. M. Lavigne. 1985. Neonatal growth and organ allometry of North- west Atlantic harp seals iPhoca groenlandica). Can. J. Zool. 63:2793-2799. 1986. Maternal investment and neonatal growth in phocid seals. J. Anim. Ecol. 55:1035-1051. 1992. Maternal investment in otariid seals and wal- ruses. Can. J. Zool. 70:1953-1964. Lima, M., and E. Paez. 1995. Growth and reproductive patterns in the South American fur seal. J. Mammal. 76:1249-1255. Lunn, N. J., and J. P. Y. Arnould. 1997. Maternal investment in Antarctic fur seals: evi- dence for equality in the sexes? Behav. Ecol. Sociobiol. 40:351-362. Lunn, N. J., and I. L. Boyd. 1993. Effects of maternal age and condition on parturi- tion and the perinatal period of Antarctic fur seals. J. Zool. (Lond.) 229:55-67. Lunn. N. J., I. L. Boyd, T. Barton, and J. P. Croxall. 1993. Factors affecting the growth rate and mass at weaning of Antarctic fur seals at Bird Island, South Georgia. J. Mammal. 74:908-919. Mathisen, O. A., R. T. Baade, and R. J. Lopp. 1962. Breeding habits, growth and stomach contents of the Steller sea lion in Alaska. J. Mammal. 43:469-477. Mattlin, R. H. 1981. Pup growth of the New Zealand fur seal Arctocepha- lus forsteri on the Open Bay Islands, New Zealand. J. Zool. (Lond. I 193:305-314. Maynard-Smith, J. 1980. A new theory of sexual investment. Behav. Ecol. Sociobiol. 7:247-251. McCann, T. S. 1981. Aggression and sexual activity of male South- ern elephant seals, Mirounga leonina. J. Zool. (Lond.) 195:295-310. McCann. T. S., M. A. Fedak, and J. Harwood. 1989. Parental investment in southern elephant seals, Mirounga leonina. Behav. Ecol. Sociobiol. 25:81-87. McLaren, I. A. 1993. Growth in pinnipeds. Biol. Rev. 68:1-79. Merrick, R. L.. R. Brown, D. G. Calkins, and T. R. Loughlin. 1995. A comparison of Steller sea lion, Eumetopias juba- tus, pup masses between rookeries with increasing and decreasing populations. Fish. Bull. 93:753-758. Merrick, R., P. Gearin, S. Osmet, and D. Withrow. 1988. Field studies of northern sea lions at Ugamak Island, Alaska, during the 1985 and 1986 breeding seasons. NOAA Tech. Memo. NMFS F/NWC-43, 60 p. Merrick, R. L.. and T. R. Loughlin. 1997. Foraging behavior of adult female and young-of- the-year Steller sea lions in Alaskan waters. Can. J. Zool. 75:776-786. Nagy, K. A. 1975. Water and energy budgets of free-living animals: measurement using isotopically labeled water. In Envi- ronmental physiology of desert organisms (N. Hadley, ed.l, p. 227-245. Dowden, Hutchinson, and Ross, Stroudsburg, PA. Nagy, K. A., and D. P. Costa. 1980. Water flux in animals: analysis of potential errors in the tritiated water method. Am. J. Physiol. 238: R454-R465. Oftedal. O. T. 1984. Body size and reproductive strategy as correlates of milk energy output in lactating mammals. Acta Zoologica Fennica 171:183-186. Oftedal, O.T., S. J. Iverson. and D. J. Boness. 1987a. Milk and energy intakes of suckling Califor- nia sea lion Zalophus californianus pups in relation to sex, growth, and predicted maintenance require- ments. Physiol. Zool. 60:560-575. Oftedal, O. T.. D. J. Boness, and R. A. Tedman. 1987b. The behavior, physiology, and anatomy of lactation in the Pinnipedia. Curr. Mammal. 1:175-245. Ono, K. A., and D. J. Boness. 1996. Sexual dimorphism in sea lion pups: differential maternal investment, or sex-specific differences in energy allocation? Behav. Ecol. Sociobiol. 38:31-41. Ono, K. A., D. J. Boness, and O. T. Oftedal. 1987a. The effect of a natural environmental disturbance on maternal investment and pup behavior in the Cali- fornia sea lion. Behav. Ecol. Sociobiol. 21:109-118. Pascual, M.A., and M. D. Adkison. 1994. The decline of the Steller sea lion in the Northeast Pacific: demography, harvest, or environment? Ecol. Appl. 4:393-403. Payne, M. R. 1979. Growth in the Antarctic fur seal Arctocephalus gazella. J. Zool. (Lond.) 187:1-20. Pitcher, K. W., D. G. Calkins, and G. W. Pendleton. 1998. Reproductive performance of female Steller sea lions: an energetics-based reproductive strategy. Can. J. Zool. 76:2075. Reilly, J., and M. A. Fedak. 1990. Measurement of body composition in living gray seals by hydrogen isotope dilution. J. Appl. Physiol. 69:885-891. Robbins, C. H.. and B. L. Robbins. 1979. Fetal and neonatal growth patterns and maternal reproductive efforts in ungulates and subungulates. Am. Nat. 114:101-116. Scheffer, V. B. 1945. Growth and behavior of young sea lions. J. Mammal. 26:390-392. 1955. Body size with relation to population density in mammals. J. Mammal. 39:493-515. Schoeller, D. A., E. van Santen, D. W. Peterson. W. Dietz, J. Jaspan, and P. D. Klein. 1980. Total body water measurement in humans with 180 and D2 labeled water. Am. J. Clin. Nutr. 33:2686-2693. Stewart, R. E. A., and D. M. Lavigne. 1980. Neonatal growth of Northwest Atlantic harp seals, Pagophilus groenlandicus. J. Mammal. 61:670-680. Trillmich, F, and D. Limberger. 1985. Drastic effects of El Nino on Galapagos pinni- peds. Oecologia, 67:19-22. Trites, A. W. 1991. Fetal growth of northern fur seals: life-his- tory strategy and sources of variation. Can. J. Zool. 69:2608-2617. 1992. Fetal growth and the condition of pregnant north- ern fur seals off western North America from 1958 to 1972. Can. J. Zool. 70:2125-2131. 1993. Biased estimates of fur seal pup mass: origins and implications. J. Zool. (Lond.) 229:515-525. Trites, A. W., and M. A. Bigg. 1992. Changes in body growth of northern fur seals Brandon et al.: Neonatal growth of Eumetopias jubatus 257 from 1958 to 1974: density effects or changes in the ecosystem? Fish. Oceanog. 1:127-136. Tveraa. T., B.-E. Saether, R. Aanes, and K. E. Erikstad. 1998. Body mass and parental decisions in the Ant- arctic petrel Thalassioca antarctica: how long should the parents guard the chick? Behav. Ecol. Sociobiol. 43:73-79. York, A. E. 1994. The population dynamics of Northern sea lions, 1975-1985. Mar. Mammal Sci. 10:38-51. Zar, J. H. 1984. Biostatistical analysis, 718 p. Prentice-Hall, Englewood Cliffs, NJ. 258 Abstract— The carpenter seabream (Argyrozona argyrozona) is an endemic South African sparid that comprises an important part of the handline fishery. A three-year study (1998-2000) into its reproductive biol- ogy within the Tsitsikamma National Park revealed that these fishes are serial spawning late gonochorists. The size at 50% maturity (L50) was estimated at 292 and 297 mm FL for both females and males, respectively. A likelihood ratio test revealed that there was no significant difference between male and female L50 (P>0.5). Both monthly gonadosomatic indices and macroscopically determined ovar- ian stages strongly indicate that A. argyrozona within the Tsitsikamma National Park spawn in the astral summer between November and April. The presence of postovulatory follicles (POFs) confirmed a six-month spawn- ing season, and monthly proportions of early (0-6 hour old) POFs showed that spawning frequency was highest (once every 1-2 days) from December to March. Although spawning season was more highly correlated to photo- period (r = 0.859) than temperature (r = -0.161), the daily proportion of spawning fish was strongly correlated (r=0.93) to ambient temperature over the range 9-22°C. These results indi- cate that short-term upwelling events, a strong feature in the Tsitsikamma National Park during summer, may negatively affect carpenter fecundity. Both spawning frequency and dura- tion (i.e., length of spawning season) increased with fish length. As a result of the allometric relationship between annual fecundity and fish mass a 3-kg fish was calculated to produce fivefold more eggs per kilogram of body weight than a fish of 1 kg. In addition to pro- ducing more eggs per unit of weight each year, larger fish also produce significantly larger eggs. Manuscript submitted 22 September 2003 to the Scientific Editor's Office. Manuscript approved for publication 30 August 2004 by the Scientific Editor. Fish. Bull. 103:258-269 (2005). Reproductive biology of carpenter seabream (Argyrozona argyrozona) (Pisces: Sparidae) in a marine protected area Stephen L. Brouwer Marc. H. Griffiths Department of Marine and Coastal Management Private Bag X2 Rogge Bay 8012, South Africa E-mail (for S. L. Brouwer): sbrouwer'5'deat gov 23 The carpenter seabream (Argyrozona argyrozona), known as "carpenter" regionally, is an endemic South Afri- can sparid found between St Helena Bay and KwaZulu-Natal (Fig.l) (Smith and Heemstra, 1986). Although the third most important species in the line-fishery in terms of landed mass, catch per unit of effort (CPUE) on traditional fishing grounds, declined by 95% during the twentieth century (Griffiths, 2000). Despite the impor- tance of this resource, little research attention has been given to this spe- cies. The only previous study on the reproductive biology of carpenter was based on specimens collected towards the western extreme of the distribu- tion range (west of Cape Agulhas), where most of the fish examined were reproductively inactive (Nepgen, 1977). As a result spawning seasonality was not accurately delineated and sizes at 50% maturity were not calculated. Assuming carpenter to be determi- nate spawners, Nepgen (1977) overes- timated batch fecundity by counting immature oocytes. The objective of the present study was to provide information on spawn- ing seasonality, size at maturity, and annual fecundity of carpenter in the Tsitsikamma National Park (TNP), a 75-km no-take marine protected area (MPA) that has existed for 38 years (Fig. 1). It was envisaged that in conjunction with other studies on carpenter (Brouwer and Griffiths1) in exploited areas this information would assist in determining the af- fects of fishing on the life history of carpenter. Materials and methods Fish were caught from a research vessel at depths between 20 and 90 m by using handlines with baited hooks of 2/o-6/o in size. An attempt was made to sample 60 fish per month between March 1996 and June 1999, although weather conditions did not always allow this number. Sampling involved measuring total and fork length (FL) (mm), whole mass (g), gutted mass (g), determining the sex of fish, and removing the gonads. Gonads were staged macroscopically according to a seven-stage maturity index (Table 1) and weighed to the nearest 0.1 g. The whole gonads were preserved in 10% neutrally buffered formalin or alter- natively fixed in Bouin's solution for 48 hours and then stored in 60% etha- nol. Preserved samples were processed for histological analysis according to the techniques described by Osborne et al. (1999). Length at maturity was modelled by using a 2-parameter logistic ogive of the form Pi 1 1 + exp -; and A = the width of the ogive. The ogive was fitted by minimizing the negative log-like- lihood. Differences in male and female L50 and a were tested by using a ratio test that minimizes the binomial log-likelihood of the form :ln Pi I- Pi) + nt xlnd-p, ) + ln where n = the number of samples in size class i\ and mt = the number of mature fish in size class i. Spawning frequency was estimated by using daily proportions of ovaries containing early postovulatory follicles (POFs), hereafter referred to as the spawn- Table 1 (continued) Stage Macroscopic Microscopic 3 Active male 4 Developing female 4 Developing male 5 Ripe female 5 Ripe male 6 Ripe, running female 6 Ripe, running male 7 Spent female 7 Spent male Testes wider and triangular in cross section. Ovary larger and orange-yellow in color. Eggs clearly discernible. Veins and arter- ies large and plentiful. Testes wider and deeper, creamy white in colour, obvious presence of sperm in main sperm duct. Ovaries are large in diameter, may have a few hydrated eggs. Yellow oocytes take up all the space. Veins and arteries large and plentiful. Sperm present in main sperm duct and in tissue. Gonad soft and breaks when lightly pinched. Ovary amber in colour. Large with sub- stantial proportion of gonad with hy- drated eggs, which fill the lumen. Veins and arteries large and plentiful. Free-flowing sperm extruded from fish when the abdomen is lightly squeezed. Testes very delicate and break easily when handled. Copious amounts of sperm pres- ent in main sperm duct and in tissue. Ovary reduced in size similar to stage-2 flaccid ovary. Few yolked oocytes remain- ing. Ovary bloodshot. Testes white in color, smal and bloodshot. shrivelled. The seminal vesicles expand and become filled with spermatogonia. Yolk vesicles are common and primary yolk oocytes begin to appear, which are characterized by the for- mation of small spherical yolk granules. The seminiferous tubules of the testes are filled with spermatozoa, which are also present in the primary sperm duct. Tertiary yolk oocytes, characteriszed by large yolk plates, appear along with primary yolk and yolk vesicles. The nucleus becomes irregular in shape and smaller in size. The nucleus migrates to the animal pole of the cell after which hydration begins, result- ing in increased transparency of the cells and an increase in cell size. The seminiferous tubules expand with copious amounts of spermatozoa that fill the lumen of the primary sperm duct. Filled with hydrated oocytes. Due to dehydration during the histological preparation, these oocytes appear as collapsed bags. Hydrated oocytes may squash and reshape the immature oocytes that sur- round them. The seminiferous tubules of the testes appear dis- tended and are filled with mature spermatozoa as is the lumen of the primary sperm duct. Cells in various stages of atresia, and some hydrated and mature oocytes may be present in the tissue. The seminiferous tubules are no longer distended and have thicker walls than stage-6 tubules. They contain few spermatozoa, which are present in the lumen of the primary sperm duct. Large blood ves- sels are apparent in the tissue. Brouwer and Griffiths: Reproductive biology of Argyrozona argyrozona 261 ing fraction (Hunter and Macewicz, 1985). POFs were aged by comparing them with known age POFs from spawning females under captive conditions. Female carpenter were held in a flow-through system at ambi- ent sea temperature (mean 16C, range 9.5-20°C) in 5000-liter circular tanks, were stimulated to ovulate with a commercially available GnRH-analogue (Davis, 1996). Three fish were sacrificed immediately after ovulation and then three fish every six hours over the following 48-hour period. Histological analysis of ova- ries revealed three clearly defined POF stages (Fig. 2). The proportions of wild-caught fish with stage-1 POFs (the spawning fraction) were inverted to produce an estimate of spawning frequency (Wilson and Nieland, 1994). Batch fecundity was estimated from counts of hy- drated oocytes from ovaries without POFs or atretic oocytes (Hunter and Macewicz, 1985). A ±1.00-g section was removed from the middle of the right ovary. This was weighed to the nearest 0.01 g and the hydrated oo- cytes were separated according to the method described by Lowerre-Barbieri and Barbieri (1993). Hydrated oocytes were suspended in water and counted at 8-10 times magnification in a Bokkeroff tray and measured to 0.1 mm with an ocular micrometer along the longest diameter. Annual fecundity was calculated as follows: Aft xfbt, where Aft = the annual fecundity for fish t; Is = the length of the spawning season (days) for fish of size class j; sf = the spawning frequency (days) for fish of size class j (all months combined); and fbt = the batch fecundity of fish t. Spawning season was established by calculating the monthly proportions of macroscopic gonad stages and mean monthly gonodosomatic index (GSI) for fish larger than L50: GSI- xlOO, where m = the gonad mass (g); and ms = the somatic mass (g) (minus gonad and stomach mass). In order to investigate the relationship between spawning and temperature, temperature data were collected at the sampling site with a Seamon Mini (Hu- grun, Iceland) recorder stationed at at a depth of 35 m on the reef from which the biological samples were col- lected. A thermistor array consisting of four underwater temperature recorders (UTRs) at depths of 12 m, 19 m, 27 m, and 35 m recorded the temperature every minute Figure 2 Postovulatory follicle (POF) stages deter- mined from carpenter [Argyrozona argyro- zona) chemically induced to spawn in an open circulating system housed at the Tsitsikamma National Park. (A) = stage 1 (0-6 hours), ( B ) = stage 2(7-24) hours and ( C I = stage 3 125-48) hours. 262 Fishery Bulletin 103(2) and stored an hourly average (Roberts-). Photoperiod data were downloaded from the South African Astro- nomical Observatory database.3 Pearson Rank correla- tion was used to measure the correlation between GSI and temperature, and GSI and photoperiod trends. Results Histological examination of the gonads revealed that although juveniles possess both testicular and ovarian tissue simultaneously (i.e., as hermaphrodites) they mature as either a male or female (Table 1) and are therefore late gonochorists (rudimentary hermaphro- dites). Gametogenesis was similar to that described for other late gonochoristic sparids e.g., Pterogymnus lania- rius (Booth and Hecht, 1997). The size at 50% maturity was estimated at 292 and 297 mm FL for females and males, respectively (Fig. 3), and in both cases is equiva- lent to an age of about five years (Brouwer and Griffiths 2004). A likelihood ratio test revealed that there was no significant difference between male and female L50 (P>0.5)or or (P>0.1). Complete ( 100% ) maturity for both sexes occurred at 480 mm FL, an age of about 15 years (Brouwer and Griffiths 2004). The sex ratio was 1F:1.3M (n=1776); a chi-square test with Yates' correction factor revealed that this sex ratio was a significant difference from unity (P<0.01). Three age-related POF stages were identified within the ovaries of captive spawned carpenter (Fig. 2). Stage- 1 POFs (0-6 hours) were very loosely arranged and ap- peared as a long convoluted string with a large clearly defined lumen. The granulosa cells were clearly visible and widely spaced and had clearly visible nuclei (Melo, 1994). Stage-2 POFs (7-24 hours) are smaller and more densely packed but still have a visible lumen. The gran- ulosa cells are closely packed and dense. Stage-3 POFs (25-48 hours) are small and densely packed. There is no lumen and the granulosa cells are closely arranged and no longer distinguishable from one another. After 48 hours at 16°C, POFs were no longer detectable. Mean GSI and the proportions of ripe (stage-5) and ripe, running (stage-6) fish increased in November and remained high until April (Figs. 4 and 5), indicating that carpenter are summer spawners. The presence of early POFs from November to March (sample numbers being too low for April) supported the macroscopically determined spawning season. The monthly spawning fraction did, however, reveal that spawning frequency was highest in January and February when the fish spawned at two-day intervals and lowest in November and April when they were found to spawn every 2-3 days (Table 2). o 1B0 230 280 330 380 430 480 530 580 1 ■ Female 0 9 ■ n=778 L50=292 • 0 8 ■ • / ■ • 07 - 0.6 ' • 05 ■ • 04 ■ 03 ■ 02 ■ 0 1 ■ n ■ — • — • — •— i- 130 180 230 280 330 380 430 480 530 580 Fork length (mm) Figure 3 The proportion of mature carpenter (Argyrozona argy- rozona) in length classes sampled in the Tsitsikamma National Park. The curves were fitted with a 2-parameter logistic ogive. 2 Roberts, M. J. 1999. CD-ROM, Tsitsikamma National Park oceanographic data, version 1.0. Marine and coastal man- agement. Private Bag X2, Rogge Bay 8012, South Africa. :l http://www.saao.ac.za [Accessed August 2000], Batch fecundity was positively correlated with both fish mass (r=0.71) and fork length (r=0.71). No correla- tion was found between fish length and relative batch fecundity (eggs/fish somatic mass) (Fig. 6). The propor- tion of fish with stage-1 POFs revealed that spawning frequency and length of the spawning season increased with fish length (Table 3). Accounting for size-related patterns in spawning season (Fig. 7) and frequency, we found that annual fecundity increased allometrically with mass (Fig. 8) and age (Table 4). Hydrated egg size was significantly smaller and more variable (average 1.0 mm ±0.16) in fish below the length at 100% maturity (480 mm FL) than those above this length (1.1 mm ±0.09) U-test, P<0.005). Brouwer and Griffiths: Reproductive biology of Argyrozona argyrozona 263 100 90 80 70 60 50 40 30 20 10 0 100 90 80 70 60 50 40 30 20 10 0 Male m m m Saj 1 ABS 12 3 4 5 9 10 11 12 12 3 4 5 6 7 Months 10 11 12 n=998 □ 1 □ 3 □ 4 Female n=778 pri f "H F E f I D1 3 - 1 3 2 s □ — : : h j III D3 □ 4 mm ^ i s : : si ni e Mr Figure 4 Monthly variation in the proportion of macroscopic gonad stages of carpenter {Argyrozona argyrozona) >L50 caught in the Tsitsikamma National Park (March 1996-July 1999). Numbers in the legend refer to the gonad stages in Table 1. 1 = juvenile, 2 = immature. 3 = active, 4 = developing, 5 = ripe. 6 = ripe, running, and 7 = spent. Table 2 Spawning frequency determined for carpenter cally determined from hydrated oocytes. (A. argyrozona) from the proportion of ovaries with stage -1 POFs or macroscopi- Month Spawning frequency (days) % ovaries with hydrated oocytes % ovaries with stage-1 POFs Macroscopic POFs November 2.1(441 4.6(23) 48 22 December 1.9(53) 3.5(21) 53 28 January 1.5(119) 1.6(5) 66 60 February 1.5(160) 1.5(35) 68 66 March 1.6(99) — 64 Not enough data April 2.6(49) — 39 Not enough data A positive relationship between temperature at the time of spawning (back-calculated from stage-1 POFs, assuming a delay of 6 hours) and the proportion of ova- ries with stage-1 POFs indicated that spawning events were positively correlated with temperature (r=0.93) over the range 9°C and 22°C (Fig. 9). GSI was how- 264 Fishery Bulletin 103(2) Male 1 . ;; 0 1 • i 1 ",T T. — — 1 1 1 1 1 1 CO o 6 5 4 3 + 2 1 Jan Feb Mar Apr May Jun Jul Aug Sept Oct Nov Dec Female i f — i 1 — i 1 1 1 — i 1 1 1 — i 1 Jan Feb Mar Apr May Jun Jul Aug Sept Oct Nov Dec Month Figure 5 Seasonal variation of the standard deviation in the gonadosomatic index (GSI) and mean values (•) for male and female carpenter (Argyrozona argyrozona) sampled in the Tsitsikamma National Park. ever strongly correlated (/-=0.86) with photoperiod but exhibited a weak negatively relationship with seasonal temperature (Fig. 10). Discussion Late gonochorism, protandry, protogyny, and hermaphro- ditism are the recognized reproductive styles of sparids (Smale, 1988; Buxton and Garratt, 1990). Although carpenter were previously described as gonochoristic (Nepgen, 1977), microscopic examination of the gonads revealed that they are late gonochorists. The sex ratio calculated during this study (1 female:1.3 male) was typical for those observed for other late gonochorists (Griffiths et al., 2002). Upon reviewing 90 species of reef fish, Sadovy (1996) concluded that although GSIs reflect the gonad maturity patterns for a species, they are poor indicators of peak spawning times. By way of example, in red hind grouper (Epmephelis guttatus) yolked oocytes are present in the ovaries for four months of the year but actual spawning Table 3 Spawning frequency (averaged over all months) and length of the spawning season calculated from the pres- ence of stage-1 POFs in carpenter (Argyrozona argyro- zona) ovaries in three size classes. Size class (mml Average spawning frequency ( days ) Spawning season (months) 250-339 340-479 480+ 9 4 3.9 is limited to a period of 10 days (Sadovy, 1996). In the case of carpenter, however, the presence of POFs from November to April supports the six-month spawning season indicated by macroscopic methods (although in some larger individuals [>480 mm FL] hydrated Brouwer and Griffiths: Reproductive biology of Argyrozona argyrozona 265 180000 160000 "> 140000 Ol Ol * 120000 "o ^ 100000 a> £ 80000 ■ Z 60000 40000 20000 0 1 = 360. 22x- 9741 r* = 0 5046 • • • • ^^ • • o *%J o • o • ©Numbefoleggs n = 51 ODavis (1997) n = 10 250 450 550 Fork length (mm) 1250 1750 2250 2750 Gonad free mass (g) 3000 ■ • • 2500 • • in >, jo • • S fc 2000 • • c >■ 3 3 3 o 1500 ■ • • I.- • £ o> to x 0) Ol C£ Ol en 1000 - • • " • ft • • • • ai 500 " • • 250 350 450 550 650 Fork length (mm) Figure 6 The relationships for carpenter (Argyrozona argyrozona) between (A) fish length and batch fecundity including data from fish spawned artificially in a previous study (Davis, 1996 I. (B) fish mass and batch fecundity and (C) fish length and relative batch fecundity. oocytes and POFs were found from October to May). Monthly spawning fraction and percentage of ovaries with hydrated oocytes nevertheless reached a peak dur- ing January and February (Table 2); these trends were not detected in the monthly GSIs. But given that the macroscopic determinations of stage followed trends in the proportions of POFs that were present, we conclude that expensive and time-consuming histological analy- sis is not necessary for determining spawning peaks for this species. 266 Fishery Bulletin 103(2) 7 - 250-339 mm FL II 6 ■ n=179 1 s 1 5 - "1 3 2 1 i o - 1 i! • • : • ' 1 1 • • i! • 1 1 • I » 2 2 3 4 5 6 7 8 9 10 11 1 7 - 6 - ■I 2i • I 1 340-479 mm FL n=421 • • | • • • 5 i 1 < • 9 1 1 | • 1 1 1 • 1 2 3 4 5 6 7 8 9 10 11 12 7 - 480+ mm FL 6 - n=67 1 5 - 4I 3 1 * • 1 : . i : * • • • • 1 • 2 - 1 - 0 - 1 1 • • . • * * 1 1 1 • 1 1 * 2 2 3 4 5 6 7 8 9 10 11 1 Months Figure 7 Seasc nal variation in the gonadosomati( index (GSI) fo r femal e carpenter (Argyrozona argyrozona ) from three size classt s samplec in the Tsitsikamma National Park. Apart from being indicators of spawning seasonality, GSI trends can provide insight into the mating patterns of a species (Sadovy, 1996). Pair-spawning sparids such as Chrysoblephus laticeps have low male GSI (±10% of female) during the spawning season (Buxton, 1990). Although the spawning behavior of carpenter has not been documented, the GSI of males (average 3.0 ±1.4) was similar to that of females (average 3.3 ±1.4) dur- ing the spawning season (Fig. 4). The large testes size suggests that carpenter are group spawners and that sperm competition is high (Sadovy, 1996). Further evi- dence for group spawning is the lack of sexual dimor- phism in this species (Smale, 1988; Mann and Buxton, 1998; Griffiths et al., 2002). Like many other South African sparids, carpenter are summer spawners (Buxton and Clarke, 1986; Buxton and Clarke, 1991; Buxton, 1993). Although various en- vironmental cues have been suggested for this seasonal spawning, it is probably a combination of events that leads to gonad maturation and spawning. Smale (1988) and Garratt (1985) speculated that increases in gonad activity of Petrus rupestris and Chrysoblephus puniceus were attributed to an increase in photoperiod and water temperature respectively; Scott and Pankhurst (1992), Brouwer and Griffiths: Reproductive biology of Argyrozona argyrozona 267 16 - 14 - | 12- I 10 - g 8 ■ ai y = 5E-06x r1 = 0.966 # • • •* n = 50 • / •^ * • ° 6 1 n E 4 ■ 2 • jS» j4* ^■» 0 1000 2000 3000 Mass (g) Figure 8 The relationship between annual fecundity and fish weight for carpenter (Argyrozona argyrozona} in the Tsitsikamma National Park. however, showed that seasonal temperature regulated gonad development for Pagrus aratus. Based on the data collected during our study, photoperiod appears to be responsible for the onset of gonad maturation in carpen- ter; when day length increases (but water temperature is variable) in September and October, and their gonads begin to develop (Fig. 10). Photoperiod was also highly correlated with GSI (r = 0.86), whereas temperature showed a weakly negative relationship (r=-0.16). Nepgen (1977) calculated spawning frequency for this species with an oocyte-size-frequency analysis of inactive females. Finding only one peak in the oocyte- size-frequency distribution, he assumed that carpen- ter spawned only once a year. In our study POFs and various yolk stage oocytes were found to occur simul- taneously, proving that carpenter are serial spawners. Accounting for monthly trends in spawning frequency and the length of the spawning season, carpenter in the Tsitsikamma National Park are estimated to spawn at least 30 times per year. This spawning frequency is similar to other predatory reef fishes, e.g., Mycteroperca microlepis (30-40 times per year) (Collins et al., 1998). Nevertheless, as with other species (Danilowicz, 1995), spawning fraction in carpenter during the spawning season was highly correlated with water temperature (r=0.931) (Fig. 9), indicating that short-term cold water upwellings, a common feature of the TNP during sum- mer (Schumann et al., 1982), may negatively impact annual carpenter fecundity in this area. Although fecundity in fishes is highly variable be- tween individuals (Sadovy 1996), absolute fecundity increases with size (Hunter et al., 1985; Davis and West, 1993; Wilson and Nieland, 1994; Collins et al., 1998). In our study absolute annual fecundity increased markedly with fish size (Table 4) and spawning season was longer for large fish (Fig. 7) (Table 3). The positive Table 4 Age-based annual fecundity of carpenter (Argyrozona argyrozona ) in the Tsitsikamma National Park. Age (yr) Number of eggs (millions I 1 0 2 0 3 0 4 0.143 5 0.288 6 0.367 7 0.441 8 0.870 9 1.014 10 1.228 11 1.498 12 1.706 13 1.763 14 2.260 15 2.233 16 2.427 17 3.132 18 3.175 19 5.363 20 6.308 21 6.308 22 7.815 23 7.430 24 6.480 25 7.421 26 8.363 27 8.363 28 8.064 29 10.397 30 11.808 correlation of batch fecundity and fish size (r=0.71), coupled with the increased length of the spawning sea- son for the older fish, greatly increases the absolute annual fecundity of larger fishes (Fig. 8). Sadovy (1996) noted that for red snapper (Lutjanus compechatius) one large female (601 mm FL) will produce as many eggs as 212 small (420 mm FL) females. Similarly, one large female carpenter of 3.3 kg will produce as many eggs as 72 small ones of 0.3 kg. In addition to higher fecundity, the larger fish produce significantly larger eggs and presumably more viable larvae (Ojanguren et al., 1996; Pepin and Anderson, 1997). Exploited populations were traditionally managed to maximize growth (Griffiths, 1997). However it is imperative to maintain sufficient numbers of reproduc- 268 Fishery Bulletin 103(2) tive adults to ensure egg production and avoid recruit- ment failure. To address proper management of line- caught fish in South Africa, spawner biomass per re- cruit models have been used (Griffiths, 1997). One as- sumption of this approach is that fecundity is linearly related to spawning biomass, regardless of individual size (Buxton, 1992). Because our study has shown that fecundity in carpenter is allometrically related to individual mass, egg-per-recruit models would be more appropriate for future stock assessment of this species. Acknowledgments We thank the staff at South African National Parks, for accommodation at the Tsitsikamma National Park and for use of their vessel. John Allen and Karoels Piterse are thanked for many hours at sea. Yolande Melo is thanked for her assistance with histological preparation and interpretation and Jeanine Van der Pol for assistance in laboratory and Tony Booth and two anonymous referees for constructive comments on this manuscript. This research was funded by the Marine Living Resources Fund. 100 ■ 90 ■ y = 3.2928X- 6.5009 80 ■ r* = 0.8672 70 ■ en .£ 60 ■ c 1 50 • Q. "1 40- 6 5 6 6 • •*-* 20 ^-^^^ 2*8 30 ■ •^^^-"^23 20 ■ ^ • 10 ■ 8 10 12 14 16 18 20 22 24 Temperature Figure 9 The relationship between proportion of carpenter (Argyrozona argyrozona) spawning (back calculated from stage-1 POFs, 0-6 hours 1 and temperature in the Tsitsikamma National Park. Num- bers above symbols refer to number offish sampled with POFs. 20 -I - 15:24 A A .A 18 ■ °v .' '. ▲' ■ 14:12 O ~Z 16 ■ =3 CD Q. 14 - 1 12 - A "-A, / \-* A" - 13 00 g o ■ 11 48 m 5" ■ 10:36 S o ■ 9 24 | ■ 8:12 - - - A- - - Temperature - 7:00 1 ' January February March April May S June o 1 July August September CD O O November December Figure 10 Monthly average temperature and photop eriod for the Tsitsi- kamma National Park. Literature cited Booth, A. J„ and T. Hecht. 1997. A description of gametogenesis in the panga Pterogymnus laniarius (Pisces: Spari- dae) with comments on changes in maturity patterns over the past two decades. S. Afr. J. Zool. 32(21:49-53. Brouwer, S. L„ and M. H. Griffiths. 2004. Age and growth of Argyrozona argyrozona (Pisces: Sparidae) in a Marine Protected Area: an evaluation of methods based on whole otoliths, sectioned otoliths and mark recapture. Fish. Res. 61:1-12. Buxton, C. D. 1990. The reproductive biology of Chrysoblephus laticeps and C. eristieeps (Teleostei: Sparidae). J. Zool. Lond. 220:497-511. 1992. The application of yield-per-recruit models to two South African sparid reef species, with special consideration to sex change. Fish. Res. 15:1-16. 1993. Life-history changes in exploited reef fishes on the east coast of South Africa. Environ. Biol. Fish. 36: 47-63. Buxton, C. D., and J. R. Clarke. 1986. Age growth and feeding of the blue hotten- tot Pachymetopon aeneum (Teleostei: Sparidae) with notes on reproductive biology. S. Afr. J. Zool. 21:33-38. 1991. The biology of the white musselcracker Sparodon durbanensis (Pisces: Sparidae) on the Eastern Cape coast. South Africa. S. Afr. J. Mar. Sci. 10:285-296. Buxton, C. D., and P. A. Garratt. 1990. Alternative reproductive styles in sea- breams (Pisces: Sparidae). Environ. Biol. Fish. 28:113-124. Collins, L. A., A. G. Johnson, C. C. Koenig. and M. S. Baker. 1998. Reproductive patterns, sex ratio, and fecun- dity in gag, Mycteroperca microlepis (Serranidael, a protogynous grouper from the northeastern Gulf of Mexico. Fish. Bull. 96:415-427. Danilowicz, B. S. 1995. The role of temperature in spawning of dam- selfish Da.icyllus albisella. Bull. Mar. Sci. 57(3 1: 624-636. Davis, J. A. 1996. Investigations into the larval rearing of Brouwer and Griffiths: Reproductive biology of Argyrozona argyrozona 269 two South African sparid species. M.S. thesis, 138 p. Rhodes Univ., Grahamstown, Eastern Cape, South Africa. Davis, T. L. O.. and G. J. West. 1993. Maturation, reproductive seasonality, fecundity, and spawning frequency in Lutjanus vittus (Quoy and Gaimard) from the North West shelf of Australia. Fish. Bull. 91:224-236. Garratt, P. A. 1985. The offshore linefishery of Natal. II: Reproduc- tive biology of the sparids Chrysoblephus puniceus and Cheimerius nufar. Invest. Rep. Oceanogr. Res. Inst. 63:1-21. Griffiths, M.H. 1997. The application of per-recruit models to Argyro- somus inodorus, an important South African sciaenid fish. Fish. Res. 30:103-115. 2000. Long-term trends in the catch and effort of commer- cial linefish off South Africa's Cape Province: snapshots of the 20lh century. S. Afr. J. Mar. Sci. 22:81-110. Griffiths, M. H., C. Wilke, A. J. Penny, and Y. Melo. 2002. Life history of white stumpnose Rhabdosargus globiceps (Pisces: Sparidael off South Africa. S. Afr. J. Mar. Sci. 281-300. Hunter, J. R., N. C. H. Lo. and R. J. H. Leong. 1985. Batch fecundity and multiple spawning in fishes. In An egg production method for estimating spawner biomass of pelagic fish: application to the northern anchovy, Engraulis mordax (R. Lasker, ed.l, 67-77. NOAA Tech. Rep. NMFS 36. Hunter, J. R., and B. J. Macewicz. 1985. Rates of atresia in the ovary of captive and wild northern anchovy Engraulis mordax. Fish. Bull. 83(2): 119-136. Lowerre-Barbieri, S. K., and L. R. Barbieri. 1993. A new method of oocyte separation and preser- vation for fish reproduction studies. Fish. Bull. 91: 165-170. Mann, B. Q., and C. D. Buxton. 1998. The reproductive biology of Diplodus sargus capen- sis andD. cervinus hottentotus (Sparidae) off the South- East Cape coast, South Africa. Cybium 22(11:31-47. Melo, Y. C. 1994. Spawning frequency of the anchovy Engraulis capensis. S. Afr. J. Mar. Sci. 14:321-331. Nepgen, C. S. V. 1977. The biology of the hottentot Pachymetopon blochii (Val.) and the silverfish Argyrozona argyrozona (Val.) along the cape south-west coast. Investl. Rep. Div. Sea Fish. Rep. S. Afr. 105:1-35. Ojanguren, A. F., and F. G. Reyes-Gavilan F. Brana. 1996. Effects of egg size on offspring development and fitness in brown trout, Salmo trutta L. Aquaculture 147(1-21:9-20. Osborne, R. F., Y C. Melo, M. D. Hofmeyer, and D. W. Japp. 1999. Serial spawning and batch fecundity of Merluc- cius capensis and M. paradoxus. S. Afr. J. Mar. Sci. 21:2111-216. Pepin, D., C. Orr, and J. T Anderson. 1997. Time to hatch and larval size in relation to tempera- ture and egg size in Atlantic cod iGadus morhua). Can. J. Fish. Aquat. Sci. 54 (suppl. 11:2-10. Sadovy, Y. J. 1996. Reproduction of reef fish species. In Reef fish- eries (N. V. C. Polunin and C. M. Roberts, eds.l, p. 15-59. Chapman and Hall, London. Schumann, E. H., L.-A. Perrins, and I. T. Hunter. 1982. Upwelling along the South Coast of the Cape prov- ince. South Africa. S. Afr. J. Sci. 78:238-242. Scott. S. G, and N. W. Pankhurst. 1992. Interannual variation in the reproductive cycle of the New Zealand snapper Pagrus aratus (Bloch & Schneider) (Sparidae). J. Fish Biol. 41:685-696. Smale, M. J. 1988. Distribution and reproduction of the reef fish Petrus rupestris (Pisces: Sparidae) off the coast of South Africa. A. Afr. J. Zool. 1988. 23(4):272-287. Smith, M. M., and P. C. Heemstra. 1986. Smiths Sea fishes, 1047 p. Southern Book Pub- lishers, Johannesburg, South Africa. Wilson, C. A., and D. L. Nieland. 1994. Reproductive biology of red drum, Sciaenops ocel- latus, from the neritic waters of the northern Gulf of Mexico. Fish. Bull. 92:841-850. 270 Abstract— During the 1990s, sea otter i En hydra lutris) counts in the Aleu- tian archipelago decreased by 70% throughout the archipelago between 1992 and 2000. Recent aerial surveys in the Aleutians did not identify the eastward extent of the decline; there- fore we conducted an aerial survey along the Alaska Peninsula for com- parison with baseline information. Since 1986, abundance estimates in offshore habitat have declined by 27-49% and 93-94% in north- ern and southern Alaska Peninsula study areas, respectively. During this same time period, sea otter density has declined by 63% along the island coastlines within the south Alaska Peninsula study area. Between 1989 and 2001, sea otter density along the southern coastline of the Alaska Pen- insula declined by 35% to the west of Castle Cape but density increased by 4% to the east, which may indicate an eastward extent of the decline. In all study areas, sea otters were primar- ily concentrated in bays and lagoon, whereas historically, large rafts of otters had been distributed offshore. The population declines observed along the Alaska Peninsula occurred at roughly the same time as declines in the Aleutian islands to the east and the Kodiak archipelago to the west. Since the mid-1980s, the sea otter population throughout south- west Alaska has declined overall by an estimated 56-68%, and the decline may be one of the most significant sea otter conservation issues in our time. Decline in sea otter (Enhydra lutris) populations along the Alaska Peninsula, 1986-2001 Douglas M. Burn Angela M. Doroff Marine Mammals Management Office U.S. Fish and Wildlife Service 10H East Tudor Road Anchorage, Alaska 99503 E-mail address (for D M Burn): douglas_burnifi'fws gov Manuscript submitted 2 September 2003 to the Scientific Editor's Office. Manuscript approved for publication 30 August 2004 by the Scientific Editor. Fish. Bull. 103:270-279 (2005). During the 1990s, the sea otter (Enhydra lutris) population in the Aleutian archipelago declined at a rate of 17.5%/yr and, overall, counts decreased by 70% throughout the archipelago between 1992 and 2000 (Doroff et al., 2003). By modeling population trends back to the mid- 1980s, Burn et al. (2003) estimated the population in the Aleutian Island chain decreased by 65,000 sea otters and was at about 10% of its carry- ing capacity in 2000. The 2000 aerial survey of Doroff et al. (2003) did not identify an eastward extent of the population decline however; therefore additional sea otter surveys along the Alaska Peninsula were needed. Historic information on population status and trends is sparse for sea otters along the Alaska Peninsula. Sea otters were exploited to near ex- tinction in the commercial fur trades (1742-1911) and were removed from large portions of their historic range worldwide (Kenyon, 1969; Lensink, 1962). At the time of their protec- tion in 1911 by an international fur seal treaty, there were 13 remnant populations remaining worldwide, 11 of which persisted and grew to re- colonize much of the former range of this species (Kenyon, 1969). Studies of both remnant native and translo- cated sea otter populations have in- dicated a pattern of colonization with high population growth rates up to 20% per year, and an expansion into adjacent, unoccupied habitat (Estes, 1990). One remnant population survived on the north side of the Alaska Pen- insula near Unimak Island (Kenyon, 1969; Schneider, 1976). Sea otter habitat in this area is unique in that shallow water (less than 100 m) ex- tends up to 50 km offshore, covering more than 10,000 km of open water. The remnant population in this ar- ea likely numbered fewer than 100 sea otters in 1911 (Kenyon, 1969). This population grew steadily and expanded its range to the northeast along the Peninsula until 1970, when extreme sea ice conditions temporar- ily reduced the range and likely the size of the population (Schneider and Faro, 1975). By 1976, most of the sea otters in this area were concentrated between Cape Mordvinof and Cape Leontovich (Schneider, 1976). In addition to the remnant popu- lation on the north side of Unimak Island, there were also two remnant populations of sea otters located to the south of the Alaska Peninsula in the Sandman Reefs and the outer Shumagin Islands (Kenyon, 1969). Sea otter habitat along the southern Alaska Peninsula differs from the northern side and comprises primar- ily rocky, mixed substrate, and ex- tensive offshore reefs (Brueggeman et al.1). In the Sandman Reefs a small number of sea otters were sighted in 1 Brueggeman, J. J., G. A. Green, R. A. Grotefendt, and D. G. Chapman. 1988. Aerial surveys of sea otters in the northwestern Gulf of Alaska and the southeastern Bering Sea. Minerals Management Service and NOAA final report, 87 p. Minerals Management Service, Anchorage, AK. [Contract no. 85-ABCV-00093.] Burn and Doroff: Decline of Enhydra lutns along the Alaska Peninsula 271 1922, and by 1962 the population had grown to an es- timated 1625 sea otters (Lensink, 1962; Kenyon, 1969). Around the same time, the population in the Shumagin Islands was estimated to be 2724 sea otters (Kenyon, 1969). The first systematic surveys of sea otter abundance along the north side of the Alaska Peninsula were con- ducted in the mid-1970s (Schneider, 1976), followed by surveys in 1982 and 1983 by Cimberg et al.2 Bruegge- man et al.1 conducted quarterly surveys of both the northern and southern Peninsula in 1986 to assess sea otter abundance and seasonal distribution. The surveys conducted in 1986 provided seasonal estimates of abundance during a single, ice-free year, and a clear picture of habitat use in the mid-1980s along the Alaska Peninsula (Brueggeman et al.1). The sea otter surveys described above were concen- trated along the western end of the Alaska Peninsula where remnant populations existed and appeared to have recovered. By the late 1980s, sea otters had also returned to the nearshore waters of the entire penin- sula as far east as Cape Douglas (DeGange et al.3). Prior to this survey in 1989, little was known about sea otter distribution and abundance on the Alaska Peninsula east of Kupreanof Point. The objectives of our study were 1) to assess current sea otter distribution and abundance along the north- ern and southern Alaska Peninsula, 2) to contrast our results with prior surveys conducted in 1986 and 1989, and 3) to relate these data to the observed sea otter population declines observed elsewhere in southwest Alaska. We repeated the aerial survey methods devel- oped by Brueggeman et al.1 for sea otter habitat along the Alaska Peninsula which consisted of a combination of strip transects in offshore habitat (to the 70-m iso- bath) and coastline surveys (si km of shore) of island groups within the study area. We also repeated the coastline surveys of DeGange et al.2 to determine the eastward extent of the decline. Materials and methods Offshore survey areas The north Alaska Peninsula (NAP) study area ranged from Cape Mordvinof on Unimak Island in the west to Cape Seniavin in the east. This area was further subdi- 2 Cimberg, R. L., D. P. Costa, and P. A. Fishman. 1984. Eco- logical characterization of shallow subtidal habitats in the north Aleutian Shelf. OCSEAP Final Rep. no. 4197, 99 p. U.S. Dept. of Commerce, National Oceanographic and Atmo- spheric Administration, Anchorage, Alaska 99501. 3 DeGange, A. R., D. C. Douglas, D. H. Monson and C. M. Robbins. 1994. Surveys of sea otters in the Gulf of Alaska in response to the Exxon Valdez oil spill. Final report to the Exxon Valdez Oil Spill Trustee Council, Marine Mammal Study 6-7, 11 p. U.S. Fish and Wildlife Service, Anchorage, Alaska 99503. vided into two subunits (NAPa and NAPb), and a line at 162°W longitude divided the two subunits (Brueggeman et al.1). The south Alaska Peninsula (SAP) study area ranged from the Ikatan Peninsula in the west to the Shumagin Islands in the east. The seaward extent of both the NAP and SAP study areas was the approximate 70-m depth contour (Fig. 1A). The strip transect method developed by Brueggeman et al.1 consisted of a series of transects oriented north- south which were spaced every three minutes of longi- tude throughout the study area. In 1986, surveys were flown in a DeHavilland Twin Otter aircraft equipped with bubble windows at an altitude of 92 m and an airspeed of 185 km/h. Two observers, one on each side of the aircraft, relayed sea otter sighting information to a data recorder seated in the aft section of the aircraft. Sea otter sightings were grouped into three distance intervals spaced at right angles to the transect line: 0.0-0.23 km, 0.23-0.46 km, and 0.46-0.93 km. These distance zones were determined by using a clinometer to place marks on the inside of the bubble windows. Environmental information on sea state, visibility, and glare was recorded throughout the survey. In May 2000 and April 2001, we repeated the survey conducted by Brueggeman et al.1 using similar meth- ods, with the exception that our survey aircraft was an Aero Commander equipped with bubble windows and we grouped sea otter sightings into five distance intervals: 0.0-0.115 km, 0.115-0.23 km, 0.23-0.345 km, 0.345-0.46 km, and 0.46-0.575 km. Coastline survey areas In 1986, Brueggeman et al.1 also surveyed the coastlines of 22 islands on the south side of the Alaska Peninsula quarterly at a distance of 0.46 km from shore, using the same aircraft, altitude, and airspeed as in the off- shore area surveys (Fig. IB). In 1989, DeGange et al.2 surveyed the coastlines of these same islands and the Alaska Peninsula from False Pass to Cape Douglas (Fig. 1C). The 1989 survey was conducted from Bell 206 and Hughes 500 helicopters at a distance of 0.2 km from shore at an altitude of 92 m and an airspeed of 130 km/h. We used similar methods (0.23 km from shore, altitude 92 m, airspeed 185 km/h) to survey the coastlines of these 22 islands and the Alaska Peninsula in April and May 2001. The area of the offshore surveys was adjacent to, but did not overlap, the area of the coastline surveys. Coastline surveys were not conducted in the NAP study area. Offshore survey analyses Prior to the analysis of the 2000-01 offshore survey data, we tested several assumptions made in the 1986 analysis regarding the detectability of sea otters as a function of 1) survey strip width, 2) survey conditions, and 3) time of day. We examined the distribution of sea otter sightings by distance zone using a chi-square analysis to determine the appropriate survey strip width to use 272 Fishery Bulletin 103(2) _Ukolnoi wosnesensk. (^opoP Korovm ^ Andronica Spectacle UiLBig Koniuji -&endfi?-l"lfi KQn'"JI TiJrner *-/V. o £? Simeonof Bird*** £jChernabura -Sanak/ A 0 25 50 100 I I I I I I I I I Kilometers Figure 1 Sea otter {Enhydra lutris) survey areas along the Alaska Peninsula. (Al Offshore areas. (Bl South Alaska Peninsula Islands. (C) Alaska Peninsula coastline. Surveyed areas in (B) and (C) include a 0.46-km zone adjacent to shore. for estimating abundance. We calculated an encounter rate as the number of sea otter groups per km of survey effort and used this rate to examine the effects of time of day and environmental conditions (wave height, and visibility) on detectability of sea otters. At the time of the surveys in 1986, re- searchers had documented a core resting period for sea otters which occurs about mid-day (Garshelis and Garshelis, 1984; Estes, 1977). As a result, Brueggeman et al.1 subset the 1986 data using only effort and observations recorded between 0830 and 1430 hours local sun time for their abundance estimates. Recent studies in- dicate activity patterns for sea otters are strongly linked to sex, age, weather condi- tion, season, and time of day (Gelatt et al., 2002). We tested the assumption that sea otters were more visible during the core resting period using a /-test of the encoun- ter rate for each transect during presumed rest and nonrest periods for the 1986 and the 2000-01 data. We measured the area of the NAP and SAP study areas using a geographic infor- mation system (Arc/Info). Our measure- ments differed from those of Brueggeman et al.,1 presumably because the original researchers had not used an equal-area map projection in their calculations. Like Brueggeman et al.,1 we estimated abun- dance of sea otters in the Alaska Peninsula offshore areas using the modified ratio of means estimator (method I) of Estes and Gilbert (1978). Noting computational er- rors in the original analysis, we recalcu- lated abundance estimates from the origi- nal 1986 data of Brueggeman et al.1 The proportion of sea otters within the survey swath that went undetected by observers was not estimated in either our survey or the surveys of Brueggeman et al.1; there- fore all abundance estimates were biased low to an unknown degree. We computed the proportional change in abundance be- tween survey periods ((Nt2—Nn)INn) as a range, using the minimum and maximum estimates from 1986 as a baseline and assuming no significant difference in the proportion of sea otters detected between surveys. Coastline survey analyses We calculated the area surveyed as the product of the coastline length and the survey strip width and calculated the den- sity of sea otters per km2 surveyed. Once Burn and Doroff: Decline of Enhydra lutris along the Alaska Peninsula 273 again assuming no significant difference in the proportion of sea otters detected between surveys, we computed the propor- tional change in density between survey periods UD,,-Dn)/Dn) of sea otter den- sity at each island within the study area between 2001 and 1986 (Brueggeman et al.1) and each Alaska Peninsula coastline segment between 2001 and 1989 (DeGange et al.2). Results Offshore surveys In 1986, Brueggeman et al.1 flew four sur- veys and an average of 3676 km of transect effort per survey. The majority (599? ) of the 1986 survey effort was conducted in Beau- fort sea state 2 (wind less than 7.4-11.1 km/h, no whitecaps) or less; and 95% of the survey effort was conducted with visibility categorized as good or better. In May 2000 and April 2001, we flew 6334 km of tran- sects and 56% of our effort was conducted in Beaufort sea state 2 or less; 97% of our effort was conducted in visibility catego- rized as good or better. In 1986, sea otter detection probability was not uniform between sighting zones (X2 = 1796, df=2, P<0.0001) and substan- tially more sea otters were observed than expected in the 0.0-0.23 km distance zone (Fig. 2A). As a result, Brueggeman et al. (1988) used only this zone in their cal- culation of sea otter abundance. In our 2000-01 surveys, sea otter detection prob- ability was also not uniform 0.14 (±0.02) n/a 60.5 (±1.9) 557.83 <0.001 0.91 141.23 11.78 Logistic 60.2 (±39.4) 0.44 (±0.05) 6.75 (±0.47) 483.00 <0.001 0.93 47.44 6.83 Female von Bertalanffy 226.2 (±18.6) 0.08 (±0.02) -3.84 (±0.40) — 612.20 <0.001 0.90 150.70 12.19 von Bertalanffy with size-at- birth 202.7 (±10.9) 0.11 (±0.01) n/a 52 1047.19 <0.001 0.88 173.07 12.78 Gompertz 220(G = 1.17±0.4) 0.16 (±0.02) n/a 60.7 (±1.6) 609.09 <0.001 0.90 151.39 12.21 Logistic 62.6 (±3.2) 0.37 (±0.03) 7.62 (±0.43) 572.84 <0.001 0.93 43.82 6.57 1 Asymptotic size for the von Bertalanffy, von logistic model is in kg. 2 t0 is the theoretical age at zero length for th increase in weight begins to decrease. Bertalanffy with size-at-birth, and Gompertz models are in cm, whereas asymptotic size for the ; von Bertalanffy whereas tu for the logistic model represents time at which the absolute rate of 286 Fishery Bulletin 103(2) | 8 8 (LUO) L|l6u3| JjJOJ a 9 r (6>|) lL)6ia/\A " >i T3 c*_ - 3 M o o J/3 +j cfi en a) M C8 O *J — J/3 t^ c (WO) I|l6u9| ijJOJ (wo) L|i6ua| jjjoj M „3 C a - 0) s J3 u Eh - Carlson and Baremore: Growth dynamics of Carcharhinus brevipinna 287 o CO - 00 o A CO re T3 ID \ o 1 bo \ ' 92 I o CO • tb \ \ co \ o \ o N „ ■ \ oob Q TJ O 0) CD n 1 ° u .2 2 x o 5 \ ■ OJ \ o - c\i .£3 cu t8 \ V °§ i> o , N 0 \ o 0 0 \ *> *S W \ • a \ o ■ o ■a ^ £ o CD ~— 15 3 5 \ en \ c o ^ " \ c o o \ en "d ■ cd o — ' c c CO -a « CO \ \ S \ \ O +J CO V o ■ cd V °° o " CD c o > "aJ.1 o ° .°\ll 1 ^ £ ° • °- V; " Tt ..X o CJ m ." CD CO CC ° V° o -V o — i -a "\ 1 -"X o CD' ' V o ' d > _>; m « 1 « 8 8 8 § co 8 8 8 8 1 1 1 1 T 1 1 ■""■ OQQQQQOC h~ co Cn ^ co oj r- E 0) LL (wo) qifjuai >(joj o to o (6>i) imBism CD i o r £ o CD < $ ° ° s> O "a iZ ^-g £ o \ cd 1 ■ . \ o \ o x o c CO .. \ 0 o N „.\ 0 ■ cj CO \ cd \ O O \ O 0 o o Q. o o \ o o o o iS \ o E \ o ^- — CO \ 03 o \ " CO s <« c \ a \ J5 ^ o \ \ +j 03 > Y oo o Y o 5h J^ CD Cfi \ cd \ " CO CQ n "JL".l o "JL'.l o - CD > J • °°v ^ . ...V. " ■** tfcH . ■ C cm 0 Wo o ooo W. o — .—i \ c\J \ ' c\j "A "A CD -^ PQ ,2 °\° o °°y o S E § SS^woffiS^ 8 88§Sj888" ? H c° (wo) qi6u8[ ^joj (luo) L|l6u8| ^JOJ 288 Fishery Bulletin 103(2) Table 4 Mean size-at- age (cm FL) for male and female spinner sharks (C arch a i hi mis brevispmna I SD = standard deviation. Age (yrl 0.0 0.5 1.5 2.5 3.5 4.5 5.5 6.5 7.5 8.5 9.5 10.5 11.5 12.5 13.5 14.5 15.5 16.5 17.5 Male Size 60.7 64.1 81.2 84.6 99.9 105.8 115.5 146.3 138.9 154.2 165.3 165.9 — 176.3 178.0 — — — — SD 5.3 5.9 11.4 13.8 17.9 11.0 4.2 23.4 13.9 13.4 3.0 — 3.1 _ _ _ _ n 29 1 12 8 21 7 7 3 6 10 7 2 — 3 1 — — — — Female Size 59.0 69.3 84.4 86.9 106.9 106.9 117.1 116.2 — 136.3 160.8 173.7 158.3 — 164.7 165.5 182.0 — 184.0 SD 4.1 11.5 2.9 8.7 13.3 13.9 19.1 20.7 — 19.6 16.4 14.8 11.2 — 21.7 21.9 — — 1.4 n 42 7 12 18 11 14 6 5 — 6 6 3 3 — 4 2 1 — 2 expected maximum size, resulting in an inflated asymp- tote and low growth coefficient. Branstetter and Stiles (1987) also encountered this problem with bull sharks (Carcharhinus leucas) but rather than fit an alterna- tive growth model, those authors hand-fitted a curve through the upper data points. Results such as these may seriously bias estimates of k and any resulting population models because several indirect estimates of natural mortality (M) and longevity rely heavily on accurate estimates of k from a growth model (Fabens, 1965; Pauly, 1980; Chen and Watanabe, 1989; Jensen, 1996). For example, the method of Jensen (1996) for estimating M yields values ranging from 0.05/yr (with results from the von Bertalanffy model) to 0.23/yr (with results from the Gompertz model). Similarly, theoretical longevity estimates determined by the method of Fabens (1965) are 115.5 years and 21.6 years from the von Ber- talanffy model and the Gompertz model, respectively. In general, our estimates of age and growth for fe- male spinner sharks from the von Bertalanffy model were similar to those reported by Allen and Wintner (2002) for spinner sharks collected off South Africa. Growth coefficients in their study were about 0.13/yr, Lx was 250 cm FL, and observed longevity for females was up to 19+ years. Branstetter (1987), in his study on sharks collected in the Gulf of Mexico, reported an observed longevity up to 11+ years (combined sexes) and growth coefficients of about 0.21/yr. Because differences in life history traits (e.g., growth rates, size and age at maturity) between populations of blacktip and bull sharks from South Africa and United States waters have been proposed (Wintner and Cliff, 1995; Wintner et al., 2002, respectively), results from our study for spinner shark may be expected to be more similar to those of Branstetter (1987) rather than those of Allen and Wintner (2002). Although techniques (e.g., counting winter bands on sagittal vertebral sections) in Brans- tetter (1987) were similar to ours, the differences are likely a result of low sample size in the earlier study. The index of average percent error (IAPE) in aging was at the higher end of the range of estimates pro- vided in other studies that also used sagittal sections for aging. Values have been reported as low as 3.0% for the oceanic whitetip shark (Carcharhinus longimanus) (Lessa et al., 1999), and up to 13.0% for the black- tip shark (Carcharhinus limbatus) (Wintner and Cliff, 1995). Although IAPE indices are most commonly used to evaluate precision among age determinations, IAPE does not test for systematic differences and does not dis- tinguish all sources of variation (Hoenig et al., 1995). In addition, comparing IAPE values among studies may not be valid unless the study species is the same and from the same geographic area (Cailliet and Goldman 2004). Although bands were readily discernible in most sam- ples, the inexperience of one of the authors (reader 2) in reading and counting vertebral bands likely led to the higher IAPE and systematic bias. Generally, most systematic bias is a shift to increasing or decreasing counts with age (Morison et al. 1998), yet the bias in this study was the result of reader 2 consistently over aging sharks from the final agreed age regardless of the band count of the sample. Percent agreement was simi- lar for samples above 115 cm FL as it was for samples below this size. Although a reference collection was aged by reader 2 prior to beginning this study, finely honed skills through experience are key elements in the technique of aging. The trend in marginal increment analysis indicated that band formation occurs once a year during winter months — a result common to most studies where rela- tive marginal increment analysis is used for carcharhi- nid sharks (e.g., Natanson et al., 1995; Carlson et al., 1999; Carlson et al., 2003). However, high variance in marginal increment analysis (MIR) within each month resulted in months not being statistically different, which is a widespread occurrence when using this meth- od. Marginal increment analysis has been criticized as one of the most abused methods for validation of band formation (Campana, 2001). Problems with differentiat- ing bands on the vertebral edge and application to older age classes may provide misleading results (Campana, Carlson and Baremore: Growth dynamics of Carcharhinus brevipmna 289 2001). Other methods have been used recently to report yearly band formation in sharks, including oxytetra- cycline marking (Simpfendorfer et al., 2002; Skomal and Natanson, 2003; Driggers et al., 2004) and bomb radiocarbon methods (Campana et al., 2002). However, validation exists for relatively few elasmobranch species (Cortes, 2000). Two-phase growth models may be more appropriate for describing the growth of sharks, especially those that are longer lived. Soriano et al. (1992) developed a biphasic growth model which they applied to the long- lived Nile perch (Lates niloticus) to better describe their change in growth from zooplanktivores as juveniles to piscivores as adults. Growth by sharks could be regard- ed as being found in two phases: a rapid juvenile growth followed by a slower adult growth. From a bioenergetic perspective, this would follow a change from energy devoted to growth to energy devoted to reproduction. The logistic model could be regarded as a two-phase model and may help to describe this change. The shift from juvenile to adult would correspond to the inflection point (fu) of the curve, which approximates biological age-at-maturity. In spinner sharks, age at maturity was reported to be about 6-7 years for males and 7-8 years for females (Branstetter, 1987). This estimate of age-at-maturity is similar to the inflection points from our logistic model of 6.75 and 7.62 years for males and females, respectively. Although each species should be evaluated separately, future studies should investigate the use of two-phase models to provide a more accurate description of the growth of elasmobranchs. There have been few other examples of fitting alter- native growth models to size-at-age data when results from the von Bertalanffy model were biologically in- correct or when models did not fit the data well. The present study represents the first attempt to dp so for a species of shark. Comparison of age and growth models by Mollet et al. (2002) and Neer and Cailliet (2001) for two species of rays revealed that the Gompertz model best described their respective data although all models they tested fitted the data fairly well. For pelagic sting- ray (Dasyatis violacea) the Gompertz model predicted a more reasonable size-at-birth and growth rate than the von Bertalanffy growth model (Mollet et al., 2002). Neer and Cailliet (2001) reported a slightly better statistical fit for the Pacific electric ray {Torpedo californica) when using the Gompertz model. However, because the differ- ence in model parameters was negligible, results were reported only for the von Bertalanffy model. The von Bertalanffy growth model is still the most common model used to describe growth in fisheries literature, despite criticism by Roff (1980) who recom- mended its retirement. As pointed out by Roff (1980), the choice of using another equation should be deter- mined by the variables that are being investigated and the results that are produced by the equation; for exam- ple, if the results appear to be biologically unrealistic. Our analysis of the growth of the spinner shark clearly demonstrates the value of this approach. Use of the von Bertalanffy growth model should continue because it permits comparison of growth curves to information al- ready published and in some cases adequately describes the growth of a given organism. However, the variety of statistical techniques and quality of each study make comparisons of von Bertalanffy growth curves between different populations difficult and results should be in- terpreted with caution regardless of what growth model is used (Roff, 1980). Acknowledgments We thank Enric Cortes, Pete Sheridan (NOAA Fisheries, Panama City Laboratory), and Miguel Arraya (Universi- dad Arturo Prat, Chile) for providing comments on ear- lier versions of this manuscript. Ken Goldman (Jackson State University) was especially helpful in discussion on precision and bias in age estimation, Miguel Arraya on the validity of the comparison of growth models, and Henry Mollet (Monterey Bay Aquarium) with the Gomp- ertz model. Many different laboratories and institutions aided with the collection of vertebrae. George Burgess and Matt Callahan (University of Florida) provided samples from the directed shark longline fishery. Lisa Natanson (NOAA Fisheries, Narragansett Laboratory) obtained samples during their longline surveys from the U.S. south Atlantic Ocean. Observers Armando de ron Santiago, Carl Greene, Matt Rayl, Bill Habich, Mike Farni, Jacques Hill, and Jeff Pulver collected samples from the directed shark gillnet fishery. Mark Grace and Lisa Jones (NOAA Fisheries, Pascagoula Laboratory) provided samples from fishery-independent longline surveys. We also thank Linda Lombardi, Lori Hale, and numerous interns who assisted with the cleaning and processing of vertebrae samples. Literature cited Allen, B. R. and S. P. Wintner. 2002. Age and growth of the spinner shark Carcharhi- nus brevipinna (Miiller and Henle, 1839) off the Kwa- zulu-Natal coast, South Africa. S. Afr. J. Mar. Sci. 24: 1-8. Beamish, R. J., and D. A. Fournier. 1981. A method for comparing the precision of a set of age determinations. Can. J. Fish. Aquat. Sci. 38: 982-983. Branstetter, S. 1987. Age and growth estimates for blacktip, Carcharhi- nus limbatus, and spinner, C. brevipinna, sharks from the northwestern Gulf of Mexico. Copeia 1987:964-974. Branstetter, S., and J. A. Musick. 1994. Age and growth estimates for the sand tiger in the northwestern Atlantic Ocean. Trans. Am. Fish. Soc. 123:242-254. Branstetter, S., and R. Stiles. 1987. Age and growth estimates of the bull shark, Car- charhinus leucas, from the northern Gulf of Mexico. En- viron. Biol. Fish. 20:169-181. Cailliet, G. M. 1990. Elasmobranch age determination and verifica- 290 Fishery Bulletin 103(2) tion: an updated review. In Elasmobranchs as living resources: advances in biology, ecology, systematic and the status of the fisheries (H. L. Pratt Jr., S. H. Gruber and T. Taniuchi. eds.), p. 157-165. NOAA Tech. Rep. NMFS 90. Cailliet, G. M.. and K. J. Goldman. 2004. Age determination and validation in chondrich- thyan fishes. In The biology of sharks and their rela- tives (J. Carrier, J. A. Musick and M. Heithaus, eds.). p. 399-447. CRC Press, Boca Raton, FL. Campana, S. E. 2001. Accuracy, precision, and quality control in age determination, including a review of the use and abuse of age validation methods. J. Fish. Biol. 59:197-242. Campana, S. E., L. J. Natanson, and S. Myklevoll. 2002. Bomb dating and age determination of large pelagic sharks. Can. J. Fish. Aquat. Sci. 59(31:450-455. Carlson, J. K., E. Cortes, and D. M. Bethea. 2003. Life history and population dynamics of the finetooth shark, Carcharhinus isodon, in the north- eastern Gulf of Mexico. Fish. Bull. 101:281-292. Carlson, J. K., E. Cortes, and A. G. Johnson. 1999. Age and growth of the blacknose shark, Carcharhi- nus acronotus, in the eastern Gulf of Mexico. Copeia 1999:684-691. Castro, J. I. 1993. The shark nurseries of Bulls bay, South Carolina, with a review of the shark nurseries of the southeast- ern coast of the United States. Environ. Biol. Fish. 38:37-48. Chen, S. B., and S. Watanabe. 1989. Age dependence of natural mortality coefficient in fish population dynamics. Nippon Suisan Gakkaishi 55:205-208. Compagno, L. J. V. 1984. FAO species catalogue: sharks of the world. An annotated and illustrated catalogue of shark species known to date. FAO Fisheries Synopsis (125), vol. 4, part 1: Hexanchiformes to Lamniformes, 249 p. FAO, Rome. Cortes, E. 1997. A critical review of methods of studying fish feed- ing based on analysis of stomach contents: application to elasmobranch fishes. Can. J. Fish. Aquat. Sci. 54:726-738. 2000. Life history patterns and correlations in sharks. Rev. Fish. Sci. 8:299-344. Driggers, W. B., J. K. Carlson, D. Oakley, G Ulrich, B. Cullum, and J. M. Dean. 2004. Age and growth of the blacknose shark, Car- charhinus acronotus, in the western North Atlantic Ocean with comments on regional variation in growth rates. Environ. Biol. Fish. 71:171-178. Fabens, A. J. 1965. Properties and fitting of the von Bertalanffy growth curve. Growth 29:265-289. Goldman, K. J. 2002. Aspects of age, growth, demographics, and thermal biology of two Lamniform shark species. Ph.D. diss., 219 p. Virginia Institute of Marine Science, Glocester Point, VA Goosen, A. J. J., and M. J. Smale. 1997. A preliminary study of the age and growth of the smoothhound shark Mustelus mustelus (Triakidae). S. Afr. J. Mar. Sci. 18:85-91. Grace, M., and T. Henwood. 1997. Assessment of the distribution and abundance of coastal sharks in the U.S. Gulf of Mexico and eastern sea- board, 1995 and 1996. Mar. Fish. Rev. 59(41:23-32. Hoenig, J. M., M. J. Morgan, and C. A. Brown. 1995. Analysing differences between two age determi- nation methods by tests of symmetry. Can. J. Fish. Aquat. Sci. 52:364-368. Jensen, A. L. 1996. Beverton and Holt life history invariants result from optimal trade-off of reproduction and survival. Can. J. Fish. Aquat. Sci. 53:820-822. Knight, W. 1968. Asymptotic growth: an example of non-sense dis- guised as mathematics. J. Fish. Res. Board Can. 25: 1303-1307. Lessa, R., F M. Santana, and R. Paglerani. 1999. Age, growth, and stock structure of the oceanic whitetip, Carcharhinus longimanus, from the southwest- ern equatorial Atlantic. Fish. Res. 42:21-30. Mollet, H. F, J. M. Ezcurra. and J. B. O'Sullivan. 2002. Captive biology of the pelagic stingray, Dasyatis violacea (Bonaparte, 1832). Mar. Freshw. Res. 53: 531-541. Morison, A. L., S. G. Robertson, and D. G. Smith. 1998. An integrated system for production fish aging: image analysis and quality assurance. N. Am. J. Fish. Manag. 18:587-598. Natanson, L. J., J. G. Casey, and N. E. Kohler. 1995. Age and growth estimates for the dusky shark, Carcharhinus obscurus, in the western North Atlantic Ocean. Fish. Bull. 193:116-126. Neer, J. A., and G. M. Cailliet. 2001. Aspects of the life history of the Pacific electric ray, Torpedo califonuca (Ayres). Copeia 2001:842-847. Neter, J., W. Wasserman, and M. H. Kutner. 1990. Applied linear statistical models. 1408 p. Richard D. Irwin, Inc. Boston. MA. NMFS (National Marine Fisheries Service). 1993. Fishery management plan for sharks of the Atlantic Ocean, 167 p. Office of Fisheries Conservation and Management, National Marine Fisheries Service, NOAA, 1334 East-West Highway, Silver Springs, MD. Pauly, D. 1980. On the interrelationships between natural mor- tality, growth parameters, and mean environmental temperature in 175 fish stocks. J. Cons. Int. Explor. Mer 39:175-192. Ricker, W. E. 1975. Computation and interpretation of biological statis- tics offish populations. Bull. Fish. Res. Board Canada 191:1-382. 1979. Growth rates and models. In Fish physiology, vol. VIII: Bioenergetics and growth (W. S. Hoar, D. J. Randall, and J. R. Brett, eds). p. 677-743. Academic Press. New York, NY. Roff, D. A. 1980. A motion for the retirement of the von Bertalanffy function. Can. J. Fish. Aquat. Sci. 37:127-129. Simpfendorfer, C. A., R. B. McAuley, J. Chidlow, and P. Unsworth. 2002. Validated age and growth of the dusky shark, Carcharhinus obscurus, from western Australian waters. Mar. Freshw. Res. 53:567-573. Skomal, G B., and L. J. Natanson. 2003. Age and growth of the blue shark (Prionance Carlson and Baremore: Growth dynamics of Carcharhmus brevipinna 291 glauca) in the North Atlantic Ocean. Fish. Bull. 101:627-639. Soriano, M., J. Moreau, J. M. Hoenig, and D. Pauly. 1992. New functions for the analysis of two-phase growth of juvenile and adult fishes, with application to Nile perch. Trans. Am. Fish. Soc. 121:486-493. Van Dykhuizen, G., and H. F. Mollet. 1992. Growth, age estimation, and feeding of cap- tive sevengill sharks, Notorynchus cepedianus, at the Monterey Bay Aquarium. Aust. J. Mar. Freshw. Res. 43:297-318. Vaughan, D. S., and M. L. Burton. 1994. Estimation of von Bertalanffy growth parameters in the presence of size-selective mortality: a simulated example with red grouper. Trans. Am. Fish. Soc. 123:1-8. von Bertalanffy, L. 1938. A quantitative theory of organic growth (inquiries on growth laws. II). Human Biology 10:181-213. Wintner, S. P., and G. Cliff. 1995. Age and growth determination of the hlacktip shark, Carcharhinus limbatus, from the east coast of South Africa. Fish. Bull. 94:135-144. Wintner, S. P., S. F. J Dudley, N. Kistnasamy, and B. Everett. 2002. Age and growth estimates for the Zambezi shark, Carcharhinus leucas, from the east coast of South Africa. Mar. Freshw. Res. 53:557-566. Zar, J. H. 1984. Biostatistical analysis, 718 p. Prentice Hall, Inc. Englewood Cliffs, NJ. 292 Abstract— Data recovered from 11 popup satellite archival tags and 3 surgically implanted archival tags were used to analyze the movement patterns of juvenile northern bluefin tuna (Thunnus thynnus orientalist in the eastern Pacific. The light sen- sors on archival and pop-up satellite- transmitting archival tags (PSATs) provide data on the time of sunrise and sunset, allowing the calculation of an approximate geographic position of the animal. Light-based estimates of longitude are relatively robust but latitude estimates are prone to large degrees of error, particularly near the times of the equinoxes and when the tag is at low latitudes. Estimat- ing latitude remains a problem for researchers using light-based geoloca- tion algorithms and it has been sug- gested that sea surface temperature data from satellites may be a useful tool for refining latitude estimates. Tag data from bluefin tuna were sub- jected to a newly developed algorithm, called "PSAT Tracker," which auto- matically matches sea surface tem- perature data from the tags with sea surface temperatures recorded by sat- ellites. The results of this algorithm compared favorably to the estimates of latitude calculated with the light- based algorithms and allowed for estimation of fish positions during times of the year when the light- based algorithms failed. Three near one-year tracks produced by PSAT tracker showed that the fish range from the California-Oregon border to southern Baja California, Mexico, and that the majority of time is spent off the coast of central Baja Mexico. A seasonal movement pattern was evident; the fish spend winter and spring off central Baja California, and summer through fall is spent moving northward to Oregon and returning to Baja California. Tracking Pacific bluefin tuna (Thunnus thynnus orientalis) in the northeastern Pacific with an automated algorithm that estimates latitude by matching sea-surface-temperature data from satellites with temperature data from tags on fish Michael L Domeier Pfleger Institute ol Environmental Research 901 B Pier View Way Oceanside, California 92054 E-mail address: Domeieng'cs com Dale Kiefer System Science Applications Inc. POBox 1589 Pacific Palisades, California 90272 Nicole Nasby-Lucas Adam Wagschal Pfleger Institute of Environmental Research 901 B Pier View Way Oceanside, California 92054 Frank O'Brien System Science Applications Inc. POBox 1589 Pacific Palisades, California 90272 Manuscript submitted 11 June 2004 to the Scientific Editor's Office. Manuscript approved for publication 21 December 2004 by the Scientific Editor. Fish. Bull. 103:292-306 (2005). Current theories indicate the presence of a single stock of northern Pacific bluefin tuna (Thunnus thynnus orien- talis) in the Pacific Ocean. Spawning adults have been recorded only from the western Pacific ( Yamanaka et al., 1963; Yabe et al., 1966; Okiyama, 1974; Okiyama and Yamamoto, 1979; Nishikawa et al., 1985) but resulting offspring are known to either inhabit the western Pacific or to travel to the eastern Pacific (Sund et al., 1981; Bay- liff, 1994; Itoh et al., 2003a) where they remain for an undetermined amount of time. Although it is believed that only a small fraction of the popu- lation migrates to the eastern Pacific, these fish are the basis for a fishery that occurs from May through Octo- ber. A recent study has documented the migration of an archival-tagged juvenile northern Pacific bluefin tuna from the western Pacific to the east- ern Pacific in about two months, where it remained for eight months before being recaptured (Itoh, et al., 2003a). Conventional tagging studies have shown that Pacific bluefin tuna in the eastern Pacific eventually return to the western Pacific where they are believed to remain as adults (Sund et al., 1981; Bayliff, 1994). We provide this cursory summary merely as an introduction to our work, deferring the known details of Pacific bluefin biology to the excellent reviews that have been previously published ( Bay- liff, 1980. 1994; Sund et al., 1981). Work presented in the present study describes the use of electronic tags (pop-up satellite-transmitting archi- val tags and archival tags obtained from fish) and a newly developed sea surface temperature (SST) based geo- Domeier et al.: Tracking Thunnus thynnus orientalis with the aid of an automated algorithm 293 location algorithm to further our understanding of blue- fin tuna movements in the eastern Pacific. The light sensors on archival and pop-up satellite tags provide data on the time of sunrise and sunset, allowing one to calculate the approximate geographic position of an animal (Delong et al., 1992; Wilson et al., 1992; Hill, 1994; Bowditch, 1995; Sobel, 1995; Welch and Eveson, 1999; Hill and Braun, 2001; Metcalfe, 2001;Smith and Goodman;1 Gunn et al.2). The accu- racy of the light-based geolocation estimates have been studied under controlled conditions (tags tethered to a moored buoy) and field conditions (tags attached to fish at a known location). Locations from tethered tags have been reported to be accurate to within ±0.2-0.9° in longitude and ±0.6-4.4° in latitude (Welch and Eve- son, 1999, 2001; Musyl et al., 2001). Tagged tuna have provided light-based geolocation estimates within ±0.5° of longitude and ±1.5-2.0° latitude (means) of known locations (Schaefer and Fuller, 2002; Gunn et al.1). Light-based estimates are not precise and comparing studies that have examined the accuracy of this method is complicated by differences in tag hardware and geo- location algorithms used by different researchers. Other physical and biological factors complicate the issue fur- ther. Day length is not a good predictor of latitude dur- ing the spring and fall equinox, therefore estimates of latitude at times surrounding the equinox contain more error than at other times of the year (Hill and Braun, 2001). Latitude estimates are also more prone to error the closer the animal is to the equator (Hill and Braun, 2001). Additional errors can be introduced into esti- mates of both latitude and longitude by the behavior of the tagged animal (e.g., diving), bio-fouling of the tag, cloud cover, and wave action (Metcalfe, 2001). Poor resolution of latitude estimates continues to be a problem for researchers using light-based geolocation algorithms. Under ideal theoretical conditions the vari- ability in latitude error cannot be less than 0.7° and the expected variability in longitude will be a constant 0.32° (Hill and Braun, 2001). Sibert et al. (2003) developed an algorithm that applies a Kalman filter to light-based geolocation estimates in an attempt to reduce the error of these estimates. Although this approach smoothes data, it does not incorporate external data (data not collected by the tag) and therefore is still affected by errors inherent in the use of light-based geolocation es- 1 Smith, P., and D. Goodman. 1986. Determining fish movements from an "archival" tag: precision of geographi- cal positions made from a time series of swimming, tem- perature and depth. NOAA. Tech. Memo. NMFS-SWFC-60, 13 p. Southwest Fisheries Science Center, La Jolla, CA 92038. 2 Gunn, J. S„ T. W. Polacheck, T. L. O. Davis, M. Sherlock, and A. Betlehem. 1994. The development and use of archival tags for studying the migration, behavior and physiology of southern bluefin tuna, with an assessment of the potential for transfer of the technology to groundfish research. In Proceedings of ICES mini-symposium on fish migration, 23 p. International Council for the Exploration of the Sea, Palaegade 2-4, DK-1261 Copenhagen K. Denmark. timates of latitude. It has been suggested that sea-sur- face-temperature (SST) and bathymetry data be used to refine light-based geolocation estimates (Block et al., 2001). These techniques are particularly useful when there is a north-to-south gradient of bathymetry or SST. The use of bathymetry to refine latitude requires an as- sumption that maximum diving depth is limited by the bottom depth; certainly this assumption introduces a new source of error. In addition, for animals that move off the continental shelf, bathymetry would be useless. The use of SST or bathymetry data to refine latitude necessitates the arduous task of matching tag data with another source of data. It was our opinion that the accuracy of tracking ma- rine animals could be improved through the develop- ment of an algorithm that automatically resolved lati- tude estimates by matching SST measurements from the tag to those taken from satellites. Here we present such an algorithm; one that was designed to operate in a geographic information system (GIS) environment, allowing for rapid analysis and display of archival and PSAT tag data. We demonstrate the algorithm and its product through the analyses of data we collected from Pacific bluefin tuna tagged in the eastern Pacific. Materials and methods Tagging in the field Pacific bluefin tuna were captured on rod and reel from a recreational fishing vessel by using live bait and circle hooks. Fishing took place 123 nmi southwest, 86 nmi southwest, and 178 nmi south of San Diego in years 2000, 2001, and 2002, respectively. Fish were lifted into the boat with a vinyl sling and then placed on a soft mat, eyes were covered with a cloth, and the gills irrigated with seawater. The fish were then measured (fork length and girth), tagged, and immediately released. Sixteen fish were tagged with Wildlife Computers Inc. (Redmond, WA) pop-up satellite archival tags (PSATs), one fish was tagged with a Microwave Telemetry Inc.! Columbia, MD) PTT-100 PSAT, and seventeen fish were tagged with Lotek Wireless Inc. (Newmarket, Ontario) LTD2310 nontransmitting archival tags. The two types of PSATs either provided data once an hour (depth, water tempera- ture, light level [Microwave Telemetry, Inc.]) or sum- marized data that had been collected every two minutes (Wildlife Computers, Inc.) — the difference being an arti- fact of the two tag manufacturers. The Lotek archival tags provided us with data every two minutes detailing the swimming depth, water temperature, internal fish temperature, and light level. Pressure sensor drift was adjusted by the tag manufacturers' software for PSAT tags and in the laboratory for the Lotek tags. The PSAT tags were rigged with 300-lb monofilament leaders and a nylon dart. In 2000 and 2001 the dart was a "bluefin-type" provided by Eric Prince (NMFS- SEFSC); in 2002 a Pfleger Institute of Environmental Research (PIER) "umbrella" dart was used (Fig. 1). 294 Fishery Bulletin 103(2) Figure 1 PIER umbrella dart used for external attachment of tags. Each style of dart was inserted through the midline of the fish at the base of the second dorsal fin according to the method of Block et al. (1998). Archival tags were surgically implanted either in the dorsal musculature below the first dorsal fin (when fork length was >110 cm) or into the peritoneal cavity (when fork length was <110 cm). The dorsal musculature im- plant was performed by making a 1-cm incision 3-5 cm below the first dorsal fin. A cold-sterilized trocar (14 mm diameter) was then inserted into the muscle, to a depth of 13-14 cm, within a plane parallel to the pterygiophores but angled 45 degrees to the anterior. The trocar was then removed and the tag was inserted so that the light stalk was angled toward the tail. The incision was then closed with a monocryl suture mate- rial. This method was similar to that used by Musyl et al. (2003). Interperitoneal implants were done according to the method of Block et al. (1998). PSAT Tracker algorithm and analysis system We have developed an automated system, called the PSAT Tracker Information System (PTIS), to improve the accuracy and minimize the subjectivity and tedium of matching data from different sources (tag and satel- lite). It is an application of the Environmental Analysis System (EASy) (System Science Applications, Redondo Beach, CA) software that is specifically designed for handling four-dimensional information (latitude, lon- gitude, depth, and time). We describe the system in terms of three processes; importing tag data and satel- lite imagery, calculation of the optimal path of the tag, and dynamic display of the path and associated tag information. Importing tag data and setting parameters The PSAT tracker information system was designed to support data formats of three tag manufacturers: Wildlife Computers, Microwave Telemetry, and Lotek. All three tag formats are imported into FIS and stored in a universal relational database format for process- ing. Key parameters used in the calculation of tracks include time and position of tag deployment, time and position of tag recovery, light-based estimates of lon- gitude (provided by tag manufacturers), maximum swimming speed of the tagged fish (estimated and determined by the user), and a bracketed range of latitude within which the program will search for SST matches. Processing involves the temporal matching of SST as recorded by the tag with that measured from satellite imagery. It is important to note that the PTIS user-defined latitude bracket is unrelated to the light- based latitude estimates provided by the tag manufac- turers; instead, it is simply a range set by the user to include all possible movement of the animal during the tag deployment. However, longitude estimates are tied to the tag manufacturers' light-based estimates; the user has the option of tying PTIS position estimates directly to the light-based estimates or allowing the algorithm to search a specified distance on either side of the light-based estimate. For this study the maximum fish velocity was set at 4 knots. This was meant to be an inclusive rather than an exclusive value, broadening the range PSAT Track- er could search for SST matches. SST matches were also constrained to remain within ±20 nautical miles (±0.33°) of the manufacturers' light-based estimates of longitude, based upon the observance by Hill and Braun Domeier et al.: Tracking Thunnus thynnus onentalis with the aid of an automated algorithm 295 Table 1 Resolution of sea-surface-temperature data from s atellites and tags (ac vanced ver\ high resolution radiometer [AVHRR1, moder- ate resolution spectroradiometer IMODIS1, multichannel sea surface temperature algorithm [MCSST], Source Accuracy (+C) Spatial scale kmi Temporal scale Availability AVHRR pathfinder 0.3-0.5 9 Daily 1985-present AVHRR pathfinder 0.3-0.5 9 8-day composite 1985-present MODIS 0.3 4.6 Daily Oct 2000-present MODIS 0.3 4.6 8-day composite Oct 2000-present MCSST (Miami) 0.5-0.7 18 Weekly composite 1981-Feb 2001 MCSST (NAVOCEANO) 0.5-0.7 18 Weekly composite Sep 2001-present Wildlife computer tag 0.05 — 1-12/day — Microwave telemetry tag 0.17 — 60 minutes — Lotek 2300 tag 0.1 — 2 minutes — (2001) that light-based longitude estimates have a year round constant error of ±0.32 degrees. Satellite imagery, temperature sensors, and land mask The PSAT Tracker code provides an interface to auto- matically download, georeference, and display SST imag- ery. As many as three different types of imagery can be layered and prioritized to produce a collage of imagery for processing and display. Higher priority layers are searched first for SST matches before "drilling down" to lower layers. The sources and types of available SST data are numerous and have varied over the time frame of this study; different sensors and algorithms produced data of differing spatial and temporal resolution or accu- racy (Table 1). To maximize the quality of the latitude estimates produced by the PSAT Tracker algorithm, we substituted better SST data as it became available. For this study SST imagery was prioritized as follows: 1) advanced very high resolution radiometer (AVHRR) or moderate resolution spectroradiometer (MODIS) daily data, 2) AVHRR or MODIS weekly data, and 3) multi- channel sea surface temperature algorithm (MCSST) weekly data. The MCSST algorithm is a weekly (or 8- day) composite that is most helpful in analyzing regions of frequent cloud cover; this algorithm was applied by the University of Miami (Miami) from 1981 through Febru- ary 2000 and has been applied by the Naval Oceano- graphic Office (NAVOCEANO) since September 2001. The MCSST algorithm provides a near complete picture of SST data for the study area; although AVHRR and MODIS data are higher resolution and more accurate. The difference in the resolution and accuracy of tem- perature sensors on the tags verses those on the satel- lites (Table 1) are worth mentioning. The accuracy of the satellite SST data, particularly for MCSST/NAV- OCEANO. is the limiting factor when attempting to match tag data to satellite data. The degree to which the satellite data and tag data must match can be set by the user in PSAT Tracker; for this study it was set between the limit of MODIS and NAVOCEANO resolu- tion (0.4°C). There is a fourth layer that is superimposed upon the imagery. This is a land mask that is used to eliminate placing a tag on land and to insure that tags move around land barriers rather than across them. Computation of the track A detailed mathematical description of the computation for the best track would take more space than is avail- able. Instead, we present a more general description of the algorithm and its logic, consisting of the following five steps that are summarized below and then subse- quently described in detail. 1 Define the daily search area found within satellite SST imagery. 2 Define appropriate tag data (termed selection set) to match to satellite SST values found within the daily search area. 3 Select candidate points within each daily search area that provide the best match to the temperatures found in the selection set. The cost of each candidate point is largely determined by the difference between the tag and satellite SST values. 4 Calculate the cost for all possible steps, called arcs, between pairs of candidate points of adjacent daily search areas. The cost of each step is a function of the length of the arc that connects adjacent candidate points (the greater the distance, the greater the cost) and the cost of each individual candidate point (see step 3). 5 Sum the costs of all tracks and identifying the track with the lowest cost. Step I: Defining the daily search area A daily search area is defined by the tag manufacturers' light-based solution for longitude, a user defined bracket for lati- tude and the value entered for maximum swimming 296 Fishery Bulletin 103(2) northern limit of habitat search lines for search area t(1) reference longitude for search area t(1) reference longitude and parallel lines for search area t(2) reference longitude and parallel lines for search area t(3) southern limit of habitat 3rd arc defining northern and southern extent of search area t(3) Figure 2 Definition of terms used to describe the PSAT Tracker algorithm. A search area is a region in a satellite thermal image where a search is conducted for pixels whose temperature values match those recorded by the tag at that time and when it is at the surface. The search area consists of a reference longitude line, defined by the daily calculation of latitude provided by the manufacturer's processed data record and par- allel search lines that provide a hedge on this determination. The search area is uniquely defined by the time at which this calculation was determined. The northern and southern bounds of the search area are determined by either the habitat range or the maximum distance that the tagged fish can swim during each time step. Those pixels underlying the reference and search line, whose temperature best match the temperatures of the selection set of points from the tag, are chosen as candidate points. One candidate point from each search area will eventually define the best track. speed of the fish. The latitudinal bounds of the daily search area are constrained in two ways, by the known (or unknown) bounds of the fish's habitat and by its maximum swimming speed. The northern and south- ern bounds of the habitat are entered by the user, and no areas are searched that are beyond these latitudes. These values are meant to be inclusive and can be determined from the literature or estimated by using latitude values provided by light-based geolocation algo- rithms. These bounds are set prior to processing and do not change throughout the processing; in this study the latitude search area was restricted to waters between 15 and 50 degrees north. Each search area is centered on the light-based lon- gitude estimate (termed the reference longitude). PSAT Tracker does not search every pixel of SST data for matches, but instead searches along parallel lines of longitude on either side of, and including, the reference longitude. These lines, termed search lines, are spaced at equal distances from the reference longitude (Fig. 2). The user establishes the extent to which PSAT Tracker searches to the east and west of the reference longitude by choosing the number of search lines as well as their distance of separation. In this study four search lines were drawn on either side of the reference longitude; these parallels were drawn 5 nmi apart resulting in a 40 nmi wide daily search area. We refer to each search ac- cording to the time at which the reference longitude was determined, t(i) (where t is the time for which the refer- ence longitude was determined and ; is the index for the sequence of daily search areas in the time series). The maximum swimming speed of the fish can also constrain the latitudinal bounds of a daily search area. The farthest a fish can swim in a given time interval is simply the product of its maximum swimming speed and the length of the time interval. Thus, all possible posi- Domeier et al.: Tracking Thunnus thynnus onentalis with the aid of an automated algorithm 297 tions that a fish can occupy when swimming in a fixed direction from the starting point of a track is the locus of points forming a circle whose center is at the starting point and whose radius is the product of its maximum swimming speed and the length of the time interval. Likewise, the farthest positions from which a fish can swim in given direction and reach the end point of the track is the locus of points forming a circle whose center is at the end point and whose radius is the product of its maximum swimming speed and the length of the time interval. The intersection of loci originating from either the start point or end point with a reference to longitude defines the most northern and southern extent of the search area for that reference longitude. Because the distance of arcs whose center lies at the start point increases with time, whereas the distance of arcs whose center lies at the end point decreases with time, the latitudinal range of the search area is usually smallest at the start of the time series and at the end of the time series and is usually largest midway through the time series. The long time series obtained from the recovered archival tags creates a situation where the latitudinal extent of the search areas is largely determined by the northern and southern bounds of the habitat rather than by swimming speed. Swimming speed does, however, constrain east-west movement on a daily basis because the reference longitudes anchor the search areas. Step 2: Selection sets for tag data The second step of processing involves selecting SST records (from the tag data set) that are coincident in time with the daily search area. The user can define the sea surface layer by entering a maximum depth of this layer; for this study the surface layer was defined as 0-1 m. The user can also determine how many values from the selection set should be used to search for SST matches. We chose a selection set consisting of three individual values for PSAT tags; however, because of the much higher frequency of measurements from the archival tags, we chose a selection set that consisted of a single average SST value for each day. The temperatures found in the selected set of points for a given daily search area would be used to calculate the location of pixels within the search area that the tag most likely visited. Step 3: Choosing candidate points Selecting candidate points from which a best track will be chosen begins by assigning a temperature cost to pixels within the search area. The temperature cost for a given pixel, j, with a search area referenced by time, t{i), AT\J, Hi)], is simply the absolute value of the difference in its temperature, Tsatij, t(i)), and that of its closest match, k, from the selected set of tag points, Ttag\k, Hi)]: AT[j,t(i)} = \Tsat[j,tii)-Ttag[k,t(i)]]\. The temperature cost, AT \j. Hi)], is an inherited trait of a pixel and will be applied to all further calculations of the best track(s). If the temperature cost of any pixel examined in a search area exceeds the cutoff value entered by the user, that pixel will be removed from further consideration. Pixels will also be removed if they lie over land. Those pixels that remain are next subjected to an evaluation to determine if they qualify as candidate points. This evaluation is based upon the value of a cost function that weighs both the pixel's temperature cost described above, AT[j, t{i)], and the pixel's contribution to spreading coverage over the search area: Cost[jMi >] = AT[j,t(i)] + Spread Factor x AL[j,t( i >]. AL [j, Hi)] is the relative contribution a pixel makes to providing even latitudinal distribution along the refer- ence longitude and search lines of the daily search area; the Spread Factor weights the relative importance of temperature costs with the benefit of obtaining an even distribution. Although the primary criterion for selecting candidate points is how well tag SST matches satellite imagery SST, we have found that this criterion alone can cause all the selected candidate points to be bunched together. Such aggregation will force the computed track into small regions of the search area without regard to the distribution of matching pixels in proceeding or succeeding search areas. To avoid this problem the Spread Factor function spreads candidate points in a north-south direction thereby providing smoother and more economical tracks. The degree to which the Spread Factor function spreads candidate points is controlled by the user by entering a weighted value. For this study we chose an intermediate value (5000 out of a possible 9999) and this value was constant for all evaluations. The number of candidate points finally determined is determined by the user. For this study, five candidate points were identified for each search area. When the user defines the number of points to be evaluated in the search areas, pixels having the lowest cost are ranked and selected accordingly. Step 4: Enumerate and calculate the cost of arcs After the candidate points have been chosen, the best track! s) is computed by choosing a single candidate point from each of the daily search areas in the time series. The best track is selected from all possible tracks by choosing the one of least cost. Thus, the solution is global rather than serial. The computation begins by calculating the cost of arcs between candidate points from adjacent search areas, and ends by summing the cost of all the arcs of a given track (Figs. 3 and 4). The cost of an arc is a function of the temperature match for the pair of candidate points that define the arc, AT\j, t(i)\ and AT[k, t(i+D], as defined above. It also depends upon the minimum swimming speed required of the fish traveling between the two candidate points, arc velocity min, where arc velocity min distance between candidate pixels {t(i + l)-t(i)) 298 Fishery Bulletin 103(2) candidate point [j, t(i)] with inherited temperature cost ATQ, t(i)] enumerate all possible Arcs between consecutive search areas arc D.t(l) >[k.t(l+1>] whose cost is a function of distance and temperature costs candidate point [k,t(h with inherited temperature cost AT|j,t(i+1)] 11] End Figure 3 Enumerating and costing arcs. An arc is defined as the arc between any two candidate points of adjacent search areas. The cost of an arc depends upon the temperature cost, AT, of the two candidate points of the arc. It is also depends upon the swimming speed required to travel the distance of the arc. The cost of the arc between candidate point j and can- didate point k is arccost({j,t(i)}->{k,Hi + l)}) = {AT(j,t(i)) + AT(k(t,i + l)) + DistFactorl - velocity where velocity = the maximum sustained swimming speed of the fish; and DistFactor = a factor that scales the cost of swim- ming at a given speed in relation to the sum of the temperature costs of the two candidate points. Values for the DistFactor and Velocity are determined by the user. The rationale for such cost is that the best track should include an assessment of variations in swimming velocity as well as the costs of temperature. If swimming speed is judged to be an insignificant cost or too difficult to quantify, the DistFactor can be set to 0. If a land barrier lies between the pair of candidate points, the distance to swim around the barrier is calculated and included in the cost of the arc. In this study an interme- diate value (5000 out of 9999) was assigned for the Dist- Factor, and this value was constant for all evaluations. Step 5: Calculating the best track Finally, the algorithm calculates the sum of the arc costs for each track: Cost of tract = £ g£, arccost({j,t(i)}- > {k,t(i + l)}). The costs for all possible tracks are then ranked, and the track(s) with the lowest cost(s) is then saved and available for display (Fig. 4). The track is saved in a table of the PSAT Tracker database; the table contains records of the latitude, longitude, time, and surface temperature of the candidate points that comprise the track, as well as records of surface temperature from the satellite imagery at regular intervals along the arcs between candidate points. Depending on the length of the time series, this process analyzes tens of thousands to hundreds of thousands of tracks and thus is the most time-consuming step of the algorithm. Analyzing position data from PSAT Tracker Location estimates provided by PSAT Tracker were subjected to spatial analysis to describe the move- Domeier et al.: Tracking Thunnus thynnus onentalis with the aid of an automated algorithm 299 End Tstart > t(1), t{1) > t(2), T(2) > t (3), t(3) > end Figure 4 Diagram to show how the best track is calculated by summing the cost of arcs for all possible paths and then choosing the track of least cost. merit patterns and habitat use of Pacific bluefin tuna in the eastern Pacific. Monthly data were combined within each tag data set prior to performing utilization distribution analyses with the Home Range Exten- sion for ArcView (version. 1.1c, BlueSky Telemetry, Aberfeldy, Scotland) that employs the fixed kernel method (Rodgers and Carr, 1998). Results were dis- played as volume contours displaying the main centers of activity for each fish during a given time period. Initial analyses allowed us to combine data so that figures could be minimized. For the archival tag data, consecutive months with similar spatial distribution were combined and individual fish with very similar tracks were combined. All data from fish that were PSAT tagged were combined by month because of the relatively sparse data compared with the data from the archival tags. PSAT tag data provided a glimpse at year-to-year variations in bluefin distribution (August 2000 through October 2002), whereas the archival tag data were for a single year and allowed for a monthly comparison within one year (August 2002 to Septem- ber 2003). The near daily position data provided through the PSAT Tracker analyses allowed us to calculate the swimming speed of each fish. This was done by simply dividing the horizontal distance between consecutive data records by the time between consecutive data re- cords ( 1-4 days). Results Tag recoveries Fifteen of the PSAT tags transmitted data after remain- ing on the fish from 2 to 191 days (Table 2). Unfortu- nately some of these tags did not transmit usable data. Fourteen of them provided a pop-up location and eleven of them transmitted enough data for some level of analy- ses of behavioral and movement patterns. The Microwave Telemetry PSAT tag provided an archival data set with a one-hour sampling schedule. The Wildlife Computer PSATs transmitted data summaries that included a daily water column profile of temperature (obtained from the deepest dive) and the percent time each fish spent within predetermined temperature and depth bins. Four archival tags were recovered after a period at liberty of 16 hours to 385 days (Table 2). The 16-hour ar- chival tag recovery was made from a recreational angler very near the point of release; this tag was not used for any analyses. The three tag recoveries made after 300 days came from a purse-seine vessel. Two of these three recaptured fish spent several weeks in a grow-out pen before the tags were discovered; the dates the fish were in the pen were not used for any analyses. The light stalks of tags 441 and 159 were damaged during recov- ery. For these tags, the internal temperature and pres- sure sensors were verified by Lotek data, but external 300 Fishery Bulletin 103(2) temperature and light level sensors could not be checked. For tag 233, none of the sensors could be verifed because the tag had to be disassembled and destroyed by Lotek personnel in order to recover the data. Table 2 Details of tagged Pacific bluefin tuna (Thunnus thynnus orientalist. WC=Wildlife Computer Tag, MT= Microwave Telemetry Tag, Lot=Lotek Tag). Fish Tag date Weight (kg) Time at liberty (days) 4WC 13 August 2002 36 23 184 WC 13 August 2002 60 62 200 WC 13 August 2002 41 51 245 WC 2 August 2000 51 19 247 WC 2 August 2000 57 38 249 WC 2 August 2000 50 102 265 WC 2 August 2000 52 33 301 WC 2 August 2000 60 191 961 WC 3 August 2001 32 9 962 WC 3 August 2001 35 4 964 WC 3 August 2001 35 23 1041 WC 3 August 2001 26 2 1042 WC 2 August 2000 42 72 283 MT 13 August 2002 41 61 114 Lot 13 August 2002 52 16 (hours) 159 Lot 13 August 2002 52 375 233 Lot 14 August 2002 43 385 441 Lot 30 August 2002 12 323 50 -I 40- 30- 20- T3 i io- to t^ III I ' I \ r-J A Fish 19203 Fish 19368 € o- o z -10- ! !J "Fish tracker latitude -20- !i i' _ . _ - Wildlife computer latitude -30- \ - Microwave telemetry latitude 4-Aug-00 c 28-Aug-OO 21-Sep-00 15-Oct-00 8-Nov-00 2-Dec-00 26-Dec-OO 19-Jan-01 20-Aug-02 13-Sep-02 7-Oct-02 Figure 5 PSAT Tracker SST-based latitude solutions vs. Wildlife Computers and Microwave Telemetry light-based latitude estimates. PSAT Tracker algorithm The archival tags provided large data sets that allowed for the comparison of the PSAT Tracker algorithm to the manufacturer's light-based geolocation solution. Because longitude estimates generated by PSAT Tracker are constrained by the light-based estimates, these values differed very little from the position estimates from the various tag manufacturers. Although similar, the PSAT Tracker latitude solutions were generally less erratic than those produced from the three light-based algorithms, particularly surrounding the times of the equinoxes (Figs. 5 and 6). The spring and fall equinoxes each produced approximately two months of unreliable latitude estimates for light-based algorithms. Pacific bluefin tuna habitat use Horizontal movement Tagged bluefin tuna ranged as far north as the California-Oregon border and nearly to the tip of Baja California, Mexico, to the south. Although this distance encompasses 2400 km of coastline, these fish spent the majority of their time in the southern part of the range, best illustrated by a home range analysis of the combined approximately year-long tracks of the three archival tagged bluefin (Fig. 7). Tagged off the northern coast of Baja California, Mexico, these three bluefin moved northward until November, followed by a southward migration to south-central Baja California where they spent the months of January through June (Fig. 8). The two larger archival-tagged fish reached the offshore waters of Oregon before turning south and the smaller fish did not venture north of San Francisco, California. The two larger fish spent much of the winter and spring (January-June) in the coastal bight between Punta Eugenia, Mexico, to the north and Cabo San Lazaro, Mexico, to the south, and the smaller fish had a more dispersed spring range north of Punta Euge- nia. In July all three fish began to move to northern Baja, back into the general area where they were originally tagged and where they were subsequently recaptured (Fig. 8). This general pattern of summer-fall move- ment northward followed by a winter migra- tion southward and a winter-spring holding pattern off south-central Baja California was supported by data from fish with PSATs in years 2000 through 2002 (Fig. 9). Although position data for the months of January through June generally placed the tagged bluefin off southern Baja, two of the three fish tagged with archival tags under- went rapid April excursions to the north be- fore returning to the south (Fig. 10). Fish 159 traveled 2130 km, one way, before return- ing by 1 May; fish 441 made a similar move but did not go as far north (1285 km) and stopped its southward return 480 km north of its original starting point. The extreme Domeier et al.: Tracking Thunnus thynnus onentalis with the aid of an automated algorithm 51)1 northern latitude estimates calculated by PSAT Tracker placed fish 441 slightly north of Point Conception, California, and fish 159 near the California-Oregon border before it returned to wintering grounds off Baja California (Fig. 10). This movement is corroborated by a westerly trend in longitudes and a dra- matic drop in SSTs. For fish 159, SST dropped from 19. TC on 4 April 2003 to 12.4°C on 18 April 2003. Similarly, fish 441 experienced an SST drop from 19.5° to 13.5°C between 1 and 18 April. Data from the archival tags provided near daily positions for each fish. The longest time between successive fixes was four days. The calculated swim- ming speeds between successive posi- tion fixes ranged from 0 to 14.7 knots for all three fish combined. The mean swimming speed for all three fish was 1.3 knots (±1.3 km. Depth and temperature ranges Vertical movement was similar to that reported for other bluefin tuna (Block et al., 1997; Block, 2001; Kitagawa et al., 2004). Detailed analyses of vertical movement and temperature preferences and tolerances are beyond the scope of this article and will be pre- sented in a future publication. In general, dives were most common during the day; maximum dive depths ranged from 341 to 382 m. Fish with archival tags spent nearly 70. 1% of the time near the surface (<20 m deep). Ambient water temperatures ranged from 5.7° to 25.0"C (mean=17.4°C). The internal temperature offish tagged with archival tags ranged from 14.1° to 29.5 C (mean=21.8°C); average internal temperatures of the fish were 4.4°C warmer than ambient waters and at times were up to 19.2°C warmer. Discussion Although we used SST matching as the sole means of estimating latitude for the fish tracks and spatial analyses presented in our study, the extent of the northward fall migration of juvenile Pacific bluefin tuna in the eastern Pacific has been corroborated by occasional commercial landings of Pacific bluefin tuna in Oregon (McCrae3). Because Pacific bluefin tuna are apparently capable of existing in the northern part of the eastern Pacific range, even during the colder months of the year, it is not clear what dictates the movement pattern of these fish. It is reasonable to speculate that the tuna are taking advantage of seasonal ocean warming to exploit distant prey when the physiological expense to maintain optimum body 50 -i 30 ■ ^ W ■V* ***VjL- /K 20 - CD ■o Fish 159 — — v — , -m \xfsh&&* S 10 - CO € -30- o z -50 - Fish tracker latitude -70 - i i i ■ 1 i i ■ 1 ' CNJ CN CM O O O C\J C\J CO o o o CO o CO o CO CO CO o o o CO CO o o Aug Sep -Oct Nov Dec Jan CD LL ro r- -Apr May ■Jun 1 1 CO CO 2 CD CD ^ CD CO CD ^ CD T- CD Figure 6 PSAT Tracker SST-b ased latitude so utions and Lotek light-based lati- tude estimates. 3 McCrae, J. 2004. Personal commun. Fish & Wildlife, Newport, OR 97365. Oregon Dept. Figure 7 Fixed kernel home range analysis illustrating relative importance of the range of juvenile Pacific bluefin tuna {Thunnus thynnus orientalist in the eastern Pacific; dis- played are all points for fish 159, 233, and 441 and volume contours of 95% (outer line) and 50% (inner line) for all three fish combined. Isolated circle to the north is a 95% contour. 302 Fishery Bulletin 103(2) Oct-Nov 441 range ▲ deployment point X recapture point Figure 8 Grayscale contours of seasonal spatial use and movement pattern for fish 159 and 233 combined, displaying "core areas" of use represented by volumes of 10-50% . The smaller total range of fish 441 is illustrated by the polygon. temperature is less. Temperature and depth tolerances and preferences indicated in our study are similar to those of bluefin tuna studied in other parts of the world (Carey and Teal, 1969; Carey and Lawson, 1973; Block et al., 1997; Kitagawa et al., 2000, 2004; Block et al„ 2001; Brill et al., 2002; Itoh et al., 2003b). The migration of a fish with an archival tag from the western Pacific to the eastern Pacific (Itoh et al., 2003a) provides an interesting comparison to our data. This individual, tagged off Japan, made the trans-Pacific migration in about two months and then resided in the eastern Pacific for about eight months before being recaptured by a recreational angler. The fish arrived off the coast of northern California in the month of Janu- ary— a time when fish from our study were found to be at the southern extreme of their eastern Pacific range. By the month of March, the western Pacific migrant had traveled to the winter-spring grounds where it then seemed to behave in a pattern similar to that of fish tagged for our study. Whether or not the Itoh et al. (2003a) tagged fish illustrated a typical transition from trans-Pacific migrant to eastern Pacific resident will require more tag recoveries. It will be equally interest- ing to see future descriptions, from archival-tag data, of maturing Pacific bluefin tuna making the trip back to the western Pacific. Two of the Pacific bluefin tuna with archival tags were captured and recaptured in very close proximity in both space and time of year. The computed tracks for these two fish, both relatively large for the eastern Pacific, also showed that they kept close to each other for most of the year. A smaller fish, tagged a month later, underwent a similar north to south movement, but did not range as far north, particularly, or south. Given our extremely low sample sizes, very little can be concluded, but the question is raised as to whether or not Pacific bluefin tuna of different year classes have distinct schools and migratory behaviors. It is also im- portant to point out that the two larger fish were tagged in the dorsal musculature, whereas the smaller fish was tagged in the peritoneal cavity. The orientation of the light stalk is different for these two methods, one point- ing towards the surface and the other in the shadow of the fish and pointing down. How this tag orientation may influence the detection of light and subsequent position estimates is unknown. Two of our fish with archival made rapid northward migrations into much colder water in the early spring. This northward migration is similar to that made by Itoh's fish in the early spring of 1998. Because these movements occurred at a time when the light-based latitude estimates prove unreliable, it would not have Domeier et al.: Tracking Thunnus thynnus onentalis with the aid of an automated algorithm 303 ~J August September 7J October ~J November December ~J January Figure 9 Positions for eleven Pacific bluefin tuna {Thunnus thynnus orien- talis) tagged with satellite pop-up tags from 2000, 2001, and 2002 showing the 100% minimum convex polygon for fish positions within a given month. been possible to be certain that this rapid excursion was authentic without the aid of SST matching (as was also done by Itoh et al. [2003a]). The PSAT Tracker algorithm provided relatively quick and automated geolocation estimates for data recovered from three separate types of tags deployed on Pacific bluefin tuna. Furthermore, the PSAT Tracker latitude solutions compared favorably to the light-based latitude estimates during non-equinox times of the year. The use of SSTs to resolve latitude allowed for spatial analyses of individual bluefin positions for every month of the year, whereas a strictly light-based approach would not provide reliable latitude position estimates for approxi- mately 30% of a year-long track. PSAT Tracker also results in a global, rather than serial track solution. In essence this means that no single position estimate is selected without regard to the influence this position has on the overall track. A serial track is one that is produced by selecting each position without regard to the effect each selection has on the overall track. A se- rial track is also heavily biased by the start point and may weight the location estimates based upon the pre- vious location estimate, allowing a single poor location estimate to ruin the remainder of the location estimates for the track. It is instructive to compare our SST matching algo- rithm to the Kalman filter-based algorithm developed by Sibert et al. (2003). The Sibert et al. algorithm depends solely upon light data collected by the tag to estimate latitude and longitude, whereas the PSAT Tracker algorithm depends upon the light field to pro- vide an estimate of longitude and solely upon the sea surface temperature to provide an estimate of lati- tude. The initial estimates of both approaches are then refined according to a goodness-of-fit criterion that depends upon assumptions regarding the swimming behavior of the tagged fish. In the case of the Sibert et al.'s algorithm, the behavior of the fish is modeled in terms of a biased random walk model that describes the movement of the fish in terms of an advection- diffusion equation; the advective term describes the most probable displacement of the fish during a time step and the diffusive term describes the distribution of less likely displacements. The usefulness of the ran- dom walk model is largely determined by the adequacy of describing the distribution of swimming speed and direction of the fish. The algorithm also includes for- mulations of the dependence of the accuracy and pre- cision of the estimates of latitude and longitude from the tag upon other factors. For example, around the equinox the weighting of the estimate of latitude from the tag measurements is greatly reduced (specifically an inverse cosine squared function of date.) The Sibert et al. algorithm simply searches for a track that mini- mizes discrepancies between the positions predicted from random walk model (the transition equation) and those predicted from the tag measurements (the mea- surement equation). 304 Fishery Bulletin 103(2) 40.00 35.09) -130.00 >k -125i0 y^ 30.00 \ -120.00 -115.00 25.00 <] W% Figure 10 Track showing northward excursions of fish 159 (track extending to 38°) and 441 (track extending to 34.5°) between 1 April and 10 May 2003. Displayed SST imagery is a composite for the month of April showing a 7°C temperature gradient. PSAT Tracker is similar to the Sibert et al. algorithm in that it invokes a model of fish behavior; there is a simple constraint on the maximum distance that a fish can swim during a time step, and shorter tracks that require lesser expenditures of energy by the fish are favored. Like the Sibert et al. algorithm, the PSAT Tracker also incorporates candidate points that are not limited to the initial light-based estimate of longitude but includes adjacent longitudes based upon the user's assessment of the accuracy of the initial estimate. Fi- nally, both the Sibert et al. and the PSAT Tracker al- gorithms yield a solution that provides a best fit to the time series of satellite (in the case of PSAT Tracker) and tag measurements to the model of fish behavior. Unfortunately, it is difficult to make a general assess- ment of the accuracy of either approach. In the case of the PSAT Tracker algorithm, the accuracy of the track will decrease in the absence of a north-south tempera- ture gradient. We have not found a means of quantita- tively determining the accuracy of PSAT Tracker cal- culations. However, the quality of the fit between pixel values of temperature from imagery and tag values for positions along the track is calculated as a x2 value. In the case of the Sibert et al. algorithm, the accuracy of the track will decrease during the period of the equinox when the latitudinal errors of the light-based estimates are extremely large. Our data indicate that this period can be as long as two months surrounding each equinox (skewed towards winter). At such times the estimates of position derived by the Sibert et al. algorithm depend largely on the random walk model of fish movement, which provides only a generic description of movement. Although the algorithm provides values for the mean square errors of bias and randomness for the tag es- timates of latitude and longitude, these values are not true values for error of predicting location; rather they represent of the discrepancy between the estimates of position by the random walk model, the formulation of the latitude estimation error, and the tag measure- ments. Additionally, the Sibert et al. algorithm does not exclude the possibility of placing a fish on land. The PSAT Tracker worked well for this study because of the strong north-to-south temperature gradient that is presented in the northeastern Pacific. Studies conducted in regions with poor temperature gradients will continue to rely on light-based latitude estimates and approaches like the Sibert et al. algorithm. Further development of PSAT Tracker, or other SST-based geolocation al- gorithms, should explore a means of using light-based latitude positions in combination with SST matching when light data are reliable, but excluding light-derived latitude positions when they are unreliable. Domeier et al.: Tracking Thunnus thynnus oriental* with the aid of an automated algorithm 305 Acknowledgments This study was made possible through the support of the George T. Pfleger Foundation and the Offield Family Foundation. We thank those who helped us capture the fish in our study: Tom Pfleger, Tom Fullam, Tom Roth- erie, Greg Stutzer and Chugey Sepulveda. Archival tags were recovered with the assistance of the Inter-American Tropical Tuna Commission. Literature cited Bayliff, W. H. 1980. Synopsis of biological data on the northern bluefin tuna. Thunnus thynnus (Linnaeus, 17581, in the Pa- cific Ocean. Inter-Am. Trop. Tuna Comm. Spec. Rep. 2: 261-293. 1994. A review of the biology and fisheries for northern bluefin tuna, Thunnus thynnus, in the Pacific Ocean. FAO Fish. Tech. Paper 336, part 2:244-295. Block. B. A., H. Dewar, S. B. Blackwell, T. D. Williams, E. D. Prince, A. Boustany, C. F. Farwell, D. J. Dau, and A. Seitz. 2001. Archival and pop-up satellite tagging of Atlantic bluefin tuna. In Electronic tagging and tracking in marine fishes (J. R. Sibert and J. L. Nielsen, eds.), p. 65-68. Kluwer Academic Pubis, The Netherlands. Block, B.A., H. Dewar, T. Williams, E. Prince, C. Farwell. and D. Fudge. 1997. Archival and pop up satellite tagging of Atlantic bluefin tuna, Thunnus thynnus. In Proceedings of the 48th annual tuna conference, p. 10. Inter-Am. Trop. Tuna Comm., Lake Arrowhead, CA. 1998. Archival tagging of Atlantic bluefin tuna (Thunnus thynnus thynnus). MTS Journal 32:37-46. Bowditch, N. 1995. The American practical navigator: an epitome of navigation. Originally by Nathaniel Bowdich 11802), Publ. 9., 873 p. Defense Mapping Agency Hydrographic/ Topographic Center, Bethesda, MD. Brill, R., M. Lutcavage, G. Metzger, P. Bushnell. M. Arendt, J. Lucy, C. Watson, and D. Foley. 2002. Horizontal and vertical movements of juvenile blue- fin tuna (Thunnus thynnus), in relation to oceanographic conditions of the western North Atlantic, determined with ultrasonic telemetry. Fish. Bull. 100:155-167. Carey, F. G., and K. D. Lawson. 1973. Temperature regulation in free-swimming bluefin tuna. Comp. Biochem. Physiol. 44a:375-395. Carey, F. G„ and J. M. Teal. 1969. Regulation of body temperature by the bluefin tuna. Comp. Biochem. Physiol. 28:205-213. Delong. R. I., B. S. Stewart, and R. D. Hill. 1992. Documenting migrations of northern elephant seals using day length. Mar. Mamm. Sci. 8:155-159. Hill, R. 1994. Theory of geolocation by light levels. In Ele- phant seals: population ecology, behavior, and physi- ology (B. J. LeBoeuf and R. M. Laws, eds.), p. 227- 236. Univ. California Press, Berkeley, CA. Hill, R. D., and M. J. Braun. 2001. Geolocation by light-level. The next step: lati- tude. In Electronic tagging and tracking in marine fish- erie (J. Sibert and J. Nielsen, eds. I. p. 315-330. Kluwer Academic Press, Dordrecht, The Netherlands. Itoh, T., S. Tsuji and A. Nitta. 2003a. Migration patterns of young Pacific bluefin tuna (Thunnus orientalis) determined with archival tags. Fish. Bull. 101:514-534. 2003b. Swimming depth, ambient water temperature preference, and feeding frequency of young Pacific blue- fin tuna (Thunnus thynnus orientalis) determined with archival tags. Fish. Bull. 101:535-544. Kitagawa, T., S. Kimura, H. Nakata, and H. Yamada. 2004. Diving behavior of immature, feeding Pacific bluefin tuna (Thunnus thynnus orientalis) in relation to season and area: the East China Sea and the Kuroshio-Oyashio transition region. Fish. Oceanogr. 13(31:161-180. Kitagawa, T., H. Nakata, S. Kimura, T. Itoh, S. Tsuji, and A. Nitta. 2000. Effect of ambient temperature on the vertical distribution and movement of Pacific bluefin tuna Thunnus thynnus orientalis. Mar. Ecol. Prog. Ser. 206:251-260. Metcalfe, J. D. 2001. Summary report of a workshop on daylight measure- ments for geolocation in animal telemetry. In Electronic tagging and tracking in marine fisheries (J. Sibert and J. Nielsen, eds. I, p. 331-342. Kluwer Academic Press, Dordrecht, The Netherlands. Musyl, M. K., R. W. Brill, D. S. Curran, J. S. Gunn, J. R. Hartog, R. D. Hill, D. W. Welch, J. P. Eveson, C. H. Boggs, and R. E. Brainard. 2001. Ability of archival tags to provide estimates of geo- graphical position based on light intensity. In Electronic tagging and tracking in marine fisheries (J. Sibert and J. Nielsen, eds). p. 343-368. Kluwer Academic Press, Dordrecht, The Netherlands. Musyl, M. K„ R. W. Brill, C. H. Boggs, D. S. Curran, T. K. Kazama, and M. P. Seki. 2003. Vertical movements of bigeye tuna (Thunnus obesus) associated with islands, buoys, and seamounts near the main Hawaiian islands from archival tagging data. Fish. Oceanogr. 12:152-169. Nishikawa, Y., M. Honma. S. Ueyanagi, and S. Kikawa. 1985. Average distribution of larvae of oceanic species of scombrid fishes, 1956-1981. Nat. Res. Institute of Far Seas Fisheries, Shimizu. 12:99. Okiyama, M. 1974. Occurrence of the postlarvae of bluefin tuna, Thunnus thynnus, in the Japan Sea. Japan Sea Reg. Fish. Res. Lab. Bull. 25:89-97. Okiyama, M., and G. Yamamoto. 1979. Successful spawning of some holepipelagic fishes in the Sea of Japan and zoogeographical implications. In Proceedings of the seventh Japan-Soviet joint sym- posium on aquaculture. p. 223-233. Tokai Univ. Press, Tokyo, Japan. Rodgers, A. R., and A. P. Carr. 1998. HRE: the home range extension for ArcView. Us- er's manual, 27 p. Centre for Northern Forest Ecosys- tem Research, Ontario Ministry of Natural Resources, Thunder Bay, Ontario, Canada. Schaefer, K. M., and D. W. Fuller. 2002. Movements, behavior, and habitat selection of bigeye tuna (Thunnus obesus) in the eastern equato- rial Pacific, ascertained through archival tags. Fish. Bull. 100:765-788. 306 Fishery Bulletin 103(2) Sibert, J. R., M. K. Musyl, and R. W. Brill. 2003. Horizontal movements of bigeye tuna tThunnus obesus) near Hawaii determined by Kalman filter analyis of archival tag data. Fish. Oceanogr. 12(31:141-151. Sobel, D. 1995. Longitude: the true story of a lone genius who solved the greatest scientific problem of his time, 184 p. Penguin Books, New York, NY. Sund, P. N., M. Blackburn, and F. Williams. 1981. Tunas and their environment in the Pacific Ocean: a review. Oceanogr. Mar. Biol. Ann. Rev. 19:443-512. Welch, D. W., and J. P. Eveson. 1999. An assessment of light-based geoposition esti- mates from archival tags. Can. J. Aquat. Sci. 56:13171- 1327. 2001. Recent progress in estimating geoposition using daylight. /;; Electronic tagging and tracking in marine fisheries (Sibert and J. Nielsen, eds.l, p. 3693-83. Kluwer Academic Press, Dordrecht, The Netherlands. Wilson, R. P., J. J. Ducamp, W. G. Rees, B. M. Culik, and K. Neikamp. 1992. Estimation of location: global coverage using light intensity. In Wildlife telemetry: remote monitoring and tracking of animals (I. G. Priede and S.M. Swift, eds.) p. 131-134. Ellis Horwood, New York, NY. Yabe, H., S. Ueyanagi, and H. Watanabe. 1966. Studies on the early life history of bluefin tuna Thunnus thynnus and on the larvae of the southern bluefin tuna T. maccoyii. Rep. Nankai Reg. Fish. Res. Lab. 23:95-129. Yamanaka, H., and staff (sic). 1963. Synopsis of biological data on kuromaguro Thun- nus orientalis (Temminck and Schlegell 1942 (Pacific Oceanl. FAO Fish Rep, 6(2):180-217. 307 Abstract— The gray snapper {Lutjanus griseus) is a temperate and tropical reef fish that is found along the Gulf of Mexico and Atlantic coasts of the southeastern United States. The rec- reational fishery for gray snapper has developed rapidly in south Louisiana with the advent of harvest and sea- sonal restrictions on the established red snapper iL. campechanus) fishery. We examined the age and growth of gray snapper in Louisiana with the use of cross-sectioned sagittae. A total of 833 specimens, (441 males, 387 females, and 5 of unknown sex) were opportunistically sampled from the recreational fishery from August 1998 to August 2002. Males ranged in size from 222 to 732 mm total length (TL) and from 280 g to 5700 g total weight (TW) and females ranged from 254 to 756 mm TL and from 340 g to 5800 g TW. Both edge analysis and bomb radiocarbon analyses were used to validate otolith-based age estimates. Ages were estimated for 718 individu- als; both males and females ranged from 1 to 28 years. The von Berta- lanffy growth models derived from TL at age were L, = 655.4ll-e[-°-23((|l| for males, L, = 657.3{l-el-° 21l"l| for females, and L , = 656.4)l-e[-° 22"l|l for all specimens of known sex . Catch curves were used to produce a total mortality (Z) estimate of 0.17. Esti- mates of M calculated with various methods ranged from 0.15 to 0.50; however we felt that M=0.15 was the most appropriate estimate based on our estimate of Z. Full recruitment to the gray snapper recreational fishery began at age 4, was completed by age 8, and there was no discernible peak in the catch curve dome. Age, growth, mortality, and radiometric age validation of gray snapper {Lutjanus griseus) from Louisiana Andrew J. Fischer Coastal Fisheries Institute School of the Coast and Environment Louisiana State University Baton Rouge, Louisiana 70803-7503 E-mail address: afischeigilsu.edu M. Scott Baker Jr. North Carolina Sea Grant UNC-W Center for Marine Science 5001 Masonboro Loop Rd Wilmington, North Carolina 28409 Charles A. Wilson Louisiana Sea Grant College Program Louisiana State University Baton Rouge, Louisiana 70803-7507 David L. Nieland Coastal Fisheries Institute School of the Coast and Environment Louisiana State University Baton Rouge, Louisiana 70803-7503 Manuscript submitted 19 September 2003 to the Scientific Editor's Office. Manuscript approved for publication 20 November 2004 by the Scientific Editor. Fish. Bull. 103:307-319 12005). The gray snapper {Lutjanus griseus), commonly referred to as the mangrove snapper, is a temperate and tropical reef species that is found along the southeastern Atlantic coast of the United States from North Carolina to Bermuda, throughout the Gulf of Mexico (GOM), and south to Brazil (Johnson et al., 1994; Allman and Grimes, 2002). Gray snapper are fairly common along the Louisiana coast and are usually associated with complex structures such as oil and gas platforms, artificial reefs and other hard bottom substrates. In 1991 restrictions were put on the recreational red snapper (Lutjanus campechanus) fishery; these restric- tions coincided with a rapid expansion of the gray snapper fishery in south Louisiana. Recreational anglers now typically target gray snapper once they have reached their bag limit of red snapper; thus peak gray snap- per landings generally coincide with the red snapper recreational season (April-October). As a result, recre- ational landings of gray snapper in Louisiana have increased exponen- tially from 3.25 metric tons (t) in 1983 to 175 t in 2002 (NMFS1). Currently there is a 305 mm (12 inches) mini- mum size and a recreational bag limit of 10 fish/person/day for gray snapper in the GOM. Some background information is available for gray snapper in the southeastern United States, mainly from south Florida. Scientists have reported on early life history (Ruth- 1 NMFS (National Marine Fisheries Service). 2003. Fisheries Statistics and Economics Division. Unpubl. data. Website: http://www.st.nmfs. gov/pls/webpls/MF_ANNUAL_LAND- INGS. RESULTS. [Accessed 25 August 2003.] 308 Fishery Bulletin 103(2) erford et al., 1989; Domier et al.. 1997), population dynamics (Rutherford et al., 1989), juvenile food hab- its (Hettler, 1989), juvenile distribution (Chester and Thayer, 1990), and reproduction (Domeier et al., 1997; Allman and Grimes, 2002). Few reports have been conducted on the age and growth of gray snapper. Manooch and Matheson (1981) used sectioned otoliths to age gray snappers from east- ern Florida but did not validate their methods. Johnson et al. (1994) also used sectioned otoliths to age fish sampled from Fort Pierce, FL, to Grand Isle, LA, again without validation of methods. Burton (2001) validated the periodicity of opaque zone formation in gray snapper from east coast Florida waters with the use of marginal increment analysis of distal edge measurements. But gray snapper have never been fully examined in the northern GOM and comprehensive age, growth, and mortality data from the thriving Louisiana recreational fishery are virtually nonexistent. The objectives of our study were to describe the age, growth, and mortality of gray snapper from the Loui- siana recreational fishery. We obtained age information through examination of cross-sectioned sagittal otoliths, validated our aging techniques with the use of bomb- radiocarbon 14C and edge analyses, produced mortal- ity estimates with standard procedures, and modeled growth with the von Bertalanffy growth equation. Methods and materials Gray snapper were sampled from the Louisiana recre- ational harvest from August 1998 to August 2002 by personnel from the Louisiana State University Coastal Fisheries Institute and the Louisiana Department of Wildlife and Fisheries. Fish were opportunistically sam- pled at charter boat facilities in Port Fourchon, LA, and at spearfishing and hook and line fishing tournaments in Grand Isle and New Orleans, LA. Morphometric measurements (fork length [FL] and total length [TL] in mm, total weight [TW] in g) were taken, sex was determined by macroscopic examination of the gonads, and both sagittae were removed, rinsed, and air dried, weighed to the nearest 0.1 mg, and stored in coin enve- lopes until processed. For specimens in which TL was unavailable, TL was estimated from FL with the equa- tion TL = 1.048(FL) + 8.35 (linear regression, df=275; P<0.001; r2=0.98) calculated from specimens in which both TL and FL were available. In order to estimate age of gray snapper, a transverse section (~1 mm thick) was taken containing the core of the left sagittal otolith of each specimen. Sections were made with a Hillquest model 800, thin-sectioning machine equipped with a diamond embedded wafering blade and precision grinder (Cowan et al., 1995). In in- stances where the left otolith was unavailable, the right was substituted. Examinations of otolith cross-sections were made under a dissecting microscope with trans- mitted light and polarized light filter from 20x to 64x. Opaque zones were enumerated along the ventral side of the sulcus acousticus from the core to the proximal edge (Wilson and Nieland, 2001). Two readers (AJF and MSB) performed opaque zone counts independently without knowledge of capture date or morphometric data. Otolith marginal edge condition was coded as opaque or translucent by using the criteria described by Beckman et al. (1989). Opaque zones were counted a second time when initial counts differed. In instanc- es where a consensus between readers could not be reached, counts of the more experienced reader (AJF) were used. Between-reader variation in opaque zone counts was examined after the second readings of oto- lith sections were completed. Differences in counts were evaluated with the coefficient of variation (CV), index of precision (D) (Chang, 1982), and average percent error (APE) (Beamish and Fournier, 1981). Ages of gray snapper were estimated from opaque annulus counts and capture date with the equation described by Wilson and Nieland (2001): Day age = -182 + ( opaque increment count x 365 ) + {(.m-l)x30)+d, where m = the ordinal number (1-12) of month of cap- ture; and d = the ordinal number (1-31) of the day of the month of capture. The 182 days subtracted from each age estimate are to account for the uniform hatching date of 1 July assigned for all gray snapper to coincide with peak spawning activity occurring in July (Domeier et al., 1997; Allman and Grimes, 2002). Age in years was assigned by divid- ing age (in days) by 365. Year of birth (YOB) was back calculated by subtracting our otolith-based age esti- mates from year of capture. Validation of the periodicity of opaque zone forma- tion in gray snapper otoliths was examined with two approaches. An advanced and accurate method of age validation uses a quantitative measurement of nuclear bomb-produced radiocarbon (14C) that was accumulated in carbon-containing hard parts of marine organisms before, during, and after the atmospheric testing pe- riod of nuclear weapons (1958-65) (Baker and Wilson, 2001). Elevated levels of 14C have been observed in hermatypic corals (Druffel, 1980, 1989) and this time- specific marker can be used to validate age estimates derived from hard parts in marine fishes (Kalish, 1993; Campana and Jones, 1998). Baker and Wilson (2001) recently validated red snapper otolith section age esti- mates using this technique with excellent results. This same method was applied in our study to the otolith cores of gray snapper hatched after the nuclear testing periods. Gray snapper hatched prior to 1973 were not avail- able for our study, and thus the steepest portion of the radiocarbon uptake curve could not be used to confirm age estimates. Consequently, no coral reference data for the general area were available after 1983. Because red snapper otoliths have been previously validated Fischer et al.: Age, growth, mortality, and radiometric age validation of Lut/anus griseus 309 with this same method (Baker and Wilson, 2001), we anticipated that gray snapper radiocarbon values would be roughly similar to red snapper values for a given YOB. To obtain the oldest portion of the otolith for radio- carbon analysis, right otoliths of older gray snapper with an estimated YOB after the period of atmospheric testing (1973-95) were embedded in araldite epoxy resin and thin sectioned (~1 mm in thickness) through the core with an Isomet low-speed saw. The otolith core region was isolated from the otolith section by using the technique described in Baker and Wilson (2001). Cores were rinsed in double-distilled de-ionized water, allowed to air dry, weighed to the nearest 0.1 mg, and submitted to the accelerator mass spectrometry (AMS) facility in acid-washed 20-mL glass scintillation vials. The mean sample weight submitted for analyses was 12.8 mg. At the AMS facility, otolith cores underwent acid hy- drolysis with 85% phosphoric acid to yield CO., which was then made into graphite (pure C) by reduction at high temperature under vacuum. The graphite was pressed onto a target, loaded on the AMS unit and analyzed for radiocarbon. Samples were also analyzed for 13C to correct for natural and machine fractionation effects. Radiocarbon values from individual otolith cores were reported as A14C (mean ±SD), the adjusted devia- tion from the radiocarbon activity of 19th century wood (Stuiver and Polach, 1977). The periodicity of opaque zone formation was also examined with edge analysis. The marginal edge of each otolith was examined and coded as 1 opaque zone forming on otolith margin; 4 translucent zone forming on margin up to 1/3 com- plete; 5 translucent zone forming on margin 1/3 to 2/3 com- plete; 6 translucent zone forming on margin 2/3 to fully complete. Percentages of otoliths with opaque margins were plotted by month of capture (Beckman et al., 1989; Campana, 2001; Wilson and Nieland, 2001) for all months in which specimens were available. In order to examine the predictive capacity of otolith weight (W0) to determine age in gray snapper, sex spe- cific Wo-age relationships were fitted by using a power function with least squares with the model: Age = aW0b. A likelihood ratio test (Cerrato, 1990) was used to test for differences between male and female models. Male and female TW-TL relationships were indepen- dently fitted with linear regression to the model W = aTLh from log10-transformed data. Male and female re- gression coefficients were compared with an ANCOVA. Variability in age, TL, and TW-frequency distributions of males and females were compared with Komolgorov- Smirnov two-sample tests (Tate and Clelland, 1957; Sokal and Rohlf, 1995). Growth of gray snapper was modeled by using all specimens of known sex. Von Ber- talanffy growth models of TL at age were fitted with nonlinear regression by least squares (SAS 6.11, SAS Institute, 1996, Cary, NO in the form: TL,=LM „l-* P=0.06). Male and female speci- mens ranged from 222 to 732 mm TL and from 254 to 756 mm TL, respectively (Fig. 1A). Both sexes exhibited multimodal distributions; males were represented in the greatest numbers at 450 mm TL, compared to 400 mm 310 Fishery Bulletin 103(2) J I ■ Males □ Females ill 225 275 325 375 425 475 525 575 625 675 725 Total length (mm) 6 -i 5 4 B ■ Males D Females J J lllll lllllinl . . 200 800 1400 2000 2600 3200 3800 4400 5000 5600 Total weight (g) Figure 1 Distributions of (A) total length in mm (n = 837) and (B) total weight in g (/i = 832) for gray snapper (Lutjanus griseus) sampled from the Louisiana 1998-2002 recreational harvest. TL for females. A Komolgorov-Smirnov two-sample test indicated no significant difference between male and female TL frequencies (maximum difference=9.45). Male and female TW ranged from 200 to 5700 g and 300 to 5800 g TW, respectively (Fig. IB). Both sexes also displayed multimodal distributions in TW. A Komol- gorov-Smirnov two-sample test indicated a significant difference between sexes at 1600 g TW (maximum dif- ference=9.67). A single predictive TL-TW regression was generated for both males and females: TW = 3. .31 x 10-5 (TL285) (Fj 822=9,326.54; P<0.001; r2=0.92). Significant differences were found between sexes in TL-TW relationships (ANCOVA test of homogeneity of slopes, F3 822=7.25; P=0.007; r2=0.92). Therefore, sepa- rate models were fitted for each sex: Males 2.04 x 10 -B(x2-93) 7588.29; P<0.001; r2 TW (^l,436 Females = TW = 5.5 x 10-5(7/L277> = 0.95) Gray snapper otoliths are very similar in physical struc- ture, although much smaller in actual size, to those of the red snapper. Opaque zones are easily distinguishable on the ventral side of the sulcus groove (Manooch and Matheson, 1981; Johnson et al., 1994; Shipp2) (Fig. 2, A and B). Sagittae were collected from 721 gray snapper of which 718 were aged. Readers were unable to resolve opaque zones in three otolith sections because of poor sectioning. Readers agreed on the ages of 568 indi- viduals (78.8%) after initial counts and differed by one opaque annulus for 154 specimens, two annuli for 18 specimens, and three annuli for 2 specimens. Readers agreed on 709 ages (98.7%) after the second reading. The average percent error (APE) was 0.5, coefficient of variation (CV) was 0.00078, and index of percent (D) was 0.0006. cf, 385 = 3,089.16; P<0.001; r2=0.89> 2 Shipp, R. L. 1991. Investigations of life history parameters of species of secondarily targeted reef fish and dolphin in the northern Gulf of Mexico. Proc. Fourth Annu. MARFIN Conf., San Antonio, TX, p 80-85. [Available from National Marine Fisheries Service, State/Federal Liaison Office, 9721 Executive Center DR. N., St. Petersburg, FL 33702.1 Fischer et al.: Age, growth, mortality, and radiometric age validation of Lut/anus gnseus 311 Figure 2 (A) Transverse section of a gray snapper (Lutjanus griseus) otolith with first opaque zone distant from the core, with 10 opaque zones and an edge condition of 4 and (B) transverse section of a gray snapper otolith with first opaque zone close to the core, with 8 opaque zones, and an edge condition of 4. D indicates dorsal side and V indicates ventral side of otolith section. Table 1 List of gray otolith sepai length. I.D.= snapper {Lutjanus griseus) otoliths analyzed for stable carbon and bomb radiocarbon. "AMS wt •ated from the otolith section and submitted for accelerator mass spectrometry (AMS) radiocarbon = our identification number. " is the analysi amount of 3;FL=fork NOS-AMS number ID. Date caught Otolith section age (yr) Birth date Otolith wt. (mg) AMS wt. (mg) (%o) AUC {'',, Mean ±SD OS-36337 320 2001 28 1973 639.1 9.9 -2.67 142.8 9.7 OS-36338 33 2000 25 1975 635.2 14.7 -2.55 126.2 6.7 OS-36339 5 2000 20 1980 536.7 15.0 -3.34 115.3 6.5 OS-36340 322 2001 16 1985 414.6 15.1 -5.27 113.5 11.9 OS-36341 316 2001 11 1990 306.5 9.8 -4.49 91.4 6.1 OS-36342 304 2001 6 1995 154.0 12.2 -5.73 74.5 5.9 The gray snapper (n = 6) used for the radiocarbon age validation procedure ranged from 6 to 28 years of estimated age and were collected during 2000 and 2001 (Table 1). Furthermore, YOB ranged from 1973 to 1995. Gray snapper radiocarbon values were plotted along with red snapper radiocarbon values from the northern Gulf of Mexico (Baker and Wilson, 2001) and coral radiocarbon values from Bermuda (Druffel, 1989), South Florida (Druffel, 1989), and Belize (Druffel, 1980) (Fig. 3). Radiocarbon values of gray snapper cores were 312 Fishery Bulletin 103(2) 150 - I *°S»I »x ." 28yr a 100 - o a° a" f """U 25 yr ■ x • i J « 2°y m. x5 • J I J 16yr "■ pyr. i 50- Q X ° Bermuda °°° x South Florida x ° Belize j ^o ° L eampeehanus 0 - I ♦ L griseus „« ■ Collection Date -50 - 1950 1960 1970 1980 1990 2000 2010 Date of calcification (A.D.) Figure 3 Plot of radiocarbon ( 14C ) values versus date of calcification for gray snapper (Lutjanus griseus) (present study) and red snapper [Lutjanus eampeehanus) (Baker and Wilson, 2001) from the northern Gulf of Mexico and from corals off Bermuda ( Druffel, 1989 ), South Florida ( 1989 ), and Belize 1 Druffel, 1980 ). Solid squares ■ indicate collection dates for the gray snapper samples Oi=6) and the age listed are the estimated ages as read from the otolith sections. 1 2 80 ■ Opaque o o" 40 • 60 20 J \ \206 \n3 160 82 35 83 JFMAMJJASOND Month Figure 4 Marginal edge analysis of gray snapper (Lutjanus griseus) otoliths sampled from the 1998-2002 Louisiana recreational harvest (n=718). Numbers above data points indicate the number of otoliths analyzed for each month. Fischer et al.: Age, growth, mortality, and radiometric age validation of Lut/anus gnseus 313 16 - A 14 - ■ Males 12 - □ Females £. 10 - " Frequency 1 1 ■ 4 - j~ li 1 2 - j H 1 ■"h-L_ rr 0 - ,i 1 1 ii ilkJIki r* ru^^mn ~ r~ ii i i i i i i i i i i i i i i i i i i i i 1 3 5 7 9 11 13 15 17 19 21 23 25 27 Age (yrs) 25 -J B 20 ■ £, 15 - Percent o 5 • 1 1 ill 0 - .l-..llM 1 1.. 1970 1973 1976 1979 1982 1985 1988 1991 1994 1997 2000 Year of birth Figure 5 i Al Age and (B) year of birth distributions for male in = 407 ) and female (n = 307) gray snapper {Lutjanus griseus) sampled from the 1998-2002 Louisiana recreational harvest. highest in 1973 and exhibited a steady decline to a low in 1995. The periodicity of opaque annulus formation in gray snapper otoliths was further examined by plotting the monthly percentages of otoliths with opaque margins (Fig. 4). Although little data were available for the win- ter months, one specimen sampled in January and two specimens sampled in February 2001 each exhibited opaque marginal otolith edges indicating that opaque annulus formation occurs during the winter. Minimum percentages of otoliths with opaque margins during the months of April (22%) and May (8%) followed by an absence of opaque margins during the months of June through October indicate the cessation of opaque annulus formation by early spring and the onset of translucent annulus formation beginning in April and continuing through November. Male and female gray snapper ranged in age from 1 to 28 years (Fig. 5A). There was no significant differ- ence in age distributions between males and females (maximum difference=6.92 yr), but both sexes exhibited variable multimodal distributions in age frequency. Year of birth (YOB) frequency was also multimodal, and the population was dominated by younger fish; 77% of males and 80% of females were aged at 10 years or younger (Fig. 5B). Significant differences in slopes were detected when plotting age-W0 relationships between sexes (ANCOVA test of homogeneity of slopes, F3 353= 8.06; P=0.0005). Therefore, predictive models of age-W0 were fitted sepa- rately for males and females using a power function with least squares as (Fig. 6) Male age = 0.0278 (W0)106 (F2 204=3,956.29, P<0.001, r2=0.89). Female age = 0.0460 (Wo)097 (F2 148=4,504.05, P<0.001, r2=0.90). The single von Bertalanffy growth model to describe gray snapper TL at age (Fig. 7) was Lt = 656.4(1 -el-0-22inj| s+' O * S* • " O q X ^r* >^ XO ^Kv^T* a> 15 - O °^/^ »° < x xte/to db o oo-xjgr ocr o o ^jmQkio x o ^kco o^fcStocxo *>* 10 - o a»o dm *P ° octfcarcjto o Males ■ <&§§&§fco * Females ^^Sf^^ Power (Males) 5 - £S|pP - - - Power (Females) 0 - 1 1 1 1 1 i i i 0 100 200 300 400 500 600 700 800 Otolith weight (mg) Figure 6 Observed otolith weight (mg) at age for male (n=204) and female (n = 148) gray snapper iLittjanus griseus) sampled from the 1998-2002 Louisiana recreational harvests. Plotted lines are power functions fitted to the data. 800 -1 700 ■ X * ox© qt x o * 8 I«X1 * £^ xl*o* * ° * *?°x n n yn §, * 1 "i 5 jL-lj» V1 600 - | 500 • |? 400- o Z 300- vm/m * x qj x 200 - o / o Males 100 - x Females 0 5 10 15 20 25 30 Age (yr) Figure 7 Observed tota length (mm) at age for male (n = 407) and female (n=307) gray snapper {Lutjanus griseus) sampled from the 1998-2002 Louisiana recreational harvests. Plotted lines are von Bertalanffy growth functions fitted to the data. Fischer et al.: Age, growth, mortality, and radiometric age validation of Lut/anus gnseus 315 5 - /\ A / A Gray snapper (n=732) 4 - / V V /\ Ages 5-16, Z=0. 17 QJ 3 - n E c J ^ 5 2- 1 ■ \a a VvJV^ 0 4 8 12 16 20 24 28 Age (yr) Figure 8 Catch curve for Louisiana gray snapper [Lutjanus griseus) (rc=742) sampled from 1998 to 2002 Louisiana recreational harvests. Table 2 Degrees of freedom (df), sum of squares (SS), mean square (MS), F value, and P values for the full von Bertalanffy growth model lin which sexes were fitted independently) is compared model (by fitting all specimens of known sex). the likelihood ratio test by which with the reduced von Bertalanffy Model df SS MS F P Full Reduced 4714 2714 1.9493 x 1.9489 x 108 10s 48,732,614 97,447,139 16,341 32,217.6 <0.0001 <0.0001 However, a likelihood ratio test indicated growth models for males and females were significantly different from one another i) 25H 5(K) Mil .'S "^ -so W -75 -70 JS5 <0 \ 5 \ 4 Management area 2 3 2 N 1 v^ *" E Management V s area 1 0 20 40 60 80 100 Miles Figure 1 Map of the Maine coastline, showing the two management areas and the nine study strata from the fishery-inde- pendent survey program for green sea urchins (Stro/igylocentrotus droebachiensis). 2001 the DMR began an extensive fishery-independent survey program. This program generates large, spa- tially referenced, scientific data sets each year, which can be incorporated into stock assessments by using either fisheries population dynamics models or spatial analysis techniques. Spatial statistics, also known as spatial statistics or geostatistics, encompasses a diverse group of techniques that can be used to model the spatial variability of a process, such as sea urchin density, to estimate the value at unobserved locations (Bailey and Gatrell, 1995; Petitgas, 2001). Spatial variability is routinely divided into two categories: first- and second-order effects, or similarly, large- and small-scale variability. Large-scale variability is the variation in the mean value of the process over the study area, whereas small-scale vari- ability is the spatial dependence of the process, in other words the similarity between neighboring sites (Bailey and Gatrell, 1995). Intrinsic second-order methods, along with kriging, have become the most popular geostatistical tools and are now commonly used to estimate exploited fish stock biomass (e.g., Simard et al., 1992; Petitgas, 1993; Pelle- tier and Parma, 1994; Maravelias et al., 1996; Lembo et al., 1998; Maynou et al., 1998; Rivoirard et al., 2000; Petitgas, 2001). Two assumptions must be met to use intrinsic geostatistical methods: 1) independence be- tween the variable and the region's geometry and 2) stationarity (Petitgas, 1993; Warren, 1998; Rivoirard et al., 2000). If these assumptions are violated, we can attempt to modify the data to make them more appli- cable or we must use other spatial analysis techniques to estimate the spatial patterns. Tessellation is a spatial analysis technique that in- vestigates first-order, or large-scale, spatial variability of a process (Ripley, 1981; Bailey and Gatrell, 1995). Triangulated irregular networks (TINs), or Delaunay triangulation, are the simplest and most common tes- sellation technique, in which a three-dimensional sur- face of contiguous, non-overlapping triangles is created by linear interpolation of the variable. TINs are most commonly used for visualization purposes but can be used to estimate the biomass of a process (Simard et al., 1992; Guan et al., 1999). They have received limited use in fisheries stock assessment, however, because if a stock exhibits stationarity, second-order methods tend 322 Fishery Bulletin 103(2) to provide more precise biomass estimates, as well as a quantification of their variances (Simard et al., 1992; Bailey and Gatrell, 1995; Guan et al., 1999). The objective of our study is to investigate the spatial trends in green sea urchin density using spatial analy- sis techniques to estimate stock biomass. In doing so, we address the suitability of second-order methods to analyze a fishery with a target species that is highly spatially variable over a large, complex study area. We compare biomass estimates from several techniques to address the suitability of TINs for biomass estimation in the green sea urchin fishery. Materials and methods Data collection and processing Sea urchin density and size-frequency information were obtained from the 2001 pilot study for the State's annual fishery-independent survey. The Department of Marine Resources conducted the survey in June and early July, after the fishing season had ended. The survey was restricted to rock and gravel habitats along the Maine coast and we used two modes of data collection, divers and video. In the first part of the study, divers sampled 144 sites according to a stratified random sampling design. The design consisted of 16 sites in each of 9 survey strata, where the width of a survey stratum was inversely proportional to the commercial landings in the region. At each site, SCUBA divers randomly sampled .30 quadrats (1 m2 each) along three parallel linear transects set perpendicular to shore, for a total of 90 quadrats per site. The sampling intensity was divided equally among three depth zones: 0-5 m, 5-10 m, and 10-15 m. At each site, size-frequency data were obtained by randomly subsampling one quadrat in each depth zone, in which test diameters were measured for all individuals in the quadrat. An additional 148 sites were sampled, in a 15-40 m depth zone, with a video camera that recorded 10 quadrats (0.5 m2 each) at each site. Because of the low sea urchin densities at these sites, test diameters were measured for all recorded speci- mens. Mean sea urchin density values were calculated for each site (rc=292) and for each depth zone within a site (« = 580). An analysis of variance (ANOVA) was used to test if there were significant differences in mean sea urchin density and test diameter among survey strata. Five test diameter categories were created to more accurately represent the wide range of individual sea urchin weights. The categories were based on the state's minimum and maximum size restrictions, allowing us to separately estimate the biomass of sea urchins that have not yet recruited to the fishery, sea urchins within the fishery, and sea urchins that have escaped the fishery. The minimum (50 mm) and maximum (80 mm) size limits for our study were set slightly wider than the those of the state, because, according to the fishery regulations, up to 10% of the catch can be il- legal-size sea urchins. Size-frequency data from sub- sampled quadrats were applied to the mean sea urchin density for the specific depth zone and site, to generate density values for each size category. Weight per sea urchin was calculated from the mean length of the cat- egory by using a length-weight relationship (Scheibling et al., 1999). Spatial interpolation A sample semivariogram, often abridged to variogram, was generated from mean sea urchin densities by site, to examine the second-order spatial variation in the data set. The sample variogram was calculated with the fol- lowing equation (Bailey and Gatrell, 1995): yUi)- 2n(h) !(*,-*/, (i) SiS:, where S, and S = sampling point pairs with (x,y) coor- dinates; n = the number of sample point pairs; h - the distance between pairs; and 2 = mean urchin density for the sample. Trends in the variogram provide insights into the viabil- ity of second-order methods for the sea urchin data. Representations of the large-scale trends in sea ur- chin density were created by using Delaunay triangu- lated irregular networks (TINs) (ArcView 3.2a, 3D and Spatial Analyst Extensions, Redlands, CA). First, the sample points were plotted by using sea urchin density (/m2) as the z value. Second, each point was connected to the three nearest sites by linear interpolation, form- ing a continuous surface of nonoverlapping triangles (Fig. 2) (Bailey and Gatrell, 1995; Guan et. al., 1999). Thus, the z value of any location within a triangular surface is based solely on the three nearest sites. TIN surfaces were generated for 40 different scenarios, ac- cording to the size category, depth zone, and manage- ment area, which minimizes variability and allows us to produce more realistic biomass estimates. Finally, using a customized C++ program,1 we modified each surface to include only areas of appropriate sea urchin habitat. The green sea urchin is most commonly found on rocky substrate in the shallow subtidal (Scheibling and Hatcher, 2001), and, accordingly, the original sur- vey program was limited to areas with predominately rock or gravel substrata in areas less than 40 meters deep. Therefore, we used a map of surficial geology to identify areas of the correct substrate type (1:100,000 scale) (Kelley et al., 1997) and digital gridded bathym- etry data to create a plot of 5-m isoline contours. The bathymetry data source consisted of digital bathymetry data sets from sources such as NOAA and the Naval Oceanographic Office (15 arc second resolution) (Row- orth and Signell, 2002). 1 The C++ code used in this study is available upon request from the principal author (RCG). Grabowski et al.: Estimating stock biomass of Strongylocentrotus droebachiensis 323 2H Kilometers ^ Urchin density 0-5 ■ 5 - 10 | Id- 15 H 15 ~\ No data 20 kiloaielere Urchin density I 10-5 1 5 - 111 ■ I"- 15 ^B 15 - HNo data 5 10 15 20 Kilometer; Urchin density (5-10 | HI- 15 ^B 15 J No data 4- Urchin density 1 10-5 B 5- 1(1 | 10- 15 Hi l5 No data Figure 2 Representations of the triangulated irregular networks (TINs), used to characterize the large-scale patterns in green sea urchin [Strongylocentrotus droebachiensis) density (number of sea urchins/m2), for the 50-64 mm sea urchin size category in the central portion of management area 2. Top left, 0-5 m depth zone; top right, 5-10 m depth zone; bottom left, 10-15 m depth zone; bottom right, 15-40 m depth zone. To determine total sea urchin biomass <6) for each sce- nario, the volume beneath the modified TIN surface was calculated, from Riemann sums, and multiplied by the mean weight (w) according to the following equation: »XW*. (2) where st n fis,) = the spatial location (x,y) on an ASCII grid; = the number of grids squares; = the TIN surface and corresponds to a z value for each grid cell; and = the grid cell size, which was 1.72 hectares for area 1 and 1.82 hectares for area 2. Fishable biomass is defined as the biomass of all legal-size sea urchins and is simply the subset of the total biomass corresponding to legal-size sea urchins. Exploitable biomass corresponds to the legal-size sea urchins that are available to the fishery. Some areas included in this study may not be subject to fishing pressure because of geographic isolation or low sea urchin densities. Because information on historical fishing grounds is insufficient, exploitable biomass was estimated by using a threshold density value. Only areas with densities greater than the threshold were included in the exploitable biomass estimates. Two different types of threshold values were tested: 1) a threshold based on total sea urchin density and 2) a threshold based on the density of legal-size sea ur- chins. The threshold values make different assumptions about the fishery: method 1 assumes that fishermen target areas based on total sea urchin density, whereas method 2 assumes that fishermen target areas based on the density of legal-size sea urchins. Interviews were conducted with state sea urchin biologists and fishermen to determine an appropriate threshold value. The reported threshold values, the minimum total sea 324 Fishery Bulletin 103(2) urchin density that could attract fishermen, ranged from 20-50 sea urchin/m2. For the first scenario, the mean density from the range of recommended values, 35 sea urchin/m2, was selected. Therefore, the biomass of legal-size sea urchins was calculated only in areas where total sea urchin density was equal to or greater than 35/m2. For the second scenario, we estimated that commercial divers target areas that have greater than 10 legal-size sea urchins/m2. Estimation of uncertainty and stock assessment Because information on uncertainty cannot be directly obtained from the TIN method, cross validation was employed to approximate uncertainty in the estimation process. Cross validation involves randomly removing a site from a data set and predicting its value based on the other data points using the TIN process (Bailey and Gatrell, 1995). Residuals, or prediction errors, are calculated between the predicted and true values at the site. The process is repeated n times, resulting in an observed set of n prediction errors, or residuals. The frequency distribution and spatial distribution of residu- als provide insights into the accuracy of the model; an ideal model would have a mean residual value of 0 and positive and negative residuals would be distributed randomly over the study area. Sea urchin biomass values were also calculated with the arithmetic mean to provide comparisons with the spatially derived estimates. For total biomass, mean sea urchin densities by survey strata were multiplied by a spatially derived area estimate of suitable sea urchin habitat (<40 meters in depth) in the strata and the mean sea urchin mass per strata. Fishable biomass was calcu- lated the same way but sea urchin density values were scaled by the proportion of legal-size sea urchins in the stratum. Finally, exploitation rates, or the ratio of com- mercial landings to the exploitable biomass estimates, were calculated to facilitate comparison with the results generated from the population dynamics stock assess- ment and a recent study on biological reference points (Chen and Hunter, 2003; Grabowski and Chen, 2004). ■ rable 1 Quadi at density counts (/m2) for the green sea l rchin l Strongylocen trotu s droebachiensis) by management area and survey strata Sample size, n, is the number of quad- rats observed. Area Stratum Density SD n Min. Max. Mean 1 1 0 36 0.17 1.62 1706 2 0 130 2.57 10.63 1600 3 0 141 3.20 11.29 1580 2 4 0 180 4.20 14.13 1490 5 0 127 4.24 12.52 1580 6 0 147 10.06 17.59 1530 7 0 11.3 7.90 13.85 1498 8 0 113 13.50 20.38 1570 9 0 280 34.45 44.03 1540 Table 2 Sea urchin test diameter (mm) for gree n sea urchins ( Strong ylocentrotus droebachiensis) subsampled in the fishery independent survey program. Area Stratum Density SD n Min Max. Mean 1 1 7 80 38.69 21,07 29 2 3 81 39,01 22,19 627 3 4 89 45.25 18,90 855 2 4 3 89 32,99 19,82 1148 5 3 77 29,25 17,56 1034 6 4 110 39,87 16,23 1734 7 5 92 47,23 16,07 1283 8 3 114 42,11 16,86 2567 9 3 114 28,84 12,90 5263 Results Sea urchin density and size frequency, which were used to calculate biomass, varied considerably along the coast of Maine. Density (number of sea urchins/m2) differed significantly among survey strata (P<0.05; ANOVA), showing a general large-scale trend of increasing den- sity from stratum 1 to 9 (Table 1). Density also varied by depth; the sea urchin density in the 15-40 m depth zone was 0.32 sea urchins/m2, significantly lower than those of the three shallow (<15 m) depth zones (P<0.05, t- test), which each had approximately 9.50 sea urchins/m2. Sea urchin test diameter varied from 3 mm to 114 mm (mean at 35.90 mm). Test diameter differed significantly among survey strata (P<0.05; ANOVA), in which strata 4, 5, and 9 had the smallest size sea urchins, and strata 3 and 5 had the largest (Table 2). No meaningful trend was evident in the sample variogram, which showed a pure nugget effect (Fig. 3). This result indicates that the sea urchin density data were too spatially variable to be analyzed by intrinsic small-scale methods. Total sea urchin biomass was estimated at approxi- mately 250,000 metric tons (t), and legal-size sea urchins accounted for 165,000 t (Fig. 4). Most of the biomass was found in management area 2, which ac- counted for over 75% and 80% of the total and fishable biomass, respectively (Table 3). For both estimates, bio- mass varied by depth, being highest in the 0-5 m depth zone and lowest in the 15-40 m depth zone (Fig. 5). The two methods used to estimate exploitable bio- mass produced different biomass estimates with unique Grabowski et al.: Estimating stock biomass of Strongylocentrotus droebachiensis 325 Table 3 A summary of 2001 biomass estimates and 2000-2001 landings in meti ic tons, for the Maine green sea urchin fishery. Biomass estimates for the TIN method and ari thmetic mean were generated in th is study, whereas the popula ion dynamics estimates are from Chen and Hunter (2003). Area 1 consists of strata 1-3 and area 2 consists of strata 4 :> When possible, 95^ confidence intervals are included. in italics. Area 1 Area 2 Total TIN method Total biomass 45,868 204,304 250,172 Fishable biomass 39,060 126,725 165,786 Exploitable biomass Method 1 3645 5793 9438 Method 2 10.886 12.069 22,955 Arithmetic mean Total biomass 47,933 (42,399-54,331) 290,954 (274,632-307,977) 338,887 (317.031-362,308) Fishable biomass 24,241 (27.575-27.2S7) 90,185 (85,144-95.144) 114,426 (106.719-122.723) Population dynamics 6550 (4041-9450) 8452' (5866-11,701) 15,002 (10.307-21.151) 2000-2001 landings 2148 3213 5361 ' 2000 value. euu - * 500- IB F 400- » E CO 300- 200- 100- 0- ♦ •A— \ . * ♦ * . ♦ ♦ ♦ ♦♦ -.«► * ♦ ♦ ♦/ ♦ ♦ 50,000 100.000 150.000 h(m) 200.000 Figure 3 Sample variogram of mean green sea urchin (Stron- gylocentrotus droebachiensis) density by site, showing small-scale variability, gamma (y), with respect to the distance between sample point pairs, /;. spatial distributions. Exploitable biomass estimates for method 2 were more than 2 times greater than those for method 1 (Table 3). With method 1, legal-size sea urchins were concentrated in the northeastern corner of management area 2, but with method 2, they were concentrated in the northeastern portion of area 1 and the central portion of area 2 (Fig. 6). Exploitable sea urchin biomass showed different patterns by manage- ment area and depth than did total biomass and fish- able biomass (Fig. 5). For example, management area 1 had a larger share of the total exploitable biomass, 39% or 47%, for methods 1 and 2, respectively, and 120,000 "I (/) c o 100.000 - o 80,000 " E 60,000 " O) F 40,000 " o m 20,000 " 0 - U\ ■ Area 1 □ Area 2 u. 0-29 30^19 50-64 65-80 Diameter (mm) 81 + Figure 4 Total biomass by green sea urchin (Strongylocentrotus droe- bachiensis) test diameter according to management area. Sea urchins between 50 and 80 mm were considered legal size for this study, and the biomass within these limits, indicated by the dashed lines, constitutes the fishable biomass. this biomass was almost exclusively found in the 0-5 m depth zone, accounting for 98% or 93%, respectively, of the area's biomass. TIN biomass estimates were similar to ones produced with the arithmetic mean but were higher for total biomass and lower for fishable biomass. Exploitation rates for method 1 were estimated at 0.59 and 0.55 for management areas 1 and 2, respectively, and 0.20 and 0.27 for method 2, respectively. Exploitation rates 326 Fishery Bulletin 103(2) 30,000 25,000 20.000 15,000 10,000 5000 g 140,000 CO 120,000 ■ total □ fishable □ exploitable meth. 1 □ exploitable meth. 2 0-5 5-10 10-15 15-40 ■ total □ fishable □ exploitable meth. 1 □ exploitable meth. 2 0-5 5-10 10-15 Depth (m) 15-40 Figure 5 Total, fishable, and exploitable green sea urchin iStrongylocentrotus droe- baclnensis) biomass estimates by depth zone. Top, area 1; bottom, area 2. from the population dynamics modeling approach were 0.38 and 0.57 (2000) for management areas 1 and 2, respectively. Cross validation of sea urchin density surfaces yield- ed a mean residual of 0.50 (median=0, standard de- viation^.86, skewness=2.80, ra = 60) (Fig. 7). Residuals were greatest in regions with the highest spatial vari- ability, such as sites within depth zones 1 and 2 and in the eastern survey strata. Discussion Spatial variability and distribution The objective of this study was to investigate the spatial variability in green sea urchin density to estimate the biomass of the Maine stock. However, several factors limited the choice of spatial statistical approaches that could be used to assess the fishery. In particular, the physical structure of the study area, the dependence of sea urchin variables upon the environment and a high degree of small-scale spatial uncertainty make small- scale approaches inappropriate. First, the study area was neither uniform nor con- tinuous. Because the aim of the fishery-independent survey program was to assess the whole population of sea urchins in Maine, the study area had to span the entire coastline. Consequently, the study area encom- passed many features that create discontinuities in a spatial model at varying, yet relatively small, spatial scales. These features included the highly indented coastline, the presence of several hundred islands and the exclusion of regions because of environmental con- straints. Second, green sea urchin variables were not independent of the study area; rather, they were depen- dent on several environmental, ecological, and anthro- pogenic factors. In particular, depth, substrate type, Grabowski et al.: Estimating stock biomass of Strongy/ocentrotus droebachiensis 327 -A o"T" -»'■; Urchin densit\ ■■ 0-10 ■■ 9 i. No data Figure 6 Final spatial representations of the density of exploitable green sea urchins (Strongylocentrotus droebachiensis). Top row, method 1: threshold was based on total sea urchin density. Bottom row, method 2: threshold was based on legal-size sea urchin den- sity. Left column, eastern portion of management area 1; middle column, central portion of management area 2; right column, northeastern corner of management area 2. benthic algal presence, and the presence and level of fishing or predatory activity all greatly affect urchin density, growth rates, and size frequency (Vadas et al., 1986; Scheibling and Hatcher, 2001). Mean sea urchin density and size frequency were not constant over the study area (Tables 1 and 2). Density exhib- ited large-scale spatial trends along the coast, which are related, at least, to depth and fishing activity. The eastward increase in total sea urchin density along the coast corresponded well with the historical patterns of commercial sea urchin fishing in the State of Maine (Table 1). The fishery began in the southwest, but as sea urchin densities dropped in those regions, the fish- ery steadily progressed northeastward along the coast. Spatial patterns in density by depth (0-15 m vs. 15-40 m) may have been caused, in part, by the difference in sampling techniques, yet the magnitude of the differ- ences and support from ecological studies indicate that there is a pattern. Finally, sea urchin densities varied dramatically on small spatial scales — variations on the order of one magnitude within the same habitat, and sometimes only meters apart, are not uncommon (Scheibling and Hatcher, 2001). This variability was evident in the variogram analysis, which showed no meaningful small-scale spatial structure and thus no stationarity (Fig. 3). We were interested in identifying a spatial statisti- cal approach that would generate reasonable estimates of stock biomass. The numerous discontinuities in the study area, the dependence of variables on ecological factors, and the high spatial variability indicated that an intrinsic spatial statistical approach was not ap- propriate for the investigation. Therefore, we needed an approach that was geared towards the detection and modeling of large-scale variability and that also exhibited some robustness to discontinuities caused by the indented coastline, islands, and habitat constraints. We believe the TIN approach used in this study satisfies these requirements, and, additionally, allows for vary- ing levels of resolutions, with finer resolution in high density sampling areas. Biomass estimates We calculated exploitable biomass in two different ways because of the different assumptions they make about the fishery. Method 1 assumes that fishermen target areas based on total sea urchin density, whereas method 2 assumes that fishermen target areas based on the density of legal-size sea urchins. The spatial distribu- tions of legal-size sea urchin density, which were used to calculate exploitable biomass, were distinctive and showed little overlap between methods (Fig. 6). The spatial distributions appear to reflect different aspects of the sea urchin fishery. When the threshold was based on total density (method 1), exploitable biomass was 328 Fishery Bulletin 103(2) • «> %° e° >% jOCl o o 0W ,o Residuals • - o o • + Figure 7 Spatial distribution of residuals and frequency distribution, insert (median=0, standard deviation = 1.86, skewness=2.80, n=60), from the cross-validation study that addressed uncertainty in the TIN estimation process for estimating bio- mass for the green sea urchin (Strongyloeentrotus droebachiensis) fishery. concentrated in the eastern corner of management area 2, which is the most northeastern location on the coast of Maine. This area has high total sea urchin densities, but relatively low densities of legal-size adults, and is an important location for the trawling industry. When the threshold was based on the density of legal-size sea urchins (method 2), however, exploitable biomass was concentrated in the eastern portion of management area 1 and the central portion of area 2. These regions have lower average sea urchin densities, but higher percent- ages of legal-size adults, and are key fishing grounds for the state's dive-based fishery. Because the two methods reflected different aspects of the fishery, it is not surprising that they produced different estimates of exploitable biomass (Table 3). Nevertheless, these estimates did not differ consider- ably from those of the population dynamics model. The spatial analysis estimates bordered the ones derived from the population dynamics model; method-1 esti- mates were smaller than those derived from the popula- tion dynamics model whereas method-2 estimates were larger. The biomass estimates were similar despite the fact that they were derived from different models (spatial analysis and population dynamics model) using entirely different data sources (fishery-independent and fishery-dependent). The status of a fishery is often determined by com- paring the current fishing mortality or stock biomass with biological reference points (BRPs) (Hilborn and Walters, 1992). The previous stock assessment study estimated that the sea urchin stock biomass in Maine is only about 10% of the virgin biomass, implying that the fishery has been severely overfished. A preliminary in- vestigation into BRPs recently estimated a BRP F0 l for the urchin fishery, based on a yield per recruit analysis, and concluded that estimates of the current exploitation rate are much higher than the BRP, which means that the fishery is being overfished (Grabowski and Chen, 2004). However, when we compare the TIN exploita- tion rates with the preliminary mean BRP F0 ,, which ranged from 0.37 to 0.43 depending upon uncertainty levels, we get an unclear assessment of the stock status. The fishery is being drastically overfished according to method 1, but is healthy according to method 2. We believe that the assessment generated by method 2 was unrealistically optimistic, considering the results from the stock assessment and the decade-long declining trend in landings. Uncertainty and further studies The TIN method was an appropriate spatial statistical approach for estimating biomass for the sea urchin fish- ery; however, a disadvantage of this technique is that there is no straightforward method to estimate the uncer- tainty in the biomass estimates. Because the technique does not incorporate a variance structure into the estima- tion process, we could not directly estimate uncertainty. Therefore, we used cross-validation to approximate the uncertainty associated with the TIN method (Fig. 7). We found that the mean residual did not equal zero, indicating that there is a global bias in the TIN surfaces and that biomass estimates were likely overestimated (Simard et al., 1992). This bias was most likely caused Grabowski et al .: Estimating stock biomass of Strongylocentrotus droebachiensis 329 by a combination of the underlying patterns in spatial variability, the linear interpolation method employed in TIN formation, and the effects of sample selection in the cross-validation study. There are several possible ways to reduce the bias in the estimation process, such as incorporating a smoothing function or weighting based on neighbors into the TIN model. This procedure would not completely address uncertainty, however, because it would only acknowledge uncertainty in the TIN estima- tion process. To obtain confidence intervals for biomass estimates, we needed to incorporate uncertainty in mean density and in TIN estimation. We are currently inves- tigating methods to estimate confidence intervals, such as using a Monte Carlo simulation approach. A thorough examination and quantification of uncertainty is beyond the scope of this article. In this study, we identified a basic approach for inves- tigating spatial patterns, and estimating stock biomass in situations where second-order methods are inappro- priate. The TIN technique generated realistic biomass estimates that are similar to those derived with other approaches, but before we can recommend this tech- nique for the green sea urchin fishery, several points must be addressed. First, the two methods used to es- timate exploitable biomass must be integrated because they reflect different aspects of the fishery and result in different stock assessments. Second, a process must be established to estimate threshold levels because they have a large control over exploitable biomass estimates. Finally, a technique must be developed to estimate uncertainty in biomass. We would also recommend fur- ther investigations into tracking fishing pressure and identifying its effects on the benthic ecosystem and the spatial distribution of sea urchins. Acknowledgments We would like to thank the staff at the Maine Depart- ment of Marine Resources for collecting and compiling the sea urchin fishery data. We would especially like to thank Margaret Hunter and Robert Russell from the DMR, Kathryn Wisz, our laboratory assistant, Ryan Weatherbee, for his help with the manuscript, and Oliv- ier Mette, for his technical assistance. This project was partially supported by grants from the Northeast Con- sortium (UNH SUB 302-628), the Maine Department of Marine Resources (G1102012), and the Sea Urchin Zone Council to Y. Chen and a Maine Marine Science Fellow- ship from the Marine Department of Marine Resources and the University of Maine School of Marine Sciences to R. Grabowski. Literature cited Bailey, T., and A. Gatrell. 1995. Interactive spatial data analysis, 413 p. Pearson Education, Essex, England. Chen. Y., and M. Hunter. 2003. Assessing the green sea urchin {Strongylocentro- tus droebachiensis) stock in Maine, USA. Fish. Res. 60:527-537 Guan, W., R. H. Chamberlain, B. M. Sabol, and P. H. Doering. 1999. Mapping submerged aquatic vegetation with CIS in the Caloosahatchee Estuary: evaluation of different interpolation methods. Mar. Geod. 22:69-92. Grabowski, R., and Y. Chen. 2004. Incorporating uncertainty into the estimation of the biological reference points Ffl , and Fmax for the Maine green sea urchin {Strongylocentrotus droebachiensis) fishery. Fish. Res. 68:367-371 Hilborn, R., and C. J. Walters. 1992. Quantitative fisheries stock assessment: choice, dynamics and uncertainty, 570 p. Chapman and Hall, New York, NY. Kelley, J. T., W. A. Barnhardt, D. F. Belknap, S. M. Dickson, and A. R. Kelley. 1997. The seafloor revealed: the geology of the northwest- ern Gulf of Maine inner continental shelf, 55 p. Open file source 96-6. Maine Geological Survey, Natural Resources Information and Mapping Center, Augusta. ME. Lembo, G, T. Silecchia, P. Carbonara, A. Acrivulis. and M. T. Spedicato. 1998. A geostatistical approach to the assessment of the spatial distribution of Parapenaeus longirostris (Lucas, 18461 in the central-southern Tyrrhenian Sea. Crus- taceana 72:1093-1095. Maravelias, C. D., D. G. Reid, E. J. Simmonds, and J. Haralabous. 1996. Spatial analysis and mapping of acoustic survey data in the presence of high local variability: geo- statistical application to North Sea herring (Clupea harengus). Can. J. Fish. Aquat. Sci. 53:1497-1505. Maynou, F. X., F. Sarda, and G. Y. Conan. 1998. Assessment of the spatial structure and biomass evaluation of Nephrops norvegicus (L.l populations in the northwestern Mediterranean by geostatistics. ICES J. Mar. Sci. 55:102-120. Pelletier, D., and A. M. Parma. 1994. Spatial distribution of Pacific halibut (Hippoglossus stenolepis): an application of geostatistics to longline survey data. Can. J. Fish. Aquat. Sci. 51:1506-1518. Petitgas, P. 1993. Geostatistics for fish stock assessments: a re- view and an acoustic application. ICES J. Mar. Sci. 50: 285-298. 2001. Geostatistics in fisheries survey design and stock assessment: models, variances and applications. Fish and Fisheries 2:231-249. Ripley, B. R. 1981. Spatial statistics, 252 p. John Wiley & Sons, New York, NY. Rivoirard, J.. J. Simmonds, K. G. Foote, P. Fernandes, and N. Bez. 2000. Geostatistics for estimating fish abundance, 206 p. Blackwell Science, Oxford, UK. Roworth, E., and R. Signell. 2002. Construction of a digital bathymetry for the Gulf of Maine, http://woodshole.er.usgs.gov/project-pages/ oracles/gomaine/bathy/data.htm. [Accessed 22 Octo- ber 2002.] Scheibling, R. E., A. W. Hennigar, and T. Balch. 1999. Destructive grazing, epiphytism, and disease: the dynamics of sea urchin-kelp interactions in Nova Scotia. Can. J. Fish. Aquat. Sci. 56:2300-2314. 330 Fishery Bulletin 103(2) Scheibling, R. E., and B. G. Hatcher. 2001. The ecology of Strongylocentrotus droebachiensis. In Edible sea urchin biology and ecology (J. M. Lawrence, ed.), p. 271-304. Elsevier, New York, NY. Simard, Y, P. Legendre, G. Lavoie, and D. Marcotte. 1992. Mapping, estimating biomass. and optimizing sam- pling programs for spatially autocorrelated data: case study of the northern shrimp (Pandalus borealis). Can. J. Fish. Aquat. Sci. 49:32-45. Vadas, R. L„ R. W. Elner, R E. Garwood, and I. G Babb. 1986. Experimental aggregation behavior in the sea ur- chin Strongylocentrotus droebachiensis. Mar. Biol. 90: 433-448. Vadas, R. L., Sr., B. D. Smith, B. Beal, and T. Dowling. 2002. Sympatric growth morphs and size bimodal- ity in the green sea urchin (Strongylocentrotus droebachiensis). Ecol. Monogr. 72:113-132. Warren, W. G. 1998. Spatial analysis for marine populations: factors to be considered. In Proceedings of the North Pacific symposium on invertebrate stock assessment and man- agement (G. S. Jamieson and A. Campbell, eds.), p. 21-28. Can. Spec. Publ. Fish. Aquat. Sci. 125. 331 Abstract— The abundance and dis- tribution of California sea lions iZalophus californianus) in central and northern California was stud- ied to allow future evaluation of their impact on salmonids, the eco- system, and fisheries. Abundance at-sea was estimated by using the strip transect method from a fixed- wing aircraft with a belly viewing port. Abundance on land was esti- mated from 126-mm-format aerial photographs of animals at haulouts between Point Conception and the California-Oregon border. The sum of these two estimates represented total abundance for central and northern California. Both types of survey were conducted in May-June 1998. Septem- ber 1998, December 1998, and July 1999. A haulout survey was conducted in July 1998. The greatest number of sea lions occurred near Monterey Bay and San Francisco Bay for all surveys. Abundance was high in cen- tral and northern California in 1998 when warm water from the 1997-98 El Nino affected the region and was low in July 1999 when cold water La Nina conditions were prevalent. At-sea abundance estimates in cen- tral and northern California ranged from 12,232 to 40,161 animals, and haulout abundance was 13,559 to 36,576 animals. Total abundance of California sea lions in central and northern California was estimated as 64,916 in May-June 1998, 75,673 in September 1998, 56,775 in December 1998, and 25,791 in July 1999. The proportion of total abundance to ani- mals hauled-out for the four complete surveys ranged from 1.77 to 2.13, and the mean of 1.89 was used to estimate a total abundance of 49,697 for July 1998. This multiplier may be appli- cable in the future to estimate total abundance of California sea lions off central and northern California if only the abundance of animals at haulout sites is known. Abundance and distribution of California sea lions iZalophus californianus) in central and northern California during 1998 and summer 1999 Mark S. Lowry National Marine Fisheries Service Southwest Fisheries Science Center 8604 La Jolla Shores Dr. La Jolla. California 92037 E-mail address: mark.lowryia'noaa.gov Karin A. Forney National Marine Fisheries Service Southwest Fisheries Science Center 110 Shaffer Road Sanla Cruz, California 95060 Manuscript submitted 1 October 2002 to the Scientific Editor's Office. Manuscript approved for publication 14 December 2004 by the Scientific Editor. Fish. Bull. 103:331-343 (2005). The California sea lion {Zalophus cali- fornianus) is distributed from central Mexico to British Columbia, Canada. Four islands off southern California (Santa Barbara, San Clemente, San Nicolas, and San Miguel Islands) form the reproductive center for the U.S. population, although some pupping occurs at various other haulout sites in central California (Pierotti et al., 1977; Keith et al., 1984). The number of individuals off California varies throughout the year because sea lions from Mexico enter and leave Cali- fornia waters and individuals from California migrate southward into Mexico or northward as far as Brit- ish Columbia, Canada (Bartholomew, 1967; Bigg, 1988; and Huber, 1991). In southern California, the abun- dance of California sea lions peaks during the summer breeding season (Bartholomew, 1967; Odell, 1975). In central and northern California, the number of sea lions typically increases in the autumn during the north- ward migration, declines in winter, increases in spring as sea lions move to rookeries in southern California and Mexico, and declines in summer (Orr and Poulter, 1965; Mate, 1975; Sullivan, 1980; and Griswold, 1985; Bonnell et al.1). Since the mid-1970s, the Califor- nia sea lion population in the Unit- ed States has expanded at an aver- age of 5.0% per year and was most recently estimated to be between 204,000 and 214,000 individuals in 1999 (Forney et al.2). This estimate is roughly 2.7 times greater than in 1981-83 (Bonnell et al.1). As the U.S. sea lion population has grown, concerns have arisen about potential impacts on commercially harvested fish stocks. California sea lions feed on a variety of fish and cephalopods, some of which are commercially im- portant species, such as salmonids (Oncorhynchus spp.). Pacific sardines (Sardinops sagax), northern anchovy (Engraulis mordax), Pacific mackerel (Scomber japonicus), Pacific whiting (Merluccius productus), rockfish (Se- 1 Bonnell, M. L., M. O. Pierson, and G. D. Farrens. 1983. Pinnipeds and sea otters of central and northern Califor- nia, 1980-1983: status, abundance, and distribution. Center for Marine Stud- ies, Univ. California, Santa Cruz. OCS Study MMS 84-0044, 220 p. Prepared for Pacific OCS Region, Minerals Man- agement Service, U.S. Department of Interior, Camarillo, Calif. 93010, con- tract no. 14-12-0001-29090. 2 Forney, K. A., J. Barlow, M. M. Muto, M. Lowry, J. Baker, G. Cameron, J. Mobley, C. Stinchcomb, and J. V. Carretta. 2000. U.S. Pacific ma- rine mammal stock assessments: 2000. NOAATech. Memo.: NOAA-TM-NMFS- SWFSC-300, 276 p. National Marine Fisheries Service, Southwest Fisheries Science Center, 8604 La Jolla Shores Drive, La Jolla. CA 92037. 332 Fishery Bulletin 103(2) bastes spp.), and market squid (Loligo opalescens) (Low- ry et al., 1990. 1991; Lowry and Carretta, 1999; Weise, 2000). Effects on these resources have been estimated for Monterey Bay only, where during the 1997-98 El Nino sea lions consumed an estimated 269.1 to 804.7 metric tons (t) of salmon, 988.4 to 2206.8 t of sardine, and 533.4 to 1827.4 t of rockfishes annually (Weise, 2000). Recently, salmon in central and northern Cali- fornia have experienced population declines and some stocks have been listed as threatened or endangered under the U.S. Endangered Species Act. Although a variety of factors are responsible for the decline (e.g., logging, dams, agriculture, fishing), some salmonid populations are at such reduced levels that predation by sea lions may negatively affect their recovery (NMFS3). Sea lions also have been documented as interfering with recreational fisheries by consuming bait and chum and depredating hooked fish (Fluharty4). Existing methods of population assessment have been based on pup counts obtained at California sea lion rookeries near the end of the breeding season and total population has been estimated by extrapolating data from a life history model (Barlow and Boveng, 1991; Boveng5; Barlow et al.6 "; Forney et al.2). However, this approach cannot be used outside of the breeding season or in nonbreeding areas. Previous studies of California sea lion abundance and distribution in central and northern California during 1980-82 (Bonnell et al.1) 'NMFS (National Marine Fisheries Service). 1997. In- vestigation of scientific information on the impacts of California sea lions and Pacific harbor seals on salmonids and on the coastal ecosystems of Washington, Oregon, and California. NOAA Tech. Memo. NMFS-NWFSC-28, 172 p. Northwest Fisheries Science Center, 2527 Montlake Blvd. E., Seattle, WA 98112-2097 and National Marine Fisher- ies Service, Northwest Region, 7600 Sand Point Way N.E., Seattle, WA 98115-0070. 4 Fluharty, M. J. 1999. California sea lion interactions with commercial passenger fishing vessel fisheries: a review of log book data from 1994, 1995, and 1996. California Department of Fish and Game Admin, report 99-2, 21 p. [Available from California Department of Fish and Game, Marine Region, San Diego Field Office, 4949 Viewridge Avenue, San Diego, CA 92123.] 5 Boveng, P. 1988. Status of the California sea lion popula- tion on the U. S. west coast. National Oceanographic and Atmospheric Administration admin, report LJ-88-07, 26 p. Southwest Fisheries Science Center, 8604 La Jolla Shores Drive, La Jolla, CA 92037. 6 Barlow, J., R. L. Brownell Jr., D. P. DeMaster, K. A. Forney, M. S. Lowry, S. Osmek, T. J. Ragen, R. R. Reeves, and R. J. Small. 1995. U.S. Pacific marine mammal stock assessments. NOAA Tech. Memo. NMFS, NOAA-TM-NMFS- SWFSC-219, 162 p. National Marine Fisheries Service, Southwest Fisheries Science Center, 8604 La Jolla Shores Drive, La Jolla, CA 92037. 7 Barlow, J., K. A. Forney, P. Scott Hill, R. L. Brownell Jr., J. V. Carretta, D. P. DeMaster, F Julian, M. S. Lowry, T. Ragen, R. and R. Reeves. 1997. U.S. Pacific marine mammal stock assessments: 1996. NOAA Tech. Memo. NMFS, NOAA- TM-NMFS-SWFSC-248, 223 p. National Marine Fisheries Service, Southwest Fisheries Science Center, 8604 La Jolla Shores Drive, La Jolla, CA 92037. and 1995-96 (Beeson and Hanan8) included only ani- mals on land; animals at sea were either not considered or were included as a rough estimate. An assessment approach was, therefore, needed to provide quantitative estimates of California sea lion abundance in central and northern California that included both animals at sea and on land. This study uses a combination of the strip-transect method (to estimate at-sea abundance) and aerial pho- tographic counts (to estimate abundance of sea lions on land) in order to estimate the total abundance of California sea lions in central and northern California. Abundances were estimated separately for seven lati- tudinal zones within central and northern California. This study also describes distribution of sea lions by age and sex class in central and northern California, describes offshore distribution of sea lions, and intro- duces a new multiplier that can be used to estimate the total abundance of California sea lions at sea and on land, when only an estimate of the number of animals on land is available. Methods Survey dates and areas Surveys were conducted during May-June, July, Septem- ber, and December 1998, and July 1999. The May-June survey occurred when salmonid smolts were migrating out of rivers (NMFS3), the July survey when the United States stock of California sea lions was expected to be distributed mostly in California coastal waters, and the September and December surveys when adult salmon were migrating into rivers (NMFS3). The study area encompassed the waters and shoreline of central and northern California from Point Conception (34°26.8'N, 120°28.0'W) to the California-Oregon border (42°00.0'N, 124°12'W) within approximately sixty nautical miles of the coast (Fig. 1). Strip-transect surveys A twin-engine, high-wing Partenavia PN68C- or PN68- observer model aircraft was flown at an airspeed of 185 km/h during strip-transect and coastal haulout surveys. Abundance of sea lions at sea was determined by using the strip-transect method because previous aerial sur- veys in central California indicated that densities of sea lions would be too great in some areas to obtain reliable measures of perpendicular distances for line-transect density estimation. Previous aerial surveys using line transect methods, conducted at 213 m altitude, indicated a relatively flat detection function for sea lions between Beeson, M. J., and D. A. Hanan. 1996. An evaluation of pinniped-fishery interactions in California. Report to the Pacific States Marine Fisheries Commission, 47 p. [Available from Pacific States Marine Fisheries Commission, 205 SE Spokane St., Suite 100, Portland, OR, 97202-6413.] Lowry and Forney: Abundance and distribution of Zalophus californianus 333 42° l l l l i i i i i i i i i i i \ Oregon I \/ S California 41° rV-Cape Mendocino 40 \ / \ / 39° \ \ i rsi 38° \ x / /M^San Francisco - \ / K \ / /I 37° v' — /—^.Monterey - 36° - 35° N \ / / ( A \\_ 1 1 \T — \ — -_ \ 1 * 1 \ 34° i i i i i Point Conception 127° 126° 125° 124° 123° 122° 121 120 11 9C Longitude (°W) Figure 1 Strip-transect lines (solid lines) within the study area (dashed line I used for esti- mating at-sea abundance of California sea lions (Zalophus californianus) in central and northern California. approximately 85 meters left and right of the transect line (Fig. 2; Carretta, personal commun.9). Therefore, strip transect assumptions, that all individuals within the observed strip are detected, were expected to be valid within 85 meters left and right of the transect line. In our study we lowered the altitude of the aircraft to 183 m to increase the detection probability for sea lions in the water, especially in Beaufort 3-4 sea states. At that altitude, the viewing area of a single observer view- ing from the belly window extended from directly below (90) to a declination angle of 65° on each side, resulting in a total strip width of 170 m, or 85 m on each side of the viewing window. Transects followed predetermined lines that system- atically zig-zagged the study area (Fig. 1). Surveys were conducted in Beaufort sea states of 0-4. The lines were flown from south to north to take advantage of 12 - 10 - c o (J a; 0.8 - oS ■D ° 0.6 - >, .5 « 0.4 - o ct 0.2 - 00 01 0.2 0.3 0.4 Perpendicular distance (km) Figure 2 Probability density function for California sea lion (Zalo- phus californianus) sightings from an aircraft flying at an altitude of 213 meters in Beaufort sea states 0-4. Figure was provided by J. Carretta, National Marine Fisheries Service, Southwest Fisheries Science Center, La Jolla, CA. 92037. 9 Carretta, J. 1998. Personal commun. Southwest Fisher- ies Science Center, NMFS, La Jolla, California, 92037. sun angle and to minimize sun glare, except on a few overcast days when southbound flights provided ample visibility. Geographical positions were recorded at one- minute intervals directly to a laptop computer by a se- rial cable connected to the aircraft's global positioning system (GPS). The following data were collected: num- ber of California sea lions, GPS position, percentage of cloud cover over the survey area, name of the observer and data recorder, Beaufort sea state, transect num- ber, and percentage of glare. Percentage of glare was defined as the proportion of the viewing area in which the observer could not see into the water because of surface reflection caused by sun or cloud glare. During the May-June survey we used a recorder, observer, and a resting person — the resting person rotating with the observer approximately every 30 minutes. During the July, September, and December surveys, the resting person was eliminated and the observer and recorder rotated at approximately 30-minute intervals. Abundance at sea We used the nonparametric Kruskal-Wallis test for two- way comparisons of the effects of glare and sea state on California sea lion sighting rates. For these tests, each transect segment with constant viewing conditions was randomly assigned to one of five substrata, which served as replicate samples for the tests. Viewing conditions with significantly lower sighting rates were excluded from the abundance analyses to reduce bias caused by missed animals. Two a posteriori geographic strata were created, inshore (50,546 km2 total surface area) and offshore 334 Fishery Bulletin 103(2) 42 41° - 40 39c 38c 37c - 36° 35' 34c Point Conception J I I I I I I I I I I I I I L N. nA WLg(O) (1) 127° 126° 125° 124° 123° 122° 121° 120° 119° Longitude (°W) Figure 3 A posteriori stratification of study area into "offshore" stratum and into seven zones (A through G) within the "inshore" stratum for estimating abundance of Cali- fornia sea lions (Zalophus californianus) from strip-transect data and haulout count data. (56,526 km2 total surface area), using transect intersect points as the dividing line (Fig. 3). Differences between the definition of haulout sites for the surveys in this study and during previous surveys in 1980-82 and 1995 (Bonnell et al.1, and Beeson and Hanan8) made it neces- sary to create additional zones within the inshore stra- tum to allow comparisons of the three data sets. The inshore stratum was thus divided into seven zones ("A" through "G"), separated at the following latitudes: 1) 35°25'N; 2) 36°15'N; 3) 37°20'N; 4) 38°10'N; 5) 39°30'N; and 6) 40°50'N (Fig. 3). The zones were separated where gaps occurred in the distribution of haulout areas along the coastline. Total area sizes for the seven zones were the following: A: 7647 km2; B: 7206 km2; C: 8025 km2; D: 6153 km2; E: 7790 km2, F: 6030 km2, and G: 7695 km2. At-sea abundance was obtained separately for offshore and inshore strata, and for each zone within the inshore stratum, by using a modified strip-transect formula that included a correction, g(0), for diving ani- mals that were not available to be seen: where Nc = corrected total abundance (corrected for animals below the surface); n = number of individuals sighted within the strip-transect; A = total size of study area (in km2); W = the strip width (in km); L = distance surveyed (in km) calculated as the sum of the great circle distances between position fixes', and g(0) = probability that a sea lion will be visible at the surface within the strip viewed by the observer as the aircraft passes over the water. Coefficients of variation (CV) and lognormal 95% con- fidence limits of these abundance estimates were cal- culated by using standard formulae (Buckland et al., 1993). Probability of missing submerged sea lions We estimated the probability of seeing sea lions at the surface, g(0), from dive data in Feldkamp et al. (1989) derived from 14 foraging trips made by seven lactating adult female California sea lions during late breeding- season: g<0) = t+s + r t+s+r+d (2) where t = average time (hours) spent at the surface between dives within diving bouts by an adult female sea lion; s = average time (h) spent swimming near the surface between diving bouts by an adult female sea lion; r = average time (h) spent resting at the surface between diving bouts by an adult female sea lion; and d= average time (h) spent diving during diving bouts by an adult female sea lion. From seven female sea lions, Feldkamp et al. (1989) calculated averages of 12.0 hours (no SD given) spent at the surface between dives within diving bouts (t), 21.9 hours (SD = 9.5 hours) spent swimming near the surface between diving bouts (s), 1.6 hours (SD = 1.6) spent rest- ing at the surface between diving bouts (r), and 17.3 hours (SD = 6.7) spent diving during diving bouts (d). We calculated the CV forg(0) from the standard deviations of diving data. In using these data we assumed that between dives, sea lions swam near the surface and at a depth where they would be seen by an observer in the aircraft and that sea lions were not visible to an observer in the aircraft during dives. Dive data were not available for other age and sex classes; therefore, Lowry and Forney: Abundance and distribution of Zalophus californianus 335 it was assumed that the proportion of time spent at or near the surface was similar for adult females and other age and sex classes and did not vary significantly within region, season, and year. All counts were conducted by the first author, who is an experienced counter with high intercount reliability (Lowry, 1999). Geographical positions (latitude and longitude) were assigned to each haulout site. Photographic surveys The aircraft was flown from north to south directly over the coastline or slightly offshore at an altitude of 183 to 213 m (typically 213 m) to locate sea lions onshore. The low altitude ensured that California sea lions could be detected on rocky substrates, aided in identification of different pinniped species, and enabled accurate counts from aerial photographs. All hauled-out California sea lions onshore were photographed. At the Farallon Islands, the aircraft was flown at an altitude of 366 to 457 m (typically 396 m) to prevent disturbance of nesting seabirds. Multiple passes were made over large rocks or islands to ensure that the entire rock or island was photographed. Surveys were made without regard to tidal conditions at any time of day between approximately two hours after sunrise and two hours before sunset. Sea lions were photographed with a 126-mm-format KA-76 camera (Chicago Aerial Industries, Inc., Chi- cago, IL) equipped with image motion compensation (IMC) and operated at a cycle rate that achieved 67% overlap between adjacent frames. The geographical position of each photograph was recorded by linking the camera (mounted vertically inside the belly of the aircraft) to a computer and GPS unit. A 152-mm fo- cal-length lens was used for low altitude photography (i.e., 183-213 m) and a 305-mm focal-length lens was used for higher altitude photography (i.e., 366-457 m). Kodak Aerochrome MS Film 2448, a very fine-grained, medium-speed, color transparency film, or Aerochrome HS Film SO-359, a very fine-grained, high-speed, color transparency film, was used. The camera was set at an aperture of f/5.6 and a shutter speed between 1/400 and 1/2000 second. Photographic counts Sea lions were counted from photographs illuminated with a light table by using a 7-30X zoom binocular microscope. Counts were obtained for five age and sex class categories: pups, juveniles, adult females or young males of similar size, subadult males, and adult males. Age and sex class distinctions were determined from size and other external characteristics (e.g., hair color on head, presence of sagittal crest, chest size, fore flip- per width, snout shape, and body coloration). Animals of each age and sex class were marked on a clear acetate plastic overlay with different colored pens as each was counted. Marks on the acetate were then compared and verified with overlapping photographs. The acetate was placed on another photograph at the exact position of the coastline where the count ended previously and the count was continued on the uncounted portion. One count was made for each rock, island, or mainland haulout site. Analysis of haulout data Counts of sea lions made in this study were compared to those obtained by earlier investigators in 1980-82 (Bon- nell et al.1) and 1995-96 (Beeson and Hanan8) by using nested ANOVAs and paired //-tests. The null hypothesis of no difference in zonal counts was used to examine differences in counts by zone, season, year, and survey. The counts were 0.45 power transformed (with Systat 6.0 for Windows, SPSS Inc., Chicago, IL) because their distribution was skewed toward zero. Results Sighting rates and g(Q) No difference was found (P>0.05) for number of sight- ings, total animals seen, and mean group size during Beaufort sea state conditions 1 through 4. A sharp decline in sighting rates was observed when sightings were grouped into glare categories of 0-24% (rc=27.3 sightings/1000 km), 25-49% (n = 17.5 sightings/1000 km), 50-74% (ra=10.7 sightings/1000 km), and 75-100% (?2 = 0 sightings/1000 km). Sighting rates were signifi- cantly greater at 0-49% glare than at 50-100% glare (P<0.001 for all surveys combined); therefore, only data collected in 0-49% glare were used for at-sea abundance estimation. With only data collected in 0-49% glare, we used 48-76% of kilometers surveyed and 79-89% of sightings. The probability of sighting a sea lion at the surface, g(0), was estimated as 0.67 (with a CV of g(0) = 0.12). At-sea abundance Strip-transect survey effort totaled 1272 km during 26-30 May 1998, 2856 km during 12-28 September 1998, 2993 km during 1-11 December 1998, and 1175 km during 13-21 July 1999 (Fig. 4). No transect survey was conducted in July 1998 because of persistent low clouds and high winds. Transect distances in 0-49% glare conditions are given in Table 1. Nearly all sightings were within the inshore stratum, and most were within 20 nautical miles from the mainland coast (Fig. 5). Cor- rected at-sea abundance estimates for sea lions in the study area (Table 1) were 28,340 (May 1998), 40,161 (September 1998), and 24,720 animals (December 1998). For July 1999, a corrected abundance estimate for the inshore stratum in July 1999 was 11,492 animals (Table 1). From the total abundance estimated in the three 1998 surveys, the average proportion represented by the offshore stratum was 0.073 (range: 0.000-0.204). From this proportion, we estimated that there were about 829 sea lions in the unsurveyed offshore stratum 336 Fishery Bulletin 103(2) A Oregon r r 1 A California l / A 41 l\ /■ 40 '•A B a. "" 39" \ / \ C itf \ / _J\D \ ~T[— ~ • 37 36° N / 1 G 3 5 . N A 34° \.y •©«- B \ Oregon 41 \\/\ California 40 . *9v 39" 38° . % 37° 36° A >;& 35° - 34° V' -oo> 127" 126° 125 124 123 122° 121 120° 119= 127° 126° 125 124 123 122= 121= 120= 119= c Oregon d California A \NC\ B -- " V ' \ C \ / ~fCi?~ V \ "f, and 95% confidence limits for strata noted v> ith a dash I — ). Corrected estimates are based on g(0) calcu ated from dive studies on lactat ing adult females du ring late breec ing -season (Feldkamp et al„ 1989). Corrected Kilo No. of No. of surveyed CV Abundance Lower 95% Upper 95% Stratum sightings animals (kml (An (A,.) CL CL 26-30 May 1998 Inshore: zone A — — 0 — — — — Inshore: zone B 5 6 96 — 3977 — — Inshore: zone C — — 19 — — — — Inshore: zone D 6 6 63 — 5156 — — Inshore: zone E — — 0 — — — — Inshore: zone F 4 4 118 — 1793 — — Inshore: zone G — — 6 — — — — Inshore: total 15 16 302 0.29 23,541 11,224 49,376 Offshore 2 3 310 1.01 4799 561 41,040 Inshore + offshore 17 19 612 0.32 28,340 15,237 52,713 12-28 September 1998 Inshore: zone A 1 1 121 — 556 — — Inshore: zone B 5 5 140 — 2256 — — Inshore: zone C 6 7 117 — 4235 — — Inshore: zone D 18 23 108 — 11,552 — — Inshore: zone E 16 25 146 — 11,752 — — Inshore: zone F 5 5 69 — 3852 — — Inshore: zone G 15 16 220 — 4919 — — Inshore: total 66 82 919 0.27 39,595 24,210 64,757 Offshore 1 1 877 1.1 566 82 3923 Inshore + offshore 67 83 1796 0.26 40,161 24,205 66.635 1-11 December 1998 Inshore: zone A 4 4 213 — 1262 — — Inshore: zone B 6 7 219 — 2026 — — Inshore: zone C 4 4 238 — 1185 — — Inshore: zone D 2 3 124 — 1303 — — Inshore: zone E 15 25 175 — 9773 — — Inshore: zone F 3 18 59 — 16,129 — — Inshore: zone G 6 6 175 — 2316 — — Inshore: total 40 67 1203 0.5 24,720 9333 65,479 Offshore 0 0 977 0 0 0 0 Inshore + offshore 40 67 2181 0.5 24,720 9726 62,831 13-21July 1999 Inshore: zone A 0 0 124 — 0 — — Inshore: zone B 0 0 174 — 0 — — Inshore: zone C 0 0 185 — 0 — — Inshore: zone D — — 0 — — — — Inshore: zone E 11 14 146 — 6573 — — Inshore: zone F 0 0 135 — 0 — — Inshore: zone G 7 9 128 — 4762 — — Inshore: total 18 23 888 0.5 11,492 4,358 30,304 Offshore (estimated) 0 0 23 0.9 829 183 3752 Inshore + offshore 18 23 911 0.43 12,232 5427 27,572 Lowry and Forney: Abundance and distribution of Zalophus califormanus 339 Table 2 Counts of California sea 1 ions (.Zalophus californianus) made from 126-mm-format aerial color photographs ~or five age- and sex-class categories found in seven zones along the cent ral and northern California coast during four surveys in 1998 and one survey in 1999. Adult females Subadult Adult Zone Pups Juveniles or young males males males Total 31 May-8 June 1998 A 0 299 1948 1554 528 4329 B 0 3195 1534 2371 911 8011 C 0 698 751 513 530 2492 D 11 3639 5821 1636 555 11,662 E 99 3481 2993 678 464 7715 F 5 186 380 93 52 716 G 34 684 886 32 15 1651 All 149 12,182 14,313 6877 3055 36,576 7-18 July 1998 A 0 358 206 148 22 734 B (1 2382 116 162 62 2722 C 0 320 287 190 101 898 D 55 1918 7318 1283 290 10,864 E 54 2920 3226 564 178 6942 F 12 63 510 125 50 760 G 0 779 1362 92 30 3340' All 121 8740 13,025 2564 733 26,260' 11-20 September 1998 A 0 73 1325 1548 559 4165 B 0 1136 351 938 173 2598 C 0 524 594 584 56 2028 D 18 1506 8453 1136 100 11,213 E 22 2122 8056 671 188 11,059 F 6 470 1440 78 24 2018 G 0 1224 ' 1175 29 3 2431 All 46 7985 21,394 4984 1103 35,512 14-16 December 1998 A 0 27 105 162 123 663 B 0 193 1790 2950 429 5362 C 0 54 201 995 516 1766 D 1 765 10,310 632 97 11,805 E 12 1566 8035 311 103 10,027 F 9 307 903 84 15 1318 G 0 201 831 63 19 1114 All 22 3359 22,175 5197 1302 32,055 6-11 July 1999 A 0 111 167 5 4 287 B 0 6 6 1 1 14 C 0 0 0 1 0 1 D 3 193 970 109 91 1366 E 4 1226 5652 398 65 7345 F 0 270 578 90 14 952 G 0 919 2426 186 63 3594 All 7 2725 9799 790 238 13,559 1 Includes 1077 unknown age - and sex-class sea lions that were estimated to have been missed n zone G. 340 Fishery Bulletin 103(2) Table 3 Results of four nested ANOVAs on haulout counts of California sea lions (Zalophus californianut i found in 7 zones v. ithin central and northern California (refer to text and Fig. 3 for zone descriptions) The tests of ANOVA revealed differences between zones. season, years, and surveys. 1980-82 sur veys were conducted by Bureau of Land Management (Bonnell et al.1) and 1995-96 surveys were conducted by the California Department of Fish and Game (Beeson and Hanans) Year was nested within survey, season was nested within year, and zone was nested within season. Source Sum-of-squares df Mean-squa re F-ratio P 1998-99 surveys Season 1427.2 3 475.7 2.177 0.179 Zone (season) 9157.0 24 381.5 1.746 0.229 1998-99 surveys vs. summer and autumn 1995 and winter 1996 surveys Survey 1610.7 1 1610.7 11.449 0.003 Season (survey) 2019.4 5 403.9 2.871 0.037 Zone (season) 11,008.8 24 458.7 3.260 0.003 1998-99 surveys vs. 1980-82 surveys Survey 1731.6 1 1731.6 21.224 <0.001 Year (survey) 2235.9 3 745.3 9.135 <0.001 Season (year) 3576.5 12 298.0 3.653 <0.001 Zone (season) 14,761.2 24 615.0 7.538 <0.001 1980-82 surveys vs. summer and autumn 1995 and winter 1996 surveys Survey 81.9 1 81.9 1.457 0.232 Year (survey) 649.0 3 216.3 3.849 0.013 Season (year) 3491.5 10 349.1 6.211 <0.001 Zone (season) 11,027.6 24 459.5 8.174 <0.001 In 1999, the majority of sea lions were found between the San Francisco Bay area and Point Conception (zones D through G). Zone E had the greatest number of sea lions (Table 2); the majority of these animals hauled out at Ano Nuevo Island. As in 1998, juveniles and adult- females or young-males were the most prevalent age and sex classes (Table 2). Only seven pups were counted in the study area during July 1999. The number of sea lions counted in 1999 was 52% of that counted in July 1998. Total abundance There was a significant correlation (r=0.468, P=0.024) between at-sea abundance and haulout abundance within zones. Total abundance of California sea lions in central and northern California during 1998 was estimated to be 64,916 in May-June, 75,673 in September, and 56,775 in December. Total abundance in July 1999 was estimated at 25,791 individuals. The proportion of total abundance to animals hauled-out was 1.77, 2.13, 1.77, and 1.90, respectively, with a mean of 1.89 and a CV for small samples (Sokal and Rohlf, 1995) of 0.09. Using the mean multiplier of 1.89 on haulout counts obtained in July 1998 (Table 2), when at-sea abundance could not be estimated, we estimated total abundance as 49,697 (CV=0.09) animals for that period. Discussion This abundance study of California sea lions in central and northern California successfully integrated two methods: 1) strip transect surveys to estimate abun- dance at sea; and 2) aerial photographic surveys to esti- mate haulout abundance. TheglO) detection probability derived from previously published dive data allowed esti- mation of total abundance, including animals expected to be underwater during at-sea strip transect surveys. Previous surveys where transect methods similar to ours were used in the Southern California Bight in 1975-78 and in central and northern California in 1980-83 (Bonnell and Ford, 1987; Bonnell et al.1- 10) did not have information for deriving ^(0), and, therefore, densities of sea lions at sea were underestimated in these studies. California sea lions were abundant in central and northern California during May through September Bonnell, M. L., B. J. Le Boeuf, M. O. Pierson, D. H. Dett- man, G. D. Farrens, C. B. Heath, R. F. Gantt, and D. J. Larsen. 1980. Summary of marine mammal and seabird surveys of the Southern California Bight area 1975-1978. Vol. 3: Investigators reports, part 1 — pinnipeds of the South- ern California Bight, 535 p. Univ. Calif, Santa Cruz, Calif. 95064. Final Report to the Bureau of Land Management, under Contract AA550-CT7-367. [NTIS PB81-248-71.1 Lowry and Forney: Abundance and distribution of Zalophus califormanus 341 1998 when waters were warm because of the strong 1997-98 El Nino. Increased abundance of juveniles and adult females were observed in this region during previous El Nifios (Huber. 1991; Sydeman and Allen, 19991 and during our May-June, July, and September 1998 surveys. The increase in adult females in central California in 1998 resulted in an increase in the num- ber of pups counted at Ano Nuevo and South Farallon Islands (106 pups in 1998 vs. 23 in 1997), and below normal births at rookeries in southern California (Low- ry, unpubl. data, Forney et al.2). In contrast to 1998, during the summer of 1999 fewer sea lions were found in central and northern California, especially north of San Francisco (zones A, B, and C), and greater num- bers were found at rookeries in southern California (M. Lowry, unpubl. data) when waters were cold as a result of the La Nina oceanographic condition that began in October 1998 (Hay ward et al., 1999). The abundance and distribution of California sea lions were distinctly different between El Nino and La Nina periods. During El Nino, sea lions were very abundant in central and northern California, and were distributed throughout the region. In contrast, during summer 1999 (our only survey that year [La Nina|), sea lions were less abundant than during summer 1998, and they were distributed only south of the San Fran- cisco Bay area. The abundance and distribution pattern of summer 1999 is similar to the observed abundance and distribution pattern described by earlier studies (Chambers, 1979; Griswold, 1985; Weise, 2000; Bonnell et al.1). During periods of elevated sea lion abundance in central and northern California, such as those ob- served during the 1998 El Nino, we would expect 1) increased consumption of prey species because of more sea lions feeding in the area, 2) increased pressure on coastal fisheries resources because sea lions feed on commercially valuable species (see Lowry et al., 1990, 1991; Lowry and Carretta, 1999; Weise, 2000), and 3) increased interactions with commercial and sport fisher- ies. The opposite would occur during periods of low sea lion abundance during non-El Nino years. Greater abun- dance of California sea lions in central and northern California during the 1997-98 El Nino event, therefore, would be expected to have a greater effect on salmonids and other sea lion prey species, and on fisheries than would occur during non-El Nino years. Abundance of sea lions in central and northern Cali- fornia during 1998 was greater in May- June (spring) and September (fall) and less in July (summer) and December (winter). This bimodal phenomenon, also ob- served in the past (Sullivan, 1980; Bonnell et al.1), is due to migrating subadult and adult male sea lions on their way to (in fall) and from (in spring) Oregon (Mate, 1975), Washington, and British Columbia (Bigg, 1988). However, these seasonal differences were not signifi- cantly different, likely because of low power (only one year of data), or because the animals behaved differ- ently from other years. In fact, fewer subadult and adult males were present at southern California rookeries during the 1998 July census (near the end of breeding season) than were present during 1997 and 1999 (M. Lowry, unpubl. data). The large number of sea lions in central and northern California during 1998 was the result of a more numerous population (U.S. population estimated at 204,000 to 214.000 in 1999) than existed when previous surveys were conducted in 1980-82 and 1995-96 (U.S. population estimated at 76,000 in 1982 and at 167,000 to 188,000 in 1995) (Barlow et al.7; For- ney2; Bonnell et al.1, and Beeson and Hanans>. In central and northern California, California sea lions have been sighted during aerial surveys (Carretta and Forney"; present study) and tracked with satellite tags (Melin and DeLong, 2000; Melin, 2002) up to 100 nautical miles from shore. However, our surveys indi- cated that they forage predominantly within 20 nautical miles from shore. The strip transect method assumes that all animals within a strip are sighted by the observer. Although we found no difference in sighting rate between Beaufort sea state scales 0-1, 2, 3, and 4, Carretta et al.12 found during their 1998-99 line transect survey in waters off San Clemente Island, California, that the effective strip width of pinniped sightings at 213 m altitude was slightly less in Beaufort sea states 3-4 (184 m on each side) than in Beaufort sea states 0-2 (256 m on each side). Their results suggest that if our analysis suffered from reduced detection probability at high sea states, then we may have underestimated at-sea abundance of sea lions or increased the variance of at- sea sea lion abundance. This potential negative effect was minimized in our surveys by surveying at a lower altitude (183 m) than the 213 m altitude surveyed by Carretta et al.12 The g{0) correction derived from dive and foraging studies of lactating adult-female California sea lions during late breeding season (July-August) may be an additional source of error in our at-sea abundance es- timates. It may not be representative of nonlactating adult females and other age- and sex-class sea lions, and it may not be representative for all seasons or different oceanographic cycles (e.g., El Nino and non- El Nino). Dive data from various ages and sexes are needed to test these assumptions, but existing dive data from a single age+sex group provided a rough correc- tion to account for animals underwater during at-sea 11 Carretta, J. V. and K. A. Forney. 1993. Report of two aerial surveys for marine mammals in California coastal waters utilizing a NOAA DeHavilland twin otter aircraft March 9-April 7, 1991 and February 8-April 6. 1992. NOAA Tech. Memo. NMFS, NOAA-f M-NMFS-SWFSC-185, 77 p. National Marine Fisheries Service, Southwest Fisheries Science Center, 8604 La Jolla Shores Drive, La Jolla, CA 92037. 12 Carretta. J. V., M. S. Lowry, C. E. Stinchcomb, M. S. Lynn, and R. E. Cosgrove. 2000. Distribution and abundance of marine mammals at San Clemente Island and surrounding offshore waters: results from aerial and ground surveys in 1998 and 1999. National Oceanographic and Atmospheric Administration admin, report LJ-00-02, 51 p. Southwest Fisheries Science Center, 8604 La Jolla Shores Drive, La Jolla, CA 92037. 342 Fishery Bulletin 103(2) surveys. Seasonal differences may exist, but data in Feldkamp et al., (1989, 1991) and Melin (2002) indicate that these differences are negligible. Feldkamp et al. (1991) showed differences in diving behavior during El Nino and non-El Nino, but Melin (2002) did not find as much difference in diving behavior during El Nino and non-El Nino (with the exception of longer transit time to foraging grounds during El Nino). Error in age- and sex-class abundance estimates at haulouts is also affected by subjectivity and inter-ob- server differences in age and sex classification of sea lions. Therefore, age- and sex-class counts provided in these surveys, although conducted by a single ex- perienced observer (M. Lowry), serve as approximate indices of sea lion age- and sex-class distributions in central and northern California. These indices will be useful for future attempts to estimate consumption of prey by sea lions along central and northern California, given that nutritional requirements differ among age and sex classes. By estimating abundance of sea lions on land as well as at-sea, we were able to derive a multiplier for estimating total abundance from counts of animals hauled out on land. This multiplier can be applied to future land counts of California sea lions in central and northern California to estimate total abundance, as has been done for harbor seals in California, Or- egon, and Washington (Huber et al., 2001; Barlow et al.6; Forney et al.2). It may also be useful for es- timating total abundance from counts of sea lions hauled out in Oregon, Washington, and British Co- lumbia because the age- and sex-class structure of sea lions is similar to that found in central and northern California. However, the multiplier should not be used for smaller areas (such as the zones in the inshore stratum) or for other species, because regional and interspecies differences may exist. In particular, it would not be appropriate for regions where sea lions reproduce, such as in the Southern California Bight (SCB) and in Mexico, because adult females that are rearing pups may spend a different proportion of their time at sea. For that reason, it would be judicious to conduct concurrent offshore and haulout surveys in the SCB and Mexico to derive a correction factor for each geographical region of the sea lion's range. Multipliers could also be derived for smaller areas (such as our zones) by conducting suitably designed smaller-scale at-sea surveys in conjunction with counts of animals hauled out, or by using satellite or radio telemetry tags to directly measure the relative times at sea and on land. The multiplier for deriving total abundance from haulout counts provides researchers and resource man- agers with an alternative method for estimating total population abundance or abundance of a stock. Abun- dance estimates derived with this new approach can be compared to abundance estimates obtained with more conventional methods (such as the life history model), and may provide a means for estimating to- tal abundance when life history data are unavailable. The approach used in the present study may be par- ticularly useful for estimating abundance at times and places unrelated to breeding activities, or for periods when breeding is disrupted, as with El Nino conditions. Abundance estimates and distributional data provided by these methods can be used to determine where and when the greatest effects on salmon and other prey spe- cies may occur. Diet studies at major hauling areas in conjunction with abundance surveys to derive consump- tion estimates are required to determine the effect of California sea lions on salmon and other sea lion prey species of the region. Acknowledgments This research was supported financially by the Office of Protected Resources, National Marine Fisheries Ser- vice. We greatly appreciate the assistance given by Jim Gilpatrick, Charlie Stinchcomb, and, especially, Scott Benson of Moss Landing Marine Laboratories during the surveys. Jay Barlow provided guidance. Special thanks to Morgan Lynn of the Southwest Fisheries Science Center who kept the photographic equipment functioning properly. Henry Orr of the Southwest Fisheries Science Center helped with illustrations. Research within Gulf of the Farallones National Marine Sanctuary and Monterey Bay National Sanctuary was conducted under National Marine Sanctuary Permit GFNMS/MBNMS-20-98. This research was conducted under MMPA Research Permit No. 774-1437. We greatly appreciated the reviews and comments by Jay Barlow, Jeff Laake, Jim Harvey, and three anonymous reviewers. Literature cited Barlow. J., and P. Boveng. 1991. Modeling age-specific mortality for marine mammal populations. Mar. Mamm. Sci. 7:50-65. Bartholomew, G. A. 1967. Seal and sea lion populations of the Channel Islands. In Proc. of the symp. on the biology of the Calif. Islands (R. N. Philbrick, ed.l, p. 229-243. Santa Barbara Botanic Garden, p. 229-243. Santa Barbara, CA. Bigg, M. A. 1988. Status of the California sea lion, Zalophus califor- nianus, in Canada. Can. Field Nat. 102:307-314. Bonnell, M. L., and R. G. Ford. 1987. California sea lion distribution: a statistical an- alysis of aerial transect data. J. Wildl. Manage. 51: 13-20. Buckland, S. T., D. R. Anderson, K. P. Burnham, and J. L. Laake. 1993. Distance sampling: estimating abundance of biological populations, 446 p. Chapman and Hall. London. Chambers, J. R. 1979. Population studies of California sea lions near Diablo Canyon, California. M.A. thesis, 154 p. Cali- fornia Polytechnic State Univ., San Luis Obispo, CA. Lowry and Forney: Abundance and distribution of Zalophus californianus 343 Feldkamp, S. D., R. L. DeLong, and G. A. Antonelis. 1989. Diving patterns of California sea lions, Zalophus californianus. Can. J. Zool. 67:872-883. Feldkamp, S. D., R. L. DeLong, and G. A. Antonelis. 1991. Effects of El Nino 1983 on the foraging patterns of California sea lions (Zalophus californianus) near San Miguel Island, California. In Pinnipeds and El Nino: responses to environmental stress (F. Trillmich and K. A. Ono, eds.) p. 146-155. Springer-Verlag, Berlin and Heidelberg, Germany. Griswold, M. D., Jr. 1985. Distribution and movements of pinnipeds in Hum- boldt and Del Norte counties, California. M.A. thesis, 101 p. Humboldt State University, Areata. CA. Hayward, T. L., T. R. Baumgartner. D. M. Checkley, R. Durazo, G. Gaxiola-Castro, K. D. Hyrenbach. A. W„ Mantyla, M. M. Mullin, T. Murphree, F. B„ Schwing, R E., Smith, and M. J. Tegner. 1999. The state of the California Current in 1998-1999: transition to cool-water conditions. CalCOFI Rep. 40:29-62. Huber, H. R. 1991. Changes in the distribution of California sea lions north of the breeding rookeries during the 1982-83 El Nino. In Pinnipeds and El Nino: responses to envi- ronmental stress (F. Trillmich and K. A. Ono, eds.) p. 129-137. Springer-Verlag, Berlin and Heidelberg, Germany. Huber, H. R., S. J. Jeffries, R. F. Brown. R. L. DeLong, and G. VanBlaricom. 2001. Correcting aerial survey counts of harbor seals (Phoca vitulina richardsi) in Washington and Oregon. Mar. Mamm. Sci. 17:276-293. Keith. E. O, R. S. Condit, and B. J. Le Boeuf. 1984. California sea lions breeding at Ano Nuevo Island, California. J. Mamm. 65:695. Lowry, M. S. 1999. Counts of California sea lion (Zalophus california- nus* pups from aerial color photographs and from the ground: a comparison of two methods. Mar. Mamm. Sci. 15:143-158. Lowry, M. S, and J. V. Carretta. 1999. Market squid (Loligo opalescens) in the diet of California sea lions (Zalophus californianus! in southern California (1981-1995). CalCOFI Rep. 40:196-207. Lowry, M. S., C. W. Oliver, C. Macky. and J. B. Wexler. 1990. Food habits of California sea lions, Zalophus californianus, at San Clemente Island, California, 1981-1986. Fish. Bull. 88:509-521. Lowry, M. S., B. S. Stewart, C. B. Heath, P. K. Yochem, and J. M. Francis. 1991. Seasonal and annual variabilitv in the diet of California sea lions, Zalophus californianus. at San Nicolas Island, California, 1981-86. Fish. Bull. 89:331- 336. Mate, B. R. 1975. Annual migrations of the sea lions Eumetopias jubatus and Zalophus californianus along the Oregon USA coast. Rapp. P.-V. Reun. Cons. Int. Explor. Mer 169:455-461. Melin, S. R. 2002. The foraging ecology and reproduction of California sea lion (Zalophus californianus californianus). Ph.D. diss., 150 p. Univ. Minnesota, Minneapolis, MI. Melin, S. R„ and R. L. DeLong. 2000. At-sea distribution and diving behavior of California sea lion females from San Miguel Island. California. In Proceedings of the fifth California Islands symposium (D. R. Browne, K. L. Mitchell, and H. W. Chaney), p. 407-412. Prepared by MBC Applied Environmental Sciences, 3000 Redhill Ave., Costa Mesa, CA 92626 under MMS I Minerals Management Service) contract no. 14-35-01-96-RC-30801 from the U.S. Dep. Interior, MMS, Pacific OCS Region. Odell, D. K. 1975. Breeding biology of the California sea lion, Zalo- phus californianus. Rapports et Proces-verbaux de Reunions 169:374-378. Orr, R. T, and T C. Poulter. 1965. The pinniped population of Ano Nuevo Island, California. Proc. Calif. Acad. Sci. 32:377-404. Pierotti, R. J., D. G. Ainley, and T. J. Lewis. 1977. Birth of a California sea lion on Southeast Faral- lon Island. Calif. Fish Game 63:64-66. Sokal, R. R., and F J. Rohlf. 1995. Biometry: the principles and practice of statistics in biological research, 887 p. W.H. Freeman and Co., New York, NY. Sullivan, R. M. 1980. Seasonal occurrence and haul-out use in pinni- peds along Humboldt County, California. J. Mamm. 61:754-760. Sydeman, W. J., and S. G. Allen. 1999. Pinniped population dynamics in central California: correlations with sea surface temperature and upwelling indices. Mar. Mamm. Sci. 15:446-461. Weise, M. J. 2000. Abundance, food habits and annual fish consump- tion of California sea lions (Zalophus californianus) and its impact on salmonid fisheries in Monterey Bay, California. M.S. thesis, 103 p. Moss Landing Marine Laboratories, Moss Landing, CA, and San Jose State Univ., San Jose, CA. 344 Abstract— The narrow-barred Span- ish mackerel (Seomberomorus com- merson) is widespread throughout the Indo-West Pacific region. This study describes the reproductive biology of S. commerson along the west coast of Australia, where it is targeted for food consumption and sports fishing. Development of testes occurred at a smaller body size than for ovaries, and more than 90^ of males were sexually mature by the minimum legal length of 900 mm TL compared to 50f7f of females. Females dominated overall catches although sex ratios within daily catches vary consider- ably and females were rarely caught when spawning. Seomberomorus commerson are seasonally abundant in coastal waters and most of the commercial catch is taken prior to the reproductive season. Spawning occurs between about August and November in the Kimberley region and between October and January in the Pilbara region. No spawning activity was recorded in the more southerly West Coast region, and only in the north Kimberley region were large numbers offish with spawning gonads collected. Catches dropped to a minimum when spawning began in the Pilbara region, when fish became less abundant in inshore waters and inclement weather conditions limited fishing on still productive offshore reefs. Final maturation and ovulation of oocytes took place within a 24-hour period, and females spawned in the afternoon-evening every three days. A third of these spawning females released batches of eggs on consecu- tive days. Relationships between length, weight, and batch fecundity are presented. Variability in spawning frequency and reproductive development of the narrow-barred Spanish mackerel (Seomberomorus commerson) along the west coast of Australia Michael C. Mackie Paul D. Lewis Daniel J. Gaughan Stephen J. Newman Western Australian Marine Research Laboratories Department of Fisheries Government of Western Australia West Coast Drive Waterman, Western Australia 6020. Australia E-mail address (for M C Mackie) mmackietg'fish.wa. gov.au Manuscript submitted 1 October 2002 to the Scientific Editor's Office. Manuscript approved for publication 14 December 2004 by the Scientific Editor. Fish. Bull. 103:344-354 (2005). The narrow-barred Spanish mack- erel (Seomberomorus commerson) is a prized food fish targeted by fishermen throughout its range in the Indo-West Pacific region (Collette and Nauen, 1983). Reaching over 2.4 m in length and 45 kg in weight, this pelagic spe- cies is seasonally abundant in coastal waters where it often schools in large numbers. In Australian waters, the commercial mackerel fishery targets these schools using trolling methods, and 2362 metric tons were caught in 2001-02 for domestic and overseas markets (ABARE. 2003). Seomberomorus commerson is also a premier sport fishing species, tar- geted by an increasing number of rec- reational anglers throughout its broad Australian distribution. The combined commercial and recreation take of S. commerson has put significant pres- sure on stocks in Queensland (QLD) waters, leading to a possible decline in the spawning stock abundance (McPherson and Williams, 2002). The biology of S. commerson in these waters has been well studied (e.g., Munro, 1942; McPherson, 1981, 1992, 1993). Biological information is also available for S. commerson in wa- ters of the Northern Territory (NT; Buckworth1), where stocks are still recovering from a prolonged period of exploitation by foreign gill-net opera- tors that ended in 1986. In contrast, little is known about the stock status and biology of S. commerson in West- ern Australian (WA) waters, despite the fact that catches are similar to those taken in QLD and the NT, and commercial fishermen have expressed concern about increasing fishing pres- sure on this species in WA. Recent moves to overhaul management of the mackerel fishery in WA (in which S. commerson is the dominant species) have further highlighted the need for more information on the biology and stock status of S. commerson along the WA coast. Research to enable a stock assess- ment of S. co?7imerson in WA waters was therefore commenced in 1999. Description of reproductive biology was a key focus of this study, since this information is required for stock assessment models and for manage- ment controls such as minimum legal lengths, which were previously set with little knowledge of the biology of S. commerson in WA. Information on other reproductive parameters, such as batch fecundity and spawning be- havior, which are also required for 1 Buckworth, R. C. 1999. Age structure of the commercial catch of Northern Ter- ritory narrow-barred Spanish mackerel. Final Report to the Fisheries Research and Development Corporation (FRDC) on project no. 1998/159. Fishery report 42, 27 p. Department of Business Indus- try and Resource Development, Darwin, Northern Territory, 0800, Australia. Mackie et al.: Variability in reproductive development of Spanish mackerel (Scomberomorus commerson) 345 stock assessments, is unavailable or insufficiently described in the literature for this species. The ob- jective of our study was, therefore, to provide a com- prehensive description of the reproductive biology of S. commerson in Western Australian waters. Material and methods Collection and processing of samples Scorn beromorus commerson were collected onboard vessels operating from a number of locations along the WA coast between 1998 and 2002 (Fig. 1). These locations were pooled into three regions to reflect differences in fishing methods within the mackerel fishery (Kimberley — east of 120°E, Pilbara — north of 23CS to the Kimberley border, and West Coast — south of 23°S; Fig. 1). Scomberomorus commerson are seasonally abundant in coastal waters although low numbers are caught in the Pilbara region during the "off-season." Samples were therefore collected throughout the year from this region only. Fresh S. commerson collected from commercial and recreational fishermen were measured (total length |TL] and fork length |FL] in mm) and, where possible, weighed to 0.1 kg (whole weight |WW] and clean weight (viscera and gonads removed]). Heads were removed and measured from tip of the mouth to firm edge of the operculum (mm), and weighed with gills intact (±0.1 gm). Gonads were removed from the fish within hours of capture, macroscopi- cally staged (see below), weighed where possible (±0.01 g), and preserved in 10% formalin and seawater solution. Frozen head and viscera obtained from commercial and recreational fishermen were also measured and weighed as above. The thawed gonads were macroscopically staged by using a simplified staging system (see below) that is used in less detailed reproductive analyses. Preserved gonads were blotted dry with a paper towel and weighed. A 4-mm slice from the mid-region was processed by using standard histological techniques and stained with Harris's haematoxylin and eosin for microscopic examination. Full details of methods used in the collection and analysis of S. commerson gonads are provided in Mackie and Lewis.2 Biological analyses Gonads were staged macroscopically and microscopi- cally. Macroscopic staging employed five developmental steps that were compatible with the microscopic staging system (Mackie and Lewis2): Pilbara Kimberley >y 'A Broome Port Hedland WESTERN AUSTRALIA West Coast •Geraldton NORTHERN TERRITORY Figure 1 Sampling locations used in the study of the narrow-barred Spanish mackerel {Scomberomorus commerson) reproductive biology. Juvenile (J) Females stage 1 stages 2-3 stage 4 stage 5 Males stage 1 stage 2 stage 3 stage 4 undifferentiated. immature ("virgin mature, resting; reproductively developed; spawning ("running, ripe studies). in other studies); in other - Mackie, M. C, and P. D. Lewis. 2001. Assessment of gonad staging systems and other methods used in the study of the reproductive biology of narrow-barred Spanish mackerel, Scomberomorus commerson, in Western Australia. Fisheries Research Report 136, 25 p. Department of Fisheries, Perth, Western Australia 6020, Australia, http://www.fish.wa.gov. au/res/broc/frr/frrl36/index.html. [Accessed January 15 2002.] immature ("virgin" in other studies); mature resting; reproductively developed, ripe; spawning ("running, ripe" in other studies). The microscopic staging system had more stages and allowed a more detailed description of spawning: Juvenile (J) undifferentiated. Females stage 1 immature ("virgin" in other studies); stage la immature, developing; stage 2 mature, resting; stage 3 mature, developing; stage 4 reproductively developed; stage 5a prespawning; stage 5b spawning ("running, ripe" in other studies); stage 5c postspawning; stage 6 spent. 346 Fishery Bulletin 103(2) Males stage 1 immature ("virgin" in other studies); stage la immature, developing; stage 2 mature, resting; stage 3 reproductively developed, ripe; stage 4 spawning. The immature, developing stage identified females that were immature and unlikely to spawn but had ovaries containing cortical alveoli stage oocytes (which other- wise identified mature, developing females). Division of the microscopic staging system for ovaries into three spawning stages was based on the pres- ence of migratory nucleus stage or hydrated oocytes within the ovarian lamellae (stage 5a), the presence of hydrated oocytes within the ovarian lumen (stage 5b), and the presence of postovulatory follicles (POFs) in the lamellae (stage 5c). In tropical fish species POFs may remain up to 24 hours in the ovaries before being resorbed (West, 1990), and there is evidence suggesting this is the case for S. commerson in Queensland waters (McPherson, 1993). In the present study POFs observed in the ovaries of females were categorized as either "new" or "old" based on their degree of degeneration (Mackie and Lewis2). Gonadosomatic indices (GSIs) were calculated by us- ing ratios of gonad weight to whole body weight, head weight, and head length. The latter two ratios were used to assess the usefulness of head and viscera sam- ples in future monitoring of S. commerson. Scomberomorus commerson is a serial-spawning spe- cies (Munro, 1942). Estimates of batch fecundity were made for preserved prespawning (stage 5a) ovaries from counts of hydrated oocytes within three samples taken from the anterior, middle, and posterior region of one lobe (each 130-200 mg). A section of each ovary was also processed by using histological methods to confirm suitability for estimation of fecundity. Some ovaries were subsequently rejected for fecundity estimates be- cause the most mature batch of oocytes had not fully hydrated and were less easy to distinguish from earlier stage oocytes. These ovaries tended to provide an over- estimate of batch fecundity (Mackie et al.3). The daily timing and frequency of spawning were determined for females captured in the Kimberley re- gion during September 1999 when 94% of ovaries were retained for histological analysis (« = 344). Spawning frequency was determined as the inverse of the spawn- ing fraction (the number of ovaries with hydrated or migratory nucleus stage (MNS) oocytes divided by the total number of mature ovaries in the catch). These data were compared with estimates made by using the number of ovaries macroscopically identified as having 3 Mackie, M. C, D. J., Gaughan, and R. C. Buckworth. 2003. Stock assessment of narrow-barred Spanish mackerel iScomberomorus commerson) in Western Australia. Final report to the Fisheries Research and Development Corpora- tion (FRDC) on project no. 1999/151, 242 p. Department of Fisheries, Perth, Western Australia, 6020. hydrated oocytes. Analyses of sex ratios were based on data where the whole catch or a known random sample of the catch was processed. Results The gonads of 5128 male, female, and juvenile S. com- merson were macroscopically staged during this study. Of these, 1624 were also processed with histological techniques for more detailed analyses. Biological analyses Body lengths ranged from 58 to 1720 mm FL (62 to 1840 mm TL), and whole weights ranged from 0.0015 to 40.6 kg. Regression analyses incorporating step-wise reduc- tion (using analysis of variance) of a fully parameterized model indicated that differences in length and weight relationships between regions and sex were minor com- pared to measurement error. Thus, the simplest models which adequately explain the pooled data were Whole weight (kg) = 3.40e - 9 x FL (mm)3 12 (re=2842) (SE of constants: a=2.78e-10, 6 = 0.01) TL 'mm) = 42.74 + (1.06 x FL (mm)) (« = 1679, r2=0.996). Overall sex ratios were biased towards females, with the M:F ratio varying between 1:1.2 and 1:1.6 in the three regions. However, there was considerable varia- tion between samples, from a peak M:F ratio of 1:2.6 for samples obtained in the nonreproductive period, to a male bias of 1.1:1 in pooled samples obtained during the peak spawning period. This slight male bias during the spawning period occurred in successive years; the sex bias, however, was variable between daily samples. Sex ratios also changed over from a male to female bias with increasing size class, with a 1:1 ratio occurring at about 1000-1050 mm FL. Ovarian weight ranged from 2.00 to 1908.30 g and testes from 0.84 to 840.10 g. Gonads of juvenile S. com- merson were small and contained no recognizable germ tissue. The smallest fish with differentiated gonads was a 301-mm-FL male. The smallest female was 396 mm FL. Two abnormally large juveniles (1170 and 1251 mm FL) were captured whose gonads had remained unusually small and undifferentiated. Body lengths of immature females (largest=1195 mm FL [13.8 kg WW]) overlapped substantially with those of mature females (smallest=641 mm FL [2.3 kg WW]). Estimates of the size at which 50% of females were mature were calculated by using all available data as well as data taken only during the reproductive season (October to April). Data for each area were pooled to provide sufficient samples (virtually all samples of im- mature fish were obtained from the Pilbara region). Both data sets provided similar estimates; 809 mm FL, ±9.8 SE (898 mm TL) for all data, and 788 mm Mackie et al.: Variability in reproductive development of Spanish mackerel (Scomberomorus commerson) 347 Females 150-, 400 600 1 1 1 r 000 1200 1400 1600 140- Males %^ . • ••. 120- /• • 100" 80 - • / / • * ■ * I 60 - i * 1 -•- — ■ ■ ■ ■ A • . [\ , 40 - 20 " ■e ■ JPBe / i / i fE^cnjrTkkf lfTk__ I 1 1 1 1 1 1 1 1 1 — 100 200 300 400 500 600 700 800 900 1000 O 1 0 I c - 0 4 1200 1400 Length Figure 2 Proportion of mature female and male S. commerson within samples. The number offish within each length class is indicated by the verti- cal bars and the left v-axis and the proportion of mature fish by the black circles and the right y-axis. The length at which 50% of fish within the samples were mature is indicated on the fitted maturity curve (±95% CI). Lengths are fork length in mm. Note that data for juvenile fish of undifferentiated sex are included in both graphs. The dashed lines indicate the length at 50% mature (P=0.5). FL (±14.5 SE) for data taken during the reproductive period. The size at which 10% of females were mature was 638 mm FL (±19.6 SE), with 90% mature by 981 mm FL (±7.2 SE) (Fig. 2A). There was also considerable overlap between the lengths of immature and mature males. The largest im- mature male was 1140 mm FL (11.3 kg WW), whereas the smallest mature male (stage 3) was 491 mm FL (1.0 kg WW). The size at which 10% of males were mature was 465 mm FL (±24.9 SE), the size at which 50% of males were mature was 628 mm FL (±13.8 SE) or 706 mm TL, and the size at which 90% were mature was 791 mm FL (±10.5 SE) (Fig. 2B). Development of oocytes is asynchronous and all stag- es of oocytes are present at the same time within re- productively active ovaries. This reproductive feature, along with the maturation of multiple batches of oocytes (as evidenced by presence of both POFs and hydrated or MNS oocytes in spawning ovaries), confirms that female S. commerson are serial or partial spawners (Hunter et al., 1985). Relationships between batch fecundity and body parameters were obtained from counts of hydrated oocytes within prespawning (stage 5a) ovaries. Size of females for which batch fecundity was determined ranged from 857 to 1143 mm FL and from 5.3 to 348 Fishery Bulletin 103(2) B Kimberley Month Resting (F 2 and 3) Developing (F4) Spawning (F5 a b) Water temp Figure 3 Annual cycle of Scombero?norus commerson reproduction within each region, as indicated by macroscopically staged ovaries. Mid-month sea surface temperatures are overlaid (solid line) and sample sizes are shown above each column. 12.7 kg WW. Both relationships were explained with power curves: Batch fecundity Batch fecundity ■■ 0.0011 x FL2896 (r2=0.441, n = 2l) 31087 x WW1™4 (r2=0.714, n = 19). Annual reproductive cycle Female S. commerson within the Pilbara region were non- reproductive between March and June, during the down- ward cycle of water temperatures (Figs. 3 and 4). As water temperatures reached a minimum in July and August (around 24°C), a small proportion of mature ovaries had become reproductively developed (stage 4). The proportion of developed ovaries during September (the start of the upward cycle of water temperatures) varied noticeably between years in the Pilbara region, from 18.5% to 79% in 2001 and 2000, respectively. A small number of females were also actively spawning when sampled during Sep- tember 2000. Peak reproductive activity extended from October to January, and spawning females were captured during this period in 1999 and 2000 when the sea surface temperature (SST) was rising from about 25.5° to 28.5°C. By February, when SST peaked at approximately 30°C, reproductive development was declining and the ovaries of most females were spent or resting. Mackie et al.: Variability in reproductive development of Spanish mackerel (Scomberomorus commerson) 349 A Pilbara C West Coast Month r i Resling (F 2) I I Developing (F3| Pre-spawn (F5a) f7\\1 Post spawning (F5c) Developed (F4) Spenl(F6) Water temp Figure 4 Annual cycle of Scomberomorus commerson reproduction in Western Australia for (A) Pilbara, (Bl Kimberley, and (Cl West Coast regions, as indicated by histologically staged ovaries. Mid-month sea surface temperatures are overlaid (solid line) and sample sizes are shown above each column. The annual reproductive cycle of ovaries in the Kimberley region follows a similar pattern to that in the Pilbara region (Figs. 3 and 4). However, because 30-50% of females captured in Kimberley waters dur- ing September 1999 and 2000 were actively spawning, it appears that S. commerson commence spawning at least one month earlier in this region. About 60% of females were also spawning when sampled during Oc- tober 1999 and 2000, although only 35% were spawning during this month in 2001. If spawning in the Kim- berley region commenced in August and concluded in November (same duration as in the Pilbara region), the associated SST ranged from approximately 26.5-27°C to 29-30°C (annual maximum approx 30-31cC; Figs. 3 and 4). Sampling of developed ovaries in March also indicated that the reproductive period for S. commerson in the Kimberley region may be more protracted than in the Pilbara region. In the West Coast region few reproductively devel- oped ovaries and no spawning ovaries were obtained; S. commerson are rarely captured in this region dur- ing the peak spawning period observed in the northern regions. The maximum sea surface temperature (SST) in this region of around 28°C is above the lower tem- perature range of spawning in the two northern regions. Reproductively developed ovaries obtained from the 350 Fishery Bulletin 103(2) A Pllbara (HL-97. WW-304. HW-232) g ■u £ 1 B Kimberley (HL-31 9. WW-443. HW-40) I I • 4 - C West Coast (HL-304, WW-97, HW-232) ■ 2 - 1 - 0 - .4 ^. • • • • * • •*T- • Date HL index WW index HW index Figure 5 Annual cycle of gonad indices for Scomberomorus commerson in Western Australia for (A) Pilbara, (B) Kimberley, and (Cl West Coast regions. Sample sizes are given for each index, where HL = head length index, WW=whole weight index, and HW=head weight index. West Coast region were collected over a range of SSTs, including when it was at a minimum (Figs. 3 and 4). Gonadosomatic indices calculated from whole weight, head weight, and head length exhibited similar patterns and confirmed the spawning cycle determined from ex- amination of ovaries (Fig. 5). The most complete data set was for females from the Pilbara region. In this re- gion indices were minimal between March and August and increased considerably during September as ovaries became reproductively developed. Peak indices occurred in November 1999 and October 2000, coinciding with peaks in the proportion of spawning (stage 5) ovaries in the samples. The drop in gonad indices during De- cember 1999 showed that the supplies of vitellogenic oocytes within the ovaries were reduced by this time, even though many females were still spawning (Fig. 4). This drop continued until March when all the ovaries in the samples were in the resting stage. Data for 2001 indicate that decreased GSIs during the reproductive season were comparable to data from the previous two Mackie et al.: Variability in reproductive development of Spanish mackerel (.Scomberomorus commerson) 351 Table 1 Ovarian development of Scomberomorus commerson sampled in the Kimberley region during the spawning season. POFs = postovulatory follicles. Ovaries in prespawning. spawning, postspawning, and spent stages of development are indicated by 5a, 5b, 5c, and 6, respectively. Note that data for stage 5c includes only females that had spawned on the day of capture (i.e., exclud- ing ovaries containing old POFs only). Data for "Old POFs" includes all ovaries containing old POFs as well as other evidence of recent or imminent spawning. Total Year caught Histological analysis Morning Afternoon Number Total mature Total 5a 5b 5c Old Old POFs Total 5a 5b 5c POFs 1999 344 325 306 171 59 0 0 70 135 1 1 23 51 2000 406 115 103 59 22 0 0 21 44 0 0 15 13 years. Data for the Kimberley and West Coast regions were limited but concurred with gonad staging data and also confirmed the low reproductive status of S. commerson within the West Coast region. Spawning Evidence of spawning was found in 237 of the histologi- cally processed ovaries. Thirty-eight percent (;? = 90) of these were about to spawn when captured (stage 5a), 62% (n=147) had recently spawned (stage 5c), and one was running, ripe (stage 5b). The ovaries of only two macroscopically staged females were also running, ripe. Most of these spawning fish (n=219) were captured in the north Kimberley region (eighteen from the Pilbara region). The most southern location from which a spawn- ing female was obtained was Exmouth (one recently spawned fish), and no females captured in the West Coast or more southern regions showed histological (or macroscopic) evidence of spawning. Spawning females collected during 1999 and 2000 were either prespawning (stage 5a) and caught in the morning, or had recently spawned (stage 5c) and were caught in the afternoon (Table 1). The absence of hy- drated oocytes in the afternoon and new POFs in the morning showed that the entire cycle of oocyte matu- ration, ovulation, and spawning is completed within a 24-hour period. Because no new POFs were present in ovaries sampled during the morning the transition from new to old POFs occurs during the night, within about 12 hours of spawning. The lack of evidence to show that females spawned on more than two consecu- tive days indicates that old POFs are unrecognizable after 24 hours. Spawning fraction was estimated by using data ob- tained in the Kimberley region during September 1999 when 95% (n=344) of ovaries were examined by using histological methods. Analyses were based on the num- ber of prespawning (stage 5a) ovaries sampled during the morning (usually between 0600-0900 h). Afternoon samples (usually 1500-1800 h) were not used because the number of spawning fish was likely to be under- estimated because of the low catchability of running, ripe (stage 5b) females. Thirty-five percent (n = 59) of mature females in the morning samples were about to spawn (stage 5a). Spawning frequency was therefore 2.9 days. Comparison of spawning fractions in samples of at least ten females showed higher spawning fractions (33-56%) for the Kimberley region compared with the Pilbara region (4-28%). Spawning fraction was also estimated for the morn- ing samples as the proportion of macroscopically staged mature ovaries that contained hydrated oocytes. Thirty- one of the 180 mature females were identified as such, providing an estimated spawning fraction of 17.2%, and a spawning frequency of 5.8 days. Thirty-six percent (n=54) of spawning females (stages 5, a-c) had spawned on two consecutive days. For exam- ple, 39 ovaries contained oocytes in the MNS or hydrat- ed stage of development (i.e., spawning was imminent when fish were captured) and also contained old POFs. Another 15 ovaries had both old and new POFs. Discussion Scomberomorus commerson has a gonochoristic life his- tory in which the gonad differentiates into an ovary or testis at around 300-400 mm FL. Males differentiate and reach sexual maturity at a smaller body size than females, as is the case with the congeneric species S. maculatus and S. cavalla (Beamariage, 1973; Schmidt et al., 1993). Consequently, more than 90% are sexually mature by the time the minimum legal length of 900 mm TL is reached in the fishery. In contrast only 50%> of females are mature at 898 mm TL. Although mortality of released undersize fish may be high because of dif- ficulties in removing fishing hooks, this size limit deters fishermen from targeting small fish and relatively few are captured (Mackie et al.3). Biases in sex ratios have been observed in several species of Scomberomorus (e.g. Trent et al., 1981; Sturm 352 Fishery Bulletin 103(2) and Salter, 1990; Begg, 1998). In the case of S. com- merson, females usually dominate size classes above the MLL because they grow faster and reach a larger maximum size (McPherson, 1992; Mackie et al.3). How- ever, they are rarely caught when actively spawning, despite observations by fishermen of leaping fish that might indicate that they are still present on the fish- ing grounds. Regional differences in fishing gear can also affect catchability. The lighter monofilament and reel outfits used in the Pilbara and West Coast regions likely catch larger fish than the heavier rope and thick monofilament hand-hauled rigs used by Kimberley fish- ermen that do not allow the fish to be "played" (i.e. do not allow the fish to swim) and may result in more gear failure. The spring-summer spawning pattern observed in our study is similar to that of S. commerson along the east coast of Australia (McPherson, 1981). Water tem- perature may influence spawning in fish by affecting gametogenesis, gonad atresia, and spawning behav- ior (Lam, 1983). In WA waters S. commerson spawn as water temperatures are rising and, as found in Queensland, may compensate for latitudinal differences in temperatures by spawning earlier in northern waters (McPherson, 1981). No evidence of spawning activity was found within the West Coast region although the annual range of water temperatures overlap with those in which spawning occurs farther north. Restricted spawning by S. commerson on the east coast occurs at similar latitudes to northern parts of the West Coast region, and anecdotal evidence suggests that spawning may be restricted in some years in this region. During the spawning period the average female S. commerson may spawn every three days and about one third of fish spawn on consecutive days. Female fish similarly spawn every 2-6 days and possibly on con- secutive days in Queensland waters (McPherson, 1993). Our study showed that estimates based on the fraction of histologically staged prespawning (stage 5a) ova- ries provided the best estimate of spawning frequency. However, only samples taken during the morning can be used for this analysis because of decreased catchabil- ity of running, ripe females in the afternoon. In com- parison, macroscopic staging of ovaries with hydrated oocytes underestimated spawning frequency because migratory nucleus oocytes (which comprised 54% of histologically staged, prespawning [stage 5a] ovaries) cannot be identified. It is also impossible to identify fish that have spawned on more than one occasion with mac- roscopic criteria, resulting in a further underestimate of spawning activity (by 25% for S. commerson). Maturation, ovulation, and spawning of oocytes by female S. commerson was completed within a 24-h cycle in the Kimberley region compared to 24-36 hours in Queensland waters (McPherson, 1993). Maturation of the oocytes is underway by sunrise and probably com- pleted in all spawning ovaries by mid to late morning to allow for ovulation prior to spawning in the afternoon. Few samples were obtained at or after dusk because fish are generally not catchable, indicating that a high incidence of spawning at this time because only one spawning fish was obtained during the study. Dusk spawning is prominent among pelagic spawning species that inhabit tropical reefs (Thresher, 1984). However, spawning in the afternoon is less common and may be linked to large tidal cycles and strong currents in the north of WA, as indicated for the brown stripe snapper (Lutjanus vitta) that also spawns in the afternoon in the Pilbara region (Davis and West, 19931. Batch fecundity of S. commerson has not previously been recorded and such data are rare for other Scomb- eromorus species. Fecundity estimates for S. commerson from the Indian Peninsula (Devaraj, 1983) were not comparable because those data appeared to be obtained from counts of both vitellogenic and previtellogenic oo- cytes. Although the current study provided fecundity estimates only for females up to 13 kg whole weight, it shows that S. commersorj is highly fecund (the highest estimated batch fecundity of 1.2 million eggs was ob- tained from an ovary that was less than half the weight of the heaviest ovary sampled). This study highlighted the need to histologically check that oocytes in the spawning batch are fully hydrated because fecundity may otherwise be over-estimated. Similarly, fecundity will be under-estimated if ovulation has commenced. The best time to collect gonad samples so that these biases are minimized is during the mid to late after- noon for this species. Fishing activity is also regulated by the reproductive cycle. About 3-6 months prior to the spawning sea- son catches of S. commerson by commercial fishermen increase as large numbers of smaller S. commerson appear on offshore reefs sometime between March and May, and soon after throughout the coastal waters of WA (Mackie4). By the time reproductive development in ovaries begins (approximately August and September in the Kimberley and Pilbara regions, respectively) catches have peaked or are declining. In the Pilbara region commercial catches have dropped to a minimum when spawning begins, as fish become less abundant in inshore waters and inclement weather conditions limit fishing on still productive offshore reefs. Because S. commerson generally do not make substantial long- shore movements (Buckworth et al.5), it is likely that most spawning activity occurs at offshore locations in this region (e.g., in mid to outer areas of the continen- tal shelf), although anecdotal evidence indicates that 4 Mackie, M. C. 2001. Spanish mackerel stock status report. In State of the fisheries report 1999/2000 (J. W. Penn, W. J. Fletcher, and F. Head, eds.), p. 71-75. Depart- ment of Fisheries, Perth, Western Australia, 6020. http:// www.fish.wa.gov.au/sof/1999/comm/nc/commnc26.html. Accessed 10/2/2001. 6 Buckworth, R. C, S. J. Newman, J. R. Ovenden, R. J. G. Lester, and G. R. McPherson. 2004. In prep. The stock structure of northern and western Australian Spanish mack- erel. Final report to the Fisheries Research and Development Corporation (FRDC) on project no. 1998/159. Department of Business Industry and Resource Development, Darwin, Northern Territory, 0800, Australia. Mackie et al.: Variability in reproductive development of Spanish mackerel (Scomberomorus commerson) 353 large, more solitary individuals may spawn in inshore waters. Catches of S. commerson peak and decline rapidly along the Kimberley coast during the main spawning period because of declining fish abundance and weather conditions. As in the Pilbara region, few S. commerson are caught at this time in southern or midsections of the Kimberley coast. Fishermen must therefore under- take extensive trips north to the remaining productive grounds located between 12.5° and 15°S latitude where the majority of S. commerson spawning activity was encountered in the present study. Although it is pos- sible that S. commerson in other areas of the Kimberley region may move offshore to spawn, it is also possible that some move northward, mixing and spawning with otherwise temporally and spatially discrete northern populations, in a similar manner to S. cavalla in U.S. waters (Broughton et al., 2002). Monitoring of the WA fishery for S. commerson is likely to be based on the collection of head and gonad samples because limited funding and large distances will restrict future research trips. Onboard storage of filleted frames for research purposes is also prohibited by the large body size of S. commerson. In contrast, the head of this species is relatively small and easy to store, and as shown in the present study, provides a general measure of reproductive activity through calculation of head-to-gonad ratios. These ratios can also be supple- mented by staging the gonads by using the macroscopic staging system developed for this species (Mackie and Lewis2). Head length can also be used to estimate body length of S. commerson (Mackie et al.3) and the otoliths contained in the head can be used to determine age. Although data gathered by such means is less accu- rate than that obtained from whole, fresh samples, it presents the best option for gathering ongoing data in sufficient quantities for meaningful analyses. Acknowledgments The authors thank the numerous commercial and rec- reational fishermen who assisted in the collection of samples and provided invaluable advice. The assistance of Department of Fisheries staff and volunteers on field trips is also appreciated. We also thank Rod Lenanton, Rick Fletcher, and Peter Stephenson for reviewing the manuscript, and to the Fisheries Research and Develop- ment Corporation for funding Project 1999/151, of which this study formed a part. Literature cited ABARE (Australian Bureau of Agricultural and Resource Economics). 2003. Australian fisheries statistics 2002, 69 p. ABARE. Canberra, New South Wales, Australia. Beaumariage, D. S. 1973. Age, growth and reproduction of king mackerel Scomberomorus cavalla, in Florida. Florida Mar. Res. Pub. 1:1-45. Begg, G. A. 1998. Reproductive biology of school mackerel {Scomb- eromorus queenslandicus) and spotted mackerel (S. munroi ) in Queensland east-coast waters. Mar. Freshw. Res. 49:261-270. Broughton, R. E., L. B. Stewart, and J. R. Gold. 2002. Microsatellite variation suggests substantial gene flow between king mackerel {Scomberomorus cavalla I in the western Atlantic Ocean and Gulf of Mexico. Fish. Res. 54:305-316. Collette, B. B., and C. E. Nauen. 1983. FAO Species catalogue. Vol 2: scombrids of the world. FAO Fish. Synop. 125:1-137. Davis, T. L. 0„ and G. J. West. 1993. Maturation, reproductive seasonality, fecundity, and spawning frequency in Lutjanus vittus (Quoy and Gaimard) from the North West Shelf of Australia. Fish. Bull. 91:224-236. Devaraj, M. 1983. Maturity, spawning and fecundity of the king seer, Scomberomorus commerson (Lacepede), in the seas around the Indian Peninsula. Ind. J. Fish. 30(21:203- 230. 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 bio- mass of pelagic fish: application to the northen anchovy (Engraulis mordax) (R. Lasker, ed.), p. 67-77. NOAA Tech. Rep. NMFS 36. Lam, T. J. 1983. Environmental influences on gonadal activity in fish. In Reproduction, part B, Behaviour and fertility control (W. S. Hoar, D. J. Randall, and E. M. Donaldson, eds. I, p. 65-115. Academic Press, New York and London. McPherson, G. R. 1981. Preliminary report: Investigations of Spanish mackerel (Scomberomorus commerson) in Queensland waters. In Northern pelagic fish seminar (D. J. Grant and D. G. Walter, eds. I, p. 51-58. Australian Govern- ment Publishing Service, Canberra, New South Wales, Australia. 1992. Age and growth of the narrow-barred Spanish mackerel {Scomberomorus commerson Lacepede, 1800) in north-eastern Queensland waters. Aust. J. Mar. Fresh. Res. 43:1269-1282. 1993. Reproductive biology of the narrow-barred Spanish mackerel (Scomberotnorus commerson Lacepede, 1800) in Queensland waters. Asian Fish. Sci. 6:169-182. McPherson, G. R., and L. E. Williams. 2002. Narrow-barred Spanish mackerel. In Queensland's fisheries resources: current condition and recent trends 1988-2000 (L. E. Williams, ed.), p. 88-93. Informa- tion Series QI02012, Department of Primary Industries Queensland, Brisbane, Australia. Munro, I. S. R. 1942. The eggs and early larvae of the Australian barred Spanish mackerel, Scomberomorus commersoni (Lace- pede) with preliminary notes on the spawning of that species. Proc. Royal Soc. Qld. 54:33-48. Schmidt, D. J., M. R. Collins, and D. M. Wyanski. 1993. Age, growth, maturity and spawning of Spanish mackerel, Scomberomorus maculatus (Mitchell), from the Atlantic coast of the southeastern United States. Fish. Bull. 91:526-533. 354 Fishery Bulletin 103(2) Sturm, M. G. deL., and P. Salter. 1990. Age, growth and reproduction of the king mack- erel, Scomberomorus eavalla (Cuvier), in Trinidad waters. Fish. Bull. 88:361-370. Thresher, R. G. 1984. Reproduction in reef fishes. 391 p. T.F.H. Pub- lications Pty Ltd., Neptune City. NJ. Trent, L„ R. O. Williams, R. G. Taylor, C. H. Saloman. and C. S. Manooch. 1981. Size and sex ratio of king mackerel, Scomberomorus eavalla, in the southeastern United States. NOAA Tech. Memo. NMFS-SEFC 62:1-59. West, G. 1990. Methods of assessing ovarian development in fishes: areview. Aust. J. Mar. Freshw. Res. 41:192-222. 355 Abstract— Recent research demon- strated significantly lower growth and survival of Bristol Bay sockeye salmon (Oncorhynchus nerka) during odd-numbered years of their second or third years at sea (1975, 1977, etc.). a trend that was opposite that of Asian pink salmon (O. gorbuscha) abundance. Here we evaluated sea- sonal growth trends of Kvichak and Egegik river sockeye salmon (Bristol Bay stocks) during even- and odd- numbered years at sea by measur- ing scale circuli increments within each growth zone of each major salmon age group between 1955 and 2000. First year scale growth was not significantly different between odd- and even-numbered years, but peak growth of age-2. smolts was sig- nificantly higher than age-1. smolts. Total second and third year scale growth of salmon was significantly lower during odd- than during even- numbered years. However, reduced scale growth in odd-numbered years began after peak growth in spring and continued through summer and fall even though most pink salmon had left the high seas by late July (10-18% growth reduction in odd vs. even years). The alternating odd and even year growth pattern was consis- tent before and after the 1977 ocean regime shift. During 1977-2000, when salmon abundance was rela- tively great, sockeye salmon growth was high during specific seasons com- pared with that during 1955-1976, that is to say. immediately after entry to Bristol Bay, after peak growth in the first year, during the middle of the second growing season, and during spring of the third season. Growth after the spring peak in the third year at sea was relatively low during 1977-2000. We hypothesize that high consumption rates of prey by pink salmon during spring through mid- July of odd-numbered years, coupled with declining zooplankton biomass during summer and potentially cyclic abundances of squid and other prey, contributed to reduced prey availabil- ity and therefore reduced growth of Bristol Bay sockeye salmon during late spring through fall of odd-num- bered years. Manuscript submitted 7 April 2004 to the Scientific Editor's Office. Manuscript approved for publication 14 December 2004 by the Scientific Editor. Fish. Bull. 103:355-370 (2005). Seasonal marine growth of Bristol Bay sockeye salmon (Oncorhynchus nerka) in relation to competition with Asian pink salmon (O. gorbuscha) and the 1977 ocean regime shift Gregory T. Ruggerone Natural Resources Consultants, Inc. 1900 West Nlckerson Street. Suite 207 Seattle, Washington 98(19 E-mail address GRuggeronein'nrccorp.com Ed Farley National Marine Fisheries Service 11305 Glacier Highway Juneau, Alaska 99801 Jennifer Nielsen Biological Resources Division U.S. Geological Survey Anchorage, Alaska 99503 Peter Hagen Alaska Dept. of Fish and Game P.O. Box 25526 Juneau, Alaska 99802-5526 Competition among Pacific salmon {Oncorhynchus spp.) for food resources in the North Pacific Ocean and Bering Sea is a potentially important mech- anism affecting salmon growth and population dynamics. Reduced growth at sea may lead to delayed matura- tion (Rogers, 1987), lower reproductive potential (Groot and Margolis, 1991), or greater risk of predation (Juanes, 1994). Density-dependent growth in the ocean has been observed among sock- eye (O. nerka), pink (O. gorbuscha), and chum salmon (O. keta), which are the most abundant species among Pa- cific salmon (Rogers1; Eggers et al.2). Density-dependent growth may occur during early marine life (Peterman, 1984) or during the homeward mi- gration period when the potential for high growth rate (Ishida et al., 1998) may be influenced by high concen- trations of salmon (Rogers and Rug- gerone, 1993). Since the early 1970s, salmon abundance in the North Pacific Ocean has increased, whereas body size for many populations of all salmon spe- cies has declined (Bigler et al., 1996). However, greater abundance of adult sockeye salmon returning to Bristol Bay, Alaska, was associated with in- creased growth during the first and second years at sea, followed by rela- tively low growth during the third year at sea, and greater adult size at a given abundance (Ruggerone et al.. 1 Rogers, D. E. 2001. Estimates of annual salmon runs from the North Pacific, 1951-2001. Report SAFS-UW- 0115. 11 p. School of Aquatic Sciences, Univ. Washington, Seattle, WA. 2 Eggers, D. M, J. Irvine, M. Fukawaki, and V. Karpenko. 2003. Catch trends and status of North Pacific salmon. Doc. no. 723, 34 p. North Pacific Anadromous Fisheries Commission (NPAFC), 889 Pender Street, Vancouver, Canada. 356 Fishery Bulletin 103(2) 2002). Increased growth of Bristol Bay sockeye salmon during the first two years at sea was associated with greater adult returns, but high abundance apparently led to increased competition and reduced growth during the third year. The potential for competition for food between Asian pink salmon and Bristol Bay sockeye salmon stocks is great in the North Pacific Ocean and Bering Sea. Tro- phic level, diet, and feeding behavior of pink salmon overlap significantly with sockeye salmon (Welch and Parsons, 1993; Davis et al., 2000; Kaeriyama et al., 2004). Asian pink salmon are highly abundant, averag- ing approximately 162 million adults in odd-numbered years and 104 million adults in even-numbered years, 1955 to 2000 (Rogers1). Bristol Bay sockeye salmon and Asian pink salmon overlap in the central North Pacific Ocean and the Bering Sea. Greatest overlap is with pink salmon from the eastern Kamchatka Peninsula and Sakhalin Island (French et al., 1976; Takagi et al., 1981; Myers et al.3), which are especially abundant, as shown by average harvests of 79,000 metric tons (t) in odd-numbered years and 33,000 t in even-numbered years, 1955-99 (Sinyakov, 1998; Anonymous4). Evidence for competition between Asian pink and Bristol Bay sockeye salmon was provided in a recent in- vestigation by Ruggerone et al. (2003). During 1955-97, annual sockeye salmon scale growth during the second and third years at sea was significantly reduced during odd- compared to even-numbered years. Adult sockeye salmon length was relatively low when sockeye salmon overlapped with abundant odd-year pink salmon during the year prior to homeward migration. Furthermore, smolt-to-adult survival of Bristol Bay sockeye salmon was significantly lower when they encountered odd-year pink salmon during the second year at sea. However, Bristol Bay sockeye salmon encountered relatively few pink salmon during their first year at sea and no com- petition effect was observed during this early marine period. In our study we examined the seasonal growth of Bristol Bay sockeye salmon scales in an effort to deter- mine the approximate timing and duration of reduced growth during odd-numbered years at sea that was observed by Ruggerone et al. (2003). Scale circuli in- crements and annuli are correlated with salmon body size (Clutter and Whitesel, 1956; Fukuwaka and Kaeri- yama, 1997; Fukuwaka, 1998). We compared seasonal scale growth before and after 1977 to examine seasonal growth trends associated with the twofold increase in Bristol Bay sockeye salmon abundance and the 1977 :) Myers, K. W., K. Y. Aydin, R. V. Walker, S. Fowler, and M. L. Dahlberg. 1996. Known ocean ranges of stocks of Pacific salmon and steelhead as shown by tagging experiments, 1956-1995. Report FRI-UW-9614, 159 p. School of Aquatic and Fishery Sciences, Univ. Washington, Seattle, WA 4 Anonymous. 2002. Biostatistical information on salmon catches, escapement, outmigrants number, and enhancement production in Russia in 2001. Doc. no. 646, 14 p. NPAFC, 889 Pender Street, Vancouver, Canada. ocean regime shift (Rogers, 1984; Beamish and Bouil- lon, 1993; Rogers1). We also examined the hypothesis that seasonal growth during the second growing sea- son was dependent on previous marine growth (Aydin, 2000). These hypotheses were tested by using scales from Kvichak River and Egegik River sockeye salmon, which averaged approximately 16 million fish per year or approximately 57% of the annual sockeye salmon run to Bristol Bay, 1955-2000. Methods For our study, we used scales from four age groups of Kvichak River sockeye salmon and three age groups of Egegik River sockeye salmon collected from the late 1950s through 2000 (Fig. 1). Adult salmon scales were obtained from the Alaska Department of Fish and Game (ADFG) archive in Anchorage, Alaska, and from the School of Aquatic and Fishery Sciences, University of Washington. Scales have been collected annually for measuring and quantifying age composition for manage- ment of the fisheries in Alaska. We selected scales from salmon sampled in the Kvichak and Egegik rivers rather than in the ocean fisheries to reduce the possibility of mixed stocks in the scale collection. Scale collections from the Kvichak River began in 1955, whereas collec- tions from Egegik River began in 1960. Major freshwater and ocean age groups from Kvichak (ages 1.2, 1.3, 2.2, 2.3) and Egegik (ages 1.3, 2.2, 2.3) sockeye salmon were measured. Age was designated by European notation, i.e. the number of winters spent in freshwater before going to sea (1 winter=age-l. or two winters = age-2.) followed by the number of winters spent at sea (two winters = age-.2 or 3 winters = age-.3.l. Nearly all Bristol Bay sockeye salmon mature after spending two or three winters at sea. Scales were selected for measurement in this study only when 1) we agreed with the age determination previously made by ADFG, 2) the scale shape indi- cated that the scale was removed from the "preferred area" (Koo. 1962), and 3) circuli and annuli were clearly defined and not affected by scale regeneration or sig- nificant resorption along the measurement axis. We measured up to 50 scales per year, representing equal numbers of male and female salmon from each age group within each stock. Scale measurements followed procedures described by Davis et al. (1990) and Hagen et al.5 After select- ing a scale for measurement, the scale was scanned from a microfiche reader and its image was stored as a high resolution digital file. High resolution (3352x4425 pixels) allowed the entire scale to be viewed and pro- vided enough pixels to be seen between narrow circuli 5 Hagen, P. T., D. S. Oxman, and B. A. Agler. 2001. Devel- oping and deploying a high resolution imaging approach for scale analysis. Doc. 567, 11 p. North Pacific Anadromous Fish Commision, 889 Pender Street, Vancouver, Canada. Ruggerone et al.: Seasonal growth of Oncorhynchus nerka in relation to competition with O. gorbuscha 357 Figure 1 Map of Bristol Bay, Alaska, and the location of the Kvichak and Egegik river systems. to ensure accurate measurements of circuli spacing. The digital image was loaded in Optimas 6.5 (Media Cybernetics, Inc., Silver Spring, MD) image processing software to collect measurement data with a customized program. The scale image was displayed on a digital LCD flat panel tablet. The scale measurement axis was determined by a perpendicular line drawn from a line intersecting each end of the first saltwater annulus. Distance (mm) between circuli was measured within each growth zone (i.e., from the scale focus to the outer edge of the first freshwater (FW1) annulus, between the first and second freshwater (FW2) annuli, within the spring plus (FWPL) growth zone, within each annual saltwater (SW1. SW2, SW3) growth zone, and from the last ocean annulus to the edge of the scale (i.e., the saltwater plus [SWPL] growth zone). Data analysis Mean scale circuli increments (distance between adjacent circuli pairs) of each age group and stock were calculated for each year when 10 or more scales were available. Typically, 40 to 50 scales of each age group and stock were measured in a given year. To facilitate evaluation of trends between odd- and even-numbered years at sea, scale circuli measurements were described in terms of the odd- or even-numbered year when the salmon entered the ocean. Thus, a salmon smolt entering the Bering Sea during an even-numbered year interacted with abundant odd-year Asian pink salmon during its second growing season (SW2) and less abundant even-year pink salmon during its third year, if it remained at sea. The number of circuli pairs considered in our analysis differed by growth zone, ranging from 22 circuli (SW1) to 20 cir- culi (SW2) to 15 circuli (SW3) in order to represent the majority of salmon. Analyses of seasonal scale growth trends were based on the mean of annual mean scale circuli increments, percentage change in scale circuli increments during odd- versus even-numbered years, and percentage change in odd- and even-year growth during periods before and after the 1977 ocean regime shift. A two-sample t-test was used to test for differences between odd- and even-numbered year scale growth at each cir- culi pair. Correlation was used to determine whether an individual's growth during the second growing season was related to previous growth at sea. Results First year (SW1) growth of ocean age-3 sockeye salmon Kvichak and Egegik river sockeye salmon scale growth (distance between adjacent circuli) increased rapidly 358 Fishery Bulletin 103(2) Odd year smolts Even year smolts Circuli pair Figure 2 Average seasonal scale growth for Kvichak and Egegik ocean age-3 sockeye salmon (Oncorhynchus nerka) that entered the ocean as smolts during odd ( ) and even ( ) numbered years, 1952-2000. Growth of salmon spending one (age 1.3) and two years (age 2.3) in freshwater are shown separately. Circuli pair ordering restarts at the beginning of each new growing season (SW1, SW2, SW3, SWPL). 95f7r confidence intervals (CIs) are shown at each measurement. after the fish entered Bristol Bay during May and early June, reaching peak growth near the fifth circuli (Fig. 2). Thereafter, growth declined steadily to a minimum at the first ocean annulus (circuli 18-22). Peak scale growth of age-2. smolts was significantly greater compared with that of age-1. smolts for both Kvichak (df=79, i=5.757, P<0.001) and Egegik salm- on (df=73, £=4.667, P<0.001). During the first eight circuli, age-2. smolts averaged 6.5% greater growth than age-1. smolts. Thereafter (circuli 11-20), growth of age-2. smolts declined more rapidly and averaged 2.3% (Kvichak) to 6.1% (Egegik) less than growth of age-1. smolts. Within the SW1 growth period, no statistically sig- nificant difference in circuli growth was detected be- tween smolts entering the ocean during odd- and even- numbered years (P>0.05). However, there was a trend for greater growth among even-year smolts in some Ruggerone et al.: Seasonal growth of Oncorhynchus nerka in relation to competition with O. gorbuscha 359 Pre-1977 Post 1976 Circuli pair Figure 3 Percent change in scale growth of ocean age-3 sockeye salmon (O. nerka) entering the ocean during even-numbered years compared to odd-numbered years. Growth patterns represent ocean developmental periods prior to 1977 ( ) and after 1976 t ). Even-year smolts encountered odd-year pink salmon (O. gorbuscha) during their second year at sea (SW2), but they encountered even-year pink salmon during their third year at sea (SW3). Age 1.3 = 1 year in freshwater and 3 years in saltwater; age 2.3= 2 years in freshwater and 3 years in saltwater. portions of SW1. including the annulus (circuli 18-22) and immediately after peak growth (circuli 7 to 13) (Figs. 2 and 3). SW1 growth of both even- and odd-year smolts tended to be greater after the 1977 climate shift than prior to this period, except for the last few circuli (Fig. 4). The greatest difference in growth between these two periods occurred immediately after entry into Bristol Bay (circuli 1-3) and during the last part of the SW1 growth period (circuli 13-19). This bimodal pattern of growth between the two periods was somewhat con- sistent among both stocks and freshwater age groups. However, Kvichak age 2.3 salmon experienced especially high early marine growth that was 17% greater, on average, after 1976. Following peak scale growth in spring, Egegik age 1.3 sockeye salmon experienced a 360 Fishery Bulletin 103(2) Odd year smolts Even year smolts Circuli pair Figure 4 Percent change in scale growth of ocean age-3 sockeye salmon (O. nerka) entering the ocean during 1977-97 from those entering the ocean during 1952-76. Growth patterns represent smolts entering the ocean during odd- ( ) and even-numbered years ( ). Even-year smolts encoun- ter odd-year pink salmon (O. gorbuseha) during their second year at sea (SW2), but they encountered even-year pink salmon during the their third year at sea (SW3). 15% increase in growth after 1976. In contrast, growth near the winter annulus (circuli 20-22) was up to 5% lower after the 1977 climate shift. Second year (SW2) growth of ocean age-3 sockeye salmon At the beginning of the second growing season (SW2), when Bristol Bay sockeye salmon are farthest south in the North Pacific Ocean (French et al., 1976), scale growth of both stocks and age groups increased rapidly, but the rate of increase was 59% less than that of SW1 and 377, less than SW3 growth (Fig. 2). Peak growth occurred near circuli 5 or 6 and it averaged 15% lower than that of SW1 growth. During their second year at sea, even-year sockeye smolts inhabited the North Pacific and Bering Sea when Asian pink salmon were abundant in offshore waters Ruggerone et al.: Seasonal growth of Oncorhynchus nerka in relation to competition with O. gorbuscha 361 Table 1 Summary of two sample /-tests for evaluating the circu i number at which sockeye scale growth began to differ between odd- versus even-numbered years of the second and third seasons at sea. Between-year differences in circuli growth were greater after the circuit num ser shown in this table. No consistent patt ern of difference;; between odd- and even-numbe 'ed years was observed during the first season at sea. Age ' 1.2 ' is a fish that has spent one year in fresh water and two years in ss It water. SW2=2 years in saltwater. Age Ocean period Stock Circuli no. df /-value P (two tailed I 1.2 SW2 Kvichak Cll 43 2.412 0.020 2.2 SW2 Kvichak Cll 44 3.283 0.002 2.2 SW2 Egegik Cll 39 3.434 0.001 1.3 SW2 Kvichak C12 42 3.068 0.004 SW3 Kvichak C8 42 3.126 0.003 1.3 SW2 Egegik Cll 38 2.140 0.038 SW3 Egegik C7 38 2.527 0.016 2.3 SW2 Kvichak Cll 43 2.711 0.010 SW3 Kvichak C8 43 2.384 0.022 2.3 SW2 Egegik Cll 39 3.061 0.004 SW3 Egegik C7 39 2.728 0.010 (i.e., during odd-numbered years). Initial scale growth prior to the SW2 peak in spring was the same between odd- and even-numbered years, although there was a ten- dency for greater growth following the SW1 annulus of even-year smolts (Fig. 3). Immediately after peak growth near circuli 11, scale growth of even-year smolts became significantly less than that of odd-year smolts (Table 1). The growth differential continued through the end of the SW2 growing season and it reached a maximum reduc- tion of -10% to -18% near circuli 14 to 18 (Fig. 3). This pattern was consistent before and after the 1977 climate shift and among each stock and age group. The reduced growth of even-year smolts during SW2 corresponded with high abundance of pink salmon in the central North Pacific Ocean during odd-numbered years. Scale growth during SW2 of both odd- and even-year smolts tended to be greater after the 1977 climate shift (Fig. 4), a period when abundance of Bristol Bay sock- eye salmon and Asian pink salmon was great. This pat- tern was consistent among both age groups of Kvichak and Egegik River sockeye salmon. Greatest growth dif- ferential between the two periods (up to 10%) occurred just after peak growth (circuli 5 to 15), a pattern that differed markedly from both SW1 and SW3. In contrast to the relatively large increase in growth shown in the central portion of SW2 after 1977, growth at the beginning of SW2 was similar during both periods and growth at the end of SW2 was relatively low after the climate shift. Third year (SW3) growth of ocean age-3 sockeye salmon Scale growth at the beginning of the third year at sea increased rapidly, peaked near circuli 5-6, then declined steadily through the year (Fig. 2). Peak growth during SW3 was intermediate to the relatively high peak growth during SW1 and relatively low peak growth during SW2. During their third year at sea, even-year sockeye smolts inhabited the North Pacific and Bering Sea when relatively few Asian pink salmon were in offshore wa- ters (i.e., even-numbered years). Prior to peak growth, SW3 growth of even-year smolts was similar or below that of odd-year smolts (Fig. 3), a pattern that contin- ued from the previous season. Immediately following the peak, growth of even-year smolts significantly increased in relation to odd-year smolts (Table 1), and growth re- mained relatively high throughout the remaining season (Fig. 2). Growth of even-year smolts was approximately 5% to 15% greater than that of odd-year smolts from circuli 8 to the annulus (Fig. 3). Differences in growth during even- versus odd-numbered years tended to be greater after 1976 when both pink and sockeye salmon were relatively abundant. Peak SW3 scale growth was up to 10% greater after the mid-1970 regime shift during both odd- and even- numbered years (Fig. 4). However, after the peak grow- ing season, scale growth was typically lower after 1976. The relatively low growth after 1976 was especially pronounced among odd-year smolts that inhabited the ocean during odd-numbered years when Asian pink salmon were abundant in offshore waters. Scale growth of odd-year smolts during SW3 was as much as 10% lower than that prior to 1977. Scale growth during both SW3 and SW2 were signifi- cantly reduced during odd-numbered years at sea (Table 1). However, SW3 scale growth during odd- versus even- years diverged immediately after the peak, whereas 362 Fishery Bulletin 103(2) Odd year smolts Even year smolts ~i i i i i i i i i i r 7 10131619 22 3 6 9 121518 1 4 Circuli pair Figure 5 Seasonal scale growth of Kvichak and Egegik ocean age- 2 sockeye salmon (O. nerka) that entered the ocean as smolts during odd- ( ) and even- ( ) numbered years, 1952-2000. Growth of salmon spending one (age 1.2) and two years (age 2.2) in freshwater are shown separately. Circuli pair ordering restarts at the begin- ning of each new growing season (SW1, SW2, SWPLl. 95% CIs are shown at each measurement. Age 1.2 = 1 year in freshwater and 2 years in saltwater. Pre-1977 Post 1976 o- SW1 (even yt) SW2 (odd yr) SWPL 5 " 0 - -< ,-,/^V\2C V SJ»? .c 4 w- \y Y. ■ * * ■1 June \ 20-23 < 1995 IZ 0.5 1 1.5 llm2 I ■ MM S ^ • \y 9 • m> June \ 18-21 < 1996 • 0.5 1 1.5 /m2 ! I ■ Figure 5 Horizontal distribution of S. niphonius eggs in the Sea of Hiuchi in 1995 and 1996. 376 Fishery Bulletin 103(2) $$&■:■■:■:■■ w _.::: May \ • 24-28 \ ^y^ 1 2 3 4/m2 1995 ^yv HUM Figure 6 Horizontal distribution of S. niphonius larvae in the Sea of Hiuchi in 1995 and 1996. that of clupeid fishes in the central Seto Inland Sea. Piscivorous fishes tend to spawn earlier than other fishes in freshwater ecosystems so that they attain sufficient size to enable consumption of other young fishes by the onset of piscivory (Keast, 1985). Because S. nipho- nius larvae are piscivorous from the first feeding stage, spawning that is synchronized with the seasonal peak in abundance of clupeid larvae would be advantageous for survival of S. niphonius larvae. Larvae of S. niphonius were abundant in the southern part of the Sea of Hiuchi in late May 1995 and 1996 while eggs were abundant in the northwestern waters during the same season. This difference in horizontal distribution patterns of eggs and larvae seems to be as- sociated with the drift by a residual flow (current) from northern to southern waters. In the central part of the Sea of Hiuchi, a residual flow in the middle (5-15 m) layers proceeds southward at a speed of about 5 cm/s (=4.32 km/d; Yanagi et al., 1995). Yolksac larvae of S. niphonius are abundant in the 5- to 10-m layers in the Sea of Hiuchi (Kishida, 1988) and do not exhibit diel vertical migration (Shoji et al., 1999). The horizontal distance between the stations with the highest egg and larval abundance in late May was approximately 15 km in 1995 and 20 km in 1996. Given that the yolksac stage is five days for mackerel larvae under 19°C (Shoji et al., 2001), drift distance while larvae are entrained in the southward residual flow during the yolksac stage would be estimated to be approximately 20 km. The estimate for the drift distance during the yolksac stage Table 1 Feeding incidence (percentage of stomachs with prey) and stomach contents of S. niphonius larvae collected in late May of 1995 and 1996 in the Sea of Hiuchi. No. of larvae examined No. of larvae feeding Feeding incidence {.% ) Size range (SL, mm) Stomach contents Sardinops melanostictus Konosirus punctatus Unidentified clupeids Engra ulis japon icus Unidentified Clupeiformes Mugiliidae Gobiidae Total 1995 1996 102 93 87 63 85.3 67.7 4.2-13.8 4.5-14.2 4 2 21 14 19 11 2 4 34 22 3 2 13 9 96 64 approximates the horizontal distance between the sta- tions of egg and larval highest abundance. It is there- fore plausible that the larvae were transported by the southward residual flow to the southern part of the Sea Sho|i and Tanaka: Feeding and growth of Scomberomorus niphonius 377 ^p&'^' C £7 . . . . •^S AJrii^. • • • ^rAf* .^ . Mav "BBB „ <& . .J 24-28 I V^a^1^~~^- 10Q. V .^ 100 200 3C )0 400 /m2 ^ ^ I I ■ June 20-23 1995 12 /m2 Figure 7 Horizontal distribution of clupeid larvae in the Sea of Hiuehi during the three cruises in 1995 and in 1996. of Hiuehi where clupeid larva concentration was high in late May. We suggest that spawning of S. niphonius in the northern part of the Sea of Hiuehi would enable their first-feeding larvae to meet high prey abundance in the southern part. Significance of high ichthyoplankton prey Water temperature and prey concentration would be possible factors that can influence growth rates of S. niphonius larvae. In aquaria, the mean absolute growth rate of S. niphonius larvae fluctuated between 0.87 and 1.28 mm/d depending on temperature between 18.2° and 22.6°C (Fukunaga et al., 1982; Shoji et al., 2001). In the present study, the mean surface temperature of the Sea of Hiuehi in late May was slightly higher in 1996, although the difference was not significant. The higher abundance of clupeid larvae in 1995 would better explain the higher larval growth rate in 1995. The mean larval growth rate in late May 1995, 1.05 mm/d, approximates those reported in aquaria at the same temperature (1.03 mm/d at 20.8°C; Fukunaga et al., 1982) where S. niphonius larvae were provided with an excess of prey, indicating that the prey concentration in late May 1995 met larval requirements. It is likely that the lower growth rate in late May 1996 resulted from lower prey concentration. This conclusion is sup- ported by results of the stomach content analysis: the larval feeding incidence was significantly lower in May 1996. We conclude that clupeid larvae concentration had a significant effect on growth of the S. niphonius larvae. In the Sea of Hiuehi, clupeid larvae abundance greatly increased from April to May. We suggest that the prey 378 Fishery Bulletin 103(2) 20 15 1995 L=1 .05-4-1.39 n=102 r2=0.87 1996 L=0.65A-0 .15 n=93 r2=0.80 i i 12 16 A (d) Figure 8 Relationships between standard length (L, mm) and otolith-estimated age (A, d) of S. niphonius larvae collected during the cruises in late May of 1995 and 1996 in the Sea of Hiuchi. 10 2 8 o S e to 1 4 i 2 0 1995 1996 Year Figure 9 Mean (SE) young-of-the-year S. niphonius abun- dance collected by the seine fishery in 1995 and 1996 in the Sea of Hiuchi. Asterisk indicates a significant difference between the years (Mann- Whitney I/- test, P<0.01). availability for S. niphonius larvae fluctuated depending for the most part on seasonal change in abundance of gizzard shad larvae that were dominant in the Sea of Hiuchi. The difference in clupeid larval abundance in late May between 1995 and 1996 may be explained by between-year difference in gizzard shad spawning stock biomass. The total catch of gizzard shad in the south- ern Sea of Hiuchi (coastal waters of Ehime Prefecture) in 1995 (372 t) was higher than that in 1996 (217 t: Ehime Prefecture Agriculture, Forestry and Fisheries Statistics Association, 1998). Implications for recruitment Variability in larval growth rate can influence survival during the larval period by affecting the length of the early life stages because total mortality is positively cor- related with the length of these early life stages (Houde, 1987). Campana (1996) demonstrated a significant corre- lation between growth to the end of the pelagic juvenile stage (90 d) and the year-class strength of Atlantic cod on the Georges Bank and suggested that the adult cohort strength could be predicted from growth during early life stages. In the present study, egg and larval S. niphopius abundance during their peak-occurrence period did not differ between 1995 and 1996, whereas YOY and 1-year- old S. niphonius were more abundant in 1995. These results indicate more successful recruitment and higher larval growth rate in 1995 although there are no data available for years other than 1995 and 1996. Larvae of S. niphonius initiate feeding at 5.59 mm SL at 18.5°C (Shoji et al., 2002). Given the mean larval growth rate in 1995 (1.05 mm/d) and 1996 (0.88 mm/d), the critical period (from first feeding to the onset of schooling at the early juvenile stage, 19.6 mm SL; Masuda et al., 2003) is estimated as 13.3 days in 1995 and 16.5 days in 1996. For S. niphonius, even a slight increase in larval stage duration due to retarded growth can greatly reduce larval survival because the larval daily mortality coeffi- cient is expected to be extremely high (>0.6: Grimes and Kingsford, 1996). The lower recruitment of S. niphonius in 1996 may be partly explained by the prolonged larval period (3.2 d) which could have led to lower survival (1/6.82, assuming the daily mortality coefficient is 0.6) during the larval period of that year. Acknowledgments We thank M. Fukuda, N. Suzuki, N. Kohno, and the crew of RV Shirafuji of NRIFEIS and staff of Asagi-Suisan Co. Ltd. for their assistance with field sampling. We also thank T. Maehara and N. Murata, Ehime Prefecture Chuyo Fisheries Experimental Station Toyo Branch, and Y. Maki, Kawarazu Fisherman's Association, for their help in collecting young-of-the-year Japanese Span- ish mackerel and data on the catch of 1-year-old fish. Two anonymous reviewers and M. Takahashi, National Research Institute of Fisheries Science, provided valu- able comments on the manuscript. Literature cited Campana, S. E. 1996. Year-class strength and growth rate in young Atlantic cod Gadus morhua. Mar. Ecol. Prog. Ser. 135: 21-26. DeVries, D., C. Grimes, K. Lang, and B. White. 1990. Age and growth of king and Spanish mackerel larvae and juveniles from the Gulf of Mexico and U.S. South Atlantic Bight. Environ. Biol. Fish. 29:135-143. Ehime Prefecture Agriculture and Forestry Statistics Association. 1998. Agriculture, forestry and fisheries statistics of Ehime Prefecture, fiscal 1996, 278 p. Ehime Prefec- Shoji and Tanaka: Feeding and growth of Scomberomorus niphonius 379 ture Agriculture and Forestry Statistics Association, Matsuyama, Japan. Finucane, J. H., C. B. Grimes,, S. P. Naughton. 1990. Diets of young king and Spanish mackerel off the southeast United States. Northeast Gulf Sci. 11: 145-153. Fukunaga, T., N. Ishibashi, and N. Mitsuhashi. 1982. Artificial fertilization and seedling propagation of Spanish mackerel. Saibai-giken 11:29-48. Grimes, C. B., and M. J. Kingsford. 1996. How do riverine plumes of different sizes influ- ence fish larvae: do they enhance recruitment? Mar. Freshw. Res. 47:191-208. Houde, E. D. 1987. Fish early life dynamics and recruitment vari- ability. Am. Fish. Soc. Symp. 2:17-29. Hunter, J. R. 1981. Feeding ecology and predation of marine fish larvae. In Marine fish larvae (R. Lasker, ed.l, p. 33-77. Univ. Washington Press, Seattle, WA. Jenkins, G. P., N. E. Milward. and R. F. Harwick. 1984. Food of larvae of Spanish mackerels, genus Scomb- eromorus (Telostei: Scombridae), in shelf waters of the Great Barrier Reef. Aust. J. Mar. Freshw. Res. 35, 477-482. Keast, A. 1985. The piscivore guild of fishes in small freshwater ecosystems. Environ. Biol. Fish. 12:119-129. Kishida, T. 1988. Vertical and horizontal distribution of eggs and larvae of Japanese Spanish mackerel in the central waters of the Seto Inland Sea. Bull. Jap. Soc. Sci. Fish. 54:1-8. 1991. Fluctuations in year-class strength of Japanese Spanish mackerel in the central Seto Inland Sea. Bull. Jap. Soc. Sci. Fish. 57:1103-1109. Kishida, T., and K. Aida. 1989. Maturation and spawning of Japanese Spanish mackerel in the central and western waters of the Seto Inland Sea. Bull. Jap. Soc. Sci. Fish. 55:2065-2074. Margulies, D. 1993. Assessment of the nutritional condition of larval and early juvenile tuna and Spanish mackerel (Pisces: Scomb- ridae) in the Panama Bight. Mar. Biol. 115:317-330. Masuda, R., J. Shoji, S. Nakayama, and M. Tanaka. 2003. Development of schooling behavior in Span- ish mackerel Scomberomorus niphonius during early ontogeny. Fisheries Sci. 69:772-776. Montani, S. 1996. Relationships between environment and fisheries in the Seto Inland Sea. In Resource and environment of the Seto Inland Sea (T. Okaichi, S. Komori, and H. Konishi, eds.), p. 1-37. Koseisha Koseikaku, Tokyo, Japan. Peters, J. S., and D. J. Schmidt. 1997. Daily age and growth of larval and early juvenile Spanish mackerel, Scomberomorus maculatus, from the South Atlantic Bight. Fish. Bull. 95:530-539. Sheldon, R. W„ A. Prakash, and W. H. Sutcliffe Jr. 1972. The size distribution of particles in the ocean. Lim- nol. Oceanogr. 17:327-340. Shoji, J., M. Aoyama, H. Fujimoto, A. Iwamoto, and M. Tanaka. 2002. Susceptibility to starvation by piscivorous Japanese Spanish mackerel Scomberomorus niphonius (Scombri- dae) larvae at first feeding. Fisheries Sci. 68:59-64. Shoji, J., T. Maehara, M. Aoyama, H. Fujimoto, A. Iwamoto, and M. Tanaka. 2001. Daily ration of Japanese Spanish mackerel Scomb- eromorus niphonius larvae. Fish. Sci. 67: 238-245. Shoji. J., T. Maehara, and M. Tanaka. 1999. Diel vertical movement and feeding rhythm of Japanese Spanish mackerel larvae in the central Seto Inland Sea. Fish. Sci. 65:726-730. Shoji, J., and M. Tanaka. 2001. Strong piscivory of Japanese Spanish mackerel larvae from their first feeding. J. Fish Biol. 59: 1682-1685. 2004. Effect of prey concentration on growth of piscivorous Japanese Spanish mackerel Scomberomorus niphonius larvae in the Seto Inland Sea, Japan. J. Appl. Ichthyol. 20:271-275. Tanaka, M., T. Kaji, Y. Nakamura, and Y. Takahashi. 1996. Developmental strategy of scombrid larvae: High growth potential related to food habits and precocious digestive system development. In Survival strategies in early life stages of marine resources ( Y. Watanabe, Y. Yamashita, and Y. Oozeki, eds.). p. 125-139. A. A. Balkema. Rotterdam. Watanabe, A. 1994. Fish caught by seine fisheries in the Sea of Hiuchi. Proceed, the Sea of Hiuchi Res. Meet. 2:4-9. Ehime Prefecture Chuyo Fisheries Experimental Station, Mat- suyama, Ehime, Japan. Yanagi, T, H. Tsukamoto, S. Igawa, and K. Shiota. 1995. Recruitment strategy of swimming crab, Portunus trituberculatus, in Hiuchi-nada, Japan. Fish. Ocean- ogr. 4:217-229. 380 Abstract— We consider estimation of mortality rates and growth param- eters from length-frequency data of a fish stock and derive the underlying length distribution of the population and the catch when there is individual variability in the von Bertalanffy growth parameter L„. The model is flexible enough to accommodate 1) any recruitment pattern as a function of both time and length, 2) length-spe- cific selectivity, and 3) varying fish- ing effort over time. The maximum likelihood method gives consistent estimates, provided the underlying distribution for individual variation in growth is correctly specified. Simula- tion results indicate that our method is reasonably robust to violations in the assumptions. The method is applied to tiger prawn data (Penaeus semisulcatus) to obtain estimates of natural and fishing mortality. Maximum likelihood estimation of mortality and growth with individual variability from multiple length-frequency data You-Gan Wang CSIRO Mathematical and Information Sciences 65 Brockway Road Floreat Park Western Australia 6014, Australia E-mail address. You-Gan Wangig'csiro.au Nick Ellis CSIRO Marine Research P.O.Box 120 Cleveland, Queensland 4163, Australia Manuscript submitted 5 March 2004 to the Scientific Editor's Office. Manuscript approved for publication 9 November 2004 by the Scientific Editor. Fish. Bull. 103:380-391 (20051. Estimation of growth and mortality is fundamental in fisheries because stock assessment and management rely on these population parameters. Length-frequency-based methods become important when aging tech- niques are either not possible or very expensive. Existing methods such as that of Beverton and Holt (1956) assume that recruitment is continu- ous and constant throughout the year, which leads to a population with an exponentially distributed age struc- ture. Existing modifications to Bever- ton and Holt's method comprise some simple recruitment patterns or distri- butions (Ssentongo and Larkin 1973; Ebert 1980; Hoenig 1987; Wetherall et al. 1987). As pointed out by Vetter (1988), the existing methods for esti- mating mortality in the literature have strong limitations and disadvan- tages. In particular, they require the following assumptions: 1) each individual follows the same von Bertalanffy growth curve; 2) the recruitment is either con- tinuous and constant through- out the year (as in Beverton and Holt [1956] and Wetherall et al. [1987]) or is a pulse function (as in Hoenig [1987]); 3) the total instantaneous mortality rate, z, is constant. As pointed out by Sainsbury (1980), it is more realistic to allow individual variability in growth. For example, using tag-recapture data, Wang et al. (1995) found substantial individual variability for the tiger prawn species P. semisulcatus. Estimation of mortality relies on the distribution of the lengths, which is determined by the age distribution, mortality rates, and the individual variability in growth rates. If individ- ual variability in growth is ignored, an inappropriate length distribution will be generated, leading to biases in parameter estimates. It is also biologically interesting to quantify the individual variability in growth, which has important implications in fisheries management. Although it is well understood that variability leads to increased uncertainty in estimates, it is less well recognized (among the fisheries community) that variability can also lead to bias. Wang and Ellis (1998) analyzed the effect of ignoring individual variability in a simplified context of constant recruitment and a single length-frequency record. They found that, in the presence of indi- vidual variability, existing methods gave positively biased parameter es- timates. More details about the back- ground can be found in Ebert (1973), Askland (1994), and Wang and Ellis Wang and Ellis: Maximum likelihood estimate of mortality and growth from multiple length-frequency data 381 (1998). See DeLong et al. (2001) for alternative ap- proaches to length-frequency data where individual variability is taken into account. In our study, we develop a new framework for analyz- ing length-frequency data. In particular, we incorporate 1) individual variability in growth parameters; and 2) an arbitrary recruitment function. The model is flexible enough to incorporate various sizes at recruitment and a fishing selectivity function. However, we did not use these aspects in the analysis of tiger prawn data. Some analytical expressions are derived for these generaliza- tions. A maximum likelihood approach is developed for estimation of mortality and growth parameters. Sepa- ration of fishing mortality from natural mortality is possible only when there is substantial contrast in the effort pattern. We also require a known recruitment pattern, and sampling times are spread out so that the length-frequency data will contain information on growth and mortality. Simulation studies are carried out to determine the performance of the method. The simulated data are generated from the recruitment pat- tern of the brown tiger prawn iPenaeus esculentus) in the northern prawn fishery of Australia. Finally we ap- ply the maximum likelihood method to length-frequency data from grooved tiger prawn data (P. semisulcatus) in the northern prawn fishery of Australia. where /',(/IL , =.v, L0=s) is the conditional probability density function of L at time t when Lx is known to be x and the size at recruitment is s. Note the lower limit of the inner integral is / because L t cannot be less than an individual's length. Let the age (again, relative to t0) at recruitment of an individual be A0. From Equation 1, we have age a at length / is a = -k~1\og(l-l/L_j ) and hence the conditional distribution, ft(l/Lx=x, Ln = s), which may be written as ft(l\x, s) for brevity, can be expressed by using the conditional distribution of age \\tia\Ly=x, A0=an) (see Wang et al.. 1995), as ft(l\x,s)- k( x-1) h, (-/?"' log(l-//.v)|.r,a0). (3) We now generalize assumptions 2 and 3 by introduc- ing the intensity function of recruitment, r(t ), and the total instantaneous mortality, z(t), which are arbitrary functions of time t. The total mortality would depend on time through the fishing mortality component F, where zit)=M+Fit) and M is the constant natural mortality. The age distribution satisfies h,(a ILM=.v,A0-a0)~exp|-J z(t - a + y)dy\r(t - a + a0). (4) Materials and methods The model We assume that the growth of individuals follow a von Bertalanffy curve so that the length at age a (relative to some origin t0) is given by L(a) = LJl -e-ha). (1) In this study, age is always defined to be relative to t0, i.e. t0 is absorbed into a for the purpose of identifiability. We will consider estimation of (k, lx) only because t0 is not estimable from length-frequency data with aging data. Note that this does not mean t0 is assumed to be 0. To provide a general treatment we relax each of the assumptions mentioned in the introduction. First we relax assumption 1 by letting the maximum length, L ,, vary within the population. We denote the density func- tion of L , as p(x), which has a mean of I r and a variance of a2. It is possible that recruits to the fishery have a range of sizes. To allow for this range we let the size at recruitment, L0, be a random variable with density function u(s). In practice, one may be able to use infor- mation from other studies (such as subadult abundance) to arrive at an approximate parametric form for u(s). If ft(l) is the probability density function of L at time t, then ft(.l) = J"j"p(x\L0=s)ft(l\L„=x,LQ = s)u(s)dxds, (2) This equation states that the density of individuals of age a is proportional to the intensity of recruitment at the time when these individuals were recruited, namely t-a+a0, multiplied by a reduction factor due to mortality over the intervening period. We therefore have ht(a\x,s) = ht(a\L„=x,Ao=-k~1log(.l-s/x)) =exp(-| (5) -k "Mogll-s/.rl zit-a + y )dy \r(t- a- k 1 log(l-s/x) and Equation 3 becomes (after substituting for a and shifting the dummy variable y) x-l exp -j: fta\x,s) -M^fy)dy t-k'Hog x-s\) (6l Let us consider the case of fixed recruitment length, i.e., L0=l0, and define a parameter vector, p, consisting of th, /x, s), and other parameters quantifying mortality and catchability. Equation 2 then reduces to a single integral over x, fl(l\/5)ocj°°p(x)exp -LMS)^4("r'log(^))£' (7) 382 Fishery Bulletin 103(2) A more convenient form for computation arises after changing the integration variable from the asymptotic length x to time since recruitment, t-a+a0, '-*Sf3 (8) The expression (Eq. 7) then becomes ftil\p)^j°°p(x(T))exp(-f' ^z(y)dy)r(t-T) ^- (9) In the special case of constant recruitment, i.e., r(t)=l, and constant mortality, z(.t)=z, f,U\p) becomes indepen- dent of time as first obtained by Powell (1979). Maximum likelihood estimation Let p,,(/3) be the expected proportion of individuals in the ith length class (/,_j, /) on the j,h occasion, where j=l, 2, • •• , N; and let n:] be the corresponding observed numbers. The value of Pj.ifi) can be obtained from the density function ft(l;P) given by Equation 2. Thus P,/Py- (10) in which fAl;P) is the (unnormalized) density function on the yth occasion. Under a multinomial model, estimation of the parameter vector fi relies on the procedure maximize £ nlt log Py(.fi) with respect to ft. (11) The sum is the log-likelihood function up to a constant independent of the parameters. The probability, p , can be approximated as fj(.li+i/2^ifj^i+V2^> wnich is the nor- malized value of the density function for thejth occasion at the midpoint of the ;'th length class. If sampling effort is known and expected catch is as- sumed to be a known function of effort and population abundance, the log-likelihood function in Equation 11 can be easily modified to incorporate effort informa- tion. For example, if the total number of individuals on each occasion, ni=I.i=nll, is assumed to follow a Poisson model with overdispersion parameter v, the log-likeli- hood function becomes sampling effort,

- Here v is introduced to allow for overdispersion in the Poisson model. It plays a weighting role for the two terms in Equation 12, and the second summation can be regarded as auxiliary information. If?;, is assumed to follow a Poisson distribution exactly, we have v=l. In our simulation and tiger prawn studies we specify a case of fixed, known recruitment length, /0, and fM;P) is obtained from Equation 7 or 9. For definiteness we set the constant of proportionality implicit in these equations to one. The integrals in Equations 7 and 9 present some subtleties for their evaluation, so that some details of the numerical implementation might be of inter- est. For the simulation study we used Equation 7. The integral was performed on an /-dependent grid of 41 and 81 quantiles of the Lx distribution p(x) and then improved upon by using the Richardson extrap- olation. Note that there is an apparent singularity at x = l. However, by decomposing the mortality into a mean and deviation term, z(y)=z +z(y)— z , we find that the factor involving mortality is proportional to (x-lYlk. Hence the integrand is proportional to (x-lYlk, and, because zlk— 1>— 1, the singularity is integrable (i.e., the integral is finite). We used a quadrature scheme designed for integrands of the form (x— lrf(x)£>— 1, to perform the integral in the neighborhood of x = l. For the tiger-prawn study we used Equation 9. The integral was performed on uniform grids of 41 and 81 points over the interval tE(0,1.5) years and, as before, was improved by using the Richardson extrapolation. We used our knowledge that tiger prawns live for about 18 months to determine the upper limit of integration. Note that despite appearances, this integral contains no singularity because 3c(t)-><* as t-»0), and therefore the factor p(x(T))/(l-e_,'T)^0. The effort integral within the integrand was computed by linear interpolation between cumulative totals of the weekly effort. The prototype implementation of our maximum like- lihood method was written in S-plus software (Lucent Technologies) by using the optimizer "nlminb." However, to improve the performance for a large number of simu- lations, the program was recoded in C by using Powell's optimization routine with numerical derivatives (Press et al., 1992). The C code and some relevant reports are available on request. 5X-]ogpj,(j8) + vX{n,logA,(0)-A//»}, (12) where A(/3) is the expected total number in the sample on the j-th occasion and depends on effort. One way to model this dependence is A/(/3) = 0p/(/3)e/, where ef is the Results Simulation studies We simulated length-frequency data based on the recruitment pattern of tiger prawns P. esculentus in Wang and Ellis: Maximum likelihood estimate of mortality and growth from multiple length-frequency data 383 1.0 1.0 /T^^v ! ! ! „ 0.8 1 \ 1 Eflort • 0.8 (0 E I 0.6 / i \ ill. Norma CO d Q> / I \ ! ! N ~ ~z - ^ -o o 3 u c iz — CD •— a t- 3 QJ _ H) m a CO 3 — — -t-J CO C a — £ ■* -= r~ CU > — ■* co o c CO CO cfl t*-i — J= 0) — cj ed cj i o , . «T •~-~. 3 0- CC -a 2 +J C Q. Ui +3 ^ u- co o> is CO o bo 01 CN ■5 e s ~ j2 co C O n cj CO o o o X — o ■^ „ Tf CD ■4-) > CO ^ ■a x « ^ (lulu) ssep L|i6uai u CD O T3 CJ 6 o >i CD CO -*-j C CD — CD CJ be c CD CD o Zj > CD ■^ P ~ CJ '" s CD 'a CO CD 3 — CD CO 390 Fishery Bulletin 103(2) are in the data. We fitted a variety of different models. The objective function -21og( likelihood) values in Table 2 should be used only as guidelines and should not drive the analysis or be used for model selection. Tiger prawns are subject to very high total mortality and hence are short-lived species. Our method is also ap- plicable to longer-lived species. However, for application to other fisheries, some modification of the model may be necessary to incorporate relevant information in the model. Simulation studies may have to be carried out to see how reliable the modified version is for param- eter estimation because many factors, such as growth rate and commercial effort patterns, will determine if parameter estimates can be found or how reliable they are if they can be found. We aim to obtain growth and mortality parameter estimates simultaneously. However, this may be too ambitious, especially for short-lived species unless other information can be incorporated to assist esti- mation. For instance, Ebert (1973) found estimation of even two parameters (natural and fishing mortal- ity) unreliable and had to assume one of them. This is perhaps why natural mortality is assumed to be known in traditional cohort analysis. Also Askland's method (1994), one of the most recent cohort-analysis methods, requires a known M. Nevertheless, in prac- tice, ik, l.,) may be estimated from different types of data. The results based on model 2 (assuming {k, l^) are known) indicate that both M and F can then be estimated more reliably when there is substantial contrast in the effort pattern. Another assumption is that catchability does not change over time. This may not be necessarily true when new technology is introduced into the fishery (Bishop et al., 2000). The assumption that growth parameters are known greatly reduces the complexity of estimating the remaining unknown parameters and improves the performance of the proposed methods. We have chosen to allow only lm to be random because, unlike tag-recapture data, the length-frequency data do not have multiple measures from each individual. Each individual is measured only once. Therefore, it might be problematic to allow random K and correlation between K and L^. Such an attempt using length-frequency data may lead to misleading conclusions because the conclusion will be model-driven instead of data-driven. Parameter estimates obtained by fixing M as a constant are deemed more reliable. We provided a framework for length-frequency da- ta analysis that incorporates continuous recruitment, selectivity, and time-dependent fishing mortality. We have also provided guidelines for how to compute the likelihood function, which depends on rather delicate integrals. Such a model would be very useful for many fisheries because such unified models are not available in the literature. Our work provides a sensible case study. Application of our method may require incorpora- tion of specific information in a fishery. We believe our model, which generalizes the traditional model and is somewhat complicated, has provided us with some use- ful results for future stock assessment and evaluation of management strategies. Acknowledgments This research project was partly supported by the Fisher- ies Research and Development Corporation of Australia. We gratefully acknowledge the helpful suggestions and comments of David Die, Andre Punt, Neil Loneragan, and two anonymous referees. Literature cited Askland, M. 1994. A general cohort analysis method. Biometrics 50:917-932. Beverton, R. J. H., and S. J. Holt. 1956. A review of methods for estimating mortality rates offish in exploited fish populations, with special refer- ence to sources of bias in catch sampling. Rapp. P.-V. Reun., Cons. Int. Explor. Mer 140: 67-83. Bishop, J., D. Die, and Y.-G. Wang. 2000. A generalized estimating equations approach for analysis of fishing power in the Australian Northern Prawn Fishery. Aust. N. Z. J. Stat. 42:159-177. DeLong, A. K., J. S. Collie, C. J. Meise, and J. C. Powell. 2001. Estimating growth and mortality of juvenile winter flounder, Pseudopleuronectes americanus, with a length-based model. Can. J. Fish. Aquat. Sci. 58:2233-2246. Deriso, R. B., and A. M. Parma. 1988. Dynamics of age and size for a stochastic population model. Can. J. Fish. Aquat. Sci. 45:1054-1068. Ebert, T. A. 1973. Estimating growth and mortality rates from size data. Oecologia 11:281-298. 1980. Estimating mortality from growth parameters and a size distribution when recruitment is periodic. Limnol. Oceanogr. 26:764-769. Hoenig, J. M. 1987. Estimation of growth and mortality parameters for use in length-structure stock production models. In Length-based methods in fisheries research (D. Pauly and G. R. Morgan, eds.), p. 121-128. ICLARM Conf. Proc. 13. McDonald, P. D„ and T. J. Pitcher. 1979. Age groups from size-frequency data: a versatile and efficient method of analyzing distribution mixtures. J. Fish. Res. Board Can. 36:987-1001. Pauly, D., J. Ingles, and T. Neal. 1981. Application to shrimp stocks of objective methods for the estimation of growth, mortality and recruit- ment-related parameters from length-frequency data (ELEFAN I and III. In Penaeid shrimps: their biology and management (J. A. Gulland and B. J. Rothschild, eds.), p. 220-234. Fishing News Books Ltd., Farnham, England. Powell, D. G. 1979. Estimation of mortality and growth parameters from the length frequency of a catch. Rapp. P.-V. Reun., Cons. Int. Explor. Mer 175:167-169. Wang and Ellis: Maximum likelihood estimate of mortality and growth from multiple length-frequency data 391 Press, W. H., S. A. Teukolsky, W. T. Vetterling, and B. P. Flannery. 1992. Numerical recipes in C; the art of scientific com- puting (2nd ed.). Cambridge Univ. Press, Cambridge, England. Sainsbury, K. J. 1980. Effect of individual variability on the von Ber- talanffy growth equation. Can. J. Fish. Aquat. Sci. 37:241-247. Somers I. F., and G. P. Kirkwood. 1991. Population ecology of the grooved tiger prawn, P. semisulcatus, in the North-western Gulf of Carpentaria, Australia: growth, movement, age structure and infesta- tion by the bobyrid parasite Epipenaeon ingens. Aust. J. Mar. Freshw. Res. 42:349-367. Somers, I. F, and Y.-G. Wang. 1996. A bioeconomic analysis of seasonal closures in Australia's multispecies Northern Prawn Fishery. N. Am. J. Fish. Manag. 17:114-130. Sparre, P. 1987. A method for estimating growth, mortality and gear selection/recruitment parameters from length-frequency samples weighted by catch per effort. In Length-based methods in fisheries research. ICLARM Conf. Proc. 13 (D. Pauly and G. R. Morgan, eds.), p. 75-102. Ssentongo, G. W., and P. A. Larkin. 1973. Some simple methods of estimating mortality rates of exploited fish populations. J. Fish. Res. Board Can. 30:695-698. Sullivan, P. J. 1992. A Kalman filter approach to catch-length analysis. Biometrics 48:237-257. Sullivan, P. J, H. L. Lai, and V. F Gallucci. 1990. A catch-at-length analysis that incorporates a stochastic model of growth. Can. J. Fish. Aquat. Sci. 47:184-198. Vetter, E. F. 1988. Estimation of natural mortality in fish stocks: a review. Fish. Bull. 86:25-43. Wang, Y.-G. 1998. An improved Fabens method for estimation of growth parameters in the von Bertalanffy model with individual asymptotes. Can. J. Fish. Aquat. Sci. 55:397-400. Wang, Y.-G., and D. Die. 1996. Stock-recruitment relationship of the tiger prawns (Penaeus esculentus and Penaeus semisulcatus) in the Australian Northern Prawn Fishery. Aust. J. Mar. Freshw. Res. 47:87-95. Wang, Y.-G., and N. Ellis. 1998. Effect of individual variability on estimation of pop- ulation parameters from length-frequency data. Can. J. Fish. Aquat. Sci. 55:2393-2401. Wang, Y.-G., and I. F. Somers. 1996. A simple method for estimating growth param- eters from multiple length-frequency data in presence of continuous recruitment. Fish. Res. 28:45-56. Wang, Y.-G., and M. R. Thomas. 1995. Accounting for the individual variability in the von Bertalanffy growth model. Can. J. Fish. Aquat. Sci. 52:1368-1375. Wang, Y.-G., M. R. Thomas, and I. F. Somers. 1995. A maximum likelihood approach for estimating growth parameters from tag-recapture data. Can. J. Fish. Aquat. Sci. 52:252-259. Wetherall, J. A., J. J. Polovina, and S. Ralston. 1987. Estimating growth and mortality in steady-state fish stocks from length-frequency data. In Length- based methods in fisheries research (D. Pauly, and G. R. Morgan, eds.), p. 53-75. ICLARM Conf. Proc. 13. 392 Abstract— Fisheries often target indi- viduals based on size. Size-selective fishing can create selection differen- tials on life-history traits and, when those traits have a genetic basis, may cause evolution. The evolution of life- history traits affects potential yield and sustainability of fishing, and it is therefore an issue for fishery manage- ment. Yet fishery managers usually disregard the possibility of evolution, because little guidance is available to predict evolutionary consequences of management strategies. We attempt, to provide some generic guidance. We develop an individual-based model of a population with overlapping genera- tions and continuous reproduction. We simulate model populations under size-selective fishing to generate and quantify selection differentials on growth. The analysis comprises a variety of common life-history and fishery characteristics: variability in growth, correlation between von Bertalanffy growth parameters (K andL,.), maturity rate, natural mor- tality rate (M), M/K ratio, duration of spawning season, fishing mortality rate (F), maximum size limit, slope of selectivity curve, age at 50% selectiv- ity, and duration of fishing season. We found that each characteristic affected the magnitude of selection differentials. The most vulnerable stocks were those with a short spawn- ing or fishing season. Under almost all life-history and fishery character- istics examined, selection differentials created by realistic fishing mortality rates are considerable. Effects of fishing on growth traits: a simulation analysis Erik H. Williams Kyle W. Shertzer Center for Coastal Fisheries and Habitat Research 101 Pivers Island Road Beaufort, North Carolina 28516 E-mail address EnkWilliams@noaa.gov Manscript submitted 16 April 2004 to the Scientific Editor's Office. Manuscript approved for publication 20 December 2004 by the Scientific Editor. Fish. Bull. 103:392-403 (20051. Fishing is typically size selective. It almost always targets the larger individuals of a population and can thus shift the spawning stock towards smaller, slower-growing individuals. If somatic growth has some genetic basis, size-selective fishing may cause evolu- tion toward a smaller size-at-age. Changes in somatic growth are well documented in field data, and several studies implicate fishing (Ricker, 1981; Harris and McGovern, 1997; Haugen and Vollestad, 2001; Sinclair et al., 2002). However, with typical field data, it is difficult to rule out other explanations. Changes in growth could result from fluctuations in population density or the environ- ment. Furthermore, they may not be evolutionary, but instead expressions of phenotypic variability. Because of such possibilities, the idea that fish- ing can cause evolution has often been accepted because of compelling theoretical arguments, rather than on empirical support. However, the laboratory experiments of Conover and Munch (2002) demonstrated that size selection can cause evolution of growth traits. More and more, fish- ing-induced evolution is considered not just possible, but prevalent (Law, 2000; Stockwell et al., 20031. The evolution of growth traits, de- spite wide acknowledgement of the potential for evolution of these traits, is usually a low priority in fishery management. However, it raises at least four management concerns. First, any reduction in growth rate or maximum size can decrease rec- reational and economic value (Miller and Kapuscinski, 1994). Second, size selection could reduce genetic vari- ability (Falconer and Mackay, 1996). unpredictably altering correlated traits and population fitness. Third, evolution may not easily be reversed, even with after-the-fact management. Fourth, the evolution of growth and other life-history traits can modify population dynamics (Bronikowski et al., 2002; Shertzer and Ellner, 2002) and therefore potential yield (Edley and Law, 1988; Heino 1998). Evolu- tion in fishes can be rapid (Reznick et al., 1997; Hendry et al., 2000; Quinn et al., 2001), so that evolutionary, population, and fishery dynamics oc- cur on similar time-scales (Sinervo et al., 2000; Shertzer et al., 2002; Yoshida et al., 2003). These dynam- ics imply that evolution matters for fishery management on the time-scale of years or decades. For fishing to cause evolution, two conditions must be met. There must be a selection differential on a pheno- typic trait and a genetic basis must exist for the trait's expression (i.e., the trait must be heritable). Selec- tion differential is defined as the dif- ference in the mean phenotypic trait value of parents before and after se- lection (e.g., size-selective fishing). Stokes and Law (2000) argued that, under exploitation levels in many of today's fisheries, "selection differ- entials on body size should be sub- stantial and measurable." Even so, attempts to estimate selection differ- entials of actual fish stocks have been rare (but see Law and Rowell, 1993; Miller and Kapuscinski, 1994). This lack of estimates is surprising, given that the data needed are often avail- able, as noted by Law (2001). The second necessary condition, her- itability, is defined as the proportion of phenotypic variability in offspring Williams and Shertzer: Effects of fishing on growth traits: a simulation analysis 393 that is due to the genotypes of the parents. It can range from zero to one, with a higher value potentially speeding the evolutionary response to selection. Field estimates of heritability in fish size are uncommon because in nature it is difficult (although not impossible; McAllister et al., 1992) to separate genetic and environ- mental effects on phenotypes. Almost all esti- mates come from laboratory experiments (e.g., Hadley et al., 1991; Conover and Munch, 2002; Vandeputte et al., 2002), mostly on populations from aquaculture breeding programs (e.g., Gje- drem, 1983; Jarayabhand and Thavornyutikarn, 1995; Henryon et al., 2002). One might expect laboratory experiments to over-estimate natural heritabilities, because experiments tend to re- duce environmental effects on total phenotypic variance, but estimates from the laboratory have been similar to those from the field (Wei- gensberg and Roff, 1996). The laboratory exper- iments indicate that heritabilities in fish growth traits may vary widely among populations but Repeat are high enough to allow rapid evolution, given over a large enough selection differential. time Models of evolutionary response to selec- steps (t) tive harvest have usually taken one of two for one approaches: quantitative genetics (e.g., Law, year 1991; Ratner and Lande, 2001) or life-history optimization (e.g., Blythe and Stokes, 1999). In the present study, we take a different approach. Rather than attempt to predict evolution ex- plicitly, we focus on selection differentials, a necessary (but not sufficient) condition for an evolutionary response. We use simulation analyses to compute selec- tion differentials caused by fishing. The simula- tion model is one common in fisheries. It con- sists of an age-structured population following von Bertalanffy growth, with fishing and repro- duction modeled as continuous processes. Our goal is to compare selection differentials across a variety of life-history and fishery char- acteristics. We quantify selection differentials on growth parameters and body size. If growth traits are heritable, those life-history and fish- ery characteristics with the largest selection differentials are most likely to generate an evo- lutionary response. Armed with such knowl- edge, fishery managers can weigh potential evolutionary effects when choosing a fishing strategy. Draw uniform random number to determine cohort of individual; probabilities based on stable age structure Draw bivanate normal random numbers to determine values of growth parameters L , K Draw uniform random number to determine spawning time step Unfished population Fished population Draw uniform random number to determine mortality Alive? Alive? t = spawning time step7 : = spawning time step? Draw uniform random number to determine if spawning occurred Draw uniform random number to determine if spawning occurred Store growth parameters Store growth parameters Figure 1 Flow diagram of the individual-based model. 250,000 individu- als were initialized and then duplicated; one copy entered an unfished population, the other entered a fished population. Both populations were simulated for a single year with monthly time steps. Selection differentials on the growth parameters were computed as the difference between mean trait values of the unfished and fished parents. Materials and methods To compute selection differentials caused by size-selective fishing we used an individual-based model (Fig. 1). To initialize the model, 250,000 individual phenotypes were generated. Each was assigned a set of life-history param- eters and then duplicated. One copy entered an unfished population that experienced only natural mortality; the other copy entered a fished population that experienced both natural and fishing mortality. Growth, survival, and reproductive success of individuals were simulated with monthly time steps for a single year. At the end of the simulation, selection differentials on growth parameters were computed as the percent change between the mean values of spawners in the two populations. Model structure The model comprised three basic life-history functions: growth, survival, and reproduction. For each individual. 394 Fishery Bulletin 103(2) size was assumed a function of age (a) and followed the von Bertalanffy model, Ha) LJl-e-^-V], (1) where /(a)=the length-at-age of an individual; Lr = the theoretical maximum length; K =the growth rate, and t0 =the theoretical age when size would have been zero. In our study, each individual's age and size were updated at each monthly time step. Survival was computed differently for the two popula- tions. In the unfished population, individuals survived with a probability depending only on the natural mor- tality rate (M/yr). In the fished population, individuals survived with a probability depending on both the natu- ral mortality rate and the size-specific fishing mortality rate. Size selectivity [s(/)] by the fishery increased with length according to the logistic equation s(l)- 1 + e ■Psu-l,) (2) where /3S = the slope of the selectivity curve; and Ls = the length at 50% selectivity. The function s(l) describes the proportion of the fully- selected fishing mortality rate IF) experienced by indi- viduals of length /. The size-specific fishing mortality rate, therefore, is s(l)F per year. Fishing was applied over a fishing season of duration DF. The probability of reproduction was assumed equal to the probability of maturity [m(a)\. In the model, matu- rity increases with age and is independent of length. Al- though maturity likely relates to length through bioen- ergetics, the relationship was not modeled here because it is, in general, poorly understood. Like selectivity (Eq. 2), m(a) was modeled by a logistic equation, but with a slope parameter, )3m, and age at 50% maturity, Am. In nature, values of life-history parameters K and Am are related to a stock's natural mortality rate. A higher natural mortality rate reduces the expected lifespan and consequently tends to be associated with a higher growth rate (K) and a younger age at maturity (Am). In the simulation, K and Am were related to natural mortality by life-history invariants (detailed later). Life-history invariants have a strong theoretical and empirical basis (Roff, 1984; Beverton, 1992; Charnov, 1993) and have been valuable in other fishery applica- tions (Mangel, 1996; Charnov and Skuladottir, 2000; Frisk et al., 2001; Williams and Shertzer, 2003). Simulation To initialize the simulation, individuals were assigned at random to a cohort. The number of cohorts was deter- mined as the age at which approximately 1% of the population would be expected to remain under natural mortality [-ln(0.01)/M, rounded to the nearest integer]. Probabilities of cohort membership decayed exponen- tially with age according to M; the probability of the oldest cohort was adjusted to include the remaining fraction offish (i.e., a plus group). The probabilities were scaled to sum to one, and a uniform random number was drawn to determine an individual's cohort. Next, individuals were assigned parameter values for von Bertalanffy growth. The value of t0 was fixed at 0.5 yr. Values of Lx and K were chosen uniquely for each in- dividual. Following Xiao (1994), Lx and if were assumed to follow a bivariate normal distribution with standard deviations aL and aA-, respectively, and correlation p. Finally, individuals were assigned a time step (month) within the year to attempt spawning. The time step was chosen from months distributed uniformly over a spawning season of duration, Z)s. Once assigned parameter values, each individual was duplicated. One copy entered the unfished population, the other the fished population. The populations were simulated in parallel over a single model year. The simulation iterated each individual through monthly time steps. At each step, the simulation com- puted growth and checked for survival and reproduc- tion. In the unfished population, the monthly probability of survival was exp(-M/12). In the fished population, the monthly probability of survival during the fishing season depended on natural mortality and on the size- specific fishing mortality. For simplicity, we assumed size within a month was fixed so that that the prob- ability of survival was exp[(-M/12-s(/0)F)/DF], where l0 was an individual's size at the beginning of the month. Outside the fishing season, only natural mortality ap- plied. To check for survival, a uniform random number was drawn and compared to the survival probability appropriate for the population. Each individual surviving to its assigned spawning time had the opportunity to reproduce. In that case, a uniform random number was drawn and compared to the probability of reproduction. If reproduction was suc- cessful, the individual's growth parameters went into a pool of parents used to compute selection differentials. Growth parameters Lx and K jointly determine size- at-age, and it is on these parameters that we describe selection differentials. At the end of the simulation year, we computed a selection differential on each growth parameter as the percent difference between mean trait values (Lr or K) of the unfished and fished parents. Based on the differences in Lx and K, we also computed upper and lower bounds of selection differentials on size-at-age. The bounds occur where age approaches t0 or oc. Because each population consisted of the same set of individuals at the beginning of the year, any differ- ence in growth traits between parents at the end of the year could be attributed solely to fishing. Base model and variations We began with a base model built on parameter values chosen or computed to represent common life-history and Williams and Shertzer: Effects of fishing on growth traits: a simulation analysis 395 Table 1 Parameter values used in the base model. Formulas for the growth rate (A'l and the age at 50% matu -ity (A„ ) are life-history invar lant relationships from Charnov ( 1993 ) and Beverton ( 1992 ) , respectively. The formula forLq is the length it age Am accord- ing to von Bertalanffy growth. A value of °° for slope parameters corresponds to a knife-edge curve. Parameter Description Formula Value M Natural mortality rate (per year) Fixed 0.2 F Fishing mortality rate (per year) Fixed Oto 10 £» Mean asymptotic size in growth function Fixed 1000 K Mean growth rate in growth function M/1.65 0.12 t0 Location parameter in growth function Fixed -0.5 cvL Coefficient of variation in hr Fixed 20% cvA. Coefficient of variation in A' Fixed 20% p Correlation between L, and K Fixed 0 ft Slope of the size selectivity curve Fixed 00 ft, Slope of the maturity curve Fixed ■x. K Age at 50% maturity log[(3 A + M)/M] IK 8.55 k Length at 50% selectivity L.[l-exp(-X[Am -to ')] 666 Ds Duration of spawning season (yr) Fixed 1 DF Duration of fishing season (yr) Fixed 1 fishery characteristics (Table 1). We then conducted a variety of sensitivity analyses. In the base model, the natural mortality rate (M) was set at 0.2/yr, a value common for many fish species. Sensitivity analyses used M = 0.1, 0.4, or 0.8. The value of M affects the values of A', Am, and Ls, according to the life-history invariant relationships (Table 1). The relationship between M and K is often referred to as the M/K ratio. Charnov (1993) suggested a central value for fishes of M/A"=1.65, which we used in the base model. Beverton (1992) examined the M/K ratio for fishes and found a range of 0.5 to 2.5. We used this range in our sensitivity analyses to examine the effect of the M/K ratio on selection differentials (Table 2). The base model treated Lx and K as independent variables (p=0. Table 1). Often these parameters are correlated. A meta-analysis by He and Stewart (2001) of 235 fish populations indicated a correlation value of -0.28. The negative correlation could be expected from a trade-off between growth rate (represented by A'l and maximum size (represented by Lx), as has been suggested in studies of bioenergetics (Stearns, 1992; Hutchings, 1993; Mangel, 1996). Our sensitivity analy- ses considered negative values of correlation that range from -0.25 to -1. With the base model, selectivity and maturity were assumed to be "knife-edge," a functional form often used in fisheries for convenience. Also, in the base model the size at 50% selectivity (Ls) was assumed to occur at an age equal to the age at 50% maturity (Am). Although these fishery characteristics are common, selectivity and maturity may not be knife-edge or coincide. In sensitivity analyses, we examined different shapes of selectivity and maturity curves (Fig. 2). We also ex- amined the affect of shifting the age at 50% selectivity from -2 to 2, in relation to the base case. This shift corresponds to a range in Ls values from 574 to 738. For simplicity, we held F constant for these sensitivity analyses, implying constant effort but resulting in dif- ferent amounts of removals. Under logistic selectivity, the oldest, largest fish receive the highest rate of exploitation. Yet often the largest fish are unavailable to a fishery because of mi- gration patterns or regulations (e.g., a maximum size limit). Thus our sensitivity analyses included a cap on susceptible sizes. The cap was set at 70, 80, or 907c of Using the base model, we examined the effects of an- nual fishing mortality rate over values that range from F=0 to F=10/yr, which is 0 to 50 times the natural mor- tality rate. Fishing mortality was applied continuously throughout the year (i.e., DF=1). In sensitivity analyses, we examined shorter fishing seasons ranging from one to six months. The F was still an annual rate but was applied over fewer months and adjusted so that the number of fish removed was the same as when DF=1. For seasons shorter than a full year, fishing was as- sumed to occur at the beginning of the year. Like the fishing season, the duration of the spawn- ing season was a full year in the base model (Ds=l). 396 Fishery Bulletin 103(2) Table 2 Percent selection differential on the von Bertalanffy growth coefficient (A") at fishing mortality = 0.8/yr. Columns correspond to the levels of the coefficient of variation (CV=0%, 10%, 20%) in A' and in the asymptotic length (L„). Any combination with 0% CV in A is not presented because it results in zero selection differential. The first row corresponds to the base model and subsequent rows correspond to changes in the base model: correlation between L , and K(p), slope of maturity curve (ft,), natural mortality (M), M/K ratio, duration of annual spawning season (Ds), maximum size limit (Lu), slope of selectivity curve (/} ), change in age at 50% selectivity (A J in relation to the base case, and duration of annual fishing season lDF). Parameter values L„:0%CV L/. 10%CV L,_ : 20%CV L„ : 0%CV LX:W%CV L„: 20%CV A: 10% CV A: 10%CV A 10%CV A: 20% CV A:20%CV A: 20%CV Base 0.7 0.5 0.3 2.1 1.7 1.2 P = -1 0.7 -0.7 -1.3 2.1 0.2 -2.3 p = -0.75 0.7 -0.3 -0.8 2.1 0.8 -0.9 p = -0.5 0.7 0.0 -0.4 2.1 1.1 0.1 p=-0.25 0.7 0.3 0.0 2.1 1.4 0.6 Pm = 0.25 0.2 0.2 0.1 0.7 0.7 0.6 ft, = 0.5 0.3 0.3 0.2 1.1 1.1 0.9 l\„ = 1 0.5 0.4 0.3 1.7 1.4 1.1 M = 0.1 0.4 0.4 0.3 1.6 1.5 1.2 M = 0.4 0.7 0.4 0.3 2.0 1.5 1.0 M = 0.8 0.6 0.3 0.2 1.6 1.1 0.7 M/A=0.5 0.6 0.3 0.1 1.9 1.0 0.6 M/K= 1 0.4 0.3 0.2 1.5 1.3 0.8 M/A=2 0.5 0.4 0.3 1.9 1.6 1.2 M/A=2.5 0.8 0.6 0.4 2.4 2.1 1.5 Ds= 1/12 1.6 1.0 0.6 4.5 3.6 2.3 Ds=3/12 1.4 0.9 0.5 4.1 3.3 2.2 Ds = 6/12 1.1 0.8 0.5 3.3 2.7 1.8 L„ = 700 0.0 0.0 0.0 -0.1 -0.1 0.0 Lu = 800 0.2 0.0 0.0 0.3 0.2 0.0 Lu = 900 0.4 0.2 0.1 1.1 0.8 0.3 ft = 0.01 0.3 0.2 0.2 1.1 1.0 0.8 ft = 0.05 0.6 0.4 0.3 1.9 1.6 1.2 ft = 0.1 0.6 0.5 0.3 2.0 1.7 1.2 As = -2 0.1 0.2 0.2 1.1 1.2 1.0 A, = -1 0.4 0.4 0.3 1.7 1.6 1.1 A, = l 0.6 0.5 0.3 2.1 1.7 1.2 As = 2 0.5 0.4 0.3 1.9 1.6 1.1 £>F= 1/12 1.5 1.0 0.6 4.3 3.5 2.3 DF = 3/12 1.3 0.9 0.6 3.9 3.1 2.1 £>,, = 6/12 1.2 0.7 0.5 3.3 2.6 1.8 In sensitivity analyses, the spawning season ranged from one to six months and was assumed to occur at the end of the year. A selection differential cannot exist without phe- notypic variation. The base model assumed a coeffi- cient of variation (CV) of 20% in both L^ and K. For sensitivity analyses, combinations of 09c , 10%, and 20% CV in Lx and K were examined for the influ- ence of growth variability on selection differentials of L and K. Results Changes in growth parameters L, and K affect size-at- age jointly, resulting in non-uniform selection differ- entials across ages (Fig. 3). The selection differentials on size are bounded by the differentials at the extreme ages, t0 and ». At the youngest age, the selection dif- ferential on size is limited by the sum of the selection differentials on L r and K plus their product. (At age t0, the selection differential on size is undefined.) As age Williams and Shertzer: Effects of fishing on growth traits: a simulation analysis 397 10 0.8 06 0 4 0.2 0.0' 10 Age o 1.0- B —7y — if , - ' 08- I 0.6- • 0.4- / 0.2- i i o.o- j i 200 400 600 800 Length 1000 Figure 2 Effect of the slope parameter on (A) the prob- ability of maturity and (B) the probability of selection. (A) Maturity slope parameter /3m = 0.25 (light dash), /i„, = 0.5 (light solid!, ft,, = 1.0 (heavy dash), and Pm=°° (heavy solid). (B) Selectivity slope parameter /3S = 0.01 (light dash). /3S = 0.05 (light solid), /3S = 0.1 (heavy dashl, and /3S = ^ (heavv solid). increases, the selection differential on size increases or decreases monotonically toward an asymptote, the selection differential on Lx. Thus selection differentials on size across all ages are bounded by those at L _, + K + Lx K and Lx. The selection differential on the small- est fish (age approaching t0) is an upper bound when the selection differential on K is positive, and a lower bound when negative. These properties are important for interpreting how selection differentials on size-at-age correspond to differentials on Lx and K. Using the base model, we computed selection differen- tials on Lx and K as functions of fishing mortality, over the range F=Q to F=10/yr. The selection differentials increased with F nonlinearly, resulting in a concave relationship (Fig. 4). However for F<2.0, the relation- ship is nearly linear. The alternative models also revealed linear relation- ships between selection differentials and F, for F<2.0 (figures not shown). In addition, those relationships have a zero intercept (by definition, no fishing, no selec- tion differential). Because the relationships are (nearly) linear and have a common intercept, the rank of selec- 1000 800 £ 600 400 200 ~i 1 1 r~ 5 10 15 20 B o - Ni 8 - ■-. 6 - *•-._ V. 4 - ** ~ 2 - "* .. 0 - I I I I I 0 5 10 15 20 Age Figure 3 Hypothetical changes in length, given changes in growth parameters. (Ai Growth trajectories in the base model (solid), a 59c de- crease in growth parameter K (dash), a 5r/r decrease in growth parameter L ^ (dot), and bc/< decrease in both parameters (dash-dot). (B) The corresponding reductions in length are relative to the base model. tion differentials among models does not change across values of F. A model that bears the highest selection dif- ferential at F=0.2 does so at F=2.0. We therefore present results of sensitivity analyses for a single value of F (F=0.8/yr), with the understanding that for other values of F (up to 2.0), magnitudes of selection differentials can be inferred and ranks among models are maintained. Increased variation in Lx and K tended to increase the selection differentials, and interaction between the two growth parameters (Tables 2 and 3). Selection dif- ferentials on Lx were generally larger than those on K. In the base model, the largest selection differential on each growth parameter occurred when variation in the focal parameter was highest and variation in the other parameter was zero. The selection differentials on size- at-age were largest when variation in both parameters was highest (20% CV for both Lx and K). 398 Fishery Bulletin 103(2) Life-history parameters The correlation (p) between L A and K was assumed to be zero in the base model and negative in sensitivity analyses. The effect of correlation depended on variation in the growth parameters. When the CV was zero for either parameter, correlation had no effect on selection differentials (Tables 2 and 3). When the CV was posi- tive for both, a negative correlation decreased selection differentials in relation to those from the base model (Tables 2 and 3). For decreased values of the correlation coefficient (i.e., stronger negative correlation), the per- cent selection differentials on K decreased, whereas the percent selection differentials on Lx either decreased or remained constant. The percent selection differentials on the size near age t0 ranged from 3.7% to -0.1% for values of p from 0 to -1. The percent selection differentials on Lx remained relatively constant, ranging from 2.1% to 2.5%, with the highest at p=0 (Fig. 5). Knife-edge maturity (/3m = oc) resulted in larger selec- tion differentials than did other maturity curves (Tables 2 and 3 1. As the slope of the maturity curve became more gradual, the selection differentials decreased. For /3m values greater than 1, the selection differentials on size were similar to those of the knife-edge case (Fig. 5). The effect of M on selection differentials was relative- ly small (Tables 2 and 3). Changes in M from 0.1 to 0.8 led to small changes in selection differentials (Fig. 5). The largest selection differentials tended to occur near intermediate values of M (Tables 2 and 3, Fig. 5). This nonlinear response in the selection differentials is not surprising because changes in M affected the values of K, Am, and maximum age nonlinearly (Table 1). Changes in the M/K ratio did not reveal a clear trend (Tables 2 and 3, Fig. 5). As with M, the M/K ratio af- fects other parameters; therefore changes in M/K could be expected to produce a nonlinear response in the selection differentials. The percent selection differen- tial on Lx was lowest at an intermediate value of Ml K-2 (Table 3). The percent selection differentials on K showed no consistent trend (Table 2). For M/K values from 0.5 to 2.5, the selection differentials on size across ages ranged from 2.3% to 4.0% (Fig. 5). Decreases in the spawning season duration (Z)s) caused a near linear increase in the selection differen- tials (Tables 2 and 3, Fig. 5). A compressed spawning duration of one month resulted in a range of 5.0% to 7.4% selection differential on size across ages (Fig. 5). Of all the life-history parameters examined in this analysis, spawning duration had the greatest effect. Fishery parameters A limit (Lu) on sizes susceptible to the fishery decreased the selection differentials (Tables 2 and 3, Fig. 5). The percent selection differential at all ages was zero for Lu = 800 and -0.1% for Lu = 700 (Fig. 5). In these analy- ses, F was held constant. Consequently, smaller values of Lu correspond to fewer fish removed. An alternative approach would have been to maintain constant catch 0 2 4 6 Fishing motality (per year) Figure 4 Selection differentials on growth parameters (A) K and (B) Lml computed as functions of fishing mortality. Parameter values are the same as those in the base model. by increasing F, which would have led to selection dif- ferentials larger than those in Tables 2 and 3. Knife-edge selectivity (jis = x) caused larger selec- tion differentials than did selectivity curves with more gradual slopes (Tables 2 and 3). For )3S greater than 0.1, the selection differential rapidly converged to that of the knife-edge case (Fig. 5). As with Lu, F was held constant across /3S sensitivity analyses. A change in the ages of fishery selectivity had little effect on selection differentials (Tables 2 and 3, Fig. 5). When selectivity was set to a larger age or size, the selection differential decreased slightly. In this case, selectivity was occurring after maturity, allowing more fish to reproduce before reaching sizes selected by the fishery. However if harvest had been held constant in- stead of F, the selection differentials would have been larger. When selectivity was set to a smaller age or size, the selection differential decreased slightly or remained constant. This result is due to a reduction in the time exposed to differential fishing mortality. Differential fishing mortality occurs only on the sizes where se- lectivity is less than one; otherwise fishing mortality is constant for all individuals. Under von Bertalanffy growth, younger fish grow more quickly. A decrease in the age or size of selectivity shifts the fishing pres- sure to ages with quicker growth, reducing the time Williams and Shertzer: Effects of fishing on growth traits: a simulation analysis 399 Table 3 Percent selection differential on the von Bertal anffy asymptotic length (Lr ) at fishing mortality = ).8/yr. Columns correspond to the levels of the coefficient of variation (CV=09i ,10% ,20% ) inL, and in the gro »vth coefficient (K). Any combination with 0% CV in t, is not presented because it results in zero selection differential. The first row corresponds to the base mo del and subsequent rows correspond to changes in the base model: correlation between L , and Kip . slope of maturity curve (ft, , natural mortality (M),M/K ratio duration of annual spawning season (Ds), maximum size limit CL ), slope of selectivity curve (ft), change in age at 50% selectivity (As) in relation to the base case, and duration of annual fishing season (DF). Parameter values L„:0%CV L„: 10%CV L, : 20%CV L„ : 0%>CV L j:io%cv L,:20%CV K: 10%CV K: 10%CV K 10%CV K: 20% CV K 20%CV K. 20%CV Base 1.0 2.7 0.9 2.7 0.8 2.5 P = -1 1.0 2.8 0.7 2.7 -0.1 2.3 p = -0.75 1.0 2.7 0.7 2.6 0.2 2.2 p= -0.5 1.0 2.8 0.8 2.7 0.4 2.3 p = -0.25 1.0 2.8 0.9 2.7 0.6 2.4 Pm = °'25 0.3 1.2 0.3 1.1 0.3 1.1 ft, = 0-5 0.5 1.8 0.5 1.8 0.5 1.7 ft,= l 0.8 2.4 0.7 2.4 0.7 2.2 M = 0.1 0.8 2.6 0.7 2.5 0.7 2.4 M = 0.4 1.0 2.8 1.0 2.7 0.8 2.6 M = 0.8 1.0 2.6 1.0 2.6 0.8 2.4 MIK = 0.5 1.4 3.2 1.4 3.2 1.3 3.1 MIK= 1 0.9 2.7 0.9 2.7 0.8 2.5 MIK =2 0.9 2.5 0.8 2.5 0.7 2.3 M/X = 2.5 1.1 2.8 1.0 2.7 0.8 2.5 Ds= 1/12 2.2 5.6 2.0 5.5 1.7 5.0 Ds = 3/12 2.0 5.1 1.8 5.0 1.5 4.6 Ds = 6/12 1.6 4.4 1.5 4.3 1.3 3.9 Lu = 700 -0.1 -0.1 -0.1 -0.1 -0.1 -0.1 L„ = 800 0.0 -0.1 0.0 -0.1 -0.1 -0.1 Lu = 900 0.3 0.5 0.2 0.5 0.1 0.4 ft = 0.01 0.5 1.9 0.5 1.9 0.5 1.8 ft = 0.05 0.9 2.7 0.9 2.6 0.8 2.5 ft = O-1 1.0 2.7 0.9 2.7 0.8 2.5 As = -2 0.4 2.1 0.4 2.1 0.4 2.1 As = -1 0.7 2.5 0.7 2.5 0.7 2.4 A8=l 1.0 2.8 1.0 2.7 0.9 2.5 A, = 2 1.0 2.7 0.9 2.6 0.8 2.4 DF= 1/12 2.2 5.5 2.0 5.3 1.6 4.8 £>f = 3/12 1.9 4.9 1.7 4.8 1.5 4.4 DF = 6/12 1.6 4.2 1.5 4.1 1.2 3.7 individuals experience differential fishing pressure and therefore the potential for selection differentials. If har- vest had been held constant instead of F, the selection differentials would have been larger. The fishing season duration (DF) affected selection differentials in ways similar to the spawning season duration (Tables 2 and 3, Fig. 5). A fishing season of one month resulted in an upper bound of selection dif- ferentials that ranged from 4.8% to 7.3% over all ages (Fig. 5). Of all the fishery parameters examined in this analysis, a concentrated fishing season resulted in the largest selection differentials. Discussion The individual-based simulation approach used here simplifies computation of selection differentials and 400 Fishery Bulletin 103(2) A B c 6- 6- 6- 4- 2- O o- -^ Q^-O • 2. o o ■ -//— 4" O o-» 0— _____ — o 0- rT 0- 0- .0 -0.6 -0.2 0 2 4 6 °° 0.1 0.3 0.5 0.7 P An M D E F 6- O I 4' cu j5 2- 6- --o--- •-.Q---0 2_ ^O 6- °- -ex \. -O.^ \. 4- "*• 2- • D 0- 0- 0- „ n— 8' 15 1.0 1.5 2.0 2.5 02 0.4 0.6 0.8 1.0 700 800 900 1000 M/K Ds Lu G H I 0 6- 6- 6- o 4- oo— —H— • 4" o_o— — °— o 4" o---o---»---o---o 2. 2- q»-- -•//--•„ O 2- - -• 0- 0- 0- 0.0 02 0.4 0.6 0.8 °° -2 -1 0 1 2 02 0.4 0.6 0.8 1.0 A ^s DF Figure 5 Upper and lower bounds of selection differentials on size across all ages. Solid line represents selection differential on the size near age f0; dashed line represents selec- tion differential on Ly. (A) correlation between L, and K (p); (B) slope of maturity curve (/3m); (C) natural mortality (M); ID) MIK ratio; (E) duration of annual spawn- ing season (Ds); (F) maximum size limit (Lu); (Gl slope of selectivity curve (j3s); (H) change in age at 50* selectivity (As) relative to the base case, and (I) duration of annual fishing season (DF). In all panels, CV's in K and L x are 20%. Filled circles refer to the base model. isolates the cause — fishing. Yet with any simulation analysis, one must interpret results in light of model assumptions. With our model maturity was assumed to be a function of age, and the computation of selection dif- ferentials were consequently focused to those on growth traits and size. If maturity were considered a function of size, it too would have been subject to a selection differential. Changes in size or age at maturity have been considered in other studies (Stokes and Blythe, 1993; Haugen and Vollestad, 2001; Olsen et al., 2004) and are likely connected to growth parameters through bioenergetic constraints. A central assumption is that somatic growth follows the von Bertalanffy model. That model was chosen be- cause of its successful track record (Chen et al., 1992; Quinn and Deriso, 1999). Life-history characteristics other than growth are assumed to follow life-history invariant relationships. The invariants constrain bio- logical parameters to values that represent an "average stock." Of course, no stock is truly average, and there- fore our sensitivity analyses incorporate considerable deviation from life-history invariants. In our simulation, the largest selection differentials occurred when the spawning or fishing seasons were Williams and Shertzer: Effects of fishing on growth traits: a simulation analysis 401 compressed. We modeled fishing seasons at the begin- ning of the year and spawning seasons at the end of the year, and in a single-year simulation, the annual timing of the fishing and spawning seasons will affect selection differentials. For example, if the one-month fishing season had been modeled at the end of the year, the selection differential would be smaller because of the 11 months of spawning prior to fishing mortality. Over multiple years, however, the annual timing of the fishing and spawning seasons is less important than their duration and overlap. Our model simulated selection differentials at the onset of a fishery. As a fishery progresses, selection differentials should decrease as life-history parameters shift in the direction of selection. A multiyear simula- tion of evolution would require knowledge or assump- tions about heritability and trait distributions, both of which are likely to be dynamic. Even so, a short-term simulation, where selection differentials and heritabil- ity are assumed to be static, may be an informative approximation. We simulated evolution of the base-model population, assuming a static heritability of 0.2 and selection differ- entials of 2.5% for Lv and 1.2% for K (values from Tables 2 and 3 with 20% CV's in both parameters). Two simu- lations were conducted with different values for fishing mortality. With F = AM, five years of evolution led to a 9.0% decrease in the capacity of spawning biomass. With F = M, five years led to a 2.3% decrease. With real fishery data it is often impossible to docu- ment conclusively that fishing causes a genetic change in growth. Any such change may be hard to measure, fall within the range of statistical variability due to sampling, or be masked by strong year classes. Selec- tion for reduced growth may be compensated by den- sity-dependent effects (for example, lower abundance leaving more resources for survivors to allocate towards growth). Even when a change can be demonstrated, fishing is just one potential explanation. Alternative explanations include environmentally driven evolution and reaction norms (i.e., phenotypic expressions of a genotype-environment interaction). Nonetheless, size-selective fishing is widespread and often accompanies changes in somatic growth rates (Ricker, 1981; Harris and McGovern, 1997; Haugen and Vollestad, 2001; Sinclair et al., 2002). Until recently, the question was whether fishing can cause changes in growth that are evolutionary, and the answer was "yes . . . probably." The laboratory experiments of Conover and Munch (2002) removed any doubt. However, those experiments represented an extreme fishery in terms of its potential to inflict a selection differential: high F compressed in time (90% of population removed in one day), knife-edge selectivity, non-overlapping gen- erations, and a population where all individuals are susceptible. The goal of our study was to shed light on selection differentials created by fishing under realistic ranges of life-history and fishery characteristics. Understanding how life-history characteristics affect selection differen- tials is important for identifying which stocks are most susceptible to evolution of growth traits. For example, susceptibility increases with compression of the spawn- ing season. Fish species with compressed spawning seasons, such as many anadromous species, may be at higher risk of evolution from size-selective fisheries. Understanding how fishery patterns affect selec- tion differentials has direct management implications because it is the fishery parameters that can be con- trolled. For example, our results indicate that size-selec- tive fisheries compressed in time are apt to cause high selection differentials. Managers should avoid "derby" style harvests, such as the annual Pacific herring sac- roe fisheries, which are completed in only a few days. Other management strategies could reduce selection differentials, such as slot limits, reduction in the slope of selectivity curves, and partial selectivity after the age at maturity. However, because no size-selective fishing pattern can preclude some directional selection on growth, management by area closures may be the best option for avoiding fishery-induced evolution of growth traits. As fishing technology improves, so does the ability to fully and rapidly exploit fish populations, and thus increase the potential for evolutionary responses. Still, when overfishing depletes a stock, low abundance is usually the paramount concern. With appropriate man- agement, stock abundance may recover, but pre-fishing growth capacity may recover more slowly or not at all if genetic variation is lost. Given plausible heritabili- ties of growth traits, this analysis shows that under a wide variety of life-history and fishery characteristics, selection differentials are large enough to allow for rapid evolution. Acknowledgments We thank R. Munoz, M. Prager, and D. Vaughan for comments on the manuscript. This work was supported by the National Marine Fisheries Service through its Southeast Fisheries Science Center. Literature cited Beverton, R. J. H. 1992. Patterns of reproductive strategy parameters in some marine teleost fishes. J. Fish. Biol. 41(suppl. B):137-160. Blythe, S. P., and T. K. Stokes. 1993. Size-selective haresting and at-at-maturity. I: Some theoretical implications for management of evolving resources. In The exploitation of evolving resources (T. M. Stokes, J. M. McGlade, and R. Law, eds.), p. 232-247. Lecture Notes in Biomathematics 99. Springer-Verlag, Berlin. Bronikowski, A. M., M. E. Clark, F. H. Rodd, and D. N. Reznick. 2002. Population-dynamic consequences of predator- induced life history variation in the guppy (Poecilia reticulata). Ecology 83:2194-2204. 402 Fishery Bulletin 103(2) Charnov, E. L. 1993. Life history invariants: some explorations of sym- metry in evolutionary ecology, 159 p. Oxford Univ. Press, Oxford, England. Charnov, E. L.. and U. Skuladottir. 2000. Dimensionless invariants for the optimal size (age) of sex change. Evol. Ecol. Res. 2:1067-1071. Chen, Y., D. A. Jackson, and H. H. Harvey. 1992. A comparison of von Bertalanffy and polynomial functions in modeling fish growth data. Can. J. Fish. Aquat. Sei. 49:1228-1235. Conover, D. O., and S. B. Munch. 2002. Sustaining fisheries yields over evolutionary time scales. Science 297:94-6. Edley. M. T., and R. Law. 1988. Evolution of life histories and yields in experi- mental populations of Daphnia magna. Biol. J. Linn. Soc. 34:309-326. Falconer, D. S„ and T. F. C. Mackay. 1996. An introduction to quantitative genetics, 4th ed., 245 p. Longman Group Ltd., Harlow, Essex, England. Frisk, M. G., T. J. Miller, and M. J. Fogarty. 2001. Estimation and analysis of biological parameters in elasmobranch fishes: a comparative life history study. Can. J. Fish. Aquat. Sci. 58:969-981. Gjedrem, T. 1983. Genetic variation in quantitative traits and selective breeding in fish and shellfish. Aquaculture 33:51-72. Hadley, N. H., R. T. Dillon, and J. J. Manzi. 1991. Realized heritability of growth rate in the hard clam Mercenaria mercenaria. Aquaculture 93:109-119. Harris, P. J., and J. C. McGovern. 1997. Changes in the life history of red porgy, Pagrus pagrus, from the southeastern United States, 1972- 1994. Fish. Bull. 95:732-747. Haugen, T. 0., and L. A. Vollestad. 2001. A century of life-history evolution in grayling. Ge- netica 112-113 and 475-491. He. J. X., and D. J. Stewart. 2001. Age and size at first reproduction of fishes: predic- tive models based only on growth trajectories. Ecology 82:784-791. Heino, M. 1998. Management of evolving fish stocks. Can. J. Fish. Aquat. Sci. 55:1971-1982. Henryon, M., A. Jokumsen, P. Berg, I. Lund, P. B. Pederson, N. J. Olesen, W. J. Slierendrecht. 2002. Genetic variation for growth rate, feed conver- sion efficiency, and disease resistance exists within a farmed population of rainbow trout. Aquaculture 209: 59-76. Hendry, A. P., J. K Wenburg, P. Bentzen, E. C. Volk, T. P. Quinn. 2000. Rapid evolution of reproductive isolation in the wild: evidence from introduced salmon. Science 290: 516-518. Hutchings, J. A. 1993. Adaptive life histories effected by age-specific survival and growth rate. Ecology 74:673-684. Jarayabhand, P., and M. Thavornyutikarn. 1995. Realized heritability estimation on growth rate of oyster, Saccastrea cucullata Born, 1778. Aquaculture 138:111-118. Law. R. 1991. On the quantitative genetics of correlated char- acters under directional selection in age-structured populations. Phil. Trans. R. Soc. Lond. B 331:213- 223. 2000. Fishing, selection, and phenotypic evolution. ICES J. Mar. Sci. 57:659-668. 2001. Phenotypic and genetic changes due to selective exploitation. In Conservation of exploited species (J. D. Reynolds, G. M. Mace, K. H. Redford, and J. G. Robinson, eds.l, p. 323-342. Cambridge LIniv. Press, Cambridge, England. Law, R., and C. A. Rowell. 1993. Cohort-structured populations, selection responses, and exploitation of the North Sea cod. In The exploita- tion of evolving resources (T. K. Stokes, J. M. McGlade, and R. Law, eds.), p. 155-173. Lecture Notes in Bio- mathematics 99. Springer-Verlag, Berlin, Germany. Mangel, M. 1996. Life history invariants, age at maturity and the ferox trout. Evol. Ecol. 10:249-263. McAllister, M. K., R. M. Peterman, and D. M. Gillis. 1992. Statistical evaluation of a large-scale fishing exper- iment designed to test for a genetic effect of size-selective fishing on the British Columbia pink salmon iOncorhyn- chus gorbuscha). Can. J. Fish. Sci. 49:1294-1304. Miller, L. M., and A. R. Kapuscinski. 1994. Estimation of selection differentials from fish scales: a step towards evaluating genetic alteration of fish size in exploited populations. Can. J. Fish. Sci. 51:774-783. Olsen, E. M., M. Heino. G. R. Lilly, M. J. Morgan, J. Brattey, B. Ernande, and U. Dieckmann. 2004. Maturation trends indicative of rapid evolution preceded the collapse of northern cod. Nature 428: 932-935. Quinn, T. J., II, and R. B. Deriso. 1999. Quantitative fish dynamics, 542 p. Oxford Univ. Press, New York, NY. Quinn, T. P., M. T. Kinnison, and M. J. Unwin. 2001. Evolution of Chinook salmon (Oncorhynchus tshaw- ytscha) populations in New Zealand: pattern, rate, and process. Genetica 112-113:493-513. Ratner, S., and R. Lande. 2001. Demographic and evolutionary responses to selective harvesting in populations with discrete generations. Ecology 82:3093-3104. Reznick, D. N., F. H. Shaw, F. H. Rodd, and R. G. Shaw. 1997. Evaluation of the rate of evolution in natural populations of guppies (Poecilia reticulata). Science 275:1934-1937. Ricker, W. E. 1981. Changes in the average size and average age of Pacific salmon. Can. J. Fish. Aquat. Sci 38: 1636-1656. Roff, D. A. 1984. The evolution of life-history parameters in teleosts. Can. J. Fish. Aquat. Sci. 41:984-1000. Shertzer, K. W., and S. P. Ellner. 2002. Energy storage and the evolution of population dynamics. J. Theor. Biol. 215:183-200. Shertzer, K. W., S. P. Ellner, G. F Fussmann, and N. G. Hairston Jr. 2002. Predator-prey cycles in an aquatic microcosm: testing hypotheses of mechanism. J. Anim. Ecol. 71:802-815. Sinclair, A. F, D. P. Swain, and J. M. Hanson. 2002. Disentangling the effects of size-selective mortal- Williams and Shertzer: Effects of fishing on growth traits: a simulation analysis 403 ity, density, and temperature on length-at-age. Can. J. Fish. Aquat. Sci. 59:372-382. Sinervo, B., E. Svensson, and T. Comendant. 2000. Density cycles and an offspring quantity and quality game driven by natural selection. Nature 406:985-988. Stearns, S. C. 1992. The evolution of life histories, 379 p. Oxford Univ. Press, Oxford, England. Stockwell, C. A., A. P. Hendry, and M. T. Kinnison. 2003. Contemporary evolution meets conservation biology. Trends Ecol. Evol. 18:94-101. Stokes, T K., and S. P. Blythe. 1993. Size-selective haresting and at-at-maturity. II: Real populations and management options. /;; The exploitation of evolving resources (T. K. Stokes, J. M. McGlade, and R. Law (eds.), p. 232-237. Lecture Notes in Biomathematics 99, Springer-Verlag, Berlin. Stokes, K., and R. Law. 2000. Fishing as an evolutionary force. Mar. Ecol. Prog. Ser. 208:307-309. Vandeputte, M., E. Quillet, and B. Chevassus. 2002. Early development and survival in brown trout iSalmo trutta fario L.l: indirect effects of selection for growth rate and estimation of genetic parameters. Aqua- culture 204:435-445. Weigensberg, I., and D. A. Roff. 1996. Natural heritabilities: can they be reliably esti- mated in the laboratory. Evolution 50:2149-2157. Williams, E. H., and K. W. Shertzer. 2003. Implications of life-history invariants for biological reference points used in fishery management. Can. J. Fish. Aquat. Sci. 60:710-720. Yoshida, T, L. E. Jones, S. P. Ellner, G. F Fussmann, and N. G. Hairson Jr. 2003. Rapid evolution drives ecological dynamics in a predator-prey system. Nature 424:303-306. Xiao, Y. 1994. Von Bertalanffy growth models with variability in, and correlation between, K and L„. Can. J. Fish. Aquat. Sci. 51:1585-1590. 404 Preliminary evidence of increased spawning aggregations of mutton snapper (Lutjanus analis) at Riley's Hump two years after establishment of the Tortugas South Ecological Reserve Michael L. Burton Kenneth J. Brennan Roldan C. Munoz Richard. O. Parker Jr. Center for Coastal Fisheries and Habitat Research National Marine Fisheries Service National Oceanic and Atmospheric Administration 101 Pivers Island Rd Beaufort. North Carolina 28516-9722 E-mail address Michael.BurtoniSnoaa.gov In this note we describe the re-for- mation of a spawning aggregation of mutton snapper {Lutjanus analis). A review of four consecutive years of survey data indicates that the aggre- gation may be increasing in size. Mutton snapper are distributed in the temperate and tropical waters of the western Atlantic Ocean from Flor- ida to southeastern Brazil (Burton, 2002). Juveniles and subadults are found in a variety of habitats such as vegetated sand bottoms, bays, and mangrove estuaries (Allen, 1985). Adults are found offshore on coral reefs and other complex hardbottom habitat. They are solitary and wary fish, rarely found in groups or schools except during spawning aggrega- tions (Domeier et al., 1996). Spawn- ing occurs from May through July at Riley's Hump (Domeier et al., 1996) and peaks in June, as indicated by gonadosomatic indices (M. Burton, unpubl. data). Mutton snapper are highly prized by Florida fishermen for their size and fighting ability, and the majority of landings occur from Cape Canaveral, , through the Flor- ida Keys, including the Dry Tortugas (Burton, 2002). Reports of spawning aggregations of tropical reef fishes are abundant in the fisheries literature. Most docu- mented aggregations of commercially important fishes are attributed to members of the grouper family, Ser- ranidae, including observations of spawning Nassau grouper iEpineph- elus striatus), red hind (E. guttatus), and tiger grouper {Mycteroperca ti- gris) in the Caribbean (see review in Domeier and Colin, 1997, and refer- ences therein). Eklund et al. (2000) observed black grouper (M. bonaci) aggregating during their spawning season just outside no-take zones along the Florida Keys reef tract. Samoilys and Squire (1994) and Samoilys (1997) documented spawn- ing aggregations of coral trout (Plec- tropomus leopardus) from the Great Barrier Reef, and Johannes (1988) described the aggregating behavior of squaretail coralgrouper (P. areola- tus) from the Solomon Islands. Most recently, Sala et al. (2003) observed aggregating behavior in two species of serranids — the sawtail grouper (M. prionura) and the leopard grou- per (M. rosacea) from the Gulf of California. There are fewer descriptions of spawning aggregations of the com- mercially important snappers (Lut- janidae) in the literature. Wicklund (1969) described spawning behavior of lane snapper {Lutjanus synagris) from southeast Florida, Carter and Perrine (1994) described a spawning aggregation of dog snapper (L. jocu) from Belize, and Sala et al. (2003) described spawning behavior in two lutjanids from the Gulf of California (yellow snapper, L. argentiventris; Pa- cific dog snapper, L. novemfasciatus). Mutton snapper (L. analis) are per- haps the best known snapper to form spawning aggregations. Craig (1966) observed concentrated commercial fishing on an apparent "spawning run" of mutton snapper in August at Long Cay, Belize. Domeier and Colin (1997) described an aggregation of L. analis in the Turks and Caicos Islands in April 1992, and Domeier et al (1996) identified a spawning ag- gregation at Riley's Hump. Because of their predictable nature with respect to location and time, spawning aggregations become ex- tremely vulnerable to heavy exploi- tation once discovered by fishermen. The majority of annual catches of Nassau grouper in some areas comes from annual spawning aggregations (Colin, 1992; Aguilar-Perera and Aguilar-Davila, 1996), whereas other aggregations have been completely extirpated (Olsen and LaPlace, 1978; Sadovy and Eklund, 1999; Heyman, 2003). Russ (1991) observed that uncontrolled fishing on spawning aggregations could lead to recruit- ment overfishing. During a May 1991 survey of Riley's Hump, a site of a known mutton snapper spawning ag- gregation in the Dry Tortugas, Flor- ida, Domeier and Colin (1997) noted that fish were more scattered and far less abundant than they were at the Turks and Caicos site. The authors suggested that this difference was at- tributable to heavy commercial fish- ing pressure at Riley's Hump during the several years prior to 1991. Although recent literature indi- cates that fishing pressure on Riley's Hump has been intensive for several years prior to 1991 (Domeier and Colin, 1997), anecdotal information indicates otherwise. According to a commercial hook-and-line fisherman who fished on Riley's Hump from 1978 through 2001, the first known Mansucript submitted 20 December 2003 to the Scientific Editor's Office. Manuscript approved for publication 29 December 2004 by the Scientific Editor. Fish. Bull. 103:404-410 (2005). NOTE Burton et al.: Spawning aggregations of Lut/anus analis at Riley's Hump 405 instance of commercial fishing on this area occurred in 1968 by a fisherman named Riley.1 However, the naviga- tion device in common use in 1968 was LORAN (long range navigation) A; thus, the likelihood of a fisherman finding the exact spot where he fished previously was much less likely than with today's global positioning system (GPS) receivers. Large-scale commercial fishing of Riley's Hump began in 1976, with the introduction of the improved LORAN C navigation system. Commercial fishermen began fishing the area with longline gear in 1979, and fish traps were introduced there in 1984. This was the period of the most intensive fishing; longliners harvested between 10 and 21 metric tons per trip and fish trappers typically landed an aver- age of 11.5 metric tons (Gladding1). It is necessary to rely on knowledgeable fishermen for anecdotal data such as this because the National Marine Fisheries Service (NMFS) did not separate out individual species in their data sets prior to 1986, instead consolidating all snap- pers into an unclassified snapper category. After 1986, landings from the Dry Tortugas were included with the rest of the Florida Keys in a Monroe County total; therefore it is virtually impossible to obtain an exact magnitude of the landings from the Dry Tortugas for this time frame without information from knowledge- able fishermen who were involved in the fishery at the time. In addition to the commercial effort, a small fleet of headboats ran multiday fishing trips to Riley's Hump and other areas in the Dry Tortugas (Dixon2). Fishermen began to realize declining catches in the mid-1980s and brought this to the attention of the fish- ery management councils. The Gulf of Mexico Fishery Management Council (GMFMC) enacted a spawning- season closure in 1992, prohibiting fishing on Riley's Hump in May and June (Gulf of Mexico Fishery Man- agement Council, 1992). An analysis of pre- and postclo- sure commercial landing data revealed that, as a result of the closure, there was a shift in effort to the months on either side of the period of closure, and landings during the two-month closure decreased in only one of the months while annual landings increased (Burton, 1997). After further urging by fishermen and an effort by the Tortugas Working Group (a group of stakehold- ers appointed by the Florida Keys National Marine Sanctuary [FKNMS] Advisory Council), the Tortugas South Ecological Reserve (TSER) was created in July 2001 specifically to protect the spawning aggregation and habitat of mutton snapper. Current regulations pro- hibit all uses of the reserve, except continuous transit through the reserve, for any vessels without a FKNMS research permit. The authors initiated data collection on Riley's Hump in July 2001 to document the effect of the newly designated ecological reserve on abundance of snappers and groupers. 1 Gladding, P. 2003. Personal commun. 27A 12th Avenue, Stock Island, FL 33040. 2 Dixon, R. 2003. Personal commun. CCFHR, NMFS, NOAA, 101 Pivers Island Rd., Beaufort, NC 28516-9722. Materials and Methods Study area Riley's Hump is a carbonate bank of Holocene origin located 20 km southwest of the Dry Tortugas National Park (DTNP) island of Garden Key (Ft. Jefferson). Riley's Hump sits in the northeast corner of the TSER within the FKNMS (Fig. 1). The area has a predominantly low-relief hardbottom and patchy hard coral and scat- tered gorgonian sponge-soft coral communities. Rising to within 30 m of the surface, Riley's Hump covers an area of approximately 10 km2. Habitat mapping efforts by Franklin et al. (2000), who used a nine-tier habi- tat classification scheme, and visual observations from SCUBA dives revealed that Riley's Hump consisted mostly of areas of rocky outcropping and some patchy hard bottom in sand. More detailed multibeam mapping showed that the top of the bank is relatively flat and has an escarpment on the south side of the bank dropping from 30 m to well over 50 m deep (Fig. 2) (Mallinson et al., 2003). Sampling Initial sampling stations were selected in 2001 by divid- ing the top of Riley's Hump into a grid consisting of 0.40-km2 sections and by conducting a census with the ship's depth sounder in order to identify (within as many grids as possible) reef habitat that could be reached by dives. Ten initial stations were selected according to this procedure. Five more stations were added in 2002 at the recommendation of our vessel captain, Peter Gladding (Fig. 2). Two-man dive teams conducted several 30-m visual census strip transects (Brock, 1954) at each sta- tion during the summer months of each year, enumerat- ing all species of snappers and groupers observed. Results We summarize our observations of mutton snapper abun- dance and behavior on Riley's Hump in Table 1, along with the observation's relation in time to the lunar cal- endar. The initial sighting of an unusually large group of mutton snapper occurred on 17 July 2001. A group of 10 fish was observed by the senior author at station 2 (Fig. 2). The fish were swimming 0.5-1 m apart in a group approximately 1.5 m above the seafloor. The next year, on 27 May 2002, we observed a larger group of approximately 75-100 mutton snapper on the same site, station 2 (Fig. 2). These fish were exhibiting simi- lar behavior to that observed the preceding year. The group remained schooled while the dive team completed one 30-m visual transect and then slowly dispersed as the divers returned to the aggregation location. On 15 June 2003, a team of divers discovered an aggregation of over 200 individual mutton snapper at station 12 (Fig 2). The fish repeatedly swam up to the diver doing the census transects and then slowly turned and swam 406 Fishery Bulletin 103(2) *foj ~^i N + (Enlarged below) 0 30 60 Miles ^Mp <***^ 0 " Loggerhead Key /■ Middle Key Garden Key Bush Key Riley's Hump South Tortugas Ecological Reserve 12 Miles Figure 1 Location of Riley's Hump, Tortugas South Ecological Reserve, Florida Keys National Marine Sanctuary. away. The aggregation was spread out over a wide area, was not as dense as in the previous two sightings, and exhibited the milling behavior similar to that described by Thresher (1984) for several other species of lutjanids. This aggregation remained at the site throughout the entire 20-minute census dive. Later that day, divers recording their observations at nearby station 2 reported a group of approximately 100 mutton snapper. These fish were more widely dispersed and maintained a distance of 3-5 m from divers. Finally, on 4 July 2004, the senior author and another diver encountered a large school of approximately 300 mutton snapper at station 12, exhibiting behavior similar to that observed during the preceding year. Discussion We believe that the large groups of fish encountered at station 12 in June 2003 and again in July 2004 were spawning aggregations based on their behavior and on the timing and location of the aggregation. First, behavior of the snappers themselves was not typical of nonspawning individuals. Although Humann (1997) described them as being very curious, mutton snapper are typically described as solitary animals (Domeier and Colin, 1997), cautious of divers, and not allowing close approach. Many large reef fishes exhibit simi- lar solitary behavior, such as Nassau grouper (Smith, 1972) and black grouper (Eklund et al., 2000). The NOTE Burton et al.: Spawning aggregations of Lut/anus analts at Riley's Hump 407 24*32' 24'31' 24-3CT 24*29' -83*08' ■aarg' -83*06; -25- -30- -35- -40- -83-05' Figure 2 Multibeam bathymetric image of the top of Riley's Hump showing locations of visual census stations (white circles! and mutton snapper aggregation sightings (stations 2 and 12). Bathymetric image was provided courtesy of D. Naar and B. Donahue, Univ. S. Florida, from Mallinson et al., 2003. Table 1 Observations on mutton snapper (Lutjanus analis) on Riley's Hump and their behavior as noted by the authors. Date and station Numbers observed Behavior Moon phase 28 May-1 June 1999 31 July-3 Aug 2000 Solitary L. analis observed on 3 of 11 dives Solitary L. analis observed on 5 of 6 dives 17 Jul 2001 Station 2 10 27 May 2002 Station 2 75-100 15 June 2003 Station 2 75-100 Station 12 200+ 4 July 2004 Station 12 300 Slowly swimming, diver avoidance Slowly swimming, diver avoidance Swimming in a tightly packed group, 1.5 m off bottom Swimming in tightly packed group, 1.5 m off bottom Widely dispersed, diver avoidance Widespread aggregation, actively swimming, did not avoid divers Widespread aggregation, actively swimming, did not avoid divers Full moon May 30 New moon July 30 3 days before new moon 1 day after full moon 1 day after full moon 2 days after full moon 408 Fishery Bulletin 103(2) senior author completed over 115 dives on Riley's Hump from 1995 through 2004, and the typical mutton snap- per sighting during dives made outside the spawning season (February, 5 dives; August, 5 dives; October, 7 dives) was a single fish. In these instances, the closest approach allowed by the fish was 3 m, and when an attempt was made to approach, the fish would swim away, maintaining separation. The only exceptions to this behavior were the four sightings in which groups of fish were apparently unconcerned with the presence of divers (Table 1). Johannes (1981) described a condi- tion he termed "spawning stupor" in P. areolatus from Palau. He took this term from the Palauan fishermen's description of the fish as "stupid." We do not believe that "stupid" in this context means unaware, but more closely approximates Johannes et al.'s (1999) modified description of spawning stupor as more of a lack of concern about divers. Mutton snapper in the spawning aggregation we observed seemed aware of our presence because they approached and retreated from the divers many times. Domeier and Colin (1997) asserted that spawning or courtship behavior is easily broken off by a diver's close approach or SCUBA exhalation, although Johannes et al. (1999) offered evidence showing that this is not always the case. We conducted our dive operations primarily in the day and thus did not witness spawning, which is thought to occur at dusk or later (Domeier and Colin, 1997). Courtship behavior has not been described for mutton snapper except by Domeier and Colin (1997) who observed that fish in the Turks and Caicos aggrega- tion "milled in a dense school from the bottom to within a few meters of the surface." The mutton snapper we observed exhibited this milling behavior and did not change it because of our presence. Consistent timing of spawning with respect to a spe- cific lunar phase has long been thought to be a char- acteristic of many spawning aggregations. Johannes (1978) noted that the majority of fishes with known lunar-associated spawning rhythms spawned near the full or new moon. However, the published literature does not provide strong support for a correlation be- tween spawning of most lutjanid species and any single lunar phase. The lane snapper aggregation observed by Wicklund (1969) occurred just after the new moon but has not been corroborated since this single observa- tion. Spawning of dog snapper in Belize was variable, however, occurring three days after the new moon on Cay Glory (Carter and Perrine, 1994) and just after the full moon on English Cay (Domeier and Colin, 1997). Spawning peaks for gray snapper off Key West, Florida, were also variable, occurring on the new and full moons of June-August, although the strongest spawning peak was associated with the last quarter moon of August, half way between the new and full moons (Domeier et al., 1996). Back-calculated spawning dates of gray snap- per collected in ichthyoplankton samples near Beaufort Inlet, North Carolina, have indicated that spawning takes place primarily at the time of the new moon and secondarily at the time of the full moon (Tzeng et al., 2003). Evidence of mutton snapper spawning tends to sup- port the argument that the species spawns during a full moon, in contrast to the examples of other lutjanids above. Mutton snapper aggregations off Gladden Spit, Belize, peaked during the April and May full moons and were heavily exploited by fishermen (Heyman et al., 2001). Domeier and Colin's (1997) observation of a mutton snapper aggregation off West Caicos occurred on the April 1992 full moon, and Domeier collected specimens with hydrated oocytes from the Riley's Hump location within one day of the full moon in May 1991 (Domeier and Colin, 1997). Our observation of a small group of about 10 mutton snapper at Riley's Hump in July 2001 occurred three days before the new moon. Our observations of groups of approximately 100, 200, and 300 fish, however, occurred one day after the full moons of May 2002 and June 2003, and two days after the full moon of July 2004, respectively. In contrast, the back-calculated spawning dates of mutton snapper collected in icthyoplankton samples near Beaufort Inlet, NC, indicated that spawning occurred from two days after the full moon to three days before the new moon and that peak spawning occurred between the full moon and last quarter moon phase (Hare3). These data are not inconsistent, however, with our observations of fish beginning to aggregate on or around the full moon for spawning. Our sightings of such large groups of mutton snapper around the full moon indicate activity associ- ated with a spawning aggregation. Finally, many species of reef fishes consistently aggre- gate to spawn at specific locations at regular intervals (e.g., daily, annually). The two main hypotheses as to why reef fishes do this are to offer increased chances of 1) immediate survival of eggs and larvae, and 2) en- trapment of larvae in favorable currents for transport to suitable nursery habitat (Johannes, 1978; Lobel, 1978; Gladstone, 1994), although the former hypothesis currently has more support (Hensley et al., 1994; Peter- son and Warner, 2002). Without invoking the hypothesis of local adaptation to the aggregation sites on Riley's Hump, several studies have indicated that the physical oceanography of the region is favorable for transporting larvae spawned at Riley's Hump up the Florida Keys reef tract (Lee et al., 1994; Lee and Williams, 1999) and even as far north as Vero Beach, Florida (Domeier, 2004), presumably to suitable habitat. We believe that the specific location on Riley's Hump where we observed aggregations supports our conclusion that these were spawning aggregations. In describing lutjanid behavior Thresher (1984) said, "A key feature of reproduction ... is an extensive spawning migration to select areas along the outer reef." Observations in the literature of reef fish spawn- ing aggregations occurring on the outer reef edge, on seaward extensions or promontories, near the shelf-edge 3 Hare, J. 2002. Personal commun. Center for Coastal Fisheries and Habitat Research, National Ocean Service, NOAA, 101 Pivers Island Rd., Beaufort, NC 28516-9722. NOTE Burton et al.: Spawning aggregations of Lutjanus analis at Riley's Hump 409 break, on the reef slope or near drop-offs are numer- ous (Randall and Randall, 1963; Smith, 1972; Munro, 1974; Colin, 1992; Shapiro et al., 1993; Sadovy et al., 1994a. 1994b; Samoilys and Squire, 1994; Sala et al., 2003, and others). Heyman (2003) described a single promontory on a Belize reef that harbored spawning aggregations of 26 different species throughout the year. The mutton snapper aggregation from West Caicos (Domeier and Colin, 1997) occurred on a reef near a drop-off into deep water. The south end of Riley's Hump drops quickly from 35 m to well over 50 m. The two sites where we have observed unusually large numbers of mutton snapper are in the vicinity of this drop-off. Station 2, where we observed aggregations of various sizes in all four years, is approximately 300 m inshore of the edge, whereas station 12, where we observed the largest aggregation in June 2003 and July 2004, is within 150 m of the edge (Fig. 2). We conclude from behavior, timing, and location that we are observing spawning aggregations of mutton snapper beginning to re-form on Riley's Hump follow- ing more than two decades of intensive exploitation. Although the numbers we observed are not close to anecdotal descriptions of the numbers of fish caught during the height of the commercial fishery at this location, it is encouraging to note that we have seen an increasing number of fish for each successive year that we have surveyed these stations. It is too early to say definitively whether the fish are actually becoming more abundant, but preliminary indications are that one effect of the TSER has been to increase numbers of mutton snapper. Current research plans include con- tinued annual monitoring of transects and increased exploration for additional spawning sites, as well as an expansion of our surveys to the last quarter and new- moon phases in order to continue to try to document the exact timing of spawning. Acknowledgments We gratefully acknowledge and dedicate this paper to Peter Gladding, master of the FV Alexis M, for his superb boat handling skills and knowledge of Riley's Hump; Peter recently lost his battle with cancer and we will greatly miss his guidance and company on our trips. We acknowl- edge the contributions of Richard Stoker, first mate of the Alexis M for his repeated suggestions and help that improved our research efforts; Don Field, Don Demaria, Bill Gordon, and Ian Workman for their assistance at various times with diving efforts; Lisa Wood for her help with the figures; Jon Hare, Erik Williams, Michael Prager, and three anonymous reviewers for constructive reviews of the manuscript that greatly improved it. Literature cited Aguilar-Perera, A., and W. Aguilar-Davila. 1996. A spawning aggregation of Nassau grouper Epi- nephelus striatus (Pisces: Serranidae) in the Mexican Caribbean. Environ. Biol. Fish. 45:351-361. Allen, G. R. 1985. Snappers of the world. An annotated and illustrated catalogue of lutjanid species known to date. FAO Fish. Synop. 125:1-208. Brock. V. E. 1954. A preliminary report on a method of estimating reef fish populations. J. Wildl. Manage. 18:297-308. Burton, M. L. 1997. The effect of spawning season closures on land- ings of two reef associated species. Proc. Gulf. Caribb. Fish. Inst. 50:896-918. 2002. Age, growth and mortality of mutton snapper, Lutjanus analis, from the east coast of Florida, with a brief discussion of management implications. Fish. Res. 59:31-41. Carter, J., and D. Perrine. 1994. A spawning aggregation of dog snapper, Lut- janus jocu (Pisces: Lutjanidae) in Belize, Central America. Bull. Mar. Sci. 55:228-234. Colin, P. 1992. Reproduction of the Nassau, grouper. Epineph- elus striatus (Pisces: Serranidae) and its relationship to environmental conditions. Environ. Biol. Fish. 34: 357-377. Craig, A. K. 1966. Geography of fishing in British Honduras and adjacent coastal waters. Tech. Rep. Coastal Studies Lab. No. 28, Louisiana State Univ., Baton Rouge, LA. Domeier, M. L. 2004. A potential larval recruitment pathway originating from a Florida marine protected area. Fish. Oceanogr. 13:287-294. Domeier, M. L., and P. L. Colin. 1997. Tropical reef fish spawning aggregations: defined and reviewed. Bull. Mar. Sci. 60:698-726. Domeier, M. L., C. Koenig. and F. Coleman. 1996. Reproductive biology of gray snapper (Lutjanus griseus) with notes on spawning for other western Atlantic snappers (Lutjanidae). In Biology and cul- ture of tropical groupers and snappers (F. Arreguin- Sanchez, J. L. Munro, M. C. Balgos, and D. Pauly, eds.), p. 189-201. International Center for Living Aquatic Resources Management Conference Proceedings 48, ICLARM, Makati City, Phillipines. Eklund, A. M., D. B. McClellan, and D. E. Harper. 2000. Black grouper aggregations in relation to pro- tected areas within the Florida Keys National Marine Sanctuary. Bull. Mar. Sci. 66:721-728. Franklin, E. C, J. S. Ault, S. G. Smith, J. Luo, G. A Meester, G. A. Diaz, M. Chiappone, D. W. Swanson, S. L. Miller, and J. A. Bohnsack. 2000. Benthic habitat mapping in the Tortugas region, Florida. Mar. Geodesy 26:19-34. Gladstone, W. 1994. Lek-like spawning, parental care and mating peri- odicity of the triggerfish Pseudobalistes flavimarginatus (Balistidae). Environ. Biol. Fish. 39:249-257. Gulf of Mexico Fishery Management Council. 1992. Help proposed for mutton snapper. Gulf Fishery News 12(4):2. Hensley, D. A., R. S. Appledoorn, D. Y. Shapiro, M. Ray, and R. G. Turingan. 1994. Egg dispersal in a Caribbean coral reef fish, -110 Fishery Bulletin 103(2) Thalossoma bifasciatum . I. Dispersal off the reef platform. Bull. Mar. Sci. 54:256-270. Heyman, W. 2003. Session introduction: conservation of multi-species reef fish spawning aggregations. Proc. Gulf Caribb. Fish. Inst. 54:650-651. Heyman, W. D, R. T. Graham, B. Kjerfve, and R. E. Johannes. 2001. Whale sharks Rhincodon typus aggregate to feed on fish spawn in Belize. Mar. Ecol. Prog. Ser. 215:275-282. Humann, P. 1997. Reef fish identification: Florida, Caribbean, Baha- mas, 2nd ed., 396 p. New World Publications, Inc. Jack- sonville, FL. Johannes, R. E. 1978. Reproductive strategies of coastal marine fishes in the tropics. Environ. Biol. Fish. 3: 65-84. 1981. Words of the lagoon: fishing and marine lore in the Palau district of Micronesia, 245 p. Univ. California Press, Berkeley. CA. 1988. Spawning aggregation of the grouper Plectropomus aerolatus (Ruppel) in the Solomon Islands. In Pro- ceedings of the 6th international coral reef symposium, Townsville, Australia, 1988 (J. H. Choat, D. Barnes, M. A. Borowitzka, J. C. Coll, P. J. Davies, P. Flood, B. G. Hatcher, D. Hopley, P. A. Hutchings. D. Kinsey, G. R Orrae, M. Pichon, P. F. Sale, P. Sammarco, C. C. Wallace, C. Wilkinson, E. Wolanski, and O. Bellwood, eds.), p. 751-755. Sixth International Coral Reef Symposium Executive Committee, Townsville, Australia. Johannes, R. E., L. Squire, T. Graham, Y. Sadovy, and H. Renguul. 1999. Spawning aggregations of groupers (Serranidae) in Palau. Mar. Conserv. Res. Ser. Publ. 1, 144 p. The Nature Conservancy, Arlington, VA. Lee, T. N., M. E. Clarke, E. Williams, A. F. Szmant, and T. Berger. 1994. Evolution of the Tortugas Gyre and its influence on recruitment in the Florida Keys. Bull. Mar. Sci. 54:621-646. Lee, T. N., and E. Williams. 1999. Mean distribution and seasonal variability of coastal currents and temperature in the Florida Keys with implications for larval recruitment. Bull. Mar. Sci. 64:35-56. Lobel, P. 1978. Diel, lunar and seasonal periodicity in the repro- ductive behavior of the pomacanthid fish, Centropyge potteri, and some other reef fishes in Hawaii. Pac. Sci. 32:193-207. Mallinson, D., A. Hine, P. Hallock, S. Locker, E. Shinn, D. Naar, B. Donahue, and D. Weaver. 2003. Development of small carbonate banks on the south Florida platform margin: response to sea level and climate change. Mar. Geol. 199:45-63. Munro, J. L. 1974. The biology, ecology and bionomics of Caribbean reef fishes. Mullidae (goatfishes). Res. Rept. 3, 44 p. Zool. Dept., Univ. West Indies, Kingston, Jamaica. Olsen, D. A., and J. A. LaPlace. 1978. A study of a Virgin Islands grouper fishery based on a breeding aggregation. Proc. Gulf Caribb. Fish. Inst. 31:130-144. Peterson, C. W., and R. R. Warner 2002. The ecological context of reproductive behavior. In Coral reef fishes: dynamics and diversity in a complex ecosystem (P. F. Sale, ed.l, p. 103-118. Academic Press, New York, NY. Randall, J. E., and H. A. Randall. 1963. The spawning and early development of the Atlantic parrotfish, Sparisoma rubripinne, with notes on other scarid and labrid fishes. Zoologica 48:49-60. Russ, G. R. 1991. Coral reef fisheries: effects and yields. In The ecology of fishes on coral reefs (P. F. Sale, ed.), p. 601-635. Academic Press. New York, NY. Sadovy, Y., P. L. Colin, and M. L. Domeier. 1994a. Aggregation and spawning in the tiger grou- per, Mycteroperea tigris (Pisces: Serranidae). Copeia 1994:511-516. Sadovy, Y., and A. M. Eklund. 1999. Synopsis of biological data on the Nassau grouper, Epinephelus striatus (Bloch, 1972) and the jewfish. E. itajara (Lichtenstein, 1822). NOAATech. Rep. NMFS 146, 68 p. Sadovy, Y., A. Rosario, and A. Roman. 1994b. Reproduction in an aggregating grouper, the red hind, Epinephelus guttatus. Environ. Biol. Fish. 41:269-286. Sala, E., O. Aburto-Orapeza, G. Paredes, and G. Thompson. 2003. Spawning aggregations and reproductive behavior of reef fishes in the Gulf of California. Bull. Mar. Sci. 72:103-121. Samoilys. M. 1997. Periodicity of spawning aggregations of coral trout Plectropomus leopardus (Pisces:Serranidae) on the northern Great Barrier Reef. Mar. Ecol. Prog. Ser. 160:149-159. Samoilys. M., and L. Squire. 1994. Preliminary observations on the spawning behav- ior of coral trout, Plectropomus leopardus (Pisces: Ser- ranidae) on the Great Barrier Reef. Bull. Mar. Sci. 54:332-342. Shapiro, D. Y, Y Sadovy, and M. A. McGehee. 1993. Size, composition and spatial structure of the annual spawning aggregation of the red hind, Epinephelus gut- tatus (Pisces: Serranidae). Copeia 1993:399-406. Smith, C. L. 1972. A spawning aggregation of Nassau grouper, Epinephelus striatus (Bloch). Trans. Am. Fish. Soc. 101:257-261. Thresher, R. E. 1984. Reproduction in reef fishes, 399 p. T.F.H. Pub- lications, Neptune City, NJ. Tzeng, M. W., J. A. Hare, and D. G. Lindquist. 2003. Ingress of transformation stage gray snapper, Lutja- nus griseus (Pisces: Lutjanidae) through Beaufort Inlet. North Carolina. Bull. Mar. Sci. 72:891-908. Wicklund, R. 1969. Observations on spawning of lane snapper. Underw. Nat. 6:40. 411 Feeding habits of European hake (Merluccius merluccius) in the central Mediterranean Sea Paolo Carpentieri Francesco Colloca Department of Animal and Human Biology University "La Sapienza" Viale dell'Umversita 32 00185 Rome, Italy E-mail address (for P. Carpentieri) paolo.carpentieri@uniromal it Massimiliano Cardinale Institute of Marine Research National Board of Fisheries P.O. Box 4 45 332, Lysekil, Sweden Andrea Belluscio Giandomenico D. Ardizzone Department of Animal and Human Biology University "La Sapienza" Viale dell'Universita 32 00185 Rome, Italy European hake (Merluccius merluc- cius) is an important predator of deeper shelf-upper slope Mediterra- nean communities. It is a nectoben- thic species distributed over a wide depth range (20-1000 m) throughout the Mediterranean Sea and the north east Atlantic region (Fisher et al., 1987). Notwithstanding the ecologi- cal and economic importance (Oliver and Massuti, 1995) of hake in the Mediterranean, many aspects of its biology (e.g., recruitment and repro- duction), due to multiple spawning (Sarano, 1986) and the current state of exploitation, are poorly understood (Arneri and Morales-Nin, 2000). Recent studies on hake feeding habits in the Mediterranean (Papa- costantinou and Caragitsou, 1987; Bouaziz et al., 1990; Oliver and Mas- suti, 1995) have focused on 0-3 age groups using data from trawl catch- es (Recasens et al., 1998; Colloca et al., 2000). For this reason, trophic habits of older individuals (Bozzano et al., 1997) and possible ontogen- esis-related diet changes are almost unknown. Therefore, in this study we combined samples from trawl and gillnet fisheries collected in the same fishing ground (Colloca et al., 2000) to address these issues. Materials and methods The study area is located off the cen- tral western coasts of Italy, cover- ing 13,404 km2 between 20 and 700 meters depth (outer boundaries: lati- tude 40°52'64, longitude 13°23T3; lat- itude 42°20'30, longitude 11°16'32). Monthly size-stratified samples were obtained from spring 1997 to winter 1998 both from bottom-trawls, gillnet commercial-vessels, and from commercial landings. Trawlers catch mainly 0-2 year-old juveniles; they rarely capture adults (Aldebert et al., 1993; Abella et al., 1997; Ardizzone and Corsi, 1997). The gillnet fishery exploits mainly adults of the species (>25 cm TL). Caught fish were kept on ice, subsequently frozen to prevent di- gestion of their stomach contents, taken to the laboratory, measured (total length: TL) to the nearest 1 mm, and weighed to the nearest 0.01 g. Sex and maturity stage were also recorded. Maturity state was determined by macroscopic analysis of the gonads by using the maturity scale for partial spawners (Holden and Raitt, 1974). Stomachs were removed and their contents weighed to the nearest 0.001 g. Prey items were identified and sorted into taxonomic groups to the species level whenever possible. When the state of digestion was more advanced, prey were checked and grouped into unidentified fish, cephalopods, or crustaceans. The de- gree of digestion of the prey was not considered in the analysis. Empty stomachs and those with partially everted or unidentified contents were excluded from the total sample. With the exception of the largest individuals (grouped into two het- erogeneous length classes), all re- maining hakes in the sample were grouped into 5-cm length classes. The study of size-related diet varia- tions was based on these groups. The contribution of each food item to the diet of these fish length groups was evaluated by using the index of rela- tive importance (IRI, Pinkas et al., 1971) as modified by Hacunda (1981): IRI= F{N + W). This index, expressed as IRI%=IRI - IIRI . 100, incorporates the percentage by num- ber (N%), wet weight (W7<), and fre- quency of occurrence (F% ) (Hyslop, 1980). Hierarchical cluster analysis and nonmetric multidimensional scal- ing (NMDS), based on Bray-Curtis similarity and on the IRI%, were used for classification and ordination of hake size classes (Clarke and War- wick, 1994). Manuscript submitted 27 April 2003 to the Scientific Editor's Office. Manuscript approved for publication 13 December 2004. Fish. Bull. 103:411-416(2005). 412 Fishery Bulletin 103(2) Results A total of 2761 hakes between 5 and 90 cm TL were collected (Table 1). The total number of prey was about 1700, divided into 46 different species. Cluster and NMDS analysis (stress = 0.02) based on the IRI allowed the identification of four groups below 50% similarity that were separated along a size gradient (Fig. 1). Euphausiids iNictiphan.es couchi, IRI=76%) and my- sids (Lophogaster typicus, IRI=22%) dominated the diet of group A (hake between 5 and 10.9 cm TL), and deca- pods were the secondary prey. £ 40 ■ M 80 100 VI III IV V IX VII VIII Hake size classes Stress: 0.02 Group A (<11 cm) II Group B (11 to15.9cm) Group C (16 to 35.9 cm) Figure 1 Dendrogram and NMDS (nonmetric multidimensional scaling) plot, based on IRI% values, of the nine hake (Merluccius merluc- eius) size classes using group-average clustering from Bray-Curtis similarity on diet data. (A) The four groups defined at arbitrary similarity level of 50% are indicated (dotted line); (B) NMDS showing the ordination of hake into four size classes with similar diets (the details of each size class are explained in the text). Group B (hake from 11 to 15.9 cm TL) showed a more heterogeneous diet characterized by a high occurrence of euphausiids but also with a considerable number of decapods (IRI=18%). Decapods were represented by a wide variety of species, such as Chlorotocus crassicor- nis, Alpheus glaber, Plesionika heterocarpus, Pasiphaea sivado, and Solenocera membranacea. Pisces and mysids showed lower percentages (IRI=15% and 4%, respec- tively). Sepiolidae (IRI = 0.9%), Sepietta oweniana and Alloteuthis media, dominated among cephalopods. The data suggest a gradual change towards a fully piscivorous diet (Fig. 2) which begins around 16 cm TL and is completed when sexual maturity is at- tained (TL = 32 cm for males and TL = 38.5 cm for females; Colloca et al., 2002). The importance of teleosts strongly increased in group C (hake from 16 to 35.9 cm TL), where they accounted for 91% of hake diet. The main prey were Clupeiformes (IRI=61 %), Sardina pilchardus and Engraulis encrasicolus. Fish (IRI = 96%) represented almost the entire diet of group D (>36 cm TL). In this group a shift towards Centracanthidae (Spicara flexuosa, Centracanthus cirrus) and a simultaneous de- cline in consumption of Clupeiformes was ob- served. Among decapods (IRI=4%), two species occurred most frequently: Processa spp. and S. membranacea. Euphausiids, mysids, and cepha- lopods were absent in the diet of hakes larger than 36 cm TL. Cannibalism of hake juveniles also accounted for some of the diet and increased with predator size. In hake between 36 and 40 cm TL cannibal- ism represented 12% of IRI, reaching the highest values (IRI = 17%) among larger individuals (TL >51 cm). Discussion Hake is a top predator that occupies different trophic levels during its ontogenetic develop- ment. Hake size classes are differentiated along food niche dimensions according to different prey sizes or different prey taxa. Hake diet shifted from euphausiids, consumed by the smaller hakes (<16 cm TL), to fishes consumed by larger hakes. Before the transition to the complete icthyophagous phase, hake showed more gener- alized feeding habits where decapods, benthic (Gobiidae, Callionymus spp., Arnoglossus spp.) and nectonic fish (S. pilchardus, E. encrasicolus) dominated the diet, and cephalopods had a lower incidence. Specific size-related differences in prey spectrum seem to be associated with dif- ferent spatial distributions or genetic needs (or with both) (Flamigni, 1984; Jukic and Arneri, 1984; Velasco and Olaso, 1998). The patterns observed in the present study indicated a strong partitioning among hake NOTE Carpentien et al.: Feeding habits of Merlucaus mer/uccius in the central Mediterranean Sea 413 Table 1 Number of hakes and values of IRI (index of relative importance) (% ) for the nine size classes. The four groups identified from the cluster analysis are indicated. Size group A B C D I II III IV V VI VII VIII IX Length (cm) 5.0-10.9 11.0-15.9 16.0-20.9 21.0-25.9 26.0-30.9 31.0-35.9 36.0-40.9 41.0-50.9 51.0-90.0 Number of hakes 202 430 564 454 555 224 139 107 75 Stomach contents 93 215 239 173 170 78 45 35 26 Prey Cephalopoda Alloteuthis media 0.22 0.02 0.01 Septet ta oweniana 0.02 0.02 0.01 Unid. Sepiolidae 0.35 0.30 0.03 Unid. Cephalopoda 0.42 0.10 0.01 0.02 Crustacea Alpheus glaber 0.02 0.33 0.05 0.22 0.05 0.81 1.54 Aristeidae 0.01 0.02 Aristeus antennatus 0.02 Chlorotocus crassieornis 1.61 1.83 1.09 1.10 0.48 Crangonidae 0.01 0.01 Pandalidae 0.03 0.01 Parapenaeus longirostris 0.01 Pasiphaea multidentata 0.02 0.01 Pasiphaea sivado 0.20 0.04 0.05 0.02 0.05 0.33 Plesionika heteroearpus 0.11 0.01 Plesionika sp. 0.62 0.07 0.01 0.04 0.05 Pontocaris lacazei 0.01 0.02 0.01 0.20 Pontophdus spinosus 0.01 0.01 0.03 0.05 0.20 Processa sp. 0.25 0.06 0.06 0.15 1.77 0.83 1.54 Solenoeera membranaca 0.04 0.02 0.05 0.34 0.58 3.27 3.53 Squilla sp. 0.05 Unid. Decapoda 3.05 19.91 6.19 2.84 2.73 1.45 1.58 1.32 Lophogaster typieus 28.77 4.34 0.16 0.01 Nictiphanes couchi 54.10 31.83 0.37 Unid. Euphasiacea 13.99 3.43 0.11 Unid. Isopoda 0.07 0.02 0.01 Pisces Argentina sphyraena 0.08 0.41 1.06 4.04 3.29 2.34 Arnoglossus laterna 0.01 0.01 Arnoglossus sp. 0.01 0.01 0.01 Callionymus sp. 0.01 0.01 0.01 0.06 Centracanthidae 0.03 0.11 2.60 2.43 11.23 53.97 Centracanthus cirrus 1.93 26.54 4.62 3.80 Clorophthalmus agassizi 0.01 Conger conger 0.34 0.85 Echiodon dentatus 0.05 Engraulis encrasicolus 1.95 11.61 1.28 4.45 9.91 0.87 1.27 1.86 Gadiculus argenteus 0.08 0.65 0.31 0.58 Gobiidae 0.04 0.02 0.01 0.01 0.05 Gobius quadrimaculatus 0.02 0.02 0.01 Lepidotrigla dieuzedei 0.01 0.01 0.78 Lesuerigobius friesii 0.01 0.02 0.03 Merluccius merluccius 0.07 0.18 12.00 4.10 17.95 continued 414 Fishery Bulletin 103(2) Table 1 (continued) Size group A B C D I II III IV V VI VII VIII IX Pisces (continued) Mullus barbatus 0.12 0.44 0.49 Myctophidae 0.30 0.28 0.03 0.15 Nettastoma melanurum 0.02 0.01 0.01 Sar'dina pilchardus 0.05 45.23 72.55 46.19 62.0 5.20 12.77 10.31 Sphyraena sphyraena 0.60 4.98 Spicara flexuosa 0.02 0.10 1.33 12.63 21.83 0.01 Spicara sp. 0.37 4.57 0.54 1.69 Trachurus trachurus 0.09 0.13 1.60 1.93 Trisopterus m. capelanus 0.02 0.01 0.01 0.05 Unid. Osteichthyes 0.04 34.19 33.14 21.61 43.44 15.01 23.09 22.90 4.25 Raja sp. 0.50 size classes. Two main thresholds associated with ontogenesis-related diet changes have been identi- fied. The first one was observed around 16 cm TL and corresponded to a significant change in depth distribution. The second, around 36 cm TL, cor- responded to the attainment of sexual maturity (Colloca et al., 2002). Although hakes are demersal fishes, they feed typically upon fast-moving pelagic prey that are ambushed in the water column (Alheit and Pitcher, 1995). There is evidence that hakes feed in mid-wa- ter or near the surface at night, undertaking daily vertical migrations (Hickling, 1927; Papacostanti- nou and Caragitsou, 1987; Orsi-Relini et al., 1989) which are more frequent for juveniles. Small hakes feed daily on small Euphausiacea (Nictiphanes cou- chi). This school-forming planktonic crustacean carries out vertical migrations at night (Casanova, 1970; Franqueville, 1971; Vallet and Dauvin, 2001). They rise to near the surface at night to feed on phytoplankton and sink during daylight between 50 and 800 m depth (Buchholz et al., 1995). Juveniles of M. merluccius may follow such migrations, moving from near the bottom, 100-200 m depth, to midwater at night (Froglia 1973; Papaconstantinou and Caragitsou, 1987; Orsi-Relini et al., 1989). Nocturnal vertical mi- gration behavior has been described for gadoids such as hake and cod and is considered responsible for the re- duction of trawl catches of these fish at night (Beamish, 1966; Bowman and Bowman, 1980). Considerable diet changes have been observed after the first year of life (>16 cm TL) when juveniles move from nursery areas on the shelf-break and upper slope to the middle shelf (Andaloro et al., 1985; Ardizzone and Corsi, 1997). The data indicate that such migration is induced by a change in trophic requirements. In this size class, diet changed to fish prey (Clupeiformes), and the importance of the small epiplanktonic crustaceans 100 -. *oo o° £ g 75- o°o o o o Porportion of fish in hake stomachs en o r, °° o ° o o ° 0 10 20 30 40 50 60 Hake length (cm) Figure 2 70 80 90 Proportion {%) of fish prey occurring in the diet of hake {Merluccius merluccius) during its growth. (Euphausiacea) strongly decreased. Clupeiforms S. pil- chardus and E. encrasicolus are distributed largely on the continental coastal shelf forming schools usually deeper than 25 m (Fisher et al., 1987). The size-depth distribution pattern of hake was con- firmed by experimental trawl surveys carried out in the Mediterranean (Relini and Piccinetti, 1996; Relini et al., 1999). Juveniles (modal length of 10 cm TL) are found mostly between 100 and 200 m depth. Intermedi- ate hakes reach the highest abundance mainly on the shelf (<100 m). Large hakes (>36 cm) are found in a wide depth range but concentrate on the shelf break during the spawning period (Recasens et al., 1998; Col- loca et al., 2000; Alvarez et al., 2001). Growth induces a continuous qualitative and quanti- tative change in diet that is reflected in the increasing NOTE Carpentieri et al.: Feeding habits of Merlucaus merluccius in the central Mediterranean Sea 415 mean weight of prey and decreasing mean number of prey items per stomach. The shift towards large fish prey (i.e., Centracanthidae) usually occurs slightly be- fore maturity — the life history stage with much higher energetic demands due to gonad development (Ross, 1978). A similar pattern was observed for Atlantic cod (Gadus morhua) where sexual maturation and spawn- ing are also associated with an ontogenetic change in diet (Paz et al., 1993). Thus, increased energy demands related to sexual requirements, gonad development, and breeding activity appear to be the critical factors driv- ing the changes in feeding strategy of M. merluccius. In large hakes (>36 cm), cannibalism played an important role and should be carefully considered in stock-recruitment analyses. Studies carried out in the Mediterranean (Macpherson, 1977; Bozzano et al., 1997) and in the Atlantic (Guichet, 1995; Link and Garrison, 2002) showed that cannibalism has some importance for hake. In silver hake (M. bilinearis), cannibalism notably increased with ontogeny (Link and Garrison, 2002). In the large cape hakes, M. capensis, hake is the dominant food item (50% of the diet) for individu- als larger than 60 cm (Roel and Macpherson, 1988). Conversely, a low cannibalism rate was observed for M. paradoxus in the same area (Payne et al., 1987). This could be a response to the greater accessibility of conspecifics compared to other species. As Payne et al. (1987) pointed out, small hake are not found in the vicinity of adults of the species. This is supported by the observed size segregation by depth, which is more pronounced in M. paradoxus than in M. capensis (Gor- doa and Duarte, 1991). Density-dependent cannibalism may be an important source of natural mortality that can stabilize fish populations (Smith and Reay, 1991), and for M. capensis, cannibalism has even been consid- ered the main cause of natural mortality (Lleonart et al., 1985; Payne and Punt, 1985). Our results on the trophic ecology of hake are of pri- mary importance for future management of fish assem- blages where this species plays an important predatory role. Multispecies management requires quantitative data on fish diet to elucidate the relationships between species and, consequently, to forecast temporal biomass fluctuations, under specific fishing regimes, in an inte- grated manner. Literature cited Abella, A., J. F. Caddy, and F. Serena. 1997. Do natural mortality and availability decline with age? An alternative yield paradigm for juvenile fisheries, illustrated by the hake Merluccius merluccius fishery in the Mediterranean. Aquat. Living. Resour. 10:1-14. Aldebert, Y., L. Recasens, and J. Lleonart. 1993. Analysis of gear interaction in a hake fishery: the case of the Gulf of Lions (NW Mediterranean). Sci. Mar. 57:207-217. Alheit, J., and T. J. Pitcher (eds.l. 1995. Hake: fisheries, ecology and markets, 478 p. Chap- man and Hall, London. Alvarez, P., L. Motos, A. Uriarte, and J. Egana. 2001. Spatial and temporal distribution of European hake. Merluccius merluccius (L.), eggs and larvae in relation to hydro-biological conditions in the Bay of Biscay. Fish. Res. 50:11-128. Andaloro, F., P. Arena, and S. Prestipino Giarritta. 1985. Contribution to the knowledge of the age, growth and feeding of hake Merluccius merluccius (L. 1758) in the Sicilian channel. FAO Fish Rep. 336:93-97. Ardizzone, G. D., and F. Corsi. 1997. Atlas of Italian demersal fishery resources. Biol. Mar. Medit. 4(2), 568 p. Arneri, E., and B. Morales-Nin. 2000. Aspects of the early life history of European hake from the central Adriatic. J. Fish. Biol. 53:1155- 1168. Beamish, F. W. H. 1966. Vertical migration of demersal fish in the northwest Atlantic. J. Fish. Res. Board Can. 23: 109-139. Bouaziz, A.. F. Djabali, and C. Maurin. 1990. Regime alimentaire du merlu {Merluccius merluc- cius, L., 1758) en baie de Bou Ismail. Rapp. Comm. int. Mer Medit. 32(1) :273. Bowman, R. E., and E. W. Bowman. 1980. Diurnal variation in the feeding intensity and catch- ability of the silver hake (Merluccius bilinearis). Can. J. Fish. Aquat. Sci. 37:1562-1572. Bozzano, A., L. Recasens, and P. Sartor. 1997. Diet of the European hake Merluccius merluccius (Pisces: Merlucciidae) in the Western Mediterranean (Gulf of Lions). Mar. Sci. 61(11:1-8. Buchholz, F, C. Buchholz, J. Reppin, and J. Fischer. 1995. Diel vertical migration of Meganyctiphanes nor- vegica in the Kattegat: comparison of net catches and measurements with acoustic Doppler current pro- filers. Helgol. Wiss. Meeresunters 49:849-866. Casanova, B. 1970. Repartition bathymetrique des euphausiaces dans le bassin occidental de la Mediterranee. Rev. Trav. Inst. Peches Marit. 34(2):205-219. Clarke, K. R., and R. M. Warwick. 1994. Change in marine communities: an approach to statistical analysis and interpretation, 144 p. Natural Environment Research Council, UK. Colloca, F., A. Belluscio, and G. D. Ardizzone. 2000. Sforzo di pesca, cattura e gestione dello stock di nasello (Meluccius merluccius) in un'area del tirreno centrale. Biol. Mar. Medit. 7(1):117-129. Colloca, F., P. Gentiloni, A. Belluscio, P. Carpentieri, and G. D. Ardizzone. 2002. Estimating growth parameters of the European hake [Merluccius merluccius) through the analysis and validation of annual increments in otoliths. Arch. Fish. Mar. Res. 50 (21:174-179. Fisher, W., W. M. Bauchot, and M. Schneider. 1987. Fiches FAO d'identification pour les besoins de la peche revision 1. Mediterranee et mer Noire. Zone de peche 37, vol. 2: Vertebres, p. 761-1530. FAO, Rome. Flamigni, C. 1984. Preliminary utilization of trawl survey data for hake {Merluccius merluccius. L.) population dynamics in the Adriatic Sea. FAO Fish. Rep. 290:109-115. Franqueville, C. 1971. Macroplancton profond (invertebres) de la Mediter- ranee nord-occidentale. Tethys 3(l):ll-56. 416 Fishery Bulletin 103(2) Froglia, C. 1973. Osservazioni sull'alimentazione del merluzzo (Mer- luccius merluccius L.l del Medio Adriatico. In Atti V Congresso Naz. Soc. It. Biol. Mar. (N. Salentina, ed.i, p. 327-341. Gordoa, A., and C. M. Duarte. 1991. Internal school density of Cape hake (Merluccius capensis). Can. J. Fish, aquat. Sci. 48:2095-2099. Guichet, R. 1995. The diet of European hake (Merluccius merluc- cius) in the northern part of the Bay of Biscay. ICES J. Mar. Sci. 52:21-31. Hacunda. J. S. 1981. Trophic relationships among demersal fishes in a coastal area of the Gulf of Maine. Fish. Bull. 79: 775-788. Hickling. C. F. 1927. The natural history of the hake. Parts I and II. Fish. Invest. Serv. II, 10(2), 112 p. Holden, M. J„ and D. F. S. Raitt. 1974. Manual of fisheries science. Part 2: Methods of resources investigation and their application. FAO Fish. Tech. Paper 115, 214 p. FAO, Rome. Hyslop, E. J. 1980. Stomach content analysis: a review of methods and their application. J. Fish Biol. 17:411-429. Jukic, S., and E. Arneri. 1984. Distribution of hake (Merluccius merluccius, L.) striped mullet (Mullus barbatus, L.) and pandora (Pagel- his erythrinus, L.) in the Adriatic Sea. FAO Fish. Rep. 290:85-91. Lleonart, J., J. Salat, and E. Macpherson. 1985. CVPA, an expanded VPA with cannibalism. Appli- cation to a hake population. Fish. Res. 12:119-146. Link, J. S., and L. P. Garrison. 2002. Changes in piscivory associated with fishing induced changes to the finfish community on Georges Bank. Fish. Res. 55:71-86. Macpherson, E. 1977. Estudio sobre relaciones troficas en peces bentonicos de la costa catalana. Ph.D. diss., 369 p. Universitat de Barcelona, Barcelona. Oliver, P., and E. Massuti. 1995. Biology and fisheries of western Mediterranean hake (M. merluccius). In Fish and fisheries series 15, hake, fisheries, ecology and markets (J. Alheit and T. J. Pitcher), p. 181-202. Chapman and Hall, London. Orsi Relini, L., F. Fiorentino, and A. Zamboni. 1989. Spatial-temporal distribution and growth of Merluc- cius merluccius recruits in the Ligurian Sea. Observa- tions on the O group. Cybium (131:263-270. Papaconstantinou, C, and E. Caragitsou. 1987. The food of hake (Merluccius merluccius) in Greek Seas. Vie Milieu 37:77-83. Payne. A. I. L„ B. Rose, and R. W. Leslie. 1987. Feeding of hake and a first attempt at determin- ing their trophic role in the South African west coast marine environment, in The Benguela and comparable ecosystem (A. I. L. Payne, J. A. Gulland and K. H. Brink, eds.l. S. Afr. J. Mar. Sci. 5:471-501. Payne, A. I. L., and A. E. Punt. 1985. Biology and fisheries of South African cape hakes (M. capensis and M. paradoxus). Hake — fisheries, ecol- ogy and markets. Chapman and Hall, London. Paz, J., J. M. Casas, and G Perez-Gandaras. 1993. The feeding of cod (Gadus morhua) on Flemish Cap. 1989-90. NAFO Scientific Council Studies 19, p. 41-50. Pinkas, L., M. S. Oliphant, and I. L. K. Iverson. 1971. Food habits of albacore, bluefin tuna and bonito in California waters. Calif. Dep. Fish Game Fish. Bull. 152:1-150. Recasens, L., A. Lombarte. B. Morales-Nin, and G. J. Torres. 1998. Spatiotemporal variation in the population struc- ture of the European hake in the NW Mediterranean. J. Fish. Biol., 53:387-401. Relini, G., and C. Piccinetti. 1996. Ten years of trawl surveys in Italian seas (1985- 1995). FAO Fish. Rep. 533:21-41. Relini, G, J. Bertrand, and A. Zamboni. 1999. Synthesis of the knowledge on bottom fish- ery resources in central Mediterranean (Italy and Corsica). Biol. Mar. Medit. 6. 868 p. Roel, B., and E. Macpherson. 1988. Feeding of Merluccius capensis and M. paradoxus off Namibia. S. Afr. J. Mar. Sci. 6:227-643. Ross, S. T. 1978. Trophic ontogeny of the leopard searobins. Prionotus scitulus (Pisces: Triglidae). Fish. Bull. 76:225-234. Sarano, F. 1986. Cycle ovarien du Merlu, Merluccius merluccius, poisson a ponte fractionnee. 2ev. Trav. Inst. Peches Marit. 48:65-67. Smith, C, and P. Reay. 1991. Cannibalism in teleost fish. Rev. Fish Biol. Fish 1:41-64. Vallet, C, and J. C. Dauvin. 2001. Biomass changes and bentho-pelagic transfers throughout the Benthic Boundary Layer in the English Channel. J. Plankton Research, vol. 23(9):903-922. Velasco, F., and I. Olaso. 1998. European hake Merluccius merluccius (L., 1758) feeding in the Cantabrian Sea: seasonal, bathymetric and length variations. Fish. Res. 38:33-44. 417 Biology of queen snapper (Etelis oculatus: Lutjanidae) in the Caribbean Bertrand Gobert Institut de Recherche pour le Developpement (IRD) Technopole Brest-lroise BP 70 29280 Plouzane, France E-mail address: gobertra) ird.fr Alain Guillou Institut Francais de Recherche pour I'Exploitation de la Mer (Ifremer) Boulevard Jean Monnet BP 171 34203 Sete Cedex, France Peter Murray Organization of Eastern Caribbean States (OECS) Environment and Sustainable Development Unit The Morne POBox 1383 Castries, Saint Lucia Patrick Berthou Institut Francais de Recherche pour I'Exploitation de la Mer (Ifremer) BP 70 29280 Plouzane, France Maria D. Oqueli Turcios 38, rue Desaix 75015 Paris, France Ester Lopez Departement Halieutique Ecole Nationale Supeneure Agronomique de Rennes 65, rue de Saint-Bneuc CS 84215 35042 Rennes Cedex, France Pascal Lorance Jerdme Huet Institut Francais de Recherche pour I'Exploitation de la Mer (Ifremer) BP 70 29280 Plouzane, France Nicolas Diaz Boyer 97129 Lamentin Guadeloupe, French West Indies Paul Gervain Rue Authe 2 Petit Pans 97100 Basse Terre Guadeloupe, French West Indies tation of the queen snapper is poorly documented, and very few detailed catch statistics are available; in all cases, the amounts landed in each country are small (probably not ex- ceeding a few tens of tons per year), but the potential production of these resources has never been estimated. Owing to the depth of its habitat and to the relatively small economic importance of the fisheries for queen snapper on the local scale, very little is known about the biology of E. ocu- latus. It is generally cited in species checklists or in general descriptions of deepwater fisheries. Very few stud- ies actually have focused on the spe- cies itself (Murray, 1989; Murray and Charles, 1991; Murray et al., 1992; Murray and Moore, 1992; Murray and Neilson, 2002). The objective of this study is to present new information about the biology of E. oculatus, obtained from fishing experiments undertaken since the 1980s in the French West Indies (Martinique, Guadeloupe, Saint- Barthelemy, and the French part of Saint-Martin), Dominica and Saint- Lucia, and from a study conducted in the late 1990s on the artisanal and semi-industrial fisheries off the Caribbean coast of Honduras. Material and methods Areas studied The data were collected from various research projects (Fig. 1 and Table 1): The queen snapper (Etelis oculatus) is among the deepest dwelling spe- cies of the family Lutjanidae, and the only Atlantic species of Etelis. Its dis- tribution covers the tropical western Atlantic Ocean, from North Carolina to the eastern tip of Brazil, at depths of 130 to 450 m (Allen, 1985). Although it reaches a large size and presents no risk of ciguatoxicity (Lorance1), the species is exploited by only a few fisheries in the Caribbean. Most often it is only a minor part of the catch of line fisheries that focus on the whole community of deep snap- pers, or on more abundant species such as vermilion snapper (Rhom- boplites aurorubens) or silk snapper (Lutjanus vivanus) (e.g., in Venezu- ela: Mendoza and Larez, 1996). In a few cases, however, E. oculatus is specifically sought by fishermen; for example, in Saint-Lucia within a tra- ditional fishery operating during the months when migratory pelagics are not fished (Murray et al, 1992), or in Bermuda where it has been caught irregularly (pulse fishery) since the ban on potfishing (Luckhurst, 1996). Commercial exploitation is only be- ginning in the French West Indies, but is much more developed in Barba- dos (Prescod et al., 1996) and Puerto Rico (Matos-Caraballo, 2000). Exploi- 1 Lorance, P. 1988. La ciguatoxicite des poissons sur les bancs de Saint-Barthe- lemy. Saint-Martin et Anguilla. Doc. Sci. Pole Caraibe 15, 31 p. [Available from Ifremer, Pointe Fort, 97231 Le Robert, France.] Manuscript submitted 16 September 2003 to the Scientific Editor's Office. Manuscript approved for publication 20 October 2004 by the Scientific Editor. Fish. Bull. 103:417-425 (20051. 418 Fishery Bulletin 103(2) Figure 1 Study area and locations sampled for queen snapper {Etelis oeulatus) by Caribbean fisheries 1982-2001. 1) On the French parts of the wide shelf shared by St- Martin, St Barthemely, and Anguilla (abbreviated as SMSBA shelf in the text), exploratory fishing experiments were conducted to assess the fishing potential and the risk of ciguatoxicity (Lorance2). The deep slopes of the bank (200-300 m) were fished in 1986-87, using bottom longlines, trammel nets, and secondarily bottom gill nets. 2) In Martinique, exploratory fishing experiments were conducted in 1986-87 on various parts of the shelf slope (100-300 m), and some observations were made in 1982 and 1988-91, mainly with gill nets and trammel nets (Guillou3). 3) In Saint-Lucia, observations were made in 1987 on the commercial fishery, and fishing experiments were conducted in 1992 with longlines (Guillou4). 4) In Dominica, fishing experiments were conducted in 1992 with longlines and gill nets. 5) In Guadeloupe, experiments were conducted in 2001 with gill nets in the range 200-400 m (Diaz et al., in press); some small Etelis were also caught with 10-mm-mesh traps used for a survey of deep crus- tacean resources. 6 1 In the Bay Islands, off the Caribbean coast of Hondu- ras, a fisheries survey was conducted in 1999-2000 as part of a coastal zone management project (Ber- thou et al.5). This artisanal fishery uses mainly han- dlines to catch snappers and groupers on the shelf, but a fraction of the fishing effort is directed towards the deepwater snappers on the shelf slopes. 7) In Honduras, the landings of the semi-industrial fishing fleet based in Roatan (Bay Islands) were studied, through catch statistics of the export firms and by sampling in the collecting centers (de Rodel- lec6). These fleets target snappers and groupers over the entire Caribbean shelf of Honduras, and fish with handlines. 2 Lorance, P. 1989. Ressources demersales et descriptions des pecheries des bancs de St-Martin et St Barthelemy. Rapp. Int. Dir. Ressources Vivantes Ifremer, DRV-89.039-RH/Mar- tinique, 75 p. [Available from Ifremer, Pointe Fort, 97231 Le Robert, France.] 3 Guillou, A. 1989. Ressources demersales du talus insu- laire de la Martinique. Rapp. int. Dir. Ressources Vivantes Ifremer DRV-89.037-RH/Martinique, 121 p. [Available from Ifremer, Pointe Fort, 97231 Le Robert, France.] 4 Guillou A., A. Lagin, and P. Murray. 1996. Observations realisees sur la biologie et la peche du «gros yeux« Etelis oeulatus Val. aux Petites Antilles de 1982 a 1992. Doc. Sci. Pole Caraibe 33, 137 p. [Available from Ifremer, Pointe Fort, 97231 Le Robert, France.] 5 Berthou P., M. D. Oqueli, E. Lopez, B. Gobert, C. Macabiau, and P. Lespagnol. 2001. Diagnostico de la pesca artesanal de la Islas de la Bahia, Honduras. Proyecto Manejo Ambi- ental de las Islas de la Bahia (PMAIB), Informe Tecnico PES-06, vol 1, 194 p. [Available from PMAIB, Roatan, Islas de la Bahia, Honduras.] 6 de Rodellec, A. 2001. Les debarquements de poissons destines a l'exportation dans l'ile d'Utila (lies de la Bahia, Honduras). Unpubl. report, IRD-Brest, 51 p. [Available from IRD, BP 70, 29280 Plouzane, France.] NOTE Gobert et al.: Biology of Etelis oculatus in the Caribbean 419 Table 1 Summary of sample sizes and depth ranges of queen snapper {Etelis oculatus) by area mercial fishing operations ( SMSBA=Saint Martin-Saint Barthelemy-Anguilla). and fishing gear, in exploratory or corn- Area Trammel nets (exploratory) Gill nets (exploratory Lines (exploratory) Lines (commercial) Depth range (ml Martinique 300 209 6 140-300 SMSBA shelf 249 406 230-430 Saint-Lucia 34 394 210-290 Dominica 191 20 180-300 Guadeloupe Bay Islands Honduran shelf 1133 794 3948 195-410 unknown unknown Fishing gears used In all islands but Guadeloupe, gill nets had mesh sizes of 65 mm (knot to knot) and a stretched height of 6.4 m. In Guadeloupe, mesh was 60 mm and height was 4 m; in addition, the nets were given more slack than in Martinique to increase their efficiency, and thus caught a wider size range offish. Trammel nets had mesh sizes of 40 mm (knot to knot) on the central panel and 200 mm on the outer panels, and were 2 m high. All nets (trammel nets and gill nets) were set overnight (15 to 20 hours of fishing time) in units of 200 or 300 m. Three types of longlines were used in the fishing ex- periments. Vertical longlines were derived from those used by fishermen in the Lesser Antilles and had about 20 hooks on 40 cm-long secondary lines. Pole longlines were adapted from a technique used in Florida and Puerto-Rico: poles about 2 m long are fastened to the main line lying on the bottom, each having 12 to 25 hooks on very short secondary lines (20 to 30 cm). Re- inforced longlines are horizontal longlines whose main line is heavier, in order to fish on very rough grounds. All longlines were hauled after 30 to 45 minutes fish- ing. For the analysis, no distinction was made between samples of these three types of longlines. Various kinds of longlines are used in the small-scale queen snap- per fishery in Saint-Lucia. Handlines used in the ar- tisanal and semi-industrial Hondurian fisheries are either mono- or multifilament, and bear one or several hooks. No detailed observations were made on the size of hooks or on the bait used in the commercial queen snapper fisheries. 1987) could be done reliably in the field for adults and juveniles, but had to be confirmed under the microscope for the smallest individuals (less than 10 cm). Fish length was the only information recordable from profes- sional landings (St-Lucia, Honduras); fishing experi- ments yielded more detailed data, by order of decreasing frequency: length (fork length FL, total length TL, or both; unless specified, all lengths mentioned in the text are fork lengths), weight, and sexual stage, and occasionally a few additional observations (such as unusual number or length of fin rays). Sexual stages were identified by using the macroscopic scale defined by Barnabe (1973) and were coded as follows: 1 (imma- ture, without identifiable sex), 2 (immature, of identifi- able sex), 3 (mature), 4 (prespawning), 5 (spawning), 6 (postspawning), and 7 (resting). Depth was recorded only in the fishing experiments; for gillnet and tram- mel-net stations, it was measured at each end of the net, and the depth used in the analysis was the average of these two values. Length-frequency analysis In most cases, length-frequency analysis was strongly hindered by gear selectivity and sample sizes. We attempted to estimate L, and ZIK with the method of Wetherall et al. (1987) applied to the sample of the semi-industrial Hondurian fishery. All other samples were unsuitable for length-frequency analysis because of severe violations of one or several assumptions, prin- cipally regarding constant catchability above the full selection length, which was obviously not the case for the three gears used in the fishing experiments. Data collected None of these studies was specifically designed for the study of E. oculatus, and therefore the nature and amount of available information (sampling coverage though time, space, and depth) for this species were variable. Species identification (Allen, 1985; Anderson, Results Depth distribution During the fishing experiments, E. oculatus of market- able size (i.e., larger than about 20 cm) were caught 420 Fishery Bulletin 103(2) between 140 and 430 m. In Martinique, the trammel nets were set between 100 and 300 m but did not catch any E. oculatus in the shal- lowest part of this range. In Guadeloupe, gill nets were set down to 410 m, but the deepest catch of queen snapper was 340 m. No E. oculatus were caught in shallower (<80 m) fishing experiments with any of the gears used (traps, gill nets, trammel nets, and long- lines) on the SMSBA shelf. According to some local fishermen, however, queen snappers can be caught from about 100 m down to 550 m (Lorance2). Depth-size relationship i u c /u ■ 60 - 50 - A D • A A • •...■•■ ■A ..a. * *■ ■ A A A o CO 40 - 30 - A ■ ■ 6 ■ - A > < ?0 - 100 150 200 250 300 Depth (m) 350 400 450 No clear relationship between depth and aver- age size offish was found in the fishing experi- ments (Fig. 2). This is not unexpected given the selectivity of some gears (gill nets) and the small sample sizes in most depth strata outside the main fishing range (250-300 m); 70% of the 456 fish caught by longlines were in the 290 m depth stratum, and five or fewer fish were caught in most of the other strata. A different picture emerges from the analysis of the professional fisheries of Honduras. Multivariate analy- sis (principal component analysis followed by hierar- chical classification) applied to the landings by species revealed the two different categories of fish caught by the two types of semi-industrial vessels operating from Roatan (de Rodellec6), the shelf-operating fleet and the slope-operating fleet. The first category of fish were dominated by shallow species such as Ocyurus chrys- urus (59.8%), Lutjanus analis (7.8%), and several grunts (Haemulidae), whereas E. oculatus accounted for only 2.2%. On the other hand, the second category comprised mainly deep snappers: L. vivanus (39.6%), E. oculatus (22.4%), R. aurorubens (6.9%) or L. buccanella (1.9%). The two divisions of the fleet independently exploit the continental shelf and the deep slope. Although actual depth of fishing operations is unknown, the shelf-oper- ating fleet probably catches E. oculatus in the deepest part of its working area (i.e., at the shallowest part of the species' bathy metric range), whereas the slope-op- erating fleet exploits the main habitat of the deepwater snappers. The size structures of Etelis catches (Fig. 3, A and B) strongly indicate that only the fish up to 45-50 cm live on the shelf or its edge, whereas individuals of all sizes, and particularly the largest ones, inhabit the shelf slope. A similar observation was made for the island of Roatan, where the artisanal fleet is the least developed of the archipelago: fishermen using small (<6 m) and often (57%) nonpowered canoes fish quite close to the shore and catch a large diversity of coastal reef fishes, a large proportion of which are juveniles. Etelis oculatus is rarely caught by these small-scale fishermen but is so only as individuals smaller than 50 cm, sometimes as small as 16 cm (Fig. 3C). Figure 2 Average fork length of queen snapper {Etelis oculatus) by depth (m) strata in the fishing experiments with gill nets (circles), trammel nets (squares), and lines (triangles). Sample sizes are indicated by size of the symbols: empty symbol (rc<10), filled symbol by increasing size (/i = ll-20, 21-50, 51-100. >100). Habitat of early juveniles Some observations were made on very small (smaller than 10 cm) individuals of E. oculatus. Off Guadeloupe, a few of them were entangled in gill nets at 300 m depth (Fig. 3D); on the same island, previous exploratory fish- ing operations with small-mesh traps caught six juve- niles ranging from 5.5 to 7 cm FL at 490 m depth; off Dominica, one small individual (8.5 cm TL) was found in the stomach of a predator caught at a depth greater than 200 m (see below). In spite of the general tendency of increasing size with depth found for the larger indi- viduals, these observations show that the habitat of early postsettlement juveniles is not restricted to the shallowest part of the species depth range. Morphometric relationships The main morphometric relations were computed from the fish sampled in commercial or scientific fishing operations in the Lesser Antilles (Martinique, Saint- Lucia, SMSBA shelf). Because the differences between relations for males and females were insignificant, only global equations are given (Table 2). Maximum size and weight The largest individual caught was 90 cm FL in the Lesser Antilles (Guadeloupe) and 86 cm in Honduras, and the maximum weight recorded was 6280 g, in the Lesser Antilles; fish were not weighed individually in Honduras. Sex-related length differences When sex was recorded, the largest fish were always female, and no male was found above 70 cm. The differ- ence between size-structure of male and female catches NOTE Gobert et al.: Biology of Etelis oculatus in the Caribbean 421 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 6 i 5 4 - 3 2 - 1 0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 FL (cm) Figure 3 Length-frequency distributions for queen snapper {.Etelis oculatus) catches. (A) Semi-industrial deepwater Hon- durian line fishery (rc=3415). (B) Semi-industrial shal- low-water Hondurian line fishery (n=387). (C) Artisanal line fishery of Roatan (Honduras) (ra=52). (D) Gillnet exploratory fishing in Guadeloupe (rc=779). (E) Trammel- net exploratory fishing in all areas: males (h=231). (F) Trammel-net exploratory fishing in all areas: females (n=227). 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 FL (cm) Figure 3 (continued) was particulary clear for trammel nets (Fig. 3, E and F): the mode corresponding to fish gilled in the small- mesh central panel had similar characteristics for both sexes (range 25-45 cm and peak at 36 cm), as opposed to the diffuse mode for fish >45 or 50 cm (predominantly females) entangled in the large-mesh outer panels. With many fewer fish («=23 for both sexes), the longline sam- ples showed a similar difference between sizes of males (maximum 55 cm, mean 43.8 cm) and females (maximum 71 cm, mean 51.8 cm). In Guadeloupe, the sex offish was not determined, but the existence of two modes in the overall size structure of gillnet catches (Fig. 3D) could possibly be related to this sex-related length difference. Growth and mortality The data collected in the various surveys did not allow any reliable analysis of the growth of E. ocula- tus. Because growth may be different for males and females, the length-frequency distributions of the large samples (where sex was not determined) from Honduras could not be processed rigorously to estimate life-history population parameters. However, in order to provide preliminary information on such a little known species, the regression method of Wetherall et al. (1987) was used in the modified version of FiSAT (Gayanilo et al., 1996) to estimate Lx and Z/K. With a satisfactory fit of the regression line (r=0.986), the estimates were L, = 90.57 cm and Z/K = 3.73. For the reason mentioned above (together with other weaknesses related to possible violations of the hypotheses underlying the regression 422 Fishery Bulletin 103(2) Table 2 Morph jmetric relationships established for queen snapper tEtelis oeulatus). Parameters Equation Sample size r FL(cm)-TL(cm) TL = 2.7458 + 1.1644 x FL 842 0.987 TL (cm) - FL (cm) FL = -1.0028 + 0.8368 x TL 842 0.987 FL(cm)-Wlg) W = 0.02748 x FL28348 499 0.990 TL(cm)-W(g) W = 0.03006 x FL2 a"? 487 0.990 method), these estimates have to be seen only as indica- tions that the asymptotic length of E. oeulatus is quite large and that the Hondurian population is moderately exploited (if M = 2K, as suggested by Ralston (1987) for snappers, then ZIK = 3.73 and E = F/Z = 0.46). Reproduction Macroscopic observations of gonads were recorded for 309 fish whose sex could be identified (118 males and 191 females); all stages of the reproductive cycle were observed, but only 20 individuals were in the prespawn- ing to postspawning stages, and a single one was found to be in the process of spawning. The smallest fish with developing gonads was 36 cm for females and 29 cm for males (Fig. 4). Although only part of the length range of males was adequately sampled (100 out of the 118 fish were smaller than 44 cm I, it appears that the progressive build-up of the reproductive male population occurs between about 30 cm and 45 cm. The picture is clearer for females, whose sample size was larger and more evenly spread over the length range: above 54 cm, all females were found to be in a reproductive cycle. The maturing process therefore occurs at clearly lower sizes for males (30-45 cm) than for females (35-55 cm). Females in advanced reproductive stages (postspawning and resting stages) were observed across the length range, including the smallest adult sizes, those below 45 cm (Fig. 4). A full analysis of the seasonality of reproduction is not possible because data were collected in only seven months, out of which only four (May, June, November, and December) yielded samples large enough for the analysis (21 to 72 females per month). No females were found to be spawning, but most of the pre- and post- spawning stages (14 out of 17) were observed in Novem- ber and December, and half of maturing females were fished during the last quarter of the year (Fig. 5). How- ever, 74% of females at sexual rest (resting stage) were caught in May and June. Additional pieces of informa- tion confirm this pattern: the only spawning individual, a male, was observed in November (Dominica); females gonads in advanced stage of vitellogenesis were observed in September (Guadeloupe); no mature individual was found in Honduras in April-June. These observations show that an active spawning period occurs at the end of the year (even if all fish caught at this period were not close to the spawning phase), as opposed to late spring which is a period of sexual inactivity. Such limited results leave open the overall interpre- tation of the annual reproductive cycle of E. oeulatus. In particular, according to the fishermen working on the SMSBA shelf, the species could have an extended spawning season, lasting from November to April or May. Predators and prey No systematic observations were made on the trophic relationships of E. oeulatus, but a few occasional record- ings were made of its predators and prey. The only record of a predator was that of a beardfish {Polymixia lowei: Polymixiidae) measuring 40 cm TL containing a very small queen snapper (8.5 cm TL) and which was caught deeper than 200 m. This is the first record of such a food item for this beryciform fish whose diet had so far been reported to comprise cephalopods (Cervigon, 1991). The stomachs of E. oeulatus that could be observed were most often empty; on a few occasions, unidentified squids were the only items present. This was the case for three fish (58 to 62 cm) caught at 430 m depth. Discussion Etelis oeulatus was found on the upper part of the con- tinental and insular slopes, from about 150 to 450 m; this observed range confirms previous indications (Allen, 1985), but the bathymetric distribution of the species could possibly extend beyond the maximum depth fished in these surveys. The presence of E. oeulatus in shal- lower waters of the shelf seems possible, according to a statement that juveniles can be found in less than 30 m (Appeldoorn et al., 1987) and to the reported catch of one fish (size not recorded) at 59 m depth by a trawl survey off southeastern United States (Cuellar et al., 1996). However the present data, other fishery-independent surveys focusing on snappers (i.e., Marcano et al., 1996, down to 128 m), and most studies on Caribbean coastal fisheries strongly indicate that the species is very rare on the shelf itself. Within the observed depth range, there is a ten- dency for the largest fish to be found in the deeper areas, as observed in the closely related Pacific species NOTE Gobert et al.: Biology of Etelis oculatus in the Caribbean 423 E. carbunculus and E. coruscans (Brouard and Grandperrin"), other deepwater lutjanids (Board- man and Weiler, 1980; Cuellar et al., 1996), and many reef fish species. The maximum size recorded for large samples (90 cm FL) is much greater than the 60 cm TL indicated by Allen (1985) but is con- sistent with other field observations, such as 94 cm TL in Saint-Lucia (Murray, 1989) or 100 cm TL in Venezuela (Cervigon, 1991). No reliable growth estimate could be obtained be- cause males and females showed very different size structures, and the only data suitable for length-fre- quency analysis were data for which the sexes had not been determined. Important differences were found between sexes in terms of size structure and maturation size. Male E. oculatus attain a smaller length than females, and are much rarer above 45 cm. Sex-ratios skewed in favor of females in large size classes were observed in the most complete studies of snapper populations (Grimes, 1987), including Pacific deepwater snap- pers (Brouard and Grandperrin7), and probably re- sult from a difference in growth and mortality be- tween the sexes; in Cuba, for instance, females of most snapper species have been found to grow faster than males (Claro and Garcia-Arteaga, 2001). In the present study, sex-specific growth and mortality estimates were not available, but our interpretation seems likely because other possible causes could be ruled out, such as selectivity of nets (morphometric relationships are identical for both sexes) and fish be- havior in relation to fishing gear (differences between sexes, however, were observed in trammel nets and lines whose catch mechanism is completely different). Different habitat preferences, which can lead to sex- related size structures in reef species (Garcia-Cagide et al., 2001), seem unlikely in our study because the deep slopes have fewer habitat gradients than the shallower reef environments and because no relation was found between depth and sex-ratio. A similar difference between males and females was found for reproductive size. Male snappers generally mature at a slightly smaller size than females, but sex does not appear as a significant factor of variation for relative length at first reproduction, as opposed to depth or continental or insular habitat (Grimes, 1987). In the Lesser Antilles, E. oculatus spawns at the end of the year and has a period of sexual rest during from late spring through early summer. These results are not sufficient to establish the entire annual reproduc- tive pattern, and even these partial findings cannot be applied to other parts of the Caribbean because snap- per populations of continental and insular shelves gen- erally show different seasonal patterns of reproduction (Grimes, 1987). This indication of a spawning period for E. oculatus in the cold season contrasts with the two eteline species (Aprion virescens and E. coruscans) stud- ied in Hawaii, which have a protracted spawning period extending through the summer (May or June through October or November) (Everson et al., 1989). 26 30 34 38 42 46 50 54 58 62 66 70 74 100% 80% 60° 40% 20% 0% B ~3i 26 30 34 38 42 46 50 54 58 62 66 70 74 FL (cm) Figure 4 Proportion of sexual stages by 2-cm length classes for (A) female (n=191) and (B) male (/; = 118 ) queen snapper iEtelis oculatus): immature fish (gray), maturing fish (large squares), prespawning (horizontal bars), spawn- ing (oblique bars), postspawning (vertical bars), sexual rest (black). Empty areas indicate the absence of data for the length class. □ April sexual rest postspawning spawning prespawning maturation immature as HMay 111 sss mil sssssss: DJune B August ■ September □ October U November U December iim ( Distribution o latus) by mont ) 20 40 60 80 100 % Figure 5 "sexual stages of queen snapper (Etelis ocu- h. Brouard, F., and R. Grandperrin. 1985. Les poissons pro- fonds de la pente recifale externe a Vanuatu. South Pacific Commission, r7<"me conference technique regionale des peches, Noumea (Nouvelle Caledonie) 5-9 August 1985. SPC/Fish- eries 17/WP.12 , 131 p. [Available from SPC, BP D5, 98848 Noumea Cedex, New-Caledonia, France.] 424 Fishery Bulletin 103(2) The data collected in these studies did not allow the analysis of the aggregation pattern of E. oculatus. In the Pacific, E. coruscans was found to form feeding ag- gregations near underwater promontories and these ag- gregations had important consequences for catchability (Ralston et al., 1986). For the deeper living alfonsinos (Beryx spp.) and orange roughy (Hoplostethus atlanti- cus), fisheries have shown their ability to quickly fish down aggregations once they are discovered (Lorance and Dupouy, 2001). Added to "K-selected" life-history strategies (high longevity, slow growth, late reproduc- tion) and irregular recruitment, this aggregating behav- ior reinforces the vulnerability of deepwater species to overfishing (Koslow et al., 2000). Recently gained knowledge about the exploitation of seamount and deep bank fish resources (Clark, 2001) cannot be applied directly to E. oculatus and the other slope-dwelling snappers which, although they are the deepest dwelling species of the family, are much closer in terms of demographic strategy to their shallow rela- tives (longevity 10-20 years; Manooch, 1987) than to these truly deep species (longevity 50 to more than 100 years; Koslow et al., 2000). However, less extreme life history traits do not protect deep snappers against over- fishing, as shown by the example of E. coruscans and E. carbunculus in Hawaii (Simonds, 1995). The limited fishery data available on E. oculatus in the Caribbean do not seem to show evidence of a similar situation so far, but the stocks are being increasingly fished without much scientific basis (i.e., catch statistics) for manage- ment (Mahon, 1990; FAO, 1993). Regulation measures continue to be defined (Diaz et al., in press), but so far they are based only on conservative rules of thumb because of a lack of reliable biological information. To address this lack of information, future research on E. oculatus therefore should address, in particular, sex- specific growth, reproductive biology, and fine-scale distribution patterns. Acknowledgments The data presented here were collected and processed with the help of many people; the authors particularly wish to thank E. Burgos, T. and J. Chapelle, R Galera, J. Grelot, A. Lagin, P. Lespagnol, L. Reynal, J. Robin, B. Seret, and the Chief Fisheries Officers of Dominica and Saint-Lucia. Literature cited Allen, G. R. 1985. FAO species catalogue. Vol. 6. Snappers of the world. An annotated and illustrated catalogue of lut- janid species known to date. FAO Fish. Synop. 125, vol. 6:1-208. Anderson, W. D. 1987. Systematics of fishes of the family Lutjanidae (Per- ciformes: Percoidei), the snappers. In Tropical snappers and groupers, biology and fisheries management (J. J. Polovina and S. R. Ralston, eds), p. 1-31. Westview Press, Boulder, CO. Appeldoorn, R., G. D. Dennis, and O. Monterrosa Lopez. 1987. Review of shared demersal resources of Puerto- Rico and the Lesser Antilles region. FAO Fish. Rep. 383:36-106. Barnabe, G. 1973. Contribution a la connaissance de la croissance et de la sexualite du Loup (Dieentrarehus labrax L.) de la region de Sete. Ann. Inst. Oceanogr. Paris Inouv. ser.) 49:49-75. Boardman C, and D. Weiler. 1980. Aspects of the life history of three deepwater snap- pers around Puerto Rico. Proc. Gulf Carib. Fish. Inst. 32:158-172. Cervigon, F. 1991. Los peces marinos de Venezuela, 951 p. Fundacion Cientifica Los Roques, Caracas, Venezuela. Claro, R., and J. P. Garcia-Arteaga 2001. Growth patterns of fishes of the Cuban shelf. /;; Ecology of the marine fishes of Cuba (R. Claro, K. C. Lindeman, and L. R. Parenti, eds.), p. 149-178. Smith- sonian Institution Press. Washington and London. Clark, M. 2001. Are deepwater fisheries sustainable ? The exam- ple of orange roughy (Hoplostethus atlanticus) in New Zealand. Fish. Res. 51:123-135. Cuellar, N., G. R. Sedberry, D. J. Machowski, and M. R. Collins. 1996. Species composition, distribution and trends in abundance of snappers of the southeastern USA, based on fishery-independent sampling. ICLARM Conf. Proc. 48:59-73. Diaz, N., P. Gervain, and V. Druault-Aubin. In press. Queen snapper (Etelis oculatus) experimental deep-sea gillnet fishery in Guadeloupe I F.W.I. ). Proc. Gulf Carib. Fish. Inst. 55. Everson A. R., H. A. Williams, and B. M. Ito. 1989. Maturation and reproduction in two Hawaiian eteline snappers, uku, Aprion virescens, and onaga, Etelis coruscans. Fish. Bull. 87:877-888. FAO (Food and Agricultural Organization of the United Nations). 1993. Marine fishery resources of the Antilles. FAO Fish. Tech. Paper 326:1-235. Garcia-Cagide A., R. Claro, and B. V Koshelev. 2001. Reproductive patterns of fishes of the Cuban shelf. In Ecology of the marine fishes of Cuba (R. Claro, K. C. Lindeman, and L.R. Parenti, eds.), p. 73- 114. Smithsonian Institution Press, Washington and London. Gayanilo F., P. Sparre, and D. Pauly. 1996. FAO-ICLARM stock assessment tools. User's manual. FAO Comput. Inform. Ser. Fisheries 8:1- 126. Grimes C. B. 1987. Reproductive biology of the Lutjanidae: a review. In Tropical snappers and groupers: biology and fisheries management (J. J. Polovina and S. R. Ralston, eds.), p. 239-294. Westview Press, Boulder, CO. Koslow J. A., G. Boehlert, J. D. M. Gordon, R. L. Haedrich, P. Lorance, and N. Parin. 2000. Continental slope and deep-sea fisheries: impli- cations for a fragile ecosystem. ICES J. Mar. Sci. 57:548-557. NOTE Gobert et al.: Biology of Etelis oculatus in the Caribbean 425 Lorance P.. and H. Dupouy. 2001. CPUE abundance indices of the main target spe- cies of the French deep-water fishery in ICES sub-areas V-VII. Fish. Res. 51:137-149. Luckhurst B. E. 1996. Trends in commercial fishery landings of grou- pers and snappers in Bermuda from 1975 to 1992 and associated fishery management issues. ICLARM Conf. Proc. 48:277-288. Mahon. R. 1990. Fishery management options for Lesser Antilles countries. FAO Fish. Tech. Paper 313:1-126. Manooch, C. S. 1987. Age and growth of snappers and groupers. In Tropical snappers and groupers: biology and fisheries management (J. J. Polovina and S. R. Ralston, eds.), p. 329-373. Westview Press, Boulder, CO. Marcano, L. A., R. Guzman, and G. J. Gomez. 1996. Exploratory fishing with traps in oceanic islands off eastern Venezuela during 1992. ICLARM Conf. Proc. 48:331-336. Matos-Caraballo, D. 2000. Overview of Puerto Rico's small-scale fisheries statistics: 1994-1997. Proc. Gulf Carib. Fish. Inst. 51:215-231. Mendoza, J. J., and A. Larez, 1996. Abundance and distribution of snappers and grou- pers targeted by the artisanal medium range fishery off northeastern Venezuela (1981-19921. ICLARM Conf. Proc. 48:266-276. Murray, P. A. 1989. A comparative study of methods for determin- ing mean length-at-age and von Bertalanffy growth parameters for two fish species. M. Phil, thesis, 222 p. Univ. West Indies, Cave Hill, Barbados. Murray, P. A., and A. V. Charles. 1991. Some considerations for increasing landings of the queen snapper, Etelis oculatus Val., in the Saint Lucian fishery. FAO Fish. Rep. 431 (suppl):75-77. Murray P. A., L. E. Chinnery, and E. A. Moore. 1992. The recruitment of the queen snapper, Etelis ocu- latus Val., into the St Lucian fishery: recruitment of fish and recruitment of fisherman. Proc. Gulf Carib. Fish. Inst. 41:297-303. Murray, P. A., and E. A. Moore. 1992. Some morphometric relationships in Etelis ocula- tus Valenciennes (Queen snapperi, landed in St Lucia, W.I. Proc. Gulf Carib. Fish. Inst. 41:416-421 Murray, P. A., and J. D. Neilson. 2002. A method for the estimation of the von Berta- lanffy growth rate parameter by direct examination of otolith microstructure. Proc. Gulf Carib. Fish. Inst. 53:516-525. Prescod, S. D., H. A. Oxenford, and C. Taylor. 1996. The snapper fishery of Barbados: present status and a preliminary assessment of the potential for expansion. Proc. Gulf Carib. Fish. Inst. 44:159-179. Ralston, S. 1987. Mortality rates of snappers and groupers. In Tropical snappers and groupers: biology and fisheries management (J. J. Polovina and S. R. Ralston, eds.), p. 375-404. Westview Press, Boulder, CO. Ralston, S., R. M. Gooding, and G. M. Ludwig. 1986. An ecological survey and comparison of bottom fish resource assessments (submersible versus handline fishing) at Johnston atoll. Fish. Bull. 84:141-155. Simonds, K. 1995. Federal state cooperation in managing deepwater bottom fish in Hawaii. South Pacific Commission work- shop on the management of South Pacific inshore fisher- ies, Noumea (New Caledonia), 26 Jun-7 Jul 1995. South Pac. Comm. Tech. Doc. Intergr. Coast. Fish. Manag. Proj. 12:299-309. Wetherall, J. A., J. J. Polovina, and S. Ralston. 1987. Estimating growth and mortality in steady state fish stocks from length-frequency data. ICLARM Conf. Proc. 13:53-74. 426 Courtship and spawning behaviors of carangid species in Belize Rachel T. Graham Wildlife Conservation Society P.O. Box 37 Punta Gorda, Belize E-mail address: rgraham@wcs.org Daniel W. Castellanos Monkey River Village Toledo District, Belize Many species of reef fish aggre- gate seasonally in large numbers to spawn at predictable times and sites (Johannes, 1978; Sadovy, 1996; Domeier and Colin, 1997). Although spawning behavior has been observed for many reef fish in the wild (Wick- lund, 1969; Smith, 1972; Johannes, 1978; Sadovy et al., 1994; Aguilar Perera and Aguilar Davila, 1996), few records exist of observations on the courtship or natural spawning for the commercially important family Carangidae (jacks) (von Westernha- gen, 1974; Johannes, 1981; Sala et al., 2003). In this study, we present the first observations on the natural spawning behavior of the economi- cally-valuable permit (Trachinotus falcatus) (Linnaeus, 1758) from the full to new moon period at reef prom- ontories in Belize, with notes on the spawning of the yellow jack (Caran- goides bartholomaei) (Cuvier, 1833), and the courtship of five other caran- gid species. Permit belong to the family Ca- rangidae and are broadly distrib- uted in the western Atlantic Ocean from Massachusetts to southeastern Brazil, including the Caribbean Sea and Gulf of Mexico (Smith, 1997). Considered an inshore pelagic species (Valdez Munoz and Mochek, 2001) that spawns offshore, permit utilize a range of habitats that include coastal mangroves and seagrass beds, reef flats, and fore-reef areas during their life-cycle (Crabtree et al., 2002). Per- mit are reported to feed during the day and may show similar feeding characteristics to the closely related T. carolinus that displays a clear cir- cadian rhythm entrained to the light phase during its feeding period (Heil- man and Spieler, 1999). According to otolith analysis of fish caught in Flor- ida, permit live to at least 23 years and reach a maximum published fork length of 110 cm and a weight of 23 kg (Crabtree et al., 2002). 12 Permit are gonochoristic and Crabtree et al. (2002) recorded 50% sexual matu- rity for females at 547 mm FL or 3.1 years and males at 486 mm FL and 2.3 years. Permit exhibit a protract- ed spawning season from March to September in Cuba ( Garcia- Cagide et al., 2001) and from March to July in Florida (Crabtree et al., 2002). High gonadosomatic indices recorded for March and maturation of oocytes noted in late June- July (Crabtree et al., 2002) support the observations by Garcia-Cagide et al. (2001) that per- mit are batch spawners and have an asynchronous cycle of vitellogenesis. Spawning cued by the full moon has been recorded in many species of reef and inshore fish (Johannes, 1978, 1981; Moyer et al., 1983; Crabtree, 1995; Hoque et al., 1999). Macro- scopic gonadal analysis and observa- tions on the timing of courtship and spawning in several carangid species in the wild (Johannes, 1981; Sala et al., 2003), coupled with gonadal sampling observations on the cap- tive spawning behavior of the related bluefin trevally (Caranx melampygus) (Moriwake et al., 2001), further indi- cate that permit and other carangids display circa lunar periodicity when spawning naturally. Permit represent a valuable re- source for recreational fishermen throughout their range. In Florida, recreational fisheries land more than 100,000 fish per year, but declines in landings from 1991 to date prompted regulation (Crabtree et al., 2002) and a move towards catch-and-release of fish. As such, Belize is rapidly be- coming known as a world-class fly- fishing location due to its abundance of permit. The fishery is highly lucra- tive; flynshers pay up to US$500 per day in Belize to catch and release a permit. This niche tourism industry has also become an economic alter- native for local fishermen (Heyman and Graham3). Consequently, infor- mation on the timing and behavior of reproduction of permit can underpin conservation efforts that focus on a vulnerable stage in their life cycle. 1 The IGFA (International Game Fishing Association) notes a record length for permit of 122 cm FL. 2001. Database of IGFA angling records until 2001. IGFA, Dania Beach, Florida, 33004. 2 The United Nations notes a maximum weight of 36 kg for a permit. (Cervigon, F., R. Cipriani, W. Fischer, L. Garib- aldi, M. Hendrickx, A.J. Lemus, R. Marquez, J. M. Poutiers, G. Robaina and B. Rodriguez. 1992. Fichas FAO de identificacidn de especies para los fines de la pesca. Guia de campo de las especies comerciales marinas y de aquas salobres de la costa septentrional de Sur America, 513 p. FAO. Rome. 3 Heyman W. D., and R. T. Graham. 2000. The voice of the fishermen of Southern Belize, 44 p. TIDE (Toledo Institute for Environment and Devel- opment), P.O. Box 150, Punta Gorda, Belize. Manuscript submitted 9 December 2003 to the Scientific Editor's Office. Manuscript approved for publication 9 November 2004 by the Scientific Editor. Fish. Bull. 103:426-432 (2005). NOTE Graham and Castellanos: Courtship and spawning behaviors of carangid species in Belize 427 Materials and methods Turneffe Elbow (17°09'N, 87°54'W) and Gladden Spit (16°35'N, 88°00'W) are two sites located on the Belize Barrier Reef that were monitored for abundance and behavior of many species of spawning reef fish between 1999 and 2002. Both sites are promontories with a slop- ing reef shelf that drops off steeply at a depth of 35-45 m to over 1000 m into the southern tip of the Cayman Trench. According to the spawning aggregation criteria developed by Domeier and Colin (1997), Turneffe Elbow and Gladden Spit attract, respectively, an estimated 13 and 27 species of reef fish that aggregate seasonally to spawn (Graham, 2003). We logged over 270 hours of underwater monitoring of reef fish spawning aggregations at Turneffe Elbow and Gladden Spit, primarily from the full-moon to the new-moon from March to July from 2000 through 2002. Additional dives took place variously over the course of 3-5 days during the same lunar period from 1999 to 2002. Most dives for monitoring spawning aggregations took place between 0830 and 1100 hours, at midday, and between 1600 to 1730 hours of each diving day. Dives began 150-250 m north of both spawning aggre- gation sites and proceeded to the south along the reef platform edge. Dive depth usually began at about 30 m and decreased to 15 m as the dive progressed because of SCUBA decompression constraints. Dives normally lasted between 35 and 50 minutes. Horizontal and ver- tical visibility rarely dropped below 20-25 m. Results and discussion During 10 dive surveys (15 diving hours) at Turneffe Elbow, we observed a large school of 250 to 500 permit aggregating on the reef promontory (Table 1). The aggregated fish slowly swam into the south current along the south-facing sloping fore-reef shelf at 5-15 m depth and the steep drop-off located at -30-35 m. The school streamed down to the spur and groove for- mations at about 20 m depth on the reef shelf and rose up into the upper water column again. Permit were loosely grouped and displayed little fear of divers, a behavior commonly observed among a range of other fish species that aggregate to spawn (Graham, 2003). Several individuals displayed a dark patch located above and behind the pectoral fin on both flanks. Permit dis- played this same behavior coloration change during each encounter. On 22 August 2000, 7 days after the full moon, at -1730 (41 minutes before sunset at 1811 hours local time) we conducted our standard north to south fish census dive at a depth of -20-30 m along the fore-reef drop off. During all dives horizontal and vertical visibil- ity was at least 20 m and often over 40 m. We observed a school of -300 permit descend from 5-15 m depth above the fore-reef drop-off to 25 m directly on the shelf edge. At -1745 hours (26 min before sunset) within- group activity increased as permit schooled densely on the edge of the reef drop-off. At -1750 hours, a subgroup of eight permit left the dense school and ascended in the water column to -18 m depth. The lead individual initiating the ascent was -100 cm FL and was pursued by seven fish ranging from -55 to 75 cm FL. The pursu- ing fish nuzzled the larger fish's vent as it rose in the water column. All fish displayed a dark flank patch behind their pectoral fins. The lead permit then ceased its ascent at -15 m, tilted its head down slightly and convulsed rapidly, releasing a puff of gametes. Pursu- ing permit tried to position their vents as closely as possible to the lead individual's while releasing their gametes. The resulting gamete cloud was less than 50 cm in diameter and dispersed within seconds (Fig. 1). Following gamete release, all fish descended quickly to the main school still located -25 m below. Within mo- ments this behavior was repeated and observed in two smaller groups of permit before all observations ceased because of a lack of light. At Gladden Spit, we observed slightly different permit spawning behavior. On 7 April 2002 (10 days after the full moon), the aggregation remained in a restricted area -100 m north of where we previously witnessed the spawning of several species of fish and -30 m east of where we have also observed groupers Epinephelus striatus, Mycteroperca tigris, M. venenosa, and M. bonaci aggregate to spawn (Graham, 2003). Ambient water temperature was 27.7°C as measured by a temperature logger (Onset Corp. Tidbit data logger) moored at the spawning site at 30 m depth. At least 300 permit — many of them large individuals (-70-90 cm FL)— schooled densely into a ball at -1700 hours (66 minutes before sunset local time) near the reef shelf drop-off at a depth of -40-48 m. Subgroups comprised five to nine fish, and the lead fish was much larger than the pursuers. Subgroups rapidly rose up on the periphery of the school, spawned at the apex of the aggregation, and descended towards the bottom of the school again. Spawning was more frenetic than that observed at Turneffe Elbow. Permit subgroups behaved in the same manner as that observed at Turneffe dur- ing spawning, and all spawning individuals displayed a large dark flank patch behind the pectoral fins. Based on our observations of courtship and spawn- ing behavior, our estimate of spawning season for per- mit in Belize may stretch from February to the end of October, beyond the period of March to September as suggested by Garcia-Cagide et al. (2001) and Crabtree et al. (2002). Permit may also reach larger sizes than published by Crabtree et al. (2002); we estimated the largest individual permit observed at Turneffe Elbow in Belize to be -120 cm FL, which may indicate that permit exceed a lifespan of 23 years. We could not determine if the lead permit was fe- male and the pursuing permit were males because no individuals were caught for gonadal analysis. However, carangids are gonochoristic and it is highly likely that the lead fish in the spawning rush was female. Garcia- Cagide et al. (2001) noted that spawning females are often larger than mature males in several species of 428 Fishery Bulletin 103(2) Figure 1 Subgroup of eight permit [Trachinotus falcatus) immediately following spawning at Truneffe Elbow, Belize. The subgroup detached itself from the main aggregation to spawn in midwater at -15 m. The larger fish led the ascent to 15 m; all fish in the subgroup hovered at that depth, released gametes, and returned to the main school at a depth of -25 m. The arrow indicates the dark patch behind the pectoral fin that each fish sports during spawning. reef fish. This is also supported by our observations of gonochoristic spawners such as the cubera snapper (Lutjanus cyanopterus) and the dog snapper (L. jocu) that display a pattern of group, broadcast spawning where larger females are swollen with roe and lead the subgroup spawning ascents (Graham, 2003). Group spawning behavior in the yellow jack (C. bar- tholomaei) closely resembled that of permit. We recorded yellow jacks schooling at Gladden Spit on only two occa- sions (Table 1). On 7 April 2002, we observed that the yellow jacks spawned at -1705 hours (61 minutes before sunset local time) at Gladden Spit, less than 50 m south of the school of spawning permit. The jacks schooled densely at -40-45 m and subgroups of 5 to 8 fish de- tached themselves from within the school, ascending rapidly to -35 m, releasing gametes at the apex, and descending into the school again. Observations ceased shortly thereafter because of depth constraints and decreasing light. Not all species of carangids are group spawners. Pair spawning has been observed in species such as C. igno- bilis and Alectis indicus in the Pacific (von Westernha- gen, 1974) and C. sexfasciatus in the Gulf of California (Sala et al., 2003). We have also observed on numerous occasions pair courtship in crevalle jack (C. hippos), horse-eye jack (C. latus), and bar jack (Carangoides ruber) in schools exceeding 1000 fish, in rainbow run- ner iElagatis bipinnulata) in schools of up to -300 fish, and occasionally greater amberjack (Seriola dumerili) in schools numbering -120 individuals, primarily fol- lowing during the full-moon and waning moon periods between February and October (Table 1). These five species displayed extended pair courtship within and outside a large aggregation of conspecific fish as they swam along the edge of the reef drop-off. All courting pairs observed showed similar behavior. The chasing fish nuzzled the gonopore of the lead fish (whose head and upper body half had turned black but whose fins were lighter, Fig. 2, A and B) during prolonged chases, often swimming close to and at a perpendicular angle to the lead fish. Seriola dumerili also displayed dichroma- tism; the pursuing fish turned a vivid electric blue and exhibited a scrawled pattern on its upper flanks, simi- lar to that displayed by the scrawled filefish {Aluterus senptus). Occasionally, 1-10 individuals that did not display coloration changes followed the courting pairs. These five species may also pair spawn because their courtship behavior parallels that of C. sexfasciatus, observed by Sala et al. (2003) to spawn in pairs from the full moon to waning crescent periods from July to September. However, we did not observe any release of gametes during all pair courtship behavior. NOTE Graham and Castellanos: Courtship and spawning behaviors of carangid species in Belize 429 Figure 2 Pair courtship behavior in the horse-eye jack (Caranx latus) at Gladden Spit, Belize. The pursuing fish often swims slightly behind the lead and their flanks touch. The lead fish (A) remains silver colored, and the pursuing fish (Bl takes on a very dark coloration around the head and upper flank during courtship. Conclusions Our observations confirm that permit spawn offshore at reef promontories that support other reef fish spawn- ing aggregations. Permit demonstrate group broadcast spawning behavior and spawning events take place close to sunset. Further observations indicate that other species of carangids, such as yellow jack are also group broadcast spawners, occupying the same spatiotemporal spawning niche as permit. If observed courtship behav- 430 Fishery Bulletin 103(2) Table 1 Timing and lunar phase of observati Belize from April 1999 to July 2002. C = courting and color change; Spaw ans on the schooling, courtship, and spawning of seven carangids at two reef promontories in fm = full moon; dafm = days after full moon; dbfm = days before full moon. Sch = schooling; n = spawning observed. Date Dive start time Moon phase Location Species Behavior 2 Apr 1999 12:04 2 dafm Gladden Yellow jack Sch 3 Apr 1999 10:25 3 dafm Gladden Crevalle Sch 5 Apr 1999 16:40 5 dafm Gladden Crevalle, horse-eye, rainbow runner Sch 2 May 1999 16:50 2 dafm Gladden Bar jack, crevalle Sch 5 May 1999 5:40 5 dafm Gladden Horse-eye, crevalle Sch, C 30 May 1999 12:45 fm Gladden Amberjack, bar jack Sch, C 3 Jun 1999 9:10 4 dafm Gladden Horse-eye, bar jack Sch 4 Jun 1999 15:30 5 dafm Gladden Crevalle Sch 30 Jun 1999 12:00 2 dafm Gladden Bar jack, horse-eye Sch 27 Sep 1999 16:30 2 dafm Gladden Crevalle, amberjack Sch, C 28 Sep 1999 10:50 3 dafm Gladden Crevalle, bar jack, horse-eye Sch 16:30 3 dafm Gladden Horse-eye, amberjack Sch, C 24 Mar 2000 17:15 4 dafm Gladden Horse-eye Sch, C 17 Apr 2000 16:25 1 dbfm Gladden Horse-eye, crevalle Sch, C 18 Apr 2000 16:25 fm Gladden Bar jack, rainbow runner Sch 19 Apr 2000 17:10 1 dafm Gladden Horse-eye Sch, C 20 May 2000 17:00 2 dafm Gladden Crevalle Sch, C 23 May 2000 16:45 5 dafm Gladden Amberjack Sch, C 24 May 2000 16:21 6 dafm Gladden Bar jack Sch 25 May 2000 -16:30 7 dafm Gladden Crevalle Sch 26 May 2000 16:00 8 dafm Gladden Horse-eye, crevalle Sch, C 23 Jun 2000 17:30 7 dafm Gladden Bar jack Sch, C 18 Aug 2000 15:36 3 dafm Gladden Bar jack Sch 15:36 3 dafm Gladden Horse-eye, rainbow runner Sch, C 19 Aug 2000 -12:00 4 dafm Gladden Bar jack, crevalle Sch 20 Aug 2000 15:00 5 dafm Turneffe Horse-eye C 15:00 5 dafm Turneffe Permit Sch 20 Aug 2000 17:00 5 dafm Turneffe Permit Sch, C 17:00 5 dafm Turneffe Amberjack, bar jack Sch, C 21 Aug 2000 15:00 6 dafm Turneffe Crevalle, horse-eye Sch, C 15:00 6 dafm Turneffe Permit Sch 22 Aug 2000 17:30 7 dafm Turneffe Horse-eye C 17:30 7 dafm Turneffe Permit Spawn 14 Oct 2000 17:30 1 dafm Gladden Horse-eye, crevalle C 15 Oct 2000 17:30 2 dafm Gladden Rainbow runner C 17 Oct 2000 16:30 4 dafm Turneffe Horse-eye, crevalle C 16:30 4 dafm Turneffe Permit, amberjack Sch 18 Oct 2000 16:30 5 dafm Turneffe Horse-eye, crevalle, amberjack, permit C 13 Dec 2000 16:30 2 dafm Gladden Horse-eye Sch 9 Apr 2001 16:00 1 dafm Gladden Horse-eye C 8 May 2001 11:15 1 dafm Gladden Crevalle c 9 May 2001 -10:30 2 dafm Gladden Crevalle Sch 7 Jun 2001 17:00 1 dafm Gladden Crevalle, bar jack, horse-eye Sch continued NOTE Graham and Castellanos: Courtship and spawning behaviors of carangid species in Belize 431 Table 1 (continued) Dive Date start time Moon phase Location Species Behavior 8 Jun 2001 17:00 2 dafm Gladden Amberjack, crevalle horse-eye Sch, C 9Jun 2001 11:00 3 dafm Gladden Crevalle, horse-eye Sch 10 Jun 2001 17:50 4 dafm Gladden Crevalle, horse-eye Sch, C 3 Oct 2001 -17:00 1 dafm Turneffe Horse-eye C 6 Feb 2002 16:00 9 dafm Turneffe Horse-eye, permit Sch 7 Feb 2002 8:30 10 dafm Turneffe Horse-eye, permit C 16:30 10 dafm Gladden Horse-eye, crevalle. bar jack Sch 28 Mar 2002 16:48 fm Gladden Crevalle, permit Sch 16:48 fm Gladden Horse-eye C 29 Mar 2002 16:30 1 dafm Gladden Crevalle, bar jack, horse-eye Sch 30 Mar 2002 16:45 2 dafm Gladden Crevalle, horse-eye Sch 31 Mar 2002 16:40 3 dafm Gladden Horse-eye Sch 1 Apr 2002 16:35 4 dafm Gladden Bar jack, horse-eye Sch 3 Apr 2002 9:40 5 dafm Gladden Bar jack, horse-eye Sch 7 Apr 2002 10:30 9 dafm Gladden Bar jack Sch 16:30 9 dafm Gladden Permit, yellow jack Spawn 16:30 9 dafm Gladden Bar jack Sch 6 May 2002 9:40 9 dafm Gladden Horse Sch 27 May 2002 12:18 1 dafm Gladden Horse-eye C 30 May 2002 11:07 4 dafm Gladden Horse-eye Sch 31 May 2002 16:20 5 dafm Gladden Crevalle Sch 1 Jun 2002 16:15 6 dafm Gladden Bar jack, rainbow runner Sch 2 Jun 2002 16:15 7 dafm Gladden Bar jack, horse-eye Sch 29 Jun 2002 12:00 5 dafm Turneffe Permit, horse-eye Sch 1 Jul 2002 15:00 7 dafm Gladden Bar jack, crevalle, horse-eye Sch ior is included, the spawning season for permit and horse-eye jacks is protracted from February through October, and the five other carangid species described in the present study spawned within this period. Pro- tection of permit stocks throughout their life cycle, and particularly during their spawning season, underpins the associated rapidly growing and economically lucra- tive flyfishing tourism. Future directions of study should include a study of permit movement patterns between feeding and spawning grounds and mortality rates of catch-and-release fishing. Acknowledgments We would like to thank two anonymous reviewers who provided helpful suggestions for the improvement of this paper. The fieldwork and observations were supported by grants from the UK Darwin Initiative and the UK's Natural Environment Research Council. We worked under permits provided by the Belize Department of Fisheries. Literature cited Aguilar Perera, A., and W. Aguilar Davila. 1996. A spawning aggregation of Nassau grouper Epi- nephelus striatus (Pisces: Serranidae) in the Mexican Caribbean. Environ. Biol. Fishes 45:351-361. Crabtree, R. E. 1995. Relationship between lunar phase and spawning activity of tarpon, Megalops atlanticus, with notes on the distribution of larvae. Bull. Mar. Sci. 56:895-899. Crabtree, R. E., P. B. Hood, and D. Snodgrass. 2002. Age, growth and reproduction of permit (Trachinotus falcatus) in Florida waters. Fish. Bull. 100:26-34. Domeier, M. L., and P. L. Colin. 1997. Tropical reef fish spawning aggregations: defined and reviewed. Bull. Mar. Sci. 60:698-726. Garcia-Cagide, A., R. Claro, and B. V. Koshelev. 2001. Reproductive patterns of fishes of the Cuban shelf. In Ecology of the marine fishes of Cuba (R. Claro, K. C. Lindeman and L. R. Parenti, eds.), p. 73-114. Smithsonian Institution Press, Washington DC. Graham, R. T. 2003. Behavior and conservation of whale sharks on the 432 Fishery Bulletin 103(2) Belize Barrier Reef. Ph.D. diss., 408 p. Environment Department, Univ. York, York, UK. Heilman, M. J., and R. W. Spieler. 1999. The daily feeding rhythm to demand feeders and the effects of timed meal-feeding on the growth of juvenile Florida pompano, Trachinotus karolinus. Aquaculture 180:53-64. Hoque, M. M., A. Takemura, M. Matsuyama, S. Matsuura, and K. Takano. 1999. Lunar spawning in Siganus canaliculatus. J. Fish Biol. 55:1213-1222. Johannes, R. E. 1978. Reproductive strategies of coastal marine fishes in the tropics. Environ. Biol. Fishes 3: 65-84. Johannes, R. E. 1981. Words of the lagoon: fishing and marine lore in the Palau, District of Micronesia, p. 245. Univ. California Press. Berkeley, CA. Moriwake, A. M., V. N. Moriwake, A. C. Ostrowski, and C. S. Lee. 2001. Natural spawning of the bluefin trevally Caranx melampygus in captivity. Aquaculture 203(1-2): 159-164. Moyer, J. T„ R. E. Thresher, and P. L. Colin. 1983. Courtship, spawning and inferred social organi- zation of American angelfishes (Genera Pomacanthus, Holacanthus and Centropyge; Pomacanthidac). Environ. Biol. Fishes 9:25-39. Sadovy, Y. 1996. Reproduction of reef fishery species. In Reef fisheries (N. V. C. Polunin and C. M. Roberts, eds.l, p. 15-59. Chapman and Hall, London. Sadovy, Y., P. L. Colin, and M. L. Domeier. 1994. Aggregation and spawning in the tiger grou- per, Mycteroperca tigris (Pisces: Serranidae). Copeia 1994:511-516. Sala, E., O. Aburto-Oropereza, G. Paredes, and G. Thompson. 2003. Spawning aggregations and reproductive behavior of reef fishes in the Gulf of California. Bull. Mar. Sci. 72(11:103-121. Smith, C. L. 1972. A spawning aggregation of Nassau grouper, Epinephelus striatus (Bloch). Trans. Am. Fish. Soc. 101:257-261. 1997. National Audubon Society field guide to tropical marine fishes of the Caribbean, the Gulf of Mexico, Florida, the Bahamas, and Bermuda, 718 p. Alfred A. Knopf, Inc.. New York, NY. Valdez Mufioz, E., and A. D. Mochek. 2001. Behavior of marine fishes of the Cuban shelf. In Ecology of the marine fishes of Cuba (R. Claro, K. C. Lindeman and L. R. Parenti, eds.l, p. 53-72. Smith- sonian Institution Press, Washington DC. von Westernhagen, H. 1974. Observation on the natural spawning of Alec- tis indicus (Ruppelli and Caranx ignobilis (Forsk.) (Carangidael. J. Fish Biol. 6:513-516. Wicklund, R. 1969. Observations on spawning of the lane snapper. Underwater Naturalist 6:40. 433 Comparison of two approaches for estimating natural mortality based on longevity* David A. Hewitt John M. Hoenig Virginia Institute of Marine Science The College of William and Mary P.O. Box 1346 Gloucester Point, Virginia 23062 E-mail address (for D A Hewitt) dhewittiq'vimsedu Vetter (1988) noted that her review of the estimation of the instanta- neous natural mortality rate (M) was initiated by a discussion among colleagues that identified M as the single most important but least well-estimated parameter in fishery models. Although much has been accomplished in the intervening years, M remains one of the most difficult parameters to estimate in fishery stock assessments. A number of novel approaches using tagging and telemetry data provide promise for making reliable direct estimates of M for a given stock (Hearn et al., 1998; Frusher and Hoenig, 2001; Hightower et al., 2001; Latour et al., 2003; Pollock et al., 2004). However, such methods are often impracticable and fishery scientists must approxi- mate M by using estimates made for other stocks of the same or simi- lar species or by predicting M from features of the species' life history (Beverton and Holt, 1959; Beverton, 1963; Alverson and Carney, 1975; Pauly, 1980; Hoenig, 1983; Peterson and Wroblewski, 1984; Roff, 1984; Gunderson and Dygert, 1988; Chen and Watanabe, 1989; Charnov, 1993; Jensen, 1996; Lorenzen, 1996). We are concerned with two ap- proaches for predicting M based solely on the longevity of the mem- bers of a stock — an approach that can be used when data are not available to make direct estimates of the parameter. One is a linear re- gression model (Hoenig, 1983) and the other is a simple rule-of-thumb approach. Hoenig (1983) found that M was inversely correlated with lon- gevity across a wide variety of taxa and recommended use of the follow- ing predictive equation relating the maximum age observed in the stock Umax) to M: ln(M) = 1.44-0.982xln(?max). (1) The rule-of-thumb approach consists of determining the value of M such that 100(P)% of the animals in the stock survive to the age tmax; thus, M- -ln(P> (2) The challenge in this approach is determining an appropriate value for the proportion P. The rule-of-thumb approach has the potential to be used widely be- cause it is presented in Quinn and Deriso (1999) and stock assessment manuals of the Food and Agriculture Organization of the United Nations (FAO; Sparre and Venema, 1998; Cadima, 2003). The approach has re- cently been used extensively, in the specific form M~3/tmax, in work relat- ed to stock assessments for blue crab (Callinectes sapiclus). In this note, we 1) show that the regression model and the rule-of-thumb approach can be compared directly; 2) illustrate the difference in the estimates of M generated by the two approaches; 3) discuss the origins and current use of the rule-of-thumb approach; and 4) recommend that the regression model be used instead of the rule-of-thumb approach. Methods With the rule-of-thumb approach, the fraction of a population that survives to a given age is used to estimate M. This approach is equivalent to a quantile estimator (Bury, 1975). Sup- pose the fraction surviving to age / is described by the negative exponential function ~-zt (3) where Z is the total instantaneous mortality rate. The quantile estima- tor is of the form -ZrP (4) where rp is the age at which 100(P)% of the population remains. In the case where P = 0.05, the estimator, based on data from a sample of the popula- tion, is 0.05 = (5) where 595 of the animals in the sample are older than age t005. To estimate M, an empirical ap- proach is usually taken where f0 05 is replaced with tmax: 0.05: -», (6) where tma!i is either the oldest age observed in the stock or the oldest age found in the literature for the spe- cies of interest. When age composition data are used from an exploited stock. Equation 6 will provide an estimate of M only if fishing mortality is rea- sonably close to zero iM=*Z) or if there is a refuge where older animals can accumulate. If exploitation affects all * Contribution 2637 of the Virginia Insti- tute of Marine Science, The College of William and Mary, Gloucester Point, VA 23062. Manuscript submitted 25 March 2004 to the Scientific Editor's Office. Manuscript approved for publication 12 October 2004 by the Scientific Editor. Fish. Bull. 103:433-437(2005). 434 Fishery Bulletin 103(2) 0 025 - / \ - 8 "0 \ Absolute differenc RE-RT o o o o o o o Ul o / — ■ — / srcent difference (RE-RT)/RE CD T / - 2 0.005 - 1 • Percent 0.000 - 1 3 5 7 9 11 13 15 17 19 21 23 25 27 29 31 96 100 max Figure 1 The absolute and percent difference between estimates of M from the regres- sion estimator (RE) and the approximate rule of thumb, 4.22/rmax (RTl. animals in the stock, Equation 6 is unlikely to provide a reliable estimate of M. The rule of thumb for approximating M follows di- rectly from Equation 6: -ln(0.05) = M xtn M = 2.996 (7) Most importantly, note that the use of 0.05 or any other proportion in the equations is arbitrary because we have no reason to believe that tmax pertains to any particular quantile. We show in the present study that this arbitrary rule of thumb for approximating M is unnecessary, as an empirical method (Hoenig, 1983) provides an analogous estimate based on a substantial data set. Equation 1 is based on the same model as that in Equation 3 and was developed from a regression of In (AD on ln(l, although the difference is usually small (Fig. 1). Estimates from the regression estimator are typically 40-50% greater than estimates from 3/tmax (Fig. 2). For example, if a maximum age of eight years is used for blue crab in Chesapeake Bay (Rugolo et al., 1998), 3/tmax gives an estimate for M of 0.375/yr and the re- gression estimator gives 0.548/yr. Perhaps the most significant result is the finding that rearrangement of the regression model yields an esti- mate of an appropriate value for P in Equation 2. The value of 4.22 in Equation 8 approximately corresponds to -ln( 0.015), indicating that the average longevity for stocks in the data set used by Hoenig (1983) is the age at which about 1.57c of the stock remains alive (versus 5% in 3/tmax). Discussion Development of the rule-of-thumb approach The rule-of-thumb approach appears to have arisen inde- pendently in four different places. Cadima (2003) sup- ported the approach by citing the early work of Tanaka (1960). Sparre and Venema (1998) based their presen- NOTE Hewitt and Hoenig: Estimating natural mortality from longevity 435 0) LU Absolute Percent i ■ i ■ i ■ : F^= 1 3 5 7 9 11 13 15 17 19 21 23 25 27 29 31 96 100 max Figure 2 The absolute and percent difference between estimates of M from the regression estimator (RE) and 3/imM (3M). tation on the work of Alagaraja (1984), who provided the mathematics of a method that Sekharan (1975) used without description. Interestingly, Shepherd and Breen (1992) rearranged Equation 3 to obtain the rule of thumb based on the results of Hoenig (1983). This latter presentation is provided in Quinn and Deriso (1999). In all of these cases, the proportion of animals surviving to £max is assumed to be some arbitrarily small value, typically 1% or 5%. The development and use of the specific form 3/tmax in blue crab work occurred altogether separately. Its use began with an assessment for the Chesapeake Bay stock, in which Rugolo et al. (1998) used an estimate of M based on "the ICES [International Council for the Exploration of the Sea] convention; that is, 5% survivor- ship at maximum age following negative exponential de- pletion." The approach is more explicitly denned in their original document (Rugolo et al.1) as M = (3/maximum age). The report also states that "this convention ... is widely used for many east coast finfish stocks (NMFS [National Marine Fisheries Service]/NEFSC [Northeast Fisheries Science Center], ASMFC [Atlantic States Ma- rine Fisheries Commission])." Following its introduction by Rugolo et al. (Rugolo et al.1; Rugolo et al., 1998), the 3/£max approach has been used in nearly all blue crab Rugolo, L., K. Knotts, A. Lange, V. Crecco, M. Terceiro, C. Bonzek, C. Stagg, R. O'Reilly, and D. Vaughan. 1997. Stock assessment of Chesapeake Bay blue crab (Callinectes sapi- dus), 267 p. Report of the Technical Subcommittee of the Chesapeake Bay Stock Assessment Committee of the National Marine Fisheries Service, NOAA (National Oceanic and Atmospheric Administration). NOAA Chesapeake Bay Office, 410 Severn Avenue, Suite 107, Annapolis, MD 21403. stock assessment work conducted on the east coast of the United States (Miller and Houde2; Miller, 2001; Murphy et al.3; Helser et al., 2002; Kahn4). The references used by Rugolo et al. (1998) in support of what they termed the "ICES convention" (Antho- ny5; Vetter, 1988) do not mention the 3/tmax approach. Rather than advocating a method for determining M, Anthony5 called for standardization of the range of ages to include in the calculation of yield-per-recruit for a stock; this range of ages was termed the stock's "fish- able life span." He proposed that the fishable life span should be defined such that the oldest age would be that 2 Miller, T. J., and E. D. Houde. 1999. Blue crab target setting, 167 p. Final report to the Living Resources Sub- committee of the Chesapeake Bay Program. University of Maryland Center for Environmental Science (UMCES) Technical Series No. TS-177-99. Chesapeake Bay Program, U.S. EPA (Environmental Protection Agency), 410 Severn Avenue, Annapolis, MD 21403. 3 Murphy, M. D., C. A. Meyer, and A. L. McMillen- Jackson. 2001. A stock assessment for blue crab, Ca Uinectes sapidus, in Florida waters, 56 p. FMRI (Florida Marine Research Institute) Inhouse Report Series IHR 2001-008. Florida Fish and Wildlife Conservation Commission, FMRI. 100 Eighth Avenue SE, St. Petersburg, FL 33701. 4 Kahn, D. M. 2003. Stock assessment of Delaware Bay blue crab (Callinectes sapidus) for 2003, 52 p. Delaware Department of Natural Resources and Environmental Control. Division of Fish and Wildlife, P.O. Box 330, Little Creek. DE 19961. 5 Anthony, V. C. 1982. The calculation of F0-1: a plea for standardization, 16 p. Northwest Atlantic Fisheries Organi- zation ( NAFO ) Serial Document N557, SCR 82/VI/64. NAFO Secretariat, P.O. Box 638, Dartmouth, Nova Scotia B2Y 3Y9, Canada. 436 Fishery Bulletin 103(2) at which 59c or less of the initial recruits survived. The use of Anthony's standard to approximate M makes the assumption that the fishable life span of an exploited stock is the same as the longevity of the members of the stock in an unexploited condition. It is unlikely that this assumption will be met unless the fishery is at an early stage in its development because fishing may alter the age structure of the stock (Hilborn and Walters, 1992). We note that although a limited num- ber of scientists involved with ICES have used 3/tmax in a general way, the method has not been adopted as a convention within ICES (O'Brien6). Furthermore, we did not find evidence that the approach is currently in common use in stock assessments on the east coast of the United States, with the exception of those for blue crab. Nonetheless, the rule-of-thumb approach certainly has the potential to be used widely, given its repeated presentation in fishery literature and its accumulated momentum in blue crab work. Recommendations The power of empirical relationships for predicting natu- ral mortality can be rather limited (Vetter, 1988; Pas- cual and Iribarne, 1993), and the uncertainty associated with parameter estimates should be taken into account whenever possible (Patterson et al., 2001). Further- more, methods for directly estimating M are likely to be preferable to making predictions based on life history features. Nonetheless, such estimates may be needed when available data are inadequate for making a direct estimate. Given the results of our comparison, we recom- mend that the regression estimator be used instead of the rule-of-thumb approach when longevity is used to predict M. The regression estimator is based on a least squares fit to an extensive data set and thus matches experience better than a rule-of-thumb approach based on an arbitrary constant. We recommend that use of the 3/tmax rule of thumb be abandoned, despite it being entrenched in blue crab literature. For a species like blue crab, for which tmax is less than 10 years, the differences in the estimates of M from the regression estimator and 3/tmax are not trivial (-45%). Although the regression estimator was based on data for fish, mollusks, and cetaceans (Hoenig, 1983) and may not be applicable to other exploited taxa, such as crustaceans, the model had a good fit to the data across widely disparate taxa. Finally, estimates of M for blue crab based on longevity are controversial because of continued difficulty in determining an appropriate 'max- In *"ne aDsence of data to directly estimate M for this species, we suggest that the most prudent course O'Brien, C. M. 2004. Personal commun. Chair of ICES Working Group on Methods of Fish Stock Assessments and ICES Resource Management Committee. CEFAS (Centre for Environment, Fisheries and Aquaculture Science) Lowestoft Laboratory, Pakefield Road, Lowestoft, Suffolk NR33 0HT, England. of action is a review and comparison of other methods for predicting M. Acknowledgments We thank Doug Vaughan for helping investigate the use of the rule-of-thumb approach, and Russell Burke, Romuald Lipcius, Jacques van Montfrans, and three anonymous reviewers for helpful comments on the manu- script. D.A.H. gratefully acknowledges the support of the Willard A. Van Engel (WAVE) Fellowship for Crus- tacean Research. This work was supported by fund- ing from the NOAA Chesapeake Bay Office, award no. NA03NMF4570376. Literature cited Alagaraja. K. 1984. Simple methods for estimation of parameters for assessing exploited fish stocks. Indian J. Fish. 31:177-208. Alverson, D. L., and M. J. Carney. 1975. A graphic review of the growth and decay of popula- tion cohorts. J. Cons. Int. Explor. Mer 36:133-143. Beverton, R. J. H. 1963. Maturation, growth and mortality of clupeid and engraulid stocks in relation to fishing. Rapp. P.-V. Reun. Cons. Int. Explor. Mer 154:44-67. Beverton, R. J. H., and S. J. Holt. 1959. A review of the lifespans and mortality rates offish in nature, and their relation to growth and other physi- ological characteristics. In The lifespan of animals, CIBA Foundation colloquia on ageing, vol. 5 (G. E. W. Westenholme and M. O'Connor, eds.), p. 142-177. Little, Brown and Company, Boston, MA. Bury, K. V. 1975. Statistical models in applied science, 625 p. John Wiley and Sons, New York, NY. Cadima, E. L. 2003. Fish stock assessment manual. FAO Fish. Tech. Pap. 393, 161 p. FAO, Rome. Charnov, E. L. 1993. Life history invariants: some explorations of sym- metry in evolutionary ecology. Oxford Univ. Press, New York, NY. Chen, S.. and S. Watanabe. 1989. Age dependence of natural mortality coefficient in fish population dynamics. Nippon Suisan Gakkaishi 55:205-208. Frusher, S. D., and J. M. Hoenig. 2001. Estimating natural and fishing mortality and tag reporting rate of southern rock lobster ijasus edwardsii) from a multiyear tagging model. Can. J. Fish. Aquat. Sci. 58:2490-2501. Gunderson, D. R., and P. H. Dygert. 1988. Reproductive effort as a predictor of natural mor- tality rate. J. Cons. Int. Explor. Mer 44:200-209. Hearn. W. S., K. H. Pollock, and E. N. Brooks. 1998. Pre- and post-season tagging models: estimation of reporting rate and fishing and natural mortality rates. Can. J. Fish. Aquat. Sci. 55:199-205. NOTE Hewitt and Hoenig: Estimating natural mortality from longevity 437 Helser. T. E., A. Sharov, and D. M. Kahn. 2002. A stochastic decision-based approach to assessing the status of the Delaware Bay blue crab stock. In Incor- porating uncertainty into fishery models (J. M. Berkson. L. L. Kline, and D. J. Orth, eds. I, p. 63-82. Am. Fish. Soc, Symposium 27, Bethesda. MD. Hightower. J. E., J. R. Jackson, and K. H. Pollock. 2001. Use of telemetry methods to estimate natural and fishing mortality of striped bass in Lake Gaston, North Carolina. Trans. Am. Fish. Soc. 130:557-567. Hilborn, R., and C. J. Walters. 1992. Quantitative fisheries stock assessment: choice, dynamics and uncertainty, 570 p. Chapman and Hall, New York, NY. Hoenig, J. M. 1983. Empirical use of longevity data to estimate mor- tality rates. Fish. Bull. 82:898-903. Jensen, A. L. 1996. Beverton and Holt life history invariants result from optimal trade-off of reproduction and survival. Can. J. Fish. Aquat. Sci. 53:820-822. Latour, R. J., J. M. Hoenig, D. A. Hepworth, and S. D. Frusher. 2003. A novel tag-recovery model with two size classes for estimating fishing and natural mortality, with impli- cations for the southern rock lobster (Jasus edwardsii ) in Tasmania, Australia. ICES J. Mar. Sci. 60:1075- 1085. Lorenzen, K. 1996. The relationship between body weight and natu- ral mortality in juvenile and adult fish: a comparison of natural ecosystems and aquaculture. J. Fish Biol. 49:627-647. Miller, T. J. 2001. Matrix-based modeling of blue crab population dynamics with applications to the Chesapeake Bay. Es- tuaries 24:535-544. Pascual, M. A., and O. O. Iribarne. 1993. How good are empirical predictions of natural mortality? Fish. Res. 16:17-24. Patterson, K., R. Cook, C. Darby, S. Gavaris, L. Kell, P. Lewy, B. Mesnil, A. Punt, V. Restrepo, D. W. Skagen, and G. Stefansson. 2001. Estimating uncertainty in fish stock assessment and forecasting. Fish and Fisheries 2:125-157. Pauly, D. 1980. On the interrelationships between natural mor- tality, growth parameters, and mean environmental temperature in 175 fish stocks. J. Cons. Int. Explor. Mer 39:175-192. Peterson, I., and J. S. Wroblewski. 1984. Mortality rate of fishes in the pelagic eco- system. Can. J. Fish. Aquat. Sci. 41:1117-1120. Pollock, K. H., H. Jiang, and J. E. Hightower. 2004. Combining telemetry and fisheries tagging models to estimate fishing and natural mortality rates. Trans. Am. Fish. Soc. 133:639-648. Quinn, T. J. II, and R. B. Deriso. 1999. Quantitative fish dynamics, 542 p. Oxford Univ. Press, New York, NY. Roff, D. A. 1984. The evolution of life history parameters in teleosts. Can. J. Fish. Aquat. Sci. 41:989-1000. Rugolo, L. J., K. S. Knotts. A. M. Lange, and V. A. Crecco. 1998. Stock assessment of Chesapeake Bay blue crab (Callineetes sapidus Rathbuni. J. Shellfish Res. 17: 493-517. Sekharan, K. V. 1975. Estimates of the stocks of oil sardine and mack- erel in the present fishing grounds off the west coast of India. Indian J. Fish. 21:177-182. Shepherd, S. A., and P. A. Breen. 1992. Mortality in abalone: its estimation, variability and causes. In Abalone of the world: biology, fisheries and culture IS. A. Shepherd, M. J. Tegner, and S. A. Guzman del Proo, eds.). p. 276-304. Fishing News Books, Cambridge, UK. Sparre, P., and S. C. Venema. 1998. Introduction to tropical fish stock assessment, Part 1: manual. FAO Fish. Tech. Pap. 306/1, rev. 2, 407 p. FAO, Rome. Tanaka, S. 1960. Studies on the dynamics and the management of fish populations. Bull. Tokai Reg. Fish. Res. Lab. 28:1-200. Vetter, E. F. 1988. Estimation of natural mortality in fish stocks: a review. Fish. Bull. 86:25-43. 438 Effects of current speed and turbidity on stationary light-trap catches of larval and juvenile fishes David C. Lindquist Richard F. Shaw Coastal Fisheries Institute School of the Coast and Environment Louisiana State University Baton Rouge, Louisiana 70803 E-mail address (for D C Lindquist): dlindqlig1 lsu.edu Light traps are one of a number of different gears used to sample pelagic larval and juvenile fishes. In contrast to conventional towed nets, light traps primarily collect larger size classes, including settlement-size larvae (Choat et al., 1993; Hickford and Schiel, 1999; Hernandez and Shaw, 2003), and, therefore, have become important tools for discern- ing recruitment dynamics (Sponau- gle and Cowen, 1996; Wilson, 2001). The relative ease with which multiple synoptic light trap samples can be taken means that larval distribu- tion patterns can be mapped with greater spatial resolution (Doherty, 1987). Light traps are also useful for sampling shallow or structurally complex habitats where towed nets are ineffective or prohibited (Gregory and Powles, 1985; Brogan, 1994; Her- nandez and Shaw, 2003). As with any sampling gear, there are concerns about light trap sam- pling biases and efficiency. Light traps are taxon-selective because they target fishes that are photoposi- tive and able to swim to and enter the trap (Thorrold, 1992; Choat et al. 1993; Hernandez and Shaw, 2003), and size-selective because both pho- totactic behavior and swimming abil- ities change during ontogeny (Stea- rns et al., 1994; Fisher et al., 2000). LTnlike conventional towed nets, it is difficult, if not impossible, to quan- tify the volume of water sampled by light traps. This is largely due to ex- ternal, environmental factors such as lunar phases, current speed or water clarity, which may have a large im- pact on catch rates (Doherty, 1987; Meekan et al, 2000). Few studies have attempted to ad- dress the effects of environmental factors on light trap performance. Catches have been found to be lower during full moons as compared to new moons, either because of the greater ambient illumination interfering with light trap efficiency (Gregory and Pow- les, 1985; Hickford and Schiel, 1999) or because of higher abundances of presettlement fish during the darker lunar phases (Johannes, 1978; Rob- ertson et al., 1988). Thorrold (1992) showed that catches were greater for light traps drifting with the current as compared to traps anchored in the current flow. Anderson et al. (2002) found that anchored light traps were less efficient at a high-current sam- pling site as compared with a low- current sampling site. The latter two studies, however, did not provide any information on catch rates with varia- tion in current speed. The purpose of this study was to assess the relation- ships between catch rates from sta- tionary (anchored or tethered) light traps at offshore petroleum platforms and concurrent measurements of cur- rent speed and turbidity. Materials and methods Study sites Larval and juvenile fishes were col- lected at five oil and gas platforms (platforms) in the north-central Gulf of Mexico. These platforms included: Mobil's Green Canyon 18 (27°56'37"N, 91°0'45"W; sampled from July 1995- June 1996); Mobil's Grand Isle 94B (28°30'57"N, 90°07'23"W; April-August 1996); Exxon's South Timbalier 54G (28°50'01"N, 90°25'00"W; April-September 1997); Santa Fe-Snyder's Main Pass 259A (29C19'32"N, 88°01'12"W; May- September 1999); and Murphy Oil's Viosca Knoll 203 (29°46'53"N, 88°19'59"W; May-October 2000). All platforms had similar underwater structural complexity, and had well- developed biofouling communities when sampled. Sampling procedures Sampling procedures have been described in detail elsewhere (Her- nandez and Shaw, 2003) and will be briefly described here. Fish collec- tions were made by using a modified quatrefoil light trap with a Brinkman Starfire II halogen light (250,000 can- dlepower) powered through an umbili- cal by a 12-volt marine battery. Light traps were deployed in surface waters within the platform structure along a stainless-steel guidewire (within- platform light trap), and tethered and floated in surface waters to a distance of 20 m from the down-current side of the platform (off-platform light trap). Light traps were deployed with their lights off, fished with lights on for 10-15 min, and retrieved with lights off. Sampling was undertaken general- ly twice monthly coincident with new and full moon phases. During each trip, light traps were fished during four to six sets per night, starting at least one hour after sunset and ending at least one hour before sun- rise, over two to three consecutive nights. Each sample set consisted of a within-platform light trap collec- Manuscript submitted 4 February 2004 to the Scientific Editor's Office. Manuscript approved for publication 1 December 2004 by the Scientific Editor. Fish. Bull. 103:438-444 (2005). NOTE Lmdquist and Shaw: Effects of current speed and turbidity on catches of larval and juvenile fishes 439 140 - • 1 120- ° 100 - sh per 00 o • • • • & 60 - LU g 40- § 2°- 2 •• • I • • ...% • 0 i 1" ~' r" T — — — i ™— n 1 i i • - i 0 10 20 30 40 50 60 70 80 90 Mean water current speed (cm/sec) Figure 1 Mean total CPUE per sampling set (from within- and off-platform light traps) in relation to the mean current speed per sampling set. Data from all platforms were combined. Line calculated from the regression equation: loglfl(y+l) = -0.013.V + 1.302, r2 = 0.23. tion and an off-platform light trap collection in random order. During sampling, turbidity (Nephelometric tur- bidity unit: NTU) was measured every 5 sec by using a Hydrolab DataSonde3 suspended in surface waters within the platform structure. Current speed and direc- tion were measured every 10 min with an InterOcean S4 Current Meter suspended 1-2 m below the surface on the up-current side of the platform. Because the platform structure undoubtedly reduced current speeds (Forristall, 1996), current data taken from this location should be considered as relative estimates for the light trap collections. Samples were preserved in 10% buffered formalin and transferred to ethanol within 12 hours. Fish were enu- merated and identified to the lowest possible taxonomic level. Preflexion larvae were measured to notochord length, and postflexion and juvenile fish were measured to standard length. Data from light trap catches were standardized to a catch per unit of effort (CPUE) of number of fish per 10 minutes. Data analyses We assumed that there were no inter-location differences in the relationship between light trap CPUE and current speed or turbidity; therefore, data from all platforms for the months May to September were combined. The relationship between total light trap CPUE and current speed or turbidity was analyzed by using regression analysis. Current speed and turbidity were analyzed separately, rather than in a multiple regression analysis, because there was a limited number of sampling sets where we had data for light trap CPUE, current speed and turbidity together (n = 60, or 31% and 37% of the available turbidity and current data, respectively). There were no significant differences in the regression coef- ficients of CPUE vs. current speed or turbidity between within- and off-platform light traps (P>0.15); therefore, the CPUEs from both light traps were averaged for each sampling set. Mean total CPUEs were log-transformed (log10(.y+l)) and analyzed with the mean current speed or turbidity from each respective sampling set. Mean CPUEs were also calculated for the dominant families collected; however, regression analyses could not be performed because variances remained heterogeneous after transformation. To investigate how fish size (i.e., locomotive ability) influenced light trap catches with increasing current speed, length-frequency distributions of all fishes col- lected at different current speed intervals (0-9, 10-19, 20-29, 30-39, 40-49 and >49 cm/sec) were compared by using Kolmogorov-Smirnov tests (a=0.05). The length- frequency figures were subdivided by three ecological groupings: clupeiforms (Clupeidae and Engraulidae); demersal taxa (predominantly Synodontidae and Blen- niidae); and scombrids and carangids, to further assess whether any changes in the size of fish collected over the current intervals were due to a particular group. All statistics were performed with SAS version 6.12 (SAS Institute, Cary, NO. Results Current speed Mean total CPUEs generally decreased with increasing current speed (Fig. 1). At current speeds s30 cm/sec, light trap catches were highly variable (CPUEs ranged from 0 to 138 fish per 10 min); however, CPUEs >20 fish per 10 min occurred only at these lower speeds. Although there were fewer samples at speeds >30 cm/sec, 440 Fishery Bulletin 103(2) 50 45 25 20 15 10 5 0 - 75 Clupeidae E 70 < > *- ^S -, (i) C 20 - CO — 15 - 111 ) Q_ 1(1 - O to 5 - a) i n - ]• Synodontidae • • • . " » 90- 80. 50 40 30 - 20 10 0 Engraulidae • • • • 0 -| . Carangld ae 8- 6 - • • 4 - • « • 2 - i-i. • 0 - ■ — f i i s • -•-•n — • — i — * •• i 25 n 20 15 10 5 0 Scombridae *%•_-. 1* *^- +■ -T- 0 10 20 30 40 50 60 70 80 90 0 10 20 30 40 50 60 70 80 90 Mean water current speed (cm/sec) Figure 2 Mean CPUE per sampling set (from within- and off-platform light traps) in relation to the mean cur- rent speed per sampling set for each of the dominant families collected. Data from all platforms were combined. Note changes in the scale of the .y-axis. CPUEs were mostly <5 fish per 10 min at these speeds. There was a significant linear relationship between log- transformed mean total CPUE data and mean current speed (log10(y+l) = -0.013.T + 1.302, r2=0.23; F=49.61, P<0.0001). Each of the dominant families collected by light traps showed a similar pattern of highest mean CPUEs at current speeds <30 cm/sec and relatively low mean CPUEs at higher current speeds (Fig. 2). Clupeidae, Engraulidae, and Blenniidae showed a slight trend of highest CPUEs at intermediate current speeds (10-30 cm/sec), whereas the other families generally had high- est CPUEs at the lowest speeds (<10 cm/sec). Synodon- tidae and Blenniidae were rarely collected at current speeds >40 cm/sec, and small numbers of Clupeidae, Engraulidae, Carangidae, and Scombridae were col- lected at speeds up to 80 cm/sec. As current speeds increased, light trap collections became limited to smaller size classes offish (Fig. 3). For the first three current intervals, i.e., 0-9, 10-19, and 20-29 cm/sec, a broad range of sizes were collected and the distributions had median lengths of 15-19 mm. However, beginning at the fourth current interval, 30-39 cm/sec, the size distributions shifted toward an increasingly greater proportion of the catch <10 mm in length. This trend was most pronounced at the two highest current intervals, 40-49 and >49 cm/sec, both of which had distributions with median lengths of 5 mm. The size distributions from the two highest cur- rent intervals were the only distributions that were not NOTE Lindquist and Shaw: Effects of current speed and turbidity on catches of larval and |uvenile fishes 441 0.15 0 1 - 0.05 0 0.15 o £ 01 3 O" 0) 0.15 I I Clupeiforms H Demersal taxa Current = 0-9 cm/sec 0.15 n = 1994 median=15 0.1 m 1 5 10 15 20 25 30 35 40+ Current=10-19 cm/sec n = 1009 median=19 tf+^fl 15 10 15 20 25 30 35 40+ Current = 20-29 cm/sec n = 1499 median=16 10 15 20 25 30 35 40+ 005 - Scombnds and carangids Current=30-39 cm/sec n=206 median =12 Jl ol^fllrMltorlH.Fffir^n.n 1 5 10 15 20 25 30 35 40+ 0.1 t 0.15 -1 l- Current=40-49 cm/sec n=78 median = 5 0 1 - - -, 0.05 -" 0 -=-Tn,,lfl = "iii 15 10 15 20 25 30 35 40+ 0.15 -, 0.1 0.05 0 1 Current = >49 cm/sec n=92 median = 5 nmTwHininn.Him n , ■ ■ ■ H 1 10 15 20 25 30 35 40+ Length (mm) Figure 3 Size distributions of fishes collected by light traps from all platforms at different current speed intervals. The total number offish collected [n) and the median length (mm) over each interval are included. Size distributions are further subdivided by three general ecological groupings: clupeiforms (Clupeidae and Engraulidae), demersal taxa (i.e., more substrate-oriented fishes such as synodontids and blenniids), and scombrids and carangids. significantly different from each other (P=0.11). The decrease in the frequency of fishes larger than 10 mm at the higher current intervals was not limited to any particular ecological grouping, i.e., pelagic fishes such as clupeiforms, scombrids, and carangids were as rare as demersal taxa. Turbidity Mean total CPUEs generally decreased with increasing turbidity (Fig. 4). Highest catches (CPUEs >50 fish per 10 min) predominantly occurred at turbidities below 1.0 NTU, whereas at higher turbidities catches were generally lower. There was a significant linear relation- ship between log-transformed mean total CPUE data and mean turbidity (log10(y+l) = -0.25.v + 1.48, r2=0.08; F=11.86, P=0.0007). The majority of the dominant families showed a simi- lar pattern of highest mean CPUEs at turbidities <1.0 NTU, and relatively low mean CPUEs at higher turbidi- ties (Fig. 5). Clupeidae, however, showed a pattern of high CPUEs at turbidities <0.5 NTU and between 1.0 and 2.0 NTU. Discussion Light trap catches of larval and juvenile fishes appeared to be negatively affected by increasing current speeds at platforms. This was expected because stronger currents may interfere with a fish's ability to swim to and enter a light trap (Doherty, 1987; Thorrold, 1992; Anderson et al., 2002). Doherty (1987) predicted that, for station- ary (anchored or tethered) light traps, catches should increase initially with current speed as more water is sampled, but then decrease as current speed inter- feres with catchability. Although mean total CPUEs clearly decreased with increasing current speed, they 442 Fishery Bulletin 103(2) 350 - | 300 - o >- 250 - • • • s. : • sz 200 - • * in W 150 - . * o ioo- l:'tm , S t M I t - . . ^* • • • 11*1-1 1.' I.J'Jl'1,' . -t* 0 H ( 112 3 Mean water turbidity (NTU) Figure 4 Mean total CPUE per sampling set (from within- and off-platform light traps) in relation to the mean turbidity per sampling set. Data from all platforms were combined. The line was calculated from the regression equation: log1(1(y+l) = -0.25.r + 1.48, r- = 0.08. Included in the analysis, but not shown in the plot, were three points from 583 to 878 CPUE between 0.2 to 0.5 NTU. did not appear to peak at some intermediate current level. These results, however, represented the total catch of all fishes, and the relationship between cur- rent speed and light trap catches may be more taxon specific (Doherty, 1987). When analyzed at the family level, a bell-shaped relationship may have occurred for Clupeidae, Engraulidae, and Blenniidae; however, the pattern was indistinct and there was generally little difference among families. The lack of any strong differences in the relationship between light trap CPUEs and current speed among the dominant families was unexpected, considering the potential differences in swimming abilities. Be- cause larvae and juveniles of demersal fishes are gener- ally believed to have lower swimming speeds (Blaxter, 1986), it was anticipated that catches of synodontids and blenniids would have been more negatively affected by increasing current speed than relatively stronger- swimming pelagic taxa (e.g., scombrids and carangids). Perhaps larvae of demersal taxa have greater swim- ming capabilities than previously considered, as has been recently found for certain settlement-stage larval reef fishes (sustained swimming speeds of 20-60 cm/ sec; Stobutzki and Bellwood, 1994; Leis and Carson-Ew- art, 1997). However, despite possible strong swimming abilities, few larval and juvenile demersal or pelagic fishes were collected at current speeds >40 cm/sec, and of these the majority were preflexion larvae that were undoubtedly passively entrained in the light trap. It is possible that the larvae and juveniles of taxa collected at platforms were unable to maintain the metabolic power required to swim against the stronger currents over extended distances from the light trap (Fisher and Bellwood, 2002). Currents may have interfered with the functioning of the light traps. Assuming that larval and juvenile fishes were able to swim against the stronger currents, their ingress into the light trap may have been impeded by turbulence created by the current flow around the trap. If turbulence occurred after some critical current speed, then this may explain the lower CPUEs beginning at around 30 cm/sec observed for each of the dominant families. Higher turbidity also appeared to have a negative ef- fect on light trap catches at platforms. Light trap catch efficiency should be greatly impaired by highly turbid waters because greater light attenuation would reduce the effective sampling radius of the trap. In addition, the phototactic response of larval and juvenile fishes may be lower at lower light intensities (Gehrke, 1994; Stearns et al., 1994). However, it is uncertain whether the relatively small range of turbidities (0.1-2.6 NTU) sampled during this study would result in a significant decrease in light trap catch efficiency, particularly given the intensity of the light source used (250,000 candle- power). The observed patterns may have been a reflec- tion of intrusions of turbid coastal and Mississippi River plume water at the platforms, during which light trap catches comprised large numbers of coastal clupeids and relatively few other taxa (Fig. 5). Although they were treated separately for the purpos- es of this study, the effects of current speed and turbid- ity also may have been interrelated. A positive relation- ship between turbidity and current speed was found for a limited data set where both variables were available (r-=0.28, P<0.0001). It is unlikely that this relationship was caused by the resuspension of benthic sediments, given the water depth at the platforms (20-230 m), but NOTE Lindquist and Shaw: Effects of current speed and turbidity on catches of larval and juvenile fishes 443 140 n 120 100 - 80 60 40 20 - 0 z> Q. o Clupeidae * i • tf ••*•» _••••■ •• ~ 60 c 0 50 1 40 I 30 20 10 - Synodontidae • • • 5 4 3 2 - 1 - 0 Carangidae 200 150 100 50 80 70 60 50 40 30 20 10 0 Engraulidae • ! • • i • • irfl I. »■!■.'. iifci mi * qt Blenniidae t »i*tm—*»t — ■ *}•+ 30 25 20 15 10 - 5 0 Scombndae t • • ■ i * : • , • ; i • jgfej Hi i. y.t.iJ.l%.\ ■•!»;. 0 1 2 3 0 1 2 3 Mean water turbidity(NTU) Figure 5 Mean CPUE per sampling set (from within- and off-platform light traps) in relation to the mean turbidity per sampling set for each of the dominant families collected. Data from all platforms were combined. Note changes in the scale of the y-axis. Not shown in the Engraulidae plot were three points from 551 to 606 CPUE between 0.2 and 0.5 NTU. particles may have been flushed from the platforms and their associated biofouling communities by currents. In a comparison of light trap catches between adjacent beach and rocky shore habitats, Hickford and Schiel (1999) attributed lower catches at the beach to lower water clarity caused by sediment resuspension by wave action. Therefore, high current speeds at platforms may have indirectly affected light trap catch efficiency by reducing water clarity. Results from this study have clear implications for future studies with light traps. At platforms, light trap CPUEs began to decline noticeably at current speeds of 30 cm/sec, and by 40 cm/sec catches of active swim- ming larval stages (i.e., all but preflexion stages) were rare. This finding suggests that, for comparison studies. estimates of relative abundance from light traps may be biased where there is considerable variation in current flow (Doherty, 1987; Anderson et al., 2002). Drifting traps may be used to avoid the confounding effect of differential water flow (Thorrold, 1992); however such a deployment method may not be applicable when habi- tats of interest are fixed (e.g., platforms, coral reefs). In such cases, the best course may be to not consider light trap samples at high current speeds (240 cm/sec). For turbidity, study results were not as clear; however, temporal or spatial variation in turbidity also would undoubtedly bias light trap results. Short of using light traps at times or locations of similar water clarity, an adjustable light source may be incorporated into light trap design so that equivalent light intensities, and 444 Fishery Bulletin 103(2) therefore sampling fields, can be maintained across a variety of water conditions. The alternative would be to standardize the volumes of water sampled by light traps; however, considering the suite of external factors that affect light trap efficiency, such attempts may be fruitless (Meekan et al., 2000). Acknowledgments We would like to thank A. Scarborough-Bull, C. Wilson, D. Stanley, J. Ditty, F. Hernandez Jr., J. Cope, J. Plun- ket, T. Farooqi, and all of those who assisted in the field and laboratory for their assistance and efforts during this research. We also thank Exxon USA, Inc., Mobil USA Exploration and Production, Inc., Santa Fe-Snyder Oil Corp., and Murphy Oil Corp. for access to their oil and gas platforms and logistical support, the crews of GC 18, GI 94B, ST 54G, MP 259A and VK 203 for their assistance and hospitality, and two anonymous review- ers for their helpful comments on this manuscript. This research was funded by the Minerals Management Ser- vice-Louisiana State University-Coastal Marine Insti- tute (contract no. 14-35-0001-30660-19961). Literature cited Anderson, T. W., C. T. Bartels, M. A. Hixon, E. Bartels, M. H. Carr, and J. M. Shenker. 2002. Current velocity and catch efficiency in sampling settlement-stage larvae of coral reef fishes. Fish. Bull. 100:404-413. Blaxter. J. H. S. 1986. Development of sense organs and behaviour of tele- ost larvae with special reference to feeding and predator avoidance. Trans. Am. Fish. Soc. 115:98-114. Brogan. M. W. 1994. Distribution and retention of larval fishes near reefs in the Gulf of California. Mar. Ecol. Prog. Ser. 115:1-13. Choat, J. H., P. J. Doherty, B. A. Kerrigan, and J. M. Leis. 1993. A comparison of towed nets, purse seine, and light- aggregation devices for sampling larvae and pelagic juve- niles of coral reef fishes. Fish. Bull. 91:195-209. Doherty, P. J. 1987. Light-traps: selective but useful devices for quan- tifying the distributions and abundances of larval fishes. Bull. Mar. Sci. 41:423-431. Fisher, R., and D. R. Bellwood. 2002. The influence of swimming speed on sus- tained swimming performance of late-stage reef fish larvae. Mar. Biol. 140:801-807. Fisher, R., D. R. Bellwood, and S. D. Job. 2000. Development of swimming abilities in reef fish larvae. Mar. Ecol. Prog. Ser. 202:163-173. Forristall, G. Z. 1996. Measurements of current blockage by the Bullwinkle platform. J. Atmos. Ocean. Technol. 13:1247-1266. Gehrke, P. C. 1994. Influence of light intensity and wavelength on phototactic behaviour of larval silver perch Bidyanus bidycmus and golden perch Mocquaria ambigua and the effectiveness of light traps. J. Fish Biol. 44:741-751. Gregory, R. S., and P. M. Powles. 1985. Chronology, distribution, and sizes of larval fish sampled by light traps in macrophytic Chemung Lake. Can. J. Zool. 63:2569-2577. Hernandez, F J., Jr., and R. F. Shaw. 2003. Comparison of plankton net and light trap meth- odologies for sampling larval and juvenile fishes at offshore petroleum platforms and a coastal jetty off Louisiana. In Fisheries, reefs and offshore develop- ment (D. R. Stanley and A. Scarborough-Bull, eds.), p. 15-38. Am. Fish. Soc. Symp. 36. Hickford, M. J. H., and D. R. Schiel. 1999. Evaluation of the performance of light traps for sampling fish larvae in inshore temperate waters. Mar. Ecol. Prog. Ser. 186:293-302. Johannes. R. E. 1978. Reproductive strategies of coastal marine fishes in the tropics. Environ. Biol. Fish. 3:65-84. Leis, J. M., and B. M. Carson-Ewart. 1997. In situ swimming speeds of the late pelagic larvae of some Indo-Pacific coral reef fishes. Mar. Ecol. Prog. Ser. 159:165-174. Meekan, M. G., P. J. Doherty, and L. White Jr. 2000. Recapture experiments show the low sampling efficiency of light traps. Bull. Mar. Sci. 67:875-885. Robertson, D. R., D. G. Green, and B. C. Victor. 1988. Temporal coupling of production and recruitment of larvae of a Caribbean reef fish. Ecology 69:370-381. Sponaugle, S., and R. K. Cowen. 1996. Nearshore patterns of coral reef fish larval supply to Barbados, West Indies. Mar. Ecol. Prog. Ser. 133:13-28. Stearns, D. E., G. J. Holt, R. B. Forward, and P. L. Pickering. 1994. Ontogeny of phototactic behavior in red drum larvae (Sciaenidae, Sciaenops ocellatusK Mar. Ecol. Prog. Ser. 104:1-11. Stobutzki, I. C, and D. R. Bellwood. 1994. An analysis of the sustained swimming abilities of presettlement and postsettlement coral-reef fishes. J. Exp. Mar. Biol. Ecol. 175:275-286. Thorrold, S. R. 1992. Evaluating the performance of light traps for sampling small fish and squid in open waters of the central Great Barrier Reef lagoon. Mar. Ecol. Prog. Ser. 89:277-285. Wilson, D. T. 2001. Patterns of replenishment of coral-reef fishes in the nearshore waters of the San Bias Archipelago, Carib- bean Panama. Mar. Biol. 139:735-753. 445 Can a change in the spawning pattern of Argentine hake (Merluccius hubbsi) affect its recruitment?* recruitment of this stock in different years between 1988 and 2001. Materials and methods Gustavo J. Macchi Conseio Nacional de Investigaciones Cientificas y Tecnicas (CONICET) Rivadavia 1917 1033 Buenos Aires. Argentina Present address: Instituto Nacional de Investigacion y Desarrollo Pesquero (INIDEP) Paseo Victoria Ocampo N° 1. CC. 175 Mar del Plata, 7600, Argentina E-mail address gmacchnaHnidepeduar Marcelo Pajaro Adrian Madirolas Instituto Nacional de Investigacion y Desarrollo Pesquero (INIDEP) Paseo Victoria Ocampo N° 1 CC. 175 Mar del Plata, 7600, Argentina Argentine hake (Merluccius hubbsi) inhabit waters of the Southwest Atlantic Ocean between 22° and 55°S, at depths ranging from 50 to 500 m (Cousseau and Perrota, 1998). This species has historically been among the more abundant fish resources in the Argentine Sea, where its biomass has ranged between one and two million metric tons annually since 1986 (Aubone et al., 2000). In this area, there are two identified fish- ing stocks, limited by the 41°S paral- lel. The southern group (Patagonian stock) is the more important with an abundance of about 85% of the total biomass estimated for this species in 1999 (Aubone et al., 2000). During the late 1990s, the spawning biomass of both stocks and their recruitment indices declined drastically, both of which were attributed to an increase in exploitation (Aubone et al., 2000). The Patagonian stock of Argentine hake spawns from November through March and peak spawning occurs in January (Macchi et al., 2004). This species is a batch spawner and has indeterminate annual fecundity, which is to say that unyolked oocytes continuously mature and are spawned throughout the reproductive season (Macchi and Pajaro, 2003). Thus, to estimate total fecundity, it is neces- sary to determinate the number of eggs released at one spawning (batch fecundity) and to estimate the num- ber of batches spawned in a repro- ductive season (spawning frequency). Macchi et al. (2004) estimated these parameters for the southern stock of M. hubbsi. They analyzed total egg production during the reproductive season and determined that the size composition of the spawning fraction influences the reproductive potential of the stock. Reproductive activity of the Pata- gonian hake historically has taken place mainly in coastal waters off the Chubut province at depths near 50 m, in the area known as Isla Escon- dida (43°30-44°S) (Ciechomski et al., 1983). Since 1997-98, a movement of reproductive hake to deeper waters and a decrease in fish density have been observed (Ehrlich et al.1). These changes, mainly in the location of the spawning area, may have affected the reproductive potential of this spe- cies, reducing the survival of eggs and larvae. If so, we would expect a negative effect on the number of juve- niles recruited after this period. In this note, we hypothesize that a change in spawning site for Patago- nian hake can affect species recruit- ment. We studied temporal changes in the location and density of spawn- ing aggregations, egg production, and Samples of M. hubbsi were collected from the area where the Patagonian stock is known to reproduce during four acoustic surveys in December 1988, 1993, 1996, and 2000 and during six trawl cruises carried out in January between the years 1996 and 2001. Acoustic surveys covered the Isla Escondida area between 43° and 45°S (Fig. 1). A SIMRAD EK400/QD echointegrator was used for the 1988 survey and a SIMRAD EK500 echo- sounder and BI500 postprocessing program were employed for subse- quent surveys. To avoid possible bi- ases due to the presence of fish in the near-bottom, acoustic transects were carried out at night when hake as- sume a more pelagic behavior. Trawl catches were carried out during the day, when fish are concentrated close to the bottom, and immediately af- ter each acoustic transect. Because trawls were intentionally biased to those areas of higher fish density, their positions were different be- tween 1988 and 2000. Nevertheless, the study area, transect design, and sampling effort were similar for all cruises covering the main spawning shoals. In January, information was col- lected from trawl surveys to assess the Patagonian stock of juvenile hake between 1996 and 2001. These cruises covered a wide area between 43° and 47°S that included a section ' Contribution 1357 from the Instituto Nacional de Investigacion y Desarrollo Pesquero, Mar del Plata, Argentina. Ehrlich, M. D., P. Martos, A. Madirolas, and R. P. Sanchez. 2000. Causes of spawning pattern variability of anchovy and hake on the Patagonian shelf. ICES CM 2000/N:06. Manuscript submitted 2 July 200.3 to the Scientific Editor's Office. Manuscript approved for publication 20 December 2004 by the Scientific Editor. Fish. Bull. 103:445-452 (2005). 446 Fishery Bulletin 103(2) 62 Longitude °W Figure 1 Distribution and density (sA) of Argentine hake iMerluccius hubbsi) estimated from acoustic surveys carried out during December (1988, 1993, 1996, and 2000) in the Isla Escondida area. The size of the symbols is proportional to the percentage of spawning females (with hydrated oocytes). Vertical shaded scale represents scattering coefficient values (sA), where 7.14 sA units = 1 t/nautical mile2. of the main spawning ground of hake. Thus, to ana- lyze spawning individuals we used only data from 33 fish stations located offshore within the spawning area between 43°30' and 46°S (Fig. 2). Trawl station sites were the same during all cruises. In January of 1996 and 2001 additional information from catches obtained inshore near Isla Escondida was analyzed (Fig. 2). Argentine hake were collected with a bottom net with a mouth width of about 20 m, a height of about 4 m, and with 20-mm mesh at the inner cover of the codend. Total length (TL) in cm, total weight (TW) in g, and sex were recorded for each fish sampled; for females a subsample was randomly selected from different trawl stations (Table 1) and the maturity stage was deter- mined for each individual. A macroscopic maturity key of five stages designed for biological studies was em- ployed: 1) immature; 2) developing and partially spent; 3) spawning (gravid and running); 4) spent; and 5) resting (Macchi and Pajaro, 2003). This scale was vali- dated by the histological analysis of ovaries collected during December 2000 and January 2001 (Macchi et al., 2004). Females were classified as reproductively active or inactive, according to the presence of yolked oocytes and atresia stages following the criteria of Hunter et al. (1992). When we consider the codes used in the visual assessment of maturity, stages 2 and 3 corresponded to active females, which were capable of spawning at the time of capture or in the near future (Hunter et al., 1992). Abundance of active females was estimated from data collected during each survey. Information obtained from sampling the trawl catch was expanded to obtain esti- mates of the number of individuals per length class, fol- lowing the method described by Macchi et al. (2004). During December, information from acoustic surveys was used to assign a different weight to each trawl station, based on the relative density and size of the school targeted by the trawl. The transect segment that contained a given trawl was determined and the average value of the water column scattering coefficient NOTE Macchi et al.: Effect of the spawning pattern of Merlucaus hubbsi on recruitment 447 68 67 66 1? 43- 44 45 4fr 1996 Argentina 63 62 - 1 — i — i — i — i — i — | — i — i — i — i — i — | — i — i — i — i — r 67 66 65 64 63 62 62 68 Longitude "W ; 2001 i i.i i -42 43 Argentina / o 44 > o . 4- 45 / ■ . i \'\ i 1 1 1 1 i i i i i ^46 67 66 65 64 63 62 Figure 2 Distribution and density of Argentine hake iMerluccius hubbsi) estimated from trawl surveys carried out during January (1996-2001) in the offshore area. The size of the symbols is proportional to the percentage of spawning females (with hydrated oocytes). The square shows the Isla Escondida area. Vertical shaded scale represent biomass in t/nautical mile2. (sA) was calculated and weighed by the corresponding number of acoustic observations. The number of active females for each survey was estimated by multiplying the number of hake within each length class by the proportion of females and the proportion of active females for that length class ob- tained for that survey (Marshall et al., 1998). The sum of values estimated across the size range was an index of the number of reproductive females in the sampled area during that survey. Egg production of the Patagonian hake in Decem- ber during the period 1988-2000 and in January from 1996 to 2001 was based on estimates of three variables: the abundance of active females per length class, the 448 Fishery Bulletin 103(2) Table 1 Number of Argentine hake (Merluccius hubbsi) s ampled during research surveys carrie d out in the north Patagonian area in December and January, bet ween 1988 and 2001. Number of Number of females Period Number of trawls ndividuals sampled subsampled December 1988 18 9527 2054 1993 6 2156 1060 1996 9 1563 690 2000 12 4390 708 January 1996 38 17,715 1509 1997 33 12,687 842 1998 33 15,804 1092 1999 33 14,987 817 2000 33 14,389 958 2001 37 17,944 856 batch-fecundity-size relationship, and spawning fre- quency. The batch fecundity-total-length relationship and the spawning frequency values used for December (1988-2000) and for January (1996-2001) were those estimated in December 2000 and January 2001, re- spectively (Macchi et al., 2004). We assumed that these values were applicable to all previous years, because in general, annual differences of these variables were not significant for hake females of the same length range (Macchi et al., 2004). Egg production by length for each month was esti- mated by multiplying the number of active females in each length class by the batch fecundity corresponding to that length class and by the number of spawnings es- timated for each month. The sum of the egg production values estimated across the size range was the total number of eggs produced in the sampled area during each month (December or January) in different years. To analyze the relationship between egg production and recruitment, estimates of the relative abundance at age 1 (number of individuals per trawl hour) of Argen- tine hake were used as a recruitment index. These data were obtained from samples to assess hake juveniles collected from the whole area covered during the cruises carried out in January 1997-2001. In 2002, this index was estimated with samples collected in the same area, but in a different month (March) (GEM, unpubl. data2). The number of age-1 individuals in year t+1 was the recruitment index corresponding to the year t. Results (with hydrated oocytes) in the Isla Escondida area during December 1988, 1993, 1996, and 2000. A decline in hake abundance from 1988 to 2000 was observed — in particu- lar, a drastic decrease in 2000, when the mean density value (14.6 t/nautical mile2) was thirty times less than that estimated in 1988 (469.2 t/nautical mile2). During December 1988-96, spawning females were mainly located in the northern area (between 43° and 44°S) inshore at depths lower than 50 m. In 2000 reproduc- tive activity was concentrated at the same latitude as in previous years, but offshore (Fig. 1). In January 1996 the highest densities of M. hubbsi and the spawning females of this species were located in the Isla Escondida area (Fig. 2). Between 1997 and 2000 we did not obtain data from this zone, but the increase in the proportion of spawning hake in deep waters observed since 1998 indicates a spatial change in the reproductive area. During January 2000 and 2001, in addition to the increase of reproductive females offshore, the abundance of hake was higher than that estimated previously for the same area (Fig. 2). In Jan- uary 2001, trawl stations located near Isla Escondida showed very low values of hake density, in contrast to that observed offshore. This contrast could be attributed to the movement of individuals from the traditional spawning area near the coast to deeper water. Egg production Egg production estimated for December in the Isla Escondida area showed a considerable decrease from 1988 to 2000 (Fig. 3). The number of eggs produced Abundance of hake and location of spawning females Figure 1 shows the acoustic densities estimated for Argentine hake and the distribution of spawning females - GEM (Grupo de Evaluacibn Merluza). 2002. Evaluacion del estado del recurso merluza (Merluccius hubbsi) al sur de 41° S, ano 2002. Unpubl report. INIDEP, CC. 175, Mar del Plata (7600), Argentina. NOTE Macchi et al.: Effect of the spawning pattern of Merlucaus hubbsi on recruitment 449 1600 2 1200 800 400 - XZL 2000 1600 1200 800 400 1988 1993 1996 2000 Figure 3 Egg production of Argentine hake (Merluccius. hubbsi) estimated for December (1988. 1993, 1996, and 2000) in the Isla Escondida area (bars), and production by unit-weight of reproductively active female (line) for the same month. per unit of weight (kg of active females) declined from 1988 to 1996, and to a value of around 1700 eggs/kg in the last year (Fig. 3). During December 2000, however, relative egg production increased to 2000 eggs/kg, which can be attributed to the effect of a higher proportion of larger females in reproductive activity. In fact, when the percentage of eggs produced by length class was analyzed, the distribution obtained for December 2000 was different from that for 1988, 1993, and 1996 (Fig. 4). During the earlier years, production mainly depended on young females (<50 cm TL), whereas in December 2000 most of the eggs produced (about 70%) where spawned by females larger than 50 cm TL. Egg production estimated for the offshore area in Jan- uary increased from 1996 to 2001 (Fig. 5), in contrast to that observed during December in shallow water near Isla Escondida. The number of eggs produced per unit of weight of active females was similar in 1996 and 1997 (about 1600 eggs/kg), but increased in 1998-2001 to about 1800 eggs/kg. This increase was similar to that observed for December 2000, which was attributed to the higher proportion of larger females within the spawning fraction of hake. In fact, percentage-distribu- tion of eggs produced by length class showed a change beginning in 1998 (Fig. 6). In 1996 and 1997, 70% of the eggs were produced by young females (<50 cm TL), but subsequent production of old females increased to 60% in 1998-99 and to 70% in 2000-01. 40 - ^^1988 o a -»— 1993 ! 3°- /V n^ // T4 T3 t 20- \ if \ | 10- al // \^^\ 0) JJ \bs=53£liS^~X 20 30 40 50 60 70 80 90 100 Total length (cm) Figure 4 Relative egg production {%) by length class estimated for Argentine hake (Merluccius hubbsi) from December 1988, 1993, 1996, and 2000. 800 - - 2000 7 o 700 - j** ' o S~ 600 - o ~ 500 - o " 400 - ion per kg of o o o o CD CM 2 300- D3 ■ 800 £ < S 200 - CD* -400 | 100 - n CD n O 1996 1997 1998 1999 2000 2001 Years Figure 5 Egg production of Argentine hake [Merluecius hubbsi) estimated for January (1996-2001) in the offshore area (bars), and production by unit-weight of reproductively active female (line) for the same month. from the offshore area was used; thus, the number of eggs estimated was a fraction of that produced by all spawning females in January. However, the increase in egg production observed offshore for the parental stock in 2000 and 2001 was coincident with higher values of age-1 recruitment estimated one year later during 2001 and 2002, respectively (Table 2). Recruitment Relative abundance data for hake at age 1 (year t+1) in the north Patagonian area were contrasted with the egg production obtained in January from the previous year (t). To estimate egg production, only information Discussion The spatial pattern of M. hubbsi spawning aggrega- tions inshore and offshore of the north Patagonian area between 1988 and 2001 has changed since 1998. This 450 Fishery Bulletin 103(2) Table 2 Egg production estimates for Argentine hake (Merluccius hubbsi) for January cruises 1 1996-2001 ) taken offshore of the north Patagonian area, and indices of abundance at age 1 corresponding to these annual classes. Year Egg production (1012) Index of age-1 hakes (individuals per trawl hour) 1996 116.625 1997 81.774 347 1998 270. 512 438 1999 228.020 133 2000 627.484 250 2001 572.485 1367 2002 2444 change was characterized by a decrease in density on shoals and a movement of spawning females to deeper water, withand a more scattered distribution than in the early 1990s. Our results confirm previous observations reported by Ehrlich et al.1, who analyzed ichtyoplankton samples collected from 1973 to 1999, in the traditional spawning area of Isla Escondida. These authors did not observe significant environmental anomalies that might have affected the spawning of hake and associated the change with the high levels in fishing exploitation in the 1990s. These shifts in the pattern of reproduction led to the following question: "How does the movement of the center of spawning affect the recruitment of Patagonian hake?" — given that different environmental conditions could be present in the new spawning area. Our analyses show that the abundance of active fe- males offshore of the north Patagonian area increased from 1998 to 2001, coinciding with a significant de- crease in hake biomass in the shallow waters of Isla Escondida. During these years, demographic changes in the offshore area were characterized by an increase of larger females (>50 cm TL) compared to previous years. The increase in proportion of older individuals in spawning condition may result in a greater contribu- tion to egg production because of the higher fecundity produced by larger females (Mairteinsdottir and Thora- rinsson, 1998). In fact, egg production estimated for the offshore Patagonian hake during January showed an in- crease since 1998, with the highest values in 2000 and 2001 (400% more than those estimated in 1996-97). A high proportion (70%) of these eggs were spawned by females larger than 50 cm TL (s5-year old, Otero et al., 1986), whereas in January 1996 and 1997 eggs were mainly produced by young females. Because of the displacement of active females to deep water, the offshore north Patagonian area from 43°30' to 45°S and between 50 m and 100 m depths was con- sidered an important section of the spawning ground for Patagonian hake after 1998. The comparison between the January 1996 and 2001 surveys, in which inshore 1996 1997 1998 -1999 -2000 -2001 40 50 60 70 80 Total length (cm) 90 100 Figure 6 Relative egg production (%) by length class estimated for Argentine hake (Merluccius hubbsi) in January from 1996 to 2001. and offshore samples of the north Patagonian area were collected, demonstrated this change. In January 1996, spawning of Patagonian hake was concentrated inshore (Isla Escondida), whereas in January 2001 reproduction of this stock took place mainly offshore (Fig. 2). For this reason, the offshore egg production value obtained after 1998 was considered a representative index of the spawning area. Relative abundance of hake at age 1 (number of indi- viduals/hour) in the north-Patagonian area, showed a decline from 1996 to 2000 and an increase in 2001 and 2002, reaching the highest values of the study period. The recruitment index obtained for 2002 (2444 individuals/h) was about twice that estimated for 2001 (1367 individu- als/h). According to Santos et al.,3 it is possible that this value has been overestimated, because it was determined 3 Santos, B. A., E. B. Louge, and R. Castrucci. 2003. Estu- dio de las variaciones conjuntas de la temperatura y de la salinidad del area de cria de la merluza con los indices de abundancia de los grupos de edad 0. 1 y 2. (enero 1995-enero 2002). Tech. Rep. 10/03. 6 p. INIDEP, CC. Plata (7600), Argentina. 175. Mar del NOTE Macchi et al.: Effect of the spawning pattern of Merluccius hubbsi on recruitment 451 from samples collected two months later (March) than those during 1996-2001. These authors suggested that the spatial distribution or catchability of juvenile hake could have changed from January to March, resulting in a greater abundance index during 2002. The higher recruitment levels observed for Patago- nian hake during 2001 and 2002 were coincident with higher indices of egg production estimated offshore in January during the two previous years (2000 and 2001). Therefore, in principle we concluded that the change in spatial location of spawners in the Patagonian stock did not appear to negatively affect the recruitment of this species. The next question to be answered is: "Why were recruitment indices in the early 2000s higher than in previous years?" Several authors have analyzed the spawner-recruit relationship in different species and have concluded that recruitment is often positively correlated with spawner biomass estimated from virtual population analysis (VPA) (Myers and Barrowman, 1996). In the case of Patagonian hake, the increase in abundance at age 1 observed in 2001 and 2002 was not associated with higher values of the VPA-based spawner biomass in previous years (GEM, unpubl. data2). Thus, envi- ronmental and ecological factors affecting prerecruit mortality should be considered, mainly in association with a no-fishing area implemented in 1997. Moreover, the demographic composition and the nutritional state of spawning females (maternal effect) are other factors that have been related to recruitment levels (Trippel et al., 1997; Kjesbu et al., 1998; Cardinale and Arrhenius, 2000). Analysis of hydrographic characteristics from the north-Patagonian waters in the 1980s and 1990s indi- cated that the Patagonian shelf, including the Isla Es- condida area, is a relatively stable environment (Erhlich et al.2). On the other hand, analysis of temperature and salinity data collected from 1995 to 2002 in the nursery area of the Patagonian stock (San Jorge Gulf), showed that higher values of salinity and temperature during the time of hatching were associated with higher indices of abundance at age 1, one year later (Santos et al.3). The high proportion of larger females in the offshore area mainly in 2000 and 2001 may have affected the quality as well as the quantity of hake progeny. In gen- eral, older females produce larger eggs and larger lar- vae with higher rates of survival, in combination with more egg batches over a longer spawning season (Kjesbu et al., 1996; Trippel, 1998). Previous reports showed that M. hubbsi older than 5-years have a longer spawn- ing season (Macchi et al., 2004) and produce heavier eggs than young females do (Pajaro et al.4). Thus, an increase in the proportion of older spawning females in 4 Pajaro, M, E. Louge, G. J. Macchi, N. Radovani, and L. Rivas. 2002. Calidad de los ovocitos de la poblacion patagonica de merluza {Merluccius hubbsi) durante la epoca de puesta estival. Tech. Rep. 55/02, 13 p. INIDEP, CC. 175, Mar del Plata (7600), Argentina. the stock may result in improved recruitment, as has been reported for other species (Mairteinsdottir and Thorarinsson, 1998). The fishing regulation for Patagonian hake imple- mented in the late 1990s mainly affected bottom trawl- ers and the factory freezer fleet, which applied greater fishing effort in the north Patagonian area during the 1990s. It is possible that this decline in harvesting pressure by trawlers on Patagonian hake after 1997 in- fluenced the reproductive success of this species. Stress can have a negative impact on fish reproduction (Camp- bell et al, 1994; Clearwater and Pankhurst, 1997). The potential effects of trawl avoidance can affect the repro- ductive physiology and behavior during spawning, which could lead to the production of fewer viable juveniles (Morgan et al., 1999). Finally, other factors, such as predation and feeding conditions within the new spawning ground of Patago- nian hake, can affect survival of the early life stages. In addition, future studies should include a comparison between the inshore and offshore waters of the north- Patagonian area with respect to the abundance of jel- lyfish (i.e., Medusae and Ctenophora), which are known to be major predators of fish eggs and larvae (Bailey, 1984; Fancett, 1988). Acknowledgments We thank Jorge Hansen for assistance with the method used to estimate fish abundance. We would also like to thank Hector Cordo for reading and making suggestions to improve the manuscript. Literature cited Aubone, A., S. Bezzi, R. Castrucci, C. Dato, P. Ibanez, G. Irusta, M. Perez, M. Renzi, B. Santos, N. Scarlato, M. Simonazzi, L. Tringali, and F. Villarino. 2000. Merluza {Merluccius hubbsi). In Sintesis del estado de las pesquerias maritimas argentinas y de la Cuenca del Plata. Anos 1997-1998, con una actualizacion de 1999 (S. Bezzi, R. Akselman and E. Boschi, eds.), p. 29-40. INIDEP, Mar del Plata, Argentina. Bailey. K. M. 1984. Comparison of laboratory rates of predation on five species of marine fish larvae by three planktonic inver- tebrates. Effects of larval size on vulnerability. Mar. Biol. 79:303-309 Campbell, P. M„ T. G. Pottinger, and J. P. Sumpter. 1994. Preliminary evidence that chronic confinement stress reduces the quality of gametes produced by brown and rainbow trout. Aquaculture 120:151-169. Cardinale, M., and F. Arrhenius. 2000. The influence of stock structure and environmen- tal conditions on the recruitment process of Baltic cod estimated using a generalized additive model. Can. J. Fish. Aquat. Sci. 57: 2402-2409. Ciechomski, J. D„ R. P. Sanchez, C. A. Lasta, and M. D. Ehrlich. 1983. Distribucion de huevos y larvas de anchoita (En- 452 Fishery Bulletin 103(2) graulis anchoita) y de merluza iMerluceius hubbsi), evaluacion de sus efectivos desovantes y analisis de los metodos empleados. Contrib. Inst. Nac. Invest. Desarr. Pesq. (Mar del Plata) 432:3-37. Clearwater, S. J., and N. W. Pankhurst. 1997. The response to capture and confinement stress of plasma Cortisol, plasma sex steroids and vitellogenic oocytes in the marine teleost, red gurnard. J. Fish Biol. 50:429-441. Cousseau, M. B., and R. G. Perrota. 1998. Peces marinos de Argentina. Biologia distribucidn y pesca, 163 p. INIDEP, Mar del Plata, Argentina. Fancett, M. S. 1988. Diet and prey selectivity of scyphomedusae from Port Phillip Bay, Australia. Mar. Biol. 98:503-509. Hunter, J. R., B. J. Macewicz, N. C. H. Lo, and C. A. Kimbrell. 1992. Fecundity, spawning, and maturity of female Dover sole Microstomas 13 pacificus, with an evaluation of assumptions and precision. Fish. Bull. 90:101-128. Kjesbu, O. S„ P. Solemdal, P. Bratlan, and M. Fonn. 1996. Variation in annual egg production in individual captive Atlantic cod iGadus morhua). Can. J. Fish. Aquat. Sci. 53:610-620. Kjesbu. O. S„ P. R. Witthames , P. Solemdal, and M. Greer Walker. 1998. Temporal variations in the fecundity of Arcto- Norwegian cod iGadus morhua) in response to nat- ural changes in food and temperature. J. Sea Res. 40:303-321. Macchi G. J., and M. Pajaro. 2003. Comparative reproductive biology of some com- mercial marine fishes from Argentina. Fisken Og Havet 12:69-77. Macchi, G. J., M. Pajaro, and M. Ehrlich. 2004. Seasonal egg production pattern of the Patagonian stock of Argentine hake iMerluceius hubbsi). Fish. Res. 67:25-38. Mairteinsdottir, G., and K. Thorarinsson. 1998. Improving the stock-recruitment relationship in Ice- landic cod (Gadus morhua L.) by including age diversity of spawners. Can. J. Fish. Aquat. Sci. 55:1372-1377. Marshall, T. O. S. Kjesbu. N. A. Yaragina, P. Solemdal, and O. Ulltang. 1998. Is spawner biomass a sensitive measure of the reproductive and recruitment potential of Northeast Arctic cod? Can. J. Fish. Aquat. Sci. 55:1766-1783. Morgan, M. J., C. E. Wilson, and L. W. Crim. 1999. The effect of stress on reproduction in Atlantic cod. J. Fish Biol. 54 (31:477-488. Myers, R. A., and N. J. Barrowman. 1996. Is fish recruitment related to spawner abun- dance? Fish. Bull. 94:707-724. Otero, H. O., M. S. Giangiobe, and M. A. Renzi. 1986. Aspectos de la estructura de poblacion de la merluza iMerluceius hubbsi). II Distribucion de tallas y edades. Estadios sexuales. Variaciones estacionales. Publ. Com. Tec. Mix. Fr. Mar. 1 (11:147-179. Trippel, E. A. 1998. Egg size and viability and seasonal offspring pro- duction of young Atlantic cod. Trans. Am. Fish. Soc. 127:339-359. Trippel, E. A., O. S. Kjesbu, and P. Solemdal. 1997. Effects of adult age and size structure on repro- ductive output in marine fishes. In Early life history and recruitment in fish populations (R. C. Chambers and E. A. Trippel, eds.), p. 31-62. Chapman & Hall, New York, NY. 453 Feeding habits of the dwarf weakfish (Cynoscion nannus) off the coasts of Jalisco and Colima, Mexico Alma R. Raymundo-Huizar Centro Universitano de la Costa, Departamento de Ciencias Universidad de Guadalaiara Av. Universidad 203 Puerto Vallarta, Jalisco, CP 48280 Mexico Horacio Perez-Espana Centro de Ecologia y Pesquerias Universidad Veracruzana Dr, Castelazo s/n. Xalapa Veracruz, CP 91 190 Mexico Maite Mascaro Laboratono de Ecologia y Conducta Unidad Academica Sisal Universidad Nacional Autonoma de Mexico Sisal, Yucatan, CP 97355 Mexico Xavier Chiappa-Carrara Unidad de Investigacion en Ecologia Marina, FES-Z Mexico, DF, CP 09230 Mexico Present address: Unidad Academica Sisal Universidad Nacional Autonoma de Mexico Sisal, Yucatan, CP 97355 Mexico E-mail address (for X. Chiappa-Carrara, contact author) chiappaig'servidor unam mx Sciaenids from the Pacific coast of Mexico are used as a second-class fish species for human consumption (Aguilar-Palomino et al., 1996). The dwarf weakfish (Cynoscion nannus) (Castro-Aguirre and Arvizu-Mar- tinez, 1976) is often caught as bycatch in the shrimp fishery but, because of its small size (<27 cm TL, total length), it is not considered a valuable resource. This species can be found in great numbers in waters between 100 and 812 m (Allen and Robert- son, 1994; Fischer et al., 1995) asso- ciated with the soft-bottom regions off the coast of Jalisco and Colima (Gonzalez-Sanson et al., 1997). Previous studies of the trophic bi- ology of the Sciaenidae (Chao and Musik, 1977; Campos and Corrales, 1986; Chao, 1995; Pelaez-Rodriguez, 1996; Cruz-Escalona. 1998; Lucena et al., 2000) have shown that they feed on a variety of small fish and benthic invertebrates (Allen and Rob- ertson, 1994). However, there are few studies concerning the feeding habits of C. nannus, and its dietary prefer- ences are not known. Considering its abundance, C. nannus must play an important role in the trophic rela- tionships of soft-bottom ecosystems in this region. Most studies describing the feeding habits of fish have used the normal- ized version of the breadth niche in- dex proposed by Levins (1968). This index is based both on the number of food resources and on the proportion of prey used by a species. The appro- priate distribution function for this index ensures sample independence among prey found in any particular stomach. Distribution functions based either on the number or the relative biomass or volume of dietary items do not ensure such independence, given that all items found in any particular stomach are statistically associated (Hurlbert, 1984). Therefore, neither the number nor the relative biomass or volume of dietary items should be used to calculate the Levins index. The only distribution function that ensures statistical independence is that which is based on the proportion of stomachs in which a certain food resource is found (Krebs, 1999). Considering the ecological impor- tance of studying the feeding habits of this abundant fish species, we ex- amined trophic breadth variations (temporally and ontogenetically) of C. nannus. When attempting to cor- rectly apply the Levins index, we used the distribution function of prey that ensures statistical independence among sampling units. Materials and methods The sampling area was located in the central region of the continental shelf off the Pacific coast of Mexico, where the mouth of the river Cuitz- mala, in Punta Farallon, Jalisco (19°22'N, 105°01'W), is the northern limit, and Cuyutlan, Colima (18:55'N, 104°08'W), is the southern limit. Sam- ples of C. nannus were collected on a monthly basis from January to Decem- ber 1996 (except February, August, and September) on the research vessel BIP V, equipped with a trawl net with a pair of codends. Sampling was car- ried out over seven transects perpen- dicular to the coast, each comprising four bathymetric strata: 20, 40, 60, and 80 m mean depth. Fish were individually identified, measured (TL, ±1 mm), and the total weight of each fish was recorded to the nearest 0.1 g. The stomachs of individual fish were dissected and preserved in 10% neutralized forma- lin. Stomach contents were analyzed Manuscript submitted 16 May 2003 to the Scientific Editor's Office. Manuscript approved for publication 20 December 2004 by the Scientific Editor. Fish. Bull. 103:453-460 (2005 1. 454 Fishery Bulletin 103(2) with a stereoscopic microscope and dietary items were identified to the lowest taxonomic level possible by using specialized keys. Garth (1958), Rodriguez de la Cruz (1987), Hendrickx and Salgado-Barragan (1991), and Hendrickx (1996), were consulted for crustacean iden- tification, whereas Jordan and Evermann (1896-1900), Castro-Aguirre (1978), Allen and Robertson (1994), Thomson et al. (2000), and FAO guides were used for fish identification (Fischer et al., 1995). Both the number of individuals and weight of each dietary category were quantified, and mean proportions in terms of number (%7VV ) and biomass C7rWj ) were cal- culated according to Tirasin and Jorgensen (1999): IX %X, = j=i ■xlOO, where X = the number or weight of each taxa i in the jth stomach; and k = the number of dietary components found in all stomachs analyzed, /;,. The percent frequency of occurrence of each component was also obtained (c7cF). Finally, the index of relative importance for each dietary category was calculated (IRI, Pinkas et al., 1971; Rosecchi and Nouaze, 1987): IRI, =(%Ni+%Pi)x%Fi. Relative importance index values were expressed as a percentage of the total items analyzed (Cortes, 1997) and results were graphically represented as a rectangle of base %F and height 7c N + 7cW. Variance analysis was applied on transformed W'= sin_1(VW)l gravimetric proportions of the dietary components (Zar, 1999) to evaluate both monthly and ontogenetic variations in the feeding habits of C. nan- nus. The number [q,=l+3.322(Log]0n)] and width of size classes (w=RTL/q) were considered for analysis, where ;; is the sample size and RTL=TLmRX-TLmm. For the analysis of trophic niche breadth, the nor- malized version of the index proposed by Levins (1968) was used. This index combines both the number of prey resources used (k) (i.e., the trophic spectrum) and the relative frequency with which each prey resource is consumed (J). This represents the distribution function of prey proportions in diet (Hespenheide 1975; Hurlbert, 1978): (n, \ Ba- k-1 Because the ensemble of prey found in any given stomach does not constitute independent samples (Hurl- bert, 1984), pf was calculated as the proportion of indi- vidual fish (iV*) that consumed a certain food resource in relation to the number of resources used by the total number of fish: N YTn sothatI^ = 1- Ba values range between 0 and 1. Zero values indi- cate that fish feed on only one prey type, representing the minimum diet breadth and high feeding special- ization. Unity values, on the other hand, indicate that the species consumed all k food resources in the same proportion (p; = l/&), representing no selection among prey types and the widest possible trophic niche (Gibson and Ezzi, 1987; Labropoulou and Eleftheriou, 1997). Ba values were calculated on the basis of matrix resources (Colwell and Futuyma, 1971) both for each month and for each size class. The percentage similarity measure (/?) between size classes q' and q" (Renkonen, 1938; Schoener, 1970; Hurlbert, 1978) was calculated as V-=l"2 H\pJq--p,A where p is the proportion of individual fish in each size class that consumed a certain food resource, calculated over the total number of stomachs per size class. Confidence intervals (CI95%) of Ba were obtained by means of the bootstrap method (Mueller and Altenberg, 1985; Efron and Tibshirani, 1986) by considering two thousand resamplings of the data (Hamilton, 1991). Results The 311 Cynoscion nannus examined ranged from 7.5 to 20.6 cm TL. Food was found in 287 (92%, rang- ing from 85% to 98% among size classes) stomachs. The trophic spectrum of C. nannus is composed of 29 dietary items (Table 1), which were classified into four general categories: penaeid shrimp, fish, stomatopods. and cephalopods. Penaeid shrimp constituted the principal dietary cat- egory of C. nannus (A^ = 82.5%, Wf = 35.4%; Fj=43.9%, 77?/j = 74.6%; Fig. 1), of which juvenile stages were the most frequent (Ff=23.4%). Fish were the second most important category (7Vf = 6.5%, Wf = 36.5%, Fv = 37.7%, 7ff/j = 14.5% ), followed by stomatopods of the Squilla genus (iV—5.8%, W==8.6%, FI=25.5%, 7/^ = 6.6%). The cephalopod Loliopsis diomedae was the last category in order of importance (JV==1.0%, Wf = 12.4%, 7^ = 4.2%, IRI; = 1.87c). Overall, significant differences in diet were found between individuals of different size classes (F=1.03; P<0.05). Values of the percentage similarity of diet (i?) between size classes were, in general, <50% (Table 2). R -values were relatively high only among size classes 2 NOTE Raymundo-Huizar et al.: Feeding habits of Cynoscton nannus 455 Table 1 Composition of the trophic spectrum of Cynoscion nannus (7.5 cm 5 TL s20.6 cm; n =287) from the coast of Jalisco and Colima (mean percentage by weight [g; %W], frequency of occurrence [%F'_ , number [%N], and index of relative importance [%/ft/| of prey). Dietary categories 7c W 9cF %N %IRI Cephalopods Loliopsis diomedae 12.4 4.2 1.0 1.8 Remains 0.1 0.4 0.1 0.0 Stomatopods Squilla sp. 5.9 19.7 4.0 6.4 S. panamensis 1.5 1.7 0.8 0.1 S. hancocki 0.5 1.7 0.4 0.0 S. mantoidea 0.7 2.5 0.7 0.1 Penaeid shrimps Solenoeera sp. 7.6 10.9 4.3 4.2 S. florea 2.2 1.3 0.6 0.1 S. mutator 3.8 2.9 4.6 0.8 Traehypenaeus brevisuturae 3.5 5.4 2.3 1.0 Juvenile shrimps 18.2 23.4 70.8 68.4 Other crustaceans Carideans 1.3 4.2 0.9 0.3 Panulirus sp. larvae 0.1 0.4 0.1 0.0 Other crustacean larvae 0.1 0.4 0.3 0.0 Euphausiids 0.5 1.3 2.3 0.1 Microcrustaceans 0.2 0.4 0.1 0.0 Unidentified remains 4.2 13.8 — 1.9 Fish Cynoscion nannus 2.1 0.8 0.4 0.1 Cherublemma emmelas 1.3 1.7 1.0 0.1 Polydactylus opercularis 1.6 1.3 0.4 0.1 Ophidium sp. 0.7 0.4 0.8 0.0 Monolene sp. 0.9 0.4 0.1 0.0 Symphurus sp. 0.1 0.4 0.1 0.0 Bregmaceros bathymaster 1.6 2.5 0.7 0.2 Anguilliformes 1.2 2.1 0.5 0.1 Leptocephalus larvae 8.9 4.2 1.1 1.4 Other fish larvae 1.4 1.7 1.0 0.1 Unidentified remains 16.6 22.2 0.3 12.3 Anelids Polychaeta 0.62 1.26 0.38 0.0 Table 2 Percentage similarity values iR) of the diet between size classes (cm, TL) of Cyn oscion nannus (ra=287) from the coast of Jalisco and Colima. Size class (cm) 7.0-8.9 9.0-10.9 11.0-12.9 13.0-14.9 15.0-16.9 17.0-18.9 Size class (cm) 9.0-10.9 37.0 11.0-12.9 44.2 54.2 13.0-14.9 33.5 65.1 51.1 15.0-16.9 31.0 47.6 51.4 64.1 17.0-18.9 12.5 31.6 39.6 38.0 40.4 19.0-20.9 29.3 45.8 44.3 47.9 40.1 21.6 456 Fishery Bulletin 103(2) 100- 80- % Weight o o 1 3 20- 0- 20- 5 — W-^ 2 II I 4 6 1 40- E % 60- 80- 0 20 40 60 80 100 120 140 160 Cumulative frequency of occurence (%) Figure 1 Graphical representation of the percentage values of the index of relative importance (IRI) of the main dietary components found in the stomachs of Cynoscion nannus (n=287). 1: penaeid shrimps; 2: stomatopods; 3: fish; 4: crustacean remains; 5: cephalopods; 6; other crustaceans. 1 oo- n=22 n=45 n=53 n=59 n=36 n=37 n = 35 0.75- i i ( » 0.50- < i I i i i 0.25" 4 > i » .«»■ „c/ .c <*■ J^ Size class (cm) Figure 2 Ontogenetic variations of the diet diversity index (B<7±CI9Vi ) of Cynoscion nannus (n is the number of stomachs contain- ing food). through 5 (51.1%-65.1%). The trophic spectrum of the smallest C. nannus (7 cm sTL slO.9 cm, n = 67) was composed by crustaceans (W^ = 68%), mostly carideans and stomatopods (W^ = 2Q9c). The diet of intermediate individuals (11 cm sTL sl6.9 cm; /; =148 ) was composed by penaeid shrimp, fish, and stomatopods. Only fish of the size classes grouped in this range showed percentages of diet similarity >50%. Among C. nannus between 17 and 18.9 cm TL (« = 37), the value of consumed fish biomass at- tained 69%, whereas that of penaeid shrimp reached 20%. Only among the larger individuals (19 cm sTL s20.9 cm; n = 35) did cephalopods attain high gravi- metric values (W^=457() followed by penaeid shrimp (W£=38%). Values of trophic niche breadth for each size class indicated ontogenetic variation in the diet (Fig. 2). The smallest individuals fed on a smaller number of prey species and showed a trend towards higher trophic specialization. Larger individuals, however, had a wider trophic spectrum and fed on a greater number of different prey species. Temporal variations in the dietary composition of C. nannus were significant (F=3.58; P<0.05). During the first months of the year, C. nannus consumed a higher percentage offish ( WJ = 37.2'7f ), NOTE Raymundo-Huizar et al.: Feeding habits of Cynoscion nannus 457 1.00- n = 20 n- 25 II 075- II n= 39 n=34 n - 35 n - 24 n=24 iS 0.50- II n= 58 " , , n = 28 0.25- " uuu"l i i i I I I I I I I I Month Figure 3 Monthly mean values of the diet diversity index [Ba± CI95,; 1 of Cynoscion nannus. The overall mean value of the index (Bo; ) and its confidence intervals (CI95%; ....) are shown in is the number of stomachs containing food). whereas towards the end of the year, penaeid shrimp were eaten in higher proportions (Wf = 50.6%). During May, stomatopods and carideans were found with higher biomass values than during the rest of the year lW^ = 68.2% and 20%, respectively). Cephalopods were found in most months with biomass values ranging from 4% to 34% of consumed biomass. The mean value of diet diversity was 0.41 (±0.18 CIg5<7t). Although the number of dietary categories for C. nannus that were identified was high (29 prey types), there were a few items with significant importance. Monthly variations in Ba ranged from 0.1 to 0.8 (Fig. 3). During most of the period analyzed, Ba values were not significantly different from each other as shown by the lack of overlap between CI95,-. The only exceptions were January and April, when CI95% was above the mean Ba ±CI95Q value, and October when ClS59 was below the mean Ba. Discussion Cynoscion nannus is a carnivorous fish that feeds on at least 29 different prey types. Although cannibal- istic behavior has been reported for several fish spe- cies in a variety of habitats and life-history strategies (Smith and Reay, 1991), C. nannus as a prey type was found in only 0.8% (two individuals >15.0 cm TL) of all stomachs analyzed. According to the IRI values, crus- taceans— specifically juvenile shrimp and stomatopods of the genus Squilla — appear to be the most important items in the diet. The type of substrate can influence the feeding habits of these fish. For example, Minello and Zimmerman (1984) observed that under experimen- tal conditions, the feeding preferences of C. nebulosus (16 cm^TL<;21 cm) for Farfantepenaeus aztecus varied depending on the substrate. These authors suggested that substrate characteristics determine the burrow- ing capacity of F. aztecus and thus predator avoidance. In the study area, juvenile shrimp and stomatopods of the genus Squilla can be abundantly found in soft- bottom habitats (Gonzalez-Sanson et al., 1997). Both the cephalopod iLoliopsis diomedae) and the fish spe- cies found in the stomach contents of C. nannus are pelagic or demersal species, indicating that the feeding activities of C. nannus are not exclusively limited to the benthos, and that this species can forage throughout the water column. Results in the present study provide evidence that fish feeding at different water depths have access to a broader variety of prey types. This has been shown both for other Sciaenidae (Chao and Musik, 1977; Campos and Corrales, 1986; Chao, 1995; Pelaez-Rodri- guez, 1996; Cruz-Escalona, 1998), and other species of demersal fish (Lucena et al., 2000). 458 Fishery Bulletin 103(2) It should be noted that graphic representations of the IRI values are more accurate in describing the diet of fish species (Cortes, 1997). Our results (Fig. 1) demon- strate that the three indices representing the relative importance of each food item highlight the influence that the percentages of occurrence, by number and by weight, have on the overall IRI values. The temporal analysis of the tropic spectrum of C. nannus showed that during October, November, and December this species fed mainly on penaeid shrimps. Fish prey were abundant in stomachs collected only during March, April, June, and November. Stomato- pods were present all year round, but only abundant during May. Low Ba values in October were due to the prevailing consumption of Solenocera spp. Monthly dif- ferences in the diet of C. nannus were most probably in accordance with the seasonal variations in prey species abundance, which in turn determined their availability. Lucena et al. (2000) found that temporal variations in the diet of C. guatucupa from southern Brazil are related to seasonal production cycles of prey, mainly fish and crustaceans, thus supporting the view that sciaenids can generally be considered opportunistic species. Results of this study showed ontogenetic variations in the trophic spectrum of C. nannus. The smallest individuals (7 cm ^ o ^^ o o - co ^^ III I I 1 1 1 [ 1 1 1 5 15 25 35 45 55 5 15 25 35 45 55 65 75 85 Dentary body length (mm) Dentary length (mm) t o - E °> FL= 10.19 x OSL- 16.51 ^f S~ FL = 6.38 xCt- 20.87 -^ length 600 iiS** ^ ~ •V« Fork 300 I ^^ I- ^^ 1 I I I I I I I I k i I I I I I I 5 1 5 25 35 45 55 65 75 85 0 20 40 60 80 1 00 120 1 40 Opercle length (mm) Cleithrum length (mm) o O / CD FL= 10.48 x MXL- 1593 -^ §" FL = 11.11 x PMXL- 12.99 -S* O o - CD *•* o - O o - CO _^f^ CO Jf£*> I I I I I I I I I I I I I I I I I 1 5 15 25 35 45 55 65 75 85 5 15 25 35 45 55 65 75 85 Maxilla length (mm) Premaxilla length (mm) Figure 2 Fork length (mm) in relation to six skull bone measurements (mm) in bluefish (Poma- tomus saltatrix). Resulting linear regression models and trendlines are shown. ( Observed - Predicted) (Predicted) xlOO. To determine if any one bone or set of bones provided the best predictor equation, comprehensive models involving sets of bones were fitted in a stepwise linear algorithm by using the Akaike information criterion (AIC) as the criterion for model selection. Models were generated in both a forwards and backwards manner in order to con- firm that the same model was returned in all cases. Results Fork length (FL) and total length (TL) measurements were taken from 58 bluefish ranging from 110 mm to 900 mm FL. The resulting regression equations correlating skull bone measurements to FL (Fig. 2) were highly significant (P=0.005 for the dentary correlation and P<0.001 for the rest of the models). The r2 values for the FL predictive equations ranged from 0.988 to 0.997, and the mean percent predictive errors ranged from -0.03 to 1.19 (Table 1). Similarly, all of the resulting models correlating the bone measurements to total length (Fig. 3) were highly significant (P<0.001, r'2 values ranging from 0.987 to 0.996, and mean percent predictive errors ranging from -0.11 to 1.07 [Table 1]). Bones were ranked from best predictor to worst pre- dictor for both the FL and TL models by using the Akaike information criterion (AIC). In both cases the premaxilla was ranked the best predictor bone, followed by the maxilla, the opercle, the dentary, the cleithrum, and finally dentary body length. The bone measure- ments included in the stepwise multiple regression mod- el for predicting fork length were PMXL, OPL, and DN (Table 2). In the best predictor model for total length, PMXL, OPL, DN and CL were included (Table 2). 464 Fishery Bulletin 103(2) Table 1 Resulting predictive equations of fork and total length in relation to several skull bone measures with corresopnding coefficient of determination (r2) and P-values, and mean percent predictive errors OPE) for each model. Bone Fork length r2 P-value %PE Dentary body length (DBLl FL = 18.27- 22.46 0.988 <0.001 0.54 Dentary (DN) FL = 10.97(£W)- 11.27 0.996 0.005 -0.03 OperclelOPL) FL = 10.19(OPL)- 16.51 0.997 <0.001 0.28 Cleithrum(CL) FL = 6.38(CLl- 20.87 0.993 <0.001 1.19 Maxilla (MXL) FL= 10.48 (MXL)- 15.93 0.997 <0.001 0.31 Premaxilla(PMXL) FL = ll.ll(PMA'L) - 12.99 0.997 <0.001 0.26 Bone Total length r2 P-value %PE Dentary body length (DBL) TL = 20. 20(DSL) - 27.69 0.987 <0.001 0.46 Dentary (DN) TL= 12.130W) - 15.42 0.996 <0.001 -0.11 OperclelOPL) TL= 11.27IOPD- 21.13 0.996 <0.001 0.15 Cleithrum(CL) TL = 7.05ICD- 26.13 0.994 <0.001 1.07 Maxilla (MXL) TL= 11.59(MA'L)- 20.43 0.996 <0.001 0.19 Premaxilla(PMXL) TL= 12.28(PMXLl- 17.20 0.996 <0.001 0.14 Table 2 Independent variables included in the stepwise linear regression models used to estimate original bluefish fork length and total length. Variables included in forward stepwise regression model Variables included in backward stepwise regression model Fork length Total length PMXL, OPL, DN PMXL, OPL, DN, CL PMXL, OPL, DN PMXL, OPL, DN, CL Discussion This study revealed that measurements of five skull bones can be used as accurate predictors of original fork length and total length of bluefish. Although the methods of other studies were incorporated in this study, the information is the first of its kind for bluefish and may serve as a tool for the future study of this species in the North Atlantic. In recent years there has been growing concern over the stability of the bluefish stock and an increased ef- fort to gather information on the possible mechanisms affecting bluefish abundance and distribution in the western North Atlantic.4 One of the proposed mecha- In 1997 Rutgers University and the NMFS organized a work- shop to study the factors that could be contributing to the depressed state of the bluefish stock. A similar concern was expressed by Congress at this time, and the Rutgers and NMFS workshop led to a request for proposals for bluefish- related research in 1998, 1999, and 2000. nisms that could be adversely influencing the recovery of bluefish is top-down pressure by a number of apex predators in the North Atlantic. Although indiscrimi- nant predation on bluefish may not be a significant pressure on the stock, size selective predation can dra- matically alter the structure of the prey community (Mclntyre and Ward, 1986; Trippel and Beamish, 1987; Sharf et. al., 1997). In order to study the consumption rates of key preda- tors in an ecosystem it is necessary to gather informa- tion on the sizes of the prey being consumed (Elliot and Persson, 1978; Sharf et. al., 1998). However, it is often difficult to estimate the original size of a prey item from stomach content data because of the complications caused by digestion. Erosion of the prey bones from digestive juices can lead to measurement error or bias when prey sizes are back-calculated from digested parts (Sharf et al., 1998). Although bias from digestion is a concern that should be addressed in studies, internal bones and hard parts of fishes have been shown to be excellent predictors of original prey size (Trippel and NOTE Wood: Using bone measurements to estimate the size of Pomatomus saltatnx 465 TL = 20.20 x DBL - 27.69 ^f* 71 = 12.13 xDW- 15.42 ^/^ o m - r- •Z" m - ^^ o o - IT! s^1* §" y^ o Lfi — C\J **• o ^^ ■ 1 1 I 1 1 1 1 1 1 1 1 5 15 25 35 45 55 5 15 25 35 45 55 65 75 85 95 Dentary body length (mm) Dentary length (mm) TL = 11.27 x OPL- 21 13 ^^ TL = 7.05 x CL - 26 1 3 ^^ Fork length (mm 250 500 750 250 500 750 l l l ^S^ I I I I I I I I 5 15 25 35 45 55 65 75 85 95 20 40 60 80 100 120 140 160 Opercle length (mm) Cleithrum length (mm) -i 71= 11 59* MXL- 20 43 .^ H TL= 12.28 x PMXL- 17.20 ^^ o in - vrf»'» y0^ o o - IT) r*^**^ § - JiS^^ o in - CVJ ^^ l~ ^^ II 1 1 1 1 1 1 1 1 1 1 5 15 25 35 45 55 65 75 85 5 15 25 35 45 55 65 75 85 Maxilla length (mm) Premaxilla length (mm) Figure 3 Total length (mm) in relation to six skull bone measurements (mm) in bluefish (Poma- tomus saltatrix). Resulting linear regression models and trendlines are shown. Beamish, 1987; Hansel, 1988, Sharf et al., 1998). In addition, the bones used in the present study are strong bones (with the exception of the opercle), that are liable to resist digestive erosion. All the relationships generated in the present study yielded very accurate predictions of original prey size, but the jaw bones are of special interest. Bluefish can be classified as predators that exhibit a biting behavior during predation. Fish that show this type of predation behavior have very heavy, robust jaw bones (Norton, 1995). The jaw bones (maxilla, premaxilla, and dentary) of bluefish are both easily identifiable and likely resis- tant to digestion, and when combined with the adequacy with which original size can be determined from these bones (based on AIC rankings and %PE), they are the best option for researchers interested in back-calculat- ing original bluefish sizes. The results of this study provide a means to fur- ther analyze the stomach contents of bluefish preda- tors beyond identifying, and quantifying prey items. The usefulness of this type of data has been shown repeatedly for a number of species (Mclntyre and Ward, 1986; Feltham and Marquiss, 1989; Serafy et. al., 1996; Sharf et. al., 1997; Sharf et. al., 1998). The ability to back-calculate the original size of a prey leads to the enhancement of diet studies and allows for more accu- rate estimates of predator consumption rates. The lack of this kind of data and correlations for many key prey species in the Atlantic and elsewhere is surprising. Acknowledgments Funding for this study was provided by the Bluefish- Striped Bass Dynamics Research Program at Rutgers University in cooperation with the National Marine Fisheries Service (grant NA97FE0363). I am indebted to the numerous fishing tournament directors, as well as the fishermen at the tournaments for allowing me to collect many of the bluefish needed for this study. I am 466 Fishery Bulletin 103(2) also especially grateful to Abby McLean for her help with the exhausting task of measuring bones. Finally, I wish to thank Francis Juanes for encouraging me to pursue and publish this study and Jeremy Collie and two anonymous reviewers for comments that helped to improve this manuscript. Literature cited Bucket, J. A., M. J. Fogarty, and D. O. Conover. 1999. Foraging habits of bluefish, Pomatomus saltatrix, on the U.S. east coast continental shelf. Fish. Bull. 97:758-775. Chase, B. C. 2002. Differences in diet of Atlantic bluefin tuna (Thun- nus thynnus) at five seasonal feeding grounds on the New England continental shelf. Fish. Bull. 100:168-180. Crane, S. A., J. M. Fenaughty. and R. W. Gauldie. 1987. The relationship between eye diameter and fork length in the spiny oreo dory, Allocyttus sp. N.Z.J. Mar. Freshw. Res. 21:641-642. Elliott, J. M., and L. Persson. 1978. The estimation of daily rates of food consumption for fish. J. Anim. Ecol. 47:977-991. Feltham, M. J., and M. Marquiss. 1989. The use of first vertebrae in separating, and estimat- ing the size of, trout tSalmo trutta ) and salmon {Salmo salar ) in bone remains. J. Zool. 219:113-122. Fickling, N. J., and R. L. G. Lee. 1981. Further aids to the reconstruction of digested prey lengths. Fish Manage. 12:107-110. Hansel, H. C, S. D. Duke, P. T Lofy, and G. A. Gray. 1988. Use of diagnostic bones to identify and estimate original lengths of ingested prey fishes. Trans. Am. Fish. Soc. 117:55-62. Kohler, N.E. 1989. Aspects of the feeding ecology of the blue shark, Prionace glauca, in the western North Atlantic. Diss. Abst. Int. Pt. B-Sci. & Eng. 49:1-179. Mclntyre, D. B., and F. J. Ward. 1986. Estimating fork lengths of fathead minnows, Pimephales promelas, from measurement of pharyngeal arches. Can. J. Fish. Aquat. Sci. 43:1294-1297. Newsome, G. E. 1977. Use of opercular bones to identify and estimate lengths of prey consumed by piscivores. Can. J. Zool. 55: 733-736. Norton, S. F. 1995. A functional approach to ecomorphological pat- terns of feeding in cottid fishes. Environ. Biol. Fishes 44:61-78. Pikhu, E. Kb.., and E. R. Pikhu. 1970. Reconstruction of the size of fishes swallowed by predators from fragments of their vertebral column. J. Ichthyol. (USSR), vol. 10:706-709. Radke, R. J.. T. Petzoldt, and C. Wolter. 2000. Suitability of pharyngeal bone measures com- monly used for reconstruction of prey fish length. J. Fish Biol. 57:961-967. Scharf, F. S., J. A. Buckel, F. Juanes, and D. O. O'Connor. 1997. Estimating piscine prey size from partial remains: Testing for shifts in foraging mode by juvenile bluefish. Environ. Biol. Fishes 49:377-388. Scharf, F. S., R. M. Yetter, A. P. Summers, and F. Juanes. 1998. Enhancing diet analyses of piscivorous fishes in the northwest Atlantic through identification and reconstruc- tion of original prey sizes from ingested remains. Fish. Bull. 96:575-588. Serafy, J. E„ C. M. Schmitz, T. R. Capo, M. E. Clarke, and J. S. Ault. 1996. Total length estimation of red drum from head dimensions. Prog. Fish-Cult. 58:289-290. Stillwell, C. E„ and N. E. Kohler. 1982. Food, Feeding Habits, and Estimates of Daily Ration of the Shortfin Mako (Isurus oxyrinchus) in the Northwest Atlantic. Can. J. Fish. Aquat. Sci. 39: 407-414. 1985. Food and feeding ecology of the swordfish Xi- phas gladias in the western North Atlantic Ocean with estimates of daily ration. Mar. Ecol. Prog. Ser. 22: 239-247. Trippel, E. A., and F. W. H. Beamish. 1987. Characterizing piscivory from ingested remains. Trans. Am. Fish. Soc. 116:773-776. I < r.m/FBNMFMT PPTWTINr. flFFTrF- 0Cti\c. — 7Q1-110 / Q77Zf, Bo ciitoH $mt^hj<£&*®mt£}} U.S. Department of Commerce Volume 103 Number 3 July 2005 Fishery Bulletin U.S. Department of Commerce Carlos M. Gutierrez Secretary National Oceanic and Atmospheric Administration Vice Admiral Conrad C Lautenbacher Jr., USN (ret.) Under Secretary for Oceans and Atmosphere National Marine Fisheries Service William T. Hogarth Assistant Administrator for Fisheries <^T0Fe% 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 C15700, Seattle, WA 98115-0070. Periodicals postage is paid at Seattle, WA, and at additional mailing offices. POST- MASTER: Send address changes for sub- scriptions to Fishery Bulletin, Superin- tendent of Documents, Attn.: Chief, Mail List Branch, Mail Stop SSOM, Washing- ton, DC 20402-9373. Although the contents of this publica- tion have not been copyrighted and may be reprinted entirely, reference to source is appreciated. The Secretary of Commerce has deter- mined that the publication of this peri- odical is necessary according to law for the transaction of public business of this Department. Use of funds for printing of this periodical has been approved by the Director of the Office of Management and Budget. For sale by the Superintendent of Documents, U.S. Government Printing Office, Washington, DC 20402. Subscrip- tion price per year: $55.00 domestic and $68.75 foreign. Cost per single issue: $28.00 domestic and $35.00 foreign. See back for order form. Scientific Editor Norman Bartoo, Ph.D. Associate Editor Sarah Shoffler National Marine Fisheries Service, NOAA 8604 La Jolla Shores Drive La Jolla, California 92037 Managing Editor Sharyn Matriotti National Marine Fisheries Service Scientific Publications Office 7600 Sand Point Way NE, BIN C15700 Seattle, Washington 98115-0070 Editorial Committee Harlyn O. Halvorson, Ph.D. Ronald W. Hardy, Ph.D. Richard D. Methot, Ph.D. Theodore W. Pietsch, Ph.D. Joseph E. Powers, Ph.D. Harald Rosenthal, Ph.D. Fredric M. Serchuk, Ph.D. George Waiters, Ph.D. University of Massachusetts, Boston University of Idaho, Hagerman National Marine Fisheries Service University of Washington, Seattle National Marine Fisheries Service Universitat Kiel, Germany National Marine Fisheries Service National Marine Fisheries Service Fishery Bulletin web site: www.fishbull.noaa.gov The Fishery Bulletin carries original research reports and technical notes on investigations in fishery science, engineering, and economics. It began as the Bulletin of the United States Fish Commission in 1881; it became the Bulletin of the Bureau of Fisheries in 1904 and the Fishery Bulletin of the Fish and Wildlife Service in 1941. Separates were issued as documents through volume 46; the last document was No. 1103. Beginning with volume 47 in 1931 and continuing through volume 62 in 1963, each separate appeared as a numbered bulletin. A new system began in 1963 with volume 63 in which papers are bound together in a single issue of the bulletin. Beginning with volume 70, number 1, January 1972, the Fishery Bulletin became a periodical, issued quarterly. In this form, it is available by subscription from the Superintendent of Documents, U.S. Government Printing Office, Washington, DC 20402. It is also available free in limited numbers to libraries, research institutions, State and Federal agencies, and in exchange for other scientific publications. U.S. Department of Commerce Seattle, Washington Volume 103 Number 3 July 2005 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 iNOAAl or any other agency or institution. The National Marine Fisheries Service iNMFS) 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 469-488 489-500 501-515 516-523 524-535 536-543 544-552 553-558 559 Articles Marina Biotog"" Woods Hole Oeanographic *' L'b jyt * d 2005 Dressel, Sherri C, and Brenda L. Norcross Using poststratification to improve abundance estimates from multispecies surveys: a study of |uvenile flatfishes Francis, Malcolm P., and Clinton Duffy Length at maturity in three pelagic sharks (Lamna nasus, Isurus oxyrinchus, and Pnonace glauca) from New Zealand Fritz, Lowell W., and Eric S. Brown Survey- and fishery-derived estimates of Pacific cod (Gadus macrocephalus) biomass: implications for strategies to reduce interactions between groundfish fisheries and Steller sea lions (Eumetopias jubatus) Greig, Thomas W., M. Katherine Moore, Cheryl M. Woodley, and Joseph M. Quattro Mitochondrial gene sequences useful for species identification of western North Atlantic Ocean sharks Hawkins, Sharon L., Jonathan Heifetz, Christine M. Kondzela, John E. Pohl, Richard L. Wilmot, Oleg N. Katugin, and Vladimir N. Tuponogov Genetic variation of rougheye rockfish (Sebastes aleutianus) and shortraker rockfish (5. borealis) inferred from allozymes Sulikowski, James A., Jeff Kneebone, Scott Elzey, Joe Jurek, Patrick D. Danley, W. Huntting Howell, and Paul C. W. Tsang The reproductive cycle of the thorny skate (Amblyra/a radiata) in the western Gulf of Maine Notes Fey, Dariusz P., Gretchen E. Bath Martin, James A. Morris, and Jonathan Hare Effect of type of otolith and preparation technique on age estimation of larval and |uvenile spot (Leiostomus xanthurus) Piner, Kevin R., Melissa A. Haltuch, and John R. Wallace Preliminary use of oxygen stable isotopes and the 1983 El Niiio to assess the accuracy of aging black rockfish (Sebastes melanops) Subscription form 469 Abstract— Population assessments seldom incorporate habitat informa- tion or use previously observed dis- tributions of fish density. Because habitat affects the spatial distribution offish density and overall abundance, the use of habitat information and previous estimates offish density can produce more precise and less biased population estimates. In this study, we describe how poststratification can be applied as an unbiased estimator to data sets that were collected under a probability sampling design, typi- cal of many multispecies trawl sur- veys. With data from a multispecies survey of juvenile flatfish, we show how poststratification can be applied to a data set that was not collected under a probability sampling design, where both the precision and the bias are unknown. For each of four spe- cies, three estimates of total abun- dance were compared: 1) unstratified; 2) poststratified by habitat; and 3) poststratified by habitat and fish den- sity (high fish density and low fish density) in nearby years. Poststrati- fication by habitat gave more precise and (or) less design-biased estimates than an unstratified estimator for all species in all years. Poststratification by habitat and fish density produced the most precise and representative estimates when the sample size in the high fish-density and low fish-density strata were sufficient (in this study, Ha20 in the high fish-density stratum, na9 in the low fish-density stratum). Because of the complexities of statis- tically testing the annual stratified data, we compared three indices of abundance for determining statisti- cally significant changes in annual abundance. Each of the indices closely approximated the annual differences of the poststratified estimates. Selec- tion of the most appropriate index was dependent upon the species' density distribution within habitat and the sample size in the different habitat areas. The methods used in this study are particularly useful for estimating individual species abundance from multispecies surveys and for retro- spective studies. Manuscript submitted 28 December 2001 to the Scientific Editor's Office. Manuscript approved for publication 31 March 2005 by the Scientific Editor. Fish. Bull. 103:469-488 (2005). Using poststratification to improve abundance estimates from multispecies surveys: a study of juvenile flatfishes Sherri C. Dressel Brenda L. Norcross Institute of Marine Science School of Fisheries and Ocean Sciences University of Alaska Fairbanks 245 O'Neill Building Fairbanks, Alaska 99775-7220 Present Address (for S C Dressel): Alaska Department of Fish and Game Commercial Fisheries Division 802 3rd Street P.O. Box 240020 Douglas, Alaska 99824-0020 E-mail address (for S C. Dressel): shern_dressel(S)fishgame. state. ak. us Scientists must be able to assess popu- lation abundance with a high degree of confidence to achieve the goals of fishery management (Quinn, 1985). To do this, survey designs and esti- mation methods that minimize the variance in estimates of abundance are needed. Recently, the National Research Council (NRC, 2000) rec- ommended incorporating habitat information and commercial fisher- ies data in population assessments. Both of these data may result in lower variances in estimates of abundance. Habitat type and habitat quality are becoming more widely recognized as primary determinants for the dis- tribution and survival of marine fish species (Murawski and Finn, 1988; Gadomski and Caddell, 1991; Reichert and van der Veer, 1991; Norcross et al., 1999). Until recently, however, few studies have been directed to- ward defining fish habitat or using habitat associations to help decrease the variability in abundance estima- tion (Scott, 1995). In response to the growing recognition of the importance of habitat, the Magnuson-Stevens Fishery Conservation and Manage- ment Act was amended in 1996 (Pub- lic Law 104-297) so that the National Marine Fisheries Service (NMFS) and regional fishery management councils must describe and identify essential fish habitat (EFH) for managed spe- cies. Similarly, a recent report from the NRC calls for methods that link environmental data to stock assess- ments (NRC, 2000). Poststratification can be used in a number of different ways to address the NRC recommendations. Although poststratification is not a new statisti- cal method, it is one that is not com- monly used for estimating ground- fish population abundance and can be used to meet these newly defined challenges. In contrast to a stratified sampling design, poststratification is a method that allocates samples to strata after they have been col- lected. As a result, habitat data col- lected during a survey can be used for stratification. When poststratifi- cation is applied to data that have been collected under a simple random sampling design, the poststratifica- tion estimator is unbiased and may produce more precise estimates than those from a simple random sampling estimator. Poststratified estimates will be nearly as precise as strati- fied sampling with proportional al- location, in which the sample sizes in each stratum are proportional to stratum sizes, if stratum sample sizes are large (rc>20) and errors in esti- mates of strata areas are negligible (Cochran, 1977; Pollock et al., 1994; 470 Fishery Bulletin 103(3) Scheaffer et al., 1996). If poststratification is applied to data from a multispecies survey, 1) abundance data for each species can be poststratified with different habitat variables or 2) abundance data for every species can be poststratified with the same variables, but different stratum boundaries can be used for each species. Many large-scale multispecies groundfish surveys are conducted by using a stratified random sampling design (Azarovitz, 1981; Halliday and Koeller, 1981; Pitt et al., 1981; Martin1; Weinberg et al.2). Depth, distance from or along shore, latitude, distance along depth contours, or broad geographic features (such as bays, capes, banks, gullies, and slopes) are used as stratum boundaries in trawl surveys because they have been shown to be related to species distributions. These fac- tors are fixed spatially, allowing samples to be allocated to strata prior to sampling. The same boundaries are used for all species, and boundaries generally remain the same over years. When conducting a multispecies survey with a strati- fied random sampling design, optimal stratification for one species may not be optimal for others (Koeller, 1981; NRC, 2000). Because the placement of strata boundar- ies is critical for precise stratified estimates (Cochran, 1977), use of a stratified sampling design for a multispe- cies survey may result in only small gains in precision for some or all species. Poststratification is possible for data that have been collected under a stratified design. It can be used to stratify data more finely for individual species. Under stratified random sampling, a simple random sample is taken in each stratum. Thus, data within each stratum can be poststratified separately with additional variables and the abundance estimates from each of the strata can be summed. The resultant estimator is unbiased and likely will be more precise than that of the original stratified design if sample sizes in poststratified strata are large enough. Often, researchers need to estimate abundance from data sets that were not recorded under a probability sampling design (a design in which randomness is built into the survey design, such as simple random sampling or stratified random sampling). Finances and logistics, for example, may make it impossible to collect data under a probability sampling design, researchers may want to estimate species abundance from commercial fisheries or other nonsurvey data, or previously collected data sets that were not recorded under a probability 1 Martin, M. H. 1997. Data report: 1996 Gulf of Alaska bottom trawl survey. NOAA Tech. Memo. NMFS-AFSC- 82, 235 p. National Technical Information Service, U.S. Department of Commerce, 5285 Port Royal Road, Springfield, Virginia 22161. 2 Weinberg, K. L., M. E. Wilkins, R. R. Lauth, and P. A. Ray- more jr. 1994. The 1989 Pacific west coast bottom trawl survey of groundfish resources: Estimates of distribution, abundance, and length and age composition. NOAA Tech. Memo. NMFS-AFSC-33, 168 p., plus appendices. National Technical Information Service, U.S. Department of Com- merce, 5285 Port Royal Road, Springfield, Virginia 22161. sampling design may be used for retrospective studies. In this article, we refer to data collection without a probability sampling design as "haphazard sampling." The use of haphazardly collected data for estimating abundance is undesirable because they cannot be eval- uated by the theorems of probability theory (Krebs, 1989). Although undesirable, it is often necessary to analyze haphazardly collected data and effective meth- ods are needed to do so. Poststratification can be applied to data that were not collected with a probability sampling design. When poststratification is applied to data not collected under a probability sampling design, the poststratification es- timator, a design-based estimator, may be biased. When analyzing such data, it is important both to maximize the precision and to minimize the bias. Poststratifica- tion has been applied to nonprobability samples in other studies to increase the precision (Hall and Boyer, 1988) and decrease the bias of estimators (Buckland and An- ganuzzi, 1988; Hall and Boyer, 1988; Anganuzzi and Buckland, 1989). Poststratification can be useful, but has some draw- backs. With poststratification, sample sizes within strata are random variables — which are an additional source of variability over that of a stratified sampling variance estimator (Thompson, 1992; Scheaffer et al., 1996). The variance of a poststratified estimator can be estimated by using standard stratified sampling variance equations and by incorporating an additional approximate term to account for the random sample sizes present with poststratification (Scheaffer et al., 1996). Alternatively, the variance of a poststratified estimator can be estimated by conditioning on samples sizes and by applying the standard stratified sampling variance equation (Thompson, 1992). For accurate post- stratification estimates, the proportion of total possible samples in each stratum (for this study the propor- tion of the total survey area included in each stratum) must be known or approximated closely enough that the error in the approximation is negligible (Cochran, 1977). Error in estimates of stratum sizes causes bias in poststratified estimates of abundance. Because error in the estimation of stratum size is unaccounted for in the estimated variance of poststratified estimates, the estimated variances may be underestimates of the true error (Cochran, 1977). This study had two goals. The first goal was to evalu- ate the benefits and drawbacks of using poststratifica- tion to incorporate habitat and fish-density information into estimates of abundance from multispecies survey data that were not collected under a probability sam- pling design. To achieve this goal, this study compared three estimates of total abundance and variance (un- stratified, poststratified by habitat, poststratified by habitat and estimates of fish density in neighboring years) for each of four species. The comparison was made to determine whether poststratification of hap- hazardly sampled data with habitat and fish-density information increases the precision and helps account for possible bias in abundance estimates. Dressel and Norcross: Using poststrafication to improve abundance estimates from multispecies surveys 471 Because this study is an observational study with haphazard sampling, the precision and bias cannot be directly assessed. Instead, we estimated and compared the precision by using unstratified and poststratified estimators. We qualitatively estimated the relative amount of design bias (i.e., how representative the esti- mates are) with the use of habitat. In previ- ous studies (Norcross et al., 1995; 1997; 1999), depth and sediment were identified as habitat characteristics closely associated with the dis- tribution of the four species in this study. From depth, sediment, and fish abundance data col- lected in this study we were able to identify ranges of habitat characteristics associated with areas of high, low, and no fish density. By estimating the proportion of area (km2) in the study area characterized by the ranges of depth and sediment, it was possible to estimate the proportion of the survey area with high, low, and no fish density. Because samples in our study were not randomly allocated, the probability of selection was not equal among all samples in the survey area. The resulting num- bers of samples taken in areas of high, low, and no fish density were not in proportion to the size (km2) of those areas as it would have been with repeated simple random sampling. There- fore, by comparing the relative size of high, low and no fish-density areas in the survey area with the relative number of samples in those areas, we made qualitative estimates of the design bias associated with the estimators. Although an assessment of the relative amount of design bias made in this way is only an approximation, it is helpful when using haphaz- ardly collected data in order to provide some indication of the amount of design bias based on the disproportion of samples in an area to the size of that area. Because of the complexities of statistically testing the annual stratified data, the second goal of our study was to develop indices of abundance that closely approximat- ed the annual differences of poststratified estimates and that could easily be tested for statistically significant changes between years. To achieve the second objective, three indices of annual relative abundance were con- structed and compared with respect to their estimated relative precision and design bias: one from all sites in the survey area, one from all sites within the species' habitat, and one from all sites within an area of high fish density within the species' habitat. The data for this study were obtained from six years of juvenile groundfish surveys conducted in Kalsin Bay and Middle Bay, Kodiak Island, Alaska. The four spe- cies studied were age-0 rock sole (Lepidopsetta spp.), age-1 yellowfin sole (Pleuronectes asper), age-0 Pacific halibut (Hippoglossus stenolepis), and age-0 flathead sole (Hippoglossoides elassodon). The survey data were collected during the six-year survey under three dif- ferent survey designs, none of which were strictly ran- domized, but each involved some degree of haphazard Study area Alaska. Figure 1 (in black) in Middle and Kalsin Bays, Kodiak Island, sampling due to weather, sediment structure, and other logistical restrictions for beam trawling in small bays off the Gulf of Alaska (Norcross et al.3). Although many trawl survey data sets to which these methods could be applied are collected under a probability sampling de- sign where the estimator is unbiased, the haphazardly collected data set used in our study was chosen to show how poststratification can be applied when both the pre- cision and the bias of the estimator are unknown. Methods Sampling Middle and Kalsin Bays are part of Chiniak Bay, 10 nmi south of the town of Kodiak, Alaska. The total size of the study area, 87 km2, included the combined areas of both bays and the areas directly outside the mouths of the bays (Fig. 1). Middle Bay is 8 km long and has depths of 50 m at the mouth of the bay and an area of 21 km2. Kalsin Bay is 8 km long, has depths greater than 100 m 3 Norcross, B. L., B. A. Holladay, A. A. Abookire, and S. C. Dressel. 1998. Defining habitats for juvenile groundfishes in Southcentral Alaska with emphasis on flatfishes. Vol. I, Final Study Report, OCS Study MMS 97-0046, 131 p. Coastal Marine Institute, Univ. Alaska Fairbanks, Fairbanks, AK 99775. 472 Fishery Bulletin 103(3) N 57.70 ~ N 57.65 M 52.55 W 152.50 W 152.45 W 152 35 W 152.30 Longitude Figure 2 Kalsin and Middle Bay sample sites (1991-96) and bathymetry. Fixed (sampled every year) sites are noted. at the mouth of the bay, and encompasses an area of 34 km2. Rocky cliffs and islands surround the mouths of the bays, and rocks in the sediment made several areas untrawlable (Fig. 2). Although trawling was not conducted in these areas, depth and sediment data were collected. In this analysis, untrawlable areas were still considered possible flatfish habitat and were included in the measurements of the size of the total study area. Annual cruises were conducted in Middle and Kalsin Bays for two weeks in August from 1991 to 1996. Ju- venile flatfish were collected by using 3.05 and 3.66 m plumb-staff beam trawls (Gunderson and Ellis, 1986). Trawl nets were made of 7-mm square net mesh and had a 4-mm codend liner that retained flatfish as small as 11 mm. Sampling methods were consistent for all six years (Norcross et al., 1995; Norcross et al.3). Collections at each sample site included a tow of 10 minutes or less, a vertical CTD (conductivity, temperature and depth) cast, and a sediment grab (0.06-m3 Ponar grab). The sampling area of each tow was determined by the width of the beam trawl, which was 0.74 of the beam length (Gunder- son and Ellis, 1986), and distance towed was based on global positioning system (GPS) coordinates. Fish were identified to the lowest possible taxon and measured to the nearest millimeter total length. At the time of collections, all rock sole were identified as Pleuronectes bilineatus. Following Orr and Matarese's (2000) revision of the genus, we refer to these fishes as Lepidopsetta spp. in this article because both species, L. bilineata and L. polyxystra, were identified in the study area during 1996 sampling. Fish ages were determined by length- frequency analysis. Fish catch-per-unit-of-effort (CPUE) values were standardized to a 1000-m2 tow area. Sampling designs varied from year to year (Norcross et al.3). Extensive exploratory sampling was conducted from 1991 through 1994 to describe juvenile flatfish distributions in relation to habitat characteristics (Nor- cross et al., 1995; 1997). The goal in these years was to sample over the widest range of areas and habitat char- acteristics possible within the depth, sediment, weather, and logistical constraints. In 1995 and 1996, sampling was stratified by depth and percent sand in sediment. The sample allocation and the number of strata differed in 1995 and 1996 (Norcross et al.3). Because of logisti- cal constraints, samples were not randomly allocated within each stratum. Within these sampling designs, nine fixed sites were chosen, each with different depth and sediment combinations and with high abundances of one of the four species. Each of the nine fixed sites was sampled at least once in each of the six years. For this study, survey data in each year were treated as unstratified samples that were not collected under a probability sampling design. Analysis Poststratification Habitat preferences of juvenile fiat- fishes, as defined by depth and sediment variables, have been identified as affecting the distribution and abun- dance of juvenile flatfish around Kodiak Island (Norcross et al., 1995; 1997; 1999; Mueter and Norcross, 1999) and elsewhere (Pearcy, 1978; Tanda, 1990; Burke et al., 1991; Rogers, 1992; Walsh, 1992). Four areas were defined for use in estimating total and relative abundance: habitat, nonhabitat, high fish-density (HFD) and low fish-density (LFD) areas. Percent sand was used as a continuous vari- able of sediment type. Suitable habitat (habitat area) was defined for each species as ranges of depth and percent sand in which the species was caught during one or more Dressel and Norcross: Using poststrafication to improve abundance estimates from multispecies surveys 473 of the six sampling years. Unsuitable habitat (nonhabitat area) was denned for each species as ranges of depth and percent sand in which the species was never caught. Within the habitat area, the area of high fish density for each year was defined as ranges of depth and percent sand associated with CPUEs in the 75th-100th percentile of nonzero catches in the five other years. The area of low fish density was defined as the remaining habitat area not incorporated in the HFD area. In order for the poststratification method to estimate abundance accurately (high precision and low bias), the size of each stratum must be known or closely approxi- mated (Cochran, 1977; Scheaffer et al., 1996). When using habitat variables to determine stratum sizes, the accuracy of stratum sizes defined by the boundaries is heavily dependent upon the number and distribution of habitat variable measurements. For our study, 243 depth and percent sand measurements collected over the six years at trawl locations were used to determine stratum boundaries. The ranges of depth and percent sand that defined the four areas for each species were contoured over the study area by using a minimum curvature algorithm (Surfer, 1995). The size of each stratum in relation to the size of the entire study area was then visually estimated to the nearest square ki- lometer. Although not used in our study, a digital rep- resentation of the size of each stratum and the size of the study area is recommended to produce more precise estimates. To assess the advantages and disadvantages of using poststratification to estimate abundance, three esti- mates of total abundance were calculated and compared for each species in each year. An unstratified estimate of total abundance was calculated from samples across the entire survey area, with no differentiation with regard to habitat. The unstratified estimate of total abundance was calculated with the standard simple random sampling equation The estimate poststratified by habitat was calculated as isl=lN,yr where fs, = the estimated population total; L = the number of strata (here L=2, habitat and nonhabitat); N: = the total number of possible samples in stra- tum ; (samples were standardized to 1000 m2, therefore iV, x 1000 m2=stratum size); and yi = the mean CPUE in stratum i. A third estimate, poststratified by habitat and fish density, was calculated with the same poststratification estimator with L = 3. This poststratification estimator used the HFD area of that year as one stratum, the LFD area of that year as the second stratum, and the nonhabitat area as the third. An approximate variance estimator (Scheaffer et al., 1996), VPGM , N(N-n)^N, 2 £IB>- was used to estimate the variance of each poststratifica- tion estimator, where V = the estimated poststratified variance of ist, the estimated population total; N = the total number of possible samples in the survey area; n = the total number of samples taken; Nt = the total number of possible samples in stratum i; and s;2 = the sample variance in stratum i. i = Ny, where i = the estimated population total; N = the total number of possible samples in the survey area; and y = the mean CPUE of all sites sampled in a year. The estimated variance for the unstratified estimator was calculated as The first term of the variance equation is the variance of a stratified sample mean under proportional allocation. The second term shows the amount of increase in vari- ance expected from post- rather than prestratification (Scheaffer et al., 1996). Relative efficiency statistics were calculated for pair- wise comparisons of the precision of the unstratified and the two poststratified estimates. Pairwise com- parisons of the estimates were made for each species in each year. Relative efficiency was calculated as V(i) = N 2 s 2\ m R.E. Va where V(f ) = the estimated variance of the population total estimate; N = the total number of possible samples in the survey area; n = the total number of samples taken; and s2 = the sample variance. where V^ represents the variance of an unstratified estimate or a stratified sample with fewer strata than the estimate of variance represented by VB. The variance of an estimate is directly affected by the sample size (Zar, 1996). In our study, three total abun- dance estimates and their respective variances were 474 Fishery Bulletin 103(3) calculated and compared for each of the 24 species- year combinations. One of the three total abundance estimates was most precise for each of the species-year combinations. For each species-year combination, the habitat stratum sample size (used in the estimate post- stratified by habitat), the HFD stratum sample size, and the LFD stratum sample size (both used in the estimate poststratified by habitat and fish density) were plotted in relation to the total abundance estimator that was most precise in order to investigate the influence of sample size on the relative precision of the three total abundance estimators. Indices of abundance Three indices were constructed for each species in each year to determine interannual variations in relative abundance (mean CPUE): an all- site index, a habitat index, and a HFD index. For each species and year, the all-site index was the mean CPUE from all sites sampled. The habitat index was the mean CPUE from all sites sampled within the species' habitat area. The HFD index was the mean CPUE from all sites sampled within the species' HFD area. CPUE values were not normally distributed and therefore the Kruskal-Wallis nonparametric analysis of variance test was used to test the three indices for each species' differences in mean CPUE among years. For species that showed significant differences (o=0.05), a Tukey HSD (honestly significant difference) multiple comparison test for unequal sample sizes was conducted to determine which years differed (a=0.05). The Tukey multiple comparison test was used because it is robust with respect to departures from population normality and homogeneity of variance (Keselman, 1976). The results for the three indices for each species were com- pared to see how the differences in estimating abun- dance with the three indices affected conclusions of significant differences in abundance between years. Numerous sources of bias can affect estimators of abundance from survey data. The poststratification estimator and other design-based estimators may be biased when applied to data that were not collected under a probability sampling design, as done in the present study. For a qualitative estimate of possible design bias in the estimates, the annual proportion of sample sites in each stratum (habitat, nonhabitat, HFD, and LFD strata) were compared with the proportion of area (km2) in that stratum. First, we compared the size of the habitat area, in relation to the size of the total survey area, with the number of samples taken in the habitat area, in relation to the number taken in the total survey area. number of samples taken in the HFD area, in relation to the number taken in the total habitat area. Size of the HFD area Size of the habitat area Number of samples taken in the habitat area Size of the total survey area Number of samples taken in the total survey area Second, we compared the size of the HFD area, in relation to the size of the total habitat areas, with the Number of samples taken in the HFD area Size of the habitat area Number of samples taken in the habitat area Recognizing that the distribution of individuals var- ied within and across strata, two measures were used to better understand the distribution of each species in each year. The proportion of zero catches (e.g., a "zero catch" for rock sole indicates a tow in which no rock sole were caught) and the mean CPUE of nonzero catches were calculated for each species in each year over four areas: the total survey area, the habitat area, the HFD area, and the LFD area. Results Fish CPUE statistics were calculated for a total of 244 quantitative tows over the six sampling years (Fig. 2) in habitats ranging from 1 to 111 m depth and from 0% to 99% sand. Based on compiled data from all six years, the habitat area for rock sole was defined by 1-84 m depth and 2-99% sand; for yellowfin sole, by 2-43 m depth and 24-99% sand; for Pacific halibut, by 2-27 m depth and 2-99% sand; and for flathead sole, by 12-87 m depth and 8-97%. sand (Fig. 3). The HFD area, defined by depth and percent sand, was determined for each of the four species in each of the six years (Table 1, Fig. 3). Although the range of depth and the range of percent sand were determined independently in each year, they remained quite constant for each species over the six sampling years. The size of habitat area in relation to total area ranged across species from 0.62 to 0.92 and, for each species, the proportion of habitat sites to total sites varied among years (Table 2). The proportion of sample sites in habitat to sample sites in the total survey area ranged from 0.88 to 1.00 for rock sole, 0.60 to 0.87 for yellowfin sole, 0.52 to 0.93 for Pacific halibut, and 0.29 to 0.67 for flathead sole. The relative number of samples taken in each species' habitat area exceeded the relative size of their habitat area (i.e., a positive disproportion of samples in habitat), except for rock sole in 1991 and 1994, yellowfin sole in 1993 and 1994, Pacific halibut in 1993 and 1994, and all years for flathead sole. On average, rock sole had a 5% positive disproportion of samples in its habitat area, yellowfin sole and Pacific halibut had an 11% positive dispropor- tion of samples in their habitat area, and flathead sole had a 15% negative disproportion of samples in its habitat area. The size of the HFD area in relation to habitat area, and the number of sites sampled in the HFD area in relation to the number sampled in the entire habitat area, varied over the six sampling years for each of the four species (Table 2). On average over the six years, Dressel and Norcross: Using poststrafication to improve abundance estimates from multispecies surveys 475 Q. CD a 1.' 40 ..: / A AA* AA*A A A A*- • * ..... A * ■ Rock sole zero catch A nonzero catch habitat area Q. 60 CD Q Percent sand in substrate VL^yzt&m • •- , I high fish-density Yellowfin sole zero catch A nonzero catch Percent sand in substrate 0. 2 ■**"■ A.. *•■ Iff* 5^ Q- 60 Q so • I • • ; habitat area ! i high fish-density Pacific halibut zero catch A nonzero catch I habitat area i | high fish-density Percent sand in substrate & 60 Q I- - -A-AI A !*** A -•"I*" : i Flathead sole i A*A . A A* A . A*A i A A-*- -A-A_> A' ' A \ zero catch A nonzero catch [ I habitat area J ' high fish-density Percent sand in substrate Figure 3 Summary of 1991-96 tows, in relation to depth and percent sand. Tows are divided into zero and nonzero catches for each species. The dotted line separates the depth and percent sand characteristics of habitat and nonhabitat areas. The dashed line separates the depth and percent sand characteristics of high and low fish-density areas within the habitat area. rock sole had a 10% negative disproportion of samples in the HFD area, Pacific halibut had a 3% negative disproportion of samples in the HFD area, and flathead sole had a 28% negative disproportion of samples in the HFD area. For yellowfin sole, the average distribution of samples between the high and low fish-density areas was in direct proportion to the size of the areas, i.e., there was no disproportion of samples. Two measures were used to characterize the distribu- tion of a species within their habitat: the proportion of zero catches and the mean of nonzero catches in high and low fish-density areas. As expected, for all species the average proportion of zero catches over all sites was greater than the proportion of zero catches in the habitat or HFD areas (Table 3). For rock sole, yellowfin sole, and flathead sole, the average proportion of zero 476 Fishery Bulletin 103(3) Table 1 Characteristics defining 1991 -96 high fish-den ?ity areas for each species of flatf sh. Ranges of depth and percent sand, defining the high fish-density (HFD) area, and the associated spat al coverage within the bay (km2). Each year's HFD area was deter- mined as the range of depth and percent sand associated with the 75lh-100th percentile of nonzero catch from the other five years. Species Year Depth (m) Percent sand in sediment Size (km2) minimum maximum minimum maximum Rock sole 1991 3.0 27.3 31.5 99.2 52 iLepidopsetta spp. ) 1992 3.0 36.0 20.2 99.2 56 1993 3.0 27.3 31.5 99.2 52 1994 3.0 27.3 31.5 98.8 52 1995 3.0 27.3 31.5 99.2 52 1996 3.0 25.0 47.8 99.2 46 average 3.0 28.3 32.4 99.2 52 Yellowfin sole 1991 1.7 23.0 40.5 98.6 33 iPleuronectes asper) 1992 2.3 25.0 24.2 86.7 29 1993 2.3 25.0 24.2 86.7 29 1994 2.3 25.0 24.2 86.7 29 1995 2.3 25.0 24.2 86.7 29 1996 2.3 25.0 24.2 86.7 29 average 2.2 24.7 26.9 88.7 30 Pacific halibut 1991 2.5 25.0 52.3 99.3 39 l Hippoglossus stenolepis ) 1992 2.3 27.0 52.3 99.3 41 1993 2.3 27.0 52.3 99.3 41 1994 2.3 27.0 52.3 99.3 41 1995 2.0 27.0 64.6 99.3 33 1996 2.3 25.5 52.3 98.4 37 average 2.3 26.4 54.4 99.1 39 Flathead sole 1991 19.8 87.0 17.4 89.1 42 (Hippoglossoides elassodon ) 1992 25.5 87.0 10.7 89.1 38 1993 19.8 87.0 8.4 70.7 34 1994 19.8 67.5 10.7 89.1 40 1995 19.8 87.0 17.4 89.1 42 1996 19.8 64.0 17.4 89.1 39 average 20.8 79.9 13.7 86.0 39 catches in the LFD area was higher than in the HFD area. For Pacific halibut, the average proportion of zero catches remained approximately constant across the entire habitat area. The relative mean nonzero catch between the LFD and HFD areas varied across species, ranging from 37% to 82% (Table 4). In each of the 24 species-year combinations, three esti- mates of population abundance were compared, except for flathead sole in 1992 when no samples were taken in the flathead sole HFD area (Fig. 4). In every case in which the proportion of habitat stratum-size sites to total study area sites exceeded the proportion of habitat stratum size to total study area size (Table 2), the unstratified estimate was greater than the estimate poststratified by habitat (Fig. 4). In every case that the proportion of habi- tat stratum sites to total study area sites was less than the proportion of habitat stratum size to total study area, the unstratified estimate was less than the estimate poststratified by habitat. Similarly, in every case that the proportion of HFD stratum sites to habitat stratum sites exceeded the proportion of HFD stratum size to habitat stratum size (Table 2), the estimate poststratified by habitat was greater than the estimate poststratified by habitat and fish density (Fig. 4). In all but two cases in which the proportion of HFD stratum sites to habitat stratum sites was less than the proportion of HFD stra- tum size to habitat stratum size, the estimate poststrati- fied by habitat was less than the estimate poststratified by habitat and fish density. The two exceptions were for Pacific halibut in 1991 and 1996, where the difference between poststratified estimates was small. In 1991, the estimate poststratified by habitat was 2.9% (8116 fish) greater than the estimate poststratified by habitat and fish density; in 1996, it was 0.56% (4905 fish greater). Dressel and Norcross: Using poststrafication to improve abundance estimates from multispeaes surveys 477 Table 2 A comparison of the relative nt mber of sample site i and relative size ( km2 ) of the habitat area high fish-density ( HFD ) area, and total study area. Comparisons include the size of the habitat area versus the size of the study area, the number of sites sampled in the habitat area versus the number sampled in the total study area, the size of the HFD area versus ;he size of the habitat area, and the number of sites e amplec in the HFD area versus he nu mber sampled in the habitat area. Habitat sites/Total sites Year ha size/ total stuuy sizeiiuii-j Species All years 1991 1992 1993 1994 1995 1996 average Rock sole (Lepidopsetta spp. ) 0.92 0.92 1.00 1.00 0.88 1.00 1.00 0.97 Yellowfin sole (Pleuroneetes asper) 0.66 0.78 0.87 0.63 0.60 0.80 0.80 0.75 Pacific halibut (Hippoglossus stenolep s) 0.62 0.73 0.93 0.58 0.52 0.80 0.80 0.73 Flathead sole (Hippoglossoides elassodon ) 0.67 0.43 0.29 0.67 0.56 0.60 0.60 0.52 High fish-density size/Habitat size (km2) High fish-density sites/Habitat sites Species Year Year 1991 1992 1993 1994 1995 1996 average 1991 1992 1993 1994 1995 1996 average Rock sole 0.65 0.70 0.65 0.65 0.65 0.58 0.65 0.64 0.73 0.42 0.45 0.55 0.50 0.55 (Lepidopsetta spp.) Yellowfin sole 0.58 0.51 0.51 0.51 0.51 0.51 0.52 0.76 0.46 0.47 0.40 0.50 0.50 0.52 (Pleuroneetes asper) Pacific halibut 0.72 0.76 0.76 0.76 0.61 0.69 0.72 0.69 0.86 0.79 0.62 0.63 0.54 0.69 (Hippoglossus stenolepis ) Flathead sole 0.72 0.66 0.59 0.69 0.72 0.67 0.68 0.52 0.00 0.50 0.43 0.50 0.44 0.40 (Hippoglossoides elassodon ) Calculations of relative efficiency among the three to- tal abundance estimators showed increases in estimated precision with stratification (Table 5). In most cases (18 out of 24), the estimate poststratified by habitat was more precise (corresponding to a lower standard error in Fig. 5) than the unstratified estimate. Of the 16 (of 23) cases in which the precision of both poststratified estimates were greater than that of the unstratified estimate, in half the estimate poststratified by both habitat and density was more precise than the estima- tor poststratified by habitat alone. Sample sizes across the survey area and in each sub- area (habitat, high fish-density, and low fish-density areas) (Table 6) strongly influenced the precision of es- timates. Habitat sample sizes for all species-year combi- nations ranged from 4 to 45 (proportion of samples tak- en in habitat ranged from 0.286 to 1.000); HFD sample sizes ranged from 0 to 29 (proportion of samples taken in the HFD area ranged from 0.0 to 0.8); and LFD sample sizes ranged from 4 to 16 (proportion of samples taken in the LFD area ranged from 0.125 to 0.583). Although the number of samples in both the high and low fish-density areas (Fig. 6, A and B) likely affected estimates poststratified by habitat and fish density, the number of samples in the HFD area appears to have had the primary influence on the precision of estimates. The species-year combinations for which the unstrati- fied estimate was the most precise occurred when habi- tat sample sizes ranged from 4 to 22 (Fig. 7) and HFD stratum samples sizes ranged from 6 to 11 (Fig. 6A). The species-year combinations for which the estimate poststratified by habitat was the most precise occurred when habitat sample sizes ranged from 12 to 30 (Fig. 7) and when sample sizes in the HFD stratum ranged from 6 to 15 (Fig. 6A). The species-year combinations for which the estimate poststratified by habitat and fish density was most precise occurred when habitat sample sizes ranged from 15 to 45 (Fig. 7) and HFD stratum sample sizes ranged from 10 to 29 (Fig. 6A). Estimates poststratified by habitat and fish density were the most precise for all three cases in which the HFD stratum sample size was greater than 20 (corresponding to LFD stratum sample sizes ranging from 9 to 16) (Fig. 6, A and B). Both of the poststratified estimates were more precise than the unstratified estimate when habitat stratum sample sizes were greater or equal to 24 (Fig. 7) and when HFD stratum sizes were greater or equal to 12 (Fig. 6A). Statistically significant changes in annual abundance varied among indices and species. There were signifi- cant changes in annual mean CPUE in all indices for rock sole and Pacific halibut, in two indices for yellowfin sole and in no indices for flathead sole (Table 7). Rock sole abundance was significantly greater in 1992 than all other years except 1996. Individual indices indicated that rock sole 1996 abundance was greater than that 478 Fishery Bulletin 103(3) a — fa w 8a C o M o Q 5 x ■ — -a i -a ai A tin 2 X - CO o o w J3 Bo cfl n c cti = >> — > 3 s- :. 3 ~ to cti Si ■So: e. o £ o. o r a * BO 0) :*> > en -a c ft a. a ctl 53 3 tit a & < "S « - fa Q fa H S £ '3 «? a; H ? n fa E-i 3 o m CN CM CO IO CM CO CO CO 3 o CO CN o o o o o o o CO CM 1 o o o t- CD CO CO CO o in CM r CN o o o o o o to CO d CM d i-l Tf CO ^H d d o CM in d to in O CO O CO to c- d d - d o o o d CN d CN d to d o o o d o cm m d co c- -* CO CO o O »-H, CN odd o CO r-l m co cn cn co ^r odd - in CM i-l rH CM m CO o o o o o O o 05 CO O o o o o o o o o o o to o CO in o o o o o o o o to m iH CO CO I— 1 in CO CO CM o m o CO CO 00 H O o o o o o o CM CM CO CO 1— 1 m co o CO CO o in o CO CO CM O o o o o o O of 1992 and 1993. Tukey post hoc tests on the yellowfin sole all-site and habitat indices showed that 1991 yel- lowfin abundance was greater than that of 1994. All three indices showed that Pacific halibut abundance was greater in 1995 than in 1991 and 1993. Individual indices also indicated that Pacific halibut abundance was greater in 1995 than in 1992, 1994, and 1996. Discussion The Chiniak Bay multispecies survey was designed to estimate the abundance of four species with equal emphasis. Because the distribution of species varied greatly throughout the bay, what might have been an optimal stratification for individual species was com- promised to develop a stratification scheme that was as effective as possible for all target species. Because we do not believe sampling was optimal for any one of the spe- cies, a poststratification method of analysis was investi- gated to increase the precision of abundance estimates for each species individually and to account for possible bias due to the uneven and nonrandom distribution of sampling sites over space and time. The need for stratification and the concern about the distribution of sampling sites arise because of the varying distributions of species in the study region. Knowledge of the spatial distributions of species is im- portant when estimating abundance from trawl surveys. A random distribution of individuals is often taken as a starting point for defining spatial distributions in ecology (Taylor et al.. 1978). It is also a primary as- sumption for many survey sampling designs and analy- sis measures. The assumption of randomly distributed individuals often is not appropriate, however, because the concentration of fish varies over time and space in relation to environmental factors (Murawski and Finn, 1988; Gadomski and Caddell, 1991; Reichert and van der Veer, 1991; Norcross et al., 1999). If habitat (Fiedler and Reilly, 1994; Reilly and Fiedler, 1994) and related spatial population density distributions (Buckland and Anganuzzi, 1988) are not accounted for when calculat- ing abundance estimates, precision can decrease and results can be seriously biased. Inaccurate results can have strong management repercussions. In situations such as that of the present study, where the sample does not properly represent the population, poststratification is appropriate (Scheaffer et al., 1996). By comparing poststratified and unstratified estimates of abundance, we found that in every species-year com- bination for which the three estimates of abundance dif- fered (Fig. 3), the poststratified estimates reduced the effect of the disproportion of samples allocated between habitat and nonhabitat areas and between high and low fish-density areas. For instance, in 1992, a dispropor- tionately large number of samples were taken in Pacific halibut habitat (Table 2). We suspect, therefore, that the unstratified estimate of abundance was an overes- timate of true population abundance. The disproportion- ately large number of samples taken in Pacific halibut Dressel and Norcross: Using poststrafication to improve abundance estimates from multispecies surveys 479 20,000,000 15,000,000 10,000,000 5,000,000 - 0 2,000,000 • 1,500,000 ■ 1 ,000,000 ■ 500,000 -| 0 Rock sole DUnstratified ■ PoststraWied(H) ■ Poststratified(D) 1991 1992 1993 1994 1995 1996 Yellowfin sole fra^foffi D Unstratified ■ Poststratified(H) IPoststratitied(D) 2,500,000 2,000,000 1 .500,000 1 ,000,000 500,000 0 1991 1992 1993 1994 1995 1996 Pacific halibut D Unstratified ■ Poststratitied(H) ■ Poststratitied(D) 1991 1992 1993 1994 1995 1996 2,500.000 -I Flathead sole 2.000.000 • -■ -■ 1.500,000 • 1 .000.000 ■ ffirf\ jM ttI 500,000 - 0 - u i a - DUnstratified ■ Poststratified(H) ■ Poststratitied(D) 1991 1992 1993 1994 Year 1995 1996 Figure 4 Three estimates of total abundance and standard error. Estimates are unstratified, poststratified by habitat (poststratified [H]), and poststratified by habitat and fish density (poststratified [£>]). habitat was adjusted by poststratifying by habitat. The estimate poststratified by habitat was less than the unstratified estimate of abundance, as we suspect the true abundance was. Poststratification by habitat and neighboring years' halibut density adjusted not only for the disproportionately large number of samples in the habitat area but also for the disproportionately large number of samples in the HFD area (Table 2). The es- timate poststratified by habitat and halibut density was less than both the estimate poststratified by habitat and the unstratified estimate, as we suspect was the case for the true Pacific halibut abundance. In 1992, the number of samples in yellowfin sole habitat was disproportionately large, but the number of samples in the HFD area was disproportionately small (Table 2). In this case, we suspect the unstrati- fied estimate of abundance was an overestimate of true abundance because of the overabundance of samples in the habitat area. We also believe, however, that it was not a very large overestimate because of the dis- proportionately small number of samples in the HFD area. Poststratifying by habitat adjusted for the dis- proportionately large number of samples in the habitat area and produced an estimate that was less than the unstratified estimate. Poststratifying by habitat and fish density adjusted for both the disproportionately large number of samples in the habitat area and the disproportionately small number of samples in the HFD 480 Fishery Bulletin 103(3) Table 4 The mean catch per unit of effort (CPUE) of nonzero catches in the habitat, high fish-density (HFD), and low fish-density (LFD) areas and the proportion of the mean CPUE of nonzero catches in LFD and habitat areas in relation to those i n the HFD area. Species Pacific halibut Flathead sole Rock sole Yellowfin sole ( Hippoglossu s ( Hippoglossoides (Lepidopsetta spp.) iPleuroneetes asper) stenolepis) el as sod on ) Habitat nonzero mean 85.3 15.6 16.8 16.0 HFD nonzero mean 105.4 20.7 17.9 20.6 LFD nonzero mean 52.2 7.6 14.7 9.5 Habitat mean/concentration mean 0.81 0.75 0.94 0.78 LFD mean/HFD mean 0.50 0.37 0.82 0.46 Table 5 L'nstratified total abun mates poststratified by dance estimates (£/), total abundance estimates poststratified by habitat (H), and total abundance esti- habitat and fish density (D) are compared by using annual relative efficiency statistics. Species Relative efficiency comparison Year 1991 1992 1993 1994 1995 1996 Rock sole HtoU 1.076 1.081 1.084 0.985 1.083 1.084 iLepidopsetta spp.) DtoH 1.129 1.496 0.865 0.834 1.089 0.999 DtoU 1.214 1.618 0.937 0.821 1.179 1.084 conclusion D>H>U D>H>U H>U>D U>H>D D>H>U H>D>U Yellowfin sole HtoU 1.318 1.317 0.988 0.922 1.266 1.324 iPleuronectes asper) DtoH 1.111 0.969 0.890 0.715 0.936 0.967 DtoU 1.465 1.277 0.880 0.659 1.186 1.280 conclusion D>H>U H>D>U U>H>D U>H>D H>D>U H>D>U Pacific halibut HtoU 1.272 1.521 1.007 1.369 1.607 1.440 [Hippoglossus stenolepi s) DtoH 1.029 0.936 0.996 0.792 1.340 0.949 DtoU 1.309 1.424 1.003 1.084 2.155 1.366 conclusion D>H>U H>D>U H>D>U H>D>U D>H>U H>D>U Flathead sole HtoU 0.726 0.449 1.075 0.973 1.025 1.056 [Hippoglossoides elassodon ) D to H 0.786 — 0.976 0.705 0.746 0.992 DtoU 0.571 — 1.049 0.686 0.765 1.047 conclusion U>H>D U>H H>D>U U>H>D H>U>D H>D>U area. As a result, the estimate poststratified by habitat and fish density was greater than the estimate post- stratified by habitat, but lower than the unstratified estimate. According to our results, it is unlikely that the estimates poststratified by habitat and fish density were the most representative estimates of abundance because poststratification adjusted for the disproportion- ate distribution of samples between areas. Another reason to poststratify the data is to increase the precision of abundance estimates. Poststratified estimates in our study were generally more precise than unstratified estimates, given sufficient sample sizes (Table 5). Poststratification by habitat character- istics increased the precision of abundance estimates in three-quarters of all species-year combinations. This finding indicates a close link between habitat type and fish abundance and agrees with poststratification re- sults in other studies (Pollock et al., 1994; Reilly and Fiedler, 1994). Estimates poststratified by both habitat and fish density were also generally more precise than Dressel and Norcross: Using poststrafication to improve abundance estimates from multispecies surveys 481 Rock sole □ Unstratified ■ Poststratified(H) ■ Poststratitied(D) 1991 1992 1993 1994 1995 1996 Yellowfin sole D Unstratified ■ Poststratilled(H) ■ Poststratified(D) 1991 1992 1993 1994 1995 1996 Pacific halibut D Unstratified ■ Poststratified(H) ■ Poststratified(D) 1991 1992 1993 1994 1995 1996 Flathead sole 1991 1992 1993 1994 1995 1996 Year Figure 5 Three standard error estimates of annual total abundance. Standard error estimates are for the unstratified, poststratified by habitat (poststratified [H]), and poststratified by habitat and fish density (poststratified [D]) estimates. unstratified estimates but were not consistently more precise than the estimates poststratified by habitat alone. The six cases in which estimates poststratified by habitat and fish density were the most precise show that some species have strong density gradients within habitat areas and that the incorporation of fish density information from neighboring years can be beneficial for increasing precision. Being able to predict the distribu- tion of fish density in one year from that of neighboring years indicates annual consistency in species distribu- tion in relation to habitat characteristics. The present study indicates that when estimating abundance from haphazardly sampled data, the estima- tor poststratified by habitat is superior to the unstrati- fied estimator regardless of sample size. The estimate poststratified by habitat was more precise than the un- stratified estimate in 18 of the total 24 species-year combinations. These 18 species-year combinations oc- curred across nearly the full range of habitat stratum sample sizes, from 12 to 45. The six cases in which the estimate poststratified by habitat was less precise than the unstratified estimate were affected by the propor- 482 Fishery Bulletin 103(3) 35 -i 0) A N > 15- A c A JA . 8- AA AA to A oj 6 A A ■a A ■g 4- A A **— A O 2 A _l 0 T 1 1 1 ' Unstratified Poststratified (H) Poststratified (D) Most precise estimate of total abundance Figure 6 High and low fish-density stratum sample size in relation to the most precise estimate of total abundance. The (A) high fish-density and IB) low fish-density stratum sample size for each species-year com- bination is plotted in relation to the most precise estimate of total abundance — the unstratified estimate, the estimate poststratified by habitat (poststratified [H]), or the estimate poststratified by habitat and fish density (poststratified [D]). tion of samples in unsuitable habitat. As a measure of variability, the magnitude of the variance is dependent on the magnitude of the data (Zar, 1996). Thus, the variances of trawl catches decrease as the observed means decrease (Taylor, 1953). A lower variance, there- fore, does not necessarily indicate a better estimator, but instead may reflect lower population abundance. In the six cases in this study where the variance of the unstratified estimate was less than the variance of the estimate poststratified by habitat, the unstratified abun- dance estimate was less than the abundance estimate poststratified by habitat. The low unstratified abundance estimates in these six cases were the result of a dis- proportionately large number of samples in nonhabitat areas in relation to the size of the nonhabitat areas. Therefore, although the unstratified estimate was more precise, it was also likely to be an underestimate of the true abundance. Thus, we suggest that the estimate poststratified by habitat is the most desirable estimator in these situations, despite the decrease in precision in relation to the unstratified estimator. In many cases, small sample size was likely the rea- son that the estimates poststratified by habitat and fish density were not the most precise of the three estimates. Poststratification produces precise estimates when the overall sample size and the sample size in each stratum are large (Scheaffer et al., 1996). In our study, the esti- mator poststratified by habitat and fish density was the most precise estimator of the three when sample size in the HFD stratum was 20 or greater and the sample size in the LFD stratum was 9 or greater. The number of samples in the HFD stratum appears to have had a larger influence on the precision of estimates stratified by habitat and fish density than the number of samples Dressel and Norcross: Using poststrafication to improve abundance estimates from multispecies surveys 483 su - 45 ■ A 40- A 35 ■ A 30- A 25- AAA 20 ■ i A A is- t AA i le- s' 0- A f ,. . , Unstratified Poststratified (H) Poststratified (D) Most precise estimate of total abundance Figure 7 The habitat stratum sample size for each species-year combination is plotted in relation to the most precise estimate of total abundance — the unstrati- fied estimate, the estimate poststratified by habitat (poststratified [H]), or the estimate poststratified by habitat and fish density (poststratified [D]). Table 6 Annual number of tows made across all strata in habitat and nonhabitat strata, and in the high and low fish-density strata within the habitat stratum. Species Stratum Year 1991 1992 1993 1994 1995 1996 Rock sole all 49 15 24 25 20 30 iLepidopsetta spp.) habitat and nonhabitat 45 and 4 15 and 0 24 and 0 22 and 3 20 and 0 30 and 0 high fish density and 29 and 16 11 and 4 10 and 14 10 and 12 11 and 9 15 and 15 low fish density Yellowfin sole all 49 15 24 25 20 30 iPleuronectes habitat and nonhabitat 38 and 11 13 and 2 15 and 9 15 and 10 16 and 4 24 and 6 asper ) high fish density and low fish density 29 and 9 6 and 7 7 and 8 6 and 9 8 and 8 12 and 12 Pacific halibut all 49 15 24 25 20 30 I Hippoglossus habitat and non-habitat 36 and 13 14 and 1 14 and 10 13 and 12 16 and 4 24 and 6 stenolepis) high fish density and low fish density 25 and 11 12 and 2 11 and 3 8 and 5 10 and 6 13 and 11 Flathead sole all 49 14 24 25 20 30 (Hippoglossoides habitat and non-habitat 21 and 28 4 and 10 16 and 8 Wand 11 12 and 8 18 and 12 elassodon ) high fish density and low fish density 11 and 10 0and4 8 and 8 6 and 8 6 and 6 8 and 10 in the LFD stratum (Fig. 6, A and B). This study sup- ports the conclusion of Scheaffer et al. (1996) but also indicates that the sample size in the HFD stratum may have a larger influence on the precision of the resultant estimate. As concluded in other studies (Fiedler and Reilly, 1994; Pollock et al., 1994; Reilly and Fiedler, 1994; Bernard et al., 1998), we found that poststratification can provide increased precision and decreased bias for estimates. Small stratum sample sizes, however, can make it impossible to detect heterogeneity among strata and fail to give increased precision (Powell et al., 1995; Friedland et al., 1999). The wide range of sample sizes among strata across species-year combinations exempli- 484 Fishery Bulletin 103(3) Table 7 Kruskal-Wallis test statistics for differences in annual relative abundance and, for significant Krust corresponding significant Tukey post hoc pairwise differences. Statistics were calculated for the all-sitf density indices. al-Wallis statistics, the , habitat, and high fish- Species Index Kruskal-Wall ('^indicates statistically signi is ficant difference) Tukey post hoc significant differences Rock sole All-site P=0.0003* 1992>1991(P<0.0006) (Lepidopsetta spp. ) 1992>1993(P<0.0001) 1992>1994(P<0.0009) 1992>1995(P<0.0124) 1996>1993(P<0.0301) Habitat P=0.0008* 1992>1991(P<0.0012) 1992>1993(P<0.0001) 1992>1994(P<0.0022) 1992>1995(P<0.0149) 1996>1993(P<0.0351) High fish density P=0.0035* 1992>1991 (P<0.0005) 1992>1993(P<0.0003) 1992>1994(P<0.0127) 1992>1995(P<0.0206) 1996>1992 (P<0.0145) Yellowfin sole All-site P= 0.0033* 1991>1994(P<0.0096) {Pleuronectes asper) Habitat P= 0.0022* 1991>1994lP<0.0374) High fish density P=0.1240 Pacific halibut All-site P= 0.001* 1995>1991(P<0.0013) tHippoglossus stenolepis) 1995>1993(P<0.0012) 1995>1994(P<0.0359) Habitat P=0.0004* 1995>1991(P<0.0018) 1995>1993(P<0.0077) High fish density P=0.0002* 1995>1991(P<0.0002) 1995>1992(P<0.0127) 1995>1993(P<0.0004) 1995>1996(P<0.0249) Flathead sole All-site P=0.1955 (Hippoglossoides elassodon) Habitat P=0.2950 High fish density P=0.5151 fies an important drawback to using the poststratifica- tion method. Because strata criteria are unknown when sampling, it is not possible to insure that there will be sufficient samples in each poststratified stratum. When resulting sample sizes in some strata are small, post- stratification may be ineffective at increasing precision. If the resulting sample size in one or more strata is one, the poststratification variance will be inestimable. If the resulting sample size in one or more strata is zero, poststratification may not be possible. Because sample size is a limiting factor for increased precision with poststratification, there are strong impli- cations for survey design. Many multispecies surveys are conducted by using a stratified random sampling de- sign. There are two ways to apply poststratification to a stratified survey. First, for an unbiased estimator, each stratum of the stratified survey can be poststratified individually (Cochran, 1977). For the poststratification estimator to have increased precision beyond that of stratified random sampling, each of the original strata must have a large number of samples to allow suffi- cient samples in each poststratified stratum. Therefore, investigators who intend to poststratify data within a stratified random survey for unbiased estimates need to construct large strata with many samples in the original sampling design. Second, if poststratification is applied to data that were not collected under a prob- ability sampling design, the estimator may be more precise, but may be biased. For the analysis of data that were not collected under a probability sampling Dressel and Norcross: Using poststrafication to improve abundance estimates from multispecies surveys 485 design, developing an index of relative abundance from all samples, or samples in the habitat or HFD areas, is an easy and effective way to estimate statistically significant changes in abundance among years. To de- termine which tows should be included in an index to effectively approximate the variations in the annual total abundance estimates, it is helpful to compare the size of the habitat area over years and to study the dis- tribution of species density within the habitat area. The goal of creating an index should be to include the most information possible, while avoiding undue influence from the haphazard distribution of sample sites. If the total study area is the same in each year, the choice of whether to use the all-site index should depend on whether the size of the habitat area is constant over the compared years. In this study, the defined habitat area for each species was the same over the six years compared. Therefore, for an index of relative abundance, the habitat index retained all necessary information and reduced possible bias due to the disproportionate distribution of haphazard samples between habitat and nonhabitat areas. When a temporally dependent strati- fication variable, such as temperature, is used to define the placement of stratum boundaries, however, the size of the habitat area may vary between years. If the an- nual size of the habitat area varies, some common size would need to be chosen for the relative index to ap- proximate the annual changes in the total abundance estimates. The all-site index could be used for this purpose, but the index will be affected by any dispro- portionate distribution of samples between habitat and nonhabitat areas. Another possible way to do this would be to include all tows from the habitat area each year, plus as many zero catches from the nonhabitat area necessary to be proportional to the annual size of the nonhabitat area. Such an approach would not depend on actual tows in nonhabitat area but would depend on the estimated size of the habitat and nonhabitat areas and the sample size in the habitat area. If the size of habitat area is the same in each year, the choice of whether to use the habitat index should depend on whether the distribution of species density is constant throughout the habitat area. If a species' density distribution is approximately constant across the habitat area, a haphazard distribution of sample sites should have little influence. Constructing an index from all habitat tows may then be desired to retain the largest sample size and the most information possible. Alternatively, if a species has a strong density gradient within its habitat area, a disproportionate distribution of sites in relation to the size of high and low fish-den- sity areas may provide an unrepresentative estimate of abundance from the habitat index. In this case, if a suf- ficient number of samples are taken in the HFD area, constructing an index from samples within the species' HFD area alone may provide an effective index while minimizing the effect of a disproportional distribution of haphazard samples within the habitat area. A comparison of the number of zero catches and the mean nonzero catch between the high and low fish- density areas provides information about the density distribution of species within a habitat area. The pro- portion of zero catches of rock sole, yellowfin sole, and flathead sole and the mean nonzero catch between high and low fish-density areas indicated density gradients within the habitat areas. Unlike these three species, the proportion of Pacific halibut zero catches was ap- proximately the same in the HFD area as across the entire habitat area and the difference in mean nonzero catch between low and high fish-density areas was only approximately half that of the other species. Therefore, it appears that the Pacific halibut density distribution across the defined habitat area varied little compared with the other three species. In this study, we suggest that the habitat index was the most appropriate for all four species. For each spe- cies in our study, the size of the habitat area remained the same across all six years. Thus, the habitat index eliminated the influence of disproportionately allocated samples in habitat and nonhabitat areas. For Pacific halibut, the relatively homogenous distribution of abun- dance across the habitat area indicates that the effect of disproportionate samples between high and low fish- density areas is small and that samples across the entire habitat area are helpful in describing annual differences in abundance. For rock sole, yellowfin sole, and flathead sole, the difference in the proportion of zero catches and nonzero mean abundance between the high and low fish-density areas was considerable. As a result, differences in annual abundance suggested by the habitat index may be affected by the inconsistent disproportion of samples between high and low fish-den- sity areas over years. Although it would be preferable to use the HFD index in these cases, annual sample sizes in the HFD area were so small that we recommend the habitat index instead. Recognizing that the habitat index will not account for the annual disproportion of samples between the high and low fish-density areas, we used the comparison of the size and the number of samples taken in high and low fish-density areas to flag differences in annual index abundance estimates that might be over- or underestimates. If this method is ap- plied in a management context, the levels of the factors describing the density distribution of the species (i.e., difference in the percent of zero catches and the percent difference in mean nonzero catch between years) can be set as criteria and kept constant over years to elimi- nate subjectivity between years or between species. For example, if the percent of zero catches in high and low fish-density regions differ by 40% and the mean nonzero catch in the HFD area is 30% greater than that in the LFD area, the HFD index should be used. Otherwise, the habitat index should be used. For many surveys, identifying habitat and fish-density areas for poststratification and index construction is pos- sible with currently available information. The estima- tion methods used in the present study can be applied to any survey for which abundance and environmental measurements are available for each sampled site and the environmental measurements are related to species 486 Fishery Bulletin 103(3) abundance in a consistent way. For example, the NMFS Bering Sea trawl survey includes measurements of depth and surface and bottom temperatures at all trawl sites (Goddard and Walters4) that could be used for post- stratification. Similarly, the Pacific West Coast trawl survey includes measurements of surface and bottom temperature and salinity at all stations (Lauth et al.5) that could be used. Poststratification allows for use of a wide range of stratification variables, including tempo- rally dependent variables that are not available before sampling is complete, e.g., temperature and salinity. For surveys where habitat information is not collected at trawl sites, habitat information from other sources can be paired with fish distribution information after collections have been made. For instance, when habitat information is available, but has not been collected at each site, spatial statistics can be used to krige the habitat information over the study area and to predict the specific habitat data value at the sampling sites. If there is a consistent relationship between species abun- dance and the habitat variable, the catch and habitat data paired at sample sites can then be used to identify areas of suitable habitat and areas of high fish density within suitable habitat. How well habitat and HFD areas are estimated will depend on the number and distribution of habitat measurements, the contouring algorithms used, and the estimates of areas within contours. Even if species are not distributed in direct response to particular environmental characteristics, the characteristics may serve as proxies for effects that are more difficult to measure (Perry and Smith, 1994). Once habitat and HFD areas are identified, poststrati- fication can be conducted for total abundance estimates, and statistically significant changes between years can be assessed with an index of relative abundance. These methods could yield more accurate estimates of abun- dance for use by managers. The goal of most sampling plans is to provide statistical estimates with the small- est possible confidence limits at the lowest cost (Krebs, 1989). Thus, being able to use data collected indepen- dently of a survey should be appealing. The NRC (2000) recommends using data from com- mercial or sportfishing vessels in scientific assessments of abundance. A primary difficulty in using commercial fisheries data for scientific estimates of abundance is that the data do not represent random samples of the fish population. As a result, commercial fisheries data 4 Goddard, P., and G. Walters. 1998. 1995 bottom trawl survey of the eastern Bering Sea continental shelf. AFSC Processed Report 98-08, 170 p. Resource Assessment and Conservation Engineering Division, Alaska Fisheries Science Center, NMFS, NOAA, 7600 Sand Point Way N.E., Seattle, Washington, 98115. 5 Lauth, R. R., M. E. Wilkins, and P.A. Raymore Jr. 1997. Re- sults of trawl surveys of groundfish resources of the West Coast upper continental slope from 1989 to 1993. NOAA Tech. Memo. NMFS-AFSC-79, 342 p. National Technical Information Service, U.S. Department of Commerce, 5285 Port Royal Road, Springfield, Virginia 22161. present a biased perspective of the population that may change over time and may not correlate well with ac- tual fish abundance (NRC, 2000). Although commercial fishery-dependent data may provide biased estimates of abundance, fishery-dependent data also provide large sample sizes and a wide range of information not avail- able from other sources. For example, commercial and sportfishing data often provide broader geographic and temporal coverage. Poststratification of haphazard data from commercial and sportfishing sources may be one way to reduce inherent bias and provide useable scien- tific information. For instance, Buckland and Anganuzzi (1988) described how data collected on commercial tuna fishing vessels can be used to estimate dolphin abun- dance when survey data are not sufficient. The data collection sites were not randomly selected. Instead, the sampling sites were directly related to dolphin sightings, because dolphins and tuna schools are often closely associated. As a result, areas of high dolphin density corresponded with areas of high fishing effort. Poststratification was used to decrease the bias result- ing from nonrandom distribution of both search effort and dolphin schools. A second example is a retrospec- tive study that combined survey and commercial fishing data. In this study (Halliday8), 1958-60 poststratified survey data were used to develop a relationship between the survey abundance of the 1954-1959 year classes and their abundance estimates from commercial fishery data. This relationship was then used, along with 1969 survey data, to predict the size of the 1966-68 year classes. The same process was used to predict the size of later year classes with later years of survey data. Poststratification also facilitates the use of a single data set for multiple objectives. Collecting data is costly and many data sets are collected and analyzed for a single objective and then not used again. Although it is preferable to use data for multiple objectives, it can be difficult to meet statistical assumptions when the data are re-used for a different purpose. For example, a multispecies survey may be stratified according to the distribution of one or more of the most commercially valuable species collected. An example is the stratifica- tion of Pacific west coast bottom trawl surveys in 1980, 1983, and 1986, which were focused to improve the precision of canary and yellowtail rockfish abundance estimates (Weinberg et al.2). If the stratification used was not effective for decreasing the variance of abun- dance estimates for other species, treating the data as if they were haphazardly collected, recognizing that the estimator may be biased, and poststratifying the data by habitat variables that are closely related to the 6 Halliday, R. G. 1970. 4T-V-W haddock: recruitment and stock abundance in 1970-72. ICNAF Res. Doc 70/75, 12 p. Approved for citation by Tissa Amaratunga, Deputy Executive Secretary, Northwest Atlantic Fisher- ies Organization. [Available from the Secretariat Library, Northwest Atlantic Fisheries Organization, 2 Morris Drive, Burnside Industrial Park, Dartmouth, Nova Scotia, Canada, B3B 1K8.] Dressel and Norcross: Using poststrafication to improve abundance estimates from multispecies surveys 487 distribution of the other species may be a beneficial way to make multiple uses of the data. Although the post- stratified estimator may be biased, poststratification may provide large gains in precision and a decrease in bias in relation to an unstratified estimator. Large increases in precision may be worth the acceptance of some bias. Multispecies surveys are often not optimal for es- timating the abundance of individual species but are often necessary because of limited time and financial resources. As a result, researchers need to explore al- ternative sampling and analysis designs to increase the precision of individual species abundance estimates (NRC, 2000). Poststratification is a method that can be applied to any number of species by using a wide range of habitat and other variables that can be stratified. Because of the dramatic increase in habitat information that is likely to be collected in response to the expanded emphasis in the Magnuson-Stevens Act (NRC, 2000) and because of the adaptability of poststratification for handling a multitude of types of data sets, the method of poststratification may provide increased usefulness for scientific researchers. Acknowledgments We thank Eric Munk and National Marine Fisheries Service Kodiak Laboratory for the vessel and field assis- tance from 1993 to 1996 and Bruce Short for field assis- tance in 1991 and 1992. Additionally, we thank Brenda Holladay, Franz Mueter, Brad Allen, Ed Roberts, and Cindy VanDamm, who helped with the fieldwork for this project, and Franz Mueter, Michael Simpkins, Robert Foy, and Amy Blanchard for constructive advice. For critical review of this article, we thank Milo Adkison, Alison Banks, Allison Barns, Cathy Coon, Judy Ham- ilton, Sue Hills, Heather Patterson, Andy Seitz, Dana Thomas, Albert Tyler, and other anonymous review- ers. This project was funded by Saltonstall-Kennedy NOAA (contracts number NA16FD021601, NA26FD0156, NA47FD0351), Minerals Management Service through the University of Alaska Coastal Marine Institute (task order numbers 11983, 12041, 18445), and the Rasmuson Fisheries Research Council. Literature cited Anganuzzi, A. A. and S. T. Buckland. 1989. Reducing bias in trends in dolphin abundance, estimated from tuna vessel data. Rep. Int. Whal. Comm. 39:323-334. Azarovitz, T. R. 1981. A brief historical review of the Woods Hole Labora- tory trawl survey time series. In Bottom trawl surveys (W. G. Doubleday and D. Rivard, eds.), p. 62-67. Can. Spec. Publ. Aquat. Sci. 58. Bernard, D. R., A. E. Bingham and M. Alexandersdottir. 1998. Robust harvest estimates from on-site roving-access creel surveys. Trans. Am. Fish. Soc. 127:481-495. Buckland, S. T. and A. A. Anganuzzi. 1988. Estimated trends in abundance of dolphins asso- ciated with tuna in the eastern tropical Pacific. Rep. Int. Whal. Comm. 38:411-437. Burke, J. S., J. M. Miller and D. E. Hoss. 1991. Immigration and settlement pattern of Paralich- thys dentatus and P. lethostigma in an estuarine nurs- ery ground, North Carolina, U.S.A. Neth. J. Sea Res. 27(3/41:393-405. Cochran, W. G. 1977. Sampling techniques, 3rd ed., 413 p. John Wiley & Sons, Inc., New York, NY. Fiedler, P. C, and S. B. Reilly. 1994. Interannual variability of dolphin habitats in the eastern tropical Pacific. II: Effects on abundances esti- mated from tuna vessel sightings, 1975-1990. Fish. Bull. 92:451-463. Friedland, K. D., J.-D. Dutil, and T. Sadusky. 1999. Growth patterns in postsmolts and the nature of the marine juvenile nursery for Atlantic salmon, Salmo salar. Fish. Bull. 97:472-481. Gadomski, D. M., and S. M. Caddell. 1991. Effects of temperature on early-life-history stages of California Halibut Paraliehthys californicus. Fish. Bull. 89:567-576. Gunderson, D. R. and I. E. Ellis. 1986. Development of a plumb staff beam trawl for sam- pling demersal fauna. Fish. Res. 4:35-41. Hall, M. A. and S. D. Boyer. 1988. Incidental mortality of dolphins in the eastern tropical Pacific tuna fishery in 1986. Rep. Int. Whal. Commn. 38:439-441. Halliday, R. G., and P. A. Koeller. 1981. A history of Canadian groundfish trawling surveys and data usage in ICNAF Divisions 4TVWX. In Bottom trawl surveys (W.G. Doubleday and D. Rivard, eds.), p. 27-41. Can. Spec. Publ. Fish. Aquat. Sci. 58. Keselman, H. J. 1976. A power investigation of the Tukey multiple com- parison statistic. Educ. Psychol. Meas. 35:97-104. Koeller, P. A. 1981. Distribution and sampling variability in the south- ern Gulf of St. Lawrence groundfish surveys. In Bottom trawl surveys (W.G. Doubleday and D. Rivard, eds.), p. 194-217. Can. Spec. Publ. Fish. Aquat. Sci. 58. Krebs, C. J. 1989. Ecological methodology, 654 p. HarperCollins Publishers, New York, NY. Mueter, F. J., and B. L. Norcross. 1999. Linking community structure of small demersal fishes around Kodiak Island, Alaska, to environmental variables. Mar. Ecol. Prog. Ser. 190:37-51. Murawski, S. A., and J. T. Finn. 1988. Biological bases for mixed-species fisheries: species co-distribution in relation to environmental and biotic variables. Can. J. Fish. Aquat. Sci. 45:1720-1735. NRC (National Research Council). 2000. Improving the collection, management, and use of marine fisheries data, 222 p. Ocean Studies Board, Com- mission on Geosciences, Environment, and Resources. National Academy Press, Washington, D.C., MD. Norcross, B. L., A. Blanchard, and B. A. Holladay. 1999. Comparison of models for defining nearshore flat- fish nursery areas in Alaskan waters. Fish. Oceanogr. 8(11:50-67. 488 Fishery Bulletin 103(3) Norcross, B. L., B. A. Holladay, and F. J. Muter. 1995. Nursery area characteristics of pleuronectids in coastal Alaska, USA. Neth. J. Sea Res. 34(1-3):161- 175. Norcross, B. L., F. J. Muter, and B. A. Holladay. 1997. Habitat models for juvenile pleuronectids around Kodiak Island, Alaska, USA. Fish. Bull. 95:504- 520. Orr, J. W., and A. C. Matarese. 2000. Revision of the genus Lepidopsetta Gill, 1862 (Teleostei: Pleuronectidae) based on larval and adult morphology, with a description of a new species from the North Pacific Ocean and Bering Sea. Fish. Bull. 98:539-582. Pearcy, W. G. 1978. Distribution and abundance of small flatfishes and other demersal fishes in a region of diverse sediments and bathymetry off Oregon. Fish. Bull. 76:629-640. Perry, R. I., and S. J. Smith. 1994. Identifying habitat associations of marine fishes using survey data: an application to the Northwest Atlantic. Can. J. Fish. Aquat. Sci. 51:589-602. Pitt, T. K„ R. Wells, and W. D. McKone. 1981. A critique of research vessel otter trawl surveys by the St. John's Research and Resource Services. In Bottom trawl surveys (W. G. Doubleday and D. Rivard, eds.), p. 42-61. Can. Spec. Publ. Fish. Aquat. Sci. 58. Pollock, K. H., C. M. Jones, and T. L. Brown. 1994. Angler survey methods and their applications in fisheries management, 371 p. Am. Fish. Soc. Spec. Publ. 25, Bethesda, MD. Powell, L. A., W. R. Clark, and E. E. Klaas. 1995. Using post-release stratification to detect het- erogeneity in mallard survival. J. Wildl. Manage. 59(41:683-690. Quinn, T. J., II. 1985. Sampling considerations. /;; Fisheries dynamics: harvest management and sampling (P. R. Mundy, T. J. Quinn II, and R. B. Deriso, eds.), 60 p. Washington Sea Grant Tech. Rep. 85-01, Seattle, WA. Reichert, M. J. M., and H. W. van der Veer. 1991. Settlement, abundance, growth and mortality of juvenile flatfish in a subtropical tidal estuary (Georgia, U.S.A.). Neth. J. Sea Res. 27(3/4):375-391. Reilly. S. B., and P. C. Fiedler. 1994. Interannual variability of dolphin habitats in the eastern tropical Pacific. I: Research vessel surveys, 1986-1990. Fish. Bull. 92:434-450. Rogers, S. I. 1992. Environmental factors affecting the distribution of sole (Solea solea (L.)) within a nursery area. Neth. J. Sea Res. 29(1-3):153-161. Scheaffer, R. L., W. Mendenhall III, and L. Ott. 1996. Elementary survey sampling, 5,h ed., 501 p. Dux- bury Press, Belmont, CA. Scott, B. 1995. Oceanographic features that define the habitat of Pacific Ocean perch, Sebastes alutus. Fish. Oceanogr. 4(21:147-157. Surfer. 1995. Surfer user's guide, vers. 6, 530 p. Golden Soft- ware, Inc., Golden, CO. Tanda, M. 1990. Studies on burying ability in sand and selec- tion to the grain size for hatchery-reared marbled sole and Japanese flounder. Nippon Suisan Gakkaishi 56:1543-1548. Taylor, C. C. 1953. Nature of variability in trawl catches. Fish. Bull. 54:145-166. Taylor, L. R., I. P. Woiwod, and J. N. Perry. 1978. The density-dependence of spatial behaviour and the rarity of randomness. J. Anim. Ecol. 47:383-406. Thompson, S. K. 1992. Sampling, 343 p. John Wiley & Sons, Inc., New York, NY. Walsh, S. J. 1992. Factors influencing distribution of juvenile yellowtail flounder (Limanda ferruginea) on the Grand Bank of Newfoundland. Neth. J. Sea Res. 29(1-3):193- 202. Zar, J. H. 1996. Biostatistical analysis, 662 p. Prentice Hall, Upper Saddle River, NJ. 489 Abstract— Reproductive data col- lected from porbeagle, shortfin mako. and blue sharks caught around New Zealand were used to estimate the median length at maturity. Data on clasper development, presence or absence of spermatophores or sper- matozeugmata, uterus width, and pregnancy were collected by observers aboard tuna longline vessels. Direct maturity estimates were made for smaller numbers of sharks sampled at recreational fishing competitions. Some data sets were sparse, par- ticularly over the vital maturation length range, but the availability of multiple indicators of maturity made it possible to develop estimates for both sexes of all three species. Porbeagle shark males matured at 140-150 cm fork length and females at about 170-180 cm. New Zealand por- beagles therefore mature at shorter lengths than they do in the North Atlantic Ocean. Shortfin mako males matured at 180-185 cm and females at 275-285 cm. Blue shark males matured at about 190-195 cm and females at 170-190 cm: however these estimates were hampered by small sample sizes, difficulty obtaining rep- resentative samples from a popula- tion segregated by sex and maturity stage, and maturation that occurred over a wide length range. It is not yet clear whether regional differences in median maturity exist for shortfin mako and blue sharks. Length at maturity in three pelagic sharks (Lamna nasus, Isurus oxyrinchus, and Prionace glauca) from New Zealand Malcolm P. Francis National Institute of Water and Atmospheric Research 301 Evans Bay Parade Greta Point Wellington, New Zealand E-mail address m francisifi'niwa co nz Clinton Duffy Department of Conservation Private Bay 68908 Auckland, New Zealand Manuscript submitted 20 April 2004 to the Scientific Editor's Office. Manuscript approved for publication 30 March 2005 by the Scientific Editor. Fish. Bull. 103:489-500 (2005). The attainment of sexual maturity in sharks is a major developmental mile- stone which has a large impact on their distribution, behavior, and biology. Immature sharks often associate with each other regardless of sex, but after maturity sexual segregation is the norm. Mature males and females may come together only to mate, resulting in movements that may range from small-scale aggregation of dispersed individuals to long-range migra- tions over thousands of kilometers. The process of maturation, and the subsequent need to channel energy into reproduction, affect the growth rate of at least some shark species. Immature male and female porbea- gles grow at the same rate, and the growth rate of both sexes slows at maturity; however females mature at a greater age than males and there- fore their period of fast immature growth lasts longer and they grow larger than males (Natanson et al., 2002). The maximum reproductive lifes- pan of a shark species is the time elapsed between the age at maturity and the maximum age. In conjunction with the duration of the reproductive cycle, the reproductive lifespan deter- mines the maximum number of litters a female shark can produce in her lifetime. Population modeling indi- cates that shark species that mature at a young age have a greater capac- ity to recover from exploitation than sharks that mature later (Smith et al., 1998). Thus age at maturity is a crucial factor influencing the produc- tivity of a species. Age at maturity can be estimated directly from paired age-and-matu- rity estimates taken from the same shark, but often such data are not available, or are too few to provide precise estimates. Consequently it is often necessary to estimate age at maturity indirectly from length at maturity and a growth curve. In the present study we estimate the length at maturity for three spe- cies of large pelagic sharks in New Zealand waters: porbeagle (Lamna nasus (Bonnaterre, 1788)), shortfin mako (Isurus oxyrinchus Rafinesque, 1810), and blue (Prionace glauca (Linnaeus, 1758)) sharks. These spe- cies are commonly caught by tuna longliners fishing around New Zea- land (Francis et al., 2001). Longline fishing effort declined from a high of over 25 million hooks per year in the early 1980s, to a low of 2—4 million hooks in 1995-98, largely because of a reduction in the number of foreign licensed vessels (Francis et al., 2001). Since then, the do- mestic longline fleet has expanded, and fishing effort exceeded 10 mil- lion hooks in 2001-02 (Ayers et al., 2004). Because of concern over the sustainability of the catches of both target and nontarget species in this fishery, the New Zealand Ministry 490 Fishery Bulletin 103(3) of Fisheries introduced individual trans- ferable quotas for a number of pelagic species, including the three sharks, in October 2004. Despite the panglobal distributions of porbeagle, shortfin mako, and blue sharks, and their importance in the catches of pelagic longline fisheries worldwide, com- paratively little effort has been devoted to estimating their length (or age) at matu- rity. In the northwest Atlantic Ocean, the length at maturity of male and female por- beagles has been well determined (Jensen et al., 2002), but preliminary data from the southwest Pacific Ocean indicate that females mature at a much smaller length there (Francis and Stevens, 2000). Mol- let et al. (2000) found significant differ- ences in the length at maturity of female shortfin makos between the Northern and Southern hemispheres; however there is little information on the length at matu- rity of male makos (Stevens, 1983). Blue sharks have been studied in a number of regions worldwide (Pratt, 1979; Ste- vens, 1984; Hazin et al., 1994; Nakano, 1994; Castro and Mejuto, 1995), but size and sex segregation have made it difficult to obtain representative samples of both sexes from which to determine length at maturity. In the southwest Pacific Ocean, esti- mates of length at maturity are lacking or uncertain for at least one sex of all three species. Although all species make long distance movements, and presumably have wide-ranging stocks, the interhemispheric differences in length at maturity reported for female porbeagles and shortfin makos indicate that it is not safe to transfer esti- mates from one region to another. The aim of the present study is to develop region- specific estimates of length at maturity for male and female porbeagle, blue and shortfin mako sharks, and to determine whether this parameter var- ies globally. These results will contribute to efforts to determine the productivity and stock status of pelagic sharks in New Zealand waters. i — i — TTri — i — i — i — i — i — i — i — i — i — i — i — r ; D Norfolk/^ Island M ■3 Figure 1 Start-of-set positions of tuna longline sets during which observers sampled porbeagle [Lamna nasus), shortfin mako (Isurus oxyrin- chus), and blue [Prionace glauca) sharks. Also shown are the North Island ports where sharks landed during fishing competitions were sampled. Fisheries observers aboard commercial tuna longline vessels (Fig. 1). Sharks obtained from fishing competi- tions provided the opportunity to measure a wide range of reproductive parameters on relatively small samples, whereas sharks observed on tuna longline vessels pro- vided large samples but limited reproductive data. Materials and methods Sharks obtained from fishing competitions Data sources Reproductive data were collected from two main sources. The first consisted of sharks sampled by the authors at recreational fishing competitions, or occasionally sharks retained by commercial fisheries or research vessels. The second source consisted of data and occasionally embryos and female reproductive tracts collected by Ministry of Competition sharks consisted mainly of makos and blue sharks sampled at fishing competitions around the North Island (Fig. 1). Most sharks were sampled from the Hawke Bay competition held annually in February or March from the port of Napier. Other significant competitions were sampled at Castlepoint, Raglan, and New Plymouth. All except two of the competition sharks were collected in summer (January-March) and samples Francis and Duffy Length at maturity in three pelagic sharks 491 spanned the period from 1986 to 2004. In the early years, only data on length, sex, weight, and maturity were collected. In later years, detailed reproductive data were also collected. The following length measurements were made as point-to-point straight line distances to the whole centimeter below actual length: Total length (TLnat): tip of snout to a perpendicular dropped from the tip of tail to the midline (with the tail in the natural position); Total length (TLflex): tip of snout to tip of tail (with the tail flexed towards the midline to provide maximum extension); Fork length (FL): tip of snout to fork in the tail; Precaudal length (PCL): tip of snout to the upper pre- caudal pit (mako and por- beagle sharks) or the origin of the upper caudal lobe (blue sharks). Total weight was measured on accurate scales provided at the fishing competitions, on research vessels, or in commercial fish processing sheds. In males, inner clasper length was measured between the anterior margin of the cloaca and the posterior clasp- er tip, and expressed as a percentage of fork length: Clasper length index (CLI) = 100 (clasper length I FL). The degree of clasper calcification and development was determined and included an assessment of whether the terminal cartilages could be splayed open, whether a spur was present and erupted, and whether the en- tire clasper could be rotated. In some males sampled in later years, the degree of development of the testes, epididymis, and ampulla at the posterior end of the epididymis was also recorded, and occasionally testes were weighed and measured (following dissection from the epigonal organ if necessary). The presence or ab- sence of spermatophores or spermatozeugmata in the ampulla epididymis was noted. (Spermatophores are masses of encapsulated sperm, and they are found in porbeagle and mako sharks; spermatozeugmata are unencapsulated masses of naked sperm that are found in blue sharks [Pratt and Tanaka, 1994]). In females, the reproductive system was examined, and in later years a number of measurements were taken. Uterus width was measured near the middle of the body cavity and expressed as a percentage of fork length: Uterine width index (UWI) = 100 (uterus width/FL). sured after dissection (if necessary) from the epigonal organ. Any contents of the uteri were noted; embryos were measured and sex was determined. The presence or absence of a hymen (cloacal membrane occluding the vaginal opening) was recorded. For both males and females, a direct assessment of maturity (hereafter called direct maturity) was made by using all the available reproductive data. A three- stage classification scheme was used: immature, ma- turing, and mature. Mature sharks were defined as those in which the reproductive system was judged to be fully functional and capable of delivering reproduc- tive products. For analysis purposes, maturing sharks were grouped with immature sharks. Sharks sampled by observers Observers sampled tuna longline sets from around the New Zealand region (Fig. 1). Data from blue and mako sharks were obtained throughout the sampled area, but porbeagles came mainly from the southwestern South Island. Most sharks were sampled in autumn-winter (April-July) over the period 2001-2003. The "standard" length measurement for sharks was FL, but frequently observers also recorded TLnat or PCL. Observers were provided with instructions and pho- tographs indicating the reproductive data they needed to collect, but they were not provided with any practi- cal training. The main data they collected were the following: inner clasper lengths, presence or absence of spermatophores or spermatozeugmata in the ampulla epididymis (for males); uterus width, maximum ovum diameter, and whether the shark was pregnant or not (for females). Examination of observer pregnancy records for blue sharks indicated numerous probable errors: uterus widths from sharks scored as pregnant were frequently less than 18 mm, which seems implausible considering that ova are ovulated at about 18 mm, and pregnant sharks are unlikely to have such small uteri (Pratt, 1979; Natanson1). This problem was apparent for sev- eral observers, some of whom were very experienced (although they had no previous experience examining shark reproductive systems). We suspect that they may have scored some female blue sharks as pregnant if the ovary contained large yolky ova (this problem did not occur for mako and porbeagle sharks, which have much smaller ovarian ova). We therefore used observer blue shark pregnancy records only if they were supported by appropriate comments on the data sheet (e.g., mention of embryos or ovulated eggs in uteri), or if the observers retained embryos or intact uteri for us to examine. Observers did not assess direct maturity; therefore we were unable to derive direct maturity ogives for observer sharks. The maximum diameter of ova, where they were suf- ficiently developed to be visible in the ovary, was re- corded, and the diameter of the oviducal gland was measured. Ovarian dimensions and weight were mea- 1 Natanson, L. 2004. Unpubl. data. National Marine Fish- eries Service, 28 Tarzwell Drive, Narragansett, Rhode Island 02882-1152. 492 Fishery Bulletin 103(3) Table 1 Regression equations used to convert shark length size. Measurement method acronyms are denned in and CTL = curved total length (both measured over s reported in the literature, the "Materials and methods' the curve of the body). r2=the coefficient of determination; « = sample section, except that CFL = curved fork length Species Regression equation r- n Data range (cm) Source Porbeagle FL = -6.943 + 0.893 TLna, 0.997 103 61-181 FL This study FL = 0.90 + 0.95 CFL 0.997 172 83-253 FL S. Campana' Mako CFL = -1.7101 + 0.9286 CTL 0.997 199 65-338 CFL Kohler et al., 1995 FL = 0.973 + 0.968 CFL 0.999 30 113-287 FL This study FL = 0.766 + 1.100 PCL 0.997 999 61-346 FL This study FL = 0.821 + 0.911 TLnat 0.993 399 70-346 FL This study Blue FL = -0.90 + 0.98 CFL 0.99 789 123-286 FL S. Campana' FL = -1.615 + 0.838 TLmt 0.990 273 50-270 FL This study FL = 0.745+ 1.092 PCL 0.998 12,657 34-326 FL This study ' Refers to footnote 2 in the general text. Data analysis For each shark species and sex, we were interested in determining the length at which 50% of the individuals in a population reached full sexual maturity. That length is the median length at maturity, hereafter referred to as "median maturity." Many shark species show abrupt transitions in the sizes of reproductive organs near length at maturity. To locate such transitions in clasper length, we fitted "split" linear regressions to CLI data plotted against FL. Split regressions consist of two simple linear re- gressions fitted to different nonoverlapping data ranges that meet at a point called the breakpoint (Kovac et al., 1999). A split regression has the form CLI = f(FL CLI = g(FL -p) + h for FL < p p) + h for FL a p, where f and g are slope parameters for the two limbs of the regression, and h andp are they-axis and .r-axis coordinates of the breakpoint, respectively. The param- eters f, g, h, and p were estimated by least squares by using the curve fitting routine in the Sigmaplot sta- tistical and graphing package (Sigmaplot, vers. 9.01, Systat Software Inc., Richmond, CA). The length at the breakpoint was corrected for downward rounding of FL by adding 0.5 cm. Maturity ogives were fitted to the direct maturity data separately by sex by using probit analysis (Pear- son and Hartley, 1962). The analyses were performed on individual FL measurements, but we also calcu- lated the proportions of mature individuals in 10-cm length classes to illustrate the trends. Probit analysis assumes that the length at which a randomly selected fish reaches maturity is normally distributed. Two pa- rameters, the mean and standard deviation of the nor- mal distribution, were fitted. Each maturity ogive is the cumulative distribution function for the associated normal distribution. The probit function was fitted by maximum likelihood, and 95% confidence limits were estimated by the bootstrap method. The mean of the normal distribution is an estimate of the median ma- turity, and it was corrected for downward rounding of FL by adding 0.5 cm. All shark length measurements provided in the pres- ent study are FL, unless otherwise stated. For com- parison with our results, we converted measurements from the literature to FL where necessary using the regression equations in Table 1. Literature reports of total length were assumed to be TLnat unless otherwise stated. Scientists working on sharks in the northeast- ern United States, and eastern Canada have typically measured lengths over the curve of the body rather than as straight line distances (Natanson1; Campana2; Pratt3), notwithstanding some published statements to the contrary (Pratt, 1979; Kohler et al., 1995). Results Porbeagle shark In male porbeagles, CLI showed two strong inflection points: the first at about 110 cm, and the second, esti- 2 Campana, S. E. 2004. Personal commun. Bedford Insti- tute of Oceanography, P.O. Box 1006, Dartmouth, Nova Scotia, Canada B2Y 4A2. 3 Pratt, H. L. 2004. Personal commun. Mote Marine Laboratory, 24244 Overseas Highway, Summerland Key, FL 33042. Francis and Duffy Length at maturity in three pelagic sharks 493 mated by split linear regression fitted to sharks longer than 110 cm, at 142.7 cm (95% confidence interval (CI) 140.7-144.7 cm) (Fig. 2). Thus rapid elongation of the claspers began at about 110 cm and was completed by 143 cm. Spermatophores first appeared in the posterior reproduc- tive tract at 135 cm and by about 152 cm 50% of males contained spermatophores. The percentage of sharks with spermato- phores peaked at 165 cm (82% of males) and then declined to about 50% , although sample sizes were small in the longer length groups (Table 2). In females, UWI began increasing at a length of about 145 cm, but many larger, nonpregnant sharks showed no expansion of the uteri (Fig. 3). Three females with UWI of about 4-5% were postpartum, and two with UWI about 11% and one with UWI of about 4% were pregnant. Pregnant females measured 167-202 cm (mean 184 cm, n = 55). Of 19 females longer than 175 cm that were scored by observers for pregnancy, 10 (53%) were pregnant, two (11%) were postpartum, and seven (37%) were rest- ing (or possibly immature). Apart from a 185-cm pregnant fe- male, all whole porbeagles examined by us were immature; therefore no at- tempt was made to estimate maturity directly. Shortfin mako shark 20 -i 15- 10 ♦ Claspers • embryos (n=6) ° Claspers - free living (n=322) Claspers (split regression) Spermatophores present %, 100 80 t/> 60 40 ^ 80 100 120 Fork length (cm) 140 160 180 200 Figure 2 Maturation of male porbeagle sharks iLainna nasus): variation in clasper development and presence of spermatophores in the reproduc- tive tract. 12- 10- 6- I I Pregnant females (n=55) o Uterus width index (n=63) CLI showed two strong inflection points in male makos; the first at about 140 cm and the second (estimated by split linear regression) at 185.1 cm (CI 182.5-187.7 cm) (Fig. 4). The smallest male with spermatophores was 136 cm, but this measurement was an outlier and may have been an error; the next smallest was 156 cm. Fifty percent of males contained spermatophores by 178 cm, and 100% by about 235 cm. Sample sizes were reason- able over the transition range but small above 230 cm (Table 2). Male makos examined by us showed little overlap in length between immature and ma- ture sharks (Fig. 4), but sample sizes were small in all length classes (Table 2). The smallest mature male was 182 cm and the largest immature male was 183 cm long. The median maturity estimated by probit analysis was 182.9 cm (CI 180.7-185.1 cm) (Fig. 4). In females, UWI began increasing at a length of about 275 cm, and all larger sharks had expanded uteri (Fig. 5). Only one pregnant female mako has been recorded from New Zealand waters, and it was 290 cm FL (Duffy and <& Qtfffooo o°q£8 ^"l&pfT 6C - 12 10 - 6 -2 0 25 50 75 100 125 150 175 200 225 Fork length (cm) Figure 3 Maturation of female porbeagle sharks lLamna nasus): variation in uterus width index, and length-frequency distribution of pregnant females. Francis, 2001); no uterus width measurement was avail- able for that shark. The remaining makos over 275 cm were either postpartum or resting. The maximum ovum diameter began increasing in sharks longer than 250 cm (in shorter sharks, ova were barely visible or were invisible) (Fig. 6). The diameter of the oviducal gland increased abruptly between 250 and 270 cm, but ovary dimensions showed no abrupt change in size (Fig. 6). Median maturity was estimated directly from a sam- ple of 88 females (Table 3). The smallest mature female 494 Fishery Bulletin 103(3) Table 2 Sample sizes by 10-cm length class for the assessment of maturity in male porbeagle, mako, and blue sharks. Porbeagle shark Shortfin mako shark Blue shark Direct Direct Length class midpoint (cm) Spermatophores Spermatophores maturity Spermatozeugmata maturity 45 0 0 0 0 1 55 0 0 0 0 3 65 0 0 0 0 2 75 2 0 0 0 0 85 15 0 1 1 0 95 4 0 0 2 1 105 0 3 1 2 0 115 2 3 3 0 0 125 8 3 0 1 0 135 23 4 8 1 0 145 28 4 11 3 1 155 30 9 4 6 5 165 17 10 6 13 3 175 18 19 3 4 6 185 6 16 7 20 4 195 4 15 1 21 6 205 1 27 1 18 6 215 0 19 4 15 4 225 0 14 1 12 1 235 0 8 0 20 8 245 0 5 1 26 2 255 0 3 0 19 1 265 0 0 0 11 1 275 0 1 0 6 2 285 0 0 0 2 0 295 0 0 0 1 1 Total 158 163 52 204 58 Table 3 Sample sizes by 10-cm length class for the assessment of maturity in female mako and blue sharks. Shortfin mako shark Blue shark Length class Shortfin mako shark Blue shark Length class Direct Direct Direct Direct midpoint (cm) maturity 0 maturity midpoint (cm) maturity maturity 55 6 215 10 2 65 0 3 225 10 0 75 0 1 235 6 0 85 0 2 245 9 0 95 0 0 255 4 0 105 0 0 265 3 0 115 2 0 275 1 0 125 2 0 285 2 0 135 2 0 295 3 0 145 10 1 305 1 0 155 6 1 315 0 0 165 3 0 325 3 0 175 1 3 335 2 0 185 3 5 345 1 0 195 0 0 Total 88 26 205 4 2 Francis and Duffy Length at maturity in three pelagic sharks 495 was 274 cm and the longest immature female was 300 cm. Median maturity was estimated by probit analysis to be 280.1 cm (CI 267.5-292.9 cm), but sam- ple sizes were very small over the tran- sitional range (Fig 5). The nonoverlap of the CIs between males and females showed that median maturity differs significantly between the sexes. Blue shark The relationship between CLI and FL was essentially linear in blue sharks, and no apparent inflections were evident (Fig. 7). The smallest male with sperma- tozeugmata was 164 cm; 50% of males contained spermatozeugmata by 194 cm, and 100% by about 260 cm. Samples of male blue sharks examined by us were small (Table 2). Maturation occurred over a wide length range: the smallest mature male was 157 cm and the largest immature male was 237 cm long. The direct estimate of median maturity was correspondingly variable (192.1 cm, CI 178.1-206.3 cm) (Fig. 7). The UWI increased abruptly above about 170 cm in some sharks, all of which were pregnant (Fig. 8). Other non- pregnant sharks up to about 220 cm FL, which were presumably subadults, had UWIs less than 2%. Pregnant females ranged from 166 to 252 cm (mean 203 cm) (Fig. 8). Only 26 females were available for di- rect maturity estimation (Table 3). The smallest recorded mature female was 142 cm, but this seems exceptional; the next smallest was 172 cm. The longest immature female was 185 cm. The num- ber of sharks in the maturation length range was inadequate for determining median maturity (Table 3), although we have shown the probit analysis ogive in Figure 8. Discussion 20-i - 100 ♦ Claspers - embryos (n=3) o Claspers - tree living (n=236) ° ,._. ^ \ 1 ^ Claspers (split regression) Spermatophores present o M8 o -80 co "O Ti o • Percentage mature (direct) o ^ oH 'to o > 3 3 Clasper length index Percentage mature (titled curve) cP'q ■60 -40 -20 CD tZ- 3 O CO X CD " -j CD zj CO 21 -a $ CD CD CO —. ft> . O 3 o- "-*■ 2 o o o c ♦ 8 A-' * ° °fc \ / o .' O • 1 .i<5 £> / 0 20 40 60 80 100 120 140 160 180 200 220 240 260 280 Fork length (cm) Figure 4 Maturation of male shortfin mako sharks Usurits oxyrinchus): van ation in clasper development, presence of spermatophores in the reproduc- tive tract, and direct maturity estimation determined from a su ite of maturity indicators. 8- • • • - 2 - 100 i i Pregnant females (n=1 ) o Uterus width index (n=79) °l ° 5" 6- C"" X CD "D C £ 4- "D c/) 3 CD 5 2- • Percentage mature (direct) Percentage mature (fitted curve) c o •< o -a -1 S CO D) 3 » CD Percentage mature (%) D O O O / ^ o ( o o ; 8&>0 %aggg5^ ° -0 0 25 50 75 100 125 150 175 200 225 250 275 300 325 350 375 Fork length (cm) Figure 5 Maturation of female shortfin mako sharks (Isurus oxyrinchus): varia- tion in uterus width index, and direct maturity estimation from a suite of maturity indicators. The only pregnant female recorded from New Zealand waters is also indicated. Maturity estimates To be sexually mature, a male shark must be able to produce viable sperm and have the means to deliver them to a female. Similarly, females must be able to produce viable eggs and nourish the developing embryos through to parturition. An assessment of the degree of development of all parts of the reproductive system and the presence or absence of reproductive products is the best way to determine sexual maturity. We used this approach to score the maturity status of individual sharks and thereby derive direct median maturity esti- mates. However, the sample sizes available for this approach were sometimes small, and confidence limits ranged from unrealistically low (because of lack of over- lap of immature and mature sharks) to high; therefore it was not possible to rely entirely on these estimates. We supplemented our direct maturity estimates with measurements or assessments (made by observers) of some key components and products of the reproductive 496 Fishery Bulletin 103(3) system. The presence or absence of spermatophores or spermatozeugmata is a good indicator of the ability of a male to produce viable sperm, but it is not infallible: such structures sometimes lack viable sperm (Pratt and Tanaka, 1994). Furthermore, male reproductive products may not be present year-round: blue sharks appear to have a seasonal cycle of spermatozeugmata production in the western central Atlantic (Hazin et al., 1994), although Pratt (1979) found no evidence of a cycle in the western North Atlantic. Thus the presence of spermatophores and spermatozeugmata does not 90- ■ -10 80 - ■ Ovary thickness (n=47) A Oviducal gland diameter (n=38) 5 | 70- o Maximum ovum diameter (n=50) o -8 0) X £ I 3 E B 60- ■— CO 3 S« 50- A m -6 o c c -a m m 3 £ £ 40- ■ >, O) A O A A L" 3 ra ra 30 - A AO > o ;» O O > 20- O . -r-8 o ° "2 1 10- rf A***-* o o 0 25 50 75 100 125 150 175 200 225 250 275 300 325 350 375 Fork length (cm) Figure 6 Maturation of female shortfin mako sharks (Jsurus oxyrinchus): relation- ship between fork length and ovary thickness, oviducal gland diameter, and maximum ovum diameter. Clasper length (%) en o oi i i 1 • jAf * Spermatozeugmata present (%) Percentage mature (%) o o o o o o ^ CO CD ''t OJ C o Claspers - free-living (n=286) Spermatozeugmata present • Percentage mature (direct) Percentage mature (fitted curve) o ft ^sF®t o? o ° ° c% cSfc ° • u -i ( Matui develc and d ) 20 40 60 80 100 120 140 160 180 200 220 240 260 280 300 3J Fork length (cm) Figure 7 ation of male blue sharks (Prionace glauca): variation i pment, presence of spermatozeugmata in the reproduct trect maturity estimation from a suite of maturity indii 0 n clasper ive tract, ;ators. necessarily confirm reproductive viability, and their absence does not confirm immaturity. Similarly, fully calcified claspers that can be rotated, splayed open, and possess an anchoring mechanism may confer an ability to mate, but they do not necessarily confirm an ability to deliver viable products; however the lack of fully developed claspers presumably does prevent suc- cessful copulation. In the present study, either the length at which clasper development was completed in half the male population, or the length at which 50% of males pos- sessed spermatophores or spermatozeug- mata, whichever was higher, provided an estimate of the lower bound of the median maturity. The actual median maturity may be higher than this es- timate if some males had reproductive products that lacked viable sperm, or if some other feature of the reproductive system (e.g., the siphon system) was in- sufficiently developed to enable delivery of sperm to the female. An analogous argument applies to fe- male sharks. Full development of the uterus and oviducal gland, and produc- tion of vitellogenic ovarian ova, are all required for successful reproduction. Expansion of the uterus, as measured here by UWI, may not be a sufficient condition by itself. Thus the length at which half the female population had expanded uteri places a lower bound on the median maturity. The smallest length at which females were pregnant, and the length-frequency distributions of pregnant females, are not by themselves good indicators of me- dian maturity. A better indicator would be the length at which half the females in a population first become pregnant, but this is impossible to determine. Fur- thermore, pregnancy estimates could be confounded by unrepresentative sam- pling of a population that may be seg- regated by reproductive status and by nonparticipation of some females during breeding because they are "resting" be- tween pregnancies. Nevertheless, preg- nancy absolutely confirms maturity; therefore it is a useful adjunct to other measures of maturity. The presence or absence of a hymen has been used in some studies to indi- cate maturity. However it should not be used for that purpose because adolescent (premature) mating occurs in at least some species of sharks, including blue sharks (Pratt, 1979). Furthermore, the absence of a hymen may not even be a good indicator of mating: we observed Francis and Duffy Length at maturity in three pelagic sharks 497 Pregnant females (n=40) Uterus width index (n=650) Percentage mature (direct) Percentage mature (fitted curve) ■8 _Q -08 125 150 175 200 Fork length (cm) Figure 8 Maturation of female blue sharks iPrionace glauca): variation in uterus width index, direct maturity estimation from a suite of maturity indicators, and length-frequency distribution of pregnant females are shown. Table 4 Summary of length at maturity indicators in porbeagle, shortfin mako, and blue sharks, and estimates of median length at maturity. Table entries are fork lengths in centimeters. Direct maturity estimates were derived by examination of a suite of maturity indicators. Italics indicate estimates based on small sample sizes over the maturation length range. " — " indicates that an estimate was not possible. Sex Maturity indicator Porbeagle shark Shortfin mako shark Blue shark Males Females 50% with spermatophores Rapid clasper elongation complete Direct maturity estimate Median length at maturity Rapid expansion of uterus begins First females pregnant Direct maturity estimate Median length at maturity 152 143 140-150 145 167 170-180 178 194 185 — 183 192 180-185 190-195 275 170 i 166 280 — 275-285 170-190 Only one pregnant female (290 cm FL) has been recorded from New Zealand. some female shortfin makos in which the membrane was very thin and partially perforated, but had clearly not been damaged by copulation. We believe that the hymen disintegrates naturally with growth in makos; the same possibility was proposed for blue sharks by Pratt (1979). Using a combination of our direct maturity estimates, and other indicators of maturity based on larger sam- ples of sharks, we generated overall estimates of me- dian maturity for both sexes of the three pelagic sharks (Table 4). Porbeagle shark In male porbeagles, the length at which 50% of sharks had spermatophores (152 cm) was longer than the length at which clasper elongation was complete (143 cm) (Table 4). However the percentage of males having sper- matophores did not reach 100% in the longer length groups (Fig. 2), indicating that some mature males were reproductively inactive. This finding is consistent with reports from the western North Atlantic that male por- beagles have a seasonal cycle of spermatophore produc- tion, with a minimum in winter-spring (Jensen et al., 2002). If some mature males lacked spermatophores, the length at which 50% of males had spermatophores in our study was probably greater than the median maturity. The lack of a direct maturity estimate limits our ability to estimate the median maturity, but it is likely in the range 140-150 cm. Similarly, we have no direct estimate of female por- beagle maturity. There was a considerable gap between 498 Fishery Bulletin 103(3) the length at which rapid expansion of the uterus be- gan (145 cm) and the length of the smallest pregnant female (167 cm). UWI values less than 2% occurred for females up to about 185 cm (Fig. 3), but this does not mean that a high proportion of females in this length group had narrow uteri; uterus width measurements were not available for most of our pregnant females and therefore large UWI values are underrepresented in Figure 3. Most pregnant females were 170-200 cm long. We estimate that median maturity in females is about 170-180 cm, but this estimate requires confirmation. It is essentially the same as that provided by Francis and Stevens (2000) for New Zealand and Australian porbeagles (their New Zealand data were a smaller subset of the data used in the present study). Although our estimates of median maturity for both males and females are uncertain, it is clear that por- beagles from New Zealand mature at considerably smaller lengths than they do in the North Atlantic. In the western North Atlantic, males mature at about 166 cm and females at 208 cm (Jensen et al., 2002). Data from the eastern North Atlantic (Gauld, 1989; Ellis and Shackley, 1995) are insufficient to estimate length at maturity, but the pregnant females reported by Gauld (1989) were considerably longer (199-248 cm) than those from New Zealand. Porbeagles from the North Atlantic also grow larger than those from New Zealand: in the North Atlantic, sharks longer than 200 cm are common (Gauld, 1989; Campana et al., 2001), whereas around New Zealand and Australia they are very rare (Francis et al., 2001; Stevens and Wayte4). Differences in length at maturity between the North Atlantic and New Zealand and the proportion of sharks in the longer length classes indi- cate the existence of separate populations in the two regions — a conclusion that is supported by the disjunct distribution of porbeagles in the Northern and Southern Hemispheres (Compagno, 2001). Shortfin mako shark Our direct maturity estimate for male makos (183 cm) was based on a small sample size, and the small overlap between the lengths of immature and mature sharks is implausible. However, the lengths at which clasper development was completed, and at which 50% of males had spermatophores, were similar to the direct estimate (Table 4). Median maturity for males is therefore about 180-185 cm. Our direct maturity estimate for female makos (280 cm) was based on few sharks over the matura- tion length range but was consistent with the length at which rapid uterus expansion began (275 cm). Our best estimate of median maturity in females is 275- 285 cm. Stevens (1983) used the degree of clasper calcifica- tion and an inflection in clasper length to estimate the length at maturity of males from New South Wales as 176 cm. In South Africa, males were estimated to mature at 177-188 cm (Cliff et al., 1990), but very few immature sharks were available. Our estimate of me- dian maturity in New Zealand males (180-185 cm) is therefore similar to those from elsewhere. Mollet et al. (2000) reported lengths at maturity for female makos of 298 cm total length in the Northern Hemisphere and 273 cm total length in the Southern Hemisphere. However, some of the 25 cm difference was due to Northern Hemisphere measurements having been taken over the curve of the body and Southern Hemi- sphere measurements having been taken in a straight line. Using appropriate conversion regressions, their Northern Hemisphere median maturity is equivalent to 267 cm FL, and their Southern Hemisphere median maturity is equivalent to 248 cm FL. When Mollet et al.'s Southern Hemisphere data are analysed separately for two subregions, South Africa and Australia, the estimated lengths at maturity are 244 cm (n = 50) and 254 cm (n = 32) respectively (Mollet5). The former is con- sistent with Cliff et al.'s (1990) estimate of 243 cm for South Africa, and the latter is consistent with Stevens's (1983) estimate of 255 cm for eastern Australia (both those estimates were made from subsets of the data used by Mollet et al. [2000]). Our estimate of median maturity in New Zealand females (275-285 cm) is substantially higher than Mol- let's5 estimate for Australia (254 cm). Because tagged makos have moved between New Zealand and eastern Australia in both directions (Chan, 2001; Hartill and Davies, 2001; Holdsworth and Saul, 2003), we think it is unlikely that the difference is due to the presence of distinct populations in the two regions. We suspect that the difference is a result of possible length estima- tion errors (some of the Australian shark lengths were calculated from recorded weights, with a length-weight regression [Stevens, 1983; Mollet et al., 2000]), and the result of small sample sizes over the length range at maturation. For our direct maturity estimate, we had only 19 New Zealand sharks over the length range 240-290 cm, and Mollet5 had 15 sharks. Interestingly, our estimate of median maturity in New Zealand females is also greater than Mollet et al.'s (2000) estimate for the western North Atlantic, thus removing the reported between-hemisphere difference. We believe that larger, accurately measured samples of female ma- kos are required before definitive statements can be made about length at maturity in the various regions. Blue shark In male blue sharks from New Zealand, CLI lacked an inflection near the length of maturity — a feature that 4 Stevens, J. D., and S. E. Wayte. 1999. A review of Aus- tralia's pelagic shark resources. FRDC Proj. Rep. 98/107, 64 p. [Available from CSIRO Marine Research, PO Box 1538, Hobart, Tasmania 7001, Australia.] 5 Mollet, H. 2004. Personal commun. Moss Landing Marine Laboratories, 8272 Moss Landing Road, Moss Landing, CA 95039. Francis and Duffy: Length at maturity in three pelagic sharks 499 has also been reported elsewhere (Pratt, 1979; Hazin et al., 1994). Thus clasper length was not useful in estimat- ing length at maturity. Our direct maturity estimate was similar to the length at which 50% of sharks had spermatozeugmata and indicated that median maturity occurs at about 190-195 cm (Table 4). In females, maturation occurred over a wide length range, as reported elsewhere (Hazin et al., 1994). Taking into account the length distributions of pregnant females and females with low UWI values (Fig. 8), we believe median maturity is likely in the range 170-190 cm. In other blue shark studies, estimation of the length at maturity has also been hindered by small sample sizes, or even a complete absence of immature or mature sharks. In the western North Atlantic, males mature at about 178 cm, and females at around the same length, although few mature females have been available (Pratt, 1979). In the Gulf of Guinea, Atlantic Ocean, 50% of females were pregnant at 180 cm (Castro and Mejuto, 1995). In Australian studies, a lack of immature sharks made it impossible to estimate maturity adequately (Ste- vens, 1984; Stevens and McLoughlin. 1991). In the North Pacific Ocean, 50% of males had spermatozeugmata at 166 cm and 50% of females were pregnant at 174 cm (Nakano, 1994). Thus worldwide estimates of maturity in blue sharks are similar to ours from New Zealand, except perhaps for a smaller length at maturity of males in the North Pacific. Unlike females in most species of sharks, female blue sharks do not appear to mature at a length greater than that for mature males. Acknowledgments We thank the Ministry of Fisheries for funding this study under research project TUN2002/01, and provid- ing access to data collected by observers. Lynda Griggs (NIWA) assisted with data extracts and interpretation, and Chris Francis (NIWA) carried out the probit analy- ses. Lisa Natanson, Wes Pratt, Steve Campana, and Henry Mollet kindly provided unpublished data and advice on their interpretation. Literature cited Ayers, D., M. P. Francis, L. H. Griggs, and S. J. Baird. 2004. Fish bycatch in New Zealand tuna longline fish- eries, 2000-01 and 2001-02. N.Z. Fish. Assess. Rep. Rep. 2004/46, 47 p. Campana, S. E., L. Marks, W. Joyce, and S. Harley. 2001. Analytical assessment of the porbeagle shark [Lamna nanus) population in the Northwest Atlantic, with estimates of long-term sustainable yield. Can. Sci. Advisory Secretariat Res. Doc. 2001/067, 59 p. Castro, J. A., and J. Mejuto. 1995. Reproductive parameters of blue shark, Prionace glauca, and other sharks in the Gulf of Guinea. Mar. Freshw. Res. 46:967-973. Chan, R. W. K. 2001. Biological studies on sharks caught off the coast of New South Wales. Ph.D. diss., 323 p. Univ. New South Wales, Sydney, New South Wales, Australia. Cliff, G., S. F. J. Dudley, and B. Davis. 1990. Sharks caught in the protective gill nets off Natal, South Africa. 3. The shortfin mako shark Isurus oxyrin- chus (Rafinesque). Sth Afr. J. Mar. Sci. 9:115-126. Compagno, L. J. V. 2001. Sharks of the world. An annotated and illustrated catalogue of shark species known to date. FAO Species Cat. Fishery Purposes 1. vol. 2. 269 p. FAO, Rome. Duffy, C, and M. P. Francis. 2001. Evidence of summer parturition in shortfin mako {Isurus oxyrinchus) sharks from New Zealand waters. N. Z. J. Mar. Freshw. Res. 35:319-324. Ellis, J. R., and S. E. Shackley. 1995. Notes on porbeagle sharks, Lamna nasus, from the Bristol Channel. J. Fish Biol. 46:368-370. Francis, M. P., L. H. Griggs, and S. J. Baird. 2001. Pelagic shark bycatch in the New Zealand tuna longline fishery. Mar. Freshw. Res. 52:165-178. Francis, M. P., and J. D. Stevens. 2000. Reproduction, embryonic development and growth of the porbeagle shark, Lamna nasus, in the south-west Pacific Ocean. Fish. Bull. 98:41-63. Gauld. J. A. 1989. Records of porbeagles landed in Scotland, with observations on the biology, distribution and exploitation of the species. Scot. Fish. Res. Rep. 45, 16 p. Hartill, B., and N. M. Davies. 2001. New Zealand billfish and gamefish tagging, 1999-2000. NIWA Tech. Rep. 106, 29 p. Hazin, F. H. V., K. Kihara, K. Otsuka, C. E. Boeckman, and E. C. Leal. 1994. Reproduction of the blue shark, Prionace glauca, in the southwestern equatorial Atlantic Ocean. Fish. Sci. 60:487-491. Holdsworth, J., and P. Saul 2003. New Zealand billfish and gamefish tagging 2001-02. N.Z. Fish. Assess. Rep. 2003/15, 39 p. Jensen, C. F, L. J. Natanson, H. L. Pratt, N. E. Kohler, and S. E. Campana. 2002. The reproductive biology of the porbeagle shark, Lamna nasus, in the western North Atlantic Ocean. Fish. Bull. 100:727-738. Kohler, N. E., J. G. Casey, and P. A. Turner. 1995. Length-weight relationships for 13 species of sharks from the western North Atlantic. Fish. Bull. 93:412-418. Kovac, V., G. H. Copp, and M. P. Francis. 1999. Morphometry of the stone loach, Barbatula bar- bat ula: do mensural characters reflect the species' life history thresholds? Environ. Biol. Fish. 56:105-115. Mollet, H. F, G. Cliff, H. L. Pratt, and J. D. Stevens. 2000. Reproductive biology of the female shortfin mako, Isurus oxyrinchus Rafinesque, 1810, with comments on the embryonic development of lamnoids. Fish. Bull. 98:299-318. Nakano, H. 1994. Age, reproduction and migration of blue shark in the North Pacific Ocean. Bull. Nat. Res. Inst. Far Seas Fish. 31:141-256. Natanson, L. J., J. J. Mello, and S. E. Campana. 2002. Validated age and growth of the porbeagle shark {Lamna nasus) in the western North Atlantic Ocean. Fish. Bull. 100:266-278. 500 Fishery Bulletin 103(3) Pearson, E. S., and H. O. Hartley. 1962. Biometrika tables for statisticians. Vol. 1, 2nd ed., 240 p. Cambridge Univ. Press, Cambridge, UK. Pratt, H. L. 1979. Reproduction in the blue shark, Prionace glauca. Fish. Bull. 77:445-470. Pratt, H. L„ and S. Tanaka. 1994. Sperm storage in male elasmobranchs: a descrip- tion and survey. J. Morph. 219:297-308. Smith, S. E., D. W. Au, and C. Show. 1998. Intrinsic rebound potentials of 26 species of Pacific sharks. Mar. Freshw. Res. 49:663-678. Stevens, J. D. 1983. Observations on reproduction in the shortfin mako Isurus oxyrinehus. Copeia 1983:126-130. 1984. Biological observations on sharks caught by sport fishermen off New South Wales. Aust. J. Mar. Freshw. Res. 35:573-590. Stevens, J. D., and K. J. McLoughlin. 1991. Distribution, size and sex composition, repro- ductive biology and diet of sharks from Northern Australia. Aust. J. Mar. Freshw. Res. 42:151-199. 501 Abstract — Survey- and fishery- derived biomass estimates have indicated that the harvest indices for Pacific cod iGadus macrocepha- lus) within a portion of Steller sea lion (Eumetopias jubatus) critical habitat in February and March 2001 were five to 16 times greater than the annual rate for the entire Bering Sea-Aleutian Islands stock. A bottom trawl survey yielded a cod biomass estimate of 49,032 metric tons (t) for the entire area surveyed, of which less than half (23,329 t) was located within the area used primarily by the commercial fishery, which caught 11,631 t of Pacific cod. Leslie deple- tion analyses of fishery data yielded biomass estimates of approximately 14,500 t (95% confidence intervals of approximately 9,000-25,000 t), which are within the 95f>r confidence inter- val on the fished area survey estimate (12,846-33,812 t). These data indicate that Leslie analyses may be useful in estimating local fish biomass and harvest indices for certain marine fisheries that are well constrained spatially and relatively short in dura- tion (weeks). In addition, fishery effects on prey availability within the time and space scales relevant to foraging sea lions may be much greater than the effects indicated by annual harvest rates estimated from stock assessments averaged across the range of the target species. Survey- and fishery-derived estimates of Pacific cod (Gadus macrocephalus) biomass: implications for strategies to reduce interactions between groundfish fisheries and Steller sea lions (Eumetopias jubatus) Lowell W. Fritz National Marine Mammal Laboratory Alaska Fisheries Science Center National Marine Fisheries Service 7600 Sand Point Way NE Seattle, Washington 98115 E-mail address lowell.fntz@noaa gov Eric S. Brown Resource Assessment and Conservation Engineering Alaska Fisheries Science Center National Marine Fisheries Service 7600 Sand Point Way NE Seattle, Washington 98115 Manuscript submitted 20 May 2004 to the Scientific Editor's Office. Manuscript approved for publication 23 March 2005 by the Scientific Editor. Fish. Bull. 103:501-515 (20051. For the past 30 years, the Steller sea lion (Eumetopias jubatus) popula- tion in western Alaska has declined (Braham et al., 1980; Sease and Gud- mundson1). The species was listed as threatened under the U.S. Endangered Species Act (ESA) in 1990 after evi- dence of a major decline in abundance in the core of its range from the Kenai Peninsula in south-central Alaska to Kiska Island in the western Aleutian Islands (Braham et al., 1980; Merrick et al., 1987). After the decline was first observed in the eastern Aleutian Islands in the early 1970s (Braham et al., 1980), it spread eastward to Prince William Sound and west- ward through Russia during the next decade (Merrick et al., 1987; Loughlin et al., 1992). From the early 1970s to 1990, counts of adult and juvenile Steller sea lions declined by over 70%, but annual rates of decline were most severe between 1985 and 1989 (-15%/ yr; Loughlin et al., 1992). During the 1990s, the decline slowed to approxi- mately -5%/yr and may have tempo- rarily abated in many areas by 2002 (Sease and Gudmundson1). Understanding the causes for the decline and lack of recovery in the Steller sea lion population has large- ly eluded scientists and managers. despite the millions of dollars spent on scientific research (Ferrero and Fritz2) and numerous reviews by aca- demic (Alaska Sea Grant3; DeMaster and Atkinson4; NRC, 1996; 2003) and governmental panels (Kruse et al.5; NMFS6'7'8-9). Although recent reviews 1 Sease, J. L., and C. J. Gudmundson. 2002. Aerial and land-based surveys of Steller sea lions (Eumetopias jubatus) from the western stock in Alaska, June and July 2001 and 2002. NOAA Tech. Memo. NMFS-AFSC-131, 45 p. Alaska Fisheries Science Center, 7600 Sand Point Way NE, Seattle WA 98115. 2 Ferrero, R. C. and L. W. Fritz. 2002. Steller sea lion research coordination: a brief history and summary of recent progress. NOAA Tech. Memo. NMFS- AFSC-129, 34 p. Alaska Fisheries Sci- ence Center, 7600 Sand Point Way NE, Seattle WA 98115. 3 Alaska Sea Grant. 1993. Is it food?: Addressing marine mammal and sea- bird declines. Workshop summary rep. AK-SG-93-01, 59 p. Univ. Alaska Fairbanks, Alaska Sea Grant College Program, Fairbanks AK 99775. 4 DeMaster, D., and S. Atkinson, (eds.l. 2002. Steller sea lion decline: Is it food? II. Workshop summary, rep. AK-SG- 02-02, 80 p. Univ. Alaska Fairbanks, Alaska Sea Grant College Program, Fairbanks AK 99775. 5. 6. 7, 8, 9 gee nexf page. 502 Fishery Bulletin 103(3) (Kruse et al.5; DeMaster and Atkinson4; NRC, 2003) concluded that "top-down" forces, such as predation or illegal shooting, are greater threats to recovery of the Steller sea lion population, they could not eliminate "bottom-up" factors from consideration. NRC (2003) suggested that NMFS conduct an adaptive manage- ment experiment to determine the magnitude of one such "bottom-up" force, nutritional stress resulting from competition with fisheries for prey (NMFS67-89; NRC. 2003). The North Pacific is home to some of the largest fisheries in the world, particularly those for groundfish such as Pacific cod (Gadus macrocephalus) and walleye pollock (Theragra chalcogramma). Steller sea lions eat a wide variety offish and cephalopods, including Pacific cod, walleye pollock, Atka mackerel (Pleurogrammus monopterygius), arrowtooth flounder (Atherestes sto- rnias), salmon (Oncorhynehus spp.), herring (Clupea pal- lasi), capelin (Mallotus villosus), eulachon (Thaleichthys pacificus), sand lance {Ammodytes hexapterus), squid, and octopus (Sinclair and Zeppelin, 2002). A large pro- portion of their diet, however, is composed of semide- mersal or pelagic schooling fish, particularly fish in spawning migrations or aggregations nearshore. These same species are often targeted at the same time and in the same areas by groundfish fisheries, particularly those fisheries that use trawl gear. Concerns about the potential of fisheries to create localized depletions of prey in important sea lion foraging habitats have led to controversial groundfish fishery restrictions throughout most of Alaska (NMFS8-9). 5 Kruse, G. H., M. Crow, E. E. Krygier, D. S. Lloyd, K. W. Pitcher, L. D. Rea, M. Ridgway, R. J. Small, J. Stinson and K.M.Wynne. 2001. A review of proposed fishery manage- ment actions and the decline of Steller sea lions lEumetopias jubatus) in Alaska: a report by the Alaska Steller sea lion restoration team. Regional information report 5J01-04, 106 p. Alaska Dep. Fish and Game, P.O. Box 25526. Juneau AK 99802. H NMFS (National Marine Fisheries Service). 1998. En- dangered Species Act Section 7 Consultation on an Atka mackerel fishery under the BSAI groundfish FMP between 1999 and 2002; authorization of a walleye pollock fishery under the BSAI FMP between 1999 and 2002; and under the GOA FMP between 1999 and 2002, 189 p. NMFS Protected Resources Division, Alaska Region, P.O. Box 21668, Juneau, AK 99802. 7 NMFS. 2000. Endangered Species Act. Section 7: Con- sultation, biological opinion and incidental take statement on the authorization of the Bering Sea-Aleutian Islands and Gulf of Alaska groundfish fisheries based on the Fishery Management Plans, 352 p. NMFS Protected Resources Divi- sion, Alaska Region, P.O. Box 21668, Juneau. AK 99802. 8 NMFS. 2001. Endangered Species Act. Section 7: Con- sultation, biological opinion and incidental take statement on the authorization of the Bering Sea-Aleutian Islands and Gulf of Alaska groundfish fisheries based on the Fishery Management Plans as modified by Amendments 61 and 70, 206 p. NMFS Protected Resources Division, Alaska Region, P.O. Box 21668, Juneau, AK 99802. 9 NMFS. 2003. Supplement to the Endangered Species Act. Section 7: Consultation, biological opinion and incidental take statement of October 2001, 179 p. NMFS Protected Resources Division, Alaska Region, P.O. Box 21668, Juneau, AK 99802. Assessment models and fisheries harvest strategies have determined the overall fishing mortality rate that can be allowed for the stock and the amount of biomass that can be removed. In practice, however, catches are not uniformly distributed across the range of the as- sessed stock nor are they distributed equally through- out the year. Although there is evidence that the Atka mackerel trawl fishery has created localized depletions of its target species (NMFS6 Lowe and Fritz, 1997; NRC, 2003), this finding has not been generally applied to fisheries for other sea lion prey. Trawl fisheries in the Aleutian Islands may have, in certain instances, reduced local abundances of Atka mackerel by as much as 90% (Lowe and Fritz, 1997). Atka mackerel and its fishery have characteristics that permitted analysis of fishery data in this way. The species does not possess a swim bladder and thus makes a poor acoustic target. As a consequence, the Atka mackerel fishery does not target on an acoustic signal, but instead trawls in ar- eas where the species is known to congregate. Through the analysis of time series of catch and effort statis- tics from local fisheries with Leslie's equation (Ricker, 1975; Hilborn and Walters, 1992; Gunderson, 1993), estimates of the initial abundance of Atka mackerel (prefishery) and its catchability (proportion of the stock caught with one unit of effort) were made within the context of certain assumptions, which included the following: 1) the population being fished is closed, or alternatively that immigration and growth are equal to emigration plus natural mortality, 2) catchability over the course of the fishery remains constant, and 3) changes in catch per unit of effort (CPUE) are di- rectly related to changes in fish density. These assump- tions may be met for marine species if the area fished is well defined (e.g., is surrounded by habitat that is unsuitable for the species), the duration of the fishing season is relatively short, or the species is relatively sedentary (Polovina, 1986; Ralston, 1986; Joll and Penn, 1990; Miller and Mohn, 1993). Although they indicate that fisheries have created local depletions of Atka mackerel, these models are difficult to apply to other North Pacific fisheries because of a lack of fishery-independent estimates of biomass and by cir- cumstances unique to the Atka mackerel fishery (e.g., the fishery trawls in areas where the species is known to congregate rather than uses acoustic signal, Atka mackerel are patchily distributed, and patches are separated by areas with low fish density). To obtain information on the winter distribution of groundfish in areas used by foraging Steller sea lions and groundfish fisheries, the Alaska Fisheries Science Center of the National Marine Fisheries Service con- ducted a bottom trawl survey for groundfish in the southeastern Bering Sea north of Unimak Island in February-March 2001 (Fig. 1). This area is important to the Pacific cod fishery in winter because cod ag- gregate in this area to spawn (Shimada and Kimura, 1994). It is also recognized as an important foraging area for Steller sea lions because it is designated as critical habitat under the ESA (NMFS7-8). Fritz and Brown: Interactions between the Pacific cod fishery and Steller sea lions 503 Figure 1 The four areas (high and low sampling-effort survey areas, the area east of the survey area, and the area south of the survey area) in the southeastern Bering Sea that were surveyed in February- March 2001 for groundfish with a bottom trawl and used for analysis of Pacific cod iGadus mac- rocephalus) fishery data. Steller sea lion {Eumetopias jubatus) critical habitat is also shown. In this article, estimates of Pacific cod biomass from Leslie depletion analyses of fishery data are compared with those derived from a bottom trawl survey con- ducted in the same area at the same time. These two methods are independent because they use completely different data to estimate the same parameter, Pacific cod biomass. If they yield similar results, they would support each other in the estimate of local area cod bio- mass and support the use of Leslie depletion analyses of data from relatively short and spatially well-defined fisheries operations for making such estimates. Fur- thermore, these comparisons increase our understand- ing of the potential local effects of a fishery in areas important for sea lion foraging and permit compari- son with the results of assessments of the Pacific cod stock in the entire eastern Bering Sea (Thompson and Dorn, 2002). In this instance, if the change in Pacific cod abundance attributable to the fisheries north of Unimak Island is not greater than what would have occurred if catch were evenly distributed throughout the year and across the range of the stock, then it could be argued that no localized depletion occurred. However, if the local change in abundance is greater than expected, does this constitute a localized deple- tion of the species? The answer ultimately depends on the extent to which the fishery negatively affects the target species (e.g., by reducing recruitment) or, as in our case, by reducing the foraging success of sea lions, which, in turn, could lead to reduced survival or reproductive rates. Although we do not know what the threshold levels of change in local prey densities are for foraging Steller sea lions, it is first necessary to determine the level of change in local abundance that may be attributable to fisheries. There are several aspects of Pacific cod life history in the eastern Bering Sea that make it difficult to use fishery data and the Leslie depletion method to estimate local area biomass. The most important may be that the population in the area fished may not be closed over the time period analyzed. Pacific cod spawn north of Unimak Island in late winter but apparently arrive in groups and, after spawning, leave the area and spread out on the eastern Bering Sea shelf to feed during the remainder of the year (Shimada and Kimura, 1994; Thompson and Dorn, 2002). Seasonal emigration from and immigration into spawning areas in critical habi- tat, modeled with a combination of fishery and survey data by NMFS scientists10 (Fig. 2), provide a baseline 111 NMFS. 2000. Estimation of monthly Pacific cod biomass inside Steller sea lion critical habitat. In Biological opinion questions, NMFS-AKC analytical team. Unpubl. manuscript, 112 p. Alaska Fisheries Science Center, 7600 Sand Point Way NE, Seattle WA 98115. 504 Fishery Bulletin 103(3) 1 o 09 08 07 06 05 04 03 02 0 1 0.0 Feb Mar Apr May Jun Aug Sep Oct Nov Dec Figure 2 Proportion of maximum (in February) biomass of Pacific cod iGadus macrocephalus) within Steller sea lion (Eumetopias jubatus) critical habitat in the eastern Bering Sea by month (see Footnote 10 in the general text). against which possible changes related to local fisheries can be compared. The model results indicate that the highest biomass in critical habitat (largely on the shelf north of Unimak Island) occurs in February, declines to about 10% of the peak in June, and then slowly rebuilds through the summer and fall. Changes in the behavior of Pacific cod immediately prior to or after spawning, such as the formation of dense aggregations or the tem- porary cessation of feeding, would affect catchability by both trawl and fixed gears. However, abrupt changes in catchability due to the formation of aggregations should be evident within the time series of catch and effort data, and changes in feeding habits would not affect the catchability by trawl gear. Methods Bottom trawl survey Stations sampled during the bottom trawl survey were selected by using a stratified random scheme. Two strata were defined: one with a high and another with a low degree of sampling effort, based on the expected distri- bution and abundance of Pacific cod from fishery infor- mation. In the nearshore or high sampling-effort stratum (7765 km2), 38 stations were sampled, whereas 19 sta- tions were sampled in the larger (12,112 km'2), offshore low sampling-effort stratum (Fig. 1). All survey tows were conducted during daylight hours from 16 Febru- ary to 1 March 2001 aboard the FV Northwest Explorer and the FV Ocean Harvester. The 49-m FV Northwest Explorer was equipped with two 1800-hp engines, and the 33-m FV Ocean Harvester had a single 1250-hp engine. Both vessels were house-forward trawlers that had stern ramps, multiple net storage reels, and paired hydraulic trawl winches with 1280-2190 m of 2.54-cm diameter steel cable. Each vessel carried a full comple- ment of navigation and fishing electronics, including global positioning systems (GPS), video position plotters, radars, and depth sounders. A Poly-Nor'eastern high-opening bottom trawl rigged with roller gear was used to sample the groundfish community at each selected location. The trawl net was constructed of 12.7-cm stretched-mesh polyethylene web and had a 3.2-cm stretched-mesh nylon liner in the codend. Accessory gear for the Nor'eastern trawl included three 54.9 m, 1.6 cm diameter galvanized wire rope bridles, and 1.8 x 2.7 m steel V-doors weighing ap- proximately 850 kg each. Biomass (S) estimates for each stratum surveyed were computed by multiplying the average CPUE (in units of kg/km2) for all hauls (n) in a stratum by its area (A). Haul CPUE was calculated as the weight of cod caught (kg) divided by the area swept (a), which was the length of the tow multiplied by the average net width determined by sonic mensuration equipment: kg ■xA. B Confidence bounds on stratum biomass estimates were computed from the standard deviation of the haul CPUEs. For haul CPUEs we assumed a catchability11 of 1 for Pacific cod (all cod within the area swept by Note that catchability within the survey biomass estima- tion procedure has a different literal definition than in the Leslie equation. Fritz and Brown: Interactions between the Pacific cod fishery and Steller sea lions 505 Bottom Trawl Survey - Pacific cod WGTCPUE o Cod CPUE=0 ° Cod CPUE < Mean * Mean < Cod CPUE < Mean + 2 SDs • Mean + 2 SDs < Cod CPUE < Mean + 4 SDs #Cod CPUE > Mean + 4 SDs Figure 3 Catch per unit of effort (CPUE=kg/km2l of Pacific cod {Gadus macrocephalus) during the February- March 2001 bottom trawl survey of the southeastern Bering Sea. "Wgtcpue" refers to the CPUE of Pacific cod from individual hauls (Table 2). Area shading is the same as that in Figure 1. the net are captured) and that it is constant over the course of the survey. This assumption is also made in the Leslie analyses of fishery data. In addition, each haul is assumed to be a random, normally distributed estimate of the density of cod within the stratum. Therefore, the average of the haul CPUEs of cod was assumed to be an unbiased estimate of the true density of cod, allowing linear extrapolation from the CPUE within the area swept to a biomass estimate for each stratum. Analysis of fishery data Fishery observers record a wide variety of information about each haul taken by a fishing vessel, including retrieval location, depth, date and time of catch, and total catch weight (all referred to hereafter as "haul data"). In addition, the catch of a randomly chosen subset of hauls was sampled to determine the species composi- tion of the haul and the length distribution of the target species (see Nelson et al. 1981 and NMFS12 for observer sampling methods). Observer data were queried for any 12 NMFS. 1996. Manual for biologists aboard domestic groundfish vessels, 431 p. U.S. Dep. Commer., NOAA. NMFS, Alaska Fisheries Science Center, 7600 Sand Point Way, NE, Seattle, WA 98115. hauls with any gear in which Pacific cod were caught in the eastern Bering Sea and Aleutian Islands region in 2001. The geographic distribution of the observed Pacific cod catch was used to estimate the distribution of the actual catch of Pacific cod from January-April 2001 in four areas of the southeastern Bering Sea (Fig. 1): the high and low sampling-effort areas surveyed in February-March 2001, and two areas outside of the area surveyed — one to the east, and one to the south. To account for Pacific cod catches in both unsampled hauls and on unobserved vessels, the observed catch of cod was multiplied by the ratio of total-to-observed catch by processing sector and gear type (Table 1). For this procedure, the catch of the unobserved portion of the fleet is assumed to be similar to the observed por- tion. Ratios of total-to-observed catch by sector and gear ranged from 1.02 to 33.94, but for the majority of the catch, the ratios were less than 2 (Table 1). A simple Leslie analysis of fishery catch and effort data was conducted on data collected by observers on- board vessels targeting groundfish. For the basic Les- lie model (Ricker, 1975; Hilborn and Walters, 1992; Gunderson, 1993) a deterministic linear relationship between CPUE and cumulative catch is assumed: Ct ■■qB0-qKt, 506 Fishery Bulletin 103(3) Table 1 Observed and total estimate 5 of total catches of Pacific cod by processor and gear type in the Be ring Sea-Aleutians Island region in 2001, and the ratio o f Total h- Observed catches. CP= =catcher processor; CV=catcher vessel Catches Processor type Gear and ratio CP CV Other Trawl Total (t) 29,398 21,354 734 Observed ( t ) 19,316 8590 720 Ratio 1.52 2.49 1.02 Hook and Total (t) 96,238 637 11,331 line Observed (t) 52,920 19 11,109 Ratio 1.82 33.94 1.12 Pot Total (t) 16,506 478 Observed (t) 4741 469 Ratio 3.48 1.02 directly related to vessel length. With increasing vessel length, horsepower would increase as would the vessel's ability to use larger nets. Vessel length (a surrogate vari- able for horsepower) could be a significant covariate in the relationship between CPUE and cumulative catch. Results Bottom trawl survey Mean CPUE (kg/km2) of Pacific cod in the smaller HSE survey stratum was almost three times higher than in the larger LSE stratum, resulting in mean biomass estimates of 31,312 t and 17,720 t of Pacific cod, respec- tively (Table 2 and Fig. 3). The highest recorded CPUE of cod was recorded for a haul on the northeast side of Unimak Pass (Fig. 3). Hauls with CPUEs above the mean were distributed throughout the HSE stratum in depths less than 200 m. Only one of the 18 hauls in the LSE stratum had a CPUE larger than the mean. For the HSE stratum, the 95% confidence interval on the mean biomass estimate was 19.284-43,339 t. where C, = catch in time period t; ft = effort in t; q = catchability;11 B0 = underlying (or initial) biomass; and Kt = cumulative catch through /. Current catch, effort, and cumulative catch are required by the model, whereas catchability and initial biomass are estimated from it. The catch and effort time series used in these analyses were 1) daily aggregates of observed cod catch in metric tons (t) and effort by ves- sels targeting cod by area (i.e., the high sampling-effort [HSE] area, the low sampling-effort area [LSE], the area east [AE] and the area south [AS] of the survey area), and 2) daily cumulative catch of cod by area for all vessels. CPUE metrics were defined for each gear: 1) trawl as the catch of cod (t) per hour of observed trawl- ing per day; 2) pot as the catch of cod (t) per 20 pots observed per day; and 3) hook and line as the catch of cod (t) per 1000 hooks observed per day. These metrics were chosen so that the CPUE for each gear would be in approximately the same range to permit being plotted together on the same axis. Changing the unit-of-effort definition (number of pots or hooks fished, for instance) has no effect on the significance of the results. Hauls for which cod was the target species were defined as those in which the catch of cod was at least 20% of the total groundfish catch; target levels of 40% and 60% were also explored for trawl fisheries. Catch and effort from these hauls alone, in which cod was the target species, were used for CPUE calculations, whereas cumulative catch was derived from the total catch of cod from all vessels regardless of their target species. The relationship between trawl vessel length and CPUE was investigated but was not included in the Leslie analyses. It was expected that CPUE would be Fishery data Total catch of Pacific cod Approximately 30,500 t of Pacific cod were caught in the four areas of the south- eastern Bering Sea from 1 January to 30 April 2001 (Table 3 and Fig. 4). Almost 60% of this total catch was collected in the HSE survey stratum, whereas 25% and 12% of the total catch were collected in the AE and AS of the survey area, respectively; only 4% was collected in the LSE survey stratum. Based on the distribution of the observed catch of cod by gear, approximately half of the total catch was collected by trawls, a third by hook and line (=longline), and 14% by pots. The distribution of cod catch by area primarily re- flects the distribution of the fishery targeting Pacific cod (Fig. 4). Of the 5813 t of cod that was observed caught by the cod trawl fleet (with at least 20% of each haul composed of cod), 86% was caught in the HSE stratum in over 4600 hours of observed trawling. Most of the remainder (13% or 781 t) was caught east (AE) of the survey area, primarily between the HSE stratum and the 20 nautical mile (nmi) radius trawl exclusion zone encompassing sea lion critical habitat around Sea Lion Rocks and Amak Island (Figs. 1 and 4). There was little trawl effort targeting Pacific cod in the LSE stratum (only 10 observed hours of trawling) or south (17 hours observed) of the survey area. The cod pot fleet worked primarily south of the survey area (57% of their catch) and in the HSE stratum (31%) in areas where conflicts with trawl gear would be minimized. The cod longline fleet worked in both the HSE stratum and to the east of the survey area, and had only trace amounts of catch in the other areas (Table 3). Percentage of Pacific cod in the haul The distribution of the percentage of cod in the total catch of each haul Fritz and Brown: Interactions between the Pacific cod fishery and Steller sea lions 507 0 25 50 Kilometers I i I Figure 4 Locations of groundfish fishery catches of Pacific cod tGadus macrocephalus) in the south- eastern Bering Sea, January-April 2001. The cod target fishery is separated by gear type (trawl = at least 20% of the haul by weight was cod). "All catches of cod" refers to bycatch in trawl fisheries targeting other species. Area shading is the same as that seen in Figure 1. Table 2 Results (catch and biomass of Pacific cod and haul data from the bottom trawl survey of the southeastern Bering Sea con- ducted in February-March 2001 Low and high sampling-effort strata are shown in Figure 1. (CPUE = =catch per unit of effort; CI=confidence interval). Survey stratum Low sampling effort High sampling effort Total Number of hauls 19 38 57 Number of hauls with cod 19 37 56 Mean CPUE (kg cod/km2) 1463 4032 3176 Range in CPUE 65-12,681 0-21,299 0-21,299 Standard deviation of CPUE 2776 4676 4292 Area of stratum (km2) 12,112 7765 19,877 Area of stratum sampled (km2) 0.472 0.927 1.399 % of stratum area sampled 0.004% 0.012% 0.007Q Biomass (t I 17,720 31,312 49.032 95% CI on biomass (t) 1513-33,928 19,284-43,339 20,796-77,267 indicates that the vast majority of the fleet using pots or longline gear were targeting Pacific cod. The total catch of 350 of 351 observed hauls of pots and 777 of 797 observed hauls of longlines was composed of at least 60% cod (Table 4). Therefore, use of a 20% threshold to identify the cod fleet for the longline and pot vessels was unnecessary. For the trawl fleet, however, more than half the observed hauls had less than 10% cod, and 508 Fishery Bulletin 103(3) 63% had less than 20% cod. These trawl vessels were targeting fish species other than Pacific cod, such as rock sole, and caught some cod (as bycatch) in the process. The distribution of hauls that had greater than 20% cod (by 10% bins) was relatively flat, varying only from 4% to 7% between bins and having no clear threshold or breakpoint. Use of a low threshold proportion of cod (such as 20%) would likely include some hauls in which Table 3 Catch and effort statistics for Pacific cod fisheries in the southeastern Bering sea by strata (Fig. 1) in January-April 2001. Statistics include total catch estimates (in metric tons (t); all gear and fisheries), observed catch by all fisheries (by gear type), and observed catch and effort by fisheries targeting Pacific cod (by gear type). Three levels of Pacific cod catches from trawl gear are listed and are based on the minimum proportion of cod in each haul. Strata East of sampling area High sampling effort Low sampling effort South of sampling area Total Catch Total catch Observed catch — all fisheries Trawl Pot Longline Total Observed catch — Pacific cod fisheries Trawl (20% cod in each haul) Trawl (40% cod in each haul) Trawl (60% cod in each haul) Pot Longline Effort Trawl (hours; 20% cod in each haul) Trawl (hours; 40% cod in each haul) Traw] (hours; 60% cod in each haul) Pot (number of pots) Longline (no. of hooks) 7691 17,875 1,200 3724 30,491 1628 5737 324 32 7720 85 655 152 1198 2091 2493 2001 45 116 4654 4205 8393 521 1345 14,465 781 4993 4119 3364 7 32 5813 85 655 152 1198 2090 2493 2001 45 116 4654 677 4644 3768 2903 10 17 5348 1857 10,130 1119 14,816 27,922 220,051 3,265.606 88,880 165,585 7,740,122 Table 4 Frequency distribution of the percentage of cod in each haul by gear for the groundfish fi shery in the four areas of the eastern Bering Sea (Fig. 1) in January-Apri 2001 %cod Trawl Longline Pot No. of hauls % of total No. of hauls % of total No. of hauls % of total <10% 1810 52 0 0 0 0 10-20% 371 11 1 0 0 0 20-30% 237 7 2 0 1 0 30-40% 169 5 1 0 0 0 40-50% 126 4 5 1 0 0 50-60% 126 4 11 1 0 0 60-70% 151 4 40 5 2 1 70-80% 166 5 120 15 4 1 80-90% 181 5 334 42 37 11 90-100% 161 5 283 36 307 87 Total 3498 797 351 Fritz and Brown: Interactions between the Pacific cod fishery and Steller sea lions 509 other species were targeted. On the other hand, the use of a high threshold (such as 60%) might exclude hauls where Pacific cod was the target species. Therefore, a range of trawl target definitions from 20% to 60% was used. The cod trawl fleet distribution shown in Figure 4 was defined by the 20% threshold. If the 40% or 60% thresholds are used, most of the cod trawl effort shown in the HSE area remains, whereas some of the effort in the eastern portions of the AE of the survey area is not coded as the effort of a cod-target fishery. Distribution of Pacific cod catch Cod catches accu- mulated differently in the three primary areas fished (Fig. 5). In the HSE area, cod catches rose steadily from 1 January through early April, and totaled approxi- mately 13,000 t. There was a brief increase in the rate of cod catch in mid-April, but by approximately 20 April, the cod fishery in the HSE area had essentially finished with a catch total of 17,875 t. In the AE of the survey area, cod catches accumulated steadily from 1 Janu- ary through 2 March, and totaled 6340 t. There was a brief increase in catch rates for 6 days from 25 through 30 March, after which the cod fishery in the AE of the survey area was finished with a catch total of 7691 1. In the AS of the survey area, there was little cod fishing effort prior to 22 February, and it lasted only through 27 March, by which time almost 3500 t had been caught; catches through 30 April from the AS of the survey area totaled 3724 t. There was very little cod fishery effort in the LSE area (Table 3), and only 1200 t of cod were caught (principally as bycatch in other fisheries) through 30 April 2001. The longline fleet began fishing for Pacific cod in both the HSE area and AE of the survey area on 1 January (Fig. 5). In the HSE area, daily average longline CPUE (t cod per 1000 hooks per day) remained relatively low and steady, ranging from 0.3-0.7 through January. The longline fleet left the HSE area for approximately two weeks, resuming effort again on 13 February and con- tinuing through 6 March. Longline CPUEs were gener- ally higher in late February than they were in January, ranging from approximately 0.7 to 1.2. The longline fleet again returned to the HSE area on 19-24 March, but daily average CPUEs were <0.5. There was sporadic longline fishing for cod in the HSE area through April, and CPUEs ranged from 0.3 to 1.0. In the AE of the survey area, the longline fleet fished continuously from 1 January through 2 March, and daily average CPUE declined from a range of 0.7-1.0 on 1-7 January to a range of 0.3-0.5 on 24 February-2 March. The trawl fishery for cod began on 20 January in both the HSE area and AE of the survey area (Fig. 5). In the HSE area, trawl CPUE (t cod per hour trawled per day) increased from a range of 0.7-1.4 on 20-27 January to a range of 1.3-2.5 on 6-15 February. From 16 February- 1 March, trawl CPUEs were slightly lower, ranging from 0.8 to 2.0, after which they declined further, ranging only from 0.5 to 1.3 from 2-24 March. On 26 March, the average CPUE increased substantially to over 12 but quickly declined to less than 1.0 by 1 April. This was followed by another short-lived increase in CPUE on 11 April, after which daily average CPUEs remained below 1.0 through April. In the AE of the survey area, CPUEs were highly variable (between 0.4 and 2.3) and there was little observable trend between 20 January and early March. On 25 March, however, average CPUE increased to over 4 and ranged between 0.4 and 3.9 through 2 April, after which there was only sporadic effort and daily average CPUEs were less than 1. The pot fishery for cod began on 22 February south of the survey area and on 24 February in the HSE area (Fig. 5). In the AS of the survey, pot CPUE (t cod per 20 pots per day) decreased from a range of 0.3-1.0 from 22 February-1 March, to a range of 0.2-0.5 on 8-17 March. However, on 18 March, pot CPUE increased to 1.1, and remained between 0.5 and 0.8 through 22 March, after which it quickly declined to very low lev- els. In the HSE area, pot CPUE ranged between 0.7 and 1.7 from 24 February to 23 March. However, on 24-25 March, CPUE was greater than 2. Pot cod fishing oc- curred on only three more days through the end of April in the HSE area: on 27 March, 6 April, and 12 April. Although daily average CPUEs on the last two days were the highest recorded in the pot fishery in 2001, observed catches on these days totaled only 4 and 5 t of cod, respectively. Leslie depletion analyses Leslie depletion analyses were conducted on four sets of Pacific cod fishery data collected in the HSE area and on two sets of data col- lected in the AE of the survey area (Table 5). In the HSE area, longline fishery data collected prior to 13 February and trawl fishery data collected prior to 6 February were excluded from the analyses because CPUE data indicated that fish were immigrating into the area in January in preparation for spawning (Fig. 5). It is unlikely that the increase in CPUE was due to a change in catchability because the increase was evident whether bait was used (pots and longlines) or not (trawls). Data indicating an increase in the abundance of cod north of Unimak Island in January and a peak in February were in agreement with a generalized model of cod abundance in Steller sea lion critical habitat in the eastern Bering Sea (Fig. 2) and seasonal cod movements from tagging data (Shimada and Kimura, 1994). The time series was truncated at 24 March because of the evidence within the fisheries data (increase in CPUE) that another group of cod had immigrated to the HSE area and AE of the survey area in late March or that catchability had increased substantially (Fig. 5). In addition, daily average CPUEs from hauls that had at least 20%, 40%, and 60% Pacific cod by weight were regressed against cumulative catch to see what effect the target definition might have on the regression results. All Leslie regressions with longline or trawl fish- ery data from the HSE area were highly significant (P<0.000001; Table 5 and Fig. 6). Coefficients of de- termination (r2) for the longline and the trawl-20% data were both greater than 0.6. Regression coefficients 510 Fishery Bulletin 103(3) A High sampling-effort area 1-Jan-01 16-Jan-01 31-Jan-01 15-Feb-01 2-Mar-01 17-Mar-01 1-Apr-01 16-Apr-01 1-May-01 1-Jan-01 16-Jan-01 31-Jan-01 15-Feb-01 2-Mar-01 17-Mar-01 1-Apr-01 16-Apr-01 1-May-01 ' C South of survey area 35 0.0 1-Jan-01 x2KJ^5cJ^bfe. 4.000 3,500 3.000 2.500 2.000 1,500 1.000 500 16-Jan-01 31-Jan-01 15-Feb-01 2-Mar-01 17-Mar-01 1-Apr-01 16-Apr-01 1-May-01 Figure 5 Daily average catch per unit effort (CPUE on left y-axis) for the observed Pacific cod (Gadus macrocephalus) fishery by gear (see legend for units) and area (Fig. 1) from 1 January-30 April 2001 in the southeastern Bering Sea. Estimated cumulative catch (t) of cod by all gear types by area is also shown (right y-axis). (slopes) in all cases were negative and significantly different from zero. Collectively, these results strongly indicate that cod fishery CPUE was negatively corre- lated with cumulative catch. Initial biomass estimates (B0) from the four regressions were similar and ranged between 14,119 and 14,806 t, with 95% confidence in- tervals ranging from approximately 9000 to 25,000 t. Use of different fishery catch levels (20%, 40%, 60% cod in each haul) had little effect on the initial biomass estimate but changed the estimate of q, which increased directly with the threshold proportion of cod in each haul (Table 5 and Fig. 7). Fritz and Brown: Interactions between the Pacific cod fishery and Steller sea lions 511 Table 5 Results of Leslie depletion analyses on cod trawl and longline fishery data collected in the (Ai high sampling- effort (HSE) survey- area and (B) east of the survey area • Fig. 2). Dates when data were collected are listed, along with the regression parameters (q = slope and y-intercept= =QB„ and statistics (P=probability that slope is not sig nificant y different from 0, r= Pearson correlation coefficient, and 95% con idence interval (CI) on S„) Fo r the trawl fishery in the HSE area, three different levels catch for the target fishery were used 20', . 40c , or60% of the total catch per haul was cod I. Cumu ative catches in each area are defined as the catch from 1 January through he end of the per iod analyzed. A High sampling-effort survey area Gear Longline Trawl 20% Trawl 40% Trawl 60% 13 Feb-24 Mar 6 Feb-24 Mar 6 Feb-24 Mar 6 Feb-24 Mar Cumulative catch (t) 11,631 11,631 11,631 11,631 B0(t) 14,251 14,806 14.119 14,410 95%CIonB0(t) 9608-22,195 10,549-21,570 9526-21,942 8989-24,860 9 0.000115 0.000172 0.000207 0.000212 v-intercept 1.6395 2.5442 2.9246 3.0573 P <0.000001 <0.000001 <0.000001 <0.000001 No. of days (n) 27 47 46 46 r2 0.712 0.635 0.577 0.479 B East of survey area Gear Longline Trawl 20% 1 Jan-2 Mar 20 Jan-21 Mar Cumulative catch (t) 6340 6837 B0(t) 14,671 95%CIonB0(t) 10,934-20,936 B b ♦ ♦ ♦ ♦ o.o - 6 Daily a\ (CPUE; fishery ] April 2C 0 80 100 120 140 160 180 200 Average vessel length (feet) Figure 8 'erage Pacific cod {Gadus macroeephalus) catch per unit of effort t/h) plotted against daily average vessel length for the trawl cod n the high sampling-effort area in two time periods: 20 January-30 01, and 6 February-24 March 2001. initial influx that peaked in early February because it sustained the fishery for only 1-2 weeks, and resulted in cod catches of only approximately 7500 t from all four areas. In the stock assessment for Pacific cod in the eastern Bering Sea and Aleutian Islands (BSAI; Thompson and Dorn, 2002), the estimate of age 3+ biomass in 2001 was approximately 1.284 million t, whereas the female spawning biomass was approximately 359,000 t. Dou- bling the latter to account for male spawner biomass, the survey and fishery data discussed in the present study indicate that only 4% of the adult spawning and 3% of the age 3+ biomass was in the HSE area, and only about 1% and 4%, respectively, in the entire area surveyed. The area north of Unimak Island is thought to be one of the principal spawning grounds for Pacific cod in the eastern Bering Sea (Shimada and Kimura, 1994; Thompson and Dorn, 2002). The results reported in the present study may indicate that either 1) this is not one of the principal spawning grounds for Pacific cod in the eastern Bering Sea and most spawning oc- curs elsewhere, 2) the stock assessment estimates are too high, or 3) Pacific cod aggregated in the area after the survey occurred. Biomass estimates from the assessment are approxi- mately twice those derived directly from bottom trawl 514 Fishery Bulletin 103(3) lil D. U i.u - 0.9 - 0.8 - 0.7 ■ 0.6 ■ 0.5 - 0.4 - 0.3 ■ n ? - ^""WCI . +-+ • + ' . + . ++ > en No fishing model © Fishing model + Longhne fishery index • Trawl fishery index ■■■■% •+ \ 1/15/2001 1/29/2001 2/12/2001 2/26/2001 3/12/2001 3/26/2001 Figure 9 Comparison of relative abundance of Pacific cod (Gadus maerocephalus) in portions of Steller sea lion lEumetopias jubatus) critical habitat from 15 Janu- ary-24 March 2001 based on 1) no fishing model: the proportion of the maxi- mum biomass (on 15 February) in critical habitat each day; 2) the fishing model: subtracting catch per day from 15 January-24 March 2001 in high and low sampling-effort areas from the no fishing model (total of 12,800 t); 3) longline fishery catch-per-unit-of-effort (CPUE) index of abundance from the high sampling-effort area, 13 February to 24 March (assigned a value of 1 on 13 February); and 4) trawl fishery (20% threshold) CPUE index of abundance from the high sampling-effort area, 6 February to 24 March (assigned a value of 1 on 6 February). surveys of the entire Bering Sea shelf conducted in summer (Thompson and Dorn, 2002). This difference stems from highly domed-shaped selectivity-at-length schedules for the summer surveys and most fishery catches of cod (Thompson and Dorn, 2002). As a conse- quence, the model "assumes" that fewer cod are caught in proportion to their actual abundance at lengths greater than 45 cm for the survey catch and 80 cm for the fishery catch. However, it is unclear how large cod avoid capture during surveys or by longline, pot, and trawl fishery gear as implied by the dome-shaped selectivity-at-length schedules. A seasonal model of Pacific cod movement patterns into and out of Steller sea lion critical habitat (Fig. 2) indicates that relative Pacific cod biomass inside criti- cal habitat is highest in February, then drops 13% in March and 44% by April. If these values are assigned to the middle of each month and daily values are ex- trapolated linearly, the relative change from 15 Febru- ary through 24 March is 23% (Fig. 9). Fishery indices of abundance in the HSE area in January and Febru- ary are consistent with this seasonal pattern, with both trawl and longline CPUEs increasing from Janu- ary to February. According to Figure 2 and the 2001 age 3+ biomass estimate (Thompson and Dorn, 2002), catches through 24 March within the entire survey area (12,806 t) represented only 1% of the BSAI stock and should have reduced the relative biomass of cod within critical habitat by only an additional 2%. Thus, the total reduction in relative cod biomass within critical habitat from mid-February through late March after accounting for fishing and emigration should have been 25% (Fig. 9). Longline and trawl fishery CPUE data in the HSE area provide an independent estimate of relative cod biomass. Both indices indicate that the re- duction in relative cod biomass within the HSE survey area through 24 March was 71-46% greater than that predicted by the model. Catches and biomass estimates of Pacific cod for dif- ferent time periods and areas can be used to compute harvest indices (catch divided by observed biomass). For instance, the harvest index within the entire sur- vey area (based on the catch from 1 January through 24 March and the survey biomass estimate) was 26% (12,806 or-=-49,032). If the focus is narrowed to only the HSE survey area through 24 March, the harvest index was 37% (11,631 or^-31,312). However, both the fish and the fishery were concentrated within the HSE area. The eastern two-thirds of the HSE survey area had survey and fishery-derived biomass estimates of 23,418 t and -14,500 t, respectively. With the area of fishery effort more precisely defined, local harvest indices increase even further, ranging from 50% (11,631 or-23,329) to 80% (11,631 or-r 14,500). Fritz and Brown: Interactions between the Pacific cod fishery and Steller sea lions 515 The annual harvest rate of BSAI cod in 2001 was es- timated to be approximately 11% (Thompson and Dorn, 2002). The total catch of cod in the BSAI through 24 March represented only 44% of the total catch of Pa- cific cod in 2001. Therefore, the harvest rate through 24 March should only have been 44% of 11%, or about 5%. The local harvest indices estimated in the present study, which ranged from 26% to 80%, were five to 16 times greater than that on the BSAI Pacific cod stock as a whole in 2001. Much of the area used by the fish- ery is designated as critical habitat for the endangered Steller sea lion, primarily because of the prey resources available within it. In addition, the fisheries occurred in the winter and early spring, when sea lions are most likely to consume Pacific cod (Sinclair and Zeppelin, 2002). It is not known how or if cod fishery catches in this area affect Steller sea lion foraging success. One objective of the Pacific cod fishery management regulations is to minimize the competitive interactions between locally intense fisheries and Steller sea lions. The suite of groundfish fishery regulations enacted in 2001 and 2002 work together to avoid adverse modifica- tion of critical habitat under the ESA. However, based on the observations during 2001 discussed in the pres- ent study, regulations for the eastern Bering Sea Pacific cod fishery should be reviewed to ensure that they meet these management objectives. Acknowledgments We thank D. DeMaster, G. Duker, B. Fadely, J. Lee, T Loughlin, S. Lowe, S. Moore, and especially M. Sigler for their reviews of early versions of the manuscript. We also give heartfelt thanks to the captains and crews of the FV Northwest Explorer and FV Ocean Harvester, AFSC personnel (E. Acuna, T Buckley, W. Floering, L. Haaga, R. Harrison, E. Jorgensen, G. Lang, D. Nebanzahl, D. Nichol, and K. Smith) who conducted the bottom trawl survey in February-March 2001, and the numerous fishery observers working onboard commercial vessels at that time. Literature cited Braham, H. W., R. D. Everitt, and D. L. Rugh. 1980. Northern sea lion population decline in the eastern Aleutian Islands. Fish. Bull. 44:25-33. Gunderson, D. R. 1993. Surveys of fisheries resources, 248 p. John Wiley & Sons, Inc. New York, NY. Hilborn, R., and C. J. Walters. 1992. Quantitative fisheries stock assessment: choice, dynamics and uncertainty, 570 p. Chapman and Hall, New York, NY. Joll, L. M., and J. W. Penn. 1990. The application of high-resolution navigation systems to Leslie-DeLury depletion experiments for the measurement of trawl efficiency under open-sea conditions. Fish. Res. 9:41-55. Loughlin, T. R., A. S. Perlov, and V. A. Vladimirov. 1992. Range-wide survey and estimation of total number of Steller sea lions in 1989. Mar. Mam. Sci. 8: 220- 239. Lowe, S. A., and L. W. Fritz. 1997. Atka mackerel. In Stock assessment and fish- ery evaluation report for the groundfish resources of the Bering Sea-Aleutian Islands regions as projected for 1998, 653 p. North Pacific Fishery Management Council, P.O. Box 103136, Anchorage, AK 99501. Merrick, R.L., T. R. Loughlin, and D. G. Calkins. 1987. Decline in abundance of the northern sea lion, Eume- topias jubatus, in 1956-86. Fish. Bull. 85: 351-365. Miller, R. J., and R. K. Mohn. 1993. Critique of the Leslie method for estimating sizes of crab and lobster populations. N. Am. J. Fish. Manage. 13:676-685. NRC (National Research Council). 1996. The Bering Sea ecosystem, 307 p. The National Academies Press, Washington, DC. 2003. Decline of the Steller sea lion in Alaskan waters: untangling food webs and fishing nets, 204 p. The National Academies Press, Washington, DC. Nelson, R., Jr., R. French, and J. Wall. 1981. Sampling by U.S. observers on foreign fishing vessels in the eastern Bering Sea and Aleutian Islands region, 1977-78. Mar. Fish. Rev. 43(5):1-19. Polovina, J. J. 1986. A variable catchability version of the Leslie model with application to an intensive fishing experiment on a multispecies stock. Fish. Bull. 84:423-428. Ralston, S. 1986. An intensive fishing experiment for the caridean shrimp, Heterocarpus laevigatus, at Alamagan Island in the Mariana archipelago. Fish. Bull. 84:927-934. Ricker, W. E. 1975. Computation and interpretation of biological statis- tics offish populations. Bull. Fish. Res. Board Canada 191, 382 p. Shimada, A. M., and D. K. Kimura. 1994. Seasonal movements of Pacific cod (Gadus mac- rocephalus) in the eastern Bering Sea and adjacent waters based on tag-recapture data. Fish. Bull. 92: 800-816. Sinclair, E. H., and T. K. Zeppelin. 2002. Seasonal and spatial differences in diet in the west- ern stock of Steller sea lions (Eumetopias jubatus). J. Mammal. 83: 973-990. Thompson, G. G., and M. W. Dorn. 2002. Assessment of the Pacific cod in the eastern Bering Sea and Aleutian Islands area. In Stock assessment and fishery evaluation report for the groundfish resources of the Bering Sea-Aleutian Islands regions, p. 121- 206. North Pacific Fishery Management Council, 605 W 4th Avenue, Suite 306, Anchorage, AK 99501. 516 Abstract— Molecular-based approach- es for shark species identification have been driven largely by issues specific to the fishery. In an effort to estab- lish a more comprehensive identifica- tion data set, we investigated DNA sequence variation of a 1.4-kb region from the mitochondrial genome cover- ing partial sequences from the 12S rDNA. 16S rDNA, and the complete valine tRNA from 35 shark species from the Atlantic fishery. Generally, within-species variability was low in relation to interspecific divergence because species haloptypes formed monophyletic groups. Phylogenetic analyses resolved ordinal relation- ships among Carcharhiniformes and Lamniformes, and revealed support for the families Sphyrnidae and Tri- akidae (within Carcharhiniformes) and Lamnidae and Alopidae (within Lamniformes). The combination of limited intraspecific variability and sufficient between-species divergence indicates that this locus is suitable for species identification. Mitochondrial gene sequences useful for species identification of western North Atlantic Ocean sharks Thomas W. Greig M. Katherine Moore Cheryl M. Woodley National Ocean Service National Center for Coastal Ocean Science Center for Coastal Environmental Health and Biomolecular Research at Charleston 219 Fort Johnson Road Charleston, South Carolina 29412-9110 E-mail address (for T W. Greig) Thomas Greig (ginoaa gov Joseph M. Quattro Department of Biological Sciences School of the Environment University of South Carolina Columbia, South Carolina 29208 Manuscript submitted 22 June 2004 to the Scientific Editor's Office. Manuscript approved for publication 28 March 2005 by the Scientific Editor. Fish. Bull. 103:516-523 (2005). Seventy-three species of sharks inhabit the United States territorial waters of the Atlantic Ocean, Gulf of Mexico, and Caribbean Sea (Compagno, 1984a, 1984b). All but one (spiny dogfish, Squalus acanthias, managed separately) are managed under the current Fisheries Management Plan (FMP) for highly migratory species (NMFS1). Thirty-three species are of lesser commercial importance and are relegated to the "deepwater and other" species management group, and 19 species cannot be landed commercially or recreationally ("prohibited species" group). The remaining 20 species are of interest to the commercial shark fishery and are categorized as large coastal species (LCS), small coastal species (SCS), and pelagic species management units in the current FMP. Although these management units are practical, it is clear that species respond uniquely to exploita- tion and therefore should be managed on a species-by-species basis (Castro et al., 1999; NMFS2). Species-level management is widely recommended (e.g., FAO Marine Resource Service, 2000) but is complicated by the pau- city of species-specific fisheries data, stemming, in part, from an inability to accurately identify species. Many commercially important spe- cies (e.g., within Carcharhiniformes) are difficult to identify whole, and this task is more daunting if indi- viduals are processed (head, entrails, and fins are removed); unfortunately, at-sea processing is widespread in the industry (Castro3). Although current U.S. legislation prohibits the practice of "finning" (where fins are retained and carcasses are discarded at sea). 1 NMFS (National Marine Fisheries Ser- vice). 2003. Final amendment 1 to the fishery management plan for Atlantic tunas, swordfish and sharks, 599 p. Of- fice of Sustainable Fisheries, Highly Migratory Species Management Division, NMFS, NOAA, 1315 East West Highway, SSMC3, Silver Spring, MD 20910. 2 NMFS (National Marine Fisheries Ser- vice). 2001. Final United States na- tional plan of action for the conservation and management for sharks, 90 p. Of- fice of Sustainable Fisheries, Highly Migratory Species Management Division, NMFS, NOAA, 1315 East West Highway, SSMC3, Silver Spring, MD 20910. 3 Castro, J. I 1993. A field guide to the sharks commonly caught in com- mercial fisheries of the southeastern United States. NOAA Tech. Memo. NMFS-SEFSC-338, 47 p. Southeast Fisheries Science Center, NMFS, NOAA, 75 Virginia Beach Dr., Miami, FL 33149. Greig et al.: Gene sequences useful for identification of western North Atlantic shark species 517 the landing of fins is allowed where carcasses and fins are off-loaded at the same time in a no more than 1:20 (fin-to-carcass) weight ratio. However, serious problems can arise in matching off-loaded fins to processed car- casses. In and of itself, the landing of shark fins can be lucrative; fins accounted for more than 50% of the total Atlantic shark fishery value in 2002 (NMFS4). Because preferences exist for fins from certain species, exvessel prices for specific types of fin vary consider- ably (e.g., Weber and Fordham, 1997). It is perhaps not surprising that augmenting the fin-to-carcass ratio with spoiled meat or "finning" target species out of season (and subsequently attributing the fins to fish that are allowed to be caught during the season) might not be uncommon (Vannuccini, 1999). Clearly, these possibili- ties lead to the challenge of matching collected fins to processed carcasses. Therefore, accurate and reliable species identification methods are paramount for law enforcement and sound species management. Molecular species identification research on sharks has been driven largely by resolution of specific prob- lems associated with the fishery. For example, Heist and Gold (1999) used mtDNA sequence data to develop restriction fragment assays that differentiate 11 species of carcharhiniform sharks commonly encountered in the LCS fishery. Similarly, Pank et al. (2001) used multiplex PCR to differentiate two morphologically similar shark species (Carcharhinus obscurus and C. plumbeus) — an approach that was expanded by Shivji et al. (2001) to include five additional species (with some overlap of species included by Heist and Gold 1999). Both ap- proaches are relatively rapid, inexpensive, and easily implemented; however, they appear most applicable when the number of species investigated is limited. In sum, of the thirty-nine species of sharks that are not in the "deepwater and other" management group, molecu- lar species identification assays have been developed for fifteen species (9 LCS, 3 pelagic, and 3 in the prohibited species management groups) (Heist and Gold, 1999; Pank et al., 2001; Shivji et al., 2001), leaving 24 species without molecular methods for identification. Some investigators have instead turned to DNA se- quence analysis to resolve issues of species identification (Takeyama et al., 2001; Akimoto et al., 2002; Jerome et al., 2003). This approach is exemplified best by the recent development of computer interfaces that allow access to and analysis of large DNA databases (DNA Surveillance, Ross et al., 2003; ARB, Ludwig et al., 2004). Simply put, these databases circumvent the te- dious process of scanning large taxonomically diverse DNA repositories (e.g., GenBank) by allowing the user to access (DNA Surveillance) or maintain (ARB) taxo- nomically restricted sets of reference sequences. Users 4 NMFS (National Marine Fisheries Service). 2003. Stock assessment and fishery evaluation report for Atlantic highly migratory species (SAFE), 274 p. Office of Sustainable Fisheries, Highly Migratory Species Management Division, NMFS, NOAA, 1315 East'West Highway, SSMC3. Silver Spring, MD 20910. can submit "unknown" sequences to compare against specified sequence subsets; subsequent analyses are returned as genetic distances (between unknown and reference sequences) and include a phylogenetic hy- pothesis. The power of this approach lies in the ease with which reference sequences can be added to the data- base, in the "quality-control" that can be exerted over subsequent additions to the reference sequences, and in the ease with which geographic variation within species can be included. The success of this approach, however, hinges on the information contained in the gene in the reference database. The inception of this approach, as applied to commercially important sharks, requires a sufficiently informative set of reference sequences against which searches can be made. The aforemen- tioned molecular approaches (RFLP, multiplex PCR) include a diversity of gene regions (mitochondrial DNA, nuclear ITS); thus no comprehensive data set exists for commercially landed Atlantic shark species. Fortu- nately, recent work with a 2.4-kb fragment of the mi- tochondrial genome (spanning 12S rDNA to 16s rDNA) to examine the phylogenetic relationships among shark orders has shown that this region may be useful in re- solving relationships at this taxonomic level (Douady et al., 2003). Unfortunately, sampling within orders was limited, and it is thus unknown whether this region contains sufficient phylogenetic signal at lower taxo- nomic levels. We present here mtDNA sequence data of a smaller fragment of the same region containing partial se- quence information for the mitochondrial 12S rDNA, 16S rDNA, and the complete valine tRNA from 35 shark species (including all 20 commercially exploited species, 12 of 19 prohibited species, the spiny dogfish, and two species of Mustelus). We suggest that a suitable locus for species-identification purposes will permit identifica- tion of unequivocally distinct species (i.e., large genetic differentiation between species compared to within spe- cies) and offer the potential for meaningful phylogenetic comparisons (important when "query" animals are ab- sent or not adequately represented in a molecular data- base). Keeping in mind issues of species identification and fisheries management, we examine this mtDNA region for patterns of genetic variability and assess its utility in phylogenetic reconstruction. We then discuss the use of this region for the underpinnings of a vali- dated reference DNA database suitable for forensic and fisheries management applications. Methods Sample collection Voucher Atlantic Ocean shark samples (muscle, fin, or blood) were obtained from the CCEHBR Marine Foren- sics archive in Charleston, SC (Table 1). Samples were accompanied by species certification and chain-of-cus- tody forms. Muscle and fin samples were either frozen at 518 Fishery Bulletin 103(3) Table 1 Scientific and common names of samples, number of individuals sampled (n), species codes, and Genbank accession numbers. Taxonomy follows Campagno (1984. 2001). Species codes correspond to a representative individual in the National Ocean Service Marine Forensics Program (CCEHBR. Charleston, SO tissue archive with that particular haplotype (except for Heterodontus franeisei Hfral). Order Family and species Common name Code ( ;i) Accession Carcharhiniformes Lamniformes Carcharhinidae Carcharhinus acronotus Blacknose Cacr003(3) AY830721 C. altimus Bignose Calt001(2) AY830722 C. brevipinna Spinner Cbre001(3) AY830723 C. falciformis Silky Cfal003(l) AY830725 Cfal006(l) AY830726 C. isodon Finetooth Ciso004(l) AY830727 CisoOlO(l) AY830728 Ciso015(l) AY830729 C. leucas Bull Cleu003(3) AY830730 C. limbatus Blacktip Clim004(l) AY830731 Clim006(2) AY830732 C. longimanus Oceanic whitetip ClonOOO(l) AY830736 Clon002(l) AY830733 Clon005(l) AY830734 Clon006(l) AY830735 C. obscurus Dusky CobsOOO(l) AY830737 Cobs001(3) AY830738 C. perezi Caribbean reef Cper001(2) AY830739 Cper002(2) AY830740 C. porosus Smalltail CporOOKl) AY830743 C. plumbeus Sandbar Cplu004(2) AY830741 Cplu023(l) AY830742 C. signatus Night Csig002(l) AY830744 Galeocerdo cuvier Tiger Gcuv003(3) AY830746 Negaprion brevirostris Lemon Nbre005(l) AY830756 Prionace glauca Blue Pgla004(l) AY830760 Pgla0020(l) AY830761 Pgla0022(l) AY830762 Rhizoprionodon terraenovae Sharpnose Rter001(2) AY830763 Rter026(ll AY830764 Sphyrnidae Sphyrna lewini Scalloped hammerhead Slew003(2) AY830768 S. mokarran Great hammerhead Smok003(3) AY830769 S. tiburo Bonnethead Stib016(2) AY830770 Stib018(l) AY830771 S. zygaena Smooth hammerhead Szyg681(6) AY830772 Triakidae Mustelus eanis Smooth dogfish Mcan003(3) AY830754 M. norrisi Florida smoothhound Mnor001(2) AY830755 Alopiidae Alopias superciliosus Bigeye thresher AsupOOKll AY830718 Asup006(l) AY830719 A. vulpinus Thresher Avul002(l) AY830720 Lamnidae Careharodon carcharias White Ccar002(3) AY830724 Isurus oxyrinchus Shortfin mako Ioxy005(l) AY830747 Ioxy032(l) AY830748 Ioxy051(l) AY830749 I. paucus Longfin mako Ipau002(2) AY830750 lpau005(l) AY830751 Lamna nasus Porbeagle Lnas001(2) AY830752 Lnas003(l) AY830753 continued Greig et al.: Gene sequences useful for identification of western North Atlantic shark species 519 Table 1 (continued) Order Family and species Common name Code i n I Accession Odontaspididae Carcharius taunts Sand tiger Otau004(ll Otau005(l) Otau007(l) AY830757 AY830758 AY830759 Orectolobiformes Ginglymostomatidae Ginglymostoma cirratum Nurse Gcir001(2) AY830745 Hexanchiformes Hexanchidae Hexanchus vitulus Bigeye sixgill Hvitlll) AY830716 Heptranchias perlo Sevengill Hperl(l) AY830715 Squaliformes Squalidae Squalus acanthias Spiny dogfish Saca002(l) Saca003(2) AY830765 AY830766 Squatiniformes Squatinidae Squatina dumeril Atlantic angel Sdum001(3) AY830767 Heterodontiformes Heterodontidae Heterodon tus fra n cisci Horn shark Hfra(l) NC003137 -80°C, dried, or stored in 70% EtOH. Blood was stored at room temperature in sodium dodecyl sulfate-urea (SDS- urea: 1% SDS, 8M urea, 240 mM Na2HP04, ImM EDTA pH 6.8). Total nucleic acids were extracted from frozen, dried, and EtOH-preserved samples by using DNeasy Tissue Kits and following manufacturer's recommenda- tions (Qiagen, Valencia, CA). DNA was isolated from blood in SDS-urea according to White and Densmore (1992; protocol 11). Extracted DNA was visualized by electrophoresis in a 1% agarose gel stained with 0.4 ng/mL of ethidium bromide in lx Tris-borate-EDTA (TBE: 89 mM Tris-borate, 2 mM Na2EDTA, pH 8). A 1-kb DNA ladder (Promega, Madison. WD was used as a size standard. Amplification and sequencing Primers 12SA-5' and 16SA-3' (Palumbi, 1996) were used to amplify an approximately 1400-bp region spanning the 3' end of the 12s rDNA, the valine tRNA, and the 5' end of the 16s rDNA region of mitochondrial DNA (mtDNA). Samples were amplified in 50 uL reactions containing -50 ng of template DNA, 20 mM Tris-HCl pH 8.4, 50 mM KC1, 0.2 mM each dNTP, 2 mM MgCl2, 20 mM each primer, and 2.5 units Taq DNA polymerase (Gibco BRL, Rockville, MD). Thermal cycling consisted of an initial denaturation at 94°C for 1.5 minutes, followed by 30 cycles of 40 seconds at 94°C, 40 seconds at 52°C, and 50 seconds at 72°C, and a final extension step of 15 minutes at 72°C. Negative controls (no template) were included in each set of reactions. PCR products were gel-purified as described in Rosel and Block (1996) and 20-50 ng were used as template for ABI Big Dye Ter- minator (v. 1.0, Applied Biosystems, Foster City, CA) cycle sequencing reactions. Sequence was obtained with amplification primers 12SA-5', 16SA-3' and two addi- tional internal sequencing primers. Sequencing reaction products were precipitated with ethanol, washed accord- ing to sequencing kit instructions, dried in a Savant Speedvac Plus, and resuspended in 4 j

0.05) over the sampling period. However, the aver- age shell gland weight (Fig. 1C) from skates captured in October was greater (P<0.05) than those captured in September. Because all shell glands from skates captured in February were in the process of encapsulating ovulated eggs, we were unable to obtain accurate individual shell gland weights. There were no differences (P>0.05) observed in the average diameter of the two largest follicles (Fig. ID), and no pattern of follicle dynamics was discerned. Also, fully formed egg cases, or those in the process of formation, were found in the uteri of skates captured during all months of the year, except June and September. Additional analysis revealed that GSI was correlated to shell gland weight (r=0.53) and average follicle diameter (r=0.4). Further- more, HSI was also correlated to shell gland weight (r=0.53) and average follicle diameter (r=0.7). Sulikowski et al.: The reproductive cycle of Amb/yra/a radiata 539 Assessment of morphological parameters in the male reproductive tract Histological stages III through VI (SIII-SVI) of spermatogenesis were examined, and GSI and HSI were determined for the 48 males col- lected during 24 months of sampling. Although the relative proportion of these four stages did not differ among months, it is notable that the production and maintenance of mature sper- matocysts (SVI) within the testes persisted throughout the year (Fig. 2A). Similarly, no significant seasonal differences were found in HSI or GSI (Fig. 2, B and C, respectively). In addition, there were weak to no correlations between spermatogenesis and either HSI or GSI (r=-0.07 and 0.13, respectively). Synchronicity between male and female reproductive cycles Results from the male and female morpho- logical reproductive parameters indicated that thorny skates are capable of reproducing throughout the year in the western Gulf of Maine. When GSI, follicle diameter in relation to percent composition of SVI, or shell gland weight in relation to percent composition of SVI were compared between male and female thorny skates, no apparent correlation was detected (Fig. 3, A-C). In contrast, when per- cent composition of SVI (spermatogenesis) was plotted against percentage of captured female skates with egg cases, a strong synchronicity (r=0.51) was observed (Fig. 4). Discussion Elasmobranchs display a wide range of repro- ductive strategies with morphological and physiological specializations for oviparous or viviparous reproduction (Wourms and Demski, 1993; Hamlett and Koob, 1999). These strategies are associated with one of three basic types of reproductive cycles: 1) reproduction throughout the year, 2) a partially defined annual cycle with one or two peaks, and 3) a well-defined annual or biennial cycle (Wourms, 1977; Hamlett and Koob, 1999). Among oviparous elasmobranchs, some species exhibit cycles with clearly delineated period* s) of reproductive activity interspersed between periods of little or no activity. For example, in the clearnose skate (Raja eglanteria), the patterns of estradiol concentrations and follicle dynamics indicate the presence of a well-defined annual reproduc- tive cycle, in which mating and egg deposition take place from December to mid May (Rasmussen et al., 1999). Likewise, hormone and morphological data also indicate a defined annual cycle in the epaulette shark (Hemiscyl- lium ocellatum) (Heupel et al., 1999) and that reproduc- tive activities take place from July to December. 90 - c B 80 - 70 - 60 - 50 - ■/K 40 - I 1 / 30 - A 4 0 4 2 3 8 6 3 3 3 5 5 Jan Feb Mar April May June July Aug Sept Oct Nov Dec 34 - 32 31 Z 30 - 29 D 8 6 Jan Feb Mar April May June July Aug Sept Oct Nov Dec Figure 1 (continued) In contrast, other oviparous elasmobranchs exhibit re- productive activity year round. For example, the present study revealed that female thorny skates are capable of reproducing throughout the year. This conclusion was based on GSI, shell gland weight, diameter of the larg- est preovulatory follicles, and the presence of egg cases in specimens collected over the course of the study. We also observed that GSI and shell gland weight were highest in October. Thus, the period (or periods) of enhanced reproductive activity appears to be an in- tegral part of continuous cycles, although the specific measured parameters or when these periods occur may vary between species. In a study of thorny skates sampled from August to December in NAFO Division 3N, females were found to be reproductively active over the entire sampling inter- val, and peak egg case production occurred in September (Del Rio, 2002). In contrast, although large preovulatory follicles were present and oviposition occurred through- out the reproductive cycle of the lesser spotted dogfish 540 Fishery Bulletin 103(3) Stage in I : Stage IV ^■i Stage V i ! Stage VI i 0.8 - c 07 ■ 06 - K^^ r-" -Kj 0.5 - 0.4 ■ 3 2 3 3 ; 5 5 4 7 4 5 4 Jan Feb Match April May June July Aug Sept Ocl Nov Dec Figure 2 Monthly changes in male thorny skates (A. radiata): (A) The mean percent of each stage of spermatogenesis (stages III through VI) found along a transect line across one representative full lobe cross section of a testis; (B) hepatosomatic index (HSI) and; (C) gonadosomatic index (GSI). Sample sizes are indicated above each month. Values are expressed as mean ±SEM. (Scyliorhinus canicula) (Henderson and Casey, 2001), ovary weight and egg deposition peaked during spring. Similarly, several morphological parameters and steroid hormones have been shown to peak in female winter skates (Leucoraja ocellata) during the summer, and egg-case production is highest in the fall (Sulikowski et al., 2004). Lastly, in L. erinacea, examination of fol- licle dynamics and egg-case production indicated that a higher proportion of females are reproductively active during two periods of time in the reproductive cycle: in the winter and in the summer (Richards et al., 1963). The fairly consistent pattern of HSI in female thorny skates over the reproductive cycle indicated that liver reserves (such as lipids and proteins used for oocyte growth) were stored and metabolized continuously throughout the year without a significant change in whole organ biomass. This is in contrast to other ovip- arous species, such as S. canicula, which displayed seasonal variations in liver mass as a result of lipid deposition occurring during different times of the re- productive cycle (Craik, 1978). The continual presence of mature spermatocysts with- in the testes over the entire sampling period indicateded that male thorny skates are also capable of reproducing throughout the year. Information describing the annual reproductive cycles of oviparous male elasmobranchs is Sulikowski et al.. The reproductive cycle of Amblyro/a radiata 541 0 60 CD 28 80 75 70 s 65 73 C a 60 55 CO 50 45 40 35 30 Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec 40 35 25 S 20 Figure 3 Comparisons between male and female thorny skate (A. radiata) reproductive parameters over the course of the sampling period: (A) GSI; (B) diameter of the two largest follicles and percentage of spermatocysts (SVIl and; (C) shell gland weight and percentage of spermatocysts (SVI). very limited because studies have focused on changes in morphological parameters (i.e., Richards et al., 1963; Craik, 1978) or steroid hormone analyses (i.e., Sumpter and Dodd, 1979; Rasmussen et al., 1999) in females. To our knowledge, the only two species in which quantita- tive methods were used to describe annual reproductive patterns in males were H. ocellatum (Heupel et al., 1999) and L. ocellata (Sulikowski et al., 2004). These two species exhibit contrasting strategies in their re- spective reproductive cycles. For example, similar to male thorny skates from the present study, male winter skates appear capable of continuous production of ma- ture spermatocysts throughout the year (Sulikowski et al., 2004). In contrast, examination of the testes and circulating hormone concentrations in H. ocellatum indicated that sperm production and androgen concen- tration display a concurrent seasonal cycle that peaks from June to October (Heupel et al., 1999). The lack of correlation between GSI or HSI and stage of spermatogenesis in the thorny skate was not surpris- ing because studies do not support the assumption that relative gonad size (or storage products in the liver) and reproductive readiness are positively correlated (Teshima, 1981; Parsons and Grier, 1992; Maruska et al., 1996). For instance, neither peak sperm production (Maruska et al., 1996) nor the pattern of testosterone concentration was correlated with GSI in Dasyatis sa- bina (Snelson et al. 1997) or L. ocellata (Sulikowski et al., 2004). Relatively few studies have assessed whether cycli- cal patterns of reproductive morphological parameters or hormone concentrations are coordinated between 542 Fishery Bulletin 103(3) Comparisons percentage of males and females over the course of their reproductive cycles. Among them, coordinated peaks in gonad weight and steroid hormone concentrations in win- ter skates (Sulikowski et al., 2004) and epaulette sharks (Heupel et al., 1999) were observed in males and females over an annual cycle. In the present study, mature spermatocysts (SVI) and per- centage of female thorny skates with egg cases were also synchronized over the course of the study. In contrast, Henderson and Casey (2001) found that the gonadal cycles of male and female lesser spotted dogfish were asynchro- nous, which they hypothesized to be due to the storage of sperm by females. Sperm storage has been documented in other female elasmobranch species as well (e.g., Pratt, 1993; Maruska et al., 1996) and is thought to be a feature pri- marily of species that are nomadic or segregated by sex (Pratt, 1993). In the current study, A. radiata was neither segregated by sex (both genders were captured in the same area and in the same trawls) nor found to be no- madic in their movement patterns (Templeman, 1987; Sulikowski, unpubl. observ. ). Moreover, because males are capable of producing viable sperm and females ap- pear to be reproductively active throughout the year, there is probably no need for the population of thorny skates that we sampled to store sperm. On the basis of the above information, we believe that the reproduc- tive cycle in the sampled population of thorny skates is coordinated over an annual cycle. In summary, according to the reproductive strategies outlined by Wourms (1977) and later by Hamlett and Koob (1999), the results of the present study indicate that thorny skates have a reproductive cycle that is con- tinuous throughout the year. For females, this conclu- sion was based on ovary weight, shell gland weight, and diameter of the largest follicles (the preovulatory fol- licles). For males, this conclusion was based on the pres- ence of mature spermatocysts within the testes over the course of the sampling period. Moreover, comparisons between the proportion of mature spermatocysts within the testes and the percentage of egg-case-bearing fe- males indicate that the reproductive cycles of male and female thorny skates are synchronized. Currently, analyses of circulating steroid hormone concentrations are in progress for the thorny skates used in the pres- ent study, which may provide additional insight into the regulation and timing of reproductive events in this species. Acknowledgments Collection of skates was conducted on the FV Mystique Lady. We thank Noel Carlson for maintenance of the fish 25 20 Jan Feb March April May June July Aug Sept Oct Nov Dec Figure 4 between the percentage of spermatocysts (SVI) and the female thorny skates (A. radiata) with egg cases. at the U.N.H. Coastal Marine Laboratory. This project was supported by a Northeast Consortium grant (no. NA16FL1324) to PCWT, JAS, and PDD. Literature cited Brander, K. 1981. Disappearance of common skate Raja batis from Irish Sea. Nature 290 (58011:48-49. Casey, J. M., and R. A. Myers. 1998. Near extinction of a large widely distributed fish. Science 28:690-692. Collette, B., and G. Klein-MacPhee 2002. Fishes of the Gulf of Maine, 3rd ed., p. 62-66. Smithsonian Institution Press, Washington, D.C. Compagno, L. J. V., D. A. Ebert, and M. J. Smale. 1989. Guide to the sharks and rays of southern Africa, 158 p. New Holland (Publ.) Ltd., London. Conrath, C. L., J. Gelsleichter, and J. A. Musick. 2002. Age and growth of the smooth dogfish, Mus- telus canis, in the northwest Atlantic. Fish. Bull. 100:674-682. Craik, J. C. A. 1978. An annual cycle of vitellogenesis in the elasmo- branch Scyliorhijius canicula. J. Mar. Biol. Assoc. UK. 58:719-726. Del Rio, J. L. 2002. Some aspects of the thorny skate, Amblyraja radi- ata, reproductive biology in NAFO Division 3N. NAFO SCR Doc. 02/118, serial no. N4739, 14 p. Dulvy, N. K., J. D. Metcalfe, J. Glanville, M. G. Pawson, and J. D. Reynolds. 2000. Fishery stability, local extinctions, and shifts in community structure in skates. Cons. Biol. 14: 283-293. Francis, M., C. O., Maolagain, and D. Stevens. 2001. Age, growth, and sexual maturity of two New Zealand endemic skates, Dipturus nasutus and D. Suhkowski et al.: The reproductive cycle of Amblyra/a radiota 543 innominatus. N.Z.J. Mar. Freshw. Res. 35:831- 842. Frisk, M. G., T. J. Miller, and M. J. Fogarty. 2001. Estimation and analysis of biological parameters in elasmobranch fishes: a comparative life history study. Can. J. Fish. Aquat. Sci. 58:969-981. Hamlett W. C, and T. J. Koob. 1999. Female reproductive system. In Sharks, skates and rays; the biology of elasmobranch fish (W. C. Hamlett, ed. I, 515 p. Johns Hopkins Univ. Press, Baltimore, MD. Henderson, A. C. and A. Casey. 2001. Reproduction and growth in the lesser-spotted dogfish Scyliorhinus canicula (Elasmobranchii: Scyli- orhinidael, from the west coast of Ireland. Can. Biol. Mar. 42:397-405. Heupel M. R., J. M. Whittier, and M. B. Bennett. 1999. Plasma steroid hormone profiles and reproduc- tive biology of the epaulette shark, Hemiscyllium ocellatum. J. Exp. Zool. 284:586-594. Hoenig, J., and S. H. Gruber. 1990. Life history patterns in the elasmobranchs: Impli- cations for fisheries management. In Elasmobranchs as living resources: advances in the biology, ecology, systematics and the status of the fisheries (H. L. Pratt Jr, S. H. Gruber, and T. Tanuichi, eds. I, p. 1-16. NOAA Technical Report, NMFS 90. Koob T. J.. P. Tsang. and I. P. Callard. 1986. Plasma estradiol, testosterone and progesterone levels during the ovulatory cycle of the little skate, Raja erinacea. Biol. Reprod. 35:267-275. Maruska K. P., E. G. Cowie, and T. C. Tricas. 1996. Periodic gonadal activity and protracted mating in elasmobranch fishes. J. Exp. Zool. 276: 219-232. Parsons G. R., and H. J. Grier. 1992. Seasonal changes in the shark testicular structure and spermatogenesis. J. Exp. Zool. 261:173-184. Pratt, H. L . 1993. The storage of spermatozoa in the oviductal glands of western North Atlantic sharks. Environ. Biol. Fish. 38:139:149. Rasmussen L E. L, D. L. Hess, and C.A. Luer. 1999. Alterations in serum steroid concentrations in the clearnose skate, Raja eglanteria: correlations with season and reproductive status. J. Exp. Zool. 284:575-585. Richards S. W., D. Merriman, and L. H. Calhoun. 1963. Studies in the marine resources of southern New England. IX. The biology of the little skate Raja erinacea, Mitchill. Bull. Bingham. Oceanogr. Coll. 18:311-407. Robins, C. R„ and G. C. Ray. 1986. A field guide to Atlantic coast fishes of North America, 354 p. Houghton Mifflin Co., Boston, MA. Simpfendorfer, C. A. 1993. Age and growth of the Australian sharpnose shark. Rhizoprionodon taylori, from north Queensland, Australia. Environ. Biol. Fish. 36:233-241. Snelson F. F. Jr, S. E. Williams-Hopper, and T. H. Schmid. 1988. Reproduction and ecology of the Atlantic stingray, Dasyatis sabina, in Florida coastal lagoons. Copeia 1988:729-739. Snelson F. F. Jr, L. E. L. Rasmussen, M. R. Johnson, and D. L. Hess. 1997. Serum concentrations of steroid hormones during reproduction in the Atlantic stingray, Dasyatis sabina. Gen. Comp. Endocrinol. 108:67-79. Sulikowski, J. A.. M. D. Morin, S. H. Suk and W. H. Howell. 2003. Age and growth of the winter skate, Leueoraja ocel- lata, in the Gulf of Maine. Fish. Bull. 101:405-413. Sulikowski J. A., P. C. W. Tsang, and W. Huntting Howell. 2004. An annual cycle of steroid hormone concentrations and gonad development in the winter skate, Leueoraja oeellata. from the western Gulf of Maine. Mar. Biol. 144:845-853. Sumpter J. P., and J. M. Dodd. 1979. The annual reproductive cycle of the female lesser spotted dogfish, Scyliorhinus canicula, and its endocrine control. J. Fish. Biol. 15:687-695. Templeman, W. 1982. Development, occurrence and characteristics of egg capsules of the thorny skate, Raja radiata, in the North- west Atlantic. J. Northw. Atl. Fish. Sci. 3:47-56. Templeman, W. 1987. Differences in sexual maturity and related char- acteristics between populations of thorny skate Raja- radiata in the northwest atlantic. J. Northw. Atl. Fish. Sci. 44 (11:155-168. Teshima, K. 1981. Studies on the reproduction of the Japanese smooth dogfishes, Mustelus manazo and Mustelus griseus. J. Shimonoseki. Univ. Fish. 29:113-199. Tricas T. C, K. P. Maruska, and L. E. L. Rasmussen. 2000. Annual cycles of steroid hormone production, gonad development, and reproductive behavior in the Atlantic stingray. Gen. Comp. Endocrinol. 118:209-225. Tsang P., and I. P. Callard. 1987. Morphological and endocrine correlates of the repro- ductive cycle of the aplacental viviparous dogfish, Squa- lus acanthias. Gen. Comp. Endocrinol. 66:182-189 Winemiller, K. O., and K. A. Rose. 1992. Patterns of life history diversification in North American fishes: implication for population regu- lation. Can. J. Fish. Aquat. Sci. 49:2196-2218. Wourms, J. P. 1977. Reproduction and development in chondrichthyan fishes. Am. Zool. 17:379-410. Wourms, J. P., and L. S. Demski. 1993. The reproduction and development of sharks, skates, rays and ratfishes: introduction, history, overview and future prospects. Environ. Biol. Fish. 38:7-21. Zeiner, S. J., and P. G. Wolf. 1993. Growth characteristics and estimates of age at maturity of two species of skates (Raja binoculata and Raja rhina) from Monterey Bay. California. In Con- servation biology of elasmobranchs, p. 87-90. NOAA Technical Report NMFS 115. 544 Effect of type of otolith and preparation technique on age estimation of larval and juvenile spot (Leiostomus xanthurus) Dariusz P. Fey Sea Fisheries Institute Dept. of Fisheries Oceanography and Marine Ecology ul Kollataja 1 81-332 Gdynia, Poland E-mail address dfeyig mirgdynia pi Gretchen E. Bath Martin James A. Morris Jonathan A. Hare NOAA National Ocean Service Center for Coastal Fisheries and Habitat Research 101 Pivers Island Road Beaufort, North Carolina 28516-9722 Otoliths of larval and juvenile fish provide a record of age, size, growth, and development (Campana and Neil- son, 1985; Thorrold and Hare, 2002). However, determining the time of first increment formation in otoliths (Campana, 2001) and assessing the accuracy (deviation from real age) and precision (repeatability of incre- ment counts from the same otolith) of increment counts are prerequisites for using otoliths to study the life his- tory offish (Campana and Moksness, 1991). For most fish species, first increment deposition occurs either at hatching, a day after hatching, or after first feeding and yolksac absorp- tion (Jones, 1986; Thorrold and Hare, 2002). Increment deposition before hatching also occurs (Barkmann and Beck, 1976; Radtke and Dean, 1982). If first increment deposition does not occur at hatching, the stan- dard procedure is to add a predeter- mined number to increment counts to estimate fish age (Campana and Neilson, 1985). Accuracy and precision of incre- ment counts is in part determined by the increment formation rate, which has been reviewed elsewhere (Campana and Neilson, 1985; Jones, 1986; Geffen, 1987), and by the type of otolith (asteriscus, sagitta, or la- pillus) and the preparation tech- nique used for aging. In most age and growth studies of larval and juvenile fish, the sagitta, the larg- est of the three otoliths, has been used (Campana and Neilson, 1985), but there are many examples of fish species that can be aged accurately by using the lapillus (e.g., Hoff et al., 1997; Bestgen and Bundy, 1998; Escot and Granado-Lorencio, 1998; Morioka and Machinandiarena, 2001). Although infrequently used, the asteriscus has provided age in- formation with similar or even bet- ter precision and accuracy than the sagitta and lapillus (David et al., 1994). However, the microstructure of asterisci is usually not as clear as that of sagittae or lapilli, and the extraction of asterisci is relatively time consuming and laborious (Cam- pana and Neilson. 1985; Neilson and Geen, 1985). As for otolith prepa- ration, two general techniques are common: 1) polishing of one or both sides of a sectioned otolith in trans- verse view, and 2) polishing of one side of the whole sagitta (Secor et al., 1992). Sagittae and lapilli pro- vide the same accuracy and preci- sion for age estimation; however, la- pilli may be easier to process for age determination and may not require processing at all (e.g., Ichimaru and Katsunori, 1995). Spot (Leiostomus xanthurus) is an important fishery species along the southeast coast of the United States (Mercer, 1987) and is a dominant species in coastal ecosystems owing to its abundance (Walter and Aus- tin, 2003). Studies of spot have il- luminated processes that affect the abundance of estuarine-dependent species ( Warlen and Chester, 1985; Flores-Coto and Warlen, 1993; Ross. 2003). Further, spot has been used as an experimental organism for ex- amining larval ecology (Govoni et al., 1985; Govoni and Hoss, 2001) and otolith chemistry (Bath Martin et al., 2000, 2004; Bath-Martin and Thor- rold, 2005). Although spot has been widely studied and is an important ecological and fishery species, basic information necessary for otolith analyses is not available. Our goal was to provide a founda- tion for the use of otolith increment counts in examining the ecology of larval and juvenile spot. Our specific objectives were 1) to determine the timing of first-increment formation of spot (Leiostomus xanthurus) and 2) to assess the accuracy and precision of age estimates from increment counts made with different combinations otoliths and preparation techniques. Specifically, four combinations of oto- liths (sagittae and lapilli) and prepa- ration techniques were compared: 1) a transverse section of the sagitta (polished on one side TSS-1); 2) a transverse section of the sagitta (pol- ished on two sides TSS-2); 3) a whole sagitta (polished on one side WS-1); and 4) a whole lapillus (polished on one side WL-1). Materials and methods First increment formation Six male and six female spot were induced to spawn by injection of human chorionic gonadotropin (HCG) hormone at the NOAA Beaufort Labo- ratory. Eggs were incubated in a 100-L Manuscript submitted 14 May 2004 to the Scientific Editors Office. Manuscript approved for publication 29 March 2005 by the Scientific Editor. Fish. Bull. 103:544-552 (2005). NOTE Fey et al.: Effect of type of otolith and preparation technique on age estimation of Leiostomus xanthurus 545 tank at constant temperature (20°C) and salinity (30%<->), under 12 h light:12 h dark photoperiod. These conditions were maintained throughout the rearing period. Hatch- ing occurred three days after spawning. Larvae were fed rotifers throughout the experiment and supplemented with enriched Artemia from day 20 through day 30. Larvae were collected 4 days (n = 5), 12 days (n=7), and 27 days (n = 5) after hatching, and live total length (LT) measurements were made. Larvae were then preserved in 95% ethanol. Sagittae and lapilli were dissected with fine-tipped forceps and embedded on microscope slides. The incre- ments were clearly visible and otoliths did not require any additional preparation. All increment counts were conducted three times by one person on different occa- sions with a lOOx oil objective and a Nikon E600 micro- scope with transmitted light. The light was polarized to obtain better visibility. The reader did not know the ages of the fish. Known fish age and the number of observed incre- ments were used to determine the time of first incre- ment formation on both the sagittae and lapilli. The number of increments deposited between sampling dates divided by elapsed days indicated periodicity of incre- ment formation. Accuracy and precision The experimental protocol and conditions were the same as in the previous examination of first increment for- mation, except that fish were reared for 53 days and artificial diet was added after day 30. Larvae (?? =24, 8.8-16.1 mm LT, mean=11.8 mm LT) were collected 34 days after hatching, and juveniles (rc=34, 19.4-28.1 mm LT, mean=24.3 mm LT) were collected 53 days after hatching. Sagittae and lapilli were dissected from fish with fine-tipped forceps and embedded for sectioning on the transverse plane (right sagitta) or polishing on the sag- ittal plane (left whole sagitta and lapillus). Priority was given to transverse sections, and if the right sagitta was damaged during preparation, the left sagitta was used (n = 8). Otoliths were sectioned with a slow-speed saw with dual diamond wafering blades. Sections were then ground on one side with 1000-grit sandpaper and polished with 0.3-|i0.05; Fig. 4A). For juveniles, however, there was a significant difference in the number of counted incre- ments among sagitta preparation methods (ANOVA, P<0.001) (Fig. 4B). A lower number of increments were enumerated from transverse sections of sagittae (with one side polished) (post hoc: Tukey HSD for unequal n, P<0.001). Moreover, -25% of otoliths within this group were not readable. All the otolith preparation techniques, except the PIS transverse sections of sagitta from juveniles, underesti- mated the age from hatching by 9-10 days. A 6-7 day difference was expected between known age and lapilli increment counts, owing to the time of first-increment formation. Thus, actual fish age was underestimated by approximately 2-4 days with lapilli increment counts. A 5-day difference was expected between known age and 548 Fishery Bulletin 103(3) Table 1 The distance from otolith core to first increment in the sagitta (first increment formed on the first day after hatching lapilli (first increment formed six days after hatching) of laboratory-reared spot iLeiostomus xanthurus). ) and in the Otolith n Distance to the first increment (um) Mean SD Range Sagittae — experiment on first-increment formation Sagittae' — experiment on accuracy and precision of aging technique Lapilli — experiment on first-increment formation Lapilli — experiment on accuracy and precision aof aging technique 17 36 17 25 8.3 0.76 7.8 0.91 12.3 0.54 12.2 0.61 6.7-9.9 6.7-8.8 11.5-13.2 11.0-14.2 ' Data for both whole sagittae (polished on one side) along sagittal view, and transverse sections of sagittae polished on two sides. Table 2 Number of increments deposited on the otoliths of laboratory-reared spot (Leiostomu parison with number of days between sampling days. s xanthurus ) between sampling days in corn- Otolith Sampling days (days after hatching) Days betweer sampling Number of increments between sampling2 Sagittae — experiment on first-increment formation 12 and 27 15 14.3 Sagittae' — experiment on accuracy and precision of age determination 34 and 53 19 18.3 Lapilli — experiment on first-increment formation 12 and 27 15 14.1 Lapilli — experiment on accuracy and precision of age determination 34 and 53 19 18.6 ' Data for both whole sagittae (polished on one side) ind for transverse sections of sagittae (pol shed on two sides). 2 No variance is given because the value re presents difference between two average increment numbers obta ined for two different groups offish. whole-sagittae increments counts, owing to the initia- tion of increments from a second check, which formed approximately 5 days after hatching. With whole-sag- ittae increment counts, actual fish age was underesti- mated by approximately 5 days. The coefficients of variation (CV), which indicates the precision of age estimates, varied from 1.4% to 8.3% (Fig. 5). CVs were statistically different among age estimation methods for both larvae and juveniles (Kruskal-Wallis ANOVA, P<0.001). Lapilli from both larvae and juveniles had lowest CVs, indicating high precision. Whole sagittae and P2S transverse sections for juveniles were comparable, but lower precision for larvae was observed. However, if transverse sections are used for aging, the preparation of both sides is important in the case of larvae (with regard to preci- sion; see Fig. 5) and mandatory in the case of juveniles (with regard to accuracy; see Fig. 4B). In addition, the confidence of the otolith reader in increment recognition (Fig. 5) indicated that the most clear and easy to count increments were found in the lapilli. Discussion First-increment formation In prior studies, the age of larval and juvenile spot was estimated by adding five days to the number of incre- ments counted from sagittae (e.g., Warlen and Chester, 1985; Flores-Coto and Warlen, 1993; Ross, 2003). Our research indicated that increment formation in sagit- tae occurred at hatching. The only study validating first-increment formation in spot used linear regres- sion analysis for laboratory-reared fish (Peters et al.1). The intercept of their regression line (age in relation to number of increments) indicated that the first increment Peters, D. S, Jr, J. C. DeVane, M. T. Boyd, L. C. Clements, and A. B. Powell. 1978. Preliminary observations on feed- ing, growth and energy budget of larval spot iLeiostomus xanthurus). In Ann. Rep. Southeast Fish. Cent., Beaufort Lab. to U.S. Dep. Energy, p. 377-397. Beaufort Laboratory, National Marine Fisheries Service, Beaufort, NC. NOTE Fey et al.: Effect of type of otolith and preparation technique on age estimation of Leiostomus xanthurus 549 Method (source of increment counts) Sagittae tranverse section one-side polished Sagittae tranverse section two-sides polished Sagittae whole •i-' r 30 28 26 24 % 22 E § 20 c o 54 a> § 52 * 50 48 46 44 42 40 32 (22) (22) (22) 1 2 3 B (21) (21) (21) (32) (32) (32) (13) (13) (13) 12 3 12 3 Increment counts within method Lapilli whole (19) (19) (19) (17) (17) (17) (23) (23) (23) 12 3 12 3 12 3 (27) (27) (27) Figure 4 Age of laboratory-reared spot [Leiostomus xanthurus) estimated from daily otolith growth increments counted at three different occasions for each preparation method: (A) larvae (34 d, 11.8 mm LT); and (B) juveniles (53 d, 24.3 mm LT). Mean and 95% confidence interval minimum, and maximum values are presented. Values in parentheses indicate sample number. Dashed line indicates the real age. formed five days after hatching, which corresponds to a time of exogenous feeding initiation in spot (Powell and Gordy, 1980; Powell and Chester, 1985). The other validation experiments on spot (Hettler, 1984; Siegfried and Weinstein, 1989) provided no information on first increment deposition time. In lapilli, increment depo- sition occurred six days after hatching, but no other studies are available for spot to compare and evaluate these results. The inconsistency in the time of first increment for- mation on the sagittae between the present study and Peters et al.'s study1 may be the result of underestima- tion by the latter because they did not section or pol- ish the otoliths. Spot otoliths are relatively large and thick and both sagittae and lapilli are difficult to read without otolith preparation for fish older than 25-30 days (-7-9 mm TL). Peters et al.1 found no increments in sagittae of four- to five-day-old fish. Although in the present study increments were not clear in sagit- tae of four-day-old spot, fish collected from the same tanks, 8 and 23 days later, had visible increments since hatching. Even if it is difficult to explain why the increments in sagittae of four-day-old-fish were not visible, results presented in the present study support the conclusion that first increment formation occurred at hatching. 550 Fishery Bulletin 103(3) s? 7 r o 6 10 <0 > 5 o *~ 4 •V o 3 11) o o ■A 1 £ (•) (..) (....) B I 1 (-) (*•) (—) (• TSS-1 TSS-2 WS-1 WL-1 TSS-1 TSS-2 WS-1 WL-1 Method (source of increment counts) Figure 5 Precision evaluation for different aging methods employed for larval (A) and early-juvenile (B) laboratory-reared spot: transverse section of sagitta (polished on one side) (TSS-1), transverse section of the sagitta (polished on two sides) (TSS-2), whole sagitta (polished on one side) (WS-1), and a whole lapillus (polished on one side) (WL-1). The coefficients of variation (CV) values were calculated for three independent increment counts per otolith. Additionally, the confidence of the otolith reader in increment recognition has been indicated by numbers of stars, i.e., poor (*), relatively good (**), good (***), and very good (****). Accuracy and precision of age estimates among different types of otoliths and preparation techniques Lack of distinct patterns in daily growth increments in otoliths of laboratory-reared fish (e.g., David et al., 1994) could make it difficult to conduct laboratory- based ecological experiments with larval fish. Hettler (1984) attempted to validate increment formation rate in the sagittal otoliths of laboratory-reared spot (13- 16 mm SL). Within eight days after tetracycline mark- ing, otolith radii increased approximately 18%, but no increments were observed. Siegfried and Weinstein (1989) confirmed daily increment formation in the sag- ittae of field-reared spot larvae, but those reared in the laboratory produced 17 increments instead of the expected 30. Our results, on the other hand, provided direct validation of daily increment formation in the sagittae and lapilli of laboratory-reared spot (Table 2). Even though increment formation was found to occur daily, there were inaccuracies in the estimate of age from otolith increment counts. Twenty-four increments were counted on the sagittae of 34-day-old larvae; if five increments were added for time between first-incre- ment formation and formation of the second check (the starting point of counts used in the present study), age was still underestimated by 4-5 days. Similarly, 24 increments were counted on the sagittae of 34-day-old larvae; if 6-7 days were added to account for the tim- ing of increment formation in the lapillus, age was un- derestimated by 3-4 days. Similar inaccuracies in age estimates were derived for 53-day-old juveniles. Peters et al.1 also found age inaccuracies of five days from sagittal increments and concluded that first-increment formation occurred five days after hatching. Given our results and those of Hettler (1984) and Siegfried and Weinstein (1989), we conclude that the likely explana- tion for age inaccuracies is that the increments near the core of the otolith become harder to read as more otolith material is laid down and this process results in the appearance of fewer increments. These inaccuracies would contribute to a 10-15% underestimation of age from sagittae and a 3-11% underestimation of age from lapilli. To account for these inaccuracies, five increments should be added to increment counts to estimate age. Lapilli, compared with sagittae, exhibited very clear patterns with increments (Fig. 2) and provided more precise results for the ages of larval and juvenile spot. Although there is no study presenting age data obtained from lapilli for larval or juvenile spot, lapilli have been used successfully for aging many other fish species. Ichimaru and Katsunori (1995) preferred the lapillus as a source of age data for two species of flyingfishes larvae (Cypselurus heterurus doederleini and Cypselurus hiraii) because increments were as clear as those in the sagittae, yet the lapilli did not require any preparation. Bestgen and Bundy (1998) reported increments depos- ited on sagittae of Colorado squawfish (Ptychocheilus lucius) were difficult to distinguish after fish were 30 days old and thus lapilli were used to age older fish. Lapilli were the preferred otoliths for age determination of young Lost River sucker (Deltistes luxatus) and short- nose sucker (Chasmistes brevirostris) because of their readability and conservative growth pattern (Hoff et al., 1997). Escot and Grando-Lorencio (1998) concluded NOTE Fey et al.: Effect of type of otolith and preparation technique on age estimation of Leiostomus xanthurus 551 that increments in lapilli of Barbus sclateri (Pisces: Cy- prinidae) were more clearly defined than in sagittae and asterisci. Similarly, our results demonstrate the utility of lapilli for larval and juvenile fish age estimates. In addition to the choice of the most suitable type of otolith, the choice of the most appropriate preparation method is an important aspect of larval and juvenile fish age determination (Secor et al., 1992). Analysis of PIS whole sagittae provided in the current study similar precision and confidence in age determination as transverse sections. Although analysis of sagittal transverse sections have been applied to spot ( Siegfried and Weinstein, 1989), the most frequently used method has been the analysis of whole sagittae in sagittal view (Hettler, 1984; Warlen and Chester, 1985; Powell et al., 1990; Flores-Coto and Warlen. 1993; Ross, 2003). Re- cently, Ross (2003) was able to age 40-160 day-old spot juveniles, analyzing whole sagittae along the sagittal view; however, polishing on both sides was frequently necessary. For whole lapilli, however, only one prepara- tion method (i.e., polishing along the sagittal plane) was used in the present study and the results were more satisfactory then those obtained for sagittae and hence no other preparation method (i.e., sectioning) seemed to be required. In conclusion, first-increment formation occurs at hatching in the sagittae and at 6-7 days after hatching in the lapilli. Increment formation rate occurs daily in both the sagittae and the lapilli. With sagittal and lapil- lar increment counts, age was underestimated and the cause appeared to be difficulty in discerning increments near the core. Whole lapilli (prepared by polishing one side along the sagittal section) provided age accuracy similar to that of the three sagittal preparations, but higher precision. Future studies would benefit from us- ing the lapillus for ecological studies of the early life history of spot. Acknowledgments The authors thank Elisabeth Laban for consultation during otolith preparation and analysis, as well as Dean Ahrenholz and Jennifer Potts for reviewing the earlier version of the manuscript. This research was performed while the first author held a National Research Council Research Associateship Award at the NOAA Beaufort Laboratory. Literature cited Barkmann, R. C, and A. Beck. 1976. Incubating eggs of the Atlantic silverside on nylon screen. Prog. Fish-Cult. 38:148-150. Bath Martin, G. E., and S. R. Thorrold. 2005. Temperature and salinity effects on magnesium, mananese, and barium incorporation in otoliths of larval spot {Leiostomus xanthurus). Mar. Ecol. Prog. Ser. 293:233-240. Bath Martin, G. E., S. R. Thorrold, and C. M. Jones. 2004. Temperature and salinity effects on strontium incorporation in otoliths of larval spot (Leiostomus xanthurus). Can. J. Fish. Aquat. Sci. 63:34-42. Bath Martin. G. E., S. R. Thorrold, C. M. Jones, S. E. Campana, J. W. McLaren, and J. W. H. Lam. 2000. Strontium and barium uptake in aragonitic otoliths of marine fish. Geochim. Cosmochim. Acta 64:1705-1714. Bestgen, K. R., and J. M. Bundy. 1998. Environmental factors affect daily increment deposition and otolith growth in young Colorado squawfish. Trans. Am. Fish. Soc. 127:105-117. Campana, S. E. 2001. Accuracy, precision and quality control in age deter- mination, including a review of the use and abuse of age validation method. J. Fish Biol. 59:197-242. Campana, S. E., and E. Moksness. 1991. Accuracy and precision of age and hatch date esti- mation from otolith microstructure examination. ICES J. Mar. Sci. 48:303-316. Campana, S. E., and J. D. Neilson. 1985. Microstructure of fish otoliths. Can. J. Fish. Aquat. Sci. 42:1014-1032. Chang, W. Y. B. 1982. A statistical method for evaluating the reproduc- ibility of age determination. Can. J. Fish. Aquat. Sci. 39:1208-1210. David. A. W., J. J. Isley, and C. B. Grimes. 1994. Differences between the sagitta, lapillus, and aster- iscus in estimating age and growth in juvenile red drum, Sciaenops ocellatus. Fish. Bull. 92:509-515. Escot, C, and C. Granado-Lorencio. 1998. Morphology of the otoliths of Barbus sclateri (Pisces: Cyprinidae). J. Zool. 246:89-94. Flores-Coto, C, and S. M. Warlen. 1993. Spawning time, growth, and recruitment of larval spot Leiostomus xanthurus into a North Carolina estuary. Fish. Bull. 91:8-22. Geffen, A. J. 1987. Methods of validating daily increment deposition in otoliths of larval fish. In Age and growth offish (R. C. Summerfelt and G. E. Hall, eds.), p. 223-240. Iowa State Univ. Press. Ames, IA. Govoni, J. J., A. J. Chester, D. E. Hoss, and P. B. Ortner. 1985. An observation of episodic feeding and growth of larval Leiostomus xanthurus in the northern Gulf of Mexico. J. Plank. Res. 7:137-146. Govoni, J. J., and D. E. Hoss. 2001. Comparison of the development and function of the swimbladder of Brevoortia tyrannus (Clupeidae) and Leiostomus xanthurus ( Sciaenidae t. Copeia 2001(2):430-442. Hettler, W. F 1984. Marking otoliths by immersion of marine fish larvae in tetracycline. Trans. Am. Fish. Soc. 113:370-373. Hoff, G. R., D. J. Logan, and D. F Markle. 1997. Otolith morphology and increment validation in young Lost River and shortnose suckers. Trans. Am. Fish. Soc. 126:488-494. Ichimaru, T., and T Katsunori. 1995. Otolith increment formation of flyingfishes larvae, Cypselurus heterurus doederleini and Cypselurus hiraii 552 Fishery Bulletin 103(3) under rearing conditions. Bull. Nagasaki Prefect. Inst. Fish. 21:1-6. Jones, C. M. 1986. Determining age of larval fish with the otolith increment technique. Fish. Bull. 83:289-298. Mercer, L. P. 1987. Fishery management plan for spot (Leiostomus xanthurus). ASMFC Fishery Management Report 11:1-81. Atlantic States Marine Fisheries Commis- sion, Washington, D.C. Morioka, S., and L. Maehinandiarena. 2001. Comparison of daily increment formation pattern between sagittae and lapilli of ling (Genypterus blacodes) larvae and juveniles collected off Argentina. N.Z.J. Mar. Freshw. Res. 35:111-119. Neilson, J. D., and G. H. Geen. 1985. Effects of feeding regimes and dial temperature cycles on otolith increment formation in juvenile salmon (Occorhynchus tschawytscha). Fish. Bull. 83:91-101. Powell, A. B., and A. J. Chester. 1985. Morphometric indicis of nutritional conditions and sensitivity to starvation of spot larvae. Trans. Am. Fish. Soc. 114:338-347. Powell, A. B., A. J. Chester, J. J. Govoni, and S. M. Warlen. 1990. Nutritional condition of spot larvae associated with Mississippi River Plume. Trans. Am. Fish. Soc. 119:957-965. Powell, A. B., and H. R. Gordy. 1980. Egg and larval development of the spot, Leiostomus xanthurus (Seiaemdae). Fish. Bull. 78:701-714. Radtke, R. L., and J. M. Dean. 1982. Increment formation in the otoliths of embryos, larvae, and juveniles of the mummichog, Fundulus heteroclitus. Fish. Bull. 80:201-215. Ross, S. W. 2003. The relative value of different estuarine nursery areas in North Carolina for transient juvenile marine fishes. Fish. Bull. 101:384-404. Secor, D .H., J. M. Dean, and E. H. Laban. 1991. Otolith removal and preparation for microstructural examination: a user's manual. Technical Publ. 1991- 01, 85 p. The Electronic Power Research Institute and the Bell W. Baruch Institute for Marine and Coastal Sciences, Columbia, SC. Siegfried, R. C. II, and M. P. Weinstein. 1989. Validation of daily increments deposition in the otolith of spot {Leiostomus xanthurus). Estuaries 12:180-185. Thorrold, S. R., and J. A. Hare. 2002. Otolith applications in reef fish ecology. In Coral reef fishes: dynamics and diversity in a complex eco- system (P. F. Sale, ed.), p. 243-264. Academic Press, San Deigo, CA. Walter, J. F, and H. M. Austin. 2003. Diet composition of large striped bass (Morone saxa- tilis) in Chesapeake Bay. Fish. Bull. 101:414-423. Warlen, S. M., and A. J. Chester. 1985. Age, growth, and distribution of larval spot, Leiostomus xanthurus, off North Carolina. Fish. Bull. 83:587-599. 553 Preliminary use of oxygen stable isotopes and the 1983 El Nino to assess the accuracy of aging black rockfish (Sebastes melanops) Kevin R. Piner Southwest Fisheries Science Center National Marine Fisheries Service, NOAA 8604 La Jolla Shores Drive La Jolla, California 92037 E-mail address Kevin. Pinena'noaa. gov Melissa A. Haltuch School of Aquatic and Fishery Sciences University of Washington, 1122 NE Boat Street Seattle. Washington 98105 John R Wallace Northwest Fisheries Science Center National Marine Fisheries Service, 2725 Montlake Blvd East Seattle, Washington 98112 (Campana, 2001). Recently, stable oxygen isotopes from Pacific halibut (Hippoglossus stenolepis) otoliths were used to examine regime shifts in the Northeast Pacific for the iden- tification of changes in bottom wa- ter temperatures (Gao and Beamish, 2003). In addition, otolith chemistry may be used to identify environmen- tal events that serve as natural tags for such studies (Campana and Thor- rold, 2001). We used a strong region- al environmental event, the 1983 El Nino, as a time marker to judge the accuracy of age assignment for black rockfish <15 years of age. The 1983 El Nino produced anomalously warm oceanic conditions along the coast- lines in the eastern Pacific; therefore the stable oxygen isotope ratio from 1983 should reflect this change in oceanic conditions. Materials and methods Black rockfish (Sebastes melanops) range from California to Alaska and are found in both nearshore and shal- low continental shelf waters (Love et al., 2002). Juveniles and subadults inhabit shallow water, moving deeper as they grow. Generally, adults are found at depths shallower than 55 meters and reportedly live up to 50 years. The species is currently man- aged by using information from an age-structured stock assessment model (Ralston and Dick, 2003). In many studies, ages are assumed to be accurate and there is no effort to validate the accuracy of the ages (Beamish and McFarlane, 1983). Re- cent methods of age validation rely upon environmental events that serve as time markers (Campana, 2001). Bomb radiocarbon released during nuclear bomb testing has been used to validate fish ages (Kalish et al., 1996; Campana, 1997; Kalish et al., 1997). Unfortunately, bomb radiocar- bon can be used only for fish that lived during the informative period (-1960-70); thus the technique has been used primarily on older ages. For many stock assessments, the validation of younger ages is more critical because of their importance in estimating vital rates, such as growth and maturity schedules. In this note we apply the well- studied relationship between water temperature and the ratio of oxygen stable isotopes in otoliths to assess the accuracy of young black rockfish ages. Oxygen isotope ratios serve as a record of past water temperatures because the isotope ratio is incorpo- rated into the otolith in near equi- librium with the ratio found in the environment (Patterson et al., 1993; Thorrold et al., 1997) and ambient water temperatures are inversely cor- related with 180/160 ratios (Gao et al., 2001). Calcified structures have a strong history of being used in en- vironmental reconstructions based on incorporated trace elements and iso- topes (Chivas et al., 1985; Holmden et al., 1997). Otolith microchemistry has been used to successfully recon- struct the environmental history of fish and to answer questions about natal homing (Thorrold et al., 2001) and population mixing (Campana et al.. 1999). Variation in oxygen iso- topes has been used to confirm vi- sually observed growth increments We obtained nine pre-aged black rock- fish otoliths collected during 1987-91 from recreationally caught fish off Cape Lookout, Oregon (~45.25°N, 145°W), from approximately 15-30 m water depth. One otolith was aged by Oregon Department of Fish and Wildlife scientists by using the tra- ditional break-and-burn method; the matching otolith was used in the stable isotope analysis. Fish from a range of years and ages (Table 1) were selected to include the 1983 El Nino year. A time series of annual summer bottom water temperatures from the same region and depth where the black rockfish otolith samples were obtained, were provided by the Pacific Hindcast from the Columbia Univer- sity International Research Institute for Climate Prediction, Palisades, New York. To estimate the year containing the warmest oceanic conditions, we examined otolith material from all opaque growth zones within each oto- lith for oxygen isotopes (180/160) and Manuscript submitted 13 February 2004 to the Scientific Editor's Office. Manuscript approved for publication 8 February 2005 by the Scientific Editor. Fish. Bull. 103:553-558 (2005). 554 Fishery Bulletin 103(3) Table 1 Age-specific o^O values for each black rockfish tSebastes by "N/A" indicates samples that were not reported by the melanops), along with annulus age stable isotope laboratory. and collection year. Values replaced Age Sample number 1987-33 1987-86 1987-98 1988-7 1988-100 1991-27 1991-86 1991-168 1991-178 0+ -0.96 N/A -0.633 -0.357 1.209 N/A -0.128 -2.06 0.756 1+ -1.17 0.914 -0.981 -0.513 0.725 -0.53 -0.114 -1.77 0.948 2 + -0.42 0.654 -0.644 -0.504 -0.544 -0.42 -0.358 -1.09 0.869 3 + -0.04 0.461 -0.518 -0.579 1.037 -0.39 -0.409 -0.64 0.741 4 + 0.42 1.006 -0.202 -0.320 N/A -0.09 -0.082 -0.85 0.338 5 + 0.28 0.946 -0.037 -0.257 N/A -0.22 -0.218 -0.73 N/A 6+ 0.957 0.630 0.137 N/A -0.05 0.1067 0.35 0.785 7+ 0.962 1.0 0.512 1.08 1.054 8+ N/A 0.805 1.07 1.113 9+ N/A 1.053 N/A 1.404 10 + N/A 1.164 N/A 1.510 11+ N/A 1.881 Annulus age (yr) 6 7 7 8 7 12 12 11 11 Collection year 1987 1987 1987 1988 1988 1991 1991 1991 1991 assigned to a year of formation based on estimated age and capture year. Chemical assay and otolith processing were completed at the Stable Isotope Laboratory of the University of Michigan. Each otolith was embedded in epoxy resin and cut transversally with a low-speed dia- mond-bladed saw. Three or four thin sections -150 /urn thick were removed from the center of each otolith. The thin sections were then glued with cyanoacrilate glue to petrographic glass slides. Samples from multiple thin sections were combined for a single assay. Each opaque growth zone was sampled by using a Merchantek Micro- milling system and assays were completed with a Finni- gan 251 MAT mass spectrometer. All measurements were reported in standard Vienna Pee Dee Belemnite (VPDB) and notation as 6%c (per mil), where 10 years. Confirmation of the annual banding pattern in the ololiths of other Sebastes species has been accomplished by using a variety of methods. Woodbury (1999) con- firmed the accuracy of age assignment in widow {Se- bastes entomelas) and yellowtail (S. flavidus) rockfish, using the change in growth increment width associated with El Nino. Piner et al. (in press) has used bomb ra- diocarbon to confirm the annual pattern of otolith band- ing in canary rockfish (S. pinniger) and has reported a possible underaging bias for older fish. Andrews et al. (1999) used radiometric age determination to confirm the longevity of long-lived species. However, a larger study on black rockfish with stable isotopes is necessary to conclusively determine age estimates accurately and potential underaging bias. The 1983 El Nino was chosen for the present study because it was one of the strongest recorded in the century (Sharp and McClain, 1993). Warm water condi- tions associated with the 1983 El Nino were sufficient to slow growth (MacLellan and Saunders, 1995; Wood- bury, 1999) and alter reproductive patterns (VenTresca et al., 1995) in species occupying similar geographic ranges. In contrast, this study attempted to indirectly measure the environment experienced by black rock- fish without the need to infer changes to biological processes. Nevertheless, our results appear to support the conclusions of MacLellan and Saunders (1995) and Woodbury (1999) that the anomalous oceanic conditions in 1983 are identifiable. The analysis of model residuals rather than raw iso- tope ratios is more appropriate because of the obvious d^O temporal trend in some samples. Otolith process- ing difficulties also may have contributed to the trend. The opaque region of the otolith decreases in size with increasing age. The narrowing of the otolith region as- sociated with older ages made precise sampling more dif- ficult and may have resulted in accidental sampling from otolith material outside the opaque region. The sampling of otolith material from outside the opaque region may have contributed otolith material formed in cooler waters in contrast to the sampling of areas of the otolith associ- ated with younger ages. The increasing trend in &H0 was not explained solely by the decreasing temporal trend of summer water temperatures. However, an additional com- ponent of that trend may be the result of age-dependent fish movement to cooler waters that are deeper or more 556 Fishery Bulletin 103(3) ID CQ > o 1991-27 1991-86 . 0 q ^-"0 1 ■"* 2 1987-98 1991-168 1 • 0 1 p D a 6 D D 2 1988-7 1991-178 1988-100 1.0 0.5 0.0 -0.5 -1.0 1.0 0.5 0.0 -0.5 -1.0 1.0 0.5 0.0 -0.5 -1.0 1.0 0.5 0.0 -0.5 -1.0 1978 1980 1982 1984 1986 1988 1990 0.5 00 -0.5 -1 0 1978 1980 1982 1984 1986 1988 1990 Year Figure 2 The time series of 6180 (•, and left axis) and the linear model residuals (□, and right axis) taken from each black rockfish [Sebastes melanops) otolith used in the present study. The solid line is the linear model fit to the 6180 data. A residual value of zero indicates perfect agreement between observed and predicted year-specific average residuals. Sample numbers corresponding to Table 1 are given inside each graph. northerly. Furthermore, the isotope variability between fish may be due to fish inhabiting different areas in the early periods of life or to temporal differences in growth. Finally, changes in calibration of the spectrometer be- tween assays may be a source of uncertainty. A critical assumption behind the present study was that the lowest 6180 corresponds to the warmest water temperature, and consequently the 1983 El Nino that serves as the time marker. The 6180 values may be impacted by salinity in addition to water temperature (Dorval, 2004), and we assumed that salinity was con- stant and that the changes in <5180 values were largely influenced by changes in temperature. A further con- founding element to this kind of study is the ability of fish to move and potentially select microhabitats with different temperatures than that of the average local environment. Natural date-specific markers also must be monitored over a number of years to ensure that they remain identifiable within the otolith (Campana, 2001). We addressed this concern by selecting fish of NOTE Piner et al : Use of oxygen stable isotopes and the 1983 El Nino to assess accuracy of aging Sebastes melanops 557 0.6 0.4 0.2 0.0 -0.2 -0.4 -0.6 -0.8 1978 1980 1982 1984 1986 1988 1990 1992 a | 0.6 • < 0.4 0.2 0.0 -0.2 ■ -0.4 • -0.6 -0.8 1978 1980 1982 1984 1986 1988 1990 1992 Year Figure 3 The year-specific average model residual from (A) all nine black rockfish (Sebastes melanops) and (B) ages 6-8 (•, solid line) and ages 11-12 (D, dashed line) fish. Error bars are ±1 SE. various ages and from various collection dates and by performing the same analysis on each fish. Lastly, this method of age validation can be difficult to implement for fish with small otoliths or for long-lived fish be- cause of the difficulties in obtaining sufficient otolith material from small growth increments. The detection of El Nino events using oa80 may allow the use of this reoccurring climatic event as a natural tag for age validation. Because previous studies that used El Nino events as time markers were forced to measure biological reactions to environmental chang- es, the use of dmO may be an improvement because it avoids assuming the intermediate step, namely that environment affects a biological process. The results from this study, however, were based on a small sample size and are, therefore, only preliminary. Further work in this area is warranted. Acknowledgments We are grateful to Maria Marcano of the University of Michigan geochemical laboratory for her help in setting up the assays. Bill Miller of the Oregon Department of Fish and Wildlife provided aging expertise.We thank the Sea Grant Fellowship in Population Dynamics and the National Marine Fisheries Service Northwest Fish- eries Science Center for financial support. Finally, the anonymous reviewers and Christian Reiss greatly helped our efforts. Literature cited Andrews, A. H., K.H. Coale, J. L. Nowicki, C, Lundstrom, Z. Palacz, J. E. Burton, and G. M. Cailliet. 1999. Application of an ion-exchange separation tech- nique and thermal ionization mass spectrometry to 226Ra determination in otoliths for radiometric age determi- nation of long-lived fishes. Can. J. Fish. Aquat. Sci. 56:1329-1338. Beamish, R. J., and G. A. McFarlane. 1983. The forgotten requirement for age validation in fisheries biology. Trans. Am. Fish. Soc. 112(6):735- 743. Campana, S. E. 1997. Use of radiocarbon from nuclear fallout as a dated 558 Fishery Bulletin 103(3) marker in the otoliths of haddock (Melanogrammus aeglefinus). Mar. Eeol. Prog. Ser. 150(l-3):49-56. 2001. Accuracy, precision and quality control in age deter- mination, including a review of the use and abuse of age validation methods. J. Fish. Biol. 59(2):197-242. Campana, S. E., G. A. Chouinard, J. M. Hanson, and A. Frechet. 1999. Mixing and migration of overwintering Atlan- tic cod (Gadus morhua) stocks near the mouth of the Gulf of St. Lawrence. Can. J. Fish. Aquat. Sci. 56(10):1873-1881. Campana, S. E., and S. R. Thorrold. 2001. Otoliths, increments, and elements: keys to a com- prehensive understanding offish populations? Can. J. Fish. Aquat. Sci. 58(l):30-38. Chivas, A. R., P. DeDeckker, and J. M. G. Shelley. 1985. Strontium content of ostracods indicates lacustrine paleosalinity. Nature 316(60251:251-253. Dorval, E. 2004. Relating water and otolith chemistry in Chesapeake Bay, and their potential to identify essential seagrass habitats for juveniles of an estuarine-dependent fish, spotted seatrout (Cynoscion nebulosus). Ph.D. diss., 134 p. Old Dominion Univ., Norfolk , VA. Gao, Y. W., and R. J. Beamish. 2003. Stable isotope variations in otoliths of Pacific hali- but iHippoglossus stenolepis) and indications of the pos- sible 1990 regime shift. Fish. Res. 60(2-):393-404. Gao, Y., H. P. Schwarcz, U. Brand, and E. Moksness. 2001. Seasonal stable isotope records of otoliths from ocean-pen reared and wild cod, Gadus morhua. Environ. Biol. Fish. 61(4):445-453. Holmden, C, R. A. Creaser, and K. Muehlebachs. 1997. Paleosalinities in ancient brackish water systems determined by Sr-87/Sr-86 ratios in carbonate fossils: a case study from the Western Canada Sedimentary Basin. Geoch. Cosmochim. Acta. 61(10):2105-2118. Kalish, J. M., J. M. Johnston, J. S. Gunn, and N. P. Clear. 1996. Use of the bomb radiocarbon chronometer to determine age of southern bluefin tuna {Thunnus maceoyii). Mar. Ecol. Prog. Ser. 143(1-31:1-8. Kalish, J. M., J. M. Johnston, D. C. Smith, A. K. Morison, and S. G. Robertson. 1997. Use of the bomb radiocarbon chronometer for age validation in the blue grenadier iMacruronus novaezelandiae). Mar. Biol. 128(4):557-563. Love, M. S., M., Yoklavich, and L. Thorsteinson. 2002. The rockfishes of the Northeast Pacific, 404 p. Univ. California Press, Los Angeles, CA. MacLellan, S. E., and M. W. Saunders. 1995. A natural tag on the otoliths of Pacific hake {Mer- luccius productus) with implications for age validation and migration. In Recent developments in fish otolith research (D. H. Secor, J. M. Dean, and S. E. Campana, eds.), p. 567-580. Univ. South Carolina Press, Colum- bia, SC. Patterson, W. P., G. R. Smith, and K.C. Lohmann. 1993. Continental paleothermometry and seasonality using isotopic composition of aragonitic otoliths in fresh- water fishes. Geophys. Monogr. 78:191-202. Piner, K. R, O. S. Hamel, J. L. Menkel, J. R. Wallace, and C. E. Hutchinson. In press. Age validation of canary rockfish (Sebastes pin- niger) from off the Oregon coast (USA) using the bomb radiocarbon method. Can. J. Fish. Aquat. Sci. Ralston, S., and E. J. Dick. 2003. The status of black rockfish (Sebastes melanops) off Oregon and northern California in 2003. In Status of the Pacific coast groundfish fishery through 2003 and stock assessment and rishery evaluation, 75 p. Pacific Fishery Management Council, Portland, OR. Sharp, G. D., and D. R. McClain. 1993. Fisheries, El Nino-Southern oscillation and upper ocean temperature records: an eastern Pacific example. Oceanography 6:13-22. Thorrold, S. R„ S. E. Campana, C. M. Jones, and P. K. Swart. 1997. Factors determining delta super! 13 IC and delta super! 18 )0 fractionation in aragonitic otoliths of marine fish. Geochim. Cosmochim. Acta 61(141:2909-2919. Thorrold, S. R., C. Latkoczy, P. K. Swart, and C. M. Jones. 2001. Natal homing in a marine fish metapopulation. Sci- ence 291(55021:297-299. VenTresca, D. A., R. H. Parrish, J. L. Houk, M. L. Gingas, S. D. Short, and N. L. Crane. 1995. El Nino effects on the somatic and reproductive condition of blue rockfish, Sebastes mystinus. Calif. Coop. Oceanic Fish. Invest. Report 36:167-174. Woodbury, D. 1999. Reduction of growth in otoliths of widow and yel- lowtail rockfish (Sebastes entomelas and S. flavidus) during the 1983 El Nino. Fish. Bull. 97:680-89. U.S. Department of Commerce Volume 103 Number 4 October 2005 Fishery Bulletin U.S. Department of Commerce Carlos M. Gutierrez Secretary National Oceanic and Atmospheric Administration Vice Admiral Conrad C. Lautenbacher Jr., USN (ret.) Under Secretary for Oceans and Atmosphere National Marine Fisheries Service William T. Hogarth Assistant Administrator for Fisheries ^ates o* The Fishery Bulletin (ISSN 0090-0656) is published quarterly by the Scientific Publications Office, National Marine Fish- eries Service, NOAA, 7600 Sand Point Way NE. BIN C15700, Seattle, WA98115-0070. Periodicals postage is paid at Seattle. WA, and at additional mailing offices. POST- MASTER: Send address changes for sub- scriptions to Fishery Bulletin, Superin- tendent of Documents, Attn.: Chief, Mail List Branch, Mail Stop SSOM, Washing- ton, DC 20402-9373. Although the contents of this publica- tion have not been copyrighted and may be reprinted entirely, reference to source is appreciated. The Secretary of Commerce has deter- mined that the publication of this peri- odical is necessary according to law for the transaction of public business of this Department. Use of funds for printing of this periodical has been approved by the Director of the Office of Management and Budget. For sale by the Superintendent of Documents, U.S. Government Printing Office, Washington, DC 20402. Subscrip- tion price per year: $55.00 domestic and $68.75 foreign. Cost per single issue: $28.00 domestic and $35.00 foreign. See back for order form. Scientific Editor Adam Moles, Ph.D. National Marine Fisheries Service, NOAA 8604 La Jolla Shores Drive La Jolla, California 92037 Managing Editor Sharyn Matriotti National Marine Fisheries Service Scientific Publications Office 7600 Sand Point Way NE, BIN C15700 Seattle, Washington 98115-0070 Editorial Committee Harlyn O. Halvorson, Ph.D. Ronald W. Hardy, Ph.D. Richard D. Methot, Ph.D. Theodore W. Pietsch, Ph.D. Joseph E. Powers, Ph.D. Harald Rosenthal, Ph.D. Fredric M. Serchuk, Ph.D. George Watters, Ph.D. University of Massachusetts, Boston University of Idaho, Hagerman National Marine Fisheries Service University of Washington, Seattle National Marine Fisheries Service Universitat Kiel, Germany National Marine Fisheries Service National Marine Fisheries Service Fishery Bulletin web site: www.fishbull.noaa.gov The Fishery Bulletin carries original research reports and technical notes on investigations in fishery science, engineering, and economics. It began as the Bulletin of the United States Fish Commission in 1881: it became the Bulletin of the Bureau of Fisheries in 1904 and the Fishery Bulletin of the Fish and Wildlife Service in 1941. Separates were issued as documents through volume 46; the last document was No. 1103. Beginning with volume 47 in 1931 and continuing through volume 62 in 1963, each separate appeared as a numbered bulletin. A new system began in 1963 with volume 63 in which papers are bound together in a single issue of the bulletin. Beginning with volume 70, number 1, January 1972, the Fishery Bulletin became a periodical, issued quarterly. In this form, it is available by subscription from the Superintendent of Documents, U.S. Government Printing Office, Washington, DC 20402. It is also available free in limited numbers to libraries, research institutions, State and Federal agencies, and in exchange for other scientific publications. U.S. Department of Commerce Seattle, Washington Volume 103 Number 4 October 2005 Fishery Bulletin Contents Articles N0>/ 0 2 2005 561-573 Kingsford, Michael J., and Julian M. Hughes Patterns of growth, mortality, and size of the tropical damselfish Acanthochromis polyacanthus across the continental shelf of the Great Barrier Reef 574-587 Kotwicki, Stan, Troy W. Buckley, Taina Honkalehto, and Gary Walters Variation in the distribution of walleye pollock (Theragra chalcogramma) with temperature and implications for seasonal migration 588-600 Luthy, Stacy A., Robert K. Cowen, Joseph E. Serafy, and Jan R. McDowell Toward identification of larval sailfish Ustiophorus platypterus), white marlin (Tetrapturus albidus), and blue marlin (Makaira nigricans) in the western North Atlantic Ocean The conclusions and opinions expressed in Fishery Bulletin are solely those of the authors and do not represent the official position of the National Marine Fisher- ies Service iNOAA) or any other agency or institution. The National Marine Fisheries Service (NMFSl 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. 601-619 McDonough, Christopher J., William A. Roumillat, and Charles A. Wenner Sexual differentiation and gonad development in striped mullet (Mugil cephalus L.) from South Carolina estuaries 620-634 Megalofonou, Persefoni, Constantinos Yannopoulos, Dimitrios Damalas, Gregorio De Metrio, Michele Deflorio, Jose M. de la Serna, and David Macias Incidental catch and estimated discards of pelagic sharks from the swordfish and tuna fisheries in the Mediterranean Sea 635-647 Narimatsu, Yoji, Daiji Kitagawa, Tsutomu Hattori, and Hirobumi Onodera Reproductive biology of female Rikuzen sole (.Dexistes nkuzenius) Fishery Bulletin 103(4) 648-658 Porter, Steven M. Temporal and spatial distribution and abundance of flathead sole (Hippoglossoides elassodon) eggs and larvae in the western Gulf of Alaska 659-669 Prince, Eric D., Robert K. Cowen, Eric S. Orbesen, Stacy A. Luthy, Joel K. Llopiz, David E. Richardson, and Joseph E. Serafy Movements and spawning of white marlin (Tetrapturus albidus) and blue marlin (Makaira nigricans) off Punta Cana, Dominican Republic 670-684 Stanley, Richard D., and Allen R. Kronlund Life history characteristics for silvergray rockfish (Sebastes brevispmis) in British Columbia waters and the implications for stock assessment and management 685-696 Weise, Michael J., and James T. Harvey Impact of the California sea lion (Zalophus californianus) on salmon fisheries in Monterey Bay, California 697-711 Welsford, Dirk C, and Jeremy M. Lyle Estimates of growth and comparisons of growth rates determined from length- and age-based models for populations of purple wrasse (Notolabrus fucicola) Notes 712-719 Bishop, Melanie J., Charles H. Peterson, Henry C. Summerson, and David Gaskill Effects of harvesting methods on sustainability of a bay scallop fishery: dredging uproots seagrass and displaces recruits 720-724 Diaz, Guillermo A., and Joseph E. Serafy Longline-caught blue shark (Pnonace glauca). factors affecting the numbers available for live release 725-727 Fey, Dariusz P., and Jonathan A. Hare Length correction for larval and early-|uvenile Atlantic menhaden (Brevoortia tyrannus) after preservation in alcohol 728-736 Hare, Jonathan A., and John J. Govoni Comparison of average larval fish vertical distributions among species exhibiting different transport pathways on the southeast United States continental shelf 737 Acknowledgment of reviewers 738 List of titles 741 List of authors 743 List of subjects 747 Subscription form 561 Abstract — Age-based analyses were used to demonstrate consistent dif- ferences in growth between popula- tions of Acanthochromis polyacanthus (Pomacentridae) collected at three dis- tance strata across the continental shelf (inner, mid-, and outer shelf) of the central Great Barrier Reef (three reefs per distance stratum). Fish had significantly greater max- imum lengths with increasing dis- tance from shore, but fish from all distances reached approximately the same maximum age, indicating that growth is more rapid for fish found on outer-shelf reefs. Only one fish col- lected from inner-shelf reefs reached >100 mm SL, whereas 38-67% offish collected from the outer shelf were >100 mm SL. The largest age class of adult-size fish collected from inner- and mid-shelf locations comprised 3-4 year-olds, but shifted to 2-year- olds on outer-shelf reefs. Mortality schedules iZ and S) were similar irre- spective of shelf position (inner shelf: 0.51 and 60.0%; mid-shelf: 0.48 and 61.8%; outer shelf: 0.43 and 65.1%, respectively). Age validation of captive fish indicated that growth increments are deposited annually, between the end of winter and early spring. The observed cross-shelf patterns in adult sizes and growth were unlikely to be a result of genetic differences between sample populations because all fish collected showed the same color pattern. It is likely that cross-shelf variation in quality and quantity of food, as well as in turbidity, are fac- tors that contribute to the observed patterns of growth. Similar patterns of cross-shelf mortality indicate that predation rates varied little across the shelf. Our study cautions against pooling demographic parameters on broad spatial scales without consid- eration of the potential for cross-shelf variability. Patterns of growth, mortality, and size of the tropical damselfish Acanthochromis polyacanthus across the continental shelf of the Great Barrier Reef Michael J. Kingsford Julian M. Hughes Reef and Ocean Ecology Laboratory School of Marine Biology and Aquaculture James Cook University James Cook Drive Townsville, Queensland, Australia 4811 E-mail address (for M J Kingstord) Michael Kingsforda<|cii edu au Manuscript submitted 10 June 2004 to the Scientific Editor's Office. Manuscript approved 30 March 2005 by the Scientific Editor. Fish. Bull. 103:561-573 120051. Coral reefs are spatially diverse and heterogeneous marine environments. The Great Barrier Reef (GBR) is the largest reef system and represents a near-continuous matrix of over 2400 individual reefs spanning a distance of some 2000 km along the coast of Queensland, eastern Australia (Fig. 1). Coral reef habitats are subject to the influences of environmental (e.g., exposure and proximity to coastlines), as well as biotic processes (e.g., avail- ability of food). Strong cross-shelf abi- otic and biotic gradients (Wilkinson and Cheshire, 1988) have the potential to influence patterns of abundance and demographic characteristics of fishes associated with coral reefs. Sev- eral studies have examined the broad- scale abundance and distribution of a wide variety of organisms across the continental shelf of the GBR, includ- ing hard corals (Done, 1982), soft corals (Dinesen, 1983), crustaceans (Preston and Doherty, 1990, 1994), algae (McCook et al., 1997), and reef fishes (Williams, 1982, 1983; Wil- liams and Hatcher, 1983; Russ, 1984a, 1984b; Newman and Williams, 1996; Newman et al., 1997; Gust et al., 2001, 2002). Great cross-shelf differences in abundance are common within and among taxa. Although environmental gradients have often been implicated as causing these patterns and it is also known that environmental features influence demographic characteris- tics (e.g., growth), there have been few comparisons of demographic charac- ters by geography and spatial scale. Demographic measures are cru- cial to understanding population dynamics. Population demographics of a number of many fish species have been shown to vary at spatial scales ranging from 100's of m to 100's of km (Gillanders, 1995; Meekan et al., 2001; Gust et al., 2002). With the ex- ception of data on a few commercially important taxa (Munro and Williams, 1985; Williams et al., 2003) and some others (e.g., acanthurids and scarids; Choat and Axe, 1996), there are few data on demographic parameters of coral reef fishes and even less on spatial variation within these para- meters. Variation in demographics may be common across the shelf. For example, significant differences in the size frequency, growth, mor- tality, and longevity in populations of three scarids (Scarus frenatus, S. niger, and Chlorurus sordidus) and an acanthurid (Acanthurus lineatus) have been shown between mid- and outer-shelf locations on the northern GBR (Gust et al., 2001, 2002). Dud- geon et al. (2000) found evidence that high levels of genetic exchange oc- curred between populations of these fishes on mid- and outer-shelf reefs and concluded that observed differ- 562 Fishery Bulletin 103(4) Distance strata and i Reef near Townsville polyacanthus. Table 1 eefs sampled during September and October 2001 over the continental shelf of the central Great Barrier Australia, for analyses of growth patterns, mortality, and size of the tropical damselfish Aeanthochromis Distance strata Reef sampled Date(s) sampled Av ?rage distance (km) to coast of the three sites ±SE Inner shelf Orpheus Island Pandora Reef Havannah Island 4 and 5 Sep 2001 3 Sept 2001 3 and 4 Sep 2001 15.3 ±0.6 16.4 ±0.3 25.1 ±0.3 Mid-shelf Bramble Reef Britomart Reef The Slashers 15 Oct 2001 16 Oct 2001 20 Oct 2001 41.1 ±0.5 38.7 ±3.0 85.3 ±2.4 Outer shelf Pith Reef Barnett Patches Myrmidon Reef 18 Oct 2001 17 Oct 2001 19 Oct 2001 74.4 ±0.9 62.6 ±1.8 110.4 ±0.8 ences in the demographic and life history features rep- resented phenotypic plasticity. Aeanthochromis polyacanthus (Bleeker) is one of a few species of fish that are found in abundance at all distances across the Great Barrier Reef (Williams. 1982, 1983) and, therefore, was ideal for comparisons of cross-shelf patterns of demographic characteristics. Aeanthochromis polyacanthus is a polymorphic gono- choristic pomacentrid and site-attached planktivore that inhabits reefs of the Indo-Australian Archipelago and adjacent regions (Allen, 1975). It is extremely wide spread and abundant along (north-south) the GBR (Wil- liams, 1982, 1983). It is unusual among marine reef fishes and unique among damselfishes in that it lacks a dispersive planktonic larval stage (Robertson, 1973). Instead, adult A. polyacanthus lay demersal eggs and after hatching, both parents defend a brood of larvae and juveniles for several months (Robertson, 1973; Al- len. 1975; Thresher, 1985a, 1995b; Kavanagh, 2000). In contrast to other taxa, therefore, dispersal is likely to be slow within and among reefs. Aeanthochromis polyacanthus is one of the best studied coral reef fishes on the GBR with respect to predation (Connell, 1996, 1998, 2000), genetics and evolution (Doherty et al., 1994, 1995; Planes and Doherty, 1997a, 1997b), be- havior (Robertson, 1973; Allen, 1975; Thresher, 1985a. 1995b; Nakazono, 1993; Kavanagh, 1998), reproductive success (Thresher, 1983). and early life history (Ka- vanagh, 2000), but no data exist on age, growth, and demographic parameters, such as mortality rates (but see estimates of juvenile mortality while larvae and juveniles are brooded by adults; Connell, 1996). The objective of this study was to compare the demo- graphic characteristics of A. polyacanthus across the continental shelf. Our approach was to sample replicate reefs in the central region of the GBR at multiple dis- tance strata from shore (inner-, mid- and outer-shelf distances). In addition, we chose a section of the GBR where A. polyacanthus exhibited the same color pattern (brown anterior and white posterior) and are known to be genetically isolated (Planes and Doherty, 1997b). Any variation in demographic parameters, therefore, could be largely attributed to phenotypic plasticity. The specific objectives of this study were the following: 1) to validate the deposition of annual growth increments for fish of a wide range of sizes and ages by using tetracy- cline, 2) to describe patterns of growth of populations of A. polyacanthus within and among distance strata; 3) to describe the age and size structures of populations of A. polyacanthus within and among distance strata, and; 4) to calculate the instantaneous mortality and survival rates (Z) of populations of A. polyacanthus within and among distance strata. Materials and methods Study sites and sampling design Spatial variation in demographics and structures of cross-shelf populations of A. polyacanthus was deter- mined by using a partially hierarchical sampling design. Individuals of a wide range of sizes were collected from three replicate reefs within each of three distance strata (inner-, mid- and outer-shelf) spanning the width of the continental shelf of the central Great Barrier Reef near Townsville, Australia (Fig. 1, Table 1). At least 16 fish were collected with hand spears from each of three sites on each reef during September and October 2001. All fish collected were the same brown and white morph (Allen, 1975). Sample processing All fish were measured (standard length [SL] to the nearest mm) and weighed (to the nearest 0.01 g). Sag- ittal otoliths were extracted, cleaned in freshwater to remove the sagittal membrane, and allowed to dry Kmgsford and Hughes Growth, mortality, and size of Acanthochromis polyacanthus 563 Figure 1 Map of the nine reefs on the central Great Barrier Reef where Acanthochromis poly- acanthus were collected. Distance strata from the mainland (i.e. inner-, mid- and outer-shelf distances I are also indicated. overnight. One otolith from each fish was then imbed- ded in Struers Epofix resin that was allowed to harden overnight in a drying oven at 60°C. A thin (250-300 jmi) transverse section perpendicular to the long axis of the otolith was then taken through the core (primordium) of the otolith with a Buehler low-speed saw with two spaced diamond blades. This section was polished by hand with 9-,«m lapping film to remove saw blade marks, thereby making the internal structure of the otolith more clearly visible. The polished section was then fixed to a labelled glass microscope slide with Crystal bond thermoplastic glue. Analysis of growth increments The opaque zones visible in the internal structure of the otolith were counted along a radius from the pri- mordium to the outer edge of the largest sagittal lobe of the otolith with a compound microscope (Leica DMLB) and white incident light source. Alternating translucent and opaque increments were interpreted as annuli. Sec- tions were coded and examined in random order and the opaque increments counted on two occasions by the same observer (JMH) separated by four weeks. Counts of annuli were compared between these two occasions in order to assess the confidence that could be placed in the interpretation of otolith structure. If increment counts differed by more than two between counting occa- sions, then the otoliths were re-examined. If, following a third reading, agreement between the third and one of the two other counts was not reached (all matching counts were used in analyses), then the otolith was not included in the analysis; 4.6% of otoliths were rejected on this basis (n=715 fish). Validation of growth increments The periodicity of growth increment formation was vali- dated by marking a group offish (of various sizes) reared in captivity with the antibiotic tetracycline hydrochlo- ride (Sigma-Aldrich, Ballerup, Denmark). Small (known to be 0+ fish) and large fish were chosen to determine if annuli are formed early and late in life. Fish were held at the MARFU Aquarium Facility, James Cook University. For the duration of the experiment, the fish were held in several 70-500 L aquaria at this facility. 564 Fishery Bulletin 103(4) Adult fish were injected in the coelomic cavity with 0.05 g/mL tetracycline in sterile saline solution at concentra- tions equivalent to 0.05 g/kg body weight (McFarlane and Beamish, 1987). The approximate weight of each individual was estimated from the relationship between weight and SL. Juveniles were mass marked by immer- sion in a tetracycline solution (concentration: 0.5g/L) in seawater for 12 hours (overnight). The tetracycline generally forms a very effective time-marker in oto- liths; it fluoresces when viewed under ultraviolet light (Geffen, 1992). The experiment commenced in May 2002 and fish were sacrificed after six months, one year (June 2003), and one-and-a-half years (November 2003). Ten fish had readable otoliths for which validation was attempted. Otolith sections were viewed with a compound micro- scope and incident ultraviolet light in a darkened room. When a fluorescing tetracycline band was identified, its position in relation to the edge was measured. The section was then examined under reflected white light and measurements of increment widths and marginal increments were recorded. Known time at liberty, ex- pressed as a proportion of one year, was then compared with estimated time at liberty by using the growth of the otoliths. If estimated time at liberty equalled ac- tual time at liberty, it supported the hypothesis that opaque increments were deposited annually. Juveniles and adults were collected on each occasion to determine whether increments were deposited annually, early and late in life. The length of time for increment formation was also estimated by calculating the number of days after tetra- cycline treatment. The number of days after treatment was estimated by comparing the position of the tetra- cycline mark with that of the last (marginal) opaque increment and the width of a full annual increment with the following formula: L,=L„[l-e-A'"-'»'], Number of days after treatment TE-MI IW x365. where TE = otolith growth after treatment; MI = the marginal increment; and IW = the final full increment width.1 where La = the asymptote of the growth curve (average maximum length); L, = length at age t\ K = the rate at which the growth curve approaches the asymptote (Lj\ t = age of fish in years; t0 = the theoretical origin of the growth curve (i.e., the hypothetical age of the fish when it has no length); and e = the base of the natural logarithm. Differences in growth curves for A polyacanthus from each reef sampled were visualized by using the tech- nique of Kimura (1980), where 95% confidence ellipses were generated around the parameter estimates of K and Lx. Confidence ellipses that did not overlap indicated dif- ferences in growth parameters and enabled the pooling of data from sites within reefs at each distance stratum. The parameter t0 was constrained to minus 0.05 to take into account the approximate size of A. polyacanthus at hatching (5 mm: Kavanagh, 1998, 2000). Mortality The instantaneous rate of mortality (Z) was calculated by using log-linear regression analyses of age-frequency data sets for A. polyacanthus populations from each reef (Pauly, 1984). With this method, recruitment was assumed to be consistent over time at each reef. The natural logarithm of the number of fish sampled from each age class was compared with their corresponding age. Year classes to the left of the age-frequency mode were excluded from the analysis because our sampling technique was biased against small A. polyacanthus. Fish greater than 60 mm were collected. The slope of the regression line between year classes estimated the instantaneous mortality rate (Z): Z = F +M, where F = fishing mortality; and M = natural mortality (Gust et al. 2002). Growth It was hypothesized that patterns of growth would vary with distance from the coast. Growth rates were described by using von Bertalanffy growth functions that provided the best fit to size-at-age data when com- pared with estimates of the Schnute growth function (Schnute, 1981). The von Bertalanffy expression for length at age t (Lt), as a function of time is 1 We assumed similar IWs for fish older than 3 years. For fish 3 years or younger the IW was calculated as an average from all experimental fish. Because there is no fishery for A. polyacanthus on the GBR, F equals zero and therefore Z estimates natural mortality only. Annual survival rate estimates were then calculated according to the equation S = e~z (Ricker, 1975). Comparisons of the slopes of age-frequency rela- tionships (for estimates of Z) were made by using analy- sis of covariance (ANCOVA) according to the procedures of Zar (1999). Data from each site were pooled for each reef because in many cases sample sizes were too small to provide reliable estimates of mortality at the site level. Similarities in mortality rates among replicate reefs within distance strata allowed us to pool data at the strata level so that comparisons of mortality between shelf positions could be made. Kingsford and Hughes Growth, mortality, and size of Acanthochromis polyacanthus 565 Figure 2 Photographs of sectioned Acanthochromis polyacanthus (age = 5 years) otolith showing: (upper) alternating opaque (annuli) and translucent band pattern and (lower) the fluorescent tetracycline mark. Note the single opaque band following the tetracycline mark (time at liberty=380 days). OTC = oxytetracline. Results Age validation All fish treated with tetracycline had clear fluorescent marks in their otoliths (Fig. 2). The positions of the fluorescent tetracycline bands in relation to the otolith margin were consistent with the deposition of opaque zones on an annual basis (Table 2). In general, percent agreement was over 75% (7/10 fish). Differences between actual and estimated time at liberty were probably related to slight variation in the small measurements that were made (i.e., fractions of a mm). The timing of deposition of the opaque increment was estimated to occur in spring because new increments were found at the edge of otoliths offish that had been marked in May and sacrificed about 200 days later. Size and age structures There were large differences in the size-frequency dis- tributions of fish sampled across the shelf (Fig. 3). At 566 Fishery Bulletin 103(4) Inner shelf Orpheus (n=41) 10 10- Jiki I Hi IIMl I 0 Pandora (n=45) ^L_ 40 60 80 100 120 40 60 80 100 120 40 60 80 100 120 Mid-shelf 10 o Bramble (n=105) 15r : l : I Britomart (n=89) 40 60 80 100 120 40 60 80 100 120 40 60 80 100 120 Outer shelf 10 is Pith (n=100) hi imB 10 - 40 60 80 100 120 0 Barnett Patches (n=117) , .nill 40 60 80 100 120 Standard length (mm) Figure 3 Size-frequency distributions for Acanthochromis polyacanthus collected from three reefs at each distance stratum from shore. Data were pooled for the three sites sampled at each reef. inner-shelf reefs (77 =155), only one fish >100 mm was collected. In contrast, between 38% and 54% offish col- lected from outer-shelf reefs were >100 mm. A mix of inner- and outer-shelf size-frequency distributions was evident for mid-shelf reefs. Bramble and Britomart reefs had 1% and 7% offish >100 mm, respectively, whereas The Slashers had the highest proportion of fish >100 mm collected of any reef (67%) including the largest individual fish collected (120 mm); however, this result was more characteristic of outer-shelf reefs. Another conspicuous feature of the cross-shelf size frequencies was the very narrow size range of adult fish collected on inner-shelf reefs in comparison to the size range of fish collected from mid- and outer-shelf locations (Fig. 3). Size selectivity due to the collection technique (hand spear) restricted the numbers offish <60 mm that could be collected. Maximum age of A. polyacanthus was similar at all reefs sampled (Fig. 4; inner shelf: 9-10 yr, mid-shelf: 9-10 yr, outer shelf: 10-11 yr). The largest age class of fish on the inner-and mid-shelf reefs comprised 3-4 year olds, whereas on the outer-shelf reefs, 2-year-old fish made up the largest proportion of the populations. The two oldest fish were both collected from outer-shelf reefs (Myrmidon and Barnett Patches) and were both 11 years old. Strong age-structured cohorts offish were found at some reefs within the same distance stratum and these cohorts were found only at these reefs and distance stratum. For example, there were strong year classes at Pith and Barnett Patches in years 5 and 6 that were not found at Myrmidon (Fig. 4). Growth Variation in patterns of growth was greater among dis- tance strata across the shelf than among reefs within a distance strata (Fig. 5). There was variation in growth between individuals from reefs within each shelf posi- tion and this resulted in variable size-at-age relation- ships (Fig. 5). From inner-shelf reefs, fish from Pandora showed small asymptotic sizes and thus had lower aver- age Lx, (Lx=77.4 mm) compared to fish from Orpheus and Havannah (L, =87.0, 84.2 mm, respectively; Table 3). Distinct, non-overlapping ellipses formed in 95% confi- dence interval plots of Lx in relation to K confirmed that growth curves for fish from Pandora differed from those at Orpheus and Havannah (Fig. 5). Fish collected from mid-shelf reefs (Bramble, Britomart, and The Slashers) showed differences in growth among all reefs (non-over- lapping 95% confidence ellipses; Fig. 5). Growth offish Kingsford and Hughes: Growth, mortality, and size of Acanthochronvs polyacanthus 567 Inner shelf Orpheus (n=35) Jin!.. 30 0 4 8 12 Mid-shelf Bramble (n=100) £ 20 cr .2 10 0 1 0 4 8 Outer shelf Pith (n=99) 20 Pandora (n=43) 30 lllll.l-. Havannah (n=67) .11 ll^ 12 12 The Slashers (n=91) 30 r 30 Barnett Patches (n=113) 20- iii 10 ll- - Myrmidon (n=81) lllll- 048 12 048 12 048 12 Age (years) Figure 4 Age-frequency distributions for Acanthochromis polyacanthus collected from three reefs at each distance stratum from shore. Data are pooled from the three sites sampled at each reef. All age estimates were derived from counts of otolith annuli. Table 2 Validation data with the use of tetracycline to deter mine the per odicity and timing of opaque ring deposition for Acantho- chromis polyacanthus with the use of tetracycline as a time marker. TAL = time at liberty expi- sssed as a proportion of one year and derived from growth measurements from reared fi sh treated with tetracycline re = = tetrac ^cline. Fish age TC to marginal TAL TAL as propoi •tion Estimated days Actual day? from Percent agreement = lyrl increment (mm) (mm) of year from TC marking TC mark ing ( estimated/actual x 100 ) 1 0.0423 0.30 0.43 110 158 69 1 0.0463 0.33 0.43 120 158 76 1 0.1784 0.94 0.97 344 355 97 5 0.0686 0.80 1.04 291 380 76 5 0.0739 0.83 1.04 304 380 80 5 0.1077 1.00 1.52 365 556 66 5 0.0805 0.81 1.52 295 556 53 6 0.1471 1.46 1.47 532 537 99 7 0.1034 0.90 1.12 327 409 80 7 0.0919 1.30 1.52 474 556 85 from the outer reefs (Pith, Barnett Patches, and Myr- midon), however, was similar for fish from each of these reefs (overlapping 95% confidence ellipses; Fig. 5). Average maximum length (L.,) varied across the shelf and differences among strata were generally greater than within-distance strata. The K values for all three 568 Fishery Bulletin 103(4) Inner-shelf reefs (n=147) ■ Orpheus Pandora " Havannah Mid-shelf reefs (n=273) • Bramble * Britomart * The Slashers x 9 * * Oufer-shelf reefs (n=296) ° Pith * Barnett Patches 10 Myrmidon 4 6 Age (years) 10 12 05 075 1 125 15 175 K 115 110 105 100 95 90 65 12 Figure 5 Von Bertalanffy growth curves for Acanthochromis polyacanthus collected from three reefs within each distance stratum. 95% confidence ellipses are given for the parameters K (growth coefficient) and Lx (mean asymptotic length). shelf positions were similar and indicated that K val- ues for A. polyacanthus converge at asymptotic sizes at approximately the same rate of growth, irrespective of proximity to the coast (Fig. 5 and Table 3). However, an obvious trend for increased L , occurred with increasing distance from the coast (inner shelf: -83 mm, mid-shelf: -99 mm, outer shelf: -102 mm). The growth parameters offish from The Slashers were more similar to those of fish taken from the outer-shelf reefs than to those we defined a priori as mid-shelf (Fig.6). The Slashers are in fact much farther from the coast (85 km), as are Pith Reef (74 km) and Barnett Patches (63 km) on the outer shelf, than the other two mid-shelf reefs (Britomart: 39 km. Bramble: 41 km) (Fig. 1, Table 1). Mortality Mortality rates for A. polyacanthus did not differ sig- nificantly between replicate reefs within inner-shelf (test for slopes df(2 19„ F=0.982, P=0.39), mid-shelf (test for slopes df(2 19l, F=1.334, P=0.29) or outer-shelf (test for slopes df(219), F=0.658, P=0.53) locations (Table 4). Kingsford and Hughes: Growth, mortality, and size of Acanthochromis polyacanthus 569 Age frequencies, therefore, were pooled at the shelf level (within distance strata; Fig 7). Acanthochromis polyacanthus mortality rates did not differ significantly between the inner-, mid- and outer-shelf strata (test for slopes df,s 63), F=0.367, P=0.70) (Fig. 6). Although mortality estimates were progressively lower with increased distance from the coast, this trend was not significant (inner shelf: -0.51, mid- shelf: -0.48. outer shelf: -0.43; Fig. 6, Ta- ble 4). Associated survival rate estimates (S) varied between reefs by -9% per an- num at inner- and mid-shelf strata and by -6% per annum on the outer shelf (Table 3). The mean difference in survival rates for A. polyacanthus between the inner and mid-shelf was ~29c and between the mid- and outer shelf was -3% (Table 4). Discussion ■ Orpheus o Pandora A Havannah • Bramble 0 Bntomart * The Slashers ° Pith a Barnett Patches x Myrmidon 110 105-1 100 95 ■ 90- 85 80 -I 75 The Slashers Myrmidon Barnett Patches Havannah Pandora 0.5 0.75 1 1.25 K 1.5 1.75 The demographic parameters of L x and patterns of growth for populations of A. polyacanthus varied across the shelf on the central GBR. Although there was varia- tion in body size and growth among reefs within a distance stratum, it was minor compared to overall cross-shelf patterns. In this study, mortality estimates and maximum age were similar for populations of fish across the shelf. Thus, in order to explain the cross-shelf trend in body size, fish must have grown faster with increasing distance from shore (Fig. 7, Table 1). Despite the relative paucity of age-based studies on reef fishes (Choat and Robertson, 2002), variable rates of growth have been previously demonstrated for fish at local scales (hundreds of metres to kilometers: Fowler and Doherty, 1992), medium scales (kilometers to tens of kilometers: Choat and Axe, 1996; Hart and Russ, 1996; Newman et al., 1996; Meekan et al., 2001; Gust et al., 2002), and large scales (thousands of kilometers: Choat and Robertson, 2002). Gust et al. (2002) found that growth patterns of scarids varied between the reef crests of mid- and outer-shelf sampling locations on the northern GBR. In contrast to the results from the cur- rent study, however, outer-shelf populations of scarids had smaller asymptotic sizes and slower growth rates than mid-shelf populations. The factors influencing pat- terns of growth, therefore, vary by group. Differences in the shape of growth curves between geographic regions or areas may be determined by both genetic and environmental influences (Sebens, 1987). Populations of reef fish are generally considered to be genetically open systems (Sale, 1991) and it is consid- ered unlikely that adaptation of such populations to local conditions through genetic selection can occur (Warner, 1991). Acanthochromis polyacanthus, how- Figure 6 95*^ confidence ellipses for the von Bertalanffy growth parameters K (growth coefficient) and L, (mean asymptotic lengthl for Acantho- chromis polyacanthus from all reefs sampled. Table 3 Parameters from von Bertalanffy growth models on the fishes collected from different dis tance strata and reefs. Shelf location and reef n La A" r- Inner shelf Orpheus Island 36 87.03 0.77 0.83 Pandora Reef 44 77.43 1.39 0.92 Havannah Island 67 84.23 1.07 0.81 Mid-shelf Bramble Reef 97 92.24 1.04 0.83 Britomart Reef 85 96.37 0.95 0.87 The Slashers 91 106.73 0.98 0.75 Outer shelf Pith Reef 100 101.98 1.11 0.76 Barnett Patches 114 100.27 1.13 0.78 Myrmidon Reef 82 103.66 1.15 0.70 ever, possesses a unique life history trait among reef fishes in that it lacks a dispersive larval phase. The major implication of this characteristic is the potential for genetic isolation of populations of these fish. Even reefs that are in relatively close proximity to one an- other (100's of m) may become "genetic islands" isolated by any barrier that proves impassable to adults (e.g., deep water). Without gene flow, reproductively isolated 570 Fishery Bulletin 103(4) Table 4 Estimates of mortality (M) for fishes collected from dif- ferent distance strata and reefs. Pooled va ues are for all reefs within one distance s tratum n = number of fish in sample. S = animal survival rate. Pooled S Pooled Reef n M M (%) S(%) Inner shelf 0.51 60.0 Orpheus Island 30 0.29 74.8 Pandora Reef 34 0.40 67.0 Havannah Island 45 0.42 65.7 Mid-shelf 0.48 61.8 Bramble Reef 83 0.44 64.4 Britomart Reef 63 0.48 61.9 The Slashers 73 0.34 71.2 Outer shelf 0.43 65.1 Pith Reef 91 0.32 72.6 Barnett Patches 96 0.40 67.0 Myrmidon Reef 79 0.38 68.4 populations are expected to diverge over time with re- spect to their genetic composition (Doherty et al., 1994). Numerous studies have examined the genetic relation- ships between populations of A. polyacanthus on the GBR (Doherty et al., 1994, 1995; Planes and Doherty, 1997a, 1997b). Isozyme analyses of populations of dif- ferent color morphs at various spatial scales have shown significant genetic variation at both the regional ( 1000's of km) and local (100's of m) level, which under normal circumstances would suggest separate species for each color morph (Doherty et al., 1994; Planes and Doherty, 1997a). However, differences in the growth rates of A. polyacanthus across the continental shelf in this study are unlikely to reflect genetic differences between the populations sampled because all individuals collected were of the same color morph and were from a rela- tively small area (about 400 km2, cf. 450,000 km2 for the entire GBR). Environmental influences that can affect patterns of growth include predation pressure, temperature, and related effects on metabolism, variations in resources (e.g., abundance of planktonic food), and variation in water condition (e.g., turbidity). High rates of predation may "drive" faster growth (Werner, 1984), or conversely, select for early matura- tion and smaller adult size (Reznick et al., 1990; Hutch- ings, 1997). It is unlikely that the cross-shelf patterns in growth that we found were determined by differences in mortality rates. Some data on serranid abundance (Williams, 1982) and anecdotal accounts have indicted that predator abundance is greatest on mid- and outer reefs of the GBR (Gust et al., 2001). Our measures of instantaneous mortality (Z) and age maximum, how- ever, did not vary with distance from the mainland. Furthermore, in contrast to the patterns that Gust et Inner shelf (n=109) Mid-shelf (n=21 9) 4 ■ 3 y=-0.48x+6.04 r2=0.89 10 12 5 1 4 3 2 1 Outer shelf ( n=266) y=-0.43x+5.49 c2=0.95 4 6 Age (years) 10 12 Figure 7 Age-based catch curve estimates of Acanthochro- mis polyacanthus mortality rates for reefs pooled by distance strata. al. (2001) found for scarids, L.y increased with distance from the coast. Mortality rates have been shown to vary among locations within reefs for several species of coral reef fish (Aldenhoven. 1986; Eckert, 1987; Sale and Ferrell, 1988; Beukers and Jones, 1997) including A. polyacanthus juveniles (Connell, 1996), as well as over larger spatial scales (Meekan et al., 2001; Gust et al., 2002). In contrast to these last two studies, particularly that of Gust et al. (2002), mortality rates for A. poly- acanthus were similar at all three cross-shelf strata. We acknowledge, however, that no data were available on mortality rates of fish from zero to two years of age. It is possible that mortality rates do vary with distance from shore over this age range. An increase in adult size may occur when individu- als experience a decline in average temperature during development (Atkinson, 1994). It is also well established that metabolism and growth are increased at higher ambient temperatures in ectotherms (Schmidt-Nielsen, 1990). Differences in temperature between the water bodies spanning inner-, mid- and outer-shelf positions in the central GBR do occur; relatively shallow near- Kmgsford and Hughes: Growth, mortality, and size of Acanthochromis polyacanthus 571 shore waters are the warmest and outer-shelf waters are the coolest (Wolanski, 2001). The opposite pattern of growth to the one observed in this study would be predicted by this cross-shelf gradient in water tempera- ture. It is also considered unlikely that local upwelling events on outer-shelf reefs could produce the observed differences, but they could influence primary produc- tivity and abundance of food (zooplankton) through nutrient-rich waters. An increase in average annual temperature correlates with maximum age in some fishes (review Choat and Robertson, 2002), but we found no differences in age maximum across the shelf. We conclude that any differences in temperature across the shelf are not persistent enough to affect cross-shelf patterns of growth of A. polyacanthus. Differences in growth profiles can be more realisti- cally attributed to cross-shelf variation in some limiting resource! s). This variation in resources may influence the quality and quantity of food, suitable nest sites, ref- uges from predators and (or) wave exposure, and density of conspecifics and (or) other species that compete with A. polyacanthus for resources. Correlative studies have concluded that the distribution and abundance of coral reef fishes is strongly influenced (directly and indirectly) by physical factors such as wave exposure, sediment loads, water depth, and topographical complexity, as well as by biological factors (Williams, 1982). These factors also have the potential to affect growth rates. A combination of reduced resource levels and high population densities on outer-shelf reefs strongly indi- cated that growth profiles represent density dependence in scarids (Gust et al., 2001, 2002). Density of con- and hetero-specifics was not recorded for our study, but densities of A. polyacanthus were clearly greatest on the mid- and outer-shelf reefs. This observation is con- trary to the pattern noted by Williams ( 1982 ) who found greatest abundances of A. polyacanthus on inner- and mid-shelf reefs. Thresher (1983) suggested that food abundance is a limiting resource for A. polyacanthus and interspecific competition for food does occur. Thus, it is plausible that variation in abundance of and com- petition for food across the shelf may have influenced the growth rates observed in the present study. The large differences in cross-shelf densities and LJs of A. polyacanthus indicate that competition may be less important than variation in quantity and quality of food across the shelf. Biomass of planktivores is generally highest at mid- shelf reefs on the central GBR (Williams and Hatcher, 1983). Although data on cross-shelf abundance and dis- tribution of plankton are limited, Williams and Hatcher attributed this pattern to the increased availability of food (zooplankton) in mid-shelf waters. Upwelling of cold, nutrient-rich water from the edge of the continental shelf results in high biomasses of phytoplankton. Aging of the water (time since upwelling) is accompanied by a shift in dominant planktonic biomass to herbivorous and then carnivorous zooplankton. This shift in biomass composition occurs simultaneously with the prevail- ing wind-driven passage of water across the shelf and ultimately leads to the greatest biomass of zooplankton occurring in mid-shelf waters (Andrews and Gentien, 1982; Sammarco and Crenshaw, 1984; Williams et al., 1988). Food quality has also been previously shown to limit growth and reproduction in herbivorous coral reef fishes (Horn, 1989; Choat, 1991). Despite a high abundance of zooplankton near shore, these waters also have higher turbidity than mid- and outer-shelf reefs. Visual impairment caused by very tur- bid waters may hinder the ability offish to feed on plank- tonic organisms and this hypothesis has been suggested as a factor contributing to the low relative abundances of planktivorous fish on inner-shelf reefs (Williams et al., 1986). It is possible that this factor may retard the growth and influence the maximum size of planktivores like A. polyacanthus by effectively reducing food avail- ability. Interestingly, lowest L r values were found at the most turbid inshore reef, Pandora. Lower visibility near shore, however, did not appear to affect the mortality rates of A. polyacanthus at inner-shelf reefs. There were clear differences in growth, size maxima, and age structures for populations of A. polyacanthus across the continental shelf of the central GBR. Al- though Acanthochromis polyacanthus grew faster and to a larger size with increasing distance from the main- land, cross-shelf mortality rates and maximum ages were similar. Because these populations of fish are un- likely to be genetically distinct, we suggest that biotic and physical processes are the most plausible cause of these cross-shelf patterns. Increased abundance of zoo- plankton in mid- and outer-shelf waters, coupled with potential visual impairment associated with high tur- bidity levels on the inner shelf, are likely mechanisms that explain the observed patterns, but multifactorial manipulative experiments are required to determine the relative contribution of these factors to variation in demographic parameters. Our study therefore cautions against pooling demographic parameters over broad spa- tial scales without considering cross-shelf variation. Acknowledgments We would like to thank H. Patterson, C. Bunt, W. Rob- bins, and the crew of the RV Orpheus for field assistance during this study. We also thank J. Ackerman for analyt- ical advice and expertise and J. H. Choat for constructive comments on the manuscript. We also thank John Mor- rison and the staff of MARFU for assistance with the maintenance of aquarium fish. The project was partly funded by an ARC Grant to MJK. This is a contribution from Orpheus Island Research Station. Literature cited Aldenhoven, J. M. 1986. Local variation in mortality rates and life history estimates of the coral reef fish Centropyge bicolor (Pisces: Pomacanthidae). Mar. Biol. 92:237-244. 572 Fishery Bulletin 103(4) Allen, G. R. 1975. Damselfishes of the South Seas, 240 p. Tropical Fish Hobbyist Publications, Neptune City, NJ. Andrews J. C, and P. Gentien. 1982. Upwelling as a source of nutrients for the Great Barrier Reef ecosystems: a solution to Darwin's ques- tion? Mar. Ecol. Prog. Ser. 8:257-269. Atkinson, D. 1994. Temperature and organism size — a biological law for ectotherms? Adv. Ecol. Res. 25:1-58. Beukers, J. J., and G. P. Jones. 1997. Habitat complexity modifies the impact of pisci- vores on a coral reef fish population. Oecologia 114: 50-69. Choat, J. 1991. The biology of herbivorous fishes on coral reefs. In The ecology of fishes on coral reefs (P. F. Sale, ed.), p. 120-155. Academic Press, San Diego, CA. Choat, J., and L. Axe. 1996. Growth and longevity in acanthurid fishes: an analysis of otolith increments. Mar. Ecol. Prog. Ser. 134:15-26. Choat, J., and D. R. Robertson. 2002. Age-based studies. In Coral reef fishes (P. F. Sale, ed.), p. 57-80. Academic Press, Amsterdam, The Netherlands. Connell, S. D. 1996. Variations in mortality of a coral reef fish: links with predator abundance. Mar. Biol. 126:347-352. 1998. Effects of predators on growth, mortality and abun- dance of a juvenile reef fish: evidence from manipula- tions of predator and prey abundance. Mar. Ecol. Prog. Ser. 169:251-261. 2000. Is there safety-in-numbers for prey? Oikos 88(31:527-532. Dinesen, Z. D. 1983. Patterns in the distribution of soft corals across the central Great Barrier Reef. Coral Reefs 1:229-236. Doherty, P. J., S. Planes, and P. Mather. 1994. Acanthochromis polyacanthus, a fish lacking larval dispersal, has genetically differentiated popula- tions at local and regional scales on the Great Barrier Reef. Mar. Biol. 121(1):11-21. 1995. Gene flow and larval duration in seven spe- cies of fish from the Great Barrier Reef. Ecology 76(8):2373-2391. Done, T J. 1982. Patterns in the distribution of coral communities across the central Great Barrier Reef. Coral Reefs 1:95-107. Dudgeon, C. L., N. Gust, and D. Blair. 2000. No apparent genetic basis to demographic differ- ences in scarid fishes across the continental shelf of the Great Barrier Reef. Mar. Biol. 137:1059-1066. Eckert, G. J. 1987. Estimates of adult and juvenile mortality for labiid fishes at One Tree Reef, Great Barrier Reef. Mar. Biol. 95:167-171. Fowler, A. J., and P. J. Doherty. 1992. Validation of annual growth increments in the otoliths of two species of damselfish from the south- ern Great Barrier Reef. Aust. J. Mar. Freshw. Res. 43:1057-1068. 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. Aquat. Sci. 117. Gillanders, B. M. 1995. Feeding ecology of the temperate marine fish Acho- erodus viridis (Labridae): size, seasonal and site-specific differences. Mar. Freshw. Res. 46(71:1009-1020. Gust, N., J. H. Choat, and J. L. Ackerman. 2001. Demographic plasticity in tropical reef fishes. Mar. Biol. 140:1039-1051. Gust, N„ J. H. Choat, and M. McCormick. 2002. Spatial variability in reef fish distribution, abun- dance, size and biomass: a multi-scale analysis. Mar. Ecol. Prog. Ser. 214:237-251. Hart, A. M., and G. R. Russ. 1996. Response of herbivorous fishes to crown of thorns starfish Acanthaster planci outbreaks III. Age, growth, mortality and maturity indices of Acanthurus nigrofuscus. Mar. Ecol. Prog. Ser. 136:25-35. Horn.M. H. 1989. Biology of marine herbivorous fishes. Oceanogr. Mar. Biol. Annu. Rev. 27:167-272. Hutchings, J. A. 1997. Life history responses to environmental variability in early life. In Early life history and recruitment in fish populations (R. C. Chambers, and E. A. Trippel, eds.), p. 139-168. Chapman & Hall, London, UK. Kavanagh, K. 1998. Notes on the frequency and function of glanc- ing behavior in juvenile Acanthochromis (Poma- centridae). Copeia 1998(21:493-496. 2000. Larval brooding in the marine damselfish Acan- thochromis polyacanthus (Pomacentridae) is correlated with highly divergent morphology, ontogeny and life- history traits. Bull. Mar. Sci. 66(21:321-37. Kimura, D. K. 1980. Likelihood methods for comparison of von Berta- lanffy growth curves. Fish. Bull. 77:765-776. McCook, L. J., I. R. Price, and D. W. Klumpp. 1997. Macro algae on the Great Barrier Reef: causes or con- sequences, indicators or models of reef degradation? Proc. 8th Int. Coral Reef Symp. 2:1851-1856. McFarlane, G. A., and R. J. Beamish. 1987. Selection of dosages of oxy tetracycline for age valida- tion studies. Can. J. Fish. Aquat. Sci. 44:905-909. Meekan, M. G., J. L. Ackerman, and G. M. Wellington. 2001. Demography and age structures of coral reef dam- selfishes in the tropical eastern Pacific Ocean. Mar. Ecol. Prog. Ser. 212:223-232. Munro, J. L., and D. M. Williams. 1985. Assessment and management of coral reef fisheries: biological, environmental and sociological aspects. Proc. 5th Int. Coral Reef Congr. 4:545-581. Nakazono, A. 1993. One-parent removal experiment in the brood-caring damselfish, Acanthochromis polyacanthus, with pre- liminary data on reproductive biology. Aust. J. Mar. Freshw. Res. 44(51:699-707. Newman, S. J., and D. M. Williams. 1996. Variation in reef associated assemblages of the Lutjanidae and Lethrinidae at different distances off- shore in the central Great Barrier Reef. Environ. Biol. Fish. 46:123-138. Newman, S. J., D. M. Williams, and G. R. Russ. 1996. Age validation, growth and mortality rates of the tropical snappers (Pisces: Lutjanidae) Lutjanus adctii (Castelnau, 1873) and L. quinquelineatus (Bloch, 1790) Kmgsford and Hughes Growth, mortality, and size of Acanthochromis polyacanthus 573 from the central Great Barrier Reef, Australia. Mar. Freshw. Res. 47:575-584. 1997. Patterns of zonation of assemblages of the Lut- janidae, Lethrinidae and Serranidae (Epinephelinae) within and among mid shelf and outer shelf reefs in the central Great Barrier Reef. Mar. Freshw. Res. 48:119-128. Pauly. D. 1984. Fish population dynamics in tropical waters. ICLARM Stud. Rev. 8:1-325. Planes, S., and P. J. Doherty. 1997a. Genetic and color interactions at a contact zone of Acanthochromis polyacanthus: a marine fish lacking pelagic larvae. Evolution 51(41:1232-1243. 1997b. Genetic relationships of the color morphs of Accrn- thochromis polyacanthus (Pomacentridael on the north- ern Great Barrier Reef. Mar. Biol. 130(11:109-117. Preston, N. P., and P. J. Doherty. 1990. Cross-shelf patterns in the community structure of coral dwelling Crustacea in the central region of the Great Barrier Reef. 1. Agile shrimps. Mar. Ecol. Prog. Ser. 66:47-61. 1994. Cross-shelf patterns in the community structure of coral dwelling Crustacea in the central region of the Great Barrier Reef. 2. Cryptofauna. Mar. Ecol. Prog. Ser. 104:27-38. Reznick, D. N., H. Bryga, and J. A. Endler. 1990. Experimentally-induced life history evolution in a natural population. Nature 346:357-359. Ricker, W. E. 1975. Computation and interpretation of biological sta- tistics offish populations, 392. Bull. Fish. Res. Board Canada 191, Ottawa, Canada. Robertson, D. R. 1973. Field observations on the reproductive behavior of a pomacentrid fish, Acanthochromis polyacanthus. Zeit- schrift fur Tierpsychologie 32:319-324. Russ, G. 1984a. Distribution and abundance of herbivorous graz- ing fishes in the central Great Barrier Reef I. Levels of variability across the entire continental shelf. Mar. Ecol. Prog. Ser. 20:23-34. 1984b. Distribution and abundance of herbivorous grazing fishes in the central Great Barrier Reef II. Patterns of zonation of mid-shelf and outer-shelf reefs. Mar. Ecol. Prog. Ser. 20:35-44. Sale, P. F. 1991. Reef fish communities: open non equilibrial systems. In The ecology of fishes on coral reefs (P. F. Sale, ed.), p. 564-598. Academic Press, San Diego, CA. Sale, P. F., and D. J. Ferrell. 1988. Early survivorship of juvenile coral reef fishes. Coral Reefs 7:117-124. Sammarco, P. W., and H. Crenshaw. 1984. Plankton community dynamics of the central Great Barrier Reef lagoon: analysis of data from Ikeda et al. Mar. Biol. 82:167-180. Schmidt-Nielsen, K. 1990. Animal physiology: adaptation to environment, 4th ed., 602 p. Cambridge Univ. Press, Cambridge, UK. Schnute, J. 1981. Versatile growth model with stable parameters. Can. J. Fish. Aquat. Sci. 38:1128-1140. Sebens, K. P. 1987. The ecology of indeterminate growth in ani- mals. Annu. Rev. Ecol. Syst. 18:371-407. Thresher, R. E. 1983. Habitat effects on reproductive successes in the coral reef fish, Acanthochromis polyacanthus. Ecology 64:1184-1199. 1985a. Distribution, abundance and reproductive suc- cess in the coral reef fish Acanthochromis polya- canthus. Ecology 66:1139-1150. 1985b. Brood-directed parental aggression and early brood loss in the coral reef fish Acanthochromis poly- acanthus. Anim. Behav. 33:897-907. Warner, R. R. 1991. The use of phenotypic plasticity in coral reef fishes as tests of theory in evolutionary ecology. In The ecology of fishes on coral reefs (P. F. Sale, ed.), p. 387-398. Academic Press, San Diego, CA. Werner, E. E. 1985. The mechanisms of species interactions and com- munity organization in fish. /;; Ecological communities (D. R. Strong, ed.), p. 360-382. Princeton Univ. Press, Princeton, NJ. Wilkinson, C. R„ and A. C. Cheshire. 1988. Cross shelf variations in coral reef structure and function- influences of land and ocean. Proc. 6th Int. Coral Reef Symp. 1:227-233. Williams, D. M. 1982. Patterns in the distribution of fish communities across the central Great Barrier Reef. Coral Reefs 1:35-43. 1983. Longitudinal and latitudinal variation in the struc- ture of reef fish communities. In The inaugural Great Barrier Reef conference (J. T. Baker, R. M. Carter, P. W. Sammarco, and K. P. Stark, eds.l, p. 265-270. James Cook Univ. Press, Townsville, Queensland, Australia. Williams, D. M., and A. I. Hatcher. 1983. Structure of fish communities on outer slopes of inshore, mid-shelf and outer shelf reefs of the Great Barrier Reef. Mar. Ecol. Prog. Ser. 10:239-250. Williams, D. M., P. Dixon, and S. English. 1988. Cross-shelf distribution of copepods and fish larvae across the central Great Barrier Reef. Mar. Biol. 99:577-589. Williams, D. M., G. Russ, and P. J. Doherty. 1986. Reef fish: large-scale distribution and recruit- ment. Oceanus 29(21:76-82. Williams, D. M., C. R. Davies, B. D. Mapstone, and G. R. Russ. 2003. Scales of spatial variation in demography of a large coral reef fish: an exception to the typical model? Fish. Bull. 101: 673-683. Wolanski, E. 2001. Oceanographic processes of coral reefs: physical and biological links in the Great Barrier Reef, 356 p. CRC Press, Boca Raton, FL Zar,J. H. 1999. Biostatistical analysis, 4th ed., 663 p. Prentice Hall, Upper Saddle River, NJ. 574 Abstract — Aspects of the feeding migration of walleye pollock iTher- agra ehalcogramma) in the eastern Bering Sea (EBS) were investigated by examining the relationship be- tween temperatures and densities of fish encountered during acoustic and bottom trawl surveys conducted in spring and summer between 1982 and 2001. Bottom temperature was used as an indicator of spring and summer warming of the EBS. Clus- ters of survey stations were identified where the density of walleye pollock generally increased or decreased with increasing water temperature. Infer- ences about the direction and magni- tude of the spring and summer feeding migration were made for five length categories of walleye pollock. Gener- ally, feeding migrations appeared to be northward and shoreward, and the magnitude of this migration appeared to increase with walleye pollock size up to 50 cm. Pollock larger then 50 cm showed limited migratory behavior. Pollock may benefit from northward feeding migrations because of the changes in temperature, zooplank- ton production, and light conditions. Ongoing climate changes may affect pollock distribution and create new challenges for pollock management in the EBS. Variation in the distribution of walleye pollock (Theragra ehalcogramma) with temperature and implications for seasonal migration Stan Kotwicki Troy W. Buckley Taina Honkalehto Gary Walters Resource Assessment and Conservation Engineering Division Alaska Fisheries Science Center National Marine Fisheries Service. NOAA 7600 Sand Point Way NE Seattle, Washington 981 1 5-6349 E-mail (for 5 Kotwicki) Stan kotwicki'5'noaa gov Manuscript submitted 20 November 2004 to the Scientific Editor's Office. Manuscript approved for publication 30 March 2005 by the Scientific Editor. Fish. Bull 103:574-587 (20051. Walleye pollock (Theragra ehalco- gramma; referred to as "pollock" in this article) migrate seasonally. Such migrations have been described for the northern Sea of Japan (Maeda, 1986; Maeda et al., 1988, 1989; Kooka et al., 1998), Korean waters (Shuntov et al., 1993), the Okhotsk Sea (Shuntov et al., 1987), and the western and central Bering Sea (Fadeyev, 1989; Bulatov and Sobolevskiy, 1990; Efim- kin, 1991; Radchenko and Sobolevskiy, 1993; Shuntov et al., 1993; Balykin, 1996). Generally, these authors have described a spring and summer migration from spawning grounds to forage areas (referred to as a "feed- ing migrations" by many authors) and a winter migration of pollock returning to spawning grounds (e.g., Maeda et al., 1988; Radchenko and Sobolevskiy, 1993). This pattern of migration is believed to occur in the eastern Bering Sea (EBS) where it has received considerable attention (Takahashi and Yamaguchi, 1972 Francis and Bailey, 1983; Pola, 1985 Shuntov, 1992; Shuntov et al., 1993 Stepanenko, 2001), but the evidence for this pattern of migration is sparse. In addition, there is a lack of infor- mation on the magnitude of, routes of, and size-dependent differences in seasonal migrations. Temperature (and other factors closely related to temperature) af- fects the distribution and movements of pollock. Pola (1985) simulated temperature-induced migrations of pollock in the EBS occurring dur- ing May and June. Pollock appear to avoid some temperatures (Swartzman et al., 1994) and prefer environmen- tal conditions that are linked to food availability associated with tempera- ture gradients and fronts along the EBS slope (Swartzman et al., 1995). Water temperature is an especially important indicator of the transi- tion from winter conditions to those supporting a spring bloom of phyto- plankton and then zooplankton. In the EBS. the simulated onset of the feeding migration of pollock was de- layed in colder years (Pola, 1985). Annual surveys documenting the spatial distribution of fishes in re- lation to water temperatures can be used to infer details about their mi- gratory behavior. Using annual sur- vey data, Mountain and Murawski (1992) found that the relationship between the distribution of season- ally migrating species and water temperature could indicate a change in the overwintering location of the fish, or a change in the timing of the spring migration, or both. In the east- ern Bering Sea, bottom trawl (BT) surveys and echo-integration-trawl (EIT) surveys are conducted in late spring and summer (Honkalehto et al1; Acuna et al.2), when water tem- • 2 See next page for footnote texts. Kotwicki et al Variation in the distribution of Theragra chalcogramma 575 peratures are generally rising on the eastern Bering Sea shelf (Overland et al., 1999; Stabeno et al., 2001). Interannual vari- ability in climatic conditions and survey timing create vari- ability in mean water tempera- tures encountered during the surveys (Acuna et al.2). We describe the variability in distribution of pollock with tem- perature and propose that this variability may be explained by the fact that pollock migrate to feeding grounds during spring and summer. Temperature is used in our study as an indica- tor of how far into an idealized seasonal warming cycle each survey has occurred. Thus, the distribution of pollock observed in a warm year would be con- sidered to be representative of that seen later in a seasonal warming cycle in a cold year. Generally, feeding migrations appeared to be northward and shoreward, and the magnitude of this migration appeared to increase with walleye pollock size up to 50 cm. Pollock larger then 50 cm showed limited mi- gratory behavior. Pollock may benefit from northward feed- ing migrations because of the changes in temperature, zoo- plankton production, and light conditions. Materials and methods 64 N - 62"N 60"N 58' N 56'N - 54"N - 52 N BERING SEA rf* °Sk gg *- ~ssg£ 66 N 64 "N 62"N 60N 58"N 56°N 54-N 176°W 172°W 168W 164°W 1 60"W Figure 1 Locations of AFSC bottom trawl stations (dots) and echo-integration survey transects (lines) in the eastern Bering Sea where walleye pollock {Theragra chalcogramma) were collected during bottom trawl surveys and echo-inte- gration trawl surveys in spring and summer between 1982 and 2001. Data used in this investigation were collected by BT and EIT surveys conducted by the Alaska Fisheries Science Center. Since 1982, BT surveys have been conducted annu- ally over a standard area of the EBS, at the centers of 20x20 nautical-mile grids (Fig. 1). The corners of the grid block were also sampled in areas surrounding St. Matthew Island and the Pribilof Islands. The same 1 Honkalehto, T., N. Williamson, and S. de Blois. 2002a. Echo integration-trawl survey results for walleye pollock (Theragra chalcogramma) on the Bering Sea shelf and slope during summer 1999. U.S Dep. Commerce, NOAA Tech. Memo. NMFS-AFSC-125,77 p. - Acuna, E., P. Goddard, and S. Kotwicki (compilers). 2003. 2002 bottom trawl survey of the eastern Bering Sea continental shelf. AFSC Processed Report 2003-01, 169 p. Alaska Fish. Sci. Cent., NOAA Natl. Mar. Fish. Serv., 7600 Sand Point Way NE, Seattle, WA 98115. standard trawl (83-112 eastern otter trawl) was used every year (Acuna et al.2) and surveys usually began in late May or early June, and ended in August. Surveys always began in the northeastern corner of the Bristol Bay and proceeded westward. Samples were collected by towing for 30 minutes at 1.54 m/s (intended speed). Temperature data were collected during each tow us- ing an expendable bathythermograph (XBT) until 1992 and after 1992 with a micro-bathythermograph (MBT) attached to the headrope of the trawl. Catches were sorted by species and weight; number of fish caught and length-frequency data were collected for each tow. Echo integration trawl survey transects were de- signed to coincide with north-south lines of BT sta- tions. Similar to the BT survey, the EIT survey began also in the eastern Bristol Bay and proceeded west- ward. The time lag between the survey varied from 0 to 30 days. Acoustic data were collected with a Simrad EK500 quantitative echo sounding system. Biological 576 Fishery Bulletin 103(4) data were collected by midwater trawl, bottom trawl, and Methot trawl (see Honkalehto et al.1 for details). Pollock length data from trawls were aggregated into analytical strata based on echosign type, geographic proximity of hauls, and similarity in size composition of hauls. Estimates of numbers of pollock by size were de- rived by scaling acoustic measurements with the target strength-to-length relationship described in Traynor (1996). Temperature data were collected with an MBT mounted on the headrope of the trawl, although many of the profiles did not reach bottom because the trawls usually targeted midwater fish aggregations. For that reason, we elected not to use the temperature data collected during the EIT survey. Because both surveys were conducted at approximately the same time of year, we used the mean bottom temperature from the BT survey as an index temperature for the EIT survey. We used EIT data collected in years 1994, 1996, 1997, 1999, and 2000. Because of the semidemersal nature of pollock (Bailey et al., 1999a) and assuming that pollock do not dive as a boat and trawl approaches, BT data are assumed to describe the demersal part of the pollock stock within 3 m of the bottom. EIT data represented the midwa- ter part of the stock from 3 m above the bottom to 14 m below the surface. In our calculations, we used two density measures: CPUE in kg/ha for the BT data and biomass (tons) per 20-mile square for EIT data (the term "density" will be used in the present study to refer to both of these measures). Echo integration trawl survey 20-mile squares were centered on the BT survey stations, so that both sets of data could be easily compared (the term "station" will be used here to refer to BT survey stations as well as EIT survey squares). Because of known age-dependent behavioral differences between pollock (e.g., Shuntov et al., 1993; Bailey et al., 1999a), we investigated five different length classes of pollock; up to 20 cm (mostly 1-year-old pollock), 21-29 cm (mostly 2-year-old pollock), 30-39 cm, 40-49 cm, and pollock >50 cm. Because of differences in the year- class strengths between years, we scaled the data by dividing the density data for each station by the aver- age fish density for each year within each length class. Thus, a station with a density value of 1 has an average density for a given year and a station with a value of 5 has a density 5 times larger for a given year. If the pollock distribution in the EBS is assumed to be dynamic and related to temperature, the relationship between temperature and pollock density will be differ- ent at each spatial location. This means that if pollock moved from location A to location B over a period of rising temperatures, we expected a negative relation- ship between density and temperature in location A and an offsetting positive relationship in location B. To study these relationships in the EBS, we applied a two- step approach. In the first step, we identified possible locations where pollock density may be changing with temperature. In the second step, we identified locations of most significant biomass changes with temperature and quantified these changes. First step— identifying areas of change in fish density with temperature For both types of surveys, we calculated the slope of the linear regression of scaled density against bottom tempera- ture for each station over the time series (e.g., a slope value of 1 indicates an increase of 1 unit of density per degree increase of temperature). Slopes in the range between -0.3 and 0.3 were ignored because they represented areas of low fish density or areas of no significant changes in fish density between years. Each station slope was then plot- ted on a map to visualize the spatial relationship between these two variables for the BT and EIT surveys. To contour areas with similar slopes, we interpolated the data using inverse distance-weighted squared inter- polation (IDW). This method was chosen because IDW is an exact interpolator, where the maximum and mini- mum values in the interpolated surface can occur only at sample points and values at all sampling points are true measured values (ArcGIS, Geostatistical Analyst Help, 2003, ESRI, Redlands, CA). Using these maps, we identified the main spatially correlated clusters of stations with positive or negative slopes of the linear regression of pollock density against temperature (Figs. 2 and 3). Stations were assigned to clusters visually by using slope maps that overlapped the stations map. For practical reasons we investigated only clusters with four stations or more. Twenty-eight clusters were identified for BT survey and 17 clusters were identified for EIT survey (Figs. 2 and 3). Second step— identifying areas of most significant changes in biomass with temperature and quantifying these changes For each cluster, we calculated mean temperature and percentage of total biomass of pollock present in this cluster in each year. Total biomass and biomass within clusters were calculated as outlined in Wakabayashi et al. (1985). The relationship between mean bottom temperature and percentage of pollock biomass within each cluster was then fitted to a linear regression model. Because the error variances for the BT survey were not constant (variance increased with fish density), we weighted the regression by the inverse of the variance (Neter et al., 1996). For the EIT survey, we made no assumptions about the variance that was due to a small number of observations (only five years of data). The relative strength of the relationship between the percentage of pollock biomass and temperature within each cluster was characterized by the P-value of the slope (Table 1) (the P-values are not a true measure of statistical significance because the stations were not chosen randomly). Only clusters with the stron- gest relationships were used in the interpretation of results. Because the number of data points (years) in each analysis was equal within the survey (BT sur- veys— 20 points, EIT surveys — 5 points), P-values in- dicate relative strength of the temperature-biomass relationship. We plotted histograms of P-values for Kotwicki et a\ Variation in the distribution of Theragra cholcogromma 577 Biomass decrease Biomass increase with temperature 180 w 176°w 172°w 168°w 164°w with temperature Clusters: Slope < - 0.3 (biomass decreases with temperature) ^m Slope > 0.3 (biomass increases with ^™ temperature) Columns: mm Predicted % of total biomass in the ^^ area during warmest year Predicted % of total biomass in the area during coldest year Standard error bar 172°W 168 W 164:W Figure 2 Clusters of positive and negative slopes of the linear regression of pollock (Theragra chalcogramma) density (detected by echo-integration trawl survey) when plotted against temperature. Columns represent predicted per- cent biomass offish in these clusters within the observed range of temperatures. Predicted percent of biomass is shown only for clusters with the strongest relationship between temperature and fish density with the exception of cluster Fl (see results for explanation). Labels are located at the geographic centers of the clusters. 578 Fishery Bulletin 103(4) Biomass decrease with temperature . I8tm 176^ 172-W I68"W 0.3 (biomass increases with temperature) Columns: I Predicted % of total biomass in the area during warmest year 1 Predicted % of total biomass in the area during coldest year Standard error bar Figure 3 Clusters of positive and negative slopes of the linear regression of pollock IT. chalcogramma) density (detected by bottom trawl survey) when plotted against temperature. Columns represent predicted percent biomass of fish in these clusters within the observed range of temperatures. Predicted percent of biomass is shown only for clusters with the strongest relationship between temperature and fish density. Kotwicki et al.: Variation in the distribution of Theragra chalcogramma 579 Table 1 Results of linear regression analyses and predicted percent of total bioma ss in each clustei within an ob served range "1 temperatures. Standard Percentage Standard error Percentage Standard error error at min. of min. at max. of max. Cluster Slope of slope Intercept r2 P temperature percentage temperature percentage Botton trawl survey Al 3.0706 4.579 7.2013 0.024 0.511 8.82 5.07 15.49 4.89 A2 1.1579 0.382 1.6145 0.338 0.007 0.55 0.36 4.31 1.01 A3 -24.0232 6.812 98.8454 0.409 0.002 47.04 5.33 5.79 6.38 A4 3.3081 1.254 2.9747 0.279 0.017 3.14 1.22 13.75 2.84 A5 2.7928 0.837 -0.5577 0.382 0.004 0.08 0.47 10.13 2.65 Bl 6.2744 1.655 4.5969 0.444 0.001 1.98 1.21 18.30 3.13 B2 -5.7916 3.445 24.2379 0.136 0.110 21.73 6.37 5.44 3.34 B3 -0.2449 0.839 4.9690 0.005 0.774 4.86 2.07 4.07 0.70 B4 3.6720 1.880 1.4972 0.175 0.066 2.83 0.52 12.05 4.33 Bo -26.2623 7.642 106.0603 0.396 0.003 70.17 14.89 5.08 4.06 B6 1.9462 0.756 2.9913 0.269 0.019 0.62 1.12 9.27 2.31 B7 3.4809 1.408 -0.5292 0.253 0.024 0.77 0.23 13.03 4.94 B8 0.5026 0.350 0.0979 0.103 0.168 0.10 0.44 2.47 1.30 CI 10.9174 2.355 in 7957 0.544 0.000 2.28 0.31 32.32 6.32 C2 -17.9173 5.809 60.49S7 0.346 0.006 42.49 7.89 0.51 5.72 C3 1.1807 0.731 1.5962 0.127 0.124 0.84 1.20 4.68 1.26 C4 -26.8246 7.771 116.1162 0.398 0.003 68.32 10.03 12.06 6.27 C5 5.3395 1.413 -3.3847 0.442 0.001 0.11 1.79 16.77 2.65 C6 -1.4934 0.843 6.3588 0.148 0.094 4.77 2.07 0.89 0.28 Dl 5.4677 1.368 7.2510 0.470 0.001 -0.15 0.44 15.42 3.55 D2 -1.9038 0.850 5.7227 0.218 0.038 4.50 1.39 0.87 0.45 D3 0.9668 2.954 2.9573 0.006 0.747 4.89 1.28 6.65 4.19 D4 -14.3609 5.356 65.9267 0.285 0.015 41.08 8.45 10.70 2.89 D5 3.6882 1.411 2.7672 0.275 0.018 4.92 1.18 15.81 3.02 D6 2.3974 0.830 -8.0125 0.317 0.010 0.04 0.31 2.60 0.81 El 4.5778 0.733 6.1399 0.684 0.000 -1.45 1.13 14.86 1.56 E2 -5.7776 1.910 26.8244 0.337 0.007 22.13 3.71 7.32 1.24 E3 0.6479 0.217 0.9447 0.332 0.008 0.91 0.28 3.80 1.01 Echo-integration trawl survey Fl 16.0479 13.649 -1.2116 0.315 0.324 10.35 21.80 51.49 20.57 F2 -27.5255 9.304 79.2218 0.744 0.059 59.39 14.86 -11.18 14.02 F3 11.7970 3.622 -13.4672 0.779 0.047 -4.97 5.76 25.28 5.46 F4 5.1896 8.463 -3.8753 0.111 0.583 -0.13 13.52 13.17 12.75 Gl 32.3770 4.047 -12.0017 0.955 0.004 11.32 6.46 94.33 6.10 G2 -25.5850 10.151 75.6302 0.679 0.086 57.20 16.22 -8.40 15.30 HI 28.0501 4.012 -16.7011 0.942 0.006 3.51 6.41 75.42 6.05 H2 -10.6961 2.999 32.2026 0.809 0.037 24.50 4.79 -2.93 4.52 H3 -11.0388 3.249 38.1076 0.793 0.042 30.16 5.19 1.85 4.90 H4 -5.0998 2.465 13.8907 0.587 0.130 10.22 3.94 -2.86 3.71 11 19.1934 7.184 -13.8822 0.704 0.075 -0.06 11.48 49.15 10.83 12 -21.2245 8.449 73.6602 0.677 0.086 58.37 13.50 3.95 12.73 13 2.3497 1.169 -0.0335 0.573 0.138 1.66 1.87 7.68 1.93 Jl -15.6292 2.465 50.1504 0.930 0.007 38.89 3.94 -1.18 3.72 J2 9.6097 4.374 -6.9678 0.616 0.115 -0.04 6.99 24.59 6.59 J3 9.7424 1.675 1.5600 0.918 0.010 8.58 2.68 33.56 2.53 ■J4 -5.8055 2.571 17.1422 0.629 0.109 12.96 4.11 -1.92 3.88 580 Fishery Bulletin 103(4) BTS ■MM, EITS 1 04 06 P-value Figure 4 Histograms of P-values of linear regression between fish density and temperature calculated for all clusters. Circled bars represent the clus- ters with strongest relationship. each survey (Fig. 4) and the groups of clusters with the strongest relationships between fish biomass and temperature were chosen for further investigations. These groups consisted of 21 clusters from BT surveys with P-values between 0.000 and 0.066 and 15 clusters from EIT surveys with P-values between 0.004 and 0.138. Using linear regression models (biomass against temperature), we calculated the predicted percentage of the total pollock biomass for each of these clusters (Table 1) within the temperature range observed during surveys (Fig. 5). To evaluate a spatial scale on which biomass redistri- bution occurred for the EIT surveys, we calculated mean distance between clusters of negative and positive slope (Table 2). To obtain these values, we generated 100 ran- dom points within each of the clusters and calculated the mean distance between all possible pairs of points from both clusters. We did not attempt to calculate this distance for the BT surveys because of the much more complicated nature of the BT cluster maps. Results Northward and inshore shifts in pollock distribution in warmer years were found in the EBS for all length cat- egories. The location and magnitude of these shifts and distance between clusters differed with the survey type and length categories. In the present study we address changes in pollock distribution by length category within each survey. Table 2 Mean distance between largest echo-integration trawl (EIT) survey clusters. Clusters for pollock >50 cm were not calculated because of low selectivity of the EIT survey for these fish. Clusters Mean distance (km) 99 % confidence interval (km) F2-F1 241.3 2.3 G2-G1 217.5 2.5 H4, H3, H2-H1 368.3 4.7 12-11 453.7 3.9 Echo-integration trawl survey The biomass of pollock <20 cm in cluster F2 near Zem- chung Canyon at latitude 59°N decreased (with increas- ing temperature) from about 59% of the total biomass of pollock in the coldest year to 0% in the warmest year (Fig. 2A). This decrease was partially offset by the increase in pollock biomass in area F3, northwest of the Pribilof Islands. The relatively weak relationship (P-value=0.324) between pollock biomass and temper- ature in cluster Fl (north of F2) was caused by the extremely high abundance of <20 cm pollock within cluster F4 during 1997. Therefore the percentage of total Kotwicki et al.: Variation in the distribution of Theragra cholcogramma 581 biomass was particularly low in clusters Fl, F2, and F3 for that year. In cluster Fl we observed an increase in biomass from 10% to 51%. For pollock 21-29 cm, changes between cluster G2 and Gl resembled changes between clusters F2 and Fl. The percentage of total biomass in these two clusters changed from 57% to 0% and from 11% to 94%, respec- tively (Fig. 2B). A slightly different situation was observed for pol- lock 30-39 cm (Fig. 2C). We identified three clusters of decreasing biomass with temperature: H2, H3, and H4 located, respectively, northwest of Zhemchug Can- yon, northwest and east of the Pribilof Islands. Overall predicted biomass change in H2, H3, and H4 decreased from 65% to 2%. The offset for this negative change was found in cluster HI, where we noted a positive change from 4% to 75%. Areas with decreasing fish biomass for pollock 40-49 cm were located within cluster 12 (Fig. 2D). Biomass de- creased from 58% in the coldest year to 4% in the warm- est year. We observed temperature-related increases in biomass mostly north of 12 in cluster II (0%-49%). A quite different situation was observed for pollock >50 cm (Fig. 2E). Although pollock of this size seemed to concentrate northwest and northeast of the Pribilof Islands (similar to pollock 30-49 cm) during cold years; in warm years they were found in EIT surveys mainly in the southeast, as opposed to the smaller fish that are found mainly in the north. Results for pollock >50 cm should be treated cautiously because only a very small part of the entire population of pollock this size can be detected with the EIT survey (Ianelli et al.3). Because of the benthic habits of pollock >50 cm (Shuntov et al., 1993), most were detected in BT surveys. Overall, our analysis of EIT survey data indicated a northward temperature-related shift of 50-80% of pollock <50 cm in two major areas. With increasing temperature, the density of pollock <40 cm decreased northwest of Zhemchug Canyon in a large area at 100 m to 200 m depths. Similarly, the density of pollock 30- 49 cm decreased northwest of the Pribilof Islands. Off- setting these decreases, pollock density increased in the northernmost area of the survey (close to the U.S. -Rus- sia Convention Line). Although the direction of the shift was the same for all length categories up to 50 cm, the mean distance between the clusters with negative slopes and clusters with positive slopes increased with fish size (Table 2). Bottom trawl survey For pollock <20 cm, we observed a decrease in pollock biomass with temperature in cluster A3 covering the 3 Ianelli, J. N., T. Buckley, T. Honkalehto, N. Williamson, and G. Walters. 2001. Bering Sea-Aleutian Islands wall- eye pollock assessment for 2002. In Stock assessment and fishery evaluation report for the groundfish resources of the Bering Sea/Aleutian Islands regions, p. 1-105. North Pac. Fish. Manag. Council. Anchorage, AK. area west of the Pribilof Islands and north to Zhemchug Canyon (Fig. 3A). We observed an increase in pollock biomass in shallower areas north of Pribilof Island (A4), as well as in the areas of 50-100 m depth east from the Pribilof Islands (A5). The magnitude of change was somewhat smaller than that observed for the EITS survey (see Fig. 3A for details). For pollock 20-29 cm, we observed a decrease in biomass from 70% to 5% in the area northwest of the Pribilof Islands (cluster B5). A cumulative increase in biomass from 7% to 52% of total biomass was observed in clusters Bl and B4 north of B5, and in clusters B6 and B7 in shallower waters (Fig. 3B). Relatively weak relationships were found between pollock biomass and temperature for clusters B2, B3, and B8. For pollock 30-39 cm, we observed a temperature- related decrease in biomass in clusters C2 and C4 (42% to 1%, and 68% to 12% accordingly) (Fig. 2C). Increase in biomass was observed in cluster CI (2-32%) north from C2. Positive change was also observed in cluster C5 (0-17%) within the shallow (<100 m) part of the southeastern Bering Sea shelf. Clusters D2 and D4 represented areas where we ob- served a significant decrease in biomass for pollock 40-49 cm (from 5% to 1%, and from 41% to 11%) (Fig. 3D). Increased biomass was detected in cluster Dl lo- cated north from D4 and in D5 located to the east of D4 in shallower waters. Very small changes were detected for pollock >50 cm. Although three clusters had a relatively strong pollock biomass and temperature relationship, the magnitude of biomass changes within the range of observed tem- peratures was quite small (Fig. 3E). Overall, as with the EIT surveys, northward shifts in distribution in warmer years were found in the BT survey data for pollock <30 cm. The magnitude of these northward shifts was somewhat smaller (15-30%) than those detected by EIT surveys. In addition, these data suggested an inshore eastward redistribution of pollock in warmer years. Changes for pollock >50 cm were evi- dent but small (in the range of 15%). Discussion Inferring seasonal pollock migration from interannual variations in distribution Interannual differences in the timing of the migration from spawning grounds to forage areas are related to water temperatures. The relationship between tem- perature and the spatial distribution of a seasonally migrating species could represent either a change in the winter location of the stock or a change in the timing of the migration or both (Mountain and Murawski, 1992). Although the evidence is not conclusive, data suggest that most pollock populations spawn in late winter or early spring in the same locations year after year (Bailey et al., 1999a). For example, large, prespawn- ing aggregations of pollock have been surveyed around 582 Fishery Bulletin 103(4) Bogoslof Island every year since 1988 in the winter (Honkalehto et al.4). Further support that temperature is related to the timing of the postspawning migration may come from temperature effects on physiological aspects of spawning. Cold water temperatures may delay the onset of spawning and extend the spawning period of walleye pollock as has been found for another gadid (Kjesbu, 1994) and for flatfish (Lange and Greve, 1997) in the Atlantic. The surveyed distribution of pollock in warmer years should be more representative of that seen later in a typical spring-summer warming cycle than the distri- bution of pollock seen in colder years. Bottom tempera- tures generally increased over the EBS and northern Bering Sea (NBS) during spring and summer (Overland et al., 1999; Khen et al., 2001; Stabeno et al., 2001). Our results show that the warmer the bottom water during spring-summer groundfish surveys, the farther away pollock <50 cm are found from their major spawn- ing grounds. Thus, we interpret areas having lower pollock density with increasing temperature (clusters with negative slope) to be areas from which pollock are emigrating, and areas having higher pollock density with increasing temperature (clusters with positive slope) to be areas to which pollock are immigrating (Figs. 2 and 3). Routes and directions of the migrations As the water warms during spring and summer, pol- lock generally migrate northward, northwestward, and inshore to shallower waters. Larger pollock (>30 cm) begin their feeding migration from spawning grounds. In many areas (white areas — Figs. 2 and 3) we did not detect a significant increase or decrease in pol- lock abundance in relation to temperature, e.g., in the major pollock spawning area north of Unimak Island (Hinckley, 1987; Bulatov, 1989), and this finding may indicate that migration progressed beyond this area before it was surveyed, even in the coldest years, or that migrations were not pronounced in this area. However, we observed a very large decrease in biomass with increasing temperature in the Pribilof Islands area (i.e., within clusters A3, B5, C4, D4, E2, H3, and 12), which is another important pollock spawning location (Maeda and Hirakawa, 1977; Hinckley, 1987; Bulatov, 1989; Bailey et al., 1999a). An offsetting increase in biomass was observed in the northernmost part of the survey area (clusters Bl, CI, Dl, Fl, Gl, HI, and ID and in shallower waters (clusters A4, A5, B6, B7, C5, and D5), which may indicate that pollock migrate north and inshore during the warming season. Echo integration trawl data indi- Honkalehto, T., N. Williamson, D. Hanson. D. McKelvey, and S. de Blois. 2002b. Results of the echo Integration-trawl survey of walleye pollock (Theragra chalcograma) conducted on the southeastern Bering Sea shelf and in the southeastern Aleutian Basin near Bogoslof Island in February and March 2002. AFSC Processed Report 2002-02, 49 p. Alaska Fish. Sci. Cent., NOAA, Natl. Mar. Fish. Serv., 7600 Sand Point Way NE, Seattle, WA 98115. cate that smaller pollock (<29 cm) probably begin their migration from overwintering areas (clusters F2 and G2) located mainly northwest of the Zhemchug Canyon. These results agree with observations made by Bailey et al. (1999b) that small age-0. age-1, and age-2 pollock are distributed farther north than larger age-3 and older pollock. Migrations continued generally northward to the U.S. -Russia Convention Line. The near-bottom part of the pollock population (detected in the BT survey) also migrates northeastward into shallower waters. At this point we cannot describe the exact starting and ending points of migration but only the general direc- tion, because surveys are performed after most of the spawning has been completed, and we lacked data for the NBS, where part of the pollock EBS population is probably migrating. The direction of movements indicated by the EIT survey data and the BT survey data were somewhat different because of the effect of depth on the avail- ability of pollock to each survey. As pollock migrate into shallower water they become more available to the BT survey and less available to the EIT survey. Therefore the BT survey indicates greater movement into shal- lower water, whereas the EIT survey indicates greater movement in a northerly direction. Seasonal migrations by pollock in the EBS are broad- ly recognized as occurring but have not been well sub- stantiated; however, most of the general observations and descriptions are in agreement with our results. It is generally recognized that the feeding migration of some EBS pollock takes them northwestward beyond our survey area and into Russian waters (Shuntov et al., 1992; 1993; Stepanenko, 2001). Pola (1985), in her numerical simulation of pollock migrations in the EBS identified two types of pollock feeding migration. One was temperature induced in the northward direction, and the other was seasonal in the northeastern direc- tion toward shallower waters. Shuntov et al. (1993) considered migrational activity to start with the on- set of sexual maturity, but our findings indicate that immature pollock do undergo feeding migrations in a northwestward direction, but over shorter distances than those traveled by mature pollock. Stepanenko (2001) also recognized migration by immature pollock. Only a few pollock tagged in the EBS have been recov- ered (Yoshida, 1985), but the relationships between the release and recovery locations are consistent with our findings of a northwestward feeding migration during the spring and summer over most of the EBS shelf and a northeastward migration into shallower water on the southeast EBS shelf. Length-based differences in migration patterns Our analysis of the EIT surveys indicates that the migrations of pollock <30 cm are shorter than those of pollock 30-50 cm. The distance pollock need to cover from clusters F2 and G2 to clusters Fl and Gl (241.3 km and 217.5 km) is much shorter than the distance to be covered by larger fish from clusters H4, H3, H2, Kotwicki et al : Variation in the distribution of Themgro chalcogrommo 583 and 12 to clusters HI and II (368.3 km and 453.7 km, respectively). Similar size-dependent differences in the distance of seasonal migrations were reported for Pacific hake (Merluccius productus), another gadoid from the north Pacific (Dorn, 1995). These observations may support the length-based hypothesis of Nottestad et al. (1999) for feeding migrations in pelagic fish. Focusing on the energetic cost-benefit relationship of long distance migration, they concluded that migration distance is a function of length, weight, and age. Smaller fish may undergo shorter feeding migrations because the ener- getic cost of migration can exceed their total energy intake resulting from the of greater hydrodynamic drag associated with smaller fish size. Migrations of the largest pollock (>50 cm), detected from the BT survey data, were of much lower magnitude then those of smaller fish. Our models indicate that only about 15% offish in this length category move between clusters in the northeastern direction toward shallower waters. These small changes detected in BT data con- tradict those seen in EIT data. Whereas a small north- ward shift in biomass (mostly from cluster E2 to cluster E4) was detected with BT data, a southeastward shift was detected with EIT data. However, because the EIT survey is not well suited for estimating the distribution of pollock >50 cm, we are inclined to put more weight on the BT data to explain temperature-related changes in biomass distribution for this length category. Larger pollock (>50 cm) appear to change their migratory be- havior. Shuntov (1992) noticed that the distribution of larger pollock (>54 cm) fundamentally differs from that of smaller pollock and that larger pollock are more benthic in behavior and feeding. Stepanenko (2001) did not observe any migrations to the Russian zone for pol- lock six years or older. We propose that the difference in the migratory behavior between pollock <50 cm and pollock >50 cm is linked to a well-known shift toward a diet offish with increasing pollock size (Bailey and Dunn, 1979; Dwyer et al., 1987). Why do pollock migrate? Pollock feeding migrations in the EBS may be driven by a combination of four factors: temperature, zooplankton production, currents, and length of daylight. Changes in the water temperature may affect pol- lock migrations. Bottom water temperature over the Bering Sea shelf rises between April and September (Pavlov and Pavlov, 1996; Overland et al., 1999; Khen et al., 2001; Stabeno et al., 2001). Our results indicate that with rising temperature pollock generally migrate northward and inshore. Pollock appear to avoid tem- peratures below 0°C (Swartzman et al., 1994); therefore a seasonal increase in temperature above 0°C can open new geographic areas for migration. Temperature was presented as one of several important stimuli affect- ing fish movements by Harden Jones (1968) and by Wielgolaski (1990), who noticed that capelin (Mallotus villosus), Atlantic cod (Gadus morhua), and haddock (Melanogrammus aeglefinus) in the Barents Sea migrate north towards a preferred temperature, either directly to satisfy metabolic requirements, or indirectly, as when attracted by food organisms. Seasonal patterns in zooplankton production and prey availability largely coincide with seasonal patterns in pollock migration and distribution. The role of food availability in driving fish-feeding migrations has been described for other zooplanktivores such as Pacific hake (Dorn, 1995), Atlantic herring (Clupea harengus), blue whiting iMieromesistius poutassou), mackerel (Scomber scombrus) and capelin (Nottestad et al., 1999). In the Bering Sea, the abundance of zooplankton is high on the EBS and NBS shelf throughout spring and sum- mer, but it remains high in autumn only in the NBS (Springer et al., 1989; Chuchukalo et al., 1996; Coyle et al., 1996). Copepods and euphausiids are major prey groups for pollock during spring and summer in the northwest area of the EBS shelf, but in autumn, 30-49 cm pollock increase their feeding on fish and decapods (Dwyer et al., 1987) which may be related to a decrease in the availability of these prey (Willette et al., 1999) in this area. Further north in the Navarin-Anadyr area, copepods and euphausiids remain major prey compo- nents in the diet of pollock <50 cm through summer and autumn (Shuntov et al., 2000). The migration pattern of pollock indicates they may follow their food supply as the production and abundance of zooplankton proceeds northward. Pollock larger than 50 cm do not undergo northward feeding migrations because small pollock, other fish, and benthos are the main components of the diet (Dwyer et al., 1987; Yoshida, 1994; Shuntov at al., 2000). In the area of pollock migrations northwest of Pribilof Islands current speeds are in the range of 1-5 cm/s at the 100 m depth and they generally run in the north- west direction (Stabeno et al., 2001). Current direction coincides with the direction of pollock migrations, so that the cost of the migration may be offset by swim- ming in the same direction as the transporting cur- rent (Nottestad et al., 1999). Water currents can also influence fish migration indirectly by providing visual stimuli arising from the moving background (Harden Jones, 1968) or by transporting food. Springer et al. (1989) suggested that the transport of zooplankton by the northwest current may cause greater levels of zoo- plankton concentration in the NBS. Because of the lack of data on current speed, he speculated that a current velocity in the range of 20 cm/s was needed to explain these high levels of zooplankton in the NBS if the high levels of zooplankton are based only on currents. The latest observations of current on the Bering Sea shelf do not support these hypotheses (Stabeno et al., 2001). However northwestern currents may contribute to high- er zooplankton biomass in the NBS. Nottestad et al. (1999) suggested that light conditions may play a role in fish feeding migrations because dur- ing summer day-length increases the farther north fish travel, thus potentially increasing feeding duration for pelagic visual predators. Pollock are visual predators and light conditions affect feeding efficiency of pollock 584 Fishery Bulletin 103(4) (Ryer and Olla, 1999; Ryer et al., 2002); therefore it may be that longer days at northern latitudes make a north- ward feeding migration beneficial by possibly providing an extended window of search time if the pollock happen to be in a locally depauperate area. However, day-length remains long enough in the entire Bering Sea for pollock to feed to satiation, and their gastric evacuation rate is slow (Dwyer et al., 1987), making the need to entirely fill their stomachs every day very unlikely. 62" N 6CTN 58:N - 56:N - 54°N 62°N - 60°N 58 N - 56"N 54°N 180°W 175°W 1 1 70"W t — r 165°W 180°W At this time it is impossible to assess which factor is most important in driving pollock migrations, but in summary we can conclude that pollock, as visual pelag- ic predators, benefit from northward feeding migrations during seasonal warming. Because three of the factors (excluding current) are similar throughout the Northern Hemisphere, we should see similar migration patterns for other pelagic fish of the north. Other examples in- clude Pacific hake migrating along the North American west coast from California to British Columbia (Francis and Bailey, 1983; Dorn, 1995). Her- ring in the Norwegian Sea undergo seasonal feeding migrations in the northwestern direc- tion from the south-central coast of Norway to the areas located northeast of Iceland (Ferno, 1998). Blue whiting, mackerel, and capelin from the north Atlantic undergo northward feeding migrations (Nottestad et al., 1999). Pacific saury (Cololabis saira), chub mackerel (Scomber japonicus). Pacific sardine (Sardinops sagax melanosticta), and Japanese anchovy (Engraulis japonicus) from the western North Pacific are reported to migrate northwards during the summer (Novikov, 1986). Capelin, Atlantic cod, and haddock in the Barents Sea migrate north towards a "preference" tempera- ture during summer (Wielgolaski, 1990). All these species have characteristics similar to those of Bering Sea pollock — that is, a pelagic or semipelagic life style, a diet of zooplankton, winter or spring spawning activity, and feed- ing migrations that take place during spring and summer. 64°N 62°N 60°N 58°N 56° N - 64 'N _ 62- N - 60"N -\-- 58°N - 56°N 175'W 170 W 165 -W 1 60"W Figure 5 Bottom water temperature contours during the bottom trawl survey in the coldest year (1999 — upper map I and warmest year (1996 — lower map). Why is temperature important? Temperature may affect the proportion of the stock that is in the standard EBS survey area. Ianelli et al.,3 using population modeling, esti- mated that fewer pollock were detected during the BT survey in the EBS with increasing tem- perature, and fewer pollock would indicate that pollock are probably leaving the survey area during seasonal migrations. We conclude that a significant part of the EBS pollock popula- tion migrates into the Navarin-Anadyr area, which can have an impact on the way the EBS stock is managed. We should account for land- ings of pollock in the Navarin-Anadyr area, estimate how much of these landings include pollock from the EBS stock, and use this esti- mate in determining the EBS total allowable catch. Further research is needed to quantify the proportion of the EBS stock migrating into the Russian fishing zone and to estimate the number of pollock caught there. Stokes5 suggested that the biomass estimates from the NBS are in the range of 0.5-1.0 million ' See next page for footnote text. Kotwicki et al.: Variation in the distribution of Theragra chalcogramma 585 metric tons per annum and the exploitation rate is in the range of 0.5 million metric tons (50-100% of the total estimate). Ongoing climate changes may affect pollock distri- bution between the U.S. and Russian EEZs. Stabeno and Overland (2001) reported a shift toward an earlier spring transition in the Bering Sea. This can affect the starting time of pollock migrations and the length of time fish spend in the Russian EEZ, increasing the availability of fish to the Russian fleet. This situation should encourage us to closely monitor changes in mi- gration patterns of pollock in the Bering Sea. Significant bias or error variation may be caused by the interaction of fish movement with survey protocol. For even relatively low fish migration velocities (<0.5 m/s), bias in estimated fish biomass can be very large (McAllister, 1998). Therefore, fish migration vectors should be estimated to minimize the bias created by not taking into account these migrations in biomass estimates. Acknowledgments The authors thank Angie Greig and Jan Benson for an introduction to ArcGIS and help with geospatial prob- lems that occurred during analyses of data. We also want to thank Kevin Bailey, Jerry Hoff, Jim Ianelli, Jay Orr, David Somerton, Phyllis Stabeno, Gary Stauffer, Neal Williamson, and three anonymous reviewers for discus- sions and review of earlier versions of this manuscript. Literature cited Bailey, K., and J. Dunn. 1979. Spring and summer foods of walleye pollock, Ther- agra chalcogramma, in the eastern Bering Sea. Fish. Bull. 77:304-308. Bailey, K. M„ T. J. Quinn II, P. Bentzen. and W. S. Grant. 1999a. Population structure and dynamics of walleye pollock, Theragra chalcogramma. Adv. Mar. Biol. 37:179-255. Bailey, K. M., D. M. Powers, J. M. Quattro, G. Villa, A. Nishimura, J. J. Traynor, and G. Walters. 1999b. Population ecology and structural dynamics of walleye pollock (Theragra chalcogramma). In Dynamics of the Bering Sea (T. R. Loughlin and K. Ohtani, eds.), p. 581-614. Univ. Alaska Sea Grant publ. AK-SG-99- 03, Fairbanks, AK. Balykin, P. A. 1996. Dynamics and abundance of western Bering Sea pollock. In Ecology of the Bering Sea. p. 177-182. Alaska Sea Grant Report 96-01, Fairbanks, AK. 5 Stokes, T. K. 2000. Review of Alaska pollock assessment. Appendix 3 in Eastern Bering Sea walleye pollock stock assessment for 2001 (J. N. Ianelli, L. Fritz, T. Honkalehto, N. Williamson, and G. Walters), p. 179-188. In Stock assessment and fishery evaluation report for the groundfish resources of the Bering Sea/Aleutian Islands regions. North Pac. Fish. Manag. Council, Anchorage, AK. Bulatov, O. A. 1989. Reproduction and abundance of spawning pollock in the Bering Sea. In Proceedings of the international symposium on the biology and management of walleye pollock; November 14-16, 1988, Lowell Wakefield Fisher- ies Symposium, p. 199-208. Alaska Sea Grant Report 89-1, Anchorage, AK. Bulatov, O. A., and E. I. Sobolevskiy. 1990. Distribution, stocks, and fishery prospects of wall- eye pollock in the open waters of the Bering Sea. Bi- ologiya Morya 5:297-304. Chuchukalo, V. I., N. A. Kuznetsova, and V. V. Napazakov. 1996. Seasonal distribution of euphausids in the Bering and Okhotsk Sea and adjacent waters of Pacific Ocean. Izv. TINRO, 119:256-281. Coyle, K. O.. V. G. Chavtur, and A. I. Pinchuk. 1996. Zooplankton of the Bering Sea: a review of Rus- sian-language literature. In Ecology of the Bering Sea, p. 97-134. Alaska Sea Grant Report 96-01, Fairbanks, AK. Dorn, M. W. 1995. The effects of age composition and oceanographic conditions on the annual migration of Pacific whiting, Merluccius productus. CalCOFI Rep. 36:97-105. Dwyer, D. A., K. M. Bailey, and P. A. Livingston. 1987. Feeding habits and daily ration of walleye pollock {Theragra chalcogramma) in the eastern Bering Sea, with special reference to cannibalism. Can. J. Fish. Aquat. Sci. 44:1972-1984. Efimkin, A. Y., and V. I. Radchenko. 1991. State of food resources and distribution of epipe- lagic fish in the western Bering Sea in the fall. Sov. J. Mar. Biol. (English translation from Biologiya Morya) 17(l):18-28. Fadeyev, N. S. 1989. Spatial and temporal variability of the eastern Bering Sea walleye pollock size composition in relation to its migrations. /;; Proceedings of the international symposium on the biology and management of wall- eye pollock; November 14-16, 1988, Lowell Wakefield Fisheries Symposium, p. 325-347. Alaska Sea Grant Report, 89-1, Anchorage, AK. Ferno, A., T. J. Pitcher, W. Melle, L. Nottestad, S. Mackinson, C. Hollingworth, and O. A. Misund. 1998. The challenge of the herring in the Norwegian Sea: making optimal collective spatial decisions. Sarsia 83:149-167. Francis, R. C, and K. M. Bailey. 1983. Factors affecting recruitment of selected gadoids in the northeast Pacific and east Bering Sea. In From year to year. (W. S. Wooster, ed.), p. 35-60. Washington Sea Grant WSG-No.83-3, Univ. Washington, Seattle, WA. Harden Jones, F. R. 1968. Fish migration. St. Martin's Press, New York, NY. Hinckley, S. 1987. The reproductive biology of walleye pollock, Ther- agra chalcogramma, in the Bering Sea, with reference to spawning stock structure. Fish. Bull. 85:481-498. Khen, G. V., A. I. Kafanov, and E. O. Basyuk. 2001. Chapter 7: General characteristics. In Hydrome- teorology and hydrochemistry of the seas, vol 10: Bering Sea (Terzyev, F. S., B. M. Zatuchnoy, and A. P. Alexeyev, eds.), p. 122-136. St. Petersburg Gidrometeoizdat.St. Petersburg, Russia. [In Russian.] 586 Fishery Bulletin 103(4) Kjesbu, O. S. 1994. Time of start of spawning in Atlantic cod (Gadus morhua) females in relation to vitellogenic oocyte diam- eter, temperature, fish length and condition. J. Fish Biol. 45(51:719-735. Kooka, K., T. Takatsu, Y. Kamei, T. Nakatani, and T. Takahashi. 1998. Vertical distribution and prey of walleye pollock in the northern Japan Sea. Fish. Sci. 64(51:686-693. Lange. U., and W. Greve. 1997. Does temperature influence the spawning time, recruitment, and distribution of flatfish via its influence on the rate of gonadal maturation? Dtsch. Hydrogr. Z. 49(2/3):251-263. Maeda, T. 1986. Life cycle and behavior of adult pollock {Theragra chalcogramma (Pallas)) in waters adjacent to Funka Bay, Hokkaido Island. Symposium on biology, stock assessment, and management of pollock. Pacific cod and hake in the North Pacific region. INPFC Bull. 45:39-65. Maeda, T., and H. Hirakawa. 1977. Spawning grounds and distribution pattern of the Alaska pollock in the eastern Bering Sea. Bull. Jap. Soc. Sci. Fish. 43:39-45. |In Japanese, English summary.] Maeda. T., T. Nakatani, T. Takahashi, S. Takagi, and B. Melteff. 1989. Distribution and migration of adult walleye pollock off Hiyama, Southwestern Hokkaido. In Proceedings of the international symposium on the biology and manage- ment of walleye pollock; November 14-16, 1988, Lowell Wakefield Fisheries Symposium, p. 325-347. Alaska Sea Grant Report, 89-1, Anchorage, AK. Maeda, T, T. Takahashi, and T Nakatani, 1988. Distribution, migration and spawning grounds of adult walleye pollock Theragra chalcogramma in the coastal waters of Hiyama Subprefecture, Hokkaido. Bull. Fac. Fish. Hokkaido Univ. 39(41:216-229. McAllister, M. K. 1998. Modeling the effects of fish migration on bias and variance in area swept estimates of biomass: a vector based approach. Can. J. Fish. Aquat. Sci. 55:2622- 2641. Mountain, D. G., and S. A. Murawski. 1992. Variation in the distribution of fish stocks on the northeast continental shelf in relation to their environment, 1980-1989. ICES Mar. Sci. Symp. 195:424-432. Neter, J.. M. H. Kutner, C. J. Nachtsheim, and W. Wasserman. 1996. Applied linear regression models, 3rd ed., 720 p. Irwin. Chicago, IL. Nottestad, L., J. Giske, J. C. Hoist, and G. Huse. 1999. A length-based hypothesis for feeding migra- tions in pelagic fish. Can. J. Fish. Aquat. Sci. 56(suppl. 1:26-34. Novikov, Y. V. 1986. Some trends of distribution and migration of common pelagic fishes of the northwestern Pacific Ocean. J. Ichthyol. 26(21:161-172. Overland, J. E., S. A. Salo, H. K. Lakshmi, and C. A. Clayson. 1999. Thermal stratification and mixing on the Bering Sea shelf. In Dynamics of the Bering Sea (T. R. Lough- lin and K. Ohtani, eds.), p. 129-146. Univ. Alaska Sea Grant, AK-SG-99-03, Fairbanks, AK. Pavlov, V. K., and P. V. Pavlov. 1996. Oceanographic description of the Bering Sea. In Ecology of the Bering Sea: a review of Russian literature. (O. A. Mathisen, and K. O. Coyle eds.), p. 135-141. Alaska Sea Grant College Program Report 96-01, Fairbanks, AK. Pola. N. B. 1985. Numerical simulations of fish migrations in the eastern Bering Sea. Ecol. Model. 29:327-351 . Radchenko, V I., and Y. I. Sobolevskiy. 1993. Seasonal spatial distribution dynamics of walleye pollock in the Bering Sea. J. Ichthyol. 33(21:63-76. Ryer, C. H., A. Lawton, R. J. Lopez, and B. L. Olla. 2002. A comparison of the functional ecology of visual vs. nonvisual foraging in two planktivorous marine fishes. Can. J. Fish. Aquat. Sci. 59(81:1305-1314. Ryer, C. H.. and B. L. Olla. 1999. Light-induced changes in the prey consumption and behavior of two juvenile planktivorous fish. Mar. Ecol. Prog. Ser. 181:41-51. Shuntov, V. P., E. P. Dulepova, K. M. Gorbatenko, A. M. Slabin- skii, and A. Y. Efimkin. 2000. Feeding of the walleye pollock Theragra chalco- gramma in the Anadyr-Navarin region of the Bering Sea. J. Ichthyol. 40(3):362-369. Shuntov, V P., A. F. Volkov, O. S. Temnykh, and E. P. Dupoleva. 1993. Pollock in the ecosystems of the Far East seas. 425 p. TINRO, Vladivostok. Russia. [In Russian.] Shuntov, V P. 1992. Functional structure of the distribution area of walleye pollock in the Bering Sea. Russ. J. Mar. Biol./ Biol. Morya 17(41:189-198. [A translation from Biol. Morya, 1991.1 Shuntov, V. P., A. F Volkov, V I. Matveev, L. V Cheblukova, and A. V Gudz. 1987. Development of productive zones in the Sea of Okhotsk during the autumn. Russ. J. Mar. Biol. /Biol. Morya 12(41:241-247. [A translation from Biol. Morya 1986.] Springer. A. M.. C. P. McRoy. and K. R. Turco. 1989. The paradox of pelagic food webs in the northern Bering Sea — II. Zooplankton communities. Cont. Shelf Res. 9(4):359-386. Stabeno. P. J., and J. E. Overland. 2001. Bering Sea shifts toward an earlier spring trans- ition. Eos. Trans. Am. Geophys. Union 82(291:317- 321. Stabeno. P. J., N. A. Bond, N. B. Kachel, S. A. Salo, and J. D. Schumacher. 2001. On the temporal variability of physical environment over the south-eastern Bering Sea. Fish. Oceanogr. 10(11:81-98. Stepanenko, M. A. 2001. Relationship between age and spatial distribution of walleye pollock in the eastern and northwestern Bering Sea. Izv. TINRO, 128:125-135. [In Russian.] Swartzman, G., E. Silverman, and N. Williamson. 1995. Relating trends in walleye pollock i Theragra chalco- gramma) abundance in the Bering Sea to environmental factors. Can. J. Fish. Aquat. Sci. 52(21:369-380. Swartzman, G., W. Stuetzle, K. Kulman. and M. Powojowski 1994. Relating the distribution of pollock schools in the Bering Sea to environmental factors. ICES J. Mar. Sci. 51(4):481-492. Takahashi, Y., and H. Yamaguchi. 1972. Stock of Alaska pollock in the eastern Bering Kotwicki et al Variation in the distribution of Themgra cholcogrammo 587 Sea. Bull. Jap. Soc. Sci. Fish. 38:389-399. [In Japa- nese, English summary. I Traynor, J. J. 1996. Target strength measurements of walleye pollock iTheragra chalcograma) and Pacific whiting [Merluccius productus). ICES J. Mar. Sci. 53:253-258. Wakabayashi, K.. R. G. Bakkala. and M. S. Alton. 1985. Methods of the US-Japan demersal trawl surveys. INPFC Bull. 44:7-29. Wielgolaski, F. E. 1990. A Barents Sea fish resources and migration model. FiskDir. Skr. Ser. HavUnders. 18:381-409. Willette, T. M., T. Cooney, and K. Hyer. 1999. Predator foraging mode shifts affecting mortality of juvenile fishes during the subarctic spring bloom. Can. J. Fish. Aquat. Sci. 56:364-376. Yoshida, H. 1985. Research on clarification of Alaska pollock resources in the Bering Sea and waters around the Kamchatka Peninsula. Can. Transl. Fish. Aquat. Sci. 5140:1-39. 1994. Food and feeding habits of pelagic walleye pollock in the central Bering Sea in summer, 1976-1980. Sci. Rep. Hokkaido Fish. Exp. Stn. 45:1-35. 588 Abstract — The identification of larval istiophorid billfishes from the western North Atlantic Ocean has long been problematic. In the present study, a molecular technique was used to posi- tively identify 27 larval white marlin (Tetrapturus albidus), 96 larval blue marlin (Makaira nigricans), and 591 larval sailfish (Istiophorus platyp- terus) from the Straits of Florida and the Bahamas. Nine morphometric measurements were taken for a subset of larvae (species known), and lower jaw pigment patterns were recorded on a grid. Canonical variates analysis (CVA) was used to reveal the extent to which the combination of morpho- metric, pigment pattern, and month of capture information was diagnos- tic to species level. Linear regression revealed species-specific relationships between the ratio of snout length to eye orbit diameter and standard length (SL). Confidence limits about these relationships served as defining characters for sailfish >10 mm SL and for blue and white marlin >17 mm SL. Pigment pattern analysis indicated that 40% of the preflexion blue marlin examined possessed a characteristic lower jaw pigment pattern and that 62% of sailfish larvae were identi- fiable by lower jaw pigments alone. An identification key was constructed based on pigment patterns, month of capture, and relationships between SL and the ratio of snout length to eye orbit diameter. The key yielded identifications for 69.4% of 304 (blind sample) larvae used to test it; only one of these identifications was incor- rect. Of the 93 larvae that could not be identified by the key, 71 (76.3%) were correctly identified with CVA. Although identification of certain larval specimens may always require molecular techniques, it is encour- aging that the majority (92.4%) of istiophorid larvae examined were ultimately identifiable from external characteristics alone. Toward identification of larval sailfish (Istiophorus platypterus), white marlin (Tetrapturus albidus), and blue marlin (Makaira nigricans) in the western North Atlantic Ocean* Stacy A. Luthy Robert K. Covwen Rosenstiel School of Marine and Atmospheric Science University of Miami 4600 Rickenbacker Causeway Miami, Florida 33149 Present address (for S. A. Luthy): Baruch Marine Field Laboratory PO. Box 1630 Georgetown, South Carolina 29442 Email address (for S A. Luthy). stacy@belle.banjch.se edu Joseph E. Serafy National Marine Fisheries Service Southeast Fisheries Science Center 75 Virginia Beach Drive Miami, Florida 33149 Jan R. McDowell The Virginia Institute of Marine Science School of Marine Science College of William and Mary PO Box 1346 Gloucester Point, Virginia 23062 Manuscript submitted 14 July 2004 to the Scientific Editor's Office. Manuscript approved for publication 6 April 2005 by the Scientific Editor. Fish. Bull. 103:588-600 (2005). Research on the early life history of exploited fishes benefits management efforts by elucidating the temporal and spatial distribution of spawning, cohort strength, and biological and physical factors affecting recruitment (Lasker, 1987). The ability to confi- dently identify specimens to species is necessary in any early life history study (Collette and Vecchione, 1995). This has not yet been achieved for larval billfishes of the family Istio- phoridae from the Atlantic Ocean: sailfish (Istiophorus platypterus), blue marlin (Makaira nigricans), white marlin (Tetrapturus albidus), and longbill spearfish (Tetrapturus pfluegeri). Larval istiophorids are easily dis- tinguished from larval swordfish (Xiphias gladius, family Xiphiidae). However, larval istiophorids are dif- ficult to identify below the family lev- el. Full fin-ray complements are not present until a larva reaches 20 mm in length, and even then, meristic counts are of limited use for identifi- cation because of significant overlap in counts among species. At best, spe- cies possibilities can be eliminated only for specimens with counts in the extremes of their ranges (Richards, 1974). The only definitively diagnos- tic count is the vertebral formula for Makaira (11 precaudal and 13 caudal) versus that of the other istiophorids (12 precaudal and 12 caudal) (Rich- ards, 1974). Larger blue marlin lar- * Contribution SFD-2003-0010 from NOAA Fisheries Sustainable Fisheries Division, Southeast Fisheries Science Center, 75 Virginia Beach Drive, Miami, Florida 33149. Luthy et al. Identification of larval sailfish, white marlm, and blue marlin in the western North Atlantic Ocean 589 vae may also be identified by the presence of a complex lateral line. Ueyanagi (1964) found this character in Pacific blue marlin of 20 mm standard length (SL), but the smallest SL of an Atlantic blue marlin from a recent collection in which a complex lateral line was visible was 26.9 mm. At lengths <20 mm, specific identifica- tion of istiophorids is even more uncertain. Ueyanagi (1963; 1964) based the identification of Indo-Pacific istiophorids <5 mm SL on four characters: 1) anterior projection of the eye orbit; 2) the position of the tip of the snout in relation to the middle of the eye; 3) pres- ence of pigments on the branchiostegal and gular mem- branes; and 4) whether the pectoral fins are rigid — a character that applies to larval black marlin tMakaira indica), a species not known to spawn in the Atlantic Ocean. For fish >5 mm SL, the characters of relative snout length and eye size are used. Ueyanagi (1964) described sailfish, striped marlin (Tetrapturus audax, the Pacific counterpart to white marlin), and shortbill spearfish {Tetrapturus angustirostris) between 10 and 20 mm SL as having long snouts. The short snout group comprised blue marlin and black marlin. The angles at which the pterotic and preopercular spines protrude from the body have also been useful in identifying Indo- Pacific specimens (Ueyanagi, 1974a). A troubling aspect of current larval istiophorid iden- tification methods is the difficulty in using some of the above characters. If a specimen is fixed with its mouth open, snout position with respect to eye is an unread- able character (Richards, 1974), and misidentifications can occur (Ueyanagi, 1974a). Evaluation of certain char- acters (e.g., whether the eye orbit projects anteriorly) can be highly subjective. The lack of confirming identi- fication characters compounds the problem; if just one character cannot be assessed, identification may not be possible (Richards, 1974). An additional problem is the apparently high variability in characters such as pigment locations and head spine angles in Atlantic istiophorids (Richards, 1974). Most of the larval specimens examined by Ueyanagi came from the Indo-Pacific; he assumed that the same identification characters would apply to their Atlantic counterparts (Ueyanagi, 1963, 1974a). Although recent genetic evidence supports Morrow and Harbo's (1969) opinion that Atlantic and Indo-Pacific sailfish are actu- ally populations of a global species (Finnerty and Block, 1995; Graves and McDowell, 1995), morphological dif- ferences have been noted in sailfish, especially at 90 cm. Specifically, the pectoral fin is longer, in relation to the body, in Atlantic sailfish than in Indo-Pacific sailfish. Differences in the spread of the caudal fin and maximum total length have also been observed. These characters were the impetus behind the separation of sailfish, at least to subspecies, by ocean basin (Naka- mura, 1974). Regardless of the taxonomic status of the Atlantic and Indo-Pacific billfishes, physical attributes of istiophorid species may vary by region. Therefore, the assumption that the larvae of Atlantic istiophorids can be identified by using the same characters attributed to Indo-Pacific istiophorids may not be valid. Billfishes are not the only group whose larval iden- tification has proven difficult. Species of the genus Sebastes, the rockfishes, have some morphological and pigmentation differences as larvae, but identification was difficult and uncertain until genetic methods were employed (Rocha-Olivares et al., 2000). Fulford and Rutherford (2000) solved a similar problem by combin- ing allozyme analysis of larval tissues with landmark- based morphometries to distinguish between species of the genus Morone. In each study, a molecular technique was used to confirm larval species identity, facilitat- ing the development of morphometric identification techniques. Several molecular methods for identifying adult billfishes have been developed (Chow, 1993; Innes et al., 1998; McDowell and Graves, 2002). In the present study, larval istiophorids from Atlantic waters were identified to species using restriction fragment length polymorphism (RFLP) analysis of a 1.2-kb segment of nuclear DNA, as described for adult billfishes by McDowell and Graves (2002). In this article we pres- ent data for genetically identified istiophorid larvae, analyses of morphometric and qualitative characters, and a key for the identification of larval istiophorids of the Straits of Florida and the Bahamas. Materials and methods Larval material Larval istiophorids were collected between June 1998 and April 2002 from the Straits of Florida and Exuma Sound, Bahamas. Several preservation fluids were used, but the majority of the larvae (-1000) were preserved in 70-95% ethanol. Butylated hydroxytoluene (BHT) saturated ethanol was used to preserve 150 larvae. Approximately 300 larvae were fixed in 10% unbuffered formalin and then transferred to 70% ethanol. In the laboratory, each fish was assigned a unique identification number and stored separately. Molecular identification Total DNA was extracted from the right eyeball of each larva, using either a quick-digest method (Ruzzante et al., 1996) or a standard high-molecular weight DNA extraction protocol (Sambrook et al., 1989). Larval identification was achieved by PCR amplification of the nuclear locus MN32-2 (Buonaccorsi et al., 1999), and subsequent RFLP analysis (restriction endonucle- ases Dra I and Dde I, Life Technologies, Bethesda, MD). If the restriction fragment pattern (Fig. 1) of a larva matched one of those described for a known-iden- tity adult, the larva was assigned to that species. See McDowell and Graves (2002) for detailed protocols and reaction parameters. Preliminary attempts to amplify DNA from formalin-fixed larvae failed; only ethanol- preserved specimens were used in subsequent molecular work. 590 Fishery Bulletin 103(4) 1Kb Plus DNA ladder sailfish white marlin blue marlin blue marlin 0) ■o ■o < a tn CL -O ra c as E a> S c TO E CD . **N* tea 12,000— 12,000— : S-;:;^ ^:> • ..• ■■■* 2000 — - 2000 — 1000— 850— 1000 — ■ 850— 650— 650— 500— 500— 400— 400— 300—: 300— 200— 200— Dde\ Oral Figure 1 Common Dele I and Dra I restriction patterns for the MN.32-2 1 )CUS rf positively identified larval istiophorids from the Straits of Flori da an d the Bah amas. The left lane of each gel contains a DNA size standa •d (Life Te chno ogies, Bethesda, MD), measured in base pairs. Characters A subset of the molecularly identified istiophorid larvae were examined to ascertain which morphological char- acters might aid in specific identification and possibly obviate the need for future molecular work. The measure- ments made by Richards ( 1974) served as a starting point for quantitative larval descriptions: standard length (SL); snout length (SN); tip of the snout to the center of the eyeball (SN-E); diameter of the eye (ED); diameter of the eye orbit (OD); head length (HL); and difference in length between the upper and lower jaws (JD). To this suite were added measurements of the preopercular (PRO) and pterotic (PTS) head spines. All measurements were taken with Image-Pro Plus software (version 4.5, Media Cybernetics, Silver Spring, MD), and each specimen was viewed through a CoolSNAP-PROcf monochrome digital camera (Media Cybernetics, Silver Spring, MD) which was connected to a Leica MZ12 dissecting microscope (at magnifications 0.8-10. Ox). Each larva was soaked in tap water for one minute before measurements were taken, to rehydrate the fish and facilitate handling. SL and PRO measurements were made from the dorsal view, JD measurements were made from the ventral view, and all other measurements were made from the left lateral view (Fig. 2). Because the preopercular spine often prevents an istiophorid larva from lying on its side, a side view was obtained by using the surface tension of the still-wet larva to adhere it to the side wall of a Petri dish. Care was taken to maintain the two points of measurement on a plane parallel to the microscope lens. Pigments observed on the ventral surface of the lower jaw rami, gular membrane, and branchiostegal mem- branes of each larva were drawn onto a generalized dia- gram of the larval istiophorid lower jaw (Fig. 3). A grid was then superimposed on the diagram, and the shape (pointate or stellate) and number of chromatophores in each grid cell were recorded. Pigment data were also recorded as binary presence or absence per grid cell. Two other categorical variables assessed were flexion stage (i.e., preflexion, flexing, postflexion) and the posi- tion of the tip of the snout with regard to a plane passing through the center of the eye and the mid-line of the body (i.e., below, even, above). Although the latter character is useful for identifying Indo-Pacific istiophorids (Ueyanagi, 1963. 1964), in our collection it was highly variable with- in species, and therefore it was not analyzed further. Month of capture was considered a partially discrimi- nating character based on differences in the length and timing of spawning seasons of local populations. Spawn- ing seasons were determined by de Sylva and Breder (1997) by gonad histology studies. Luthy et al : Identification of larval sailfish, white marlin, and blue marlm in the western North Atlantic Ocean 591 Figure 2 Morphometric measurements illustrated on a 10.7-mm SL sailfish. SN = snout length; SN-E = snout to mid-eye; OD = eye orbit diameter; ED = eye diameter; PTS = length of pterotic spine; PRO = length of preopercular spine. Drawings by S. Luthy. Gular membrane Branchiostegal membrane Figure 3 Lower jaw pigments were characterized by drawing chromatophores onto a generalized lower jaw diagram (A), reproduced from Richards (1974). A grid (B) was then superimposed onto the diagram and the number and shape of chromatophores were recorded for each grid cell. The numbers in diagram B are numbers used to identify the cells of the grid and not the number of chromatophores per cell. 592 Fishery Bulletin 103(4) Data analyses Canonical variates analysis (CVA) was used to visualize the separation between species and the relative impor- tance of all variables (morphometric characters, pigment patterns, and month of capture) in that separation. Results from the CVA were used to help drive charac- ter selection for subsequent analyses. The significance of the canonical axes was obtained with a Monte Carlo permutation test (499 iterations). The canonical analyses were performed with the software CANOCO (version 4.5, Microcomputer Power, Ithaca, NY), and plotted with the associated software CANODRAW. In the CVA, all the molecularly identified white mar- lin (21) and blue marlin (68) with full measurement sets (i.e., no missing values) and a subset of sailfish (135) with full measurement sets were compared. Every at- tempt was made to include fish from different locations, different years and months of collection, and across the full available size range of each species, in order to capture as much intra- and inter-species variation as possible. Forward selection was used as a guide for the creation of a reduced set of variables by retaining those that were significant for discrimination at oc=0.05 in a Monte Carlo permutation test (499 iterations). Months that were excluded by selection were restored to the variable set to insure that the entire spawning season was represented. It was assumed that pigment on the right lower jaw ramus was of equal importance as pig- ment in the corresponding location on the left lower jaw ramus; thus if a pigment grid from only one side of the jaw was selected, the corresponding grid from the other side of the jaw was added back to the reduced set. In addition to its function as an exploratory tool for character selection, CVA with the reduced set of vari- ables was used to identify unknown larvae to species. Ordination coordinates of an unknown larva were ob- tained by summing the products of the canonical coef- ficients and the character values for the unknown (stan- dardized to mean 0, standard deviation 1). The identity of an unknown larva was determined by its placement in the ordination with respect to the reference larvae. The CVA provided clues as to which individual pig- ment grid cells were important for species discrimina- tion, but cluster analysis was employed to examine overall lower jaw pigment patterns. Simple average link cluster analysis of Jaccard similarity indices was executed on pigment grid cell presence (binary coding) in the suite of lower jaw grid cells with BioDiversity Pro1 software for the 26 white marlin with undamaged lower jaws and for equal numbers of randomly chosen blue marlin and sailfish. Analyses were conducted on all larvae together, and separately by flexion stage. Pig- ment drawings of the individual larvae within single- 1 McAleece, N., P. J. D. Lambshead, G. L. J. Paterson, and J. D. Gage. 1997. The National History Museum and The Scottish Association for Marine Science. Website: http:// www.sams.ac.uk/. [Accessed 5 February 2003.] species clusters were examined visually for commonali- ties. If a pattern was detected, the entire database of pigment position, number, and shape of all molecularly identified larvae was searched for that pattern. Lower jaw pigment patterns that were confined to one species only were deemed diagnostic characters. Lower jaw pigment patterns alone did not resolve the differences among the species sufficiently for identifica- tion of all larvae. Therefore, for each species, continuous variables related linearly to SL were regressed against SL by using SAS (version 8.02, SAS Institute, Cary, NO software. Two ratios were also examined in this way — snout length divided by eye orbit diameter, and snout length divided by eye diameter. Both ratios were suggested by the results of the full-model CVA because the influence of snout length was large and opposite in sign to the large and similar vectors of orbit diameter and eye diameter. The former ratio was also considered by Ueyanagi (1963, 1964, 1974b) to be an important distinguishing character for istiophorid larvae. The same larvae that were used in the CVA analyses were used for the regressions, plus three white marlin, two sailfish, and two blue marlin that were excluded from CVA because of a missing measurement. Suitability of the characters for linear regression was assessed visually. Confidence intervals of 95%, 99%, and 99.9% were constructed around the regressions. Intersections of the three levels of confidence intervals for the three species were examined for maximum discrimination at the smallest standard length. The relationships that provided the best separation were included in the iden- tification key. The identification key was constructed from the vari- ous characters that showed differences among the three species. All of the larvae used in developing the key were tested with it, as well as 12 blue marlin and 61 sailfish that were previously excluded from the analy- ses. A set of 50 larvae were independently identified by two observers unfamiliar with the key (naive observ- ers). The only information about the fish provided to them was month of capture, so that each made his own measurements and pigment evaluations. The percent accuracy of their identifications was taken as a measure of the utility of the key. Results Molecular identification The molecular identification technique was applied to 1044 larvae. Amplification success rates appear to have been negatively affected by the addition of BHT to etha- nol and by the use of the Ruzzante et al. (1996) DNA extraction protocol. Overall, 714 (68.4%) istiophorids were successfully identified to the species level. Sailfish represented 82.8%- of this group (591 larvae), whereas 96 blue marlin (13.4%) and 27 white marlin (3.8%) were identified. No longbill spearfish were identified. Sailfish larvae (2.9 mm-18.3 mm SL) were collected from April Luthy el al Identification of larval sailfish, white marhn, and blue marhn in the western North Atlantic Ocean 593 T CONTINUOUS VARIABLES ■ rna^ch o NOMINAL VARIABLES A SAMPLES BLUE MARLIN 0 SAILFISH °o oapril a SN * ^ JD WHITE MARLIN ED V V o ©o o // A' may & -4 Afltit-1 ag ■ A pseS V w V A j sept V PRO -6 Axis 1 6 Figure 4 Canonical variates analysis with the reduced set of variables. Arrows indicate the direction of increase in continuous variables and may be extended backward through the origin of the graph to show a decrease in the value of the character. Variables that extend far- thest from the origin are most useful in the separation. SN = snout length; JD = difference in the lengths of the jaws; ED = eye diameter; PRO = length of preopercular spine; p (number i = presence of pig- ment in lower jaw grid cell (number). through September, white marlin (4.5 mm-20.3 mm SLi were collected from March through June, and larval blue marlin (3.8 mm-22.1 mm SL) were collected from June through September. Month of capture closely matched the reported spawning seasons for these species in the west- ern North Atlantic: April through October for sailfish, March through June for white marlin, and July through October for blue marlin (de Sylva and Breder, 1997). Because blue marlin larvae were also caught in June, the blue marlin spawning season was expanded to include that month for the purposes of the identification key. Canonical variates analysis In the CVA with all variables included, separation of the three species was achieved with little overlap. Sailfish larvae were separated from the marlins along canoni- cal axis 1 (eigenvalue = 5.45). The separation was driven mainly by ED, OD, and lower jaw pigmentation. White marlin larvae separated from blue marlin primarily along canonical axis 2 (eigenvalue = 0.79), largely by month of capture, as well as SN, SN-E, and JD . The overall ordination was significant at P=0.002. The forward selection process, along with the re-addi- tion of counterpart pigment grids and the full spawning season, yielded the following 21 out of 32 variables: March, April, May, June, July, August, September, SN, JD, ED, PRO, and pigment grids 1-4, 6-9, 11, and 12. The following variables were ultimately excluded from the data set: SL, SN-E, OD, HL, PTS, and pigment grids 5, 10, and 13-16. The degree of species overlap was similar to that in the full model (Fig. 4). This overall ordination was also significant at P=0.002. The eigenvalue of the first canonical axis was 4.71, whereas the eigenvalue of the second canonical axis was 0.71. Coordinates obtained from the canonical coefficients and character values, standardized by reference set character means and standard deviations (Table 1), accurately placed test "unknowns" in the ordination of the reference larvae. 594 Fishery Bulletin 103(4) 2.5 2- =5 1 5 05 D O 0 sailfish 99% CI sailfish individuals white marlin 99% CI white marlin individuals blue marlin 99% CI blue marlin individuals O v aifi i aft'' -cf' 1 1 „"' cP -" □ ° ,,* ^^^ ,' s 0 0 "23 t 0 i i i - 25 75 10 125 15 Standard length (mm) 17.5 20 225 Figure 5 Relationship of the ratio of snout length to orbit diameter with standard length. Lines represent 99% confidence intervals. Lower jaw pigment patterns Sailfish of all flexion stages with chromatophores on one or both sides of the lower jaw rami and sometimes in the middle of the gular membrane comprised single- species clusters. Examination of all molecularly identi- fied larvae showed that many sailfish had pigment on the posterior % of the lower jaw, but a few marlins also had stray pigments in that region. The minimum crite- rion to identify sailfish by lower jaw pigment without misidentifying other species was pigment in at least three of lower jaw pigment grids 1, 2, 3, 7, 8, 9, and 11. The shape and number of chromatophores within the grids was inconsequential. Not all sailfish larvae possessed the putative sailfish pattern, but 61.8% of molecularly identified sailfish (353 of 571 with intact lower jaws) could be identified by their lower jaw pig- ments alone. Preflexion and flexing blue marlin also formed single- species clusters owing to the pattern of a single, pointate chromatophore in each of lower jaw grid cells 4 and 6, but without any other pigment (except occasionally in grid cell 12 or 13). However, not all small blue marlin exhibited this pattern. Eight of the 20 (40%) preflexion, molecularly identified blue marlin with intact lower jaws could be ac- curately identified by lower jaw pigments. Although some postflexion white marlin had a similar pattern, no preflex- ion or flexing larvae of other species were misidentified as blue marlin by virtue of this pigment pattern. Linear regressions Residual plots showed no deviations from homogeneity of variance. Snout length, snout to mid-eye, ratio of snout length to eye diameter, and ratio of snout length to orbit diameter were all linearly related to SL. Jaw difference was linear and appeared to be helpful for dis- criminating istiophorids >12 mm SL, but too few larvae of this size were available for meaningful regressions. The ratio of snout length to orbit diameter provided the most separation between the species as indicated by the full model CVA. The 99% upper limit of the regression of this ratio against SL for white marlin was used to separate sailfish from both marlin species at 10 mm SL. If white marlin is ruled out as a possibility by month of capture, sailfish can be separated from blue marlin by the blue marlin upper 99% confidence limit for the regression of the ratio of snout length to orbit diameter at 8 mm SL. The lower 99% confidence limit for the regression of the ratio of white marlin snout length to orbit diameter separated them from blue marlin at 17 mm SL (Fig. 5, Table 2). Luthy et al : Identification of larval sallfish, white marhn, and blue marlin in the western North Atlantic Ocean 595 Table 1 Canonical coefficients, mean, and standard deviation of each character from the canonical variates analysis (reduced set of char- acters). The coordinate of a larva on canonical axis 1 (x can be found by x=%c,jZ, , where c = canonical coefficient and z = Ichar- acter va lue- character mean (/character standard deviation. The coordinate 'of a larva on canonical axis 2 ' vi can be found by -V=X0.030SL + 0.551 Istiophorus platypterus 6b Snout length /orbit diameter s0.030SL + 0.551 Makaira nigricans Part II: for larvae >10 mm SL la Chromatophores of any number or shape in 3 or more of lower jaw pigment grids 1, 2, 3, 7, 8, 9, 11 Istiophorus platypterus lb Without the above lower jaw pigment pattern 2 2a Snout length / orbit diameter >0.057SL + 0.427 Istiophorus platypterus 2b Snout length / orbit diameter s0.057SL + 0.427 3 3a Larva caught in March, April, or May Tetrapturus albidus 3b Larva caught in June or later 4 4a Larva caught in July, August, September, or October Makaira nigricans 4b Larva caught in June 5 5a Standard length >17 mm 6 5b Standard length <17 mm either Makaira nigricans or Tetrapturus albidus 6a Snout length / orbit diameter >0.047SL + 0.319 Tetrapturus albidus 6b Snout length / orbit diameter <0.047SL + 0.319 Makaira nigricans a measurement was missing. Thus, when the key and CVA analyses were combined, 92.4% of the tested larvae were correctly identified. One of the two naive observers found that one larva out of the test set of 50 was too damaged to be evalu- ated. He correctly identified 35 larvae and found 14 to be unidentifiable with the key. Overall, his success rate was 71.4%. The other observer correctly identified 30 larvae, misidentified one (the larva not evaluated by the other observer and the same larva misidentified by the authors), and found 19 to be unidentifiable by the key. His overall success rate was 60%. The difference in the number of larvae that could not be identified with the key was the result of differences in interpretation of the lower jaw pigment position for larvae less than 10 mm SL. Discussion Because adults of four istiophorid species are found in the Straits of Florida and Bahamian waters, a reliable larval identification technique for these species is neces- sary (Voss, 1953). Incorrect species identifications can have serious ramifications on other areas of istiophorid early life history research. For example, studies on early growth would suffer if a larval blue marlin, which is thought to reach 174 cm lower jaw fork length (LJFL) by age one (Prince et al., 1991), were to be confused with a larval sailfish, which reportedly grows to only 108.9 cm LJFL (Hedgepeth and Jolley, 1983; Prager et al., 1995) by age one. Few characters are available to separate the spe- cies of larval istiophorids (Richards, 1974). Although Luthy et al.: Identification of larval sallfish, white marlin, and blue marlin in the western North Atlantic Ocean 597 a single character may be used to separate fish into groups, early work has lacked a means to confirm the identity of the groups. Molecular techniques provided a solution to this problem. A limitation of the molecular identification technique that we used was that only those larvae preserved in ethanol could be identified. Formalin fixation does not always preclude the use of PCR-based methods, but work is usually limited to small fragments; 570 bp is considered large for success- ful amplification (Shedlock et al., 1997). In the present study, DNA quality was too low in the formalin-fixed is- tiophorid larvae for PCR to amplify the 1.2-kb MN32-2. Consequently, only ethanol-preserved larvae could be used for key development and testing. Because of likely differences in length shrinkage between larvae pre- served only in ethanol and those fixed in formalin, it is possible that the regressions presented in the present study are not valid for the latter. No longbill spearfish were among the molecularly iden- tified larvae; thus this species could not be included in the key. Very little is known about the longbill spearfish, but it is reported that larvae are found offshore (Uey- anagi et al., 1970), and that even adults are quite rare in United States and Bahamian waters (Robins, 1975). The longbill spearfish spawning season appears to range from late November to early May and peaks in Febru- ary (Robins, 1975; de Sylva and Breder, 1997). Although there is some overlap in the spawning season of longbill spearfish with the spawning seasons of other Atlantic istiophorids, because of the rarity and predominantly offshore occurrence of the longbill spearfish, its absence from the key may not pose major problems for the iden- tification of istiophorid larvae from our study area. The larval istiophorids used to create and test the identification key were all captured either in the Straits of Florida or in Bahamian waters and were all smaller than 22 mm SL. Caution must be used when apply- ing the key to larvae from other parts of the world or to larger sizes. Ueyanagi (1963) assumed that spe- cies pairs from different oceans (white marlin and striped marlin [Tetrapturus audax], longbill spearfish and shortbill spearfish [Tetrapturus angustirostris], Atlantic and Pacific blue marlin, Atlantic and Pacific sailfish]) would be identifiable by the same characters. Although these pairs exhibit the same RFLP patterns at the MN32-2 locus (McDowell and Graves, 2002), we have not tested the key with Pacific larvae and cannot be certain that their measurements would fall within the same regression limits or that they would have the same lower jaw pigment patterns. Even within the Atlantic Ocean, spawning seasons vary with location (e.g., Bartlett and Haedrich [1968] collected larval blue marlin off the coast of Brazil in February and March). Month of capture was crucial in our analyses for dis- criminating between small marlins when spawning season overlap is minimal; therefore our key may need adjustment to reflect local spawning seasons when ap- plied to other locations. As in Indo-Pacific istiophorid larvae (Ueyanagi, 1964, 1974b), snout length, eye orbit diameter, and lower jaw pigmentation are important characters for identifying larval istiophorids of the western Atlantic. However, white marlin differ markedly from their Indo-Pacif- ic counterpart, striped marlin. White marlin larvae, long-held as members of the "long-snout group" of istio- phorids, actually more closely resemble the short-snout- ed blue marlin until 17 mm standard length (Fig. 6). After they reach this size, snout length is intermediate between that of blue marlin and sailfish. This result cautions against the assumption that even large larvae with short snouts are blue marlin. Snout length may be useful as a character in phylogeny studies. The identification methods presented in the present study reduce subjectivity in the evaluation of charac- ters. This study also brings to light the caveats of using lower jaw pigment patterns as a means of identification and limits which pigment patterns qualify as diagnos- tic. Although there is a family of lower jaw pigment patterns that appears to mark sailfish only, if this char- acter were the only means of identifying sailfish, nearly 40% of our sailfish (as confirmed by RFLP analysis) would have been misidentified or escaped classification. Likewise, the preflexion blue marlin pigment pattern will not lead to misidentifications, but too many preflex- ion blue marlin lack the pattern to justify its use as a stand-alone identification character. Lower jaw pigment patterns have also been suggested as potentially useful characters for separation of subspecific populations of both sailfish (Ueyanagi, 1974a, 1974b) and striped mar- lin in the Indo-Pacific (Nishikawa, 1991). The hypoth- esis of pigment-delineated sailfish populations was not borne out (Leis et al., 1987), and the high variability of lower jaw pigments among larvae of each species from our study area casts further doubt on the notion of us- ing pigments alone to distinguish populations. Our identification key does not enable separation of species for certain classes of istiophorid larvae. For example, larvae that are caught in June, are less than 10 mm SL, and possess none of the diagnostic lower jaw pigment patterns are especially problematic. In these "dead end" cases, discriminant analysis (CVA) is useful. Although a few larvae were misidentified with the CVA, these larvae were plotted near the interface of two species groupings; this position alerts the user to the fact that misidentification is a possibility. One dis- advantage of using CVA (or any discriminant analysis) for identification is that all of the variables must have a value, meaning that a larva with broken preopercular spines, for example, cannot be entered into the analysis. When the species possibilities are narrowed down to blue marlin and either sailfish or white marlin, it may be feasible to identify larvae by vertebral formula. Rich- ards (1974) suggests that this is difficult with larvae less than 20 mm SL, but it is the method that Prince et al. (1991) used to identify blue marlin that were 5-10 mm SL. Molecular identification is always an option for resolving dead ends. The identification of larval istiophorids has never been an easy task. Molecular identification is reliable, but can be relatively more labor intensive and expensive 598 Fishery Bulletin 103(4) v. Figure 6 Size series of genetically identified representatives of each species. Top row: sailfish. Middle row: white marlin. Bottom row: blue marlin. Left column: ~5 mm SL. Middle column: -10 mm SL. Right column: -15 mm SL. Luthy et al : Identification of larval sailfish, white marlin, and blue marlin in the western North Atlantic Ocean 599 than traditional methods. The creation of a key based on characters developed from molecularly identified At- lantic larvae makes it possible to use more traditional methods to make reliable identifications. Despite the limitations of the key, it works well for larvae caught in our area. We recommend further testing with istio- phorid larvae from other waters, and the inclusion of longbill spearfish larvae. Acknowledgments The authors appreciate financial support provided by Network Miami and Anheuser Busch, the American Institute of Marine Science, the Miami Billfish Tour- nament's Captain H. Vernon Jr. Scholarship, the Inter- national Light Tackle Tournament Association, and the University of Miami's Center for Sustainable Fisheries. We thank G. Diaz, K. Gracie, L. Leist, M. Williams, O. Bowen, C. Schmitz, C. Faunce, D. Schuller, G. Meyers, and M. Feeley for volunteering their time for specimen collection and J. Post, T. Capo, J. Ault, S. Smith, and J. Luo for early instruction. Laboratory advice and com- miseration were provided by C. Campbell. P. Walsh, J. van Wye, and all the members of the VIMS genetics laboratory. We offer special thanks to J. Graves, in whose laboratory the molecular work was carried out. We are grateful to T Grothues for sharing his CVA wisdom and to J. Llopiz, D. Richardson, and K. Denit for testing our key. W. Richards, D. deSylva, and C. Paris were instrumental in the interpretation of identification characters. This work could not have been carried out without the generosity and enthusiasm of D. Frazel and his family in donating their time and the use of their boat. Larvae were collected under NMFS permits HMS- EFP-00 through 03, and under University of Miami animal care protocols (02-063). Literature cited Bartlett, M. R., and R. L. Haedrich. 1968. Neuston nets and South Atlantic larval blue marlin {Makaira nigricans). Copeia 1968:469-474. Buonaccorsi, V. P., K. S. Reece, L. W. Morgan, and J. E. Graves. 1999. Geographic distribution of molecular variance within the blue marlin (Makaira nigricans): a hierarchi- cal analysis of allozyme, single-copy nuclear DNA, and mitochondrial DNA markers. Evolution 53:568-579. Chow, S. 1993. Identification of billfish species using mitochon- drial cytochrome b gene fragment amplified by poly- merase chain reaction. Collect. Vol. Sci. Pap. ICCAT 41:549-556. Collette, B. B„ and M. Vecchione. 1995. Interactions between fisheries and systematics. Fisheries 20:20-25. de Sylva, D. P., and P. R. Breder. 1997. Reproduction, gonad histology, and spawning cycles of North Atlantic billfishes (Istiophoridae). Bull. Mar. Sci. 60:668-697. Finnerty, J. R., and B. A. Block. 1995. Evolution of cytochrome b in the Scombroidei iTele- ostei) — molecular insights into billfish (Istiophoridae and Xiphiidae) relationships. Fish. Bull. 93:78-96. Fulford, R. S.. and D. A. Rutherford. 2000. Discrimination of larval Morone geometric shape differences with landmark-based morphometries. Co- peia 2000:965-972. Graves, J. E„ and J. R. McDowell. 1995. Inter-ocean genetic divergence of istiophorid billfishes. Mar. Biol. 122:193-203. Hedgepeth, M. Y., and J. W. Jolley. 1983. Age and growth of sailfish, Istiophorus platyp- tents, using cross sections from the fourth dorsal fin spine. In Proceedings of the international workshop on age determination of oceanic pelagic fishes: tunas, billfishes, and sharks (E. D. Prince and L. M. Pulos. eds.), p. 131-135. NOAA Tech. Rep. 8. Innes, B. H., P. M. Grewe, and R. D. Ward. 1998. PCR-based genetic identification of marlin and other billfish. Mar. Freshw. Res. 49:383-388. Lasker, R. 1987. Use offish eggs and larvae in probing some major problems in fisheries and aquaculture. Am. Fish. Soc. Symp. 2:1-16. Leis, J. M., B. Goldman, and S. Ueyanagi. 1987. Distribution and abundance of billfish larvae (Pisces, Istiophoridae) in the Great Barrier Reef Lagoon and Coral Sea near Lizard Island, Australia. Fish. Bull. 85:757-765. McDowell, J. R., and J. E. Graves. 2002. Nuclear and mitochondrial DNA markers for spe- cific identification of istiophorid and xiphiid billfishes. Fish. Bull. 100:537-544. Morrow, J. E., and S. J. Harbo. 1969. A revision of the sailfish genus Istiophorus. Copeia 1969:34-44. Nakamura, I. 1974. Some aspects of the systematics and distribution of billfishes. 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. 45-53. NOAA Tech. Rep. NMFS SSRF-675. Nishikawa, Y. 1991. On the melanophore patterns on lower jaw of the larvae of striped marlin. Bull. Natl. Res. Inst. Far Seas Fish. 28:15-19. Prager, M. H., E. D. Prince, and D. W. Lee. 1995. Empirical length and weight conversion equations for blue marlin, white marlin, and sailfish from the North-Atlantic Ocean. Bull. Mar. Sci. 56:201-210. Prince, E. D., D. W. Lee, J. R. Zweifel, and E. B. Brothers. 1991. Estimating age and growth of young Atlantic blue marlin Makaira nigricans from otolith micro- structure. Fish. Bull. 89:441-459. Richards, W. J. 1974. Evaluation of identification methods for young billfishes. 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. 62-72. NOAA Tech. Rep. NMFS SSRF-675. Robins, C. R. 1975. Synopsis of biological data on the longbill spear- fish, Tetrapturus pfluegeri, Robins and de Sylva. In 600 Fishery Bulletin 103(4) Proceedings of the international billfish symposium; Kailua-Kona, Hawaii, 9-12 August 1972. Part 3: Spe- cies synopsis (R. S. Shomura and F. Williams, eds.), p. 28-37. NOAA Tech. Rep. NMFS SSRF-675. Rocha-Olivares, A., H. G. Moser, and J. Stannard. 2000. Molecular identification and description of pelagic young of the rockfishes Sebastes constellatus and Sebastes ensifer. Fish. Bull. 98:353-363. Ruzzante. D. E., C. T. Taggart, and D. Cook. 1996. Spatial and temporal variation in the genetic com- position of a larval cod (Gadus minima) aggregation: Cohort contribution and genetic stability. Can. J. Fish. Aquat. Sci. 53:2695-2705. Sambrook, J., E. F. Fritsch, and T. Maniatus. 1989. Molecular cloning: a laboratory manual, p. 9.17-9.19. Cold Spring Harbor Laboratory Press, Cold Spring Harbor, NY. Shedlock, A. M„ M. G. Haygood, T. W. Pietsch and P. Bentzen. 1997. Enhanced DNA extraction and PCR amplification of mitochondrial genes from formalin-fixed museum specimens. Biotechniques 22:394-400. Ueyanagi, S. 1963. Methods for identification and discrimination of the larvae of five istiophorid species distributing in the Indo-Pacific. Rep. Nankai Reg. Fish. Res. Lab. 17:137-150. [In Jap.. Engl, sum.] 1964. Description and distribution of larvae of five istio- phorid species in the Indo-Pacific. Mar. Biol. Assoc. India. Proc. Symp. Scombroid Fish. Mandapam Camp, Part 1:499-528. 1974a. On an additional diagnostic character for the identification of billfish larvae with some notes on the variations in pigmentation. In Proceedings of the inter- national billfish symposium, Kailua-Kona, Hawaii, 9-12 August 1972. Part 2: Review and contributed papers (R. S. Shomura and F. Williams, eds.) p. 73-78. NOAA Tech. Rep. NMFS SSRF-675. 1974b. Present state of billfish larval taxonomy. In The early life history offish (J. H. S. Blaxter, ed.), p. 649-658. Springer-Verlag, New York, NY. Ueyanagi, S., S. Kikawa, M. Uto, and Y. Nishikawa. 1970. Distribution, spawning, and relative abundance of billfishes in the Atlantic Ocean. Bull. Far Seas Fish. Res. Lab. 3:15-55. Voss. G. L. 1953. A contribution to the life history and biology of the sailfish. htiophorus americanus Cuv. and Val., in Florida waters. Bull. Mar. Sci. Gulf Carib. 3:206-240. 601 Abstract — This study examined the sexual differentiation and reproduc- tive dynamics of striped mullet iMugil cephalus L.) in the estuaries of South Carolina. A total of 16,464 specimens were captured during the study and his- tological examination of sex and matu- rity was performed on a subsample of 3670 fish. Striped mullet were sexually undifferentiated for the first 12 months, began differentiation at 13 months, and were 90% fully differentiated by 15 to 19 months of age and 225 mm total length (TL). The defining morphologi- cal characteristics for differentiating males was the elongation of the pro- togonial germ tissue in a corradiating pattern towards the center of the lobe. the development of primary and sec- ondary ducts, and the lack of any rec- ognizable ovarian wall structure. The defining female characteristics were the formation of protogonial germ tissue into spherical germ cell nests, separa- tion of a tissue layer from the outer epithelial layer of the lobe-forming ovar- ian walls, a tissue bud growing from the suspensory tissue that helped form the ovary wall, and the proliferation of oogonia and oocytes. Sexual maturation in male striped mullet first occurred at 1 year and 248 mm TL and 100% maturity occurred at age 2 and 300 mm TL. Female striped mullet first matured at 2 years and 291 mm total length and 100% maturity occurred at 400 mm TL and age 4. Because of the open ocean spawning behavior of striped mullet, all stages of maturity were observed in males and females except for functionally mature females with hydrated oocytes. The spawning season for striped mullet recruiting to South Carolina estuaries lasts from October to April; the majority of spawning activ- ity, however, occurs from November to January. Ovarian atresia was observed to have four distinct phases. This study presents morphological analysis of reproductive ontogeny in relation to size and age in South Carolina striped mullet. Because of the length of the undifferentiated gonad stage in juve- nile striped mullet, previous studies have proposed the possibility of pro- tandric hermaphrodism in this species. The results of our study indicate that striped mullet are gonochoristic but capable of exhibiting nonfunctional hermaphroditic characteristics in dif- ferentiated mature gonads. Manuscript submitted 11 March 2003 to the Scientific Editor's Office. Manuscript approved for publication 31 May 2005 by the Scientific Editor. Fish. Bull. 103:601-619 (2005). Sexual differentiation and gonad development in striped mullet iMugil cephalus L.) from South Carolina estuaries* Christopher J. McDonough William A. Roumillat Charles A. Wenner Marine Resources Research Institute South Carolina Department of Natural Resources 217 Fort Johnson Road Charleston, South Carolina 29412 E-mail address (for C J Mcdonough) mcdonoughcsdnr.scgov The striped mullet (Mugil cephalus L.) is distributed circumglobally in tropi- cal and semitropical waters between latitudes 42°N and 42°S (Thomson, 1963; Rossi et al., 1998). Even though considered a marine species, striped mullet are euryhaline and can be found year round throughout the full range of estuarine salinities in the southeastern United States (Jacot, 1920; Anderson, 1958). Striped mullet are important throughout the world for commercial fisheries and aqua- culture. In the southeastern United States there are large-scale commer- cial fisheries for striped mullet in North Carolina and Florida. South Carolina and Georgia have much more limited landings (NMFS1). The commercial effort in the south- eastern United States targets "roe" fish (fish containing roe) during the fall spawning migration. Throughout the rest of the year mullet are fished commercially for human consump- tion (particularly the west coast of Florida) and bait (Anderson, 1958). Striped mullet have a significant eco- nomic impact in the southeast where they represented a landings value of 16.4 million dollars from 1994 to 2000 (NMFS1). Striped mullet landings in the Gulf of Mexico were significantly higher with a landings value of 86.2 million dollars for the same time period. Striped mullet are also one of the most important forage fishes that occur in the estuaries of the southeast and represent a significant food source for upper level piscivores (Wenner et al.2). General information on the biol- ogy of striped mullet has been well documented (Jacot, 1920; Anderson. 1958; Thomson, 1963, 1966; Chubb et al., 1981) but limited information is available on the reproductive biol- ogy of wild populations (Anderson, 1958; Stenger, 1959; Greeley et al., 1987; Render et al., 1995). There is a large body of work concerning striped mullet reproduction in aquaculture but many of these studies have con- centrated on females by using arti- ficial manipulation of the reproduc- tive cycle. Although the maturation process of oocytes may be the same as that in wild striped mullet, the environment and conditions under which maturation occurred in these studies was artificial (Shehadeh et al., 1973; Kuo et al., 1974; Pien and Liao, 1975, Kelly, 1990; Tamaru et al., 1994; Kuo, 1995). This lack of in- * Contribution 564 of the Marine Re- sources Research Institute, South Caro- lina Dept. of National Resources, Charles- ton, SC 29412. 1 NMFS (National Marine Fisheries Service). 2001. Unpubl. data. Sta- tistics and Economic Division, 1315 East- West Highway, Silver Spring, Md. 20910. http://www.st.nmfs.gov/stl/index.html. 2 Wenner, C. A., W. A. Roumillat. J. E. Moran, M. B. Maddox, L. B. Daniel, and J. W. Smith. 1990. Investigations on the life history and population dynamics of marine recreational fishes in South Carolina, part 1, p. 2-22. Completion reports, Project F-37, Charleston, and Project F-31, Brunswick. South Carolina Marine Resources Research Institute, P.O. Box 12559 Charleston, S.C. 29422. 602 Fishery Bulletin 103(4) formation on reproductive biology is surprising given the worldwide importance of mullet. In particular, there have been very few studies where sexual differentiation of immature striped mullet has been examined in con- junction with histological confirmation of maturity stage in reproductively capable adults. One notable exception was the work of Stenger (1959), who although thorough in histological confirmation of the male and female de- velopmental stages in relation to length, did not take age into consideration at differentiation or maturity. More recent studies (Chang et al., 1995; Chang et al., 1999) have examined gonad histology and plasma sex steroids during sex differentiation in young-of-the-year striped mullet up to 12 months old, but these studies did not provide any detail on fish length during devel- opment and differentiation. Other studies have exam- ined oocyte development and relative fecundity for the reproductive assessment of female striped mullet but did not examine reproductive development in males or take into consideration an independent confirmation of fish age (Greeley et al., 1987; Render et al., 1995). Few studies have described the process of spermatogenesis in striped mullet because most efforts on the propaga- tion and enhancement of striped mullet reproduction have concentrated on female development because of their commercial value. Grier (1981) used striped mul- let in describing the cellular organization of testes and spermatogenesis as a model for synchronously spawning fishes but did not describe size and age in relation to spermatogenesis. Striped mullet are considered isochronal spawning fishes (Greeley et al., 1987; Render et al, 1995). There are only a few observations of offshore spawning activ- ity (Arnold and Thompson, 1958), and eggs and larvae have rarely been collected offshore (Anderson, 1958; Fi- nucane et al., 1978; Collins and Stender, 1989). Collins and Stender (1989) concluded that striped mullet spawn in and around the edge of the continental shelf off the coasts of North Carolina, South Carolina, Georgia, and the east coast of Florida (an area often referred to as the South Atlantic Bight), but may also spawn outside the South Atlantic Bight (SAB). They also indicated a protracted spawning season that extended from October to April. This contrasts with the estimated spawning season from previous studies (2-5 months from No- vember through March) (Jacot, 1920; Broadhead, 1956; Anderson, 1958; Arnold and Thompson, 1958; Stenger, 1959; Dindo and MacGregor, 1981; Greeley et al., 1987; Render et al., 1995; Hettler et al., 1997). Female mul- let were thought to mature at three years of age at a size of 230 to 350 mm standard length (Thomson, 1951, 1963; Greeley et al., 1987). This study had three purposes: 1) to determine at what size and age striped mullet become fully sexually differentiated and to describe the morphological char- acteristics of sexual differentiation in both male and female striped mullet; 2) to determine the size and age at first maturity for each sex; and 3) to describe the timing and process of gametogenesis in relation to size and age in both males and females in order to provide a histological baseline for the evaluation and reproductive staging of striped mullet. Materials and methods Sampling and data collection Collections of striped mullet were conducted from Octo- ber 1997 through December 2000. Collections were based on a protocol of monthly random stratified sampling conducted in the Cape Romain, Charleston Harbor, and the ACE Basin estuaries in South Carolina (Fig. 1). The Charleston Harbor estuarine system is made up of three river systems: the Ashley, Cooper, and Wando rivers. In addition, Charleston Harbor proper was sampled as a separate stratum. The ACE Basin estuary is formed by the confluence of the Ashepoo, Combahee, and Edisto rivers and was sampled as a single estuary. One of the problems initially encountered with sampling was the ability to sample striped mullet throughout their estua- rine salinity range. The primary sampling gear used was a 184-meter trammel net with 356-mm stretch mesh outside panels and a 64-mm stretch mesh inner panel. Because striped mullet use the full range of estuarine habitats and freshwater, the use of alternate gear was necessary to obtain a representative sample of the popu- lation within all salinity regimes. Specimens collected with additional gear types in low salinity and freshwater habitats supplemented those specimens sampled with a trammel net. The additional gear types were an electro- shock boat, cast nets, and gill nets. The electroshock boat samples were obtained from the South Carolina Depart- ment of Health and Environmental Control from the major coastal river basins in South Carolina, including freshwater portions of the Waccamaw, Black, Pee Dee, Sampit, Santee, Cooper, Edisto, Ashepoo, Combahee, and Broad rivers (Fig. 1). Cast nets were used primarily in different portions of the Charleston Harbor estuary in tidal creeks and in areas where the trammel net could not be used effectively . The cast nets were 1.84 meters in diameter and had 10-mm mesh. The gill net was a 200-meter net with 64-mm stretch mesh that was used to test the efficiency of the trammel net sets. Standard morphological measurements were total length (TL), fork length (FL), standard length (SL) in mm, and body weight (BW) in grams (g). Any sub- sequent mention of fish length in the remaining text will be total length unless otherwise noted. Sagittal otoliths were removed for estimating fish ages. A gross examination of the gonads was used for initial sex and maturity assessment. If the gonads were estimated to weigh more than 1 g they were also weighed. A small sample of gonad tissue was removed from the posterior portion of the gonad where the lobes were joined and was fixed in 10% neutral buffered formalin for histologi- cal examination. The tissue samples used for histologi- cal evaluation were taken from the posterior section of the gonad because earlier developmental stages and differentiation were more evident where the ductwork McDonough et al.: Sexual differentiation and gonad development in Mugil cephalus 603 Aw C^ 4 (*- - ^rr"-^, S' Helena Sound A' frills Bay Cape Remain 5^ ^LM Charleston Harbor SOUTH CAROLINA Ji_ Figure 1 Map of coastal South Carolina with estuaries where trammel net collections were made: Cape Romain, Charleston Harbor, and Ashepoo River, Combahee River, Edisto River (ACE) Basin, as well as the coastal rivers where elec- troshock collections were made. and gonad tissue joined in striped mullet (Chang et al., 1995). Comparisons of oocyte density from different sections of striped mullet ovary have also demonstrated uniform distribution throughout the ovary (Shehedeh et al., 1973; McDonough et al., 2003). A gonadosomatic in- dex (GSI) was calculated for specimens according to the method of Render et al. (1995) where GSI was expressed as a percentage of gonad weight (GW) divided by body weight (BW) minus gonad weight, such that GSI = (GW/(BW-GW))x 100. Histological processing The tissue samples were processed by using standard wax histology techniques (Humason, 1967). Tissues were embedded in paraffin and cut on a rotary micro- tome. The sections, which ranged from 5 to 7 ^m thick, were then placed on microscope slides and stained with standard haematoxylin and eosin-Y staining techniques (Humason, 1967). After staining, tissue sections were sealed under a cover slip and evaluated for sex and maturity with a compound light microscope at lOOx magnification. The sex of each specimen was determined to be male, female, or undifferentiated. Maturity was assessed according to a modified version of the sched- ule used by Wenner et al. (1986) that was adapted by the authors to work with isochronal spawning fish, as well as assessed with previous models of reproductive development (Stenger, 1959; Grier, 1981; Wallace and Selman, 1981) (Table 1). Ovarian atresia was divided into four distinct phases as described by Hunter and Macewicz (1985). For the sake of consistency, the same terminology was used to describe the four phases of ovarian atresia in striped mullet in this study: alpha, beta, gamma, and delta (see Table 2). These evaluation methods were based on identification of morphological characteristics evident in histological sections. Speci- mens were evaluated by two readers to avoid bias. Any discrepancies of maturity stage between readers were either mutually resolved or the specimen was excluded from further analysis. 604 Fishery Bulletin 103(4) Table 1 Histological criteria used to determine reproductive stage in striped mullet (Mugil cephalus) once sexual differentiation has occurred. Modified from Wenner et al. ( 1986). Reproductive stage Male Female 1. Immature Developing 3. Running, ripe 4. Atretic or spent 5. Inactive or resting Inactive testes; small transverse sections compared to those of resting male; sper- matogonia and little or no spermatocyte development. Development of cysts containing primary and secondary spermatocytes all the way through accumulation of spermatozoa in lobular lumina and ducts. Predominance of spermatozoa in lobules and ducts and little occurrence of sper- matogenesis. No spermatogenesis occurring but some residual spermatozoa in shrunken lobules and ducts. Larger transverse sections compared to those of immature males; little or no sper- matocyte development; empty lobules with well-developed secondary ductwork and some residual spermatagonia. Inactive ovary with previtellogenic oocytes and no evidence of atresia. Oocytes are <80 (ira, lamellae lack muscle, and connective tissue bundles are not as elongate as those in mature ovaries, ovary wall is very thin. Developing ovary have enlarged oocytes generally greater than 120 um in size. Cortical alveoli become present and actual vitellogenesis occurs after oocytes reach 180 .um in size and continue to increase in size. Abundant yolk globules with oocytes reach a size range of >600 um. Completion of yolk coalescence and hydration in most oocytes. More than 30f? of developed oocytes undergoing the atretic process. See Table 2 for detailed description of the atretic process. Previtellogenic oocytes with only traces of atresia. In comparison to those of immature females, most oocytes are >80 ,«m, lamellae have some muscle and connective tissue bundles; lamellae are larger and more elongated than those of immature females and the ovarian wall is thicker. Table 2 Histological criteria used to determine atretic stage in striped mullet Mugil cephalus). Criteria based on ovarian atretic process described by Hunter and Macewicz (1985) and observational data of striped mullet ovaries from this study. Atretic stage Description 1. Alpha atresia a Vitellogenic oocytes are present with distinct yolk globules, which are beginning to break down. The most developmentally advanced oocytes will undergo atresia first, followed by less developed oocytes. The oocyte will break down from the interior outward; the vitelline membrane and follicle layers are the last portion of the oocyte to decay. As the oocyte breaks down, a series of vacuoles of various sizes will appear within the oocyte. /! The oocytes continue to become reduced in size as they decay. The vacuoles that began to form during the alpha stage are now coalescing together to form one large vacuole within the oocyte. This gives the lamellae a distinct hollow matrix and just the outer layers of the oocyte and follicle are now left. This appears to be the shortest atretic phase. 7 The oocytes that were left in the hollow matrix during the beta stage now begin to shrink in size and the outer layers fold in on themselves as the oocyte collapses. The areas in and around the collapsed oocytes and lamellae become highly vascularized during this stage in order to facilitate rapid resorption of decaying cellular material. There will still be some vacuoles present within the collapsed oocytes but they have become much smaller and there are far fewer of them. This stage continues until most of the remaining oocytes that developed for spawning are no longer recognizable as oocytes. 4. Delta atresia A The remnants of old oocytes at this stage are identifiable only as decaying cellular material and will stain a distinct yellow-brown color and are still present in (approximately) 30% or more of the material within the ovary. Undeveloped oocytes have a much more distinct and numerous presence within individual lamellae. The amount of vascularization seen in the gamma stage is reduced because most of the old material has been reabsorbed. Beta atresia 3. Gamma atresia McDonough et al.: Sexual differentiation and gonad development in Mugil cephalus 605 Aging techniques Age was determined by using the left sagittal otolith, which was embedded in epoxy resin. A 0.5-mm transverse section encompassing the otolith core was cut with an Isomet low speed saw with diamond wafering blades. The thin section of otolith embedded in the epoxy was observed with a dissection microscope at 20x magnifica- tion, and age was recorded as the number of annular rings present. The otoliths were initially aged by one reader. A second reader then evaluated a subsample of specimens from 1998 and 2000 and all the otoliths from 1999. The two groups of ages were compared by the percentage of agreement between the different age determinations and by a paired Mest that allowed a com- parison of the means and variances of the two groups iCampana et al., 1995). Ages were then validated by marginal increment analysis in order to establish the timing and periodicity of increment deposition (Campana. 2001). In addition, the precision of the ages was compared by using average percent error (APE) between the two sets of ages. "Precision" was defined as the repro- ducibility of age determinations (Beamish and Fournier, 1981; Chang, 1982). Using the Levenburg-Marquardt procedure (Zar, 1984). we determined the growth curve with a nonlinear least squares regression of total length on age. Results Age structure CL 20 0123456789 10 Age Figure 2 Age-frequency distribution (expressed as a percentage) for striped mullet 'Mugil cephalus L. i from South Carolina estuaries October 1997 to December 2000. n = 3760. We recorded the age of 3760 specimens and examined these specimens histologi- cally to determine sex and maturity stage. An additional 2524 young-of-the-year (age 0) specimens were used for the nonlinear regression of total length on age. as well as the sex ratios by both size and age. The age range for striped mullet in this study was 0 to 10 years, and 1- and 2-year-olds dominated the age distribution (Fig. 2). There was 81.7% agreement for age data between the two readers, and 99.5% agree- ment within one year for both readers. A Mest indicated no significant difference between the two sets of age estimations (r=2.898. df=1.233. P<0.05). The average percent error (APE) (Beamish and Fournier, 1981) between the two sets of age estimations was 0.41%. Marginal increment analysis indicated that growth increments were deposited during July 'Fig. 3). The 0.35 0.30- E 0.25 E | 0.20 i 0.15 0.10 0.05- 0.00 — i — i — i — n— i— 1998 1999 2000 Figure 3 Mean marginal increment distance by month for striped mullet i Mugil cephalus L. I from South Carolina estuaries, October 1997 to December 2000. n = 3760. Marginal increment equals the otolith section radius minus the distance from the core to the last annular increment. total length at age regression demonstrated a strong relationship 325 mm, although still possibly sexually immature, were fully sexually dif- ferentiated. The ratio of males to females was 2:1 until the fish were larger than 325 mm 325 mm(/2„=005=352.8, df=l). The sex ratio by age class showed 98.9% of the age-0 specimens were sexually undif- ferentiated (Fig. 6). The few age-0 fish that were differentiated were all males. At first annulus deposition, 91.9% of the specimens had differentiated. There were a few speci- mens (0.8% I that remained undifferenti- ated to 3 years old, but all striped mullet age 4 or older were completely differen- tiated. The sex ratio of males to females in the one-year-old age class was 1.0:0.25 80 70 to to to to <0 ri7toto*^fo'^to'-.to'^toK~to*^.to^to''--.<£ ^ ^.V^J^ fv fv (V > > > - to -- .to o 0/ to 'v to to to to Undifferentiated Male Female Size class (mm) Figure 5 Sex by size class (25-mm size classes) for striped mullet iMugil cepha- lus L.) from South Carolina estuaries, October 1997 to December 2000. n = 6284. 200 mm had become sexually differentiated. The undifferentiated gonads in specimens >200 mm were highly vascularized and had both the presence of ductwork, rounded germ cell nests, and lobule-like structures. In some cases, germ cell nests that were characteristic of female precursors 608 Fishery Bulletin 103(4) could also be found in the center portions of lobes adja- cent to the characteristic male precursor lobule struc- tures. The primary duct was now well formed; however there were still no definitive morphological characterstic that would enable sex determination. Male differentiation The initial differentiation of males was evident in the morphological features of the germ-cell tissue located along the peripheral portions of each lobe. The germ tissue began to form elongated bands perpendicular to the edge of the lobe, whereas the somatic tissue began to form fibrous bands originating along the edges of the 600 500 - 400 300 200 100 Males n_ , j\\N Immature H"'"""1 Developing ■■ Atretic i i Resting H Pfl^T-r , W In N Q "V 'V "> T> ' ' *> <*~. IS Qn <% I55 I ' 250 it 200 - 150 100 50 B Females r$ 171 l\\\l Fern 1mm Mi^l Fern Dev ^H Male Atr Fern Resl i £5 v- C *~. 1 (3s rj Of — i 1 inn nn JAN FEB MAR APR MAY JUN JUL AUG SEP OCT NOV DEC ajo- Inactive females 150- n = 539 ■■ - 50- ~~ 1 : 1 n r nn — ,-nn JAN FEB MAR APR MAY JUN JUL AUG SEP OCT NOV DEC JAN FEB MAR APR MAY JUN JUL AUG SEP OCT NOV DEC Month Figure 9 Frequency of occurrence of each maturity stage by month for male and female striped mullet (Mugil cephalus L.) from South Carolina estuaries, October 1997 to December 2000. McDonough et al. Sexual differentiation and gonad development in Mugil cephalus 611 immature ovaries, and had distinct smooth muscle layers (Fig. 14H). Any stromal tissue left in the ovary at this point was also greatly reduced and was essentially the mesentary from which the lamellae were suspended. Discussion Age structure o The abundance of 1- and 2-year-old striped mullet in South Carolina indicated that imma- ture fish dominate the estuarine population. The importance of proper age validation in order to make comparisons of age and sexual maturity cannot be understated. The most important aspect of age validation is to obtain a degree of precision that allows repeatability in age determinations (Campana, 2001). The periodicity of growth increment formation was validated by marginal increment analysis, and the precision of these age estimates was then tested by comparing age counts of two inde- pendent readers. Marginal increment analysis showed that annual growth increments were deposited in striped mullet in July in the entire data set, as well as sepa- rately for ages 1-5. By validating increment periodicity separately for different age groups, a consistent pat- tern for the species can be determined (Campana et al., 1995; Campana, 2001). The percent agreement be- tween the two readers and a r-test for independent age determinations allowed direct comparisons of the two groups of ages for consistency (Campana et al. 1995). However, these two methods were both independent of the age of the species. Therefore, average percent error (APE) was used to compare the different sets of ages because it is not independent of the age of a species (Beamish and Fournier, 1981). The low APE (0.41%) found between the two different age estimates indicated a high degree of precision, which allowed acceptance of these age determinations. Sexual differentiation Striped mullet are gonochoristic and sex is genetically determined. In contrast to mammals, gender of the mature germ cells of teleosts present in the gonad rather than the gender of the duct system forms the basis for classifying an individual as male or female (Shapiro, 1992). Early duct structures of the undifferentiated gonad characteristic of male development regress on female development. Initial duct development, along with germ tissue placement, takes on characteristics of the eventual sex once the process of differentiation begins (Asoh and Shapiro, 1997). Because of the plasticity of their gonad development, striped mullet retain some characteristics of the opposite sex (such as singular oogonia in males or Males Female MAY JUN JUL AUG SEP OCT NOV DEC Month I \\ 111; \l \K M'K Figure 10 Mean gonosomatic index value by month for male and female striped mullet (Mugil cephalus L.) from South Carolina estuaries from 1998 to 2000. n = 455. duct-work in females) during the initial stages of differen- tiation. The term that has been used to describe this phe- nomenon is "intersex" (Yamamoto, 1969) but this state could more accurately be defined as the hermaphroditic stage of some gonochoristic species. Numerous descrip- tions of intersex exist for teleosts (Atz,1964). Previous studies have brought up the possibility of hermaphrodism in striped mullet (Stenger. 1959; Atz, 1964; Moe, 1966); however, there is only one example of a simultaneous her- maphroditic striped mullet in the literature (Franks et al., 1998). Once differentiation advances, these secondary characteristics atrophy, and the gonad develops toward the genetically determined sex. We found that at first annular increment deposition (15-19 months), most (95%) immature striped mullet were sexually differentiated. Chang et al. (1995), us- ing cultured striped mullet, found that differentiation began only after 12 months of age, and 70% to 90% of immature fish at 15 to 17 months had differentiated sexually. We found only a small percentage (1.2%) of differentiated specimens at 12 months of age. Chang et al.(1995) did not report fish sizes, and Stenger (1959) studied sizes at sexual differentiation without reporting age. Stenger (1959) concluded that striped mullet up to 150 mm generally were not differentiated sexually. We found four specimens in which differentiation had occurred in the 126-150 mm size range, which repre- sented specimens 12 months or less in age. Once our specimens reached the 176-200 mm size range, just over 60% had sexually differentiated, which was also approximately the size range at which the first annulus appeared (Wenner and McDonough3). Chang et al. (1995) found that females differentiated earlier than males; we, on the other hand, showed sex 612 Fishery Bulletin 103(4) C 'T*\Sft5 DW V »££ 'tftJ rv ■ Jr -''.'* " _I)NV .*£*. '>? &f?A ,J fj / n n ^ '-.•„ v>q 7 BV I) iilMPs31i Figure 11 Photomicrographs of histological sections of undifferentiated juvenile striped mullet (Mugil cephalus L.) (A) 35-mm specimen at lOOx (scale bar=50 um) and (B) 600x (scale bar=10 ,«m) respectively; (C) 55-mm specimen at 400x, scale bar=20 jim; (D) 135-mm specimen at 400x, scale bar=20 ,um; (E) 184-mm specimen at 400x, scale bar=20 f*Jt$'t \ \ Mm* D *ST LA I ow OL Figure 13 Photomicrographs of histological sections of sexually differentiating female striped mullet iMugil cephalus L.). (A) Early differentiation of germ cell nests in a 239-mm specimen at lOOx, scale bar=100 um: (Bl early differentiation of germ cell nests in a 205-mm female at 400x, scale bar=20 ,«m; (C) mid-differentiation in a 225-mm female at lOOx, scale bar=100 jim; (D) advanced sexual differentiation with developing lamellae and ovarian wall in a 279-mm female at lOOx, scale bars = 100 f0.05) in average size of specimens by fishing gear (Fig. 5). Larger specimens were reported from the Levantine basin area and a smaller one was re- ported from the Balearic Sea (Fig. 6). Out of 27 common thresher shark sexed, 15 were males and 12 females. Sex ratio was 1.25 male:l female. The TL-FL and TL- dressed weight relationships are given below: TL = 20. 2 +1.707 FL TL = 69.7 DW° 35i [7^ = 0.95, n=24] [^=0.99, n=18]. The remaining nine shark species observed accounted for only 0.87% of the total shark catches. In total, 26 tope sharks were measured (ranging from 35 to 190 cm), 15 porbeagles (ranging from 87 to 277 cm), 7 bigeyed thresher sharks (ranging from 146 to 353 cm) and 4 smooth hammerheads (ranging from 277 to 300 cm TL). Only three bluntnose sixgill sharks (mean weight of 10.7 kg), two sandbar sharks (mean weight of 17 kg), two longnose spurdogs (mean weight of 1.7 kg), two basking sharks, and one smoothhound were reported, but no length measurements were available for these species. A total of 571 specimens were examined for life condi- tion on capture. The majority were very active follow- ing capture and their physical condition was especially good. Only 5.1% of the specimens brought onboard were dead (Table 10). Discussion Our results show that most of the sharks caught by the swordfish and tuna fisheries in the Mediterranean Sea are typically pelagic or coastal-pelagic species of wide- spread distribution in temperate and tropical waters throughout the world. However, some sporadic catches of poorly known, deepwater species of the families Hexan- chidae and Alopiidae were also observed. The most plausible reason for these catches is that the deepwater species ascend close to the surface at night where they may be taken by longlines targeting swordfish (Castro et al., 1999). 628 Fishery Bulletin 103(4) Table 8 Fishing set s, effort 1x1000 hook 5 or 1000 m of net), and catch r ates (number offish/1000 hooks or number offish/1000 m of net ) of sharks and target species samp ed in the large pelagic fisheries of the Mediterranean Sea during 1998-99 Sampling conducted both at sea and at landing sites. PG =Pri maceglaaea, 10=Isurus oxyrinet us, AV=Alopias vulpinus, GG = Galeorhinus galeus. The target species for specific gears Xiphias gladius for SWO-LL SWO-LLA and DN; Th unnus alalunga for ALB-LL; and Thunnus thynnus for BFT-LL. Fishing Catch rate Other Total Target gear Area Sets Effort PG IO AV GG sharks sharks species SWO-LL Ionian 594 1151.0 0.53 0.00 0.001 0.00 0.003 0.53 2.67 Levantine 7 7.0 0.00 0.00 0.00 0.14 0.00 0.14 7.71 Adriatic 771 2061.6 1.00 0.00 0.004 0.00 0.00 1.00 3.59 Tyrrhenian 9 18.5 0.27 0.00 0.00 0.00 0.00 0.27 8.43 Strait of Sicily 23 46.4 0.06 0.00 0.02 0.02 0.11 0.22 14.53 Balearic 1312 1168.8 0.07 0.04 0.01 0.003 0.001 0.12 13.09 Alboran 1391 1406.7 3.59 0.19 0.008 0.007 0.004 3.80 10.59 Catalonian 290 522.1 0.17 0.004 0.004 0.004 0.004 0.18 5.99 Total 4397 6382.0 1.24 0.05 0.006 0.003 0.002 1.30 7.00 SWO-LLA Aegean 42 18.5 1.19 0.00 0.05 0.05 0.00 1.30 11.27 Levantine 211 94.9 0.31 0.08 0.01 0.00 0.01 0.41 15.40 Total 253 113.4 0.45 0.07 0.02 0.01 0.01 0.56 14.72 ALB-LL Aegean 99 151.0 0.04 0.00 0.00 0.00 0.00 0.04 5.59 Adriatic 6 15.3 0.00 0.00 0.00 0.00 0.00 0.00 22.22 Ionian 239 527.0 0.09 0.00 0.00 0.00 0.00 0.09 19.60 Strait of Sicily 7 17.5 0.00 0.00 0.00 0.00 0.00 0.00 127.14 Balearic 48 158.7 0.00 0.006 0.00 0.00 0.006 0.013 23.73 Catalonian 41 142.1 0.07 0.007 0.00 0.00 0.00 0.08 29.14 Total 440 1011.6 0.07 0.002 0.00 0.00 0.00 0.07 20.76 BFT-LL Strait of Sicily 2 2.8 0.00 0.00 0.00 0.00 0.00 0.00 5.36 Balearic 19 20.9 0.29 0.00 0.00 0.00 0.00 0.29 3.88 Total 21 23.7 0.25 0.00 0.00 0.00 0.00 0.25 4.05 DN Ionian 715 8336.3 0.03 0.00 0.002 0.00 0.001 0.04 0.21 Table 9 Fishing sets and catch rates (number of fish/fishing set) of sharks and the Ionian Sea during 1998-99. PG=Prionaee glauca, \0=hurus oxyrin The target species for specific gears: Xiphias gladius for SWO-LL and DN target species in the three fishing gears studied in chus, AV=Alopias vulpinus, GG=Galeorhinus galeus. Thunnus alalunga for ALB-LL. Fishing gear Catch rate Sets PG IO AV GG Other sharks Total sharks Target species SWO-LL ALB-LL DN 594 1.02 239 0.21 715 0.39 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.03 0.00 0.01 0.00 0.02 1.03 0.21 0.44 5.17 43.22 2.50 Onboard observations and interviews with fishermen at landing sites revealed that shark discarding is not a common practice in the large pelagic fisheries in the Mediterranean Sea. Very few shark discards were re- corded and only from Greek vessels (seven blue sharks out of 78 total). The fishermen usually retain their in- cidental catches because there is a market demand for sharks in Europe. However, wholesale shark flesh prices are quite variable, ranging from 2 to 8 euros. Moreover, the jaws and tails of some shark species are often sold Megalofonou et al.: Incidental catch and estimated discards of pelagic sharks in the Mediterranean Sea 629 100 90 13 80 a. o 70 60 50 -•- Xiphias gladius -•- Sharks 25 2.0 o 15 2 •■ 1 0 ■■05 00 Jan Feb Mar Apr May Jun Jul Months Aug Sep Oct Nov Dec Figure 2 Monthly variation in sharks and swordfish longline CPUE (catch in num- bers/1000 hooks) in the swordfish longline fishery of the Mediterranean Sea during 1998-99. 0 35 ■ 0 30 ■ E 0 25 ■ n 1. 1 LJ 0 20 ■ o 1 III 015 ■ ■> 0. o 010 ■ 0 05 0 00 Xiphias gladius Sharks Jan Feb Mar Aug Sep Oct Nov Dec Figure 3 Monthly variation in sharks and swordfish CPUE (catch in numbers/1000 m net) in the driftnet fishery of the Mediterranean Sea during 1998-99. in local markets. The very low discard rate of shark — about 1% of the sharks caught during onboard sampling was discarded — confirmed that sharks contribute to fishermen's income and may become target species with future increases in their market value. That discard- ing was observed only in the Greek swordfish fleets is probably due to the low market prices of shark meat compared to the high price of swordfish in this country. Sometimes during long trips fishermen are reluctant to retain them onboard and loose cool storage space for more valuable species such as swordfish or tuna. The analysis of catch composition by gear and areas indicated that the various gears used in the swordfish and tuna fisheries affect the shark populations dif- ferently and that the proportion of shark catches is related both to the type of fishing gear and the sam- pling area. This finding is consistent with previous findings for the Mediterranean Sea where incidental shark catch in the swordfish fisheries varied from in- significant to dominant, depending on the area studied (De Metrio et al., 1984; Di Natale, 1998; Filanti et al., 1986; Buencuerpo et al., 1998; Mejuto et. al., 2002). The highest shark incidental catches were found in the Alboran Sea and were probably related to their location (Alboran Sea), adjacent to the Atlantic Ocean. Shark bycatch in the Atlantic swordfish fishery is one 630 Fishery Bulletin 103(4) - Blue shark (Prionace glauca) 8"': ■ L n=3784 6 - n n 4% - tl " " : - J] n r I ..■k.-.-j. nnil-ftTYTTi-v.n,,- 40 60 80 100 120 140 160 180 200 220 240 260 280 300 320 340 360 Total Length (cm) Shortfm mako (Isurus oxyrinchus) n=257 B,,« n 40 60 80 100 120 140 160 180 200 220 240 260 280 300 Total Length (cm) 12 ■ Common threshe ■ shark (Alop ias vulpinus) 10' ■ 8% ■ n=50 ':" ' 4% ■ 1 2% ■ 0% 1 1 1 1 1 50 100 150 200 250 300 350 400 450 500 550 Total length (cm) Figure 4 Length-frequency distribution (in percentage by 5-cm size classes) for Prionace glauca, Isurus oxyrinchus, and Alopias vulpinus sampled in the Mediterranean Sea during 1998-2000. of the highest in the world, rarely dropping below 30% of the total catch in numbers of fish (Amorim et al., 1998; Buencuerpo et al., 1998; Hazin et al., 1998; Marin et al., 1998). The higher incidence of sharks in the Alboran Sea could also be due to the higher trophic potential of the western Mediterranean compared to the eastern part. The discrepancies in observed at-sea and at-landing data, especially in the west- ern Mediterranean Sea catch composition, could be mainly due to the discarding of "other species." In addition, the discarding of undersize target species, such as sword- fish and tunas, could be another reason for the discrepancies observed. It is reasonable that observers at landing sites were not able to record exactly the entire nonshark discards at sea from the information that fishermen provided; thus shark landings do not always reflect actual percentage of catch composition caught at sea. The shark catch rates obtained in our study were lower than those reported in previous studies for various areas of the Mediterranean Sea (Table 11) probably be- cause of the fishing pressure throughout the years. A comparison of the shark catch rates in the Mediterranean and Atlantic indicated that the catch rates are generally lower throughout the Mediterranean (Table 11). Possible reasons could be either the lower productivity of the Mediterranean Sea, or, as alluded to above, lower availability of sharks in the Mediterranean due to re- gional depletion from historical fishing, or both. The configuration and effectiveness of fishing gears used could be another rea- son for the higher CPUE in the Atlantic Ocean. Hazin et al. (1998) and Kotas et al.2 reported an increase in use of wire snoods in Atlantic swordfish fisheries to retain more sharks for the growing market for shark fins. Monthly analysis of catches indicated that maximum catch rates occur during late spring and summer (May-August) in the swordfish longline (SWO-LL) fishery, and in June in the driftnet fishery. Month- ly variations in catch rates were found also by Buencuerpo et al. (1998), who reported peaks of shark catch in April and Septem- Kotas. J. E., S. dos Santos, V. G. de Azevedo. J. H. de Lima, J. D. Neto, and C. F. Lin. 2000. Observations on shark by-catch in the monofilament longline fishery off southern Brazil and the National ban on finning, 8 p. IBAMA-REVIZEE research. [Copyright: www. wildaid.org.l Megalofonou et al : Incidental catch and estimated discards of pelagic sharks in the Mediterranean Sea 631 600 500 400 300 100 600 500 400 300 200 100 —I— Max Pnonace glauca _ Min n=3771 I I Mean+SD Mean-SD o Mean I Q □ I SWO-LL ALB-LL BFT-LL DN SWO-LL, Isurus oxyrinchus n=257 1 I □ 600 500 400 300 200 100 SWO-LL ALB-LL BFT-LL DN SWO-LL, Aloptas vulpinus n=48 I I SWO-LL ALB-LL BFT-LL DN Fishing gear SWO-LL, Figure 5 Size-range variation for Prionace glauca, Isurus oxy- rinchus, and Alopias vulpinus by fishing gear in the Mediterranean Sea during 1998-2000. See Table 1 for definitions of abbreviations for fishing gear along x axis. 600 300 200 100 600 500 400 200 100 — ^ax Prionace glauca Min n=3771 CD Mean+SD Mean-SD □ Mean I M T T I T Isurus oxyrinchus n=257 i 600 500 300 200 100 Alopias vulpinus n=48 I T 1 23456789 Area Figure 6 Size-range variation for Prionace glauca, Isurus oxy- rinchus, and Alopias vulpinus by area sampled in the Mediterranean Sea during 1998-2000. See Table 1 for definitions of area numbers along the x axis. ber in the eastern N. Atlantic and Straits of Gibral- tar. Probably, certain water temperature preferences of sharks during their biological cycle force them to shift to shallower and warmer water masses, especially in summer. At these depths sharks are more vulnerable to surface gears and that is reflected in higher catches. Higher catch rates in late spring and summer could be also attributed to juvenile recruitment (Strasburg, 632 Fishery Bulletin 103(4) Table 10 Life-status condition of 571 sharks at time of capture, by species, and per fishing gear, observed onboard commercial fishing ves- sels in the Mediterranean Sea during 1998-2000. Gear abbreviations: SWO-LL = swordfish longline, SWO-LLA=American-type swordfish longline, ABL-LL=albacore longline, DN=driftnet, BFT-LL=bluefin tuna longline. Good Fair Poor Number % Species P. glauca 364 71.0 I. oxyrinchus 7 22.6 A. vulpinus 3 18.8 G. galeus 4 80.0 A. superciliosus 1 100.0 C. plumbeus 0 0.0 H. griseus 3 100.0 Fishing gear SWO-LL 334 66.8 SWO-LL^ 34 97.1 ALB-LL 12 46.2 DN 2 40.0 BFT-LL 0 0.0 Total 382 66.9 Number % 69 13.5 10 32.3 4 25.0 1 20.0 0 0.0 2 100.0 0 0.0 76 15.2 0 0.0 6 23.1 2 40.0 2 40.0 86 15.1 Number % 57 11.1 9 29.0 8 50.0 0 0.0 0 0.0 0 0.0 0 0.0 64 12.8 0 0.0 6 23.1 1 20.0 3 60.0 74 13.0 Dead Number % 23 4.5 5 16.1 1 6.3 0 0.0 0 0.0 0 0.0 0 0.0 26 5.2 1 2.9 2 7.7 0 0.0 0 0.0 29 5.1 1958; Carey and Scharold, 1990; Nakano, 1994; Bigelow et al., 1999). The abundance and widespread distribution of blue sharks throughout the Mediterranean that we deter- mined supports previous findings. However, our ob- served catch rates were lower than those reported ear- lier for the same areas (De Metrio et al., 1984; Filanti et al., 1986; Buencuerpo et al., 1998; Di Natale, 1998; Relini-Orsi et al., 1999; De Zio et al., 2000). Varia- tion in sex ratio and size distribution between differ- ent areas studied indicated sexual or size segregation, or both. Spatial and temporal segregation of pelagic sharks by sex and size was well documented by Stras- burg (1958) and Nakano (1994) in the Pacific Ocean. Further analysis regarding distribution by latitude- longitude, time of year, and size classes of specimens is needed to establish a possible blue shark migratory pattern in the Mediterranean Sea. Pratt's estimates on the sexual maturity of blue shark (215 cm TL for males, 257 cm TL for females) from the North Atlantic Ocean (Pratt, 1979) indicate that in all areas studied in the Mediterranean Sea, albacore and swordfish longline fisheries generally capture immature to subadult speci- mens and driftnets and American type swordfish long- lines capture adults. Of all blue sharks captured in the large pelagic fisheries of the Mediterranean during our study, 91.1% were under 215 cm TL and 96.3% under 257 cm TL. This observation, which indicates that the majority of Mediterranean blue sharks caught have not reached maturity, is of concern and reinforces the need for global assessments of this species. In the Atlantic and Pacific Ocean results based on a considerable time series of data show a decrease in abundance (Cramer, 1996) and in average size (Holts et al., 1998) of blue sharks. Because blue sharks are an incidental catch in the large pelagic and highly migratory species fisheries in the Mediterranean, standardizing catch rates is very difficult. Average size may be a more sensitive indicator of shark stock status than catch rates when there is a long enough time-series of data. We found a much lower incidental catch of shortfin mako than other authors have reported in the Medi- terranean (Dai, 1997; Buencuerpo et al., 1998). This species seems more abundant in the Atlantic Ocean where in some areas it represents more than 10% of total catches (Buencuerpo et al., 1998; Stone et al., 2001). The almost equal sex ratio reflects the findings of Buencuerpo et al., (1998) and Moreno et al., (1992). As with blue sharks, larger makos were observed in the Levantine basin although in small numbers. Because males mature at 195 cm TL (Compagno, 1984) and females between 273 and 298 cm (Mollet et al., 2000), 98.4% of shortfin makos in our study were smaller than the size of first maturity. The absence of a consistent time series of abundance data did not allow us to es- timate the trend in the status of the shortfin mako population in the Mediterranean Sea. Cramer (1996) outlined a steady decline in catch indices for this spe- cies from 11.86 fish/1000 hooks in 1985, to 3.52 in 1996 for the U.S. commercial Atlantic longline fishery in the Caribbean and the Gulf of Mexico. The Azorean fleet mako landings decreased by almost 50% in numbers from 1987 to 1994 (Castro et al., 1999). Together with the low catch rates in the Mediterranean Sea, short- Megalofonou et al.: Incidental catch and estimated discards of pelagic sharks in the Mediterranean Sea 633 Table 1 1 Comparison of shark catch rates (CPUE in number offish/1000 hooks) in longline fisheries during investigations in the Mediter- ranean Sea and the Atlantic Ocean. SWO-LL= swordfish longline; Tuna-LL=tuna longline gear. Author De Metrio et al. (1984)' Filantietal.(1986l DeZioetal. (2000) DiNatale(1998) Buencuerpo et al. ( 1998 1 Present study Present study Present study Present study Buencuerpo et al. ( 1998 ) Stone and Dixon (2001) Hazinetal. (1998) Area Period Ionian Sea Ionian Sea Adriatic Sea Tyrrhenian Sea, Strait of Sicily Gibraltar Strait Ionian Sea Adriatic Sea Strait of Sicily Alboran Sea E. Atlantic NW Atlantic W. Atlantic 1984 1978-85 1984-98 1991-92 1991-92 1998-99 1998-99 1998-99 1998-99 1991-92 1999 1983-97 t ;<\u SWO-LL SWO-LL SWO-LL SWO-LL SWO-LL SWO-LL SWO-LL SWO-LL SWO-LL SWO-LL SWO-LL Tuna-LL CPUE 0.9-2.2 1.5-3.0 2.4 0.4 24.2 0.5 1.0 0.2 3.8 9.9-37.J 43.8 16.8 Blue shark catch rates only. fin makos may be one of the most over-fished pelagic sharks in the Mediterranean Sea. Our low catch rates for common thresher shark in the Mediterranean were almost identical with the findings of Buencuerpo et al. (1998) for the Gibraltar Strait re- gion. However, the abundance of this species supports directed fisheries in some areas. Such a case occurred off California waters during 1977-85, when thresher shark CPUE in the driftnet fishery ranged from 0.13 to 1.92 fish/fishing set (Holts et al., 1998). In our study, one third of the specimens caught came from the Io- nian driftnet fishery but the largest individual was captured in the Levantine basin (514 cm TL) with the swordfish longline. Pacific females mature at 315 cm TL (Strasburg, 1958) and males mature at about 333 cm TL (Cailliet and Bedford, 1983), and we calculated that 40% of the female common thresher sharks caught were below 315 cm and 50% of the males were below 333 cm. Although the above data indicate that most were caught as immature sharks, there are no data on the first maturity of common thresher sharks in the Mediterranean Sea. There is doubt, however, that fe- males mature at a smaller size than males in the same region and we therefore deduced that fishing pressure was very intense on juvenile and subadult groups. The low capture numbers for other shark species could be due either to the scarcity of these species in the Medi- terranean Sea or to the "fished-down" condition of shark populations, or both could be causes. Another reason could be the low capture efficiency of the gears used. The high proportion of sharks that were alive on cap- ture agrees with Kotas et al.2, who reported that 97% of blue sharks and 78% of shortfin makos were alive when landed on deck. These high survival rates are en- couraging and could become the basis for conservation measures in the future, such as releasing immature fish or enforcing catch quotas. Our study provides a reference point for the present status of pelagic sharks in the Mediterranean Sea, the effect of fisheries on them, and a baseline for future monitoring. Fishing for swordfish and tunas affects much of the pelagic ecosystem by taking predators of swordfish and tunas (large pelagic sharks), their prey (small tunas), and their competitors, such as other elas- mobranchs, billfishes, and tunas. Up to now, there has been little documentation and understanding of fishing effects on the wider ecosystem. To strengthen manage- ment for large pelagic fishes such as sharks, a multi- species assessment with an ecosystem approach should be adopted. To achieve this goal, long-term monitoring programs should be established and exploitation strat- egies should be linked to conservation plans for shark species in the Mediterranean Sea. Acknowledgments We thank the Greek, Italian, and Spanish fishermen who collaborated during sampling procedures. We thank also the two anonymous reviewers who improved the manuscript with their valuable suggestions. This study was performed under the financial aid of the Commis- sion of the European Communities (Project no. 97/50 DG XIV) and does not necessarily reflect the views of the European Commission and in no way anticipates the Commission's future policy in this area. Literature cited Amorim de, A. F., C. A. Arfelli, and L. Fagundes. 1998. Pelagic elasmobranchs caught by longliners off southern Brazil during 1974-97: an overview. Mar. Freshw. Res. 49(71:621-632. 634 Fishery Bulletin 103(4) Anonymous. 1999. Report of the Inter-sessional meeting of the ICCAT sub-committee on by-catch; Messina, Italy, May 11-14 1999. ICCAT Col. Vol. Sci. Pap. 51:1729-1775. Bigelow, K. A., C. H. Boggs, and X. He. 1999. Environmental effects on swordfish and blue shark catch rates in the US North Pacific longline fishery. Fish. Oceanogr. 8:178-198. Buencuerpo, V., S. Rios, and J. Moron. 1998. Pelagic sharks associated with the swordfish, Xiphias gladius, fishery in the eastern North Atlan- tic Ocean and the Strait of Gibraltar. Fish. Bull. 96:667-685. Cailliet, G. M., and D. W. Bedford. 1983. The biology of three pelagic sharks from California waters, and their emerging fisheries: a review. CalCOFI Rep. 24:57-69. Carey, F. G„ and J. Scharold. 1990. Movements of blue sharks [Prionace glauca) in depth and course. Mar. Biol. 106:329-342. Castro, J. I., C. M. Woodley, and R. L. Brudek. 1999. A preliminary evaluation of the status of shark species. FAO Fish. Tech. Pap. 380:1-72. FAO. Rome. Compagno, L. J. V. 1984. FAO species catalogue. Vol. 4: Sharks of the World: an annotated and illustrated catalogue of shark species known to date. Part 2: Carchariniformes. FAO Fish. Synop. 125:251-655. Cramer, J. 1996. Recent trends in the catch of undersized sword- fish by the U.S. pelagic longline fishery. Mar. Fish. Rev. 58:24-32. Dai.X. 1997. A preliminary analysis on the composition of catches obtained by longline fishing in the Mediter- ranean Sea. J. Shangai Fish. Univ. 6:107-111. De Metrio, G., G. Petrosino, C. Montanaro, A. Matarrese, M. Lenti, and E. Cecere. 1984. Survey on summer-autumn population of Prionace glauca L. (PISCES, CHONDRICTHYES) during the four-year period 1978-81 and its incidence on swordfish tXiphias gladius L.) and albacore (.Thunnus alalunga (Bonn)) fishing. Oebalia 10:105-116. De Zio, V, A. M. Pastorelli, and L. Rositani. 2000. Catture accessorie di Prionace glauca (L.) durante la pesca dei grandi pelagici nel basso Adriatico (1984-1998). Biol. Mar. Meditterr. 7:444-446. Di Natale, A. 1998. By-catch of shark species in surface gear used by the Italian fleet for large pelagic species. ICCAT Col. Vol. Sci. Pap. 48:138-140. Filanti, T, P. Megalofonou, G. Petrosino, and G. De Metrio. 1986. Incidenza dei Selaci nella pesca del Pesce Spada con longline nel golfo di Taranto. Nova Thalassia 8:667-669. Hazin, F H. V, J. R. Zagaglia. M. K. Broadhurst, P. E. P. Travassos, and T. R. Q. Bezerra. 1998. Review of a small-scale pelagic longline fishery off Northeastern Brazil. Mar. Fish. Rev. 60:1-8. Holts, D. B., A. Julian, O. Sosa-Nishizaki, and N. W. Bartoo. 1998. Pelagic shark fisheries along the west coast of the United States and Baja California, Mexico. Fish. Res. 39:115-125. Marin, Y. H„ F Brum, L. C. Barea, and J. F. Chocca. 1998. Incidental catch associated with swordfish long- line fisheries in the south-west Atlantic Ocean. Mar. Freshw. Res. 49(71:633-639. Mejuto, J., B. Garcia-Cortes, and J. M. de la Serna. 2002. Preliminary scientific estimations of by-catch landed by the Spanish surface longline fleet in 1999 in the Atlantic Ocean and Mediterranean Sea. ICCAT Col. Vol. Sci. Pap. 54:1150-1163. Mejuto, J., and J. M. de la Serna. 2000. Standardized catch rates by age and biomass, for the North Atlantic swordfish iXiphias gladius) from the Spanish longline fleet for the period 1983-1998 and bias produced by changes in the fishing strategy. ICCAT Col. Vol. Sci. Pap. 51:1387-1411. Mollet, H. F, G. Cliff, H. L. Pratt Jr., and J. D. Stevens. 2000. Reproductive biology of the female shortfin mako, Isurus oxyrinchus Rafinesque, 1810, with comments on the embryonic development of lamnoids. Fish. Bull. 98:299-318. Moreno, J. A., and J. Moron. 1992. Comparative study of the genus Isurus (Rafinesque, 1810), and description of a form ("Marrajo Criollo") apparently endemic to Azores. Aust. J. Mar. Freshw. Res. 43:10-122. Musick, J. A., G. Burgess, G. Cailliet, M. Camhi, and S. Fordham. 2000. Management of sharks and their relatives (Elasmo- branchii). Fisheries 3:9-13. Nakano, H. 1994. Age, reproduction and migration of blue shark in the North Pacific Ocean. Bull. Natl. Res. Inst. Far Seas Fish./Enyosuikenho 31:141-219. Relini-Orsi, L., G. Palandri, F. Garibaldi, and C. Cima. 1999. Longline swordfish fishery in the Ligurian Sea: eight years of observation on target and by catch species. ICCAT Col. Vol. Sci. Pap. 49:146-150. Pratt, H. L. 1979. Reproduction in the blue shark, Prionace glauca. Fish. Bull. 77:445-470. Stone, H., and L. Dixon. 2001. A comparison of catches of swordfish, Xiphias gla- dius, and other pelagic species from Canadian longline gear configured with alternating monofilament and mul- tifilament nylon gangions. Fish. Bull. 99:210-216. Strasburg, D. W. 1958. Distribution, abundance and habits of pelagic sharks in the central Pacific Ocean. Fish. Bull. 58: 335-361. 635 Abstract — The annual ovarian cycle, mode of maturation, age at maturity, and potential fecundity of female Rikuzen sole (Dexistes rikuzenius) from the North Pacific Ocean off the coast of Japan were studied by 1) his- tological examination of the gonads, 21 measurement and observation of the oocytes, and 3) by otolith aging. The results indicated that ovulation occurs from September to December and peaks between September and October. Vitellogenesis began again soon after the end of the current season. Maturity was divided into eight phases on the basis of oocyte developmental stages. Mature ova- ries contained developing oocytes and postovulatory follicles but no recruit- ing oocytes, indicating that this spe- cies has group-synchronous ovaries and is a multiple spawner. Almost all females matured first at an age of 1+ year and spawned every year until at least age 8+ years. Poten- tial fecundity increased exponentially with body length and the most fecund fish had 15 times as many oocytes as the least fecund fish. Potential fecun- dity and relative fecundity were both positively correlated with age from 1 to 6+ years, but were negatively corre- lated, probably because of senescence, in fish over 7 years. These results emphasize that the total productivity of aD. rikuzenius population depends not only on the biomass of females older than 1+ but also on the age structure of the population. Reproductive biology of female Rikuzen sole (Dexistes rikuzenius)* Yoji Narimatsu Daiji Kitagawa Tsutomu Hattori Tohoku National Fisheries Research Institute Fisheries Research Agency Hachinohe Branch. Same-machi Hachinohe, Aomon, 031-0841 Japan E-mail address (for Y. Narimatsu) nary@aftrc go ip Hirobumi Onodera Iwate Fisheries Technology Center Hirata, Kamaishi Iwate, 026-0001 Japan Manuscript submitted 10 January 2004 to the Scientific Editor's Office. Manuscript approved for publication 10 April 2005 by the Scientific Editor. Fish. Bull. 10.3:635-647 (2005). To understand fish population dynam- ics, reproductive information, such as the maturation of oocytes, the size and age at first maturity, and fecun- dity, is indispensable. Gonadal matu- ration is determined from the external appearance of the gonads, the gonad- osomatic index, and oocyte size, or from observations of histologically prepared gonads (West, 1990). With the former two methods it is possible to measure samples in the field and to record data on numerous samples in a short period of time; however, the mode of oocyte development can only be clarified by using observations of histologically prepared gonads (Wal- lace and Selman, 1981). The methods used to determine if an individual has spawned and to measure the number of eggs spawned in the current repro- ductive season differ with the mode of oocyte development (West, 1990). In fishery models, reproductive potentials are conventionally repre- sented by spawning stock biomass (Ricker, 1954; Beverton and Holt, 1957; Trippel et al., 1997). Howev- er, at the population level spawning stock biomass does not always corre- late with egg productivity. Length at first maturation, the frequency of oc- currence of degenerated oocytes, and fecundity (that is, the total number of offspring produced in a reproduc- tive season by an individual female) are closely related to the age and energetic conditions of an individual (Hunter and Macewicz, 1985a; Hor- wood et al., 1986, 1989; Trippel et al., 1997; Sampson and Al-Jufaily, 1999; Kurita et al., 2003). Therefore, examination of age and body size in relation to fecundity is useful in de- termining the abundance of eggs laid in a population. Oocyte development can be divided into three types (Wallace and Sel- man, 1981). In determinate fecundity, fecundity is fixed before spawning starts, such as in species which have synchronous or group-synchronous ovaries. In indeterminate fecundity (i.e., for those species whose ovaries develop asynchronously), unyolked oocytes grow to maturity after the onset of spawning (Hunter and Mace- wicz, 1985b; Hunter et al., 1992). In addition, the development of oocytes can vary even among populations of a single species (Sampson and Al-Jufai- ly, 1999) and some females classified as maturing or mature by external observation are often actually imma- ture, and vice versa (Hunter et al., 1992; Zimmermann, 1997). Hence, with a species or a population for 1 Contribution B57 from Tohoku National Fisheries Research Institute, Fisher- ies Research Agency of Japan, Miyagi, Japan. 636 Fishery Bulletin 103(4) which little information is available, it is important to determine specific reproductive traits by using the most accurate methods and to compare the results with those of simpler methods. Rikuzen sole (Dexistes rikuzenius) (also known as Rikuzen flounder, FAO) is a coastal flatfish that lives at depths of 100 to 360 m in the waters off the south coast of southern Hokkaido, Japan, and the southern Korean Peninsula (Sakamoto, 1984). It inhabits sandy bottoms and preys mainly on benthic invertebrates (Fujita et al., 1995). It is relatively abundant in the North Pacific off the coast of Japan and is an important fishery resource for bottom trawlers (Ishito, 1964; Ogasawara and Ka- wasaki, 1980). The commercial catch of flatfish such as the Rikuzen sole has fluctuated widely in this area over the past few decades (Anonymous, 2002), and therefore fisheries management is needed to maintain stable and appropriate fish-density levels. In addition to fisheries, various internal and external conditions may affect the fluctuations in abundance of fish populations. Understanding reproductive traits, or survival in the early life stages, is a step toward revealing population dynamics. Although both sexes have indeterminate growth trajectories, conspicuous sexual dimorphism occurs during the growth and life span of Rikuzen sole. Females are larger at any given time after age 1+ and live longer than males (Ishito, 1964). The spawning period of the Sendai Bay popula- tion occurs from late October to late January and peaks from November to December (Ogasawara and Kawasaki, 1980). Using measurements of oocyte diameter and the appearance of the whole ovary, Ogasawara and Kawa- saki (1980) revealed that females spawn several batches of eggs during one spawning season. However, because histological observations of the gonads have not been conducted, details of the reproductive biology, such as annual cycle of oocyte development, and body size and age at maturity, have not been determined. In addition, no information about fecundity has been reported. We examined the oogenesis of Rikuzen sole caught in the North Pacific Ocean off the coast of Japan over a period of one year. The aim was to determine the mode of maturation, annual reproductive cycle, and age at first maturity based on histological examinations, age determinations from otolith growth increments, and gonadosomatic indices (GSIs). Using these results, we were able to estimate body size and age-related potential fecundity and were able to develop a simpler method for determining potential fecundity. Materials and methods From May 2000 to April 2001, except for July and August when commercial bottom trawl fishing was pro- hibited, Rikuzen sole samples were collected once or twice a month from the fisheries market in Hachinohe, Aomori Prefecture, Japan. All samples were caught by bottom trawl nets in the coastal waters off Shitsukari (41°22\ 141°33'E) and Hachinohe (40°43'N, 144°44'E), 128° 132° 136° 140 144° 146° 44° N A / 40° •° < V 36° ^ f r -^ 1 32° 44 40° 36° 32 128' 132: 136° 140° _ I 144" 41- - 40° 39° 38L 37° Sendai Bay 140° 14V 142° 143° Longitude (E) 1 44° 145° Figure 1 Catch area for Rikuzen sole (Dexistes rikuzenius) in the Northern Pacific Ocean off the northeast coast of Japan, 2000-2001. from depths of 70-300 m (Fig. 1). During July and August, samples were collected with bottom long lines off the coast at Onezaki (39°12'N, 141°56'E) from a depth of 85-109 m. A total of 1031 females were collected and their standard lengths (SL) to the nearest mm, total body weights, eviscerated body weights, and ovary weights to the nearest 0.1 g were measured. The GSI and body condition (BC) of each specimen were calculated with the following formulas: GSI = (gonad weight/eviscerated body weight)xl00, and BC = (eviscerated body weight/ SL3)x 100. Ovaries and sagittal otoliths were removed Nanmatsu et al.: Reproductive biology of Dexistes rikuzemus 637 An (86) (60) (84) (27) (114) (77) (64) (79) (66) (75) (129) (80) (90) Jan Feb Mar Apr May Jun Jul Month Aug early late Oct Nov Dec Sep Sep Figure 2 Annual changes in the gonadsomatic index (GSI) and body condition IBC) values of female Rikuzen sole {Dexistes rikuzenius). Solid and open circles show the mean values of GSI and BC, respectively. Vertical bars represent the standard deviations of these means. Sample numbers are shown in brackets. within a day after each catch for histological observa- tions and age determination, respectively. The otoliths were washed with distilled water and left to dry until preparation for age determination. Ovaries were fixed in 10% buffered formalin for 24 hours. The middle portions of eyed-side ovaries of 309 specimens were extracted, dehydrated, embedded in paraffin, sectioned at 8 f0.05); however, the sample size was very small; therefore the tests have little power. Discussion Gonadal maturation GSI and histological examinations showed that oocytes develop rapidly from May to August and that the reproductive season lasts from Septem- ber to December; mainly from September to October in the study area. Mature females in the Sendai Bay area were also observed for four months, but the reproductive season in this area occurs from October to January and peaks in November (Ogasawara and Kawasaki, 1980), which was later than the peak documented in the pres- ent study for the area off the Hachinohe coast. The Sendai Bay catch area was located at a lower latitude (37°00'N-38°05'N; Ogasawara and Kawasaki, 1980) than that of the Hachinohe study area (Fig. 1); this difference is relevant because gonadal maturation is usually dependent on water temperature (Kruse and 250 .• * • f 200 e & * ? : : ** kj H £ 150 E 1. * T • _i • co 100 50 01 23456789 Age (years) Figure 4 Relationship between age, including maturity, and the standard length of Rikuzen sole (Dexistes rikuzenius) caught between August and December. Solid circles represent maturing or spent individuals and the open circle at age 1+ represents an imma- ture individual. Tyler, 1983; Asahina and Hanyu, 1983; Conover, 1990). In 2000, the water temperature in the Hachinohe study area decreased faster than that of Sendai Bay in 1977 and 1978 when studied by Ogasawara and Kawasaki (TNFRI1). These results indicate that gonadal matura- tion in Rikuzen sole also depends on water temperature. TNFRI (Tohoku National Fisheries Research Institute). 2004. Unpubl. data. Water temperature data. Tohoku National Fisheries Research Institute, Fisheries Research Agency of Japan. Shiogama City, Miyagi Prefecture 985- 0001 Japan. NarimatSU et a\ : Reproductive biology of Dexistes rikuzenius 643 II .JJI 100 200 300 400 500 600 700 800 900 1000 100 200 300 400 500 600 700 800 900 1000 £ 10 100 200 300 400 500 600 700 800 900 1000 100 200 300 400 500 600 700 800 900 1000 lj|ta 100 200 300 400 500 600 700 800 900 1000 100 200 300 400 500 600 700 800 900 1000 Oocyte diameter (urn) Figure 5 Oocyte diameter distributions just before the spawning season of Rikuzen sole [Dexistes rikuzenius). Oocyte diameter was divided into small-scale (less than 200 j. 350000 c 300000 3 e e o 8 o o 0 • o o J to ° 8 Sq «° o «s to -e •° * • to o i f Relativ o o o o o in rc 250000 ~ 200000 c £ 1 50000 o 0_ e fecundity o o o 100000 i 500 50000 ■ • • 0 2 4 6 8 10 Age (years) Figure 7 Relationship between age (years) and potential fecundity (solid circle. oocyte number/femal e) and relative fecundity (open circle, oocyte number/female per g ) of Rikuzen sole (Dexistes rikuzenius) in the late vitellogenic maturity phase. although they had no other vitellogenic oocytes. There are three potential hypotheses to explain the fate of these primary yolk oocytes. One explanation is that the oocytes are spawned in the current reproductive season. Maddock and Burton (1999) showed that in American plaice (Hippoglossoides platessoides), a group-synchro- nous spawner, the size frequency of oocytes during the prereproductive season was not continuous, whereas during the reproductive season the size frequency was continuous. The reason for this difference was that during the reproductive season cortical alveoli stage oo- cytes are larger than those during the prereproductive season. It is unclear, however, whether these cortical alveoli oocytes will be spawned during the reproductive season (Maddock and Burton, 1999). Although similar to those of the American plaice, all Rikuzen sole ovaries with primary yolk-stage oocytes contained no secondary or more advanced stage oocytes. In addition, oocytes that would be spawned in the current reproductive sea- son developed beyond the secondary yolk stage before the beginning of the reproductive season. Therefore, primary yolk-stage oocytes occurring late in the repro- ductive season might not be spawned that season. Primary yolk-stage oocytes were found from October to August (the late reproductive to vitellogenic season) (Table 3). From October to December only a small per- centage of individuals possessed oocytes in this stage, whereas their ratio increased from January to April. These results indicate that females begin vitellogenesis for the next reproductive season shortly after spawning. This hypothesis is supported by reports that the vitel- logenesis of flatfishes takes a long time (Yamamoto, 1954, 1956; Ishida and Kitakata, 1982; Zamarro, 1992; Harmin et al., 1995). Atretic oocytes were present in low proportions from March to April and in high proportions in May. The mature phase of ovaries with atretic oocytes did not differ from that of ovaries without atretic oocytes. In addition, developmental stage did not differ between atretic and normal oocytes in any ovary. Therefore, it seems that the primary yolk-stage oocytes observed late in the reproductive season will not selectively degener- ate, rather they will be spawned. Decisions regarding maturity and age at maturity POFs were present from September to January and all specimens caught during this period had either oocytes in the advanced yolk stage or POFs in their ovaries. All specimens caught between November and December con- tained ovaries with POFs, whereas they were observed only in a small percentage of the specimens caught in January. The spawning season lasted from September to December, but almost all spawning had finished by October. These results indicate that the duration until resorption of the POFs ranges from a few weeks to two months. For a few weeks immediately following spawn- ing, the presence of POFs can be used as a criterion for the differences between post- and prespawning individu- als. This feature is consistent with that of other flatfish in which POFs degenerate within one or two months (Barr, 1963; Janssen et al., 1995). By noting the presence of POFs and advanced yolked oocytes, we were able to classify individuals as mature or immature. All but one individual caught during the reproductive period were maturing or had spawned. The body size of the mature females ranged from 114 to 237 mm SL, which corresponded to an age from 1 to 8+ years, respectively, whereas the immature female (131 mm SL) was age 1+. These results indicated that most female Rikuzen sole in this population mature at 2 years old, or at the latest at 3 years old, and spawn every year after maturation. Almost all (99.5%) fish caught commercially are adult individuals. Nanmatsu et al : Reproductive biology of Dexistes nkuzemus 645 Fecundity The potential fecundity of group-synchronous spawning fish can be determined prior to the spawning season (Takano, 1989). In Rikuzen sole, oocyte-stage composi- tion became discontinuous beyond the late vitellogenic maturity phase, when a gap was found between secondary or tertiary yolk stages and the late perinucleolus stage. Oocyte diameter distributions in late vitellogenic maturity phase ovaries revealed that oocytes could be divided into small (less than 200 jim) and large (more than 300 jmi) scale groups. Taking into account the oocyte diameters observed in the histological sections, small-scale group oocytes corresponded to cortical alveoli or less advanced stage oocytes, whereas larger oocytes corresponded to secondary yolk or more advanced stage oocytes. The occurrence of atretic oocytes was highest in May and became lower as the season progressed until the end of the spawning season. These phenomena may cor- relate with both annual feeding cycles and maturation. Ogasawara and Kawasaki (1980) showed that in the Sendai Bay population, Rikuzen sole feed actively for a few months after spawning and then feed passively for the next few months. Gut-content weight began to increase again in June. In our study area, BC increased from about May, corresponding to the time when the oocytes begin to mature rapidly. As described before, vitellogenesis in this species takes a long time. Because oocytes are metabolically active in the season when the energetic condition of Rizuzen sole is still recovering, a higher proportion of atretic oocytes occur during this period. Potential fecundity may not correspond to annual fe- cundity because of the presence of atretic and residual oocytes (Witthames and Greer Walker, 1995; Kurita et al., 2003). Therefore, we examined the potential fecun- dity of fish in the late vitellogenic maturity phase just before the spawning season. The frequency of occurrence of atretic oocytes may be underestimated because these oocytes have shrunk and are smaller than the maturing yolked oocytes. In addition, atretic oocytes may occur in the ovaries during the premature maturity phase. How- ever, in our samples a low percentage of atretic oocytes were observed. Only a small percentage of premature ovaries were found on or before the reproductive season; this finding seems to indicate that the oocytes of this species take a short time to develop from the tertiary vitellogenic stage to maturation. These results make clear that potential fecundity differs from annual fecun- dity, but the extent of this difference was nevertheless relatively small in the samples. Moreover, ovulated, but not spawned oocytes were observed in the maturing and spent ovaries; these oocytes have the potential to cause an overestimation of annual fecundity. However, the frequency of ovaries with residual ovulated oocytes was small; therefore, such oocytes may not seriously influence annual fecundity, as with the case of Dover sole (Microstomus pacificus) (Hunter et al., 1992). Vitellogenesis in American plaice was seen to begin soon after spawning ( Zamarro, 1992), as with Rikuzen sole. Separation of oocyte diameter in this species oc- curs approximately three months before the start of the spawning season. In Rikuzen sole, potential fe- cundity was determined as being much closer to the reproductive season. Reproduction occurred from early September, but occurrence of the maturity phase in August varied largely among individuals. The potential fecundity of almost all fish (85%) could be determined until August. These results indicate that certain condi- tions and measurements are necessary when examining potential fecundity without histological methods. Potential fecundity became determinate for the first time at maturity during the late vitellogenic phase. Some of the maturity phase ovaries contained second- ary yolk-stage oocytes and all contained tertiary yolk- stage oocytes. The secondary yolk-stage oocytes ranged in diameter from 260 to 440 jim — a range that does not overlap with the diameter range of primary yolk-stage oocytes (180-220 f6 mm as feeding, based not only on size but also on the amount of yolk present, and whether prey were visible in their gut. These catego- ries were based on observations of fiathead sole larvae Porter: Temporal and spatial distribution and abundance of eggs and larvae of Hippoglossoides elassodon 651 Table 2 The number of stations used each year to assess monthly flathead sole (Hippoglossoides elassodon ) larval distribution in the western Gull of Alaska. Cruise Apr Apr May May Jun Jun year 1-15 16-30 1-15 16-31 1-15 16-30 Jul Aug Sep Oct Nov 1972 i 27 40 — — — — — — 1977 — — — — — — — — 11 48 1978 60 2 — — — 69 20 — 57 67 118 1979 — — — 58 — — — — 18 — 1981 61 123 16 136 — — — — — — — 1982 55 28 — 62 — — — — — — — 1983 — — 1 63 — — — — — — — 1984 63 66 28 — — — — — — — — 1985 87 28 62 135 54 — — — — — — 1986 185 34 89 19 — — — — — — — 1987 177 83 — 58 — 15 4 — — — — 1988 227 64 13 2 1 — — — — — — 1989 128 69 132 34 1 — — — — — — 1990 107 — 90 70 78 — — — 6 — 1991 90 150 119 97 — — — — — — — 1992 94 — 158 136 — — — — — — — 1993 96 — 141 90 24 — — — — — — 1994 4 — 89 133 6 — — — — — — 1995 — — — 98 — — — — — — — 1996 — 59 273 130 — — — — — — — 1997 — — — 10(1 — — — — — — — 1998 — — 72 128 — 26 — — — — 1999 — — 6 233 83 — — — — — — Total 1434 733 1329 1782 247 110 24 0 81 78 166 7 No stations. reared in the laboratory iS. Porter, unpubl. data). In the laboratory, flathead sole larvae hatch with pigmented eyes, three tail pigment bands, and an open mouth (S. Porter, unpubl. data). Flathead sole larvae that were collected from MOCNESS tows and that did not have these features were classified as embryos (it was sus- pected that handling during collection may have caused some of the late stage eggs to prematurely hatch), and their lengths were not included in the weighted mean depth. For eggs and larvae, a weighted mean depth was calculated for each stage or size category, and depths were compared by using ANOVA and the Tukey HSD multiple comparison test. Results Eggs Geographic distribution and abundance Eggs were col- lected as early as March but in small numbers (Figs. 2 and 3A). Most spawning began from early to mid-April (Fig. 3B) near the Kenai Peninsula and then progressed with time southwest into Shelikof Strait and along the Alaska Peninsula. There are two main areas where peak spawning (from early to mid-May) occurred: Shelikof Strait and between the Shumagin Islands and Unimak Island (Fig. 3C). In June, spawning generally declined in these areas and was most intense around Kodiak Island (3D). Eggs were collected as late as July (one station in 1978, on the eastern side of Kodiak Island). Vertical distribution There were similar trends in the vertical distribution of eggs among tows (Fig. 4). Abun- dance peaked at about 20 to 35 m below the surface, decreased at greater depth, and then slightly increased below 125 m. Because the trend of the catches of the tows were similar, we were able to increase sample size in the depth intervals by pooling data from similar depth intervals for further analyses. Eggs were pelagic and most abundant near the surface (mean depth 43 ±10 m) and at the deep sampling depths (mean depth 149 ±6 m); abundance was low in mid-water (Fig. 4). Late-stage eggs (stages 16-21) dominated the depths of 652 Fishery Bulletin 103(4) 1 00 75 ■ 50 25 ■^H egg abundance number oi stations \ . HI ' \ / \ / \ 1 / \ ■ I ' \ ~u 1 1 \ / l_ " / 1 / \ _l \ / \ / , — -■1 II 11 ■- m t r , — , — Month Jvjtt j\M S<#v O1* . high abundance. Early stage eggs were most abundant in mid-water; they accounted for 79% of the total number of eggs collected between 50 and 159 m depth. Sixty- six percent of all eggs collected above 66 m depth were middle- and late-stage A eggs. The largest numbers of late-stage B eggs were found below 124 m depth, where they accounted for 83% of all eggs collected. Mean egg stage depth showed that as the eggs developed from the early stages to the middle stages they rose toward the surface (mean depth of the eggs changed from 54 to 28 m); then in the later stages of development the eggs sank and hatched at depth. Late-stage B eggs were collected significantly deeper (mean depth 90 ±37 m) than late- stage A eggs (mean depth 35 ±7 m; ANOVA, P=0.007; Tukey HSD multiple comparison test, P= 0.006). Larvae Geographic distribution and abundance Larvae were found from early April to October, but they were most abundant from mid-May to mid- June (Fig. 5). From mid- to late April, larvae were most abundant near the Kenai Peninsula (Fig. 6A), and as spring progressed their abundance increased southwest along the Alaska Peninsula (Fig. 6B). Peak abundance occurred during the first two weeks of June in the southern portion of Shelikof Strait (Fig. 6C). From mid- to late June larvae were most abundant on the east side of Kodiak Island (Fig. 6D). Although most of the surveys were conducted in this area, it is possible that larvae may have been abundant elsewhere in the study area during this time. From July through October, only the area east of Kodiak Island was surveyed, and larval abundance there was low. Larval drift Satellite-tracked drifters released in May 1994 and drogued at 40 m indicated that the Alaska Coastal Current flow was strong and moving to the southwest — typical surface current flow for this area (Bailey4). In May 1996, drifters showed that flow was weak, disorganized and moving somewhat to the north- east (Bailey et al., 1999). In early May 1994, very few flathead sole larvae were collected; therefore the center point of the flathead sole egg distribution was used to infer the starting location of larval drift. Size-at-age data have shown that the growth rate for flathead sole larvae is 0.3 mm/day in Auke Bay, Alaska (Haldorson et al., 1989). Using this growth rate, we determined that larvae hatched in early May could have grown as much as 6 mm in the 21 days between surveys. The size class of larvae greater than 9 mm was assumed to include larvae that had hatched from the eggs present in early May. The location of the centers of distribution of the early May eggs and late May larvae indicated that the larvae had drifted southward over the continental shelf (Fig. 7). In 1996 all the larvae collected in early May were 7.1 mm and smaller (range 4.2 to 7.1 mm). The area was surveyed 26 days later, and growth of about 8 mm could have occurred between surveys. For larvae collected at the end of May, the size group longer than 12 mm was assumed to include the early May larval group. The location of the centers of distribution of the early May and late May larvae showed that the larvae were retained at nearly the same location (Fig. 8). 4 Bailey, K. M. 2002. Personal commun. NOAA, Alaska Fisheries Science Center, 7600 Sand Point Way NE, Seattle, WA 98115. Porter: Temporal and spatial distribution and abundance of eggs and larvae of Hippoglossoides elassodon 653 B 172° 168° 164' 160° 148" 144" 172' 168° 164° 160 156" 152' 148° 144" 168° 164° 160° 156 164° 160° 156" 152° 148 "W C D 172° 168" 164" 160' 156 152° 148° 144" 172° 168° 164° 160° 156° 152° 148" 144" N 60 A 58 56" 54°- y^ *>t > III' y 52° early to mid-May 164° 160° 156° 152° 148° 168° 164° 160° 156° 152 148 W egg abundance per 10 m2 ZD0 I !0-10 ; 10-50 □ 50-100 5iiS 1 00-200 I > 200 Figure 3 The geographic distribution of flathead sole (H. elassodon) eggs in the western Gulf of Alaska during the spawning season; I A) March, (B) early to mid-April, (C) early to mid-May, ID) early to mid-June. Vertical distribution There were similar trends in the vertical distribution of larvae among tows (Fig. 9). Abundance peaked at about 15 to 30 m below the sur- face, then decreased, and larvae were collected from the deepest sampling depth interval from one tow (1996A; Fig. 9). Because the tows were alike, to increase sample size in the depth intervals, we pooled data from simi- lar depth intervals but from different tows for fur- ther analyses. Larval abundance was highest near the surface and at the deepest depths sampled (Fig. 9). In Auke Bay, Alaska, flathead sole larvae migrated vertically at night no more than 15 m, ending at 20 m depth, and they were less aggregated (Haldorson et al., 1993). This depth was much shallower than the depth at which larvae and late-stage eggs were collected in tow 1996A (sampling depth interval was 174-236 m). Therefore the deep concentration of larvae in 1996 was probably due to eggs hatching rather than to vertical migration. The deepest sample comprised embryos and larvae (the larvae, however, were too damaged to deter- mine whether they were prefeeding or feeding larvae), and samples collected above 100 m were a mixture of embryos and prefeeding larvae (29%), and feeding larvae (71%). The smallest larvae (<5 mm) were found in deepest water (mean depth 166 ±32 m), and larger larvae (>5 mm) were found in shallower water (above about 60 m depth; ANOVA, P<0.001; Tukey HSD mul- tiple comparison test, P<0.001). The size distribution of the larvae indicated that soon after hatching they rise to the surface to feed. Discussion Flathead sole inhabit the continental shelf of the North Pacific Ocean, and the area used for the present study o54 Fishery Bulletin 103(4) Catch/10 m2 0 100 200 300 400 500 600 0 H 25 - iP>< __-a w_. „▼- -~ ~ . — — • 50 ■ It L — ^ — 75 - 1 10° 1 F — o— 1991 w 125 -» Q *-^^_ —a— 1996A 150 • i ~^^^^^^ t 1996B 175 - \ 200 - V 225 J Figure 4 The vertical distribution of flathead sole iH. elassodon) eggs collected from four MOCNESS tows conducted in 1991, 1993, 1996 during peak spawning. Symbols indicate the mean of the depth interval that the samples were collected in. 40 -i m^^m larva] abundance number of stations \ - 1500 30 - \ z E o ~CD £ 20 - CD "D C < \ \ ■ 1000 § D" o s ■ 500 % 10 - e*\^V^1^>*V^ ^ S°VV °* ^ Month Figure 5 The mean abundance of flathead sole (H. elassodon) larvae in the western Gulf of Alaska during the year. Standard deviation and number of stations used for each time period are also shown. Abundance in early April and October was very low, 0.06 and 0.11 larvae/10 m2, respectively. contains the highest relative abundance of adult flathead sole off the west coast of North America (Wolotira et al.1). Generally, outside the study area the abundance of adult flathead sole is low (Wolotira et al.1); therefore these areas most likely had very little effect on the abundance of eggs and larvae collected from within the study area. In the western Gulf of Alaska flathead sole spawn in three main areas during the spring: near the Kenai Peninsula, in Shelikof Strait, and in the area between the Shumagin Islands and Unimak Island. Spawning progresses in a southwesterly direction along the Alas- ka Peninsula. Flathead sole in spawning condition are abundant from March through May (Hirschberger and Porter: Temporal and spatial distribution and abundance of eggs and larvae of Hippoglossoides elassodon 655 A B 172° 168° 164 160 156 152 148' 144 172= 168° 164 160 156 152" 148° 144 \ 60 A 58 WJ- 56 54° 52° mid- to late April N A ErJ"""! ,# • jp~& #-' II 'III' |f l mid- to late May 168° 164° 160° 156 152° 148" 168" 164 160° 156 152' 148'W C D 172° 168: 164° 160° 156 152° 148° 144° 172° 168° 164 160= 156 152° 148° 144° N 60 A 58 56 ■■*;=. 54° Willi 52° early to mid-June 168° 164° 160° 156° 152° 148 168° 164° 160° 156° 152 148°W egg abundance per 10 m; 0-10 ] 10-50 ZZI 50-100 &£i 100-200 I =■ 200 Figure 6 The geographic distribution of flathead sole (H. elassodon) larvae in the western Gulf of Alaska during months of the spawning season. (A) mid- to late April. (B> mid- to late May, (C) early to mid-June, (D) mid- to late June. Smith2), which correlates with the period when eggs were collected in the present study. Peak spawning oc- curred from early to mid-May. and by the end of June spawning was nearing completion. Larval abundance peaked from early to mid-June in the southern portion of Shelikof Strait. In late July, late-stage flathead sole larvae were the most abundant of larval fish collected in the Gulf of Alaska between the Semidi Islands and Unimak Island (Brodeur et al., 1995). Flathead sole larvae have also been found on the east side of Ko- diak Island during the summer (Kendall and Dunn. 1985). Laboratory observations of the changes in density of flathead sole eggs during development are inconsistent. Results of one study showed that egg density decreased throughout development to hatching (Alderdice and Forrester, 1974). Another study found that up to 24 hours before hatching the eggs floated at the surface of a container and then sank to the bottom and hatched (Miller, 1969), indicating that density had increased late in development. A field study of the vertical distri- bution of Atlantic halibut iHippoglossus hippoglossus) eggs in Norwegian fjords showed that later stage eggs had a higher density (and were found deeper) than earlier egg stages (Haug et al., 1986). Results from the present study support the findings of Miller (1969). in that the density of flathead sole eggs in the present study appeared to increase near the time of hatch- ing. For the larvae of both the arrowtooth flounder (Atheresthes stomas) and Pacific halibut {Hippoglossus stenolepis), small larvae were found deep and larger sizes migrated towards the surface (Bailey and Pic- quelle, 2002). In the present study, flathead sole larvae had a similar vertical distribution pattern indicating that after hatching in deep water they rise to near the surface to feed. 656 Fishery Bulletin 103(4) 162° 161= 58° - 57°- 56° 55°- 160° 159° 158° 157° 156" 155" 154° 153° strong southwesterly, current flow .A* x^ ,N%' ^ 4-° ^ ! 200 m •58° *cPv 200 f "^ 57° ■56° 55°N 161° 160 159° 158° 157° 1 56" 155° 154° 153°W + center of distributionof eggs in early May + center of distribution of larvae greater than 9 mm in late May (assumed to include larvae that hatched from early May eggs) Figure 7 The drift of flathead sole (H. elassodon) larvae in Shelikof Strait during May 1994. Surface current flow was strong and southwesterly. 162° 161" 160" 159° 158° 157° 156° 155° 154° 153° 58° 57° 56°- 55° .# <** weak northeasterly or l/4r'l disorganized current fj; /[ flow s «*, * <*. ■ *•><, 200 r MB 200 r 161 160 159 158" 157° 156° 1 55" 154° ■57° 58 56° 55° N 153°W + center of distribution of all early May larvae (mean length 5.71 ±0.71 mm) ■^ center of distribution of late May larvae greater than 12 mm (assumed to include larvae present in early May) Figure 8 The drift of flathead sole (H. elassodon I larvae in Shelikof Strait during May 1996. Surface current flow was disorganized, or weak, and to the northeast. Porter: Temporal and spatial distribution and abundance of eggs and larvae of Hippoglossoides elassodon 657 Catch/10 m2 0 10 20 30 40 0 1 25 \^ ... T 50 Tgf— 1/b 75?/ E 100 i CD Q 125 I 150 H —a— 1991 — •— 1993 — o— 1996A t 1996B 175 ' 200 - 225 - Figure 9 The vertical distribution of flathead sole IH. elassodon) larvae collected from fo it MOCNESS tows conducted in 1991. 1993, and 1996 during peak sp awning. Symbols indicate the mean of the depth interva 1 that the samples were collected in. Some species of flatfish spawn offshore (e.g., ar- rowtooth flounder and Pacific halibut, Bailey and Pic- quelle, 2002), but the present study has shown that flathead sole spawn on the continental shelf. Flathead sole nursery areas have been found to be in the bays of the Alaska Peninsula and Kodiak Island (Norcross et al., 1999), and it is crucial that the larvae remain on the shelf near their nursery areas. Changes in egg density may be a mechanism for retaining flathead sole larvae on the shelf. For arrowtooth flounder and Pacif- ic halibut larvae in the western Gulf of Alaska, it has been suggested that deep water currents (100-400 m depth in sea valleys and in troughs in the continental shelf) transport these larvae from the offshore areas where they hatch to their nearshore nurseries (Bai- ley and Picquelle, 2002). By sinking when they are nearing hatching, flathead sole eggs that have drifted southwesterly (i.e. away from nursery areas) with the surface currents can be brought back (along with newly hatched larvae) toward inshore juvenile nursery areas. Alternatively, the act of sinking as they near hatching may be a way for newly hatched larvae to avoid preda- tion by keeping them out of the surface waters where they are likely to encounter predators. The physical environmental conditions of Shelikof Strait may also serve to retain flathead sole larvae on the shelf. In May 1994 when the Alaska Coastal Current flow was strong and to the southwest, larvae drifted southward but remained on the continental shelf. In May 1996 when the flow was weak, disorganized, and moving somewhat to the northeast, the larvae remained at vir- tually the same location for the entire month because surface current flow in Shelikof Strait was weakened and reversed because of anomalous atmospheric con- ditions. Under both flow regimes larvae remained on the continental shelf in southern Shelikof Strait. Ed- dies may also be an important retention mechanism for flathead sole larvae because entrainment in one of these could slow drift. Under typical conditions in Shelikof Strait (i.e., strong southwesterly current flow), eddies frequently occur and they drift slower than the water surrounding them (Kendall et al., 1996). They can also remain nearly stationary for two weeks (Schumacher et al., 1993). Both biological and environ- mental factors may work together to retain flathead sole larvae on the continental shelf and keep them near their nursery areas. Acknowledgments I would like to thank Debbie Blood and Angie Lind for determining developmental stages of flathead sole eggs, and Susan Picquelle for assistance with egg and larval distribution charts. Kevin Bailey and Jeff Napp provided helpful comments on an early draft of this manuscript. Two anonymous reviewers offered improvements. This research is contribution FOCI- 0475 to NOAA's Fisheries-Oceanography Coordinated Investigations. Literature cited Alderdice, D. R, and C. R. Forrester. 1974. Early development and distribution of the flathead sole (Hippoglossoides elassodon). J. Fish. Res. Board Can. 31:1899-1918. 658 Fishery Bulletin 103(4) Bailey, K. M., N. A. Bond, and P. J. Stabeno. 1999. Anomalous transport of walleye pollock larvae linked to ocean and atmospheric patterns in May 1996. Fish. Oceanogr. 8:264-273. Bailey, K. M., and S. J. Picquelle. 2002. Larval distribution of offshore spawning flatfish in the Gulf of Alaska: potential transport pathways and enhanced onshore transport during ENSO events. Mar. Ecol. Prog. Ser. 236:205-217. Bailey, K. M., P. J. Stabeno, and D. A. Powers. 1997. The role of larval retention and transport fea- tures in mortality and potential gene flow of walleye pollock. J. Fish Biol. 51 Isuppl. A):135-154. Blood, D. M., A. C. Matarese, and M. M. Yoklavich. 1994. Embryonic development of walleye pollock, Ther- agra chalcogramma, from Shelikof Strait, Gulf of Alaska. Fish. Bull. 92:207-222. Brodeur, R. D., M. S. Busby, and M. T. Wilson. 1995. Summer distribution of early life stages of wall- eye pollock, Theragra chalcogramma, and associated species in the western Gulf of Alaska. Fish. Bull. 93:603-618. Haldorson, L.. A. J. Paul, D. Serritt, and J. Watts. 1989. Annual and seasonal variation in growth of larval walleye pollock and flathead sole in a southeastern Alaska bay. Rapp. P.-V. Reun. Cons. Int. Explor. Mer 191:220-225. Haldorson, L., M. Prichett, A. J. Paul, and D. Ziemann. 1993. Vertical distribution and migration offish larvae in a northeastern Pacific bay. Mar. Ecol. Prog. Ser. 101:67-80. Haug, T., E. Kj0rsvik, and P. Solemdal. 1986. Influence of some physical and biological factors on the density and vertical distribution of Atlantic halibut Hippoglossus hippoglossus eggs. Mar. Ecol. Prog. Ser. 33:207-216. Kendall, A. W., Jr., and J. R. Dunn. 1985. Ichthyoplankton of the continental shelf near Kodiak Island Alaska. NOAA Tech. Rep. NMFS 20, 89 p. Kendall, A. W., Jr., and S. J. Picquelle. 1989. Egg and larval distributions of walleye pollock Theragra chalcogramma in Shelikof Strait, Gulf of Alaska. Fish. Bull. 88:133-154. Kendall, A. W., Jr., J. D. Schumacher, and S. Kim. 1996. Walleye pollock recruitment in Shelikof Strait: applied fisheries oceanography. Fish. Oceanogr. 5 (suppl. 11:4-18. Matarese, A. C., A. W. Kendall Jr., D. M. Blood, and B. M. Vinter. 1989. Laboratory guide to early life history stages of northeast Pacific fishes. NOAA Tech. Rep. NMFS 80, 652 p. Miller, B. S. 1969. Life history observations on normal and tumor- bearing flathead sole in East Sound, Orcas Island (Washington). Ph.D. diss., 131 p. Univ. Washington, Seattle, WA. Norcross, B. L., A. Blanchard, and B. A. Holladay. 1999. Comparison of models for defining nearshore flat- fish nursery areas in Alaskan waters. Fish. Oceanogr. 8:50-67. Rose, C. S. 1982. A study of the distribution and growth of flathead sole (Hippoglossoides elassodon). M.S. thesis, 59 p. Univ. Washington, Seattle, WA. Schumacher, J. D., P. J. Stabeno, and S. J. Bograd. 1993. Characteristics of an eddy over a continental shelf: Shelikof Strait, Alaska. J. Geophys. Res. 98: 8395-8404. Theilacker, G. H., and S. M. Porter. 1995. Condition of larval walleye pollock, Theragra chalcogramma, in the western Gulf of Alaska assessed with histological and shrinkage indices. Fish. Bull. 93:333-344. Watts, J. D. 1988. Diet and growth of first-year flathead sole (Hip- poglossiodes elassodon) in Auke Bay, Alaska. M.S. thesis, 80 p. Univ. Alaska, Juneau, AK. 659 Abstract — With a focus on white mar- lin {Tetrapturus albidus), a concurrent electronic tagging and larval sampling effort was conducted in the vicinity of Mona Passage (off southeast His- paniola), Dominican Republic, during April and May 2003. Objectives were 1) to characterize the horizontal and vertical movement of adults captured from the area by using pop-up satel- lite archival tags (PSATs); and 2) by means of larval sampling, to investi- gate whether fish were reproducing. Trolling from a sportfishing vessel yielded eight adult white marlin and one blue marlin (Makaira nigricans); PSAT tags were deployed on all but one of these individuals. The excep- tion was a female white marlin that was unsuitable for tagging because of injury; the reproductive state of its ovaries was examined histologically. Seven of the PSATs reported data summaries for water depth, tempera- ture, and light levels measured every minute for periods ranging from 28 to 40 days. Displacement of marlin from the location of release to the point of tag pop-up ranged from 31.6 to 267.7 nautical miles (nmi) and a mean dis- placement was 3.4 nmi per day for white marlin. White and blue marlin mean daily displacements appeared constrained compared to the results of other marlin PSAT tagging stud- ies. White marlin ovarian sections contained postovulatory follicles and final maturation-stage oocytes, which indicated recent and imminent spawn- ing. Neuston tows (/i=23) yielded 18 istiophorid larvae: eight were white marlin, four were blue marlin, and six could not be identified to species. We speculate that the constrained movement patterns of adults may be linked to reproductive activity for both marlin species, and, if true, these movement patterns may have several implications for management. Protection of the potentially impor- tant white marlin spawning ground near Mona Passage seems warranted, at least until further studies can be conducted on the temporal and spatial extent of reproduction and associated adult movement. Movements and spawning of white marlin (Tetrapturus albidus) and blue marlin (Makaira nigricans) off Punta Cana, Dominican Republic Eric D. Prince1 Robert K. Cowen2 Eric S. Orbesen' Stacy A. Luthy2 Joel K. Llopiz2 David E. Richardson2 Joseph E. Serafy' 1 Southwest Fisheries Science Center National Marine Fisheries Service 75 Virginia Beach Drive Miami, Florida 33149 E-mail address (for E D Prince): eric pnnce@noaa gov 2 Rosenstiel School of Marine and Atmospheric Science Division of Marine Biology and Fisheries University of Miami 4600 Rickenbacker Causeway Miami, Florida 33149 Manuscript submitted 24 June 2004 to the Scientific Editor's Office. Manuscript approved for publication 31 March 2005 by the Scientific Editor. Fish. Bull. 103:659-669 (2005). White marlin {Tetrapturus albidus) and blue marlin (Makaira nigricans) are widely distributed throughout the tropical and temperate waters of the Atlantic Ocean and adjacent seas; the former species is endemic only to the Atlantic Ocean (Mather et al., 1975). Genetic analyses and tag recapture data have indicated that each spe- cies has a single Atlantic-wide popu- lation (ICCAT, 1998). Several stock assessment indicators indicate that the white marlin population has been severely overfished for several decades (ICCAT, 2001, 2002). The Atlan- tic blue marlin stock is also heavily over-exploited, but to a lesser degree. The main source of adult mortality for both stocks is the multinational offshore longline fisheries that, in the process of targeting tunas (Scombri- dae) and swordfish (Xiphias gladius), land the marlins as bycatch (ICCAT, 2002, 2003). Despite their economic and ecologi- cal value, little is known about the biology and ecology of Atlantic mar- lins (Prince and Brown, 1991). This is especially true regarding the repro- ductive biology of white marlin and adult movement patterns in spawning areas (Baglin, 1979; Mather, 1975; White Marlin Status Review Team1; SEFSC2). Long-term (i.e., >40 years) commercial (Goodyear, 2003) and rec- reational (i.e., Cabeza de Toro Billfish Tournament, Graves and McDowell, 1995; Casilla3) fishing records indi- cate that, every spring, white marlin are present in relatively high numbers off the southeastern coast of Hispan- iola. This observation, coupled with 1 White Marlin Status Review Team. 2002. Atlantic White Marlin Status Review Document, 49 p. Report to National Marine Fisheries Service, South- east Regional Office, 263 13th Avenue, St. Petersburg, FL 33701-5511. 2 SEFSC ( Southeast Fisheries Science Cen- ter). 2004. Atlantic Billfish Research Plan. National Marine Fisheries Ser- vice, Southeast Fisheries Science Center, 75 Virginia Beach Drive, Miami, FL 33149-1003. 3 Casilla, W. 2003. Personal commun. Club Nautico de Santo Domingo, Calle Juan Baron Fajardo #2, Ensanche Iantini, Santo Domingo, Dominican Republic. 660 Fishery Bulletin 103(4) anecdotal information about gravid fish, prompted the present examination of adult movements in a potentially important, but as yet unconfirmed, spawning location. The present study was conducted off Punta Cana, Dominican Republic, during April and May 2003. Objec- tives were 1) to characterize the horizontal and vertical movement of adult white marlin captured from the area using pop-up satellite archival tags (PSATs) and 2) to investigate by larval sampling, whether marlin were reproducing at this location. Materials and methods Deployment of PSAT tags on adult marlin was conducted from a 17-m charter fishing vessel by using standard trolling gear (9/0 long-shaft J hooks) and dead bait. Wildlife Computers Inc. (Redmond, WA) PAT 3 model tags were used. This tag allows the user to program pop-up date, sampling interval, criteria for premature release, bin demarcations for sampling temperature and pressure (depth), as well as transmission and memory priorities. These tags were programmed to sample depth (pressure), temperature, and light once every minute and the depth and temperature records were summa- rized into histograms at 3-hour intervals. A pressure- activated mechanical detachment device was also used which severs the monofilament tether at a depth of about 1500 m — well before the 2000 m depth at which the tag is crushed and disabled. This feature helps prevents data loss in the event of fish mortality. All PSAT tags were rigged similarly according to methods described by Graves et al. (2002). Billfish han- dling and tagging procedures and associated devices reviewed by Prince et al. (2002a) were also used. The target area for tag placement was about 4 to 5 cm ven- tral to the dorsal midline, adjacent to the first several dorsal spines. An effort was made to insert the anchor through the dorsal midline, pterygiophores, and connec- tive tissue to a depth just short of the anchor exiting the opposite side of the fish. In addition, a conventional streamer tag (series PS) was placed in the fish well posterior to the PSAT tag, according to standard pro- cedures (Prince et al., 2002a). Two devices were used during tagging which tend to reduce stress in captured fish and to aid in proper tag placement. The first was a "snooter" (a wire snare housed in a 1.5-m PVC tube), which secures to the up- per bill and allows the tagger to maintain control of the fish while its head remains beneath the water during the tagging procedure (Prince et al., 2002a). The second was a small hook "gaff" (a long shaft 9/0 hook with point and barb removed) to manipulate the position of the fish in relation to the tagging vessel. Captured fish were resuscitated for 3 to 15 minutes, depending on their apparent state of exhaustion, by moving the vessel ahead at two to three knots while maintaining control of the fish with the snooter. State of exhaustion was inferred from coloration, fight time, and signs of sluggish movement. One white marlin died during tagging and was re- tained for examination of its reproductive status. Whole or quarter transverse sections of ovarian tissue were preserved in 10% formalin. Preparation for histologi- cal analysis followed McBride et al. (2002). Histologi- cal determination of spawning activity was based on oocyte classification and the presence of postovulatory follicles (Wallace and Selman, 1981; Hunter and Mace- wicz, 1985; Hunter et al., 1992). Once adult marlins were located for tagging, neuston sampling was conducted from the same fishing vessel with methods similar to those reported by Serafy et al. (2003). In the present study, ten-minute daytime tows were performed with two neuston nets. Both nets had 1000-fim mesh and were attached to 1 mx0.5 m or 2 mx 1 m rectangular aluminum frames. Water volume filtered was measured with a mechanical flow meter; station coordinates and water column depth measure- ments were obtained by using a hand-held geographical positioning system and depth sounder. Neuston collec- tions were made along a series of transects that covered the general area of the recreational fishery for white marlin at this location (Fig. 1). The neuston samples were initially stored in 150 proof white rum. Upon returning to the laboratory (i.e., within 24-96 hours) they were transferred to 95% ethanol. Billfish larvae were sorted from the samples and measured by using Image Pro image analysis software (Image Pro Plus, version 4.5, Media Cybernetics, Inc. Silver Spring, MD). Larval identification was conducted by using restriction fragment length polymorphism analysis of the nuclear MN32-2 locus following the methods of McDowell and Graves (2002). Results Seven white marlin and one blue marlin were tagged with PSAT tags off Punta Cana, Dominican Republic, between April 23-24 and May 14-17 2003 (Table 1). All but two tags were programmed to pop-up after 30 days; the exceptions were 40-day deployments for one white marlin and one blue marlin. One of eight PSATs (deployed on a white marlin) failed to transmit data and one white marlin died prior to release (see below) from hook-related injuries. The displacements of the six white marlin from the original point of release ranged from 31.7 to 267.7 nmi (58.7 to 495.8 km), whereas the dis- placement for the blue marlin was 219.3 nmi (406.2 km. Table 1, Fig. 2). Displacements per day for white marlin ranged from 1.1 to 7.2 nmi (average of 3.4 nmi). Cor- responding daily displacement for the one blue marlin was 5.48 nmi (Table 1). The minimum and maximum depth and temperatures monitored for the seven PSAT-tagged marlin during the 30- and 40-day deployments showed that on most days, marlin visited depths 2IOO m (Fig. 3). Minimum temperatures ranged from 16.8° to 20.6°C, whereas the maximum temperatures ranged from 28.2° to 30.0°C. In all cases, the minimum depths for each fish monitored Prince et al.: Movements and spawning of Tetrapturus albidus and Makaira nigricans 661 18.7 N 18.8 N 186N 18.4 N 18.2 N 18.5 N 68.3 W 68.1 W 68.6 W 68.4 W 68.2 W 68.0 W Longitude x- 18.5 N 68.3 W 68.1 W Longitude Figure 1 (A) Western part of Mona Passage off Punta Cana, Dominican Republic, showing the general area of the recreational fishery for white marlin [Tetrapturus albidus, rectangle) and larval sampling (oval); (B) April 23-24 sampling stations and (C) May 13-17 sampling stations. X = stations with no billfish larvae. □ = stations with white marlin larvae, A = stations with blue marlin [Makaira nigricans) larvae, • = stations with unidentified larval istiophorids. Larger markers indicate two billfish in sample; smaller markers indicate one billfish in sample. Depth contours are in meters. during April and May were recorded at the surface, whereas maximum depths ranged from 184 to 368 m (Fig. 3). In one case (i.e., PC-WHM01), the minimum and maximum temperatures and depths converged at the surface, indicating constrained vertical movement for this individual. However, in the majority of tracks there was a clear separation of minimum and maximum temperature and depth (e.g., PC-WHM02, Table 1), indicating that active vertical movements were made each day. Only one of the transmitting tags appeared to pop-up prematurely (PC-WHM01, Fig. 3). This tag disengaged from its white marlin host during a deep dive (368 m) after 28 days at large (two days early). Al- though the fate of this fish cannot be determined, death is a distinct possibility. In general, all marlin spent a high proportion of the time in which they were moni- tored in the upper 25 m and at temperatures a28°C. For example, marlin spent from 50% to 60% of the time in the first depth bin (0 to 25 m) and about 60% to 75% of their time in the 28° to 30°C temperature bin (Fig. 4). Both marlin species made dives down to 100-200 m or more on a fairly consistent basis but generally stayed at these depths less than 10% of the time (Fig. 4). One female adult white marlin, measuring 157 cm lower jaw fork length, could not be resuscitated during pop-up satellite tagging, presumably because of damage caused by a hook that penetrated the stomach. Based on length-weight conversion equations (Prager et al., 1995), the estimated weight of this fish was 21.6 kg (47.2 pounds). The histologically examined ovaries con- tained distinct postovulatory follicles, indicating that spawning likely occurred within the previous 24 hours (Fig. 5, upper panel). In addition, imminent spawning (likely within the following 12 hours) was indicated by 662 Fishery Bulletin 103(4) 20 N :-.fi W 60 W 0 60 120 240 360 4B0 ■ Nautical Miles Figure 2 Displacement vectors (from point of tag release to point of tag pop-up in nautical miles, nmi) for six white marlin {Tetrapturus albidus) (solid lines, 31.7-268 nmi) and one blue marlin (Makaira nigricans) (dashed line, 219 nmi) released off Punta Cana, Dominican Republic, bearing pop-up satellite archival tags during April and May 2003. Tags were programmed for either 30- or 40-day deployments. Table 1 Summary of pop-up satellite archival tag information for seven white marlin iTetrapturus albidus, WHM) and one blue marlin iMakaira nigricans, BUM I released from recreational gear in the vicinity of Punta Cana, Dominican Republic, April and May 2003. Net displacements are given in nautical miles (nmi) and kilometers (km). Compass direction (in degrees) indicates the bearing from point of tag release to point of first transmission. Dashed line indicates that no value was available. Tag number Days monitored Estimated weight pounds (kg) Location of release Location of first transmission Compass direction Net displacement nmi (km) Displacement per day nmi (km) PC-WHM-01 28 40(18.14) 18.49°N, 68.38:W 19.17°N, 68.26°W 9.52° 41.21 (76.32) 1.47(2.72) PC-WHM-02 31 40(18.14) 18.60:N, 68.27°W 19.56°N, 66.58°W 58.87° 111.87(207.18) 3.61(6.69) PC-WHM-03 31 50(22.68) 18.49°N, 68.37°W 19.14°N, 66.25°W 71.81° 126.76(234.76) 4.09(7.57) PC-WHM-04 30 35(15.88) 18.69'N, 68.27°W 18.16°N, 68.28°W 181.03° 31.68(58.67) 1.06(1.96) PC-WHM-05 30 50(22.681 18.70'N, 68.29°W 17.81°N, 66.70°W 120.11° 105.22(194.87) 2.84(5.26) PC-WHM-06 0 50(22.68) 18.29°N, 68.13W — — — — PC-WHM-07 37 60(27.22) 18.60°N, 68.30W 14.12°N, 68.38°W 181.00° 267.73 (495.84) 7.24(13.41) PC-BUM-01 40 130(58.97) 18.49°N, 68.38°W 16.75°N, 65.01°W 117.78° 219.32(406.18) 5.48(10.51) Prince et al : Movements and spawning of Tetrapturus albidus and Makaira nigricans 663 PC-WHM-01 PC-WHM-04 0 -I 100 ■ g" 200- £ 300 ■ Q. CD Q 400 • 500 • \'v' "'■' ■ ■ 37 •32 ■ 27 ■ 22 ■ 17 WW II] -1 100-lAil».JUt cr. uf\j/|wvv^ -1 "g" 200- 2. £ 300- i'.v' •J CB 1. 1> a a 400 ■ o "" 500- WW : 37 32 3 TD CD 27 S c 22 S O 17 — ■^fl^Ni^ 0 200 400 600 800 1000 0 200 400 600 800 1000 Time (h) Time (h) PC-WHM-02 PC-WHM-05 Depth (m) n c/i *- cj rv> -» 3 O O O O O 3 O O O O O O • 37 • 32 ■ 22 ■ 17 0- 1 £ 2«0- cd 3 S ^SOD- S' Q- S ^400- £} 500- wS ! _37 ti CD -32 3 CD - 27 S ■22 § ■17 O 600 ' ) 200 400 600 800 10 Time (h) 00 0 200 400 600 SOU 1000 Time (h) PC-WHM-03 °1 _. 100- c? 3 200- "O .-^ CD £ S ■=■ 300- 1 g- "00- O ° 500 ■ PC-WHM-07 T42 '37 S^ CD • 32 3 -o CD •27 S 22 3 ' 17 O 100 200 E — 300 .C 0 400 Q 500 ^^gr ■ 37 ■ 32 ■ 27 ■ 22 ■ 17 g-M^f" D 200 400 600 800 1C 00 0 200 400 600 800 1000 Time (h) Time (h) PC-BUM-01 n An Depth (m) 3 O O O O O 3 O O O O O C ^4^^^^ -37 - 32 3 -D CT> ►27 3 c -22 3 ■17 S 0 200 400 600 800 1000 Time (h) Depth Temperature Figure 3 Minimum and maximum depth and temperature per 3-hour time intervals for six white marlin (Tetrapturus albidus) and one blue marlin {Makaira nigricans) monitored with pop-up satellite archival tags. Tags were programmed to deploy for either 30 or 40 days, April and May 2003, in the vicinity of Punta Cana, Dominican Republic. WHM = white marlin, BUM = blue marlin. 664 Fishery Bulletin 103(4) White marlin Blue marlin 0-25 26-50 51-75 76-100 101-125 126-150 151-175 176-200 201-225 226-250 >250 Depth bins (m) White marlin Blue marlin 0<12 12-14 14-16 16-18 18-20 20-22 22-24 24-26 26-28 28-30 30-32 32-60 Temperature bins (m) Figure 4 Total time at depth (upper panel) and time at temperature (lower panel) for white marlin iTetrapturus albidus) and blue marlin (Makaira nigricans) tagged with popup satellite archival tags during April and May 2003. Tags monitored marlin for 30 and 40 days. some oocytes exhibiting an early state of final oocyte maturation, including migration of the nucleus towards the oocyte periphery and yolk coalescence (Fig. 5, lower panel). A total of 23 neuston net tows were made in the gen- eral area of the recreational fishery from 23 April to 17 May 2003 (Fig. 1). These tows yielded 18 larval billfishes. Molecular identification was successful for 12 larvae: 8 white marlin and 4 blue marlin (Table 2). Half of the white marlin larvae were 3-4 mm standard length (SL), two were 4-5 mm SL, one was 6.2 mm SL, and one was 12.1 mm SL (Fig. 6). The one positively identified blue marlin larva captured in April was 4.6 mm SL; the remainder taken in May were 3.5 mm SL, 5.1 mm SL, and 10.4 mm SL. Sizes of the six unidenti- fied billfish larvae ranged from 3 to 6 mm SL (Fig. 6). Discussion Larval sampling with neuston tows and histological analyses of adult ovaries confirmed spawning activity of white marlin in the vicinity of Punta Cana during April and May (2003). Co-occurrence of larval blue marlin and white marlin in samples indicated that the two species share this spawning location. White and blue marlin spawning activity in the vicinity of Punta Cana, as indicated from the data presented in our study, Prince et al Movements and spawning of Tetrapturus albidus and Makairo nigricans 665 also coincided in time and space with the fishing activity of the recreational white niarlin fishery that has operated each spring at this location for over 40 years. From PSAT tag data, adult white and blue marlin caught at this time and in this area appeared to exhibit similar vertical and horizontal movement pat- terns in terms of time at depth, time at temperature, average horizontal dis- placement per day, net horizontal dis- placement, and direction of dispersion (compass heading). Movements Average displacement per day is one possible measure to characterize daily horizontal movement activity. We exam- ined this metric in other PSAT stud- ies on marlin and compared them with our results (Fig. 7). Graves et al. (2002) monitored eight blue marlin with PSAT tags caught off Bermuda in July 2000 for periods of 5 days each and reported net displacement vectors ranging from 7.8 to 26.4 nmi/day (mean displacement for the eight fish was 17.5 nmi/day). Kerstetter et al. (2003) also monitored blue marlin during the summer months with PSAT tags in the northwest Atlantic (for 5 and 30 days) and found that displacements ranged from 15.1 to 39.2 nmi/day (mean for six fish was 22.9 nmi/day). Net dis- placement findings (17.5 and 22.9 nmi/ day), presumably for blue marlin spawn- ing times (summer months) from these two studies were roughly 5 to 6.5-fold higher than the average displacements for white marlin reported in our present study (about 3.3 nmi/day) and were 3 to 4-fold higher than the average displace- ment for the one blue marlin monitored in our study (Fig. 7). A recent report (Graves and Horodysky4) has provided displacement movements of white marlin monitored with PSAT tags for 5 to 10 day periods from three different areas in the Northwest Atlantic during May (Punta Cana, Dominican Republic), August- September (U.S. Mid-Atlantic waters), and November (La Guaira, Venezuela) 2002. Only the work in Punta Cana was conducted during the presumed spawn- ing season for white marlin. Average displacements for these areas were 9.6 Figure 5 Upper panel. A postovulatory follicle (POF), advanced yolked oocyte (AYO), and follicle (F) are shown in section of gonad from a female white marlin (Tetrapturus albidus) sampled off Punta Cana, Dominican Republic, 16 May 2003. The presence of POFs indicates recent spawning (likely within 24 hours). Lower panel. The labeled oocyte has begun final oocyte maturation, as indicated by the migration of the nucleus (Nu) to the periphery and yolk coalescence (YC). Both of these steps are among the series of events initiated hormonally that occur just prior to spawning (likely within 12 hours). 4 Graves, G. E., and A. Z. Horodysky. 2002. Progress report. Use of pop-up satellite archival tags to study sur- vival and habitat utilization of white marlin released from the recreational fishery, 34 p. Virginia Institute of Marine Science, College of William and Mary, Gloucester Point, VA 23062-1346. 666 Fishery Bulletin 103(4) Table 2 Summary of neuston tow information for larval collections of istiophorids in the vicinity of Punta Cana, April and May 2003. "Unidentified istiophorids" refers to specimens for which molecular identification was Dominican Republic, unsuccessful. 2003 dates Number of tows Volume filtered (m3) Number of positive tows Number (length range) of white marlin Number (length range) of blue marlin Number (length range) of unidentified istiophorids 23-24 April 13-17 May Total 11 12 23 9400 8413 17,813 7 5 12 7 (3.45-12.16 mm) 1 (6.20 mm) 8 1 (4.6 mm) 3 (3.49-10.45 mm 4 2 (3.98-5.28 mm) 4 ) (3.25-4.4 mm) 6 nmi/day for Punta Cana; 9.4 nmi/day for the U.S. Mid- Atlantic region; and 8.2 nmi/day for La Guaira Bank (Fig. 7). Thus, the average white marlin displacements found by Graves and Horodysky were 2 to 3-fold higher than those reported in the present study. Black marlin (Makaira indica, Gunn et al., 2003) and striped marlin (Tetrapturus audax, Domeier et al., 2003) monitored mostly outside of spawning times and areas had displace- ments per day 2 to 4-fold higher than those in the pres- ent study. Therefore, reproductively active white marlin and blue marlin monitored in our study (30- or 40-day deployments) appeared to have more constrained average displacements per day than those in other studies where similar PSAT technology was used to monitor marlin in and outside of their respective spawning seasons. Further PSAT-based research, with extended monitor- ing durations (i.e. at least s3-4 months) on white mar- lin and other billfish species in their spawning areas, will be necessary to clarify the causative factors for these findings. Interpretation of our findings also needs to be tempered by the fact that the displacement vectors (minimum straight line distances) used to characterize movements in this study were limited to beginning and end points. In theory, daily estimates of light-based geolocation would provide improved resolution of small- scale movement patterns. However, there is little sci- entific agreement (Musyl et al., 2001; Hill and Braun, 2001) as to the methods and validity of daily tracks generated from highly variable light levels, particularly for wide ranging species near the equator. Although we present no evidence that the horizontal movement patterns of blue marlin (other than possi- bly constrained displacements) reported in our study are directly related to spawning activity, the possibil- ity that white marlin could show fidelity to a spawn- ing area cannot be ruled out. For example, Pepperell (1990) examined conventional tagging results off east- ern Australia and reported that the periodic peaks in return frequency were possibly indicative of black mar- lin returning to the spawning ground as part of their annual migration cycle. The multidirectional pattern of blue and white marlin displacements found in the present study was very similar to the pattern reported by Graves et al. (2002) for blue marlin monitored with PSAT tags for five days off Bermuda. The relatively short-term duration of PSAT tags in both studies (5-40 days) generally precludes detection of directed seasonal horizontal movement patterns (including potential an- nual fidelity to a spawning area) as described by Mather et al. (1975), Pepperell (1990), and Ortiz et al. (2003). Detailed accounts of temperature and depth prefer- ences of electronically monitored white marlin have been rare and those that do exist are limited to very short (<;ten days) monitoring durations (Block et al., 1990; Horodysky et al., 2003; Graves and Horodysky4). We found that white marlin monitored with PSATs for periods of 28-40 days spent the majority of time in the upper 25 m of the water column at temperatures of 28-30°C. Similar findings were found for this species by Graves and Horodysky4 and Horodysky et al. (2003), as well as for blue marlin, black marlin, and striped mar- lin reported by Graves et al. (2002); Kerstetter et al. (2003); Gunn et al. (2003); and Domeier et al. (2003). However, we could not directly address the depth at which spawning occurs in our study from PSAT results, other than to note the preference of adults for, and capture of larvae in, surface waters. Empirical data on the depth of spawning for istiophorids are not available, although anecdotal evidence indicates that some species may spawn in surface waters (black marlin observations by Harvey, personal commun.5). Spawning Prior studies of gonads have indicated that white marlin spawn in the northwest Atlantic during the spring (Baglin, 1977, 1979; de Sylva and Breder, 1997). Spring aggregations of white marlin have been the target of the Cabeza de Toro billfish tournament off Punta Cana for over 40 years (Casilla3), and the sampling of larvae in s Harvey, G. C. McN. 2004. Personal commun. 102 Webster Drive, P.O. Box 10499 APO, Grand Cayman Island, Cayman Islands, British West Indies. Prince et al.: Movements and spawning of Tetropturus albidus and Makaira nigricans 667 4- n 3- D white marlin ■ blue marlin n unidentified istiophond 2- : i r n 1 1 [ 6-7 7-8 8-9 Length (mm) 9-10 10-11 11-12 12-13 Figure 6 Length-frequency distribution for larval white marlin (Tetrapturus albidus), blue marlin, [Makaira nigricans), and unidentified istiophorids collected in the vicinity of Punta Cana. Dominican Republic, April and May 2003. the present study is the first to provide direct evidence of springtime spawning activity in this area. Histologi- cal assessment of the captured female ovarian tissue is consistent with the premise that the adult white mar- lins in the aggregation that we located during fishing and PSAT tagging operations participated in spawning activity. This contention is strengthened by the presence of very small, presumably very young, white (and blue) marlin larvae in the same location. The presence of larvae is the most direct way of docu- menting that a spawning event has actually occurred. This is particularly relevant to highly mobile species, such as billfishes, that can cover large distances in a short time (Prince and Brown, 1991). Serafy et al. (2003) used a similar approach to identify blue marlin spawning grounds in the area of Exuma Sound, Bahamas. In their neuston collections, 90 blue marlin, no white marlin, and three sailfish larvae were captured. Because Serafy et al., (2003) sampled during the entire month of July, it seems possible that larval sampling in Exuma Sound took place after the majority of white marlin spawning had already occurred. Subsequent neuston sampling of Bahamian waters yielded white marlin larvae in Exuma Sound in April and in the Old Bahama Channel and just east of Long Island in March, but no blue marlin during these months (D. E. Richardson and S. A. Lu- thy, unpubl. data). Extensive sampling of the Straits of Florida (SOF) over four years resulted in sporadic captures of white marlin larvae in May and June. Blue marlin was the more common larval marlin in the SOF and was captured from June to September (S. A. Luthy, unpubl. data). In the present study, white marlin larvae were twice as abundant in larval samples as blue marlin larvae (which had been captured earlier in Punta Cana) than in other areas where blue marlin larvae had been found. These results are consistent with reports that white marlin is primarily a spring-time spawner but 35 -, 3 < Z3 CD < g 3 CD CD a CD CD a i | < 3 i 3 E E N > X 3 E £ o C3 E o a o I a — a CD Q I g Figure 7 Mean displacement per day (in nautical miles) for blue marlin iMakaira nigricans), white marlin {Tetrapturus albidus), black marlin iMakaira indica). and striped marlin {Tetrapturus audax) monitored with pop-up sat- ellite archival tags by Gunn et al. (2003) [Australia], Domeier et al. (2003) [Mexico], Graves et al. (2001) [Bermuda], Kerstetter et al. (2003) [Northwest Atlantic], Graves and Horodysky4 [Punta Cana, Dominican Repub- lic, La Guaira, Venezuela, U.S. Mid-Atlantic region], and present study. In all studies, displacements were computed from the point of tag release to the point of first transmission from PSAT tags and are not meant to infer tracks taken by the fish. Means are accompanied by ± one standard deviation for each species identified as follows: blue marlin (BUM, stippled bar), white marlin (WHM, empty bar), black marlin ( BLK, solid bar), and striped marlin (STM, cross hatched bar). mark an expansion of the July to October spawning season reported for blue marlin in the North Atlantic by Erdman (1968) and de Sylva and Breder (1997). 668 Fishery Bulletin 103(4) For blue marlin larvae <6.2 mm SL, Serafy et al. (2003) found problems with estimating age from size with the larval growth equations reported by Prince et al. (1991). Serafy et al. (2003) suggested an exponen- tial growth curve with an assumed size-at-hatching of 2.5 mm SL yielded more realistic larval age values for this growth stanza (<6.2 mm SL). Application of the Serafy et al. (2003) growth model to the larval blue marlin collected in the present study indicates that larvae 3 mm SL were 2 days old, 4-mm-SL larvae were 5 days old, and 5-mm-SL larvae were over 7 days old. It seems possible that blue and white marlin have similar size-at-hatching and growth rates at this early stage of development. Given this assumption, the fact that half of the white marlin larvae (4 out of 9) and a third of the blue marlin larvae sampled in this study were 3-4 mm long (i.e., only a few days old) indicates that spawning activity was taking place in the same general area where these larvae were captured and where the recreational fishery for these species was being pursued. This statement may not hold true for the larval marlin in our collections over 4 mm SL because increases in size and age add increased un- certainties concerning possible spawning locations. Providing a more precise estimate of spawning loca- tion was beyond the scope of our study, although we would expect that the upstream spawning locations (assuming minimal mobility of larvae) of both marlin species to be a function of the prevailing currents and oceanographic features in the Punta Cana area and the elapsed time between the spawning event and sample collection. Future research should focus on a more rigorous and comprehensive estimate of spawning location for all sizes and ages of larvae. This would require both a significant increase in the spatial and temporal larval sampling scheme, as well as direct aging methods for both species and sizes of marlin larvae collected. Implications for managment and future research The current stock status of Atlantic white marlin indi- cates that biomass is only at about 12% of the level nec- essary to maintain maximum sustainable yield (MSY) and continues to decline (ICCAT, 2002). The stock has been estimated to be incurring fishing mortality at a rate about eight times higher than the population can sustain to produce MSY (ICCAT, 2002). Although the Atlantic blue marlin stock is also considered to be overexploited, its status is not as precarious as that of white marlin (ICCAT, 2001). The characterization of adult movements and larval distribution in a potentially important spawning area is seen as a necessary "first step" toward improved management and rebuilding of depressed Atlantic billfish stocks, possibly with gear restrictions (e.g., use of circle hooks. Prince et al., 2002b; Horodysky and Graves, 2005). Improved management seems particularly relevant in the area of Punta Cana because the target of the 40-year-old Cabeza de Toro tournament is, and probably always has been, a repro- ductively active aggregation of white marlin. In light of the ICCAT recommendation to reduce mortality on the overexploited marlins from all Atlantic fisheries (ICCAT, 2002), a shift to catch-and-release requirements for the white marlin recreational fishery off Punta Cana, and the use of circle hooks during the spring months, may be suitable options. In terms of spawning, there is an obvious need for more detailed spatiotemporal informa- tion on the distribution of marlin reproduction and on the identification of nursery areas to help managers make informed decisions regarding conservation of the resource. In addition, more fine-scale data on adult movement patterns in relation to horizontal and verti- cal use of the water column, including identification of spawning depth, are necessary. Acknowledgments This work was made possible through Cooperative Research and Recover Protected Species Candidate Plus Program funds of the National Marine Fisheries Service and additional support from The Billfish Founda- tion and the Lmiversity of Miami, Center for Sustainable Fisheries, Billfish Research Initiative. We thank Noretta Perry at the Florida Fish and Wildlife Commission's Fish and Wildlife Research Institute for histological slide preparations. Literature cited Baglin, R. E„ Jr. 1977. Maturity, fecundity and sex composition of white marlin (Tetrapturus albidus). Int. Comm. Cons. Atlantic Tunas, Madrid, Spain. Coll. Vol. Sci. Pap. 6:408-416. 1979. Sex composition, length-weight relationship, and reproduction of the white marlin, Tetrapturus albidus, in the western North Atlantic Ocean. Fish. Bull. 76:919-926. Block, B. A. 1990. Physiology and ecology of brain and eye heaters in billfishes. Planning the future of billfishes. In Pro- ceedings of the international billfish symposium II; 1-5 August 1988, Kona, Hawaii, Part 2: Contributed papers (R. H. Stroud ed.l, p. 123-136. National Coalition for Marine Conservation, Savannah, GA. de Sylva, D. P., and P. R. Breder. 1997. Reproduction, gonad histology, and spawning cycles of north Atlantic billfishes (Istiophoridae). Bull. Mar. Sci. 60:668-697. Domeier, M. L., H. Dewar, and N. Nasby-Lucas. 2003. Mortality of striped marlin (Tetrapturus audax) caught on recreational tackle. Mar. Fresh. Res. 54: 435-445. Erdman, D. S. 1968. Spawning cycle, sex ratio, and weights of blue marlin off Puerto Rico and the Virgin Islands. Trans. Am. Fish. Soc. 97:131-137. Goodyear, C. P. 2003. Spatiotemporal distributions of longline CPUE and sea surface temperatures for Atlantic marlins. Aust. J. Mar. Freshw. Res. 45:409-417. Prince et al.: Movements and spawning of Tetrapturus albidus and Makairo nigricans 669 Graves, J. E., and J. R. McDowell. 1995. Inter-ocean genetic divergence of istiophorid billfishes. Mar. Biol. 122:193-203. Graves. J. E„ B. E. Luckhurst, and E. D. Prince. 2002. An evaluation of popup satellite tags for estimating post release survival of blue marlin (Makaira nigricans) from a recreational fishery. Fish. Bull. 100:134-142. Gunn, J. S„ T. A. Patterson, and J. G. Pepperell. 2003. Short term movement and behavior of black marlin Makaira indica in the Coral Sea as determined through a popup satellite archival tagging experiment. Mar. Freshw. Res. 54:515-525. Hill, R. D., and M. J. Braun. 2001. Geolocation by light level. In Electronic tagging and tracking in marine fisheries I J. R. Sibert and J. L. Nielsen, eds.), p. 317-330. Kluwer Academic Publish- ers, The Netherlands. Horodysky, A. Z., and J. E. Graves. 2005. Application of popup satellite archival tag technol- ogy to estimate post release survival of white marlin (Tetrapturus albidus) caught on circle and straight-shank "J" hooks in the western North Atlantic recreational fishery. Fish. Bull. 103:84-96. Horodysky, A. Z., D. W. Kerstetter, and J. E. Graves. 2003. Habitat preferences and diving behavior of white marlin (.Tetrapturus albidus) released from the recre- ational rod-and-reel and commercial pelagic longline fisheries in the western North Atlantic Ocean: implica- tions for habitat-based stock assessment models. The Int. Comm. Cons. Atlantic Tunas, Madrid, Spain. Coll. Vol. Sci. Pap, SCRS/2003/033, 13 p. Hunter, J. R., and B. J. Macewicz. 1985. Measurement of spawning frequency in multiple spawning fishes. In An egg production method for esti- mating spawning biomass of pelagic fish: application to the northern anchovy, Engraulis mordax iR. Lasker, ed.), p. 79-94. NOAA Tech. Rep. NMFS 36. Hunter. J. R., B. J. Macewicz, N. C. H. Lo, and C. A. Kimbrell. 1992. Fecundity, spawning, and maturity of female Dover sole Microstomas pacificus, with an evaluation of assump- tions and precision. Fish. Bull. 90:101-128. ICCAT ( International Commission for the Conservation of At- lantic Tunas). 1998. Report of the third ICCAT billfish workshop. ICCAT, Madrid, Spain. Coll. Vol. Sci. Pap., vol. XLVII, 352 p. 2001. Report of the fourth billfish workshop. ICCAT, Madrid, Spain. Coll. Vol. Sci. Pap., vol LIU, 375 p. 2002. Executive summary report for white marlin. ICCAT, Madrid, Spain. Report for biennial period, 2000-2001, part 2 (2001), vol. 2:76-82. 2003. Executive summary report for white marlin. ICCAT, Madrid, Spain. Report for biennial period 2002- 2003, part 1 (2002), vol.2: 97-103. Kerstetter, D. W„ B. E. Luckhurst, E. D. Prince, and J. E. Graves. 2003. Use of pop-up satellite archival tags to demonstrate survival of blue marlin {Makaira nigricans) released from pelagic longline gear. Fish. Bull. 101:939-948. Mather. F J. III. H. J. Clark, and J. M. Mason Jr. 1975. Synopsis of the biology of the white marlin, Tetrap- turus albidus Poey (1861). In Proceedings of the inter- national billfish symposium; Kailua-Kona, Hawaii, 9-12 August 1972, Part 3: Species synopses (R. S. Shomura and F. Williams, eds.), p. 55-94. NOAA Tech. Rep. NMFS SSRF-675. McBride, R. S., F. J. Stengard, and B. Mahmoudi. 2002. Maturation and diel reproductive periodicity of round scad (Carangidae: Decapterus punctatus). Mar. Biol. 140:713-722. McDowell, J. R., and J. E. Graves. 2002. Nuclear and mitochondria DNA markers for specific identification of istiophorid and xiphiid billfishes. Fish. Bull. 100:537-544. Musyl, M. K., R. W. Brill, D. S. Curran, J. S. Gunn, Jason R. Hartog, R. D. Hill, D. W. Welch, J. P. Eveson, C. H. Boogs, and R. E. Brainard. 2001. Ability of electronic archival tags to provide estimates of geographical position based on light intensity. In Electronic tagging and tracking in marine fisheries (J. R. Sibert and J. L. Nielsen, eds.), p. 343- 367. Kluwer Academic Publishers, The Netherlands. Ortiz, M., E. D. Prince, J. E. Serafy, D. B. Holts, K. B. Davy, J. G. Pepperell, M. B. Lowry, and J. C. Holdsworth. 2003. Global overview of the major constituent-based billfish tagging programs and their results since 1954. Mar. Freshw. Res. 54:489-507. Pepperell, J. G. 1990. Movements and variations in early year class strength of black marlin Makaira indica off eastern Australia. Planning the future of billfishes. In Proceed- ings of the internatioal billfish symp. II; 1-5 August 1988. Kona, Hawaii. Part 2: Contributed papers (R. H. Stroud, ed), p. 51-66. National Coalition for Marine Conservation, Savannah, GA. Prager, M. H., E.D. Prince, and D.W. Lee. 1995. Empirical length and weight conversion equations for blue marlin, white marlin, and sailfish from the North Atlantic Ocean. Bull. Mar. Sci. 56:201-210. Prince, E. D., and B. E. Brown. 1991. Coordination of the ICCAT enhanced research pro- gram for billfish. In Creel and angler surveys in fisher- ies management I Guthrie et al.. eds. i, p. 13-18. Am. Fish. Soc. Symp. 12. Prince, E. D., D. W. Lee, J. R. Zweifel, and E. B. Brothers. 1991. Estimating age and growth of young Atlantic blue marlin, Makaira nigricans, from otolith micro- structure. Fish. Bull. 89:441-459. Prince, E. D., M. Ortiz, A. Venizelos, and D. S. Rosenthal. 2002a. In-water conventional tagging techniques devel- oped by the cooperative tagging center for large, highly migratory species. Am. Fish. Soc. Symp. 30:155-171. Prince, E. D., M. Ortiz, and A. Venizelos. 2002b. A comparison of circle hook and "J" hook perfor- mance in recreational catch-and-release fisheries for billfish. Am. Fish. Soc. Symp. 30:57-65. Serafy, J. E., R. K. Cowen, C. B. Paris, T. R. Capo, and S. A. Luthy. 2003. Evidence of blue marlin. Makaira nigricans, spawn- ing in the vicinity of Exuma Sounds, Bahamas. Mar. Freshw. Res. 54:299-306. Wallace, R. A., and K. Selman. 1981. Cellular and dynamic aspects of oocyte growth in teleosts. Am. Zool. 21:325-343 670 Abstract — We summarize the life his- tory characteristics of silvergray rock- fish (Sebastes brevispinis) based on commercial fishery data and biologi- cal samples from British Columbia waters. Silvergray rockfish occupy bottom depths of 100-300 m near the edge of the continental shelf. Within that range, they appear to make a seasonal movement from 100-200 m in late summer to 180-280 m in late winter. Maximum observed age in the data set was 81 and 82 years for females and males, respectively. Maximum length and round weight was 73 cm and 5032 g for females and 70 cm and 3430 g for males. The peak period of mating lasted from December to February and parturi- tion was concentrated from May to July. Both sexes are 50% mature by 9 or 10 years and 90% are mature by age 16 for females and age 13 years for males. Fecundity was estimated from one sample of 132 females and ranged from 181,000 to 1,917,000 oocytes and there was no evidence of batch spawning. Infection by the copepod parasite Sarcotaces aretieus appears to be associated with lower fecundity. Sexual maturation appears to precede recruitment to the trawl fishery; thus spawning stock biomass per recruit analysis (SSB/R) indicates that a F50rr harvest target would cor- respond to an F of 0.072, 20% greater than M (0.06). Fishery samples may bias estimates of age at maturity but a published meta-data analysis, in conjunction with fecundity data, independently supports an early age of maturity in relation to recruitment. Although delayed recruitment to the fishery may provide more resilience to exploitation, managers may wish to forego maximizing economic yield from this species. Silvergray rockfish are a relatively minor but unavoid- able part of the multiple species trawl catch. Incorrectly "testing" the resil- ience of one species may cause it to be the weakest member of the species complex. Life history characteristics for silvergray rockfish (Sebastes brevispinis) in British Columbia waters and the implications for stock assessment and management Richard D. Stanley Allen R. Kronlund Fisheries and Oceans, Canada Pacific Biological Station Nanaimo, British Columbia, Canada V9T 6N7 Email address (lor R D Stanley) stanleyngipac dfo-mpo.gc ca Manuscript submitted 6 April 2004 to the Scientific Editor's Office. Manuscript approved for publication 31 March 2005 by the Scientific Editor. Fish. Bull. 103:670-684 (2005). Silvergray rockfish (Sebastes brevispi- nis) range from the Gulf of Alaska to central Baja California (Love et al., 2001) and are a minor part of the trawl and hook-and-line fisheries catch from northern Washington to the Gulf of Alaska (Alaska Fisheries Information Network.1 Pacific Fisheries Informa- tion Network,2 Pacific Biological Sta- tion3). Coastwide commercial landings averaged 2600 metric tons (t) from 1990 to 2000, and about two-thirds of these landings came from British Columbia (B.C.) waters, mostly from bottom trawling. Hook-and-line land- ings are the most common type in Alaskan waters (mostly from south- eastern Alaska) and have averaged less than 20 t. Combined annual trawl landings from Washington and Oregon peaked at over 1000 t from 1977 to 1979, declined to an average of 210 t from 1990 to 1998, and since 1999 have further declined to negli- gible levels. The B.C. bottom trawl fishery is currently managed through individual vessel quotas (IVQs) whereby a fixed proportion of the annual quota for each stock is allocated to each quota- holder. Because silvergray rockfish are currently assessed and managed as four separate stocks (Fig. 1: Pacific Marine Fisheries Commission areas 3CD, 5AB, 5CD, and 5E), a vessel may possess up to four area-specific quotas for silvergray rockfish. All bot- tom trawlers on the outer coastal wa- ters of British Columbia are required to have an independent observer on the vessel. Once vessels have reached their IVQ for one area and species, and have exhausted their limited op- portunity to temporarily lease quota from other lease-holders, they must cease all bottom trawling even though they may still have IVQ remaining for other species in that area. The quotas for silvergray rockfish are relatively small compared with those for other species in the fishery; thus fishermen can fully fill their sil- vergray rockfish IVQs as they target other species. However, if silvergray rockfish become difficult to avoid through increased abundance or avail- ability, or if the silvergray rockfish quota is reduced, even though catch rates remain constant, they become a nuisance in that fishermen cannot fulfill their IVQs for other species without exceeding their silvergray rockfish IVQ. Therefore, the quotas for minor species, such as silvergray rockfish, now assume more impor- tance than would be gained from their landed value. Finally, the en- actment of species-at-risk legislation in Canada has led to the requirement 1 Alaska Fisheries Information Network. 2000. AKFIN Support Center, 612 W Willoughby Ave. Suite B. Juneau, Alaska 99801. 2 Pacific Fisheries Information Network. 2000. Pacific States Marine Fisheries Commission, 205 SE Spokane Street, Suite 100, Portland, Oregon 97202. 3 Pacific Biological Station. 2000. Un- publ. data. Fisheries and Oceans Can- ada. Nanaimo, British Columbia V9T 6N7, Canada. Stanley and Kronlund: Life history characteristics for Sebastes brevispmis 671 136°W 134°W 132°W 130'W 1283W 126°W 12-f:W 1225W 54>N 52°N' 50'N' Moresby Trough Reed Trough __^.Sea Otter Trough 5 -_ V^Sfefe- ' 48'N- 0 50 100 200 i Kilometers 3CD Site A 1 Meter Station Figure 1 Coastal waters of British Columbia showing boundaries of silvergray rockfish [Sebastes brevispinis) stocks, trawl capture locations of silvergray rockfish (black dots) for 1996-2000, mooring site for the oceanographic metering of temperature at-depth (Al meter station), and 500-. 1000-, and 1500-m depth contours. to assess and protect the status of any species affected by fishing, regardless of its commercial value. Research on silvergray rockfish is an example of an area that has been neglected owing to the lack of eco- nomic importance of this species in the commercial fisheries. Even the data that are available have been collected incidentally during fishing operations target- ing other species or during generic sampling programs. Nevertheless, we show in the present article that these data, in conjunction with detailed commercial catch and effort data, can be used to provide insight into the biolo- gy, assessment, and management of silvergray rockfish. This article summarizes this information and provides estimates of the various life history parameters needed for stock assessment. Some of the estimates represent updates from previous work, but we also for the first time present estimates of fecundity and maturity at age and size. Using these values, we also derive a target reference point. Materials and methods Data sources Data for silvergray rockfish were collected from B.C. waters during port sampling, at-sea observer programs, and research cruises from 1977 through 2000. These data reside at the Pacific Biological Station, Nanaimo, B.C., Fisheries and Oceans, Canada. As of June 2000, the database contains information on over 40,000 speci- mens. Of these specimens, we aged 13,671 representing most of the specimens from which we obtained otoliths, in addition to documenting length, sex, and maturity stage. We, also used catch observations from the com- mercial trawl observer program from 1996 through 2000. Habitat Preferred depth distributions of silvergray rockfish were inferred from analyzing catch rates in the commercial data. We used all bottom tows that contained silvergray rockfish and included tow duration. Bottom depth of the tows was determined as the midpoint between beginning and end depth of the tows. We applied a nonparametric locally weighted regression smoothing function (LOESS) (Cleveland, 1979) to log-CPUE observations grouped by 20-m intervals. Depth of peak catch rates by month were compared with temperature-at-depth estimates based on data col- lected from the site Al meter station on the west coast of Vancouver Island (Fig. 1: 48°32"N by 126°12"W). These data, collected from 1986 to 2000 (excluding El Nino years), were taken from 35-, 100-, 175-, and 400-m depths. The temperatures at fixed depths were then 672 Fishery Bulletin 103(4) Table 1 Field classification of gonad maturity stages for silvergray rockfish ^Sebastes brevispinis Region Science Branch. Fisheries and Oceans, Canada. ) used by the Groundfish Section, Pacific Female ovaries Male testes 1 Immature (translucent, small, color can be clear, amber, yellow, or pink) Immature (translucent, string-like) 2 Developing (small, opaque or translucent, can be yellow, usually light pink) Developing (swelling, brown-white) 3 Developed (eggs usually white or cream white, can be yellow or orange-yellow) Not used 4 Fertilized (large, cream or orange-yellow eggs, translucent) Developed (large, white, easily broken) 5 Embryos or larvae present (includes eyed eggs) Ripe (running sperm) 6 Spent (flaccid, red, a few larvae may be present) Spent (flaccid, creamy-brown, some milt present but not free-flowing) 7 Resting (moderate size, firm, red-grey, red-grey, pink, or purple to almost black) Resting (ribbon-like, small brown) converted through interpolation to provide depth at specific temperatures (Hourston1). Aging and growth determinations Ages were determined by using the otolith burnt-section technique (MacLellan, 1997) with a minor modification. A survey directed at studying juvenile rockfish in 1991 captured two 17-cm silvergray rockfish. An examination of these otoliths indicated that the previous application of the method had incorrectly assigned the first annulus to the age count in specimens. Therefore, some previ- ously aged specimens were probably under-estimated by one year (MacLellan5). A faint first annulus is consis- tent with the late spring to mid-summer parturition of silvergray rockfish that appears to preclude significant summer growth in its first year. The method was modi- fied in August of 1992, and we added one year to all previously aged specimens in the data set. Most (85%) of the otoliths were aged by one reader. The remaining 15% were aged by two readers to moni- tor consistency. If there was a disagreement, the two readers agreed on a "resolved" age. Age and length data were fitted to a generalized growth model (Schnute, 1981) (Appendix 1). Growth dimorphism was calculated as the ratio of the mid- points of fork length (maximum observed length minus minimum observed length) between males and females (Lenarz and Wylie Echeverria, 1991). by tracking the proportions in each maturity stage by month. Lacking histological confirmation for character- izing maturity, we followed the suggestion of Wylie Ech- everria (1987) and used only those specimens collected from the reproductive or gestation period of March to August. Within this subset, we grouped female stages 1 and 2 as immature, and stages 3-7 as mature. Because most mature females exhibited fertilized eggs by March, we assumed that females with small, nondeveloped ovaries in March through August would not complete parturition in the same calendar year. We assumed that stage 1, during which testes are translucent and string-like, was the only male imma- ture stage. Subsequent stages 2 and 4-7 were grouped as mature (stage 3 was not used in the field). The pro- portion of stage-2 males (in relation to males in other mature stages) decreased rapidly during the mating season (September- January) indicating that many of the specimens classified as stage 2 would become ma- ture within the same calendar year. We emphasize, however, that without histological support for these classifications, the assumptions of maturity-at-age or maturity-at-length remain tentative. The estimated proportions of maturity at age were computed by fitting a generalized additive model (GAM) to the binomial maturity classes (0=immature, ^ma- ture) (Hastie and Tibshirani, 1990). A logistic link with a binomial error structure was applied, as well as a second-degree nonparametric LOESS smoother. Reproductive maturity Maturity stage was classified macroscopically in the field (Table 1). We examined the annual reproductive cycle 4 Hourston, R. 2003. Personal commun. Institute of Ocean Sciences. Fisheries and Oceans Canada. 9860 West Saanich Road, P.O. Box 6000. Sidney, British Columbia. V8L 4B2, Canada. 5 MacLellan S. 2000. Personal commun. Pacific Biological Station, Fisheries and Oceans Canada. Nanaimo, British Columbia. V9T 6N7, Canada. Fecundity Fecundity was estimated from a single sample (??=132) of females captured by commercial bottom trawl in Sea Otter Trough in April 1989 (Fig. 1). The catch was stored in refrigerated seawater for four days prior to sampling. Sampling was stratified by length to obtain a range of ages, and from each fish we obtained measurements of fork length, gonad weight, and somatic weight. We also collected otoliths and counted the number of cysts con- Stanley and Kronlund: Life history characteristics for Sebastes brevispims 673 taining the copepod parasite Sarcotaces arcticus in the coelomic cavity. All the oocytes of all the female gonads appeared to be in a prefertilized condition. The ovaries that were used for fecundity esti- mation were fixed and stored in modified Gilson's solution (Leaman. 1988) and shaken weekly for one year. Fecundity estimates were derived gravimetri- cally (Leaman, 1988). Each ovary was drained and filtered through stacked sieves (100-750 urn); each clump was broken manually if possible. The ovar- ian membranes and connective tissue were teased away from eggs and discarded. The oocytes were transferred to millipore filters, vacuumed-dried for 15 minutes, and the oocytes and filter were weighed to 0.01 g. Four subsamples of approximately 0.1 g and 1000 oocytes were weighed to 0.0001 g. Total fecundity was estimated for each fish by multiply- ing total vacuum-dried ovary weight by the mean density of the four samples. Fecundity relationships against age. weight, and length were examined with a generalized additive model (GAM) (Hastie and Tibshirani, 1990). An identity link with a Gauss- ian error structure was used in each case. Ovaries to be used for histological examination were fixed in Smith's formal dichromate solution and then stored in 39c formaldehyde. Histology samples were imbedded, sectioned, mounted, stained with Harris' haematoxylin, and counterstained with alcoholic eosin (Gray, 1954). Spawning stock biomass per recruit (SSB/R) We combined estimates of instantaneous natural mor- tality rate (M) of 0.06 and partial recruitment from Stanley and Kronlund (2000) with our estimates of the proportion mature at age and predicted fecundity at age in order to derive estimates of the expected population fecundity of unfished populations (Appendix 2). The impact of fishing on spawning stock biomass per recruit (SSB/R) can then be explored by comparing the ratio of predicted cumulative fecundity of a cohort under exploi- tation to predicted cumulative fecundity under no fishing pressure (Gabriel et al., 1989; Clark, 1991). Results Habitat The commercial data indicated that the highest catch rates and most of the landings of silvergray rockfish come from the edge of the continental shelf or along the edges of deep troughs (Fig. 1). These tows were typically conducted in bottom depths of 100 to 300 m, although silvergray rockfish have been reported from tows with mid-point bottom depths greater than 580 m. Monthly catch rates by depth indicate a seasonal trend wherein peak catch rates are highest in depths of 180-280 m in March and April, but highest in depths of 100-200 m in September and October (Fig. 2). Feb Mar Apr May Jun Jul Month Aug Sep Oct Nov Dec Figure 2 Silvergray rockfish (Sebastes brevispinis) seasonal depth distribution. The solid lines show the median (heavy line) and 25th and 75th percentiles (thin lines) for the number of silvergray rockfish catch observations (observed commer- cial trawl sets) at depth, between 1996 and 2003. The dots indicate the estimated depth at 7.2°C ±1 standard deviation (dotted line l. If the shift in catch rates correctly indicates sea- sonal movement, and the interpolated temperatures at site Al characterize bottom temperatures on the coast, together they indicate that silvergray rockfish tend to maintain peak densities at bottom water temperatures centered around 7.2°C (Fig. 2). The move to shallower water in the late spring, however, seems to lag behind the cooling of shallower water that results from sum- mer upwelling (Thomson6). The return to deeper water in the fall is coincident with the warming of water at greater depths. The cohabitants of silvergray rockfish were also in- ferred from commercial trawl observations. For these data, we selected tows with at least 50 kg of silvergray rockfish. Silvergray rockfish represented 12.8% of the total catch of over 36,000 t (Table 2). The dominant species by weight in these tows were Pacific ocean perch (Sebastes alutus), arrowtooth flounder (Atheres- thes stomias), yellowmouth rockfish (S. reedi), yellowtail rockfish (S. flavidus), redstripe rockfish (S. proriger), and canary rockfish (S. pinniger). The species most frequently co-occurring in the tows were arrowtooth flounder, lingcod (Ophiodon elongatus), spiny dogfish 6 Thomson, R. 2003. Personal commun. Institute of Oceans Sciences, Fisheries and Oceans Canada. 9860 West Saanich Road, P.O. Box 6000. Sidney, British Columbia V8L 4B2, Canada. 674 Fishery Bulletin 103(4) Table 2 Fish species captured in 1996-99 B.C. bottom trawl tows that contained silvergray rockfish (Sebastes brei ispinis). % of total catch °!c frequency Common name Species (36.489,773 kg) (10,820 tows) Silvergray rockfish Sebastes brevispinis 12.8 100.0 Arrowtooth flounder Atheresthes stomias 13.0 77.2 Lingcod Ophiodon elongatus 2.8 65.1 Spiny dogfish Squalus acanthias 2.5 58.4 Yellowtail rockfish Sebastes flavidus 11.3 57.4 Canary rockfish Sebastes pinniger 5.4 55.2 Redstripe rockfish Sebastes paueispinis 1.3 54.0 Pacific cod Gadus maerocephalus 1.1 53.7 Pacific halibut Hippoglossus stenolepis 0.6 48.2 Redstripe rockfish Sebastes proriger 7.2 47.3 Rex sole Errex zachirus 0.8 46.6 Sablefish Anoplopoma fimbria 0.6 46.2 Spotted ratfish Hydrolagus colliei 0.6 43.7 Pacific ocean perch Sebastes alutus 13.9 40.4 Yellowmouth rockfish Sebastes reedi 12.7 39.2 Dover sole Microstomas pacificus 1.1 36.0 Petrale sole Eopsetta Jordan i 0.4 34.5 Redbanded rockfish Sebastes babeoeki 0.9 33.7 English sole Pleuronectes vetulus 0.5 28.3 Widow rockfish Sebastes entomelas 3.9 27.1 Greenstriped rockfish Sebastes elongatus 0.3 27.0 Longnose skate Raja rhina 0.3 26.0 Others 6.2 — (Squalus acanthias), yellowtail rockfish, canary rock- fish, redstripe rockfish, and Pacific cod (Gadus maero- cephalus). All of these species were observed in more than 50% of the selected tows. The cohabitants varied with depth. Tows conducted in depths less than 200 m tended to include lingcod, dog- fish, canary rockfish, and yellowtail rockfish, whereas catches from greater than 200 m were dominated by arrowtooth flounder, Pacific ocean perch, redstripe rock- fish, and yellowmouth rockfish. Fishermen report that silvergray rockfish are typically found over relatively "hard" bottom, often in proximity to bottom that was not trawlable because it was too rough. They are rarely caught in midwater trawls. Aging and growth estimates The maximum ages observed in Canadian samples were 81 and 82 years for females and males, respectively. The corresponding ages at the 99.9% percentiles were 76 and 77 years. Although we assumed that our aging methods for silvergray rockfish provided unbiased estimates of age, agreement between readers was poor. Agreement to ±1 year was 60-80% for ages less than 20 years and then declined with age. The standard errors of the growth parameter esti- mates show that there is a significant, albeit modest, difference in growth rates; females grow faster and to a larger size (Table 3, Fig. 3). Maximum observed length was 73 and 70 cm for females and males, respec- tively. We estimated the length-weight relationship for females and males separately and combined from 476 total specimens (Table 3, Fig. 3). The ratio of the mid- point lengths for males and females was 97.2 (Table 4), indicating little sexual dimorphism. Maturation cycle The field maturity observations were congruent for females and males (Fig. 4). Testes began developing (stage 2) in September and October and were large and swollen by November and December (stage 4) (Fig. 4). January and March testes were in the late stages of mating (stage 6), whereas from April through August testes appeared to be in a resting phase for males (stage 7). The few observations of large swollen testes with running sperm (stage 5) occurred from October through February. The Stanley and Kronlund: Life history characteristics for Sebastes brevispmis 675 Table 3 Growth and fecundity parameter est mates and standard errors for si lvergrav rockfish [Sebastes brevi spinis 1 1 see Append ix 1 for parameter definitions). Equation Parameter Females Males Combined Estimate SE Estimate SE Estimate SE Length-at-age Vi 48.985 0.048 47.887 0.041 48.468 0.034 y-2 60.628 0.015 56.108 0.091 57.719 0.083 a 0.0581 0.002 0.0708 0.002 0.0709 0.002 b 1.0000 1.000 1.000 T> 15.000 15.000 15.000 T2 60.000 60.000 60.000 Length/Weight lln scale) a -4.000 0.137 -2.506 0.411 -3.634 0.157 P 2.924 0.034 2.547 0.105 2.833 0.040 Fecundity/Somatic weight (In scale) a P 3.014 1.367 0.572 0.073 Fecundity/Length a P -3.454 4.2833 1.007 0.251 Table 4 Comparison of silvergray rockfish iSebastes brevispinis) fork length (1991) (groups 2-4). ratio (group 1) with results from Lena rz and Echevarria Species group Deep (>125m) Shallow (<125m) All rockfish species combined 1 Silvergray rockfish (present study) Fork length ratio 0.97 2 Water-column species Number of species Standard length ratio 12 0.88 5 0.91 17 0.89 3 Demersal species Number of species Standard length ratio 5 0.95 12 0.98 17 0.97 4 All rockfish species combined Number of species Standard length ratio 17 0.90 17 0.96 34 0.91 peak period of mating is presumably December to Febru- ary. One sample of 109 males, collected in March 1988, was recorded entirely as maturing. This one sample accounted for all but two records of stage-4 males col- lected in March and, therefore, contradicted the results of 20 other March samples, totalling 364 specimens. Although we found no evidence of a recording error, we suggest that these specimens were misclassified and were probably recovering instead of developing males. The developing ovaries (stages 2 and 3), observed from January to April, shifted to fertilized through to resting stages (stages 3-7) in April to June. Eyed lar- vae were commonly observed from May to July although a few individuals with eyed larvae were observed in February, August, and October. We examined whether there was a relationship be- tween the size of the female and the timing of par- turition by categorizing July observations as either "parturition not completed" (stages 3-5) or "parturition completed" (stages 6-7) (Fig. 5). The results indicated a dome-shaped relationship with length wherein it ap- pears that a higher proportion of the smaller and larger females had not completed parturition. There were too few observations from June to examine the transition in more detail or to examine whether timing varied with latitude within B.C. waters. Age observations from the commercial fishery indicate that both sexes are 50% mature at about 10 years of age and over 90% are mature at age 16 for females, and age 13 for males (Table 5, Fig. 6). However, the analysis was limited by the lack of young fish in the samples. For example, there were only five 8-year old and thirteen 9-year old females in the data set. Com- parison of the age at maturity and partial recruitment at age indicates that silvergray rockfish mature prior to recruitment (Table 5, Fig. 7). 676 Fishery Bulletin 103(4) A 70 ■ 60 " ^MSm ^>r 50 ■ :••.«'" 40 - l*r+ ■ ■'• 30 " 20 " 1 — -i — ■ Female — i 1 — B 70 " 60 " \ i- E o 50 ■ -■Bv^ 3 jJJJWWT-': c _l 40 ■ 30 " 20 " Male — i 1 1 — 0 20 40 60 Age class 80 20 40 60 80 Age class c 70 ' 60 " 50 ■ _...-— — E u c _l 40 " 30 J Both sexes Female Male 20 " — i 1 1 1 — 5000 " D + + 4000 " § 3 3000 ■ '^3¥ a 2000 " 1000 - o - S* 20 40 60 Age class 80 20 30 40 50 60 Length (cm) 70 Figure 3 Observed lengths at age for (A) female and (B) male silvergray rockfish tSebastes brevispinis). Predicted length-at-age for (C) females, males, and both sexes combined; and (D) weight at length for females ("+") and males ("o"). Fecundity and stock-assessment-parameter estimates The total number of large oocytes ranged from 181,000 to 1,917,000 (Fig. 8). A general linear model (GLM) treatment of log fecundity against log somatic weight and age indicated that age was not a significant variable after accounting for somatic weight. Although size is a better predictor of fecundity than age, we also provide the predicted fecundity with age (Table 5 1 for subsequent calculation of SSB/R. We examined histological cross-sections from 11 ma- ture specimens in the sample. All appeared to be late in the process of vitellogenesis, the late stage 3 of Wylie Echeverria (1987) or stage V of Bowers (1992). The oo- cytes in each ovary were either large, with diameters ranging from 300 to 600 ftm or smaller than 150 f.i. There was little variation within ovaries in the dia- meter of the larger eggs (± 50 f % mature Fecundity (106) Length (cm) Weight (g) er mature 1 0.000 NA NA 0.000 NA NA NA 0.000 2 0.000 NA NA 0.000 NA NA NA 0.000 3 0.000 NA NA 0.010 NA NA NA 0.000 4 0.000 NA NA 0.020 NA NA NA 0.000 5 0.000 NA NA 0.041 NA NA NA 0.000 6 0.000 NA NA 0.080 NA NA NA 0.103 7 0.000 NA NA 0.143 NA NA NA 0.195 8 0.000 42.680 1158 0.235 NA 42.386 1138 0.330 9 0.000 44.750 1233 0.352 NA 43.348 1205 0.492 10 0.002 45.698 1307 0.479 NA 44.245 1270 0.647 11 0.151 46.593 1379 0.599 NA 45.080 1332 0.770 12 0.283 47.437 1448 0.700 0.496 45.858 1391 0.855 13 0.401 48.233 1516 0.776 0.536 46.583 1448 0.909 14 0.505 48.985 1582 0.833 0.576 47.258 1502 0.942 15 0.596 49.694 1645 0.875 0.616 47.887 1553 0.961 16 0.674 50.363 1707 0.906 0.656 48.473 1602 0.974 17 0.742 50.994 1766 0.928 0.696 49.019 1648 0.982 18 0.799 51.590 1823 0.944 0.736 49.528 1692 0.988 19 0.847 52.152 1877 0.955 0.776 50.002 1734 0.992 20 0.887 52.682 1930 0.962 0.817 50.444 1773 0.995 21 0.919 53.183 1980 0.968 0.857 50.855 1810 1.000 22 0.944 53.655 2029 0.971 0.898 51.238 1845 1.000 23 0.963 54.101 2075 0.967 0.939 51.595 1878 1.000 24 0.977 54.521 2119 0.962 0.981 51.928 1909 1.000 25 0.987 54.917 2161 0.953 1.022 52.238 1938 1.000 26 0.994 55.292 2201 0.949 1.057 52.527 1965 1.000 27 0.999 55.645 2240 0.960 1.087 52.796 1991 1.000 28 0.999 55.978 2276 0.972 1.117 53.046 2015 1.000 29 1.000 56.292 2311 0.985 1.145 53.280 2038 1.000 30 1.000 56.589 2345 0.992 1.166 53.497 2059 1.000 40 1.000 58.774 2598 1.000 1.252 55.002 2210 1.000 50 1.000 59.996 2747 1.000 1.228 55.743 2287 1.000 60 1.000 60.680 2832 1.000 1.069 56.108 2325 1.000 70 1.000 61.030 2881 1.000 NA 56.288 2344 1.000 relationship of silvergray rockfish to other rockfish spe- cies was examined by Gharrett et al. (2001). Growth Silvergray rockfish age estimates have not been vali- dated as they have been for other rockfish (Bennett et al., 1982; Culver, 1987; Leaman and Nagtegaal, 1987; Andrews et al. 2002; Kerr et al. 2004); however, there is evidence of a modal progression in the year classes (Stanley and Kronlund, 2000). Our estimated growth rates were similar to those reported by Archibald et al. (1981), who used a small subset of the current data. The maximum recorded size of 73 cm for silvergray rockfish is larger than that for most rockfish but smaller than that reported for the largest rockfishes, such as yelloweye rockfish (S. ruberrimus), cowcod {S. levis), shortraker (S. borealis), and bocaccio (S. paucispinis), all of which can exceed 91 cm (Haldorson and Love, 1991). The growth rate of silvergray rockfish is similar to that of other rockfishes (Haldorson and Love, 1991), and weight at length was Stanley and Kronlund: Life history characteristics for Sebastes brevispmis 679 1 o - o-tr^^ — v~Q-—±L&xr~ cxr^ 0.8" / 0/ 0 6 " / o 0/ 04" / 0> 0 2 " E ^ 00" / ° ~ 0 10 20 30 40 Proportion m CD O °y^r^ 06" 1° 04 " fo Q 1 02" J oo- o 0 10 20 30 40 Age Figure 6 The estimated proportion mature at age for (A) female and (B) male silvergray rockfish (Sebastes brevispinis). similar between sexes as is common for most rockfishes (Love et al., 1990). Lenarz and Wylie Echeverria's (1991) examination of growth dimorphism led them to categorize rockfish as demersal versus water column, and shallow (<125 m) versus deepwater species (>125 m). Table 4 shows that silvergray rockfish are consistent with other demersal rockfish in that they show relatively little sexual dimor- phism in growth. Lenarz and Wylie Echeverria (1991) suggested that the size dimorphism may result from trade-offs between fecundity and size; they suggest that among water-column species, males may optimize size solely for survival, whereas added size for a female may confer advantages in egg production. Seasonal maturation and age at maturity The difficulties in the macroscopic staging of rockfish maturity have been widely discussed (Gunderson et al., 1980; Love and Westphal, 1981; Wyllie Echever- ria, 1987; Love et al., 1990; Nichol and Pikitch, 1994). These authors are consistent in suggesting that maturity stages should be verified by histological examination of samples collected through all seasons. More problematic than the staging is the possibil- ity that commercial fishery samples may not be repre- sentative of the overall population. If only the mature fraction of an age class recruits to the fishery, then age at maturity derived from commercial samples will underestimate actual age at maturity. For the trawl nets used in the rockfish fishery in British Columbia, size at 100% retention for rockfish is about 30 cm. Sil- vergray rockfish do not begin to recruit to the fishery until about 35 cm; thus age or size at recruitment is conditioned by behavior of the silvergray rockfish and not by mesh size. Given the discussion above, our conclusions on age and length at maturity should be viewed as tentative. Nevertheless, the available observations indicate that most females are mature by age nine and most males by age nine or ten. Lenarz and Wylie Echeverria (1991) noted that in 21 of 31 rockfish species, females and males matured at similar ages. Mating appears to take place from September through January and peaks from December through January. This time range differs from the range derived from ob- servations for southeastern Alaska where ripe male sil- vergray rockfish were observed from January to March 680 Fishery Bulletin 103(4) 1.0 06 04 0.2" 00 Mature females Selectivity 10 20 Age - 1 — 30 - 1 — 40 Figure 7 Maturity at age for female silvergray rockfish iSebastes brevispi- nis) in comparison with estimated age at recruitment. (O'Connell8). Significant proportions of females with fer- tilized eggs began to appear 2-3 months later in March and peaked from April to May. This lag time does not differ noticeably from that for other rockfish. Wyllie Echeverria (1987) reported that fertilized eggs are usu- ally found 1-3 months after mating. A few specimens with eyed larvae have been observed in February and March but significant proportions are not observed until April. Parturition lasts through July and peaks in June. Westrheim (1975) suggested that the principal month of parturition was later than June for Oregon-B.C. waters, and later than May for the Gulf of Alaska. Phillips (1964) suggested that the timing of rockfish reproduction could be classified into two broad seasons: early (winter) or late (spring-summer). Silvergray rock- fish clearly fall within the latter category. A mating period from December to January and par- turition in June implies a 5-6 month process. This is longer than the average period reported for rockfish by Wyllie Echeverria (1987) but similar to those re- ported for greenstripe rockfish (S. elongatus) (Dec-Feb to June), redstripe rockfish (Nov-Jan to June) and sharpchin rockfish (S. zaeentrus) (Oct-Jan to Apr-May) (Shaw, 1999). The longer periods may reflect that these species and samples were from higher latitudes than the California observations prevalent in Wyllie Ech- everria's work. However, Shaw (1999) pointed out that 8 O'Connell, V. 1986. Spawning seasons for some Sebastes species landed in the Southeast Alaska longline fishery for nearshore rockfishes (1982-1985). Unpublished report, 21 p. Alaska Department of Fish and Game, Division of Commercial Fisheries, 304 Lake St., No. 103, Sitka, AK 99835-7563. rosethorn rockfish (S. helvomaculatus) samples from the same latitudes indicated a maturation process of 1-2 months. Batch spawning has been reported by Moser (1967a, 1967b) for some rockfish species but our his- tological examination of 11 specimens taken from the April sample provided no indication of this in silvergray rockfish. Samples taken closer to parturition would be more conclusive. The July samples indicated a dome-shaped relation- ship in the timing of parturition. As reported for dark- blotched rockfish (Nichol and Pikitch, 1994) and yel- lowtail rockfish (Eldridge et al., 1991), we observed that the smaller females tended to complete parturition later. However, unlike the results from previous stud- ies, our results indicates that the largest females also tended to complete parturition later. Fecundity Different authors have emphasized that actual fecundity at parturition may be lower than estimates derived prior to fertilization (MacGregor, 1970; Boehlert et al., 1982; Haldorson and Love, 1991; Gunderson, 1997), although this was not observed in yellowtail rockfish (Eldridge et al. 1991). Future studies could examine fecundity closer to parturition; however, it is difficult to capture specimens on the verge of parturition without inducing extrusion (Boehlert et al., 1982). We also caution that our estimates are from one sample and Guillemot et al. (1985) reported significant interannual variation in gonadal development among five species of northern California rockfish. The presence of the Sarcotaces arcticus parasite, pre- viously reported for silvergray rockfish (Sekerak, 1975), Stanley and Kronlund: Life history characteristics for Sebastes brevispmis 681 1500 ■ 1000 500 oo o °° 8o o o°oo ° °t+°8 oB + o + 10 20 30 40 Age 50 60 70 0.6 0.5 0.4 0.3 0.2 0 1 B 10 20 30 40 Age 50 60 70 5 1500 1000 500 o'<5 -Jo o0 + 2000 3000 Somatic weight (g) 4000 Figure 8 Silvergray rockfish (Sebastes brevispinis) fecundity (thousands of eggsl versus lA) age, (B) relative fecundity (thousands of eggs/g somatic weight) against age, and (C) fecundity against somatic weight. Solid circles indicate two possibly anomalous points. The plus symbols indicate females infected by the Sareotaces areticus parasite. The dashed curves represent the limits of point -wise 95% confidence intervals. The "rug" along the x-axis of each plot shows the frequency of observations of age or size classes. appears to be associated with reduced fecundity, albeit this conclusion is based on three observations. This conclusion is consistent with qualitative observations by the senior author that the gonads of infected silvergray rockfish tend to be smaller. Silvergray rockfish fecundity appears typical of the genus as summarized in the meta-data treatment by Haldorson and Love (1991). Predicted fecundity for a 40-year old female exceeds 1,250,000 oocytes, although the maximum observed fecundity in a small sample was almost 2,000,000. The slope of the relationship of log fecundity to log length from our study was 4.283, close to the mean of 4.10 reported for other rockfish (Haldorson and Love, 1991). Haldorson and Love (1991) noted that the ratio of fecundity at the age of 50% maturity versus fecundity 682 Fishery Bulletin 103(4) CD CO w 1 o - o \ 09 - \ 0 \ 0 8 - 0 7 - 0.6 " \ 0 \ 0 \ o \ ^o o^ I "^o^ I 0.00 0.02 0.04 0.06 0 08 0 10 F Figure 9 Spawning biomass per recruit (SSB/R) against instan- taneous fishing mortality (F) for silvergray rockfish r,. The parameter y, is the size of a fish at time rv and y., is the size of a fish at time T2 with .Vo>Ji>0. Parameters a and b determine the shape of the growth curve by controlling the acceleration (decelera- tion! in growth from times t1 to t2. The parameter a has units (in time), and b is dimensionless. Although the mathematical expression of the model has four cases, these four cases actually represent the limiting forms of a single equation as a or b (or both) approach 0. Appendix 2— Spawning stock biomass per recruit If Na is a vector of the numbers of females at each age under constant conditions, such that Na+l=Nae -iFS.+M) where F = the instantaneous fishing mortality rate; Sa = the partial recruitment at age a; and M = the instantaneous natural mortality rate; then the cumulative spawning potential of a cohort of females over the lifetime of the cohort (under constant FandM andSn) is SSB/R = ^NaFecaMatn, i where Fecn = fecundity at age a, and Matn = proportion mature at age a. The spawning potential per recruit (SSB/R) can then be calculated under various estimates of F and compared with the unfished SSB/R (F=0) as shown in Figure 9. 685 Abstract — To assess the impact of California sea lions (Zalophus cali- fornianus) on salmon fisheries in the Monterey Bay region of California, the percentages of hooked fish taken by sea lions in commercial and rec- reational salmon fisheries were esti- mated from 1997 to 1999. Onboard surveys of sea lion interactions with the commercial and recreational fisheries and dockside interviews with fishermen after their return to port were conducted in the ports of Santa Cruz, Moss Landing, and Monterey. Approximately 1745 hours of onboard and dockside surveys were conducted — 924 hours in the com- mercial fishery and 821 hours in the recreational fishery (commercial pas- senger fishing vessels [CPFVs] and personal skiffs combined I. Adult male California sea lions were responsible for 98. 4*5 of the observed depredations of hooked salmon in the commercial and recreational fisheries in Mon- terey Bay. Mean annual percentages of hooked salmon taken by sea lions ranged from 8.5% to 28.6% in the commercial fishery, 2. 2% to 18.36% in the CPFVs, and 4.0% to 17.5% in the personal skiff fishery. Depreda- tion levels in the commercial and recreational salmon fisheries were greatest in 1998 — likely a result of the large El Nino Southern Oscilla- tion (ENSO) event that occurred from 1997 to 1998 that reduced natural prey resources. Commercial fishermen lost an estimated $18,031-$60,570 of gear and $225.833-$498,076 worth of salmon as a result of interactions with sea lions. Approximately 1.4-6.2% of the available salmon population was removed from the system as a result of sea lion interactions with the fishery. Assessing the impact of a growing sea lion population on fisheries stocks is difficult, but may be necessary for effective fisheries management. Impact of the California sea lion {Zalophus californianus) on salmon fisheries in Monterey Bay, California Michael J. Weise James T. Harvey Moss Landing Marine Laboratories 8272 Moss Landing Road Moss Landing, CA 95039-9647 Present address (lor M. J. Weise): Department of Ecology and Evolutionary Biology University of California Santa Cruz Center for Ocean Health 100 Shaffer Rd Santa Cruz, California 95060 E-mail address (for M J Weise): weiseiu'biology ucsc edu Manuscript submitted 13 August 2004 to the Scientific Editor's Office. Manuscript approved for publication 10 June 2005 by the Scientific Editor. Fish. Bull. 103:685-696(2005). California sea lions (Zalophus cali- fornianus) interact with almost all commercial and recreational fisheries along the California coast, causing entanglement and damage to fishing gear and loss of catch (Beeson and Hanan1; NMFS2). The prey of these pinnipeds has been of interest for years because pinnipeds have been viewed as competitors with humans for a variety of fish species. Historically, this competition between pinnipeds and fishermen was of limited impor- tance because fishes and pinnipeds were harvested. However, the increas- ing specialization within the fishing industry during the twentieth century and changing attitudes toward pinni- peds have intensified this competition (Harwood and Croxall, 1988). Since the passage of the Marine Mammal Protection Act (MMPA) in 1972, the population of California sea lions has increased along the West Coast of North America (NMFS2). This increase in pinniped populations has resulted in an increase in the number of reports of pinnipeds interacting with fishing boats and depredating the catch in fisheries along the West Coast (Beeson and Hanan1; NMFS2). The California sea lion popula- tion, found from offshore islands in Mexico north to Vancouver Island, British Columbia, has increased steadily throughout the latter part of the twentieth century (NMFS2). In the early 1900s, the over-riding management philosophy was to limit the California sea lion population because of damage to commercial catches and competition for salmonid fishery resources (Everitt and Beach, 1982). Numbers of sea lions began to increase in the 1940s with curtail- ment of commercial harvests, but bounties were paid for seals and sea lions in Oregon and Washington until the early 1970s. Following passage of the MMPA in 1972, the California sea lion population increased at an annual average of 5.0-6.2% along the West Coast (Carretta et al.3). There are an estimated 204,000-214,000 sea lions in U.S. waters (Carretta et 1 Beeson, M. J., and D. A. Hanan. 1996. An evaluation of pinniped-fisheries interactions in California. Report to the Pacific States Marine Fisheries Com- mission, 46 p. Pacific States Marine Fisheries Commission, 205 SE Spokane St., Portland, OR 97202. 2 NMFS (National Marine Fisheries Ser- vice). 1997. Impacts of California sea lions and Pacific harbor seals on salmo- nids and the coastal ecosystems of Wash- ington, Oregon, and California. NOAA Tech. Memo. NMFS-NWFSC-28, 150 p. Northwest Fisheries Science Center, 2725 Montlake Blvd. East, Seattle, WA 98112-2097. 3 Carretta, J. V., M. M. Muto, J. Barlow, J. Baker, K. A. Forney, and M. Lowry, editors. 2002. U.S. Pacific Marine Mammal Stock Assessments: 2002. NOAA/NMFS Tech. Memo., NOAA-TM- NMFS-SWFSC-346, 290 p. Southwest Fisheries Science Center, 8604 La Jolla Shores Drive, La Jolla, California 92037- 1508. 686 Fishery Bulletin 103(4) al.3), and an additional 45,000-54,000 animals along Baja, Mexico (Aurioles-Gamboa and Zavala-Gonzalez, 1994). In the Monterey Bay region, sea lions do not breed but several important resting sites exist with a range of 3000 to 7500 animals during the nonbreed- ing season (Weise, 2000). In contrast to increases in numbers of sea lions, serious declines in salmonid popu- lations have occurred in recent years as a result of changes and degradation in riverine habitat, declines in water quality, overharvesting, changes in oceanic conditions, and the development of hydroelectric power systems that obstruct major riverine migration routes. Chinook salmon (Oncorhynchus tshawytscha) stocks in the Central Valley of California probably represent 85% to 95% of the chinook salmon catches south of Pt. Arena and in Monterey Bay (PFMC4). Central Valley chinook originate in the Sacramento River and San Joaquin River and have four distinct runs (portion of a salmon stock that returns to their native streams to spawn during a specific season): fall, late-fall, win- ter, and spring. Fall and late-fall runs are relatively healthy, but winter and spring runs are listed as en- dangered under the Endangered Species Act (ESA). Salmon landed in Monterey Bay during the summer fishing season are predominantly fall and late-fall run Central Valley chinook salmon. Size limits and seasonal restrictions are set to reduce retention of listed winter run Central Valley chinook and Klamath River stocks (PFMC4). By taking hooked fish, sea lions can affect salmon stocks because commercial and recreational fishermen continue to fish for salmon to replace those taken by sea lion and this activity of predation and compensatory fishing leads to greater numbers of fish being removed from the population. In the ocean com- mercial troll and recreational salmon fishery, sea lions will swim near or follow fishing boats and will depre- date fish once hooked. Consumption of hooked salmon by sea lions may not only impact salmonid stocks but impact the economic vi- ability of fisheries. Recreational and commercial salmon fishing is an important social and economic asset in California, representing $28,856,000 in revenues in 1995 (PFMC5). Concern over declining salmonid stocks has resulted in adjustments of fishing regulations, such as allocation of harvest between ocean and inland user groups, harvest quotas, and time and area closures (Beeson and Hanan1). Increasing losses offish to Cali- fornia sea lions may produce further restrictions for the recreational and commercial salmon fisheries. 4 PFMC (Pacific Fisheries Management Council). 1999. Re- view of 1998 ocean salmon fisheries. NOAA Award No. NA97FC0031, sections A1-A50 and Bl-43. Pacific Fisher- ies Management Council, 7700 NE Ambassador Place, Suite 200, Portland, OR 97220-1384. 5 PFMC (Pacific Fisheries Management Council). 1995. Re- view of 1994 ocean salmon fisheries. NOAA No. NA57FC0007, sections A1-A50 and B1-B43. Pacific Fisheries Management Council, 7700 NE Ambassador Place, Suite 200, Portland, OR 97220-1384. During the last several decades only a few research- ers have attempted to quantify the impact of sea lions on fisheries in California waters and, more specifical- ly, the Monterey Bay region. According to Beeson and Hanan,1 the recreational ocean salmon landings in 1995 were greatest in Monterey Bay and San Francisco areas and experienced the greatest amount of sea lion preda- tion (charter passenger fishing vessels and private skiff combined). In our study, we surveyed salmon fisheries in Monterey Bay because of the particularly high rates of interactions with sea lions (Beeson and Hanan1) in an effort to better understand the nature and extent of these interactions in the commercial and recreational fisheries. The purpose of this study was to estimate the per- centage of salmon taken by California sea lions from commercial and recreational salmon fisheries in Mon- terey Bay from 1997 to 1999. We hypothesized that the percentages of fish taken by California sea lions in salmon fisheries would be greater than those taken in previous years and would be part of an increasing trend in sea lion and fisheries interactions paralleling the growth of the sea lion population. Further, we esti- mated the number of fish removed from the California Central Valley chinook stock from observed percent- ages of fish taken by sea lions in fisheries. Lastly, we estimated the monetary losses associated with sea lions interacting with commercial and recreational salmon fisheries in Monterey Bay from 1997 to 1999 by quan- tifying the value of fish lost and the type and amount of gear lost or damaged. Methods From 1997 to 1999, observations of interactions between pinnipeds and salmon fisheries were conducted onboard boats, and interviews with fishermen were performed at dockside at the three major ports in the Monterey Bay region: Santa Cruz, Moss Landing, and Monterey (Fig. 1). Salmon fishing operations included commercial troll fishery and recreational fisheries consisting of com- mercial passenger fishing vessels (CPFVs) and private skiffs. The timing of the commercial and recreational salmon fishery seasons varied each year of the study, and sampling was conducted from the beginning to the end of each season (Table 1). The commercial troll fish- ery included day boats (i.e., a one-day fishing trip) and multiple-day boats. Fishing areas included in our study ranged from Pt. Sur north to Ano Nuevo Island. Data regarding fisheries interactions collected at the three different ports were pooled because fishermen from all three ports often fish as a fleet. Dockside surveys were conducted to achieve a greater sampling effort than could be obtained from onboard observations alone. Onboard surveys were conducted to test reliability of dockside surveys and to ensure that investigators fully understood the nature of the interaction. Small biases have occurred when combining onboard and dockside surveys but were attributed to Weise and Harvey Impact of Zalophus californianus on salmon fisheries 687 Table 1 Commercial and recreational salmon fishery seasons in the Monterey Bay region from 1997 to 1999. Commercial Recreational 1997 1998 1999 1-31 May, 23 June-18 July, 1-30 September 1-31 May, 16 June-30 September 1 May - 21 August, 1-30 September 15 March-19 October 14 March-7 September 14 March-6 September onboard sampling in areas where interaction was more prevalent (Miller et al.6). During this study, captains were requested during onboard observations to conduct normal fishing operations and not to intentionally seek out areas with greater or lesser rates of interaction between sea lions and fishery operations. Sampling of commercial and recreational salmon fisheries was stratified by month and approximately equal numbers of onboard and dockside surveys were conducted monthly. Sampling days and ports were se- lected randomly for onboard and dockside surveys of commercial fishing operations, but onboard surveys were limited by crew cooperation and space availabil- ity. Each onboard survey in the commercial fishery took a full fishing day onboard one boat, and dockside interviews were conducted during four-hour periods in the middle to late afternoon during the peak time that vessels returned to port. For CPFVs, which operate virtually every day but have a greater number of boats and passengers on weekends, two-thirds of onboard and dockside sampling dates were selected randomly from possible weekend dates and one-third from all possible weekdays. Onboard surveys of CPFV took a full fishing day aboard one vessel, and dockside surveys took two to three hour periods in early afternoon during peak return times for CPFVs at a randomly selected port. The goal of CPFV dockside surveys was to sample (for the sampling day) all CPFVs targeting salmon and that had returned to port. In the skiff fishery, greater num- bers of fishing trips occurred on weekends; therefore approximately three-quarters of sampling days occurred on weekends, and one-quarter occurred on weekdays. Onboard surveys in 1997 aboard one skiff took a full fishing day, and dockside surveys from 1997 to 1999 were conducted during two-hour sampling periods in late morning and early afternoon during the peak re- turn time for private skiffs. In 1997, four onboard surveys were conducted in the commercial and CPFV fishery, and five onboard pri- vate skiff surveys were conducted. Whereas in 1998 and 1999. in an effort to increase onboard sample size, survey effort was concentrated in the commercial and 6 Miller, D. J., M. J. Herder, and J. P. Scholl. 1983. Cal- ifornia marine mammal-fishery interaction study, 1979- 1981. NMFS Southwest Fish. Cent., Admin. Rep. LJ-83-13C, 233 p. Southwest Fisheries Science Center, 8604 La Jolla Shores Drive, La Jolla, CA 92037-1508. Figure 1 Primary fishing ports used by commercial and recre- ational salmon vessels, and pinniped haul-out sites in Monterey Bay, California. CPFV fisheries; 22 surveys conducted each year in each fishery. Information collected at dockside included port of call, number of fish landed, number of fish taken by pinnipeds at or below the surface, species and number of marine mammals involved in surface take, number of fish released, number of released fish taken by marine mammals, and type and amount of gear loss. Onboard surveys included the same information collected at dockside, as well as standard length of all fish landed. Commercial troll and recreational salmon fisheries use different types of fishing gear, which can affect the nature and magnitude of their interactions with pinnipeds. Commercial salmon trolls are designed to catch fast-swimming fishes by using flashy lures that are trolled behind the moving vessel on heavily 688 Fishery Bulletin 103(4) weighted fishing lines. Multiple lines are mounted on outrigger poles to ensure separation of the lines and are controlled by small hydraulic winches (Starr et al., 1998). Depending on conditions, commercial fishermen use three to fifteen lures per line and two to six lines per boat, totaling six to ninety lures with hooks per boat. In recreational boats each fisherman traditionally uses one rod, reel, line, and hook with bait. Surface takes, also termed "definite takes," were de- fined as takes when pinnipeds took a hooked salmon (and when the species and number of marine mammals involved could be determined). Surface takes were also recorded when fish were hooked and the action of the line indicated that a fish was no longer hooked, and a pinniped surfaced immediately with a fish in its mouth. Takes below the surface, or "probable takes," were de- fined as takes when fish were removed from the hook (and when the species and number of marine mammals involved could not observed directly). Evidence that indicated the occurrence of below-surface takes was in the form of bent hooks, lost gear, or a sea lion surfac- ing within several minutes with a salmon, provided no other fishing boats were in close proximity. Two types of takes were designated because takes below surface were not witnessed, and other predators including sharks take fish from lines, or fish may have escaped. However, fishermen and researchers recognized that takes by pinnipeds, specifically by sea lions, differed from takes by sharks and other predators by the action of the line, effect on the hook or lure (or both), and type of bite on fish parts remaining on the hook. Number of salmon and percentage of catch taken by pinnipeds were compared with the total catch and the legal catch in commercial and recreational fisheries. To- tal catch was defined as numbers offish hooked, includ- ing all legal-size fish, fish taken by pinnipeds, and all undersize fish. Legal catch represented only fish of legal size landed by anglers. Our rationale for using total catch was that all fish, regardless of size, have an equal probability of being taken by pinnipeds; therefore, com- parisons with total catch were a more accurate metric for quantifying the impact of pinnipeds on the salmon fishery. Comparisons with the legal catch inflated the percentage of fish taken by pinnipeds and exacerbated the perception of the problem of pinnipeds interacting with salmon fisheries. However, previous researchers have compared percentage takes by pinnipeds with legal catch; therefore we also made the comparison with legal catch to place our results in a historical context. Mean percentages of fish taken by sea lions in rela- tion to total catch (referred to as "mean percentage of fish taken by sea lions") for the commercial, CPFV, and skiff fisheries for onboard and dockside surveys from 1997 to 1999 were non-normal in distribution and were transformed by using the arcsine transformation for parametric statistical comparisons (Zar, 1996). Mean percentages of fish taken by sea lions in the three fish- eries (commercial, CPFV, and skiff) were compared between onboard and dockside surveys, among years (1997 to 1999), between seasons (sea lion breeding and nonbreeding seasons), and between takes (surface and below surface) using a Students t-test and ANOVA or a Mann-Whitney [/-test and Kruskal-Wallis test for data that were non-normal and heteroscedastic after transformation. Sea lion breeding and nonbreeding seasons from 1997 to 1999 were determined by using aerial and ground counts from Weise (2000). The breeding season was desig- nated as the time when a significant decline in the num- ber of breeding adult males was recorded at haul-out sites in the Monterey Bay region, when animals where pre- sumably heading for the breeding rookeries in southern California. Typically the breeding season is from June and July, and the nonbreeding season occurs during the months of March, April, May, August, and September. Mean catch per unit of effort, or the numbers of fish hooked per day per boat, in commercial, CPFV, and skiff fisheries data were non-normal and heterosce- dastic, therefore, were they were transformed by us- ing -J count + 1 (Harvey, 1987; Zar, 1996). Mean catch per unit of effort for the three fisheries was compared among years with an ANOVA. To estimate the impact of California sea lion depreda- tion on salmon populations in Monterey Bay we com- pared estimated numbers of hooked salmon taken by sea lions and the Central California Valley index (CVI) for chinook salmon abundance. The CVI is the numbers of ocean- and inland-harvested Chinook salmon and the sum of all runs of chinook on the Sacramento Rivers (PFMC4) and represents presumably the population of salmon passing through the Monterey Bay region during the fishery season. The estimated number of salmon taken was calculated from the observed num- ber of takes in the commercial and recreational fishery multiplied by the percentage of the total catch that was sampled. Percentage of the total catch sampled was es- timated by dividing the number of observed legal-size fish landed by the total number of legal-size fish landed (CDF&G, unpubl. data7). Monetary losses resulting from sea lion interactions with salmon fisheries were estimated by evaluating numbers of fish taken by sea lions and types and quan- tities of fishing gear damaged or lost during these inter- actions. Information for the analysis of monetary loses was collected during dockside and onboard surveys for commercial and recreational salmon fisheries. Annual monetary losses resulting from fish taken by sea lions were calculated by using total numbers of estimated takes by sea lions, average dressed mass (mass of gutted and cleaned fish) of salmon landed in Monterey from 1997 to 1999, and average exvessel price (wholesale price per pound of fish paid to fishermen) for chinook salmon in California from 1997 to 1999 (PFMC4). Estimated numbers of takes by sea lions in Monterey Bay from 1997 to 1999 were a function of 7 CDF&G (California Department of Fish and Game). 2004. Ocean Salmon Project database. CDF&G Ocean Salmon Project, 475 Aviation Blvd., Suite 130, Santa Rosa, CA 95403. Weise and Harvey: Impact of Zalophus ca/iformanus on salmon fisheries 689 numbers of observed takes (based on dockside samples) and proportions of the total catch sampled. Estimates of lost and damaged gear were calculated by using average costs for each type of gear used in commercial and recreational salmon fishing operations. A survey of the seven local retail fishing tackle stores in Santa Cruz, Moss Landing, and Monterey was used to estimate mean value of each type of fishing gear used in the recreational (CPFV and skiff combined) salmon fishery. All charter-fishing companies in the three ports in Monterey Bay were surveyed to estimate mean cost of a "setup" sold by charter boat companies to custom- ers. A "setup" was defined as a hook and leader, or a hook, leader, and a 4 oz. or 8 oz. lead sinker. Costs of commercial fishing gear were estimated by surveying 19 local fishermen from the three ports in Monterey Bay. Commercial fishermen buy the majority of their gear in bulk, and often by mail order to reduce costs. Results From 1997 to 1999. 1745 hours of onboard surveys and dockside interviews were conducted in the commercial and recreational salmon fisheries. In 1997, 337 hours of onboard and dockside surveys were conducted, 144 hours in the commercial fishery, 103 hours in the CPFV fishery, and 90 hours in the skiff fishery. In 1998, 704 hours of onboard and dockside surveys were conducted: 370 hours in the commercial fishery, 270 hours in the CPFV fishery, and 64 hours in the skiff fishery. During 1999, 704 hours of onboard and dockside surveys were conducted, 410 hours in the commercial fishery, 258 hours in the CPFV fishery, and 36 hours in the skiff fishery. Increased sampling effort in 1998 and 1999 were the result of increased onboard survey effort in the commercial and CPFV fisheries. During this study 101 onboard surveys and 2780 dockside interviews (number of boats sampled) were conducted in the commercial and recreational salmon fisheries. There were no significant differences in mean percentages of fish taken by sea lions between onboard and dockside surveys in the commercial (1997, P=0.329; 1998, P=0.623; 1999, P=0.653), CPFV (1997, P=0.276; 1998, P=0.660; 1999, P=0.327) and skiff fisheries (1997, P=0.052; Fig. 2). We assumed, therefore, that dockside surveys provided a representative measure of pinniped takes in the salmon fisheries and onboard survey data were pooled with dockside interview data for subsequent analysis. A total of 967 interviews with commercial fishermen and 1813 interviews with recreational fishermen were were conducted at dockside in Monterey Bay, account- ing for 41,895 and 15,115 hooked salmon, respectively (Table 2). In the commercial fishery a similar number of interviews were conducted in 1997 and 1998, whereas in 1999 approximately 21.2% greater numbers of inter- views were conducted with the same effort. However, the number of fish landed in 1999 was significantly less than in 1997 and 1998. In the CPFV fishery, the trend 40' Commercial ^^m Onboard 30 I- — t Dockside 20' T 10 r 1 1 0 ■ 1 CPFV 2£ 2 "g 30 Q- C C Q. 20 O O) S 10 c oj o CD Q- 0 40 J i. Skiff 30 20 ^ 10 . 0 1 n 1997 1998 1999 Figure 2 Percentage of pinniped takes in relation to the total number of salmon hooked as determined from dockside and onboard surveys for the commercial, commercial passenger fishing vessel (CPFV I, and personal skiff fisheries in Monterey Bay, California, from 1997 to 1999. Onboard survey effort concentrated in CPFV and commercial fisheries during 1998 and 1999. Error bars indicate one standard error. was similar to the commercial fishery, but the number of fish landed and the number of boats surveyed was significantly fewer overall. In the skiff fishery, there was a steady decline in the number of fishermen surveyed and the number of fish landed from 1997 to 1999. Onboard observations combined with dockside inter- views revealed that California sea lions were almost exclusively responsible for the depredation of hooked salmon in the commercial and recreational fisheries in Monterey Bay, taking 98.4% of the 1199 observed hooked salmon from 1997 to 1999. Of the estimated 2420 takes in 1997, 1072 were directly observed surface takes and sea lions were identified in 98.6% of the takes (Table 2). In 1998, approximately 501 of 5542 takes 690 Fishery Bulletin 103(4) Table 2 Yearly catch statistics and estimates of the number and percentc ge of salmon taken by pinnipeds in the commercial. commercial passenger fis hing vessel l CPFV Land skiff salmon fisheries dun ng dockside surveys in Monterey Bay in 1997, 1998, and 1999. Catch statistics Number of takes Percentage of takes Total Number of Number Number Total % Total % Number number legal-size Number of offish offish of legal of total dockside offish fish under-size taken at taken below catch catch Fishery Year interviews hooked landed fish surface surface lost lost Commercial 1997 297 17,943 12,288 4124 522 1009 12.5 8.5 1998 293 15,446 6206 4829 97 4314 71.1 28.6 1999 377 8506 6785 966 37 718 11.1 8.9 Total 967 41,895 25,279 9919 656 6041 26.5 16.0 CPFV 1997 139 5168 3157 1577 247 187 13.7 8.4 1998 179 4694 3267 569 305 553 26.3 18.3 1999 58 362 319 35 6 2 2.5 2.2 Total 376 10,224 6743 2181 558 742 19.3 12.7 Skiff 1997 723 2926 1643 828 303 152 27.7 15.6 1998 538 1564 882 409 99 174 31.0 17.5 1999 176 401 315 70 8 8 5.1 4.0 Total 1437 4891 2840 1307 410 334 26.2 15.2 occurred at the surface, and sea lions were identified in 98.4% of those takes. In 1999, 51 of the 779 takes occurred at the surface, and sea lions were responsible for 96.1% of the takes. We assumed sea lions took simi- lar percentages of fish below the surface. As evidence of takes below the surface, sea lions would come to the 35-, (/) T3 0- 0 40- \ i i 1999 30 20 10 i ' rh 0 ■ _^ ^ Commercial Charter Skiff Figure 4 Mean percentage of fish taken by pinnipeds during the California sea lion (Zalophus califor- nianus) breeding and nonbreeding seasons in the commercial, commercial passenger fishing vessel (CPFV), and personal skiff fisheries in Monterey Bay, California, from 1997 to 1999. Error bars indicate one standard error. breeding season of 1997 (P<0.000), whereas in 1998 (P=0.158) and 1999 (P=0.358) there was no significant difference. During all three years, surveys were con- ducted on commercial, CPFV, and skiff fisheries during August and September; however, there was little to no salmon fishing effort because of the perceived sea lion problem and because the remaining fishermen targeted albacore tuna or rockfishes (or both). Because of the different styles of hook-and-line fishing in the commercial troll and recreational salmon fisheries, sea lions were more likely to take fish below the surface from commercial trollers but to take fish at and below the surface from recreational vessels. In the commercial fishery, according to dockside interviews and onboard Commercial CPFV Skiff Figure 5 Mean catch per unit of effort (mean number offish caught per day! in commercial, commercial passenger fishing vessel (CPFV). and skiff fisheries in Monterey Bay, Cali- fornia, from 1997 to 1999. Error bars indicate one stan- dard error. surveys combined, percentages of takes by sea lions below the surface of the water varied throughout the season and were significantly greater than surface takes in 1997 (P=0.001), 1998 (P<0.000), and 1999 (P<0.000; Table 2). In contrast, in the recreational fishery the per- centages of takes by sea lions below the water's surface and at the surface varied by year. During 1997, greater percentages of takes by sea lions occurred at the surface than below the surface on CPFVs (P=0.082) and skiffs (P=0.001; Table 2). Whereas in 1998, significantly great- er percentages of takes occurred below the surface in the CPFV gpv>P ga,gp> v,p,u,w ga.glS,v.p.s,m ga'gp v- u< "'• s' '" ga,gp, v,p, u. w,s,m ing no seasonal growth through to u=l indicating the maximum seasonal growth effect, i.e., where growth effectively ceases at some point each year). The model was fitted by minimizing negative log-like- lihood (-A) function (Eq. 9 in Francis [1988a]). For each data set, made up of i : = 1 to n growth increments: A = X,ln[(l-p)A,+p/i?], where A, =exp -^(AL,-^, -m)2/(CT,2 + s2) [2^(cr,2 + s2)Ji (7) (8) The measured growth increment of the ;'th fish, AL;, has its corresponding expected mean growth increment, Hr as determined from Equation 5 above, where ,i(; is normally distributed with standard deviation or In this study, a, was assumed to be a function of the expected growth increment j.it (Eq. 5, Francis, 1988a): m- (9) AL- Pga-agp Sa-gp a-p ) AT-Ht sin\27r(T-w)] „ where 0, = u — - fori = 1,2. 2/r (5) (6) The parameters gu and g.t are the estimated mean an- nual growth (cm/yr) of fish of initial lengths a cm and P cm, respectively, where a0.05 indicates a high level of outliers and therefore some caution would be required in interpreting the overall model fit. The optimal model parameterization was determined by fitting five different models, comprising different Welsford and Lyle: Estimates of growth of Notolabrus fuacola from length- and age-based models 701 combinations of parameters (Table 2), with unfitted parameters held at zero. A LRT was used to deter- mine the improvement in model fit with the different parameterizations (Francis, 1988a). For models with an equal number of parameters, the model producing the lowest negative log likelihood (-A) was considered the best fit. As with the otolith models, LRTs were conducted on the GROTAG models to compare between sites and sexes, and models were also bootstrapped 5000 times. First-order corrected 959c CIs were calculated for pa- rameter estimates (Haddon, 2001), and pairwise com- parisons of growth parameters, by using CIs and ran- domization tests, as described above for otolith-based models. Results Otolith interpretation Kolmogorov-Smirnov tests showed no significant dif- ference in age-frequency distributions generated by repeat readings of 55 otoliths by the primary reader (Z)005=0.259, Dmax=0.072, not significant) or between readers (D0 05 =0.259, Z)max=0.109, not significant). The IAPE score for all three readings was calculated as 6.9%, and no systematic under- or over-estimation of ages was apparent in age bias plots within or between readers. Therefore age estimates derived from the first readings by the primary author were used for modeling. Age-based growth modeling Site comparisons No significant differences in length frequencies were detected in a Kolmogorov-Smirnov test between sites (D005=0.169, -Dmax=0.097, not significant). Length-at-age estimates showed high variability among individuals, as evidenced by the spread of data points around the fitted models (Fig. 1), and estimates of a ranged from 1.16 to 2.17 cm across all models (Table 3). However, mean lengths-at-age were adequate- ly described by the VBGF across the ages represented by the samples from the two sites. The plots of the site-specific VBGFs indicated that mean length-at-age at Lord's Bluff was higher than at Point Bailey. Because of the absence of young (0+ and 1+) fish in the samples from both sites, and fish >14+ at Lord's Bluff, the standard VBGF parameters were difficult to interpret biologically. Confidence intervals for the three standard VBGF parameters largely overlapped in comparisons between sites (Table 3). Plots of the bootstrap parameter estimates showed strong nonlinear correlations, particularly between la and k, revealing minimal overlap between sites, most easily visualized with logarithmic axes (Fig. 2A). Nonlinear correla- tion between parameter estimates and minimal over- lap between sites were also true to a lesser extent in estimates of lx versus t0 (Fig. 2B). LRTs showed that differences between sites were highly significant overall 36- 34- 32- 30 — 28 — ° ° ° /* — '' Kit,***' ■ A til . • PB ♦ °AViZ i LB ■:.*il?*T PBVBGF 26 — E a. 24 — g) 22 — 0) "" 20- r 18 — 16 — 14 — /* * 1 1 1 1 1 1 1 1 1 1 1 2 4 6 8 10 12 14 16 18 20 22 24 Estimated age (yr) Figure 1 Length-at-age estimates for Notolabrus fucicola, derived from otoliths (symbols), and corresponding von Berta- lanffy growth functions I VBGFs) fitted by least squares (lines). PB = Point Bailey, LB = Lord's Bluff. but could not be attributed to significant differences in individual parameters (Table 4). Confidence intervals for the Francis (1988b) repa- rameterized version of the VBGF clearly indicated sig- nificant differences in growth rates between sites in all three parameters, and no overlap between sites in the CIs of the estimates of mean length at 4, 7, or 10 years old (Table 3). These differences were also evident in plots of bootstrap parameter estimates, the two sites being clearly separated in the parameter space, and showed none of the high nonlinear correlations evident in the standard VGBF estimates (Fig. 3B). Randomization tests produced CIs of the difference between sites of 1.16-2.67, 2.48-3.50, and 2.82-4.44 cm for Z4, /7, and l10, respec- tively. Highly significant differences in all individual parameters growth parameters in the reparameterized model were also shown in LRTs between sites, but no significant difference in o was detected (Table 4). Sex comparisons Confidence intervals for the standard and reparameterized von Bertalanffy parameters sig- nificantly overlapped in all comparisons between sexes (Table 3). Likelihood ratio tests showed no significant differences between models of sexes within sites — a conclusion supported by considerable overlap in plots of bootstrap estimates (not shown). Length-based growth modeling Model parameterization Site-specific data sets were optimally parameterized under the most complex model, 702 Fishery Bulletin 103(4) Table 3 Von Bertalanffy growth function parameter estimates for Notolabrus fucicola. Numbers in bold text are parameter estimates from the original dataset. Numbers in parentheses are the proportion of parameter estimates from bootstrapped data sets that were less than the estimate from the original data set. Numbers in plain text are first-order corrected bootstrap 95% confidence intervals. LB = Lord's Bluff; PB = Point Bailey. Dataset Parameter estimate Ijcm) /;• i/yr) f0(yr) 14 (cm) (7(cml /10(cm) a (cm) LB 44.7 0.085 -3.23 20.4 25.9 30.1 1.61 (0.48) (0.51) (0.50) (0.51) (0.50) (0.51) (0.57) 35.4 to 68.4 0.036 to 0.152 -5.82 to -1.59 20.0 to 20.9 25.4 to 26.3 29.4 to 30.8 1.39 to 1.87 PB 43.3 0.065 -4.65 18.5 22.9 26.5 1.79 (0.66) (0.51) (0.50) (0.52) (0.53) (0.58) (0.32) 37.9 to 86.7 0.021 to 0.096 -8.71 to -2.83 17.9 to 19.2 22.6 to 23.2 26.1 to 26.9 1.57 to 1.92 LBS 8 52.1 0.059 -4.46 20.3 25.5 29.7 1.38 (0.51) (0.49) (0.48) (0.51) (0.48) (0.50) (0.64) 34.6 to 1210.1 0.001 to 0.157 -9.21 to -1.55 19.8 to 20.9 24.9 to 25.9 28.9 to 30.5 1.16 to 1.68 LB2 5 43.2 0.095 -2.80 20.5 26.1 30.4 1.74 (0.47) (0.51) (0.48) (0.51) (0.48) (0.49) (0.62) 33.1 to 187.8 0.007 to 0.192 -7.42 to -0.98 19.9 to 21.3 25.5 to 26.7 29.2 to 31.7 1.45 to 2.17 PB2"° a 2! — ♦ PB j^Jfl LB AJU '^^5s r«r 25 - 21 — 2= — .M |j^^.- ■ ^sn HR-? -?VA? 22 1 1 l /4 (cm) E 29- B • PB O LB 26 ~l 27 /7 (cm) Figure 3 Bootstrap estimates of reparameterized von Ber- talanffy growth function mean lengths at age for Notolabrus fucieola, by site. (A) /4 versus /- (B) l- versus /10 mean length-at-ages at 7 and 10 years. Contrasting crosses show the location of parameter estimates based on the original data set (+, PB = Point Bailey, x, LB = Lord's Bluff). 0 25 0 20 - 0 15 0 10 0 05 0 00 -0 05 -0.10 - -0 15 -020 -0 25 "I I I I T~ 2 5 3 3.5 4 4 5 g20 (cm/yr) B 0 05 1 U Figure 4 Bootstrap estimates of GROTAG parameters for Noto- labrus fucieola, by site: (A) g20 versus g30, mean annual growth at initial length 20 and 30 cm and (B) u versus w, magnitude and timing of seasonal growth. Contrasting crosses show the location of parameter estimates based on the original data set (4-, PB = Point Bailey, x, LB„ =Lord's Bluff). at or=0.05 when tested individually (Table 8A), in agree- ment with the results of the randomization tests. Sex comparisons Bootstrapped parameter estimates from sex-specific data sets were approximately sym- metrical about the original estimates (Table 7). The largest divergence from 0.5 was evident in estimates of s for females at Lord's Bluff and males at Point Bailey. Bootstrap estimates of u for Lord's Bluff males occasion- ally extended into spurious negative values, lowering confidence estimates of the extent of seasonal growth in this data set (Table 7). Based on simple overlap of CIs, no single parameter differed significantly between sexes at either site (Ta- ble 7). Plots of the bootstrap estimates of the growth parameters g.,0 and g.,8 showed minimal overlap between males and females, and separation was most evident along the g20 axis (Fig. 5A). Plots of bootstrapped es- timates of the seasonal growth parameters u and w (Fig. 5B), and the measurement error parameters m and s Welsford and Lyle: Estimates of growth of Notolabrus fuacola from length- and age-based models 705 Table 5 Parameter estimates and negative log likelihoods ( -A) of models used in likelihood ratio tests t(i determine the opti mal para- meterization of GROTAG models for Notolabrus fucicola tagging data. bv site. Bold text in -A column indicates the optimally parameter zed model for each data set. Model 4 is equivalent to model 5 withp = 0 in these instances. LBre^ = residents of Lord's Bluff: PB = Point Bailey. Parameter estimate §20 §30 w s m Data set Model (cm/yr) (cm/yr) V u (yr) (cm) (cm) P -A LB 1 1.84 1.07 0.88 — — — — 0.07 57.06 2 3.00 1.67 0.88 0.59 0.22 — — 0.07 50.46 3 2.60 1.12 0.29 — — 0.22 -0.12 0.00 20.59 4 and 5 3.30 1.42 0.26 0.45 0.14 0.22 -0.10 0.00 12.97 PB 1 1.50 1.01 0.73 — — — — 0.16 87.82 2 1.55 1.15 0.82 0.31 0.13 — — 0.07 79.02 3 1.87 1.18 0.36 — — 0.19 -0.08 0.00 36.16 4 and 5 1.53 1.01 0.35 0.57 0.91 0.18 -0.07 0.00 23.52 Table 6 Paramete • estimates and ne native log likelihoods -A) of models used in likelihood ratio tests to determine the optimal para- meterization of GROTAG models for Notolabrus fucicola tagging data. by sex within site. Bold text in - A column indicates the optimally parameterized model for each data set. * indicates the parameter estimates and likelihoods when GROTAG is fitted to the Lord's Bluff (LB) 2S data set with a single outlier removed. Model 4 is equivalent to model 5 withp = 3 in all other instances. PB = Point Bailey. Pa rameter estimate §20 §30 w s m Data set Model (cm/yr) (cm/yr) V u (yr) (cm) (cm) P -A LBSS 1 1.98 1.49 0.52 — _ — — 0.00 43.07 2 1.88 1.54 0.50 0.23 0.04 — — 0.00 39.24 3 2.09 1.62 0.27 — — 0.21 -0.05 0.00 32.21 4 and 5 2.04 1.67 0.27 0.23 0.19 0.20 -0.04 0.00 29.44 LB$S 1 2.05 1.40 0.52 — - — — 0.16 60.58 2 1.99 1.20 0.48 0.41 0.98 — — 0.15 58.19 3 2.88 1.87 0.26 — — 0.25 -0.29 0.00 41.15 4 2.75 1.75 0.25 0.32 0.96 0.24 -0.31 — 38.40 5 2.66 1.48 0.22 0.47 0.94 0.22 -0.26 0.03 36.23 4 and 5* 2.66 1.48 0.22 0.48 0.94 0.23 -0.26 0.00 30.36 PBSS 1 1.31 1.02 0.60 — — — — 0.24 21.31 2 1.15 0.96 0.61 0.41 0.90 — — 0.19 19.93 3 1.54 1.21 0.33 — — 0.19 -0.03 0.00 6.43 4 and 5 1.15 0.93 0.32 0.81 0.88 0.18 -0.04 0.00 2.49 PB?9 1 1.49 1.15 0.68 — — — — 0.16 30.55 2 1.43 1.16 0.90 0.33 0.12 — — 0.00 28.85 3 1.96 1.32 0.38 — — 0.20 -0.11 0.00 19.06 4 and 5 1.46 1.01 0.39 0.77 0.87 0.18 -0.12 0.00 15.78 (Fig. 5C) showed distinct relationships within the two sexes. Randomization tests confirmed significant dif- ferences in g20, m, and w. The CIs of these differences were estimated to be 0.2-1.09 cm/yr faster for females with an initial size of 20 cm, with an annual peak in fe- male growth 3-152 days earlier than males, and with a measurement error that overestimated female length by 2-40 mm more than the measurement error for males. 706 Fishery Bulletin 103(4) Table 7 GROTAG parameter estimates derived from Notolabrus fucicola tag-recapture data. For all data sets, ga is the mean annual growth of individuals with an initial length of 20 cm. gp represents the estimated mean annual growth of individuals with an initial length of 30 cm for Lord's Bluff ( LBre,) and Point Bailey < PB (, or the estimate for 28-cm individuals for all other data sets. Numbers in bold text are the parameter estimates from the original data sets. Numbers in parentheses are the proportion of parameter estimates from bootstrap data sets less than the original estimate. Numbers in plain text are first-order corrected bootstrap 95^ confidence intervals. Data set Parameters estimate Sa (cm/yr) SB (cm/yr) t> u w (yr) s (cm) m (cm) LBres 3.30 1.42 0.26 0.45 0.14 0.22 -0.10 (0.50) (0.48) (0.54) (0.43) (0.47) (0.60) (0.48) 2.32 to 4.34 0.80 to 2.19 0.14 to 0.40 0.23 to 0.68 -0.08 to 0.20 0.18 to 0.26 -0.18 to -0.03 PB 1.53 1.01 0.35 0.57 -0.09 0.18 -0.07 (0.51) (0.51) (0.50) (0.46) (0.54) (0.55) (0.56) 1.21 to 1.94 0.76 to 1.31 0.27 to 0.44 0.25 to 1.00 -0.14 to 0.05 0.15 to 0.22 -0.12 to -0.01 LBf ! 2.04 1.68 0.27 0.23 0.19 0.20 -0.04 (0.481 (0.51) (0.58) (0.45) (0.48) (0.57) (0.49) 1.77 to 2.31 1.32 to 2.01 0.20 to 0.40 -0.06 to 0.43 -0.02 to 0.29 0.12 to 0.28 -0.14 to 0.06 LB?? 2.66 1.48 0.22 0.48 -0.06 0.23 -0.26 (0.49) (0.50) (0.50) (0.39) (0.52) (0.62) (0.49) 2.27 to 2.98 1.18 to 1.83 0.13 to 0.30 0.16 to 0.69 -0.16 to 0.12 0.14 to 0.34 -0.41 to -0.10 PB?c? 1.15 0.93 0.32 0.81 -0.12 0.18 -0.04 (0.41) (0.43) (0.57) (0.51) (0.43) (0.62) (0.53) 0.83 tol.69 0.61 to 1.41 0.17 to 0.47 0.18 to 1.00 -0.20 to 0.10 0.14 to 0.24 -0.14 to 0.06 PB?? 1.46 1.01 0.39 0.77 -0.13 0.18 -0.12 (0.50) (0.47) (0.56) (0.46) (0.47) (0.56) (0.51) 1.08 to 2.33 0.70 to 1.01 0.23 to 0.74 0.14 to 1.00 -0.20 to 0.14 0.09 to 0.27 -0.22 to 0.00 Table 8 Likelihood ratio tests of the GROTAG models for which bootstrap parameter estimates were generated (Tables 5 and 6): (A) Point Bailey l PB) against Lord's Bluff (LBreJ (B) LB- S against LB? J. -A= negative log-likelihoods. The base case is the negative log- likelihood of the data sets fitted with two wholly separate models. * = significant at a=0.05. A Hypothesis Base case Coincident curves =£20 =£.30 = V = 11 =W -A 36.49 51.98 42.19 37.36 37.35 36.62 38.91 37.64 36.66 30.98 11.38 1.72 1.72 0.12 4.84 2.28 0.33 df <0.001* <0.001* 0.189 0.190 0.623 0.028* 0.130 0.565 B Hypothesis Base case Coincident curves =#20 =#28 = V = (/ =w df 59.80 — - — 72.17 24.75 7 ! 0 - • LB Males LB Females B 04 g (cm/yr) • LB Males LB Females -0 4 m (cm) Figure 5 GROTAG bootstrap parameter estimates for Notolabrus fucicnla from Lord's Bluff, by sex: (A)g20 versus g28, mean annual growth at initial length 20 and 28 cm; (Bl u versus w, magnitude and timing of seasonal growth and (C) m versus s, mean and standard deviation of measurement error. Contrasting crosses show the location of parameter estimates based on the original data set (+ = males, x = females). nor LRTs indicated significant difference in any of the model parameters, and bootstrap plots showed large regions of overlap (not shown). Discussion Comparisons of models In this study, two methods, based on mathematically different concepts, produced similar conclusions, namely that growth in N. fucicola was faster at Lord's Bluff than at Point Bailey. The results of length-based and age- based models also produced similar conclusions regard- ing the methods most suitable for robust comparisons of models and parameter estimates for different groups of fish. Confidence intervals were only reliable indicators of difference in cases where parameters showed low levels of correlation between estimates and where highly sig- nificant differences existed, such as in site comparisons of the reparameterized VBGF parameters, and hence were of limited utility. Likelihood ratio tests provided a robust method of testing differences between models. However, we believe 708 Fishery Bulletin 103(4) that evidence from more than one source is required be- fore conclusions can be drawn about differences between models designed to describe nonlinear processes such as growth. In the present study, bootstrapping techniques proved to be informative as a way of visualizing the behavior of the models used, and the distributions and correlations of parameter estimates that could not be determined readily from model likelihoods alone. They also provided a basis for estimating nonparametrically with randomization tests the differences, and CIs, of growth estimators between populations. Hence we rec- ommend bootstrapping, plots of parameter estimates, and randomization tests to complement the "traditional" statistical tests such as the LRTs. The standard VBGF has been criticized for the dif- ficulty it causes in extracting biological meaning from parameters (Knight, 1968; Roff, 1980; Francis, 1988b; 1992). The problem is particularly acute where only a part of the size or age range (or both ranges) of animals is available — a situation regularly faced in analyses of fisheries data (Haddon, 2001). Data sets in our study were limited, particularly by the lack of fish in the lower age classes (cf. Ewing et al., 2003). Hence, any attempt to interpret or compare la or t0 as descrip- tors of the growth of N. fucicola would be spurious. Furthermore, because k and lx are highly correlated, comparisons of k cannot be independent of the effects of size or age selectivity on a data set. Because of the limitations of such parameters, and as la and k are often inputs into population dynamics models and em- pirical models estimating parameters such as natural mortality (e.g., Pauly, 1979), extreme caution should be exercised when extrapolating these values from limited data. However, this instance exemplifies the utility of the reparameterization, because even with limited data, the useful parameters of mean lengths at age can be estimated and compared. Variability in growth Models of growth can be used to estimate length-depen- dent processes in fish populations, such as reproductive output, increases in biomass due to individual growth, selectivity of fishing gear, and the impact and appropri- ateness of size limits as management tools. The results of the present study demonstrate that growth varies significantly across individuals, seasons, sexes, and sites in N. fucicola. Although the significance of estimating the variabil- ity in growth around the population mean (v) was not explicitly tested during model parameterization, values of v around 0.2 to 0.7 were estimated for all data sets modeled. Values in this range have been estimated with GROTAG from other species of bony fishes (Francis, 1988a; b; 1988c; 1992; Francis, et al., 1999) and car- tilaginous fishes (Francis and Francis, 1992; Francis, 1997; Francis and Mulligan, 1998; Simpendorfer, 2000; Simpendorfer, et al., 2000), indicating that considerable individual variability in annual growth of size classes is common. The extent of variability in individual growth is an important factor when quantifying growth be- cause it may obscure other sources of growth variation, particularly in situations where data are limited. This effect may partially explain why age-based models failed to detect any significant effect of sex on growth rates in our study, whereas length-based modeling indicated that among smaller size classes, females grew faster than males at Lord's Bluff. On the basis of a large data set (>1000 individuals), Ewing et al. (2003) demonstrated that average length-at-age was significantly higher for females than males in N. fucicola although the magni- tude of this difference was small. No growth differences between the sexes were evident at Point Bailey but given slower growth rates, the absolute magnitude of any expected growth differences related to sex would be relatively small and difficult to detect statistically. Our study is the first to show that growth rates of N. fucicola vary significantly across small spatial scales; the two sites in our study were separated by less than 25 km. At Point Bailey, few individuals reach the mini- mum legal size limit of 30 cm until 10 years of age, whereas at Lord's Bluff they do so at least two years earlier (Fig. 1). An equivalent conclusion is evident from the GROTAG estimates, indicating that a 28-cm fish at Point Bailey will take nearly 2 years on aver- age to exceed 30 cm, whereas fish of the same size are likely to reach legal size in just over a year at Lord's Bluff. Hence relative yields and rates of replacement of recruited size and age classes are likely to be lower at Point Bailey than at Lord's Bluff. However, because N. fucicola can be sexually mature at lengths of 12 cm (Patterson, 2000), some individuals are likely to have spawned for 6-8 years before recruitment to the fishery at Lord's Bluff (Fig. 1). This size at maturity suggests that the minimum legal size limit provides effective protection of the reproductive output of the prerecruit population of TV. fucicola at both sites. Using length-at-age estimated from whole otoliths, Barrett (1999) found no growth differences between sev- eral populations of N. fucicola in southeastern Tasmania and used these findings to support the hypothesis that populations are not resource limited. Our study did not specifically address any hypothesis about resource limi- tation but has clearly demonstrated that growth rates can vary between populations at the scale of individual reefs. Notolabrus fucicola are site-attached once they settle out of the plankton, rarely having an ambit of more than 500 m on contiguous reef, and rarely cross- ing soft bottom habitat if they are resident on smaller patch reef habitat (Barrett, 1995b). Intuitively, it fol- lows that if productivity varies between reefs, then the potential for growth of individual site-attached reef fish may be limited. A variety of factors have been cited in other temperate reef species where spatial variability in length-at-age is evident, such as habitat type (Gilland- ers, 1997; Barrett, 1999), conspecific competition and variation in juvenile recruitment (Jones, 1980, 1984), and impacts of exploitation (Buxton, 1993). Further study is advocated to determine the factors that influ- ence N. fucicola growth at this scale. Welsford and Lyle: Estimates of growth of Notolabrus fucicola from length- and age-based models 709 Parameterization of seasonal growth significantly improved the fit of the GROTAG models, indicating that seasonal variability in growth is significant for N. fucicola. The estimates of seasonal growth from our study constitute the first for this species. The LRTs indicated significant differences in the timing of maxi- mum growth (h>) between sites and between sexes at Lord's Bluff. This result was repeated in the randomiza- tion tests based on the outputs of bootstrapping. Peak growth in N. fucicola at both sites is estimated to oc- cur over the austral spring-summer, during maximum water temperatures and increased productivity off the coast of Tasmania (e.g., Halpern, et al.4), and peak growth occurs significantly later in the season at Lord's Bluff than at Point Bailey. The mechanism affecting the timing of seasonal growth at this reef-by-reef scale is worthy of further investigation but is likely to include variability in seasonal cycles of oceanography, in avail- ability of food (Denny and Schiel, 2001; Shepherd and Clarkson, 2001) and in temperature effects on metabo- lism, controlling the amount and timing of resources for allocation to growth throughout the year. The estimate of the size of the difference in w be- tween the sexes at Lord's Bluff had very broad CIs, and it is difficult to propose a hypothesis that could result in seasonal growth varying between the sexes by as much as five months, although resource allocation for reproduction could be involved. It may be that the particularly small size of the female data set at this site limited our ability to estimate seasonal growth ac- curately with GROTAG, and further study is required to more precisely determine how important seasonal growth differences between the sexes are in temperate reef fishes such as N. fucicola. Sex-specific GROTAG analyses indicated a significant difference in measurement errors; females were under measured by a mean of 3 mm, compared to less than 1 mm for males at Lord's Bluff. Greater measurement errors for females have been detected in other studies with GROTAG (e.g., Simpendorfer, 20001, but a reason for greater difficulty in measuring females is difficult to determine. A possible explanation from our study is the high individual growth variability and small sample sizes. Both of these factors have been shown to affect accurate estimation of measurement error in GROTAG (Francis and Mulligan, 1998), and therefore the high estimate of m in our study may be an artifact of the data set. 4 Halpern, D., V. Zlotnicki, P. M. Woicheshyn, O. B. Brown, G. C. Feldman, M. H. Freilich, F. J. Wentz, and C. Gentemann. 2000. An atlas of monthly mean distribu- tions of SSMI surface wind speed, AVHRR sea surface tem- perature, TMI sea surface temperature, AMI surface wind velocity, SeaWIFS chlorophyll-a, and TOPEX/POSEIDON sea surface topography during 1998. Jet Propulsion Labora- tory Publication 00-08, 102 p. National Aeronautics and Space Administration, Jet Propulsion Laboratory, California Institute of Technology, 4800 Oak Grove Drive, Pasadena, CA 91109. A significant difference in growth between the sexes at Lord's Bluff indicates that under conditions of rapid growth, females may grow significantly faster than males. As discussed above, the current minimum legal size limit is effectively protecting the reproductive out- put of the prerecruit population of N. fucicola. However, any significant lowering of the legal minimum size is contraindicated where, in prerecruitment size classes, females grow more rapidly than males, because lower- ing the legal size may result in differences in sex-spe- cific fishing mortality. As demonstrated in the present study, the choice of growth model and the methods used to compare pa- rameter estimates are critical to ensuring that growth is adequately described, differences in growth are de- tected, and if detected, are interpretable. In combina- tion, the tests we employed are shown to be generally robust, even in situations where data sets are limited in sample size or by coverage across the full range of age and length classes. We recommend the use of a combination of approaches, including growth models with biologically interpretable parameters, statistical tests such as LRTs, plots of bootstrap parameters, and nonparametric randomization tests, to provide insight into the growth dynamics of fish species. Acknowledgments We wish to thank Malcolm Haddon, John Hoenig, Craig Johnson, Paul Burch, and Philippe Ziegler for their con- structive suggestions for the manuscript. Alan Jordan and Graeme Ewing made invaluable contributions to the field and laboratory analyses. This study was conducted as a part of a Ph.D. program by the primary author, through the Faculty of Science and Engineering at the University of Tasmania. Literature cited Barrett, N. S. 1995a. Aspects of the biology and ecology of six temper- ate reef fishes (Families: Labridae and Monacanthidae). Ph.D. diss., 192 p. Univ. Tasmania, Hobart, Tasmania, Australia. 1995b. Short- and long-term movement patterns of six temperate reef fishes (Families Labridae and Mona- canthidae). Mar. Freshw. Res. 46:853-860. 1999. Food availability is not a limiting factor in the growth of three Australian temperate reef fishes. En- viron. Biol. Fish. 56:419-428. Beamish. R. J., and D. A. Fournier. 1981. A method for comparing the precision of a set of age determinations. Can. J. Fish. Aquat. Sci. 38:982- 983. Buxton, C. D. 1993. Life-history changes in exploited reef fishes on the east coast of South Africa. Environ. Biol. Fish. 36:47-63. Campana, S. E., M. C. Annand. and J. I. McMillan. 1995. Graphical and statistical methods for determining 710 Fishery Bulletin 103(4) the consistency of age determinations. Trans. Am. Fish. Soc. 124:131-138. Cerrato, R. M. 1990. Interpretable statistical tests for growth com- parisons in the von Bertalanffy equation. Can. J. Fish. Aquat. Sci. 47:1416-1426. 1991. Analysis of nonlinearity effects in expected-value parameterizations of the von Bertalanffy equation. Can. J. Fish. Aquat. Sci. 48:2109-2117. Cranfleld, H. J., K. P. Michael, and R. I. C. C. Francis. 1996. Growth rates of five species of subtidal clam on a beach in the South Island. New Zealand. Mar. Freshw. Res. 47:773-784. Denny, C. M., and D. R. Schiel. 2001. Feeding ecology of the banded wrasse Notolabrus fucicola (Labridael in southern New Zealand: prey items, seasonal differences, and ontogenetic variation. N. Z. J. Mar. Freshw. Res. 35:925-933. Edgar, G. J. 1997. Australian marine life: the plants and animals of temperate waters, 544 p. Reed Books, Melbourne, Australia. Ewing, G. P., D. C. Welsford, A. R. Jordan, and C. Buxton. 2003. Validation of age and growth estimates using thin otolith sections from the purple wrasse Notolabrus fucicola. Mar. Freshw. Res. 54:985-993. Fabens, A. J. 1965. Properties and fitting of the von Bertalanffy growth curve. Growth 29:265-289. Fournier, D. A., J. R. Sibert, J. Majkowski, and J. Hampton. 1990. MULTIFAN a likelihood-based method for esti- mating growth parameters and age composition from multiple length frequency data sets illustrated using data for southern bluefin tuna iThunnus maccoyii). Can. J. Fish. Aquat. Sci. 47:301-317. Francis, M. P. 1997. Spatial and temporal variation in the growth rate of elephantfish (Callorhinchus milii). N. Z. J. Mar. Freshw. Res. 31:9-23. Francis, M. P., and R. I. C. C. Francis. 1992. Growth rate estimates for New Zealand rig iMustelus lentieulatus). Aust. J. Mar. Freshw. Res. 43:1157-1176. Francis, M. P., and K. P. Mulligan. 1998. Age and growth of New Zealand school shark, Galeorhinusgaleus. N. Z. J. Mar. Freshw. Res. 32:427- 440. Francis, M. P., K. P. Mulligan, N. M. Davies, and M. P. Beentjes. 1999. Age and growth estimates for New Zealand hapuku, Polyprion oxygeneios. Fish. Bull. 97:227-242. Francis, R. I. C. C. 1988a. Maximum likelihood estimation of growth and growth variability from tagging data. N. Z. J. Mai-. Freshw. Res. 22:42-51. 1988b. Are growth parameters from tagging and age- length data comparable? Can. J. Fish. Aquat. Sci. 45:936-942. 1988c. Recalculated growth rates for sand flounder, Rhombosolea plebeia, from tagging experiments in Can- terbury, New Zealand. N. Z. J. Mar. Freshw. Res. 22: 53-56. 1992. L-infinity has no meaning for tagging data sets. In The measurement of age and growth in fish and shell- fish. Australian Society for Fish Biology work- shop, Lome 22-23 August 1990 (D. A. Hancock, ed.), p. 182-184. Bureau of Rural Resources, Canberra, Australia. Gillanders, B. M. 1997. Comparison of growth rates between estuarine and coastal reef populations of Achoerodus viridis (Pisces: Labridae). Mar. Ecol. Prog. Ser. 146:283-287. Haddon, M. 2001. Modelling and quantitative methods in fisheries, 406 p. Chapman & Hall/CRC, Boca Raton, FL. Jones, G. P. 1980. Growth and reproduction in the protogynous her- maphrodite Pseudolabrus celidotus (Pisces: Labridae) in New Zealand. Copeia 1980:660-675. 1984. Population ecology of the temperate reef fish Pseudolabrus celidotus Bloch and Schneider (Pisces: Labridae) II. Factors influencing adult density. J. Exp. Mar. Biol. Ecol. 75:277-303. Kimura, D. K. 1980. Likelihood methods for the von Bertalanffy growth curve. Fish. Bull. 77:765-776. Knight, W. 1968. Asymptotic growth: an example of nonsense disguised as mathematics. J. Fish Res. Board Can. 25:1303-1307. Mooij, W. M., J. M. V. Rooij, and S. Winjhoven. 1999. Analysis and comparisons of fish growth from small samples of length-at-age data: detection of sexual dimorphism in Eurasian perch as an example. Trans. Am. Fish. Soc. 128: 483-490. Patterson, T 2000. Fisheries models of the temperate wrasses Noto- labrus tetricus and Notolabrus fucicola: implications of life history strategy for management. B.S. honours thesis. Ill p. Univ. Tasmania. Hobart, Tasmania, Australia. Pauly. D. 1979. On the interrelationships between natural mortal- ity, growth parameters, and mean environmental temper- ature in 175 fish stocks. J. Cons. CIEM 39:175-192. Ratkowsky, D. A. 1986. Statistical properties of alternative parameteri- sations of the von Bertalanffy growth curve. Can. J. Fish. Aquat. Sci. 43:742-747. Roff, D. A. 1980. A motion for the retirement of the von Bertalanffy function. Can. J. Fish. Aquat. Sci. 37: 127-129. Russell, B. C. 1988. Revision of the labrid fish genus Pseudolabrus and allied genera. Rec. Aust. Mus. Supplement 9:1-76. Schenker, N., and J. F. Gentleman. 2001. On judging the significance of differences by exam- ining overlap between confidence intervals. Am. Stat- ist. 55:182-186. Shepherd, S. A., and P. S. Clarkson. 2001. Diet, feeding behaviour, activity and predation of the temperate blue-throated wrasse, Notolabrus tetricus. Mar. Freshw. Res. 52:311-322. Simpendorfer, C. A. 2000. Growth rates of juvenile dusky sharks, Careha- rhinus obscurus (Lesueur, 1818) from southwestern Australia estimated from tag-recapture data. Fish. Bull. 98:811-822. Simpendorfer, C. A., J. Chidlow, R. McAuley, and P. Unsworth. 2000. Age and growth of the whiskery shark, Furgaleus macki, from southwestern Australia. Environ. Biol. Fish. 58:335-343. Welsford and Lyle: Estimates of growth of Notolabrus fucicola from length- and age-based models Wang, Y. G., D. Thomas, and I. F. Somers. 1995. A maximum likelihood approach for estimating growth from tag-recapture data. Can. J. Fish. Aquat. Sci. 52:252-259. Wang, Y. G.. and M. R. Thomas. 1995. Accounting for individual variability in the von Bertalanffy growth model. Can. J. Fish. Aquat. Sci. 52: 1368-1375. Wang, Y. G. 1998. An improved Fabens method for estimation of growth parameters in the von Bertalanffy model with individual asymptotes. Can. J. Fish. Aquat. Sci. 1998:397-400. Wang. Y G., and N. Ellis. 1998. Effect of individual variability on estimation of pop- ulation parameters from length-frequency data. Can. J. Fish. Aquat. Sci. 55:2393-2401. 712 Effects of harvesting methods on sustainability of a bay scallop fishery: dredging uproots seagrass and displaces recruits Melanie J. Bishop Charles H. Peterson Henry C Summerson David Gaskill University of North Carolina at Chapel Hill Institute of Marine Sciences 3431 Arendell St Morehead City, North Carolina 28557 E-mail address (for M J Bishop, contact author) melaniebishop-1fgiutsedu.au Present address (for M. J. Bishop): Department of Environmental science University of Technology, Sydney Corner of Westbourne St. and Pacific Highway Gore Hill, New South Wales, Australia 2065 Fishing is widely recognized to have profound effects on estuarine and marine ecosystems (Hammer and Jansson, 1993; Dayton et al., 1995). Intense commercial and recreational harvest of valuable species can result in population collapses of target and nontarget species (Botsford et al., 1997; Pauly et al., 1998; Collie et al. 2000; Jackson et al., 2001). Fishing gear, such as trawls and dredges, that are dragged over the seafloor inflict damage to the benthic habitat ( Dayton et al., 1995; Engel and Kvitek, 1995; Jennings and Kaiser, 1998; Watling and Norse, 1998). As the growing human population, over-capitalization, and increasing government subsidies of fishing place increasing pressures on marine resources (Myers, 1997), a clear understanding of the mecha- nisms by which fishing affects coastal systems is required to craft sustain- able fisheries management. Dredging, possibly the most de- structive of common fishing meth- ods (Collie et al., 2000), has been the subject of many recent ecological studies (Dayton et al., 1995; Jen- nings and Kaiser, 1998; Thrush et al., 1998). These studies indicate that dredge extraction and disturbance can have large direct effects on the abundance, biomass, and diversity of resident macrobenthic species (e.g., Caddy, 1973; Eleftheriou and Robert- son, 1992). In addition, dredging can indirectly affect macrobenthic species through disturbance of benthic habi- tat (Ramsay et al., 1998; Lenihan and Peterson, 1998). Indirect impacts of dredging may be particularly seri- ous where highly structured biogenic habitats, such as oyster reefs or sea- grass beds, are affected (Peterson et al., 1987; Lenihan and Peterson. 1998; Collie et al., 2000; Lenihan and Peterson, 2004). These habitats may be considered essential habitat for many species of fish of commer- cial or recreational value (Thayer et al., 1975), providing refuges from predators (Orth et al., 1984; Castel et al., 1989) and abundant epibiotic food (Virnstein et al., 1984; Sanchez- Jerez et al., 1999). Among fishery species dependent on biogenic habitat is the commer- cially and recreationally important bay scallop (Argopecten irradians). In the two reproductive seasons, spring and fall, bay scallop recruits settle onto hard substrates (Belding, 1910; Castagna, 1975) where they remain attached for the first few months of their lives. They then complete their 12-24 month life cycle on the estuary floor. In North Carolina, eelgrass is the only hard substrate of any abun- dance to which bay scallop recruits can attach themselves (Kirby-Smith, 1970). Commercial harvest of bay scallops in North Carolina is achieved pri- marily by toothless epibenthic dredge (22.7 kg legal limit; NCFMC1). Dredg- es have the advantage that, unlike rakes, they can be used from boats in deep as well as shallow waters. Their disadvantage is that they decrease the biomass and shoot density of sea- grass in scallop beds (Fonseca et al., 1984). Early in the North Carolina scallop season, which extends from December through May (NCMFC1), most of the juveniles from the previ- ous fall spawning are still attached to seagrass blades (Spitsbergen-). If these juveniles are displaced by habi- tat destruction, reduced numbers of scallops may be available for harvest in the subsequent year (hypothesized by Thayer and Stuart. 1974). Al- though seagrasses can recover from small-scale disturbances to shoots by vegetative growth, large-scale dis- turbances to their subsurface root and rhizome system may permanent- ly reduce the density of submerged aquatic vegetation (SAV) (Peterson et al., 1987) such that it may limit settlement of the following year's recruits or induce greater rates of predation on them (or bring about both). Although, in North Carolina, the bay scallop fishery management plan requires that the scallop sea- son be opened after fall spawning is completed (Peterson, 1990); it fails to consider how methods of harvest may indirectly effect spawning stock biomass in years to come. 1 NCMFC (North Carolina Marine Fisher- ies Commission). 2005. North Caro- lina fisheries rules for coastal waters, 210 p. North Carolina Department of Environment and Natural Resources, 1601 Mail Service Center, Raleigh, NC 27699. 2 Spitsbergen, D. 1979. A study of the bay scallop (Argopeeten irradians) in North Carolina waters. Report for Proj- ect 2-256-R, 44 p. North Carolina Divi- sion of Marine Fisheries. 3441 Arendell Street, Morehead City, NC 28557 Manuscript submitted 30 October 2004 to the Scientific Editor's Office. Manuscript approved for publication 1 April 2005 by the Scientific Editor. Fish. Bull. 103:712-71912005). NOTE Bishop et al.: Effects of harvest methods on sustainability of a bay scallop fishery 713 Implementation of gear restrictions that allow only hand methods of harvesting scallops (i.e., hand, rake, dip nets) may minimize impacts of harvesting on scallop recruits by reducing damage to seagrass and the loss of juvenile bay scallops that comprise the year class that will be fished in the following year. Although such restrictions were introduced to Bogue Sound in 1992 in response to the 1987 red tide that decimated scallop populations in that water basin (Summerson and Peter- son, 1990), this conservation-based measure was discon- tinued in 1998 because of social pressure from fisher- men. In the present study, we ascertain the impacts of dredges and hand-harvesting methods on the biomass of seagrass, as compared to undisturbed controls, 1) by measuring the biomass of seagrass directly dislodged by each method, and 2) by ascertaining, through measure- ments of biomass one month later, whether this removal affects the standing stock of seagrass over a longer temporal scale. We also tested both direct and indirect effects of seagrass removal on bay scallop recruits by measuring their density before and one month after harvesting and by ascertaining whether any document- ed difference can be explained by the numbers directly removed by uprooting of seagrass during harvesting. Such an assessment of ecological impacts of dredging on bay scallop recruits is urgently required given that North Carolina landings of bay scallops have fallen to an historic low since the relaxation of gear restrictions (Burgess and Bianchi3). Materials and methods Nine adjacent experimental plots, 25 mx8 m, were estab- lished as a research sanctuary, closed to commercial fishing activity, in western Bogue Sound, North Carolina (34°41.6'N, 76°59.1'W), prior to the opening of the scallop season in winter 2001-2002. Although this section of Bogue Sound has been closed to scallop dredging since at least 1998, its high-tide water depth of 1.5 m is well within the depth range for harvesting with this method. Plots contained continuous seagrass beds dominated by Zostera marina on a muddy-sand bottom. Three of the plots were randomly assigned to each of the experi- mental treatments: hand-harvested, dredge-harvested, and control (undisturbed). In order to ensure that our treatments were representative of harvesting methods and intensities used by the industry, they were per- formed with participation of an experienced commercial scallop fisherman (Ted Willis of Salter Path). Dredging was achieved with a standard 72-cm wide steel scallop dredge, at an intensity of five parallel tows, each run- ning along the length of the plot within a 10-minute period. This method, which mimicked commercial fishing 1 Burgess, C. C, and A. J. Bianchi. 2004. An economic profile analysis of the commercial fishing industry of North Carolina including profiles for state-managed species, 243 p. North Carolina Division of Marine Fisheries, 3441 Arendell Street, Morehead City, NC 28557. practices, minimized overlap between the dredge paths. Hand scalloping involved a single fisherman collecting scallops from the bottom by hand, also during 10-minute periods. Care was taken to ensure that the treatments were applied evenly over the entire plot to avoid creating large within-plot variance that might preclude detection of differences among plots. Seagrass and scallops collected during harvesting were retained for measurements. The number of adult scallops (>40 mm shell height; Peterson et al., 1989) obtained with each of the methods of harvest was enu- merated. The size (to the nearest 0.1 mm) and number of juvenile scallops collected as bycatch and the dry weight of seagrass removed during harvesting were quantified separately. Because not all seagrass and juvenile scallops displaced by harvesting are retained in the dredge or by a fisherman collecting scallops by hand methods, an 8-m long net with 5-mm mesh that extended from the bottom to the surface was set down- stream from each plot and perpendicular to the flow of the current during harvest. The nets were strung between stakes marking the corners of the experimental plot. Dislodged juvenile scallops and seagrass collected by the nets were added to the amounts extracted from the dredge to compute displacement totals. Nets were also set downstream of controls to determine natural rates of transport of seagrass and juvenile scallops that could not be attributed to harvesting operations. Each plot was sampled on 14 January 2002, immedi- ately prior to harvesting on that same day to determine: 1) the density of bay scallop recruits (size s40 mm; Peterson et al., 1989); 2) the size distribution of the recruits; and 3) biomass per unit of area of seagrass. These variables were resampled on 25 February 2002, over one month later, to ascertain any lasting impact of harvest. Sampling of scallops was conducted with a 0.5-m'2 cylindrical quadrat, haphazardly positioned at nine locations within each plot. A 1.2-cm tall cylinder of 6-mm nylon mesh, attached to the quadrat and sus- pended by a buoyant plastic hoop that floated on the surface of the water, isolated the volume of water above each quadrat so that it could be sampled by suction with a Venturi suction device (according to Peterson et al., 1989). The suction device forced 600 mL of water per minute through a 3-mm collecting bag. Suction sampling was necessary because scallops, which typi- cally recline on the bottom, can enter the water to swim when threatened by predators or otherwise disturbed (Peterson et al., 1982). The disturbance caused by suc- tion sampling of only nine small areas was minimal compared to the scale of harvesting disturbance. Upon returning to the laboratory, seagrass was removed from samples for measurement of dry weight biomass and live scallops were counted, measured to the nearest 0.1 mm and categorized as adults (>40 mm) or recruits (^40 mm) in the subsequent year class. Seagrass was sampled in five replicate 0.25-m2 areas within each plot by suction dredging inside a 0.56-m diameter circular quadrat to a sediment depth of 12 cm. Previous sampling has shown this method to be success- 714 Fishery Bulletin 103(4) 240 " 180 " !E O) '53 120 ■ 5 >. Q 60 " 0 ' ' ' ' Control Hand Dredge Treatment Figure 1 Mean (±1 SE) dry weight of seagrass displaced from control (undisturbedl, hand-harvested, and dredged plots of seagrass during the 10 minutes during which the treatments were applied. n = 3. ful in removing both roots and shoots in their entirety (Peterson et al., 1983a). Shoots and roots, which were collected in a 3-mm nylon mesh bag, were dried at 60°C to constant weight to calculate total dry weight biomass of seagrass. ANOVAs allowed us to test for a significant inter- action between time (before versus after) and distur- bance (dredge versus hand-harvest versus control) in the biomass of seagrass and recruit density of bay scallops (a basic BACI design; Green, 1979), indicative of an impact of harvest. The cause of any significant time x disturbance interactions was explored by using Student-Newman-Keul (SNK) tests. Prior to each analysis, Cochran's (1951) C-test was done to test for heterogeneity of variances. Where variances were hetero- geneous, data were In (x+1) transformed to remove heteroscedasticity at a = 0.05. Results Of the two methods used to harvest adult scallops, hand harvesting had by far the greater efficiency in these shallow waters (ANOVA, P<0.0001). Over a period of 10 minutes, an average of 156 ±12 (1 SE) scallops within each 25x8 m plot was harvested by hand as compared to 26 ±1 scallops with the dredge. The two methods of harvesting differed significantly in their impact on seagrass. Hand harvesting of scal- lops did not increase dislodgement of seagrass above the natural drift rate (Fig.l). Dredging, in contrast, resulted in 127 times the export of seagrass. This ex- traction did not, however, result in detectable reductions in biomass per unit of area of seagrass within dredged plots when sampled one month later. There was no sig- nificant temporal change in the biomass of seagrass in any of the three treatments from before to one month after harvesting (Table 1, Fig. 2). Fewer than 2% of the estimated total number of juve- nile scallops in a plot were directly removed by dredg- ing and none was removed by hand-harvesting. Never- theless, sampling one month after harvesting indicated depressed densities of juvenile bay scallops in dredged plots (Table 2; Fig. 3). This difference could not be at- tributed to natural change; small increases (16-55%) in numbers of juvenile bay scallops in the hand-harvested and control plots were documented over the same period (Fig. 3). A comparison of size-frequency histograms of juvenile bay scallops within each type of plot from be- fore to after harvesting revealed that the decrease in ju- venile scallop numbers in the dredged plots was primar- ily due to losses of scallops in the smallest size classes (<14 mm; Fig. 4). In the dredged plots, mean (±SE) size of juveniles (<40 mm in shell height) increased from 17.04 ±0.83 in January to 20.43 ±0.76 in February. Over the same time period, mean size changed little in the control (16.09 ±0.85 to 16.75 ±0.75 mm) or in the hand-harvested (18.19 ±0.85 to 17.95 ±0.65 mm) plots. Discussion Previous research indicates that the implementation of certain gear restrictions on estuarine bivalve fisheries can minimize habitat destruction without sacrificing harvesting efficiency (Peterson et al., 1983b; Lenihan and Peterson, 2004). In our study, which successfully mimicked the efficiency of commercial dredging and NOTE Bishop et al.: Effects of harvest methods on sustainability of a bay scallop fishery 715 Table 1 BACI (Green, 1979) analysis of variance that tes ts for ar impact of scallop harvest ng on biomass of seagrass. Nine plots of seagrass were randomly assigned to three treatments: undisturbed control hand-harvested dredged. Biomass of seagrass was determined immediately before (Jan 2002) and one month after (Feb 2002) application of treatments to plots, n = 5. Source df MS F P Before versus after treatment 1 0.14 0.78 0.41 Treatment 2 0.35 0.81 0.49 Plot (treatment) 6 0.43 3.50 0.00 Before vs. after x treatment 2 0.26 1.41 0.31 Before vs. after x plot (treatment) 6 0.18 1.49 0.19 Residual 72 0.12 Transformation ln( v+1) Cochran's test C= 3.16

0.05) hand-harvesting of bay scallops (see Burgess and Bianchi3), hand-harvesting yielded six times the bay scallop harvest obtained per unit of time by dredging, while reducing del- eterious environmental effects. Hand-harves- tikng did not result in uprooting of seagrass or displacing juvenile bay scallops, whereas dredging caused significant damage to sea- grass. Ten minutes of dredging resulted in an average dry weight loss of 200 g of seagrass per plot — 9 % of the estimated biomass of sea- grass present prior to harvest. Despite this siz- able removal of seagrass biomass, a persistent impact of dredging on seagrass biomass was not detected one month later. To the contrary, a 39% increase in seagrass biomass was seen across the dredged plots that was not repli- cated in the control plots. This result indicated that dredging had only a short-term negative impact on seagrass shoots (the necessary pro- duction of new leaves) and instead appeared to stimulate new production during the winter period that was more than sufficient to replace dredging damage. Despite the rapid recovery of seagrass from dredging injury, a sustained negative impact of dredging on the density of juvenile bay scallops within plots was detected over the one-month period of our study. In contrast to the small increases in juvenile scallop density that occurred in hand-har- vested and control plots over the course of the study, mean density of juveniles in dredged plots declined from 1.37 ±0.33 (1 SE) to 0.89 ±0.23 per 0.5 m2. This 40% reduction in juvenile scallops in dredged plots cannot be explained by the bycatch alone. Whereas total bycatch of juveniles was, on average, two scallops per dredged plot, the average reduction in the density of juvenile bay scallops was 0.5 per 0.5-m2 quadrat or 200 per 200-m2 plot. Instead, the reduction in density of juvenile scallops in dredged plots is best explained by their migration Before After Time Figure 2 Mean (±1 SE) dry weight of seagrass per 0.25-m2 quadrat in con- trol (undisturbed), hand-harvested, and dredged plots immedi- ately before and one month after the 10-minute treatments were applied. n=15. after dredging injury to seagrass habitat into adjacent undisturbed control and hand-harvested plots. Abun- dances of juvenile bay scallops in hand-harvested and control plots increased over the one month of our study by an amount more than sufficient to compensate for losses of juveniles from dredged plots. These increases in abundances in control and hand-harvested plots can- not be attributed to the settlement of new recruits: fall recruitment of juvenile scallops to seagrass beds is typically completed by the end of December (Peterson et al., 1989), spring spawning does not commence until March (Peterson and Summerson, 1992), and scallops spawned during our experiment could not possibly have grown fast enough over one month to reach a size re- 716 Fishery Bulletin 103(4) Table 2 BACI analysis of variance testing for an impact of scallop harvesting on density of scallop recruits. Nine plots of seagrass were randomly assigned to three treatments: undisturbed control, hand-harvested, dredged. Density of scallop recruits was deter- mined immediately before (Jan 2002) and one month after (Feb 2002) application of treatments to plots. »=9. Source df MS F P Before vs. after treatment 1 0.89 0.78 0.41 Treatment 2 5.57 2.74 0.14 Plot (treatment) 6 2.03 0.77 0.59 Before vs. after x treatment 2 4.57 4.01 0.08 Before vs. after x plot (treatment) 6 1.14 0.43 0.85 Residual 144 Cochran's test C = 0.13(P>0.05) SNK tests Before vs. after x treatment Before: control = hand-harvested After: control = hand-harvested > = dredged dredged tained by sieves (see Irlandi et al., 1999 for growth rates). Scallops colonizing hand-harvested and control plots were of the right size and of sufficient abundance to be those missing from dredged plots. The migration appears to have included active swimming because tidal currents were perpendicular to the direction of scallop movement. Although juvenile scallops are largely sessile, our in- terpretation that juveniles migrate in response to dredg- ing is consistent with field and laboratory observations of juvenile bay scallop behavior. During seasonal slough- ing of eelgrass blades, juvenile bay scallops break away 2.4 Dredged Hand-harvested Control Before Time After Figure 3 Mean (±1 SE) number of juvenile bay scallops (s40 mm in height) per 0.5-m2 quadrat in control (undisturbed), hand vested, and dredged plots immediately before and one month the 10-minute treatments were applied. ;i = 15. shell -har- after and re-establish byssal attachments to seagrass blades (Thayer et al., 1975). Mesocosm observations confirm that juveniles are capable of swimming distances of at least several meters when displaced (Bishop, personal observ. ). Thus, our experimental restriction on dredging to small areas may have facilitated relocation of scallops to adjacent, undisturbed habitat, where they remained one month later even after seagrass had regrown in the dredged plots. In the case of the commercial fishery, however, juvenile scallops emigrating from disturbed habitat over the extensive fished areas would be far less likely to encounter undisturbed seagrass habitat for re- attachment. Indeed, transport to unfavorable unvegetated habitat where predation risk is enhanced would likely inflate mortality. In our study, juvenile scallops lost from the dredged plots came primarily from the small- est size classes. Small juvenile scallops are more susceptible to benthic predators that forage within seagrass beds than larger ju- veniles (Pohle et al., 1991). Because the for- aging efficiency of some predators increases with decreasing biomass of seagrass (Prescott, 1990), a decrease in seagrass biomass, even for a period of weeks, would likely increase predation on juvenile scallops. Thus, small ju- veniles probably are increasing their chances of survival by emigrating away from depleted and into denser seagrass. Larger juveniles, in contrast, experience a partial size refuge from predators (e.g., Pohle et al., 1991), and thus have less incentive to emigrate. This study considered the impact of only a single bay scallop-harvesting event on sea- grass biomass and abundance of juvenile bay scallops within small experimental plots. Fishing disturbances are, however, typically chronic, occurring multiple times within a given season, and over large spatial scales. NOTE Bishop et al Effects of harvest methods on sustainability of a bay scallop fishery 717 Before Control 12 6 0 12 >, o c d 6

'■*'» it * ^ ' " Grand • , , Banks 50°N - 40°N - „ — -rsnL W 30°N - Atlantic Ocean 20°N - \J ^. — i i 1 i - 50°N - 40"N 30=N 80= W 70='W 60°W 50=W 40"W Figure 1 Locations of observed longline sets (1992-2002) recorded in the U.S. Pelagic Observers Program database and analyzed in the present study. the event that a factor was found to be nonsignificant (P>0.05), it was removed and a regression was rerun until all highest order model terms were significant (Hocking, 1976; Draper and Smith, 1981). We assumed maturity (both sexes) occurred at 185 cm FL (Pratt, 1979). The average PDA and the ratio of immature-to- mature individuals discarded in each 0.5-degree cell were estimated and plotted in order to visually examine the spatial distribution of these two variables. Table 1 Regression coefficients and associated standard error values (SE) for the estimation of proportion of blue shark released alive IPDA) in = 37), where fi0 corresponds to the intercept, and fi, and /i, are coefficients associated with blue shark fork length and set duration, respectively. Parameters Estimate SE P> If I ft ft ft 0.967 0.0021 -0.0269 0.0500 0.0002 0.0037 <0.0001 <0.0001 <0.0001 Results Data from 702 longline sets were used in analyses and resulted in size and condition (i.e., live or dead) informa- tion on 4290 individual blue shark. From these data, a total of 37 proportions (i.e., PDA values) were calculated shark size and set duration had significant effects on and used in regression analyses. PDA (r2=0.86, n=37, P<0.00001; Table 1). Plots of the Most of the sets targeted swordfish (39%) or sword- fish and tuna (36%), or unspecified tuna species (14%). Bigeye tuna and yellowfin tuna were the target of 8% and 3% of the sets, respectively. About 88% of the sets included in the analysis were characterized as "night sets" and the remaining were "day sets." Overall, more blue shark were released alive (69%) than dead. Shark sizes, water temperatures, and set durations used in the multiple linear regression ranged from 75 to 300 cm FL (median=175 cm), 8 to 29°C (median=19°C), and 6 to 14 hours (median=12), respec- tively. About 68% of all released animals measured less than the size of sexual maturity (i.e., <185 cm FL). Multiple linear regression indicated that no interac- tion terms were statistically significant and that only observed proportions and the predicted response surface illustrate how the proportion of live releases increases with shark size and decreases with duration of set (Fig. 2, A and B). Whereas set duration has a moderate im- pact on the largest size classes, the proportion of live sharks <185 FL (i.e., immature stages) is consider- ably reduced even at relatively short set durations. For example, predicted PDA for the smallest sharks (i.e., FL=75 cm) was 0.67 and 0.47 for set durations of 6 and 14 hours, respectively; for those animals measuring 250 cm FL, it was 0.94 and 0.80 for the same set durations. Maps of mean PDA values and of the proportion of imma- ture sharks caught indicated conspicuous differences off the U.S. east coast versus over the Grand Banks (Fig. 3, A and B). Specifically, the proportion of live releases 722 Fishery Bulletin 103(4) tended to be lower over the Grand Banks than off the U.S. east coast and the mean ratio of immature blue shark tended to be higher. Discussion Our results indicate that blue shark tolerance to the stresses associated with longline capture decreases with animal size at levels that vary with set duration. These results are consistent with the findings of Neilson et al. (1989) and Milliken et al. (1999) who observed greater discard mortality among the smallest sizes classes of Figure 2 (A) Observed proportions of blue shark discarded alive (ra = 37) for each fork-length set duration combination; and (B) predicted response surface. Iongline-caught Atlantic halibut (Hippoglossus hippo- glossus) and cod (Gadus morhua), respectively. In our study, set duration represented the maximum possible time a given fish was "on-hook," and thus was only the coarsest of measures of the magnitude and duration of stress experienced by hooked fishes. Nevertheless, this crude measure appears to have captured enough of the cumulative stress effects on fish survival to emerge as a significant factor. In contrast, water temperature did not emerge as important in our analysis. However, we suspect this resulted because surface water tempera- tures (the only temperature measurements available) are poor indicators of the levels and changes in temperature actually experienced by captured sharks. Presumably, better predictions of condition at boat-side (and thus live discard quantities) could be made with knowledge of time-on-hook, depth, and temperature of capture, rate of gear retrieval, sea conditions, etc. Unfortunately, many of the measurements that are likely most relevant to recording shark condition at boat-side can only be made by distributing and retrieving large quantities of electronic instruments (i.e., temperature-depth recorders and hook-timers, see Boggs, 1992) near the hooks, and for each set. Such an approach is not only costly, but also difficult to implement without a research team dedicated for this purpose. Similarly, only by directed research can questions of postrelease mortality be addressed. Clearly, the proportions of living blue shark considered in our study are minimum estimates of fishing impacts because they do not account for delayed mortality of individuals released injured or otherwise impaired. For gauging postrelease mortality of Iongline-caught blue shark, tagging studies are warranted (Neilson et al., 1989; Kohler et al., 2002). Evident in the maps is that the proportion of blue sharks available for live release was not homogeneous throughout the spatial range examined. Overall the proportion of blue shark released alive was higher (0.78) along the U.S. Atlantic east coast and decreased over the Grand Banks (0.67) (Fig. 3A). The maps also indicated that overall the proportion of immature blue sharks was highest over the Grand Banks (0.93) compared to the U.S. Atlantic east coast (0.63) (Fig. 3B). In their exami- nation of U.S. Atlantic east coast longline catches south of the present study (i.e., between 35° and 22°N latitude), Beerkircher et al. (2004) found that 0.87 of blue shark caught were alive at boat-side. It seems likely, therefore, that contributing to the relatively higher survival ob- served by Beerkircher et al. (2004) was that only about half of the blue shark in their analysis were immature (as inferred from size). Blue shark interactions over the Grand Banks deserve special attention because most in- dividuals discarded by the U.S. pelagic longline fleet are captured in that area. In 2002, for example, two thirds of the estimated 4335 blue shark mortalities attributed to U.S. Atlantic pelagic longline fleet were captured in this area (Diaz, unpubl. data3). Diaz, G. A. 2005. NMFS Pelagic longline logbook pro- gram. NMFS/SEFSC Miami, FL 33149. NOTE Diaz and Serafy: Factors affecting the number of Pnonoce glauca available for live release in fisheries 723 V ^ 1 ■*""< &-*< Atlantic fs**- Ocean 50°N 70" W 60" W 50"W 40°W 40"N 50"N 70"W 60°W 40:N 50"W 40"W 50"N - 40 N 70 W 60"W 50"W 40"W U 0 0 |m 0.4-0 6 HI 0.0-0.2 |] 0.6-0.8 H7?] 0.2-0 4 ■ 0.8-1.0 50N 40°N 70:W 60°W 50,:-W 40"W Figure 3 (Ai Average proportion of blue shark released alive and (B) average proportion of immature blue shark released in pelagic longline sets. Proportions were estimated for 0.5-degree cells where at least one longline set was deployed in the period 1992-2002. Ward et al. (2004) modeled the effect of set duration on pelagic longline catches and found that blue shark catch rates increased with set duration. According to our results, the increase in set durations also leads to increases in the number retrieved dead. In concept, a possible management measure to achieve reductions in blue shark mortality may include shortening long- line set durations. However, a regulation of this nature would be difficult to implement (let alone enforce) be- cause swordfish catch rates are also lowered when set durations are shortened (Ward et al., 2004) and there- fore result in negative economic impacts that would likely be unacceptable to the industry. Results of this analysis also have implications for blue shark stock assessment. Stock assessments based on longline fisheries data often use a hook selectivity function of a logistic form, whereby hook retention is 100% for fish larger than a certain size. In the particu- 724 Fishery Bulletin 103(4) lar case of blue shark, where most individuals caught are released (dead or alive), fishing mortality is best estimated from the number of animals released dead, rather than from all animals caught. Because larger animals have a higher probability of being released alive, a logistic selectivity function without size or age survival adjustment, could lead to overestimation of impacts on the stock. Thus, a dome-shaped selectivity function that incorporates the size-based survival infor- mation presented in the present study may represent an improvement over current techniques. Acknowledgments We thank L. Brooks, E. Cortes, S. Turner, and two anonymous reviewers for invaluable comments on the manuscript. Literature cited Bailey, K., P. G. Williams, and D. Itano. 1996. By-catch and discards in Western Pacific tuna fisheries: a review of SPC data holdings and literature. Tech. Rep. Ocean Fish. Programme. S. Pac. Comm. no 34, 148 p. Beerkireher, L. R., E. Cortes, and M. Shivji. 2004. Characteristics of shark bycatch observed on pelagic longlines off the Southeastern United States, 1992-2000. Mar. Fish. Rev. 64(41:40-49. Boggs, C. H. 1992. Depth, capture time and hooked longevity of longline-caught pelagic fish: timing bites offish with chips. Fish. Bull. 90:642-658. Draper, N., and H. Smith. 1981. Applied regression analysis, 709 p. Wiley Inter- science, New York. NY. Francis, M. P. 1998. New Zealand shark fisheries: development, size and management. Mar. Freshw. Res. 49(7):579-591. Francis, M. P., L. M. Griggs, and S. J. Baird. 2001. Pelagic shark by-catch in the New Zealand tuna longline fishery. Mar. Freshw. Res. 52(21:165-178. Hocking R. R. 1976. The analysis and selection of variables in linear regression. Biometrics 32:1-50. Kohler, N. E., P. A. Turner. J. J. Hoey, L. J. Natanson, and R. Briggs. 2002. Tag and recapture data for three pelagic sharks species: blue shark (Prionacea glauca), shortfin mako (Isurus xyrinchus), and porbeagle lLamna nasus) in the North Atlantic ocean. International Commis- sion for the Conservation of Atlantic Tuna (ICCAT) 54(41:1231-1260. Macias D., and J. M. de la Serna. 2002. By-catch composition in the Spanish Mediterra- nean longline fishery, 198 p. Proc. 4th meeting of the European Elasmobranch Association. Societe Francaise d'Ichtyologie, Paris, France. Milliken. H. O., M. Farrington, H. A. Carr, and E. Lent. 1999. Survival of Atlantic cod (Gadus morhua) in the Northwest Atlantic longline fishery. Mar. Technol. Soc. J. 33:19-24. Neilson, D. J., K. G. Waiwood, and S. J. Smith. 1989. Survival of Atlantic halibut (Hippoglossus hippo- glossus) caught by longline and otter trawl gear. Can. J. Fish. Aquat. Sci. 46:887-897. Pratt, H. L„ Jr. 1979. Reproduction in the blue shark, Prionace glauca. Fish. Bull. 77:445-470. Robins, C. R., and G. C. Ray. 1986. A field guide to Atlantic coast fishes of North America. Peterson Field Guide Series, 354 p. Houghton Mifflin, Boston, MA. Stevens, J. D. 1992. Blue and mako shark by-catch in the Japanese longline fishery off south-eastern Australia. Sharks: biology and fisheries. Aust. J. Mar. Freshw. Res. 43(11:227-236. Sokal, R. R., and F. J. Rohlf. 1981. Biometry, 2nd ed.. 859 p. W. H. Freeman. New York, NY Ward, P., R. A. Myers, and W. Blanchard. 2004. Fish lost at sea: the effect of soak time on pelagic longline catches. Fish. Bull. 102:179-195. 725 Length correction for larval and early-juvenile Atlantic menhaden (Brevoortia tyrannus) after preservation in alcohol Dariusz P. Fey Sea Fisheries Institute Dept. of Fisheries Oceanography and Marine Ecology ul. Kollataia 1 81-332 Gdynia, Poland E-mail address dfeyg'mirgdynia pi Jonathan A. Hare NOAA National Ocean Service Center for Coastal Fisheries and Habitat Research 101 Pivers Island Road Beaufort, North Carolina 28516-9722 Body length measurement is an im- portant part of growth, condition, and mortality analyses of larval and juvenile fish. If the measurements are not accurate (i.e., do not reflect real fish length), results of subsequent analyses may be affected consider- ably (McGurk, 1985; Fey, 1999; Porter et al., 2001). The primary cause of error in fish length measurement is shrinkage related to collection and preservation (Theilacker, 1980; Hay, 1981; Butler, 1992; Fey, 1999). The magnitude of shrinkage depends on many factors, namely the duration and speed of the collection tow, abun- dance of other planktonic organisms in the sample (Theilacker, 1980; Hay, 1981; Jennings, 1991), the type and strength of the preservative (Hay, 1982), and the species of fish (Jen- nings, 1991; Fey, 1999). Further, fish size affects shrinkage (Fowler and Smith, 1983; Fey, 1999, 2001), indi- cating that live length should be mod- eled as a function of preserved length (Pepin et al., 1998; Fey, 1999). The goal of this study was to ana- lyze the shrinkage of late-larval and early-juvenile Atlantic menhaden (Brevoortia tyrannus) during pres- ervation in 95% alcohol. A length correction formula is presented that allows live standard length to be calculated from preserved standard length. Materials and methods Larval and early juvenile Atlantic menhaden were collected on three different occasions during January- March 2003 with a neuston net (2-m2 opening and 947-,um mesh) deployed for 2-minute durations from a bridge to Pivers Island, located about 2 km inside Beaufort Inlet, North Caro- lina. Samples were placed in a cooler and transported to the laboratory. Live Atlantic menhaden larvae were sorted from the samples (?i=100) and their standard lengths (SL) were mea- sured to the nearest 0.01 mm with a caliper. All specimens (19.1-31.4 mm SL) were placed in individual vials filled with 95% ethyl alcohol. The fish were remeasured 3. 20, and 90 days after preservation. Repeated measures ANOVA and Tukey HSD tests were used to ana- lyze the significance of length changes during 90 days of preservation. The preserved length after 90 days was than compared with live length to test whether a single correction factor is appropriate for a calculation of live length (/-test analysis for the slope difference from one). Additionally, the precision of measurements was evaluated by two replicate measure- ments of all larvae three days after preservation. Linear regression anal- ysis was then used to describe the relationship between the two length measurements. The possible deviation of intercept from zero and slope from one was estimated (/-test) to test for the possible significant differences between the two measurements. Results Time in preservative had a significant effect on measured length of Atlantic menhaden larvae (repeated measures ANOVA, P<0.0001). Fish were sig- nificantly larger prior to preservation compared to three days after pres- ervation (Tukey HSD, P<0.001) and significantly larger three days after preservation compared to 20 and 90 days after preservation (Fig. 1) (Tukey HSD, P<0.001). When shrink- age is described as a relative value, the change in length that occurred during the first three days of preser- vation was 3.62%. Length decreased by an average of 0.22% during the fol- lowing 17 days and by 0.073% during the remaining 70 days. Although smaller fish shrank pro- portionally more than the larger ones (/-test for H0: slope = 0, P<0.001) (Fig. 2A), no size effect was observed when shrinkage was analyzed as absolute length (regression slope = 0.996, SE = 0.008; HQ: slope=l; /-test of regression slope, P=0.605). How- ever, the ^-intercept of the regres- sion of preserved length at 90 days on live length was significantly different from zero (regression in- tercepts.17; SE = 0.21; H0: y-inter- cept=0; /-test of regression intercept, P<0.001) (Fig. 2B). Therefore, the significantly different from zero in- tercept can be used as a correction factor (i.e., SLfresh = SLpreserl.ed+l.l7 mm). The shrinkage magnitude observed by Maillet and Checkley (1991) was compared to the results derived in our study (Fig 2B). Their formula indicated shrinkage of about 2% compared to approximately 4% in our study. Manuscript submitted 31 March 2004 to the Scientific Editor's Office. Manuscript approved for publication 8 February 2005 by the Scientific Editor. Fish. Bull. 103:725-727 (2005). 726 Fishery Bulletin 103(4) The two readings performed to estimate the measure- ment error were not statistically different as indicated by the parameters of the regression line fitted to the first measurement versus second measurement data (SL1 = 0.992 SL2 + 0.21, /-2 = 0.998). The slope was not statistically different from one (regression slope=0.992, SE = 0.005; H0: slope = l; ?-test of regression slope, P=0.106), and the intercept was not statistically differ- ent from zero (regression intercept=0.21; SE = 0.12; H0: •10 0 10 20 30 40 50 60 70 80 90 Time of preservation (days) Figure 1 Change in standard length of Atlantic menhaden (Brevoortia tyrannies) (n=100) during 90 days of preservation in 95% alcohol. Mean values and standard error of length measure- ments obtained from repeated measurements of 100 fish. y-intercept=0; r-test of regression intercept, P=0.418). Measurement precision, the absolute values of the dif- ference between the two series of length measurements of the same specimens averaged 0.12 mm (SD = 0.09), which corresponded to an average of 0.47% of length (SD = 0.35). Thus, the error associated with measure- ment is an order of magnitude less than the change in length due to shrinkage within the first three days of preservation. Changes in length during following 87 days were below measurement error. Discussion This research on late-larval and early-juvenile Atlan- tic menhaden shrinkage is the first for this spe- cies. Maillet and Checkley (1991) used a shrinkage correction formula (cited as unpubl. data) in their study on larval menhaden growth but did not provide additional information (e.g., range offish sizes) to accompany their formula. Their correction formula differs from ours, and the discrepancy may be related to differences in experimental procedure and differ- ent developmental stages. In the present study live fish were used, but in Maillet and Checkley 's study (1991) it was not indicated whether larvae were alive or dead prior to preservation. Further Maillet and Checkley (1991) examined larval menhaden (17-24.5 mm SL), whereas we examined late-larval to early- juvenile menhaden (19.1-31.4 mm SL). The shrinkage of larval and early-juvenile Atlan- tic menhaden after the first three days of preserva- tion was significant, but small in magnitude. Be- yond 20 days of preservation significant additional shrinkage did not occur. In fact, the length changes after day 3 were below the estimated measurement 7 A 33 6 © o o o o £ 31 5 4 <>**©—- o ■■: © CO © ©0 © ©o E x: c T3 29 27 25 3 - o o © o o OJ o o o C CO 23 •i y = -0.150(x) + 8 01 © © en CD > 21 1 rJ = 0.205 © © _l 19 0 _i 1 — 1 I 1 ' ' J 1 1 19 21 23 25 27 29 Live standard length (mm) 31 33 B - & ' 4^^ LSL = 0.996(PSL) + r2 = 0.993 17 19 21 23 25 27 29 31 Preserved standard length (mm) 33 Figure 2 Length changes of Atlantic menhaden (Brevoortia tyrannus) during preservation for 90 days in 95% alcohol (;i = 100). (A) The relationship between live standard length (LSL) and relative (%) shrinkage magnitude; (B) the relationship between live and preserved standard lengths described with linear regression. The solid line indicates the 1:1 ratio. The arrow points to the correction curve obtained from Maillet and Checkley (1991): SL/„,,. = 0.978(SL ic1 -s " NOTE Fey and Hare: Length correction for larval and early-|uvemle Brevoortia tyrannus 727 error. Additionally, decreasing shrinkage as a function of increasing fish length was present when relative (%) shrinkage was analyzed. Similar results with regard to time and fish size effect were previously reported for other fish species preserved with formalin and alcohol (see Fey, 1999, for overview). The effect of shrinkage on growth rate analysis was described by Fey (1999) for larval sprat. If growth rate is estimated by using a regression of length at age, the influence of shrinkage on growth estimates depends on the absolute value of length changes (i.e., expressed in mm) among small and large fish, and the error may be as high as 0.07 mm/d. However, if the absolute values of length decrease equally across fish lengths, even large shrinkage (on average) may have no effect on the results of growth rate analysis. In addition to length at age analysis, average growth rate (mm/d) may be calculated for individual fish. The potential error in growth estimates will then be directly proportional to both the relative and absolute magnitude of shrink- age. This potential bias in growth-rate calculations described by Fey (1999) for sprat emphasizes the im- portance of correcting for preservation. Although the relationship between otolith size and fish size may be used for length correction (Leak, 1986; Radtke, 1989). Fey (1999) showed that greater accuracy is provided when a fresh length-preserved length relationship is used. However, such a relationship may be supple- mented by additional measurements (i.e., body depth and otolith size) to improve the accuracy of the correc- tion model (Porter et al.. 2001). In the current study, absolute changes in length (expressed in mm) of alcohol- preserved menhaden were not dependent on fish size and therefore a single correction factor was sufficient for a calculation of live length. The length correction factor provided in our study will benefit future studies on the ecology of early life stages of menhaden, similar to that conducted by Warlen et al. (2002), where pre- served length measurements were used. Acknowledgments This research was performed while the first author held a National Research Council Research Associateship Award at NOAA Beaufort Laboratory. This note is also a contribution to the State Committee for Scientific Research (grant no. 2P04F 005 27). Literature cited Butler. J. L. 1992. Collection and preservation of materials for oto- lith analysis. /;; Otolith structure examination and analysis (D. K. Stevenson and S. E. Campana, eds. i. p. 13-17. Can. Spec. Pub. Fish. Aquat. Sci. 117. Fey, D. P. 1999. Effects of preservation technique on the length of larval fish: methods of correcting estimates and their implication for studying growth rates. Arch. Fish. Mar. Res. 47:17-29. 2001. Length correction of larval and early-juvenile her- ring tClupea harengus) and smelt [Osmerus eperlanus > after preservation in formalin and alcohol. Bull. Sea Fish. Inst, ll 1551:47-51. Fowler, G. M., and S. J. Smith. 1983. Length changes in silver hake (Merluccius bilinearis ) larvae: effects of formalin, ethanol, and freezing. Can. J. Fish. Aquat. Sci. 40:866-870. Hay. D. E. 1981. Effects of capture and fixation on gut contents and body size of Pacific herring larvae. Rapp. P.-V. Reun. Cons. Int. Explor. Mer 178:395-400. 1982. Fixation shrinkage of herring larvae: effects of salinity, formalin concentration, and other factors. Can. J. Fish. Aquat. Sci. 39:1138-1143. Jennings, S. 1991. The effects of capture, net retention and preserva- tion upon lengths of larval and juvenile bass, Dicentrar- chus labrax iL.). J. Fish Biol. 38:349-357. Leak. J. C. 1986. The relationship of standard length and oto- lith diameter in larval bay anchovy, Anchoa mitchilli iVal.). A shrinkage estimator. J. Exp. Mar. Biol. Ecol. 95:17-23. Maillet, G. L., and D. M. Checkley Jr. 1991. Storm-related variation in the growth of otolith of larval Atlantic menhaden Brevoortia tyrannus: a time series analysis of biological and physical variables and implications for larva growth and mortality. Mar. Ecol. Prog. Ser. 79:1-16. McGurk. M. D. 1985. Effect of net capture on the postpreservation mor- phometry, dry weight, and condition factor of Pacific herring larvae. Trans. Am. Fish. Soc. 114:348-355. Pepin. P.. J. F. Dower, and W. C. Legget. 1998. Changes in the probability density function of larval fish body length following preservation. Fish. Bull. 96:633-640. Porter, S. M., A. L. Brown, and K. M. Bailey. 2001. Estimating live standard length of net-caught walleye Pollock (Theragra chalcogramma) larvae using measurements in addition to standard length. Fish. Bull. 101:384-404. Radtke, R. L. 1989. Larval fish age, growth, and body shrinkage: infor- mation available from otoliths. Can. J. Fish. Aquat. Sci. 46:1884-1894. Theilacker, G. H. 1980. Changes in body measurements of larval northen anchovy, Engrciulis mordax, and other fishes due to handling and preservation. Fish. Bull. 78:685-692. Warlen, S. M., K. W. Able, and E. H. Laban. 2002. Recruitment of larval Atlantic menhaden [Brevoor- tia tyrannus) to North Carolina and New Jersey estuaries: evidence for larval transport northward along the east coast of the United States. Fish. Bull. 100:609-623. 728 Comparison of average larval fish vertical distributions among species exhibiting different transport pathways on the southeast United States continental shelf Jonathan A. Hare John J. Govoni Center for Coastal Fisheries and Habitat Research 101 Pivers Island Road Beaufort. North Carolina 28516 Present address (for J A. Hare): Narragansett Laboratory Northeast Fisheries Science Center 28 Tarzwell Drive Narragansett, Rhode Island 02882 E-mail address (for J A Hare) ion hareia'noaa.gov Water currents are vertically struc- tured in many marine systems and as a result, vertical movements by fish larvae and zooplankton affect horizontal transport (Power. 1984). In estuaries, the vertical movements of larvae with tidal periods can result in their retention or ingress (Fortier and Leggett, 1983; Rijnsdorp et al., 1985; Cronin and Forward. 1986; For- ward et al., 1999). On the continental shelf, the vertical movements of organ- isms interact daily and ontogeneti- cally with depth-varying currents to affect horizontal transport (Pillar et al., 1989; Barange and Pillar, 1992; Cowen et al., 1993, 2000; Batchelder et al., 2002). A suite of fish species, which use estuaries during the juvenile stage, spawn during winter on the mid- and outer continental shelf of the south- east United States (Fig. 1A): Brevoor- tia tyrannus (Atlantic menhaden), Leiostomus xanthurus (spot), Micropo- gonias undulatus (Atlantic croaker), Paralichthys albiguta (Gulf flounder), P. dentatus (summer flounder), and P. lethostigma (southern flounder). Ver- tically structured flow is a major part of proposed larval transport mecha- nisms for these species from offshore spawning areas to estuarine nurs- ery habitats (Govoni and Pietrafesa, 1994; Hare et al., 1999). Brevoortia tyrannus, however, is found higher in the water column on average than the other species that use estuaries during their juvenile stage (Miller et al., 1984; Govoni and Pietrafesa, 1994; Govoni and Hoss, 2001). Fur- ther, larvae of B. tyrannus apparent- ly exhibit a difference in horizontal transport compared to other winter- spawning species that use estuarine habitats as juveniles; B. tyrannus lar- vae spawned on the southeast U.S. shelf may be transported to estuarine nursery habitats along the northeast U.S. shelf (Warlen et al., 2002). The effects of differences in vertical lar- val distribution on cross-shelf lar- val transport are unknown, and the transport pathways from shelf spawn- ing areas to estuarine nursery areas remain unclear. Other species also spawn during winter on the southeast United States continental shelf. Some species settle to benthic habitats on the shelf (e.g., Etr'opus cyclosquamus [shelf floun- der], E. microstomus [smallmouth flounder], and E. rimosus [grayfloun- der], Leslie and Stewart, 1986) or re- main on the shelf in pelagic habitats (e.g., Etrumeus teres [round herring], Crawford, 1981; Schwartz, 1989). However, some species are regularly advected offshore, entrained into the Gulf Stream, and exported north- wards (e.g., Bothus spp. [peacock, eyed, and spotted flounders], Pepri- lus triacanthus [butterfish], Syacium papillosum [dusky flounder], Xyrich- tys novacula [pearly razorfish]; Hare and Cowen, 1991; Cowen et al., 1993; Rotunno and Cowen, 1997; Grothues and Cowen, 1999). The purpose of our study was to ex- amine associations between average larval fish vertical distributions and general larval transport pathways on the southeast United States conti- nental shelf during winter. Our goal was to determine if larval vertical distributions differed among species that exhibit different outcomes of lar- val transport: export from the local shelf, arrival at local estuaries, and retention on the shelf. Our approach, however, was unconventional. Rather than couple detailed descriptions of the flow field with detailed describi- tions of larval vertical distributions (including diel variation), we chose to compare average vertical distributions among species that exhibit overall differences in larval transport. Verti- cal distribution data were collected in three separate years, over periods of time ranging from 24 to 96 hours. If average larval vertical distributions are different among species, and these differences occur consistently among the various sampling times and in concordance with the general outcome of transport, then we conclude that larval vertical distributions are an important part of larval transport on the southeast U.S. shelf. Our specific objectives were two- fold: 1) to test the null hypothesis that there are no differences in lar- val fish vertical distributions between species, and 2) to evaluate significant differences in larval depth distribu- tion in relation to the a priori clas- sification of the outcome of transport. Vertically discrete data from six sam- pling times were analyzed, and ow- ing to differences in protocols among sampling times, comparisons of lar- val vertical distributions were made within sampling times only. The re- sults of these comparisons were then combined to evaluate whether there were consistent differences in larval vertical distributions among sampling times related to the outcome of larval transport. Manuscript submitted 5 April 2004 to the Scientific Editor's Office. Manuscript approved 30 March 2005 by the Scientific Editor. Fish. Bull 103:728-736(2005). NOTE Hare and Govoni: Larval fish transport and vertical distributions on the southeast US continental shelf 729 Cape Hatteras 16 35 34 A 1986 ■ 1989 • 1991 33 77 Longitude °W 75 Figure 1 (A) Map of the east coast of the United States rotated 18° counter-clockwise. The spatial extent of the northeast and southeast United States continental shelves is indicated by each rectangle. The area of panel B is shown as a trapezoid. (B) The northern portion of the southeast United States continental shelf showing the coastline, the 10-m, 20-m, 30-m 40-m, 50-m, and 100-m isobaths. The three prominent capes are identified and locations of stations sampled in this study are shown. Material and methods Data collection Larval fish were collected every six hours (0600, 1200, 1800, and 2400) at an offshore and an inshore station during three winters: 21-26 February 1986, 26 January- 1 February 1989, and 5-7 February 1991 (Fig. IB). Offshore stations were located on approximately the 50-m isobath, and inshore stations were located on approximately the 35-m isobath. In 1986, offshore and onshore stations were occupied for 102 and 48 h, respec- tively. Collections were taken horizontally at 1, 18, and 32 m at the offshore station and 1, 13, and 25 m at the inshore station with a 60-cm opening-closing bongo net (Weibe and Benfield, 2003) with 333-^m mesh and a 1-m2 Tucker trawl ( Weibe and Benfield, 2003 ) with 202-f by using a test of independence (Pearson chi-square. Sokal and Rohlf, 1981; McCleave et al. 1987). Depth distributions were averaged over each station. Compari- sons were then made between all pairs of taxa within a station, and a Bonferoni correction was applied to assess the significance of the tests of independence. Comparisions were not made between stations, because sampling methods varied and depth distributions were not directly comparable. The following null hypothesis was evaluated: during each station occupation, average larval depth distributions were independent of species. Column and row variables were species and depth strata; cell values were the average proportion of the larvae captured in a depth stratum at a station. Comparisons of center of mass were also made and the results were very similar to the results of the test of independence reported in the present study. The calculation of average proportion was made in two steps. First, the proportion of larvae (P) collected in each depth stratum id) at each sampling time (i) during each station occupation (J) was calculated: the individual species comparisons were pooled across station by the a priori assigned outcome of transport. The number of significant differences found between species were then compared to the number of significant differences expected with a 5% error rate by using the G-statistic (Sokal and Rohlf 1981). For example, in a comparison of B. tyrannus to exported species, five pairwise comparisons of larval depth distributions were found to be significantly different and 12 were not sig- nificantly different. At «=0.05, one significant and 16 nonsignificant differences are expected. The G-statistic demonstrates that more significant differences were found between B. tyrannus and exported species than expected by chance. The classifications of significant depth differences (shallower, deeper, different) were then examined to determine the relation between larval vertical distributions and the general outcome of larval transport. Results dij 'dij where C = larval concentration 100/m3. Then the average proportion of larvae (P) for each depth stratum (d) was calculated for each station (J): IP* where nn= the number of sampling times (;') during station occupation (J). Because the significance of a test of independence depends, in part, on the magnitude of the cell values (i.e., sample sizes), average larval concentration of each species during each station occupation (number of lar- vae/100 m3) was used as a weighting factor. The av- erage proportion of larvae at depth during a station occupation (Pd.) was multiplied by the weighting factor to derive the cell values for use in the test of indepen- dence. The weighting factor approximated the number of fish larvae collected, and incorporated the effect of variability in sampling volume. Values of the standardized residuals, which are a result of the test of independence, were used to classify significant differences in depth distribution as follows: species A shallower (<) than species B, species A deeper (>) than species B, and species A distributed differently (< or >) than species B. This last category was assigned when one species was not clearly deeper or shallower than the other species, yet its depth distributions were significantly different. To evaluate whether larval fish vertical distributions were associated with larval transport, the results of Comparison of larval vertical distributions indicated that B. tyrannus often had the shallowest larval verti- cal distribution. There were more significant differences than expected by chance between the vertical distribu- tions of B. tyrannus and exported, estuarine, and shelf- resident taxa (Table 1). For all significant differences, the standard deviates from the test of independence indicated that B. tyrannus were found in shallower water than were other taxa (Appendix 1). Exported taxa generally were higher in the water col- umn than estuarine and shelf-resident taxa. There were more significant differences than expected by chance between the vertical distributions of exported taxa and estuarine and shelf-resident taxa (Table 1). Further, 9 of 12 significant differences between exported and es- tuarine taxa indicated that exported taxa were found in shallower water; 8 of 11 significant differences between exported and shelf resident taxa indicated that exported taxa were found in shallower water (Appendix 1). The vertical distributions of estuarine and shelf-resi- dent taxa were different more often than expected by chance, but taxa of neither group were consistently found in shallower water (Table 1). Significant differ- ences in larval vertical distributions were distributed evenly among the three classifications of the direction of difference (»=4 shallower; n=2 deeper; ;; = 5 different) (Appendix 1). Discussion The results indicate an overall hierarchy of larval ver- tical distributions; B. tyrannus was found in shallower water than were exported taxa, and exported taxa were shallower than estuarine and shelf-resident taxa. Although this general pattern emerged, considerable variability in larval vertical distributions was observed, which is a common result of many studies (e.g., Boehlert NOTE Hare and Govoni: Larval fish transport and vertical distributions on the southeast US continental shelf 731 Table 1 Summary of the pairwise comparisons of larval depth distributions between species classified by the a priori outcome of trans- port. In each table cell, the number to the left is the number of significant pairwise differences, the number to the right is the total number of comparisons across the six station occupations, and the number in parentheses is the G-statistic for evaluating the null hypothesis that the number of observed differences is as expected with a 5% error rate. The critical value at «=0.05 is 5.99 and significant values are indicated in bold. Values greater than 5.99 indicate that there are more significant differences between species than expected by chance. Exported taxa are Bothus spp., Peprilus triacanthus, Syacium papillosum, Xyrichtys novacula. Estuarine taxa include Leiostomus xanthurus, Micropogonias undulatus, and Paralichthys spp. Shelf resident taxa include Etropus spp. and Etrumeus tei'es. A priori classification of the outcome of transport Brevoortia tyrannus Exported Estuarine Shelf resident Exported Estuarine Shelf resident 5/ 17(10.29) 12/15(55.35) 9/ 12(39.11) 2/17(1.17) 12/43(23.88) 11/34(25.09) 5/13(12.96) 11/30(27.96) 2/6(4.39) and Mundy, 1994; Brodeur and Rugen, 1994). Variability in larval fish vertical distributions (and zooplankton) is related to processes that influence water column mixing (e.g.. Heath et al„ 1988; Incze et al., 2001) and to spe- cies-specific responses to diel cycles and gradients in turbulence, temperature, and salinity (DeVries et al. 1995; Olla et al., 1996; Gray and Kingsford, 2003). The approach used in the present study was to average over shorter-scale variability (hours) in larval vertical dis- tributions to examine longer-time-scale patterns (days) in larval vertical distributions. Average larval vertical distributions of exported, estuarine-dependent. and shelf-resident taxa and the implied outcomes of their larval transport are consis- tent with the results of physical oceanographic models and observations of shelf circulation in the southeast United States continental shelf. The model of Janowitz and Pietrafesa (1980) (see also Miller et al., 1984) in- dicated a three-layered, cross-shelf flow during winter: surface and near-bottom offshore flow, and intermedi- ate onshore flow. Similarly, the model of Werner et al. (1999) indicated a two-layered, cross-shelf flow during winter: an offshore flow near the surface and onshore flow throughout the rest of the water column. Surface flow in the study area during winter is typically off- shore (Govoni and Pietrafesa, 1994). On the inner and middle shelf (water depths <40 m), average bottom flow is onshore; on the outer shelf (water depth 40-75 m), average intermediate flow is onshore, whereas bottom flow is offshore (Fig. 5b in Lee et al.. 1989). Modeled and observed flow fields may indicate that larvae in the surface water will move offshore (exported taxa), where the probability of entrainment into the Gulf Stream is higher. Larvae that are in the middle or lower portion of the water column will move onshore (i.e., estuarine- dependent and shelf-resident taxa). Thus, the average larval vertical distributions, the general outcome of larval transport, and the generalized observed and modeled vertical flow fields are consistent. Differences between vertical distributions of larval B. tyrannus and the other estuarine-dependent taxa (Fig. 2; see also Govoni and Pietrafesa, 1994) imply differences in cross-shelf transport. There are several possibilities, none mutually exclusive. 1) Onshore trans- port of larval B. tyrannus occurs with northeast wind events and onshore transport of other estuarine-depen- dent larvae occurs with southwest or northwest wind events. This possibility is supported by the model simu- lations of Hare et al. (1999). 2) Cross-shelf transport of B. tyrannus larvae occurs in surface Gulf Stream intrusions (Checkley et al., 1988; Stegmann and Yoder, 1996), whereas cross-shelf transport of other estuarine- dependent larvae occurs by wind-driven mechanisms. This possibility has not been adequately evaluated. 3) All estuarine-dependent larvae are transported across the shelf by the same mechanisms, but the rate of their transport differs. For example, southwest wind events cause onshore transport rates to be greater for the other estuarine-dependent taxa because B. tyrannus larvae spend less time in the intermediate portion of the wa- ter column. This possibility is also supported by Hare et al. (1999), who found that in modeled larval vertical distributions, the outcome of larval transport was modi- fied by circulation. From these alternative hypotheses, it is clear that our understanding of the cross-shelf transport of larval fishes remains incomplete and that the effective physical and biological mechanisms are complex. One approach to resolving the affect of vertical dis- tribution on cross-shelf larval transport is to develop a specific hypothesis regarding supply of larvae to inlets that is based on the above possibilities and then to test these hypotheses using the long time-series of larval ingress collected at Beaufort Inlet (see Warlen, 1994). Three alternative patterns in ingress, based on the three possibilities presented above, could be evaluated by using ingress data collected at Beaufort Inlet: 1) in- gress of B. tyrannus occurs during northeast winds, and 732 Fishery Bulletin 103(4) Inshore- 1986 K mm £ 15l q- r Bt Sp Bs Pt Xn Ps Lx Mu Et Es Inshore- 1989 Offshore- 1986 1- h £$ i-.a- r -3- ■ :t ^ §- ■f- f- g- s- T.F.. §- S- §- Bt Sp Bs Pt Xn Ps Lx Mu Et Es Offshore- 1989 iEffl EE B- i- Bt Sp Bs Pt Xn Ps Lx Mu Et Es Inshore- 1991 Bt Sp Bs Pt Xn Ps Lx Mu Et Es Offshore- 1991 u Bt Sp Bs Pt Xn Ps Lx Mu Et Es Bt Sp Bs Pt Xn Ps Lx Mu Et Es Mean proportion of larval concentration at depth Figure 2 Mean proportions of larvae sampled at depths at six stations on the southeast United States shelf. Error bars indicate standard deviation of mean proportions calculated by using all the samples collected at a station. The x-axis of all panels is the same and ranges from 0 to 1.2. The species indicated in each figure is denoted by a two letter code (P>t=Brevoortia tyrannus, Sp = Syacium papillosum, B>s=Bothus spp., Pt=Peprilus triacanthus, Xn=Xyrichtys novacula, Ps=Paralichthys spp., Lx =Leiostomus xanthurus, Mu=Micropogonias undulatus, Et=Etrumeus teres, and Es=Etropus spp.). Species are grouped by an a priori assignment of their general outcome of transport. the ingress of other species occurs during northwest, west, and southwest winds; 2) ingress of B. tyrannus is not related to wind, and ingress of the other species is related to northwest, west, and southwest winds; 3) and ingress of all estuarine-dependent species occurs during similar wind forcing. Other studies have estab- lished similar a priori predictions for relations between wind forcing and ingress, yet results have been equivo- cal (e.g., Blanton et al., 1995). One explanation is that cross-shelf larval transport and larval ingress occur through multiple steps (Boehlert and Mundy, 1988; Het- tler and Hare, 1998), effectively decoupling wind-driven, cross-shelf larval transport from larval ingress. Similarities in vertical distributions of larval B. tyrannus and exported larval taxa indicate that a great- er proportion of B. tyrannus larvae may be entrained into the Gulf Stream than larvae of other species that use southeast estuaries as juvenile nurseries. Once entrained into the Gulf Stream, larvae are transported northeastward and they either continue to move in the Gulf Stream or are returned to the shelf edge north of Cape Hatteras by warm-core ring streamers or in dis- charges of Gulf Stream water (Hare and Cowen 1991, 1996; Churchill et al., 1993; Cowen et al., 1993; Hare et al., 2002). Govoni and Spach (1999) reported offshore exchange of B. tyrannus larvae into the Gulf Stream, and Warlen et al. (2002) concluded that some B. tyran- nus larvae spawned south of Cape Hatteras do enter estuaries north of Cape Hatteras in the spring. The mechanisms of northward transport of B. tyrannus have yet to be studied, but transport to the northeast United States shelf edge by the same mechanisms as those that drive exported taxa is possible. In marine systems, larval fish interact with verti- cally structured flow with vertical motions and thereby affect their horizontal transport (Cowen et al., 1993, 2000; Grioche et al., 2000). Apart from specific trans- port mechanisms, the present study demonstrates an overall link between larval vertical distributions and transport for multiple species. Species that moved in- shore or remained on the shelf were found deeper in the water column than species that were exported from the shelf. Cowen et al. (1993) indicated that as larvae on the northeast U.S. shelf edge move deeper, they become more susceptible to onshore flows. Similarly, Cowen et al. (2000) argued that pomacentrid larvae are distributed at mid-depths off Barbados, and these mid-depth distributions resulted in larval retention NOTE Hare and Govoni: Larval fish transport and vertical distributions on the southeast US continental shelf 733 closer to the island. Peterson (1998) proposed that in upwelling systems, copepods can affect retention on the shelf through ontogenetic vertical migrations, whereby younger stages inhabit the upper offshore-flowing wa- ter and older stages inhabit the lower onshore-flowing water (see also Peterson et al., 1979). Similar models were developed by Pillar et al. (1989) and Barange and Pillar (1992) for euphausiids in the Benguela upwelling zone. Additionally, Batchelder et al. (2002) indicated that copepods can be retained nearshore in upwell- ing systems through diel vertical migrations between offshore-flowing surface waters and onshore-flowing bottom waters. From these studies and the results from the present study, a general hypothesis emerges that in many marine systems, fish larvae and zooplankton can affect onshore transport by moving deeper in the water column. Thus, similar to selective tidal stream trans- port whereby larvae use predictable tidal flows to either remain in estuaries or enter estuaries (Forward and Tankersley, 2001), general features in circulation may exist across physical oceanographic systems that allow larvae to influence their cross-shelf transport through basic changes in their vertical distribution. Acknowledgments We thank the participants of the South Atlantic Bight Recruitment Experiment for their constructive comments throughout this study. We also appreciate the contribution of those who assisted in the field and the officers and crews of the NOAA Ships Oregon II and Chapman. Dave Colby, Frank Hernandez, Patti Marraro, Allyn Powell, Larry Settle, Petra Stegmann, and six anonymous reviewers commented on earlier drafts of this manuscript. This study was completed while the senior author held a National Research Council Research Associateship at the NOAA Beaufort Laboratory. Literature cited Barange, M., and S. C. Pillar. 1992. Cross-shelf circulation, zonation and maintenance mechanisms of Nyctiphanes capensis and Euphausia han- sent (Euphausiaceal in the northern Benguela upwelling system. Cont. Shelf Res. 12:1027-1042. Batchelder, H. P.. C. A. Edwards, and T. M. Powell. 2002. Individual-based models of copepod populations in coastal upwelling regions: implications of physiologically and environmentally influenced diel vertical migration on demographic success and nearshore retention. Prog. Oceangr. 53:307-333. Blanton. J., E. Wenner, F. Werner, and D. Knott. 1995. Effects of wind-generated coastal currents on the transport of blue crab megalopae on a shallow conti- nental shelf. Bull. Mar. Sci. 57:739-752. Boehlert, G. W., and B. C. Mundy. 1988. Roles of behavior and physical factors in larval and juvenile fish recruitment to estuarine nursery areas. Am. Fish. Soc. Symp. 3:57-76. 1994. Vertical and onshore-offshore distributional pat- terns of tuna larvae in relation to physical habitat features. Mar. Ecol. Prog. Ser. 107:1-13. Brodeur, R. D., and W. C. Rugen. 1994. Diel vertical distribution of ichthyoplankton in the northern Gulf of Alaska. Fish. Bull. 92:223-235. Cheekley. D. M., S. Raman, G. L. Maillet, and K. M. Manson. 1988. Winter storm effects on the spawning and larval drift of a pelagic fish. Nature 335:346-348. Churchill, J. H., E. R. Levine, D. N. Connors, and P. C. Cornillon. 1993. Mixing of shelf, slope and Gulf Stream water over the continental slope of the Middle Atlantic Bight. Deep-Sea Res. 40:1063-1085. Cowen, R. K., J. A. Hare, and M. P. Fahay. 1993. Beyond hydrography: can physical processes explain larval fish assemblages within the Middle Atlantic Bight? Bull. Mar. Sci. 53:567-587. Cowen, R. K., K. M. M. Lwiza, S. Sponaugle, C. Paris, and D. Olson. 2000. Connectivity of marine populations: open or closed? Science 287:857-859. Crawford, R. J. M. 1981. Distribution, availability and movements of round herring Etrumeus teres off South Africa, 1964- 1976. Fish. Bull. South Africa 14:141-181. Cronin, T. W., and R. B. Forward. 1986. Vertical migration cycles of crab larvae and their role in larval dispersal. Bull. Mar. Sci. 39:192-201. DeVries, M. C, R. B. Forward, and W. F. Hettler. 1995. Behavioral response of larval Atlantic menhaden Brevoortia tyrannus (Latrobe) and spot Leiostomus xan- thurus (Lacepede) to rates of salinity change. J. Exp. Mar. Biol. Ecol. 185:93-108. Fortier, L., and W. C. Leggett. 1983. Vertical migrations and transport of larval fish in a partially mixed estuary. Can. J. Fish. Aquat. Res. 39:1150-1163. Forward, R. B., K. A. Reinsel, D. S. Peters, R. A. Tankersley, J. H. Churchill, L. B. Crowder, W. F. Hettler, S. M. Warlen, and M. D. Greene. 1999. Transport of fish larvae through a tidal inlet. Fish. Oceanogr. 8 (suppl. 21:153-172. Forward, R. B., and R. A. Tankersley. 2001. Selective tidal-stream transport of marine ani- mals. Oceanogr. Mar. Biol. Ann. Rev. 39:305-353. Govoni, J. J., and D. E. Hoss. 2001. Comparison of the development and function of the swimbladder of Brevoortia tyrannus (Clupeidaei and Leiostomus xanthurus (Sciaenidae). Copeia 2001: 430-442. Govoni, J. J., and L. J. Pietrafesa. 1994. Eulerian views of layered water currents, vertical distribution of some larval fishes, and inferred advective transport over the continental shelf off North Carolina. USA. in winter. Fish. Oceanogr. 3:120-132. Govoni, J. J., and H. L. Spach. 1999. Exchange and flux of larval fishes across the west- ern Gulf Stream front south of Cape Hatteras, USA, in winter. Fish. Oceanogr. 8 (suppl. 2):77-92. Gray, C. A., and M. J. Kingsford. 2003. Variability in thermocline depth and strength, and the relationships with vertical distributions offish larvae and mesozooplankton in dynamic coastal waters. Mar. Ecol. Prog. Ser. 247:211-224. Grioche, A.. X. Harlay, P. Koubbi, and L. Fraqa Largo. 2000. Vertical migrations of fish larvae: Eulerian 734 Fishery Bulletin 103(4) and Lagrangian observations in the Eastern English Channel. J. Plankton Res. 22:1813-1828. Grothues, T. M., and R. K. Cowen. 1999. Larval fish assemblages and water mass history in a major faunal transition zone. Cont. Shelf Res. 19:1171-1198. Hare, J. A., J. H. Churchill, R. K. Cowen, T. Berger, P. Cornillon, P. Dragos, S. Glenn, J. J. Govoni, and T. N. Lee. 2002. Routes and rates of larval fish transport from the southeastern to the mid-Atlantic North American continental shelf. Limnol. Oceanogr. 47:1774-1789. Hare, J. A., and R. K. Cowen. 1991. Expatriation of Xyrichtys novacula (Pisces: Labri- dae) larvae: evidence of rapid cross-slope exchange. J. Mar. Res. 49:801-823. 1996. Transport mechanisms of larval and pelagic juve- nile bluefish (Pomatomus saltatrix) from South Atlantic Bight spawning grounds to Middle Atlantic Bight nursery habitats. Limnol. Oceanogr. 41:1264-1286. Hare. J. A., J. A. Quinlan, F. E. Werner, B. O. Blanton, J. J. Govoni, R. B. Forward, L. R. Settle, and D. E. Hoss. 1999. Larval transport during winter in the SABRE study area: results of a coupled vertical larval behavior- three-dimensional circulation model. Fish. Oceanogr. 8 (suppl. 2):57-76. Heath, M. R., E. W. Henderson, and D. L. Baird. 1988. Vertical distribution of herring larvae in relation to physical mixing and illumination. Mar. Ecol. Prog. Ser. 47:211-228. Hettler, W. E, and J. A. Hare. 1998. Abundance and size of larval fishes outside the entrance to Beaufort Inlet, North Carolina. Estuar. 21:476-499. Incze, L. S., D. Hebert, N. Wolff, N. Oakey, and D. Dye. 2001. Changes in copepod distributions associated with increased turbulence from wind stress. Mar. Ecol. Prog. Ser. 213:229-240. Janowitz, G. S., and L. J. Pietrafesa. 1980. A model and observation of time dependent upwell- ing over the mid-shelf and slope. J. Phys. Oceanogr. 10:1574-1583. Lee, T. N., E. Williams, J. Wang, R. Evans, and L. Atkinson. 1989. Response of South Carolina continental shelf waters to wind and Gulf Stream forcing during winter of 1986. J. Geophys. Res. 94:10715-10754. Leslie, A. J., and D. J. Stewart. 1986. Systematics and distributional ecology of Etropus (Pisces, Bothidae) on the Atlantic coast of the United States, with description of a new species. Copeia 1986:140-156. McCleave, J. D., J. J. M. Bedaux, P. G. Doucet, J. C. Jager, J. L. T. Jong, W. J. van der Steen, and B. Voorzanger. 1987. Statistical methods for analysis of plankton and nekton distribution, with application to selective tidal stream transport of juvenile American eels {Anguilla rostrata). J. Cons. Int. Explor. Mer 44:90-103. Miller, J. M., J. P. Reed, and L. J. Pietrafesa. 1984. Patterns, mechanisms and approaches to the study of estuarine dependent fish larvae and juveniles. In Mechanisms and migrations in fishes (J. D., McCleave, G. P. Arnold, J. J. Dodson, and W. H. Neill, eds.), p. 210-225. Plenum Press, New York, NY. Olla, B. L., M. W. Davis, C. H. Ryer, and S. M. Sogard. 1996. Behavioural determinants of distribution and sur- vival in early stage walleye pollock, Theragra chalco- gramma: a synthesis of experimental studies. Fish. Oceaogr. 5(suppl. 11:167-178. Pearre, S. 1979. Problems of detection and interpretation of vertical migration. J. Plankton Res. 1:29-44. Peterson, W. T. 1998. Life cycle strategies of copepods in upwelling zones. J. Mar. Sys. 15:313-326. Peterson, W. T., C. B. Miller, and A. Hutchinson. 1979. Zonation and maintenance of copepod popula- tions in the Oregon upwelling zone. Deep-Sea Res. 26A:467-494. Pillar, S. C, D. A. Armstrong, and L. Hutchings. 1989. Vertical migration, dispersal and transport of Euphausia lucens in the southern Benguela Current. Mar. Ecol. Prog. Ser. 53:179-190. Power, J.H. 1984. Advection, diffusion, and drift migrations offish larval fish. In Mechanisms and migrations in fishes (J. D., McCleave, G. P. Arnold, J. J. Dodson, and W. H. Neill, eds.), p. 27-37. Plenum Press, New York, NY. Rijnsdorp, A. D., M. van Stralen, and H. W. van der Veer. 1985. Selective tidal stream transport of North Sea plaice larvae Pleuronectes platessa in coastal nursery areas. Trans. Am. Fish. Soc. 114:416-470. Rotunno, T, and R. K. Cowen. 1997. Temporal and spatial spawning patterns of the Atlantic butterfish, Peprilus triacanthus, in the South and Middle Atlantic Bights. Fish. Bull. 95:785-799. Schwartz, F. J. 1989. Zoogeography and ecology of fishes inhabit- ing North Carolina's marine waters to depths of 600 meters. NOAA Nat. Undersea Res. Program Rep. 89- 2:335-374. Sokal. R. R., and F. J. Rohlf. 1981. Biometry, 2nd ed., 859 p. W.H. Freeman and Company, New York. NY. Stegmann, P. M., and J. A. Yoder. 1996. Variability of sea-surface temperature in the South Atlantic Bight as observed from satellite: impli- cations for offshore spawning fishes. Cont. Shelf Res. 16:843-861. Warlen, S. M. 1994. Spawning time and recruitment dynamics of larval Atlantic menhaden, Brevoortia tyrannus, into a North Carolina estuary. Fish. Bull. 92:420-433. Warlen, S. M., K. W. Able, and E. H. Laban. 2002. Recruitment of larval Atlantic menhaden (Brevoor- tia tyrannus) to North Carolina and New Jersey estuaries: evidence for larval transport northward along the east coast of the United States. Fish. Bull. 100:609-623. Wiebe, P. H., K. H. Burt, S. H. Boyd, and A. W. Morton. 1976. A multiple opening/closing net and environmental sensing system for sampling zooplankton. J. Mar. Res. 34:313-326. Wiebe, P. H., and M. C. Benfiled. 2003. From the Hensen net toward four-dimensional biological oceanography. Prog. Oceanogr. 56: 7-136. Werner, F. E., B. O. Blanton, J. A. Quinlan, and R. A. Luettich. 1999. Physical oceanography of the North Carolina continental shelf during the fall and winter seasons: implications for the transport of larval menhaden. Fish. Oceanogr. 8 (suppl. 2):7-21. NOTE Hare and Govoni: Larval fish transport and vertical distributions on the southeast U S continental shelf 735 Appendix 1 Significant pairwise differences between the average larval vertical distributions of 10 species on the southeast United States continental shelf. A total of 187 comparisons were made and the 69 significant differences are listed below. Significant differ- ences between average depth distributions were determined by using a test of independence with the cell values as average proportion of larvae at depth, averaged over a station occupation and weighted by the mean larval concentration at the station. A Bonferroni correction was applied to significance tests within each station occupation. The direction of significant differences (shallower [<], deeper [>], and different [<>]) was determined from the standardized residuals from the test of independence. Year Station Species A Species B 1986 Offshore Brevoortia tyrannus < Peprilus triacanthus 1986 Offshore Brevoortia tyrannus < Paralichthys spp. 1986 Offshore Brevoortia tyrannus < Leiostomus xanthurus 1986 Offshore Brevoortia tyrannus < Etropus spp. 1986 Offshore Brevoortia tyrannus < Etrumeus teres 1986 Offshore Bothus spp. < Peprilus triacanthus 1986 Offshore Bothus spp. < Paralichthys spp. 1986 Offshore Bothus spp. < Etrumeus teres 1986 Offshore Peprilus triacanthus > Leiostomus xanthurus 1986 Offshore Peprilus triacanthus > Micropogonias undulatus 1986 Offshore Peprilus triacanthus <> Etropus spp. 1986 Offshore Peprilus triacanthus <> Etrumeus teres 1986 Offshore Paralichthys spp. > Leiostomus xanthurus 1986 Offshore Paralichthys spp. > Micropogonias undulatus 1986 Offshore Paralichthys spp. <> Etropus spp. 1986 Offshore Paralichthys spp. <> Etrumeus teres 1986 Offshore Leiostomus xanthurus < Etropus spp. 1986 Offshore Leiostomus xanthurus < Etrumeus teres 1986 Inshore Brevoortia tyrannus < Peprilus triacanthus 1986 Inshore Brevoortia tyrannus < Paralichthys spp. 1986 Inshore Brevoortia tyrannus < Leiostomus xanthurus 1986 Inshore Brevoortia tyrannus < Micropogonias undulatus 1986 Inshore Brevoortia tyrannus < Etrumeus teres 1986 Inshore Peprilus triacanthus > Paralichthys spp. 1986 Inshore Paralichthys spp. <> Etropus spp. 1989 Offshore Bothus spp. < Etropus spp. 1989 Inshore Brevoortia tyrannus < Leiostomus xanthurus 1989 Inshore Brevoortia tyrannus < Etropus spp. 1989 Inshore Brevoortia tyrannus < Etrumeus teres 1991 Offshore Brevoortia tyrannus < Bothus spp. 1991 Offshore Brevoortia tyrannus < Peprilus triacanthus 1991 Offshore Brevoortia tyrannus < Paralichthys spp. 1991 Offshore Brevoortia tyrannus < Leiostomus xanthurus 1991 Offshore Brevoortia tyrannus < Micropogonias undulatus 1991 Offshore Brevoortia tyrannus < Etropus spp. 1991 Offshore Brevoortia tyrannus < Etrumeus teres 1991 Offshore Bothus spp. < Leiostomus xanthurus 1991 Offshore Bothus spp. < Micropogonias undulatus 1991 Offshore Bothus spp. < Etropus spp. 1991 Offshore Bothus spp. < Etrumeus teres 1991 Offshore Peprilus triacanthus < Leiostomus xanthurus 1991 Offshore Peprilus triacanthus < Micropogonias undulatus 1991 Offshore Peprilus triacanthus < Etrumeus teres continued 736 Fishery Bulletin 103(4) Appendix 1 (continued) Year Station Species A Species B 1991 Offshore Paralichthys spp. < Leiostomus xanthurus 1991 Offshore Leiostom us xa n th u ru s <> Etropus spp. 1991 Offshore Leiostomu s xa n th uru s <> Etrumeus teres 1991 Offshore Etropus spp. <> Etrumeus teres 1991 Inshore Brevoortia tyrannies < Bothus spp. 1991 Inshore Brevoortia tyrannus < Paralichthys spp. 1991 Inshore Brevoortia tyrannus < Leiostomus xanthurus 1991 Inshore Brevoortia tyrannus < Micropogonias undulatus 1991 Inshore Brevoortia tyrannus < Etropus spp. 1991 Inshore Brevoortia tyrannus < Etrumeus teres 1991 Inshore Bothus spp. < Peprilus triaeanthus 1991 Inshore Bothus spp. < Paralichthys spp. 1991 Inshore Bothus spp. < Leiostomus xanthurus 1991 Inshore Bothus spp. < Micropogonias undulatus 1991 Inshore Bothus spp. < Etropus spp. 1991 Inshore Bothus spp. < Etrumeus teres 1991 Inshore Peprilus triaeanthus <> Etropus spp. 1991 Inshore Xyrichthys novacula < Micropogonias undulatus 1991 Inshore Xyrichthys novacula < Etropus spp. 1991 Inshore Paralichthys spp. < Micropogonias undulatus 1991 Inshore Paralichthys spp. < Etropus spp. 1991 Inshore Leiostomus xanthurus < Micropogonias undulatus 1991 Inshore Leiostomus xanthurus < Etropus spp. 1991 Inshore Micropogonias undulatus > Etropus spp. 1991 Inshore Micropogonias undulatus > Etrumeus teres 1991 Inshore Etropus spp. > Etrumeus teres Acknowledgment of reviewers The editorial staff of Fishery Bulletin would like to acknowledge the scientists who reviewed articles published in 2004-2005. Their contributions have helped ensure the publication of quality science. 737 Dr. David A. Ambrose Dr. Allen H. Andrews Dr. John Arnould Dr. Richard J. Beamish Dr. James L. Bodkin Dr. Richard W. Brill Dr. Jon K.T. Brodziak Dr. Nancy Brown-Peterson Dr. John K. Carlson Dr. Felicia Coleman Dr. Michael Comeau Dr. Bruce H. Comyns Dr. Roy E. Crabtree Mr. Andrew W. David Dr. Michael W. Davis Dr. Tim L.O. Davis Ms. Allison DeLong Dr. Edward E. DeMartini Dr. RJ. Doherty Dr. Michael L. Domeier Dr. Miriam J. Doyle Mr. Nick K. Dulvy Dr. Anne-Marie Eklund Dr. Alan R. Everson Mr. John W. Forsythe Dr. Clive Fox Dr. Robert Foy Mr. Michael Frick Dr. Kevin D. Friedland Dr. Stewart Frusher Dr. Jacques Gagne Dr. Fracisco J. Garcia-Rodriguez Dr. Lance R Garrison Dr. Anthony J. Gharrett Dr. Chris W. Glass Dr. Robert Grabowski Dr. John E. Graves Dr. Lewis J. Haldorson Dr. Anne Hallowed Dr. Jon Hare Mr. Christopher W. Harnden Dr. James X. Hartmann Dr. Andrew J. Harwood Dr. Kelly Hastings Dr. Fabio H.V. Hazin Dr. Thomas E. Helser Dr. Steven W. Hewett Mr. Peter B. Hood Dr. Edward D. Houde Dr. Harriet Huber Dr. George D. Jackson Dr. Stephen C. Jewett Mr. Todd Kassler Dr. John S. Kennelly Dr. Mariano Koen-Alonso Dr. Christopher C. Koenig Dr. Robert G. Kope Dr. Mary Labropoulou Dr. Thomas E. Laidig Ms. Kristen L. Laidre Ms. Kathy L. Lang Dr. Christopher M. Legault Dr. Bruno Leroy Dr. C.J. Limpus Dr. Flavia M. Lucena Mr. Sve-Gunnar Lunneryd Dr. Molly E. Lutcavage Dr. Joanne Lyczkowski-Schultz Dr. Clyde L. Mackenzie Dr. Niels Madsen Dr. Francesc Maynou Dr. John D. McEachron Dr. M.J. Meekan Dr. Richard L. Merrick Dr. Russell B. Millar Dr. Thomas J. Miller Dr. Beatriz Morales-Nin Dr. Debra J. Murie Dr. Michael Musyl Mr. Daniel G. Nichol Dr. David L. Nieland Dr. Victoria M. O'Connell Dr. Jose G. Pajuelo Dr. Donald E. Pearson Dr. Pierre Pepin Dr. R. Ian Perry Dr. John S. Peters Dr. William Peterson Dr. Kenneth W. Pitcher Dr. Dominique Ponton Ms. Jennifer C. Potts Dr. Terrance J. Quinn II Dr. Robert J. Radke Ms. Darlene Ramon Dr. Christian Reiss Dr. William J. Richards Dr. Christopher N. Rooper Dr. Susan E. Safford Dr. Eric Saillant Dr. Kurt M. Schaefer Dr. Richard F. Shaw Dr. Colin Simpfendorfer Dr. G.B. Skomal Dr. Peter J. Smith Ms. Susan E. Smith Dr. Roxanne Smolowitz Mr. John Sneva Dr. Derke Snodgrass Dr. John D. Stevens Dr. Iain M. Suthers Mr. Jesus Tomas Dr. Marc Trudel Dr. Douglas S. Vaughan Dr. Peter Ward Dr. Christopher R. Weidman Dr. Jerry A. Wetherall Dr. Erik Williams Dr. Dave T Wilson Ms. Tonya K. Zepplin Dr. Christian E. Zimmerman 738 Fishery Bulletin 103(4) Fishery Bulletin Index Volume 103(1-4), 2005 List ot titles 103(1) 1 An assessment of scup iStenotomus chrysops) and black sea bass {Centropristis striata) discards in the directed otter trawl fisheries in the Mid-Atlan- tic Bight, by Eleanor A. Bochenek, Eric N. Powell, Allison J. Bonner, and Sarah E. Banta 15 Fecundity of shortspine thornyhead iSebastolobus alascamis) and longspine thornyhead (S. altivelis) (Scorpaenidae) from the northeastern Pacific Ocean, determined by stereological and gravimetric tech- niques, by Daniel W. Cooper, Katherine E. Pearson, and Donald R. Gunderson 23 Relative pleopod length as an indicator of size at sexual maturity in slipper (Scyllarides squammosus ) and spiny Hawaiian (Panulirus marginatus) lob- sters, by Edward E. DeMartini. Marti L. McCracken, Robert B. Moffitt, and Jerry A. Wetherall 34 Seasonal changes in growth of coho salmon (Oncorhynchus kisutch ) off Oregon and Washington and concurrent changes in the spacing of scale cir- culi, by Joseph P. Fisher and William G. Pearcy 52 Escapement of the Cape rock lobster (Jasus lalandii ) through the mesh and entrance of commercial traps, by Johan C. Groeneveld, Jimmy P. Khanyile, and David S. Schoeman 63 Quantification of drag and lift imposed by pop-up satellite archival tags and estimation of the meta- bolic cost to cownose rays (Rhinoptera bonasus), by Donna S. Grusha and Mark R. Patterson 71 Effects of El Nino events on energy demand and egg production of rockfish (Scorpaenidae: Sebastes): a bioenergetics approach, by Chris J. Harvey 84 Application of pop-up satellite archival tag technol- ogy to estimate postrelease survival of white marlin {Tetrapturus albidus) caught on circle and straight shank ("J") hooks in the western North Atlantic rec- reational fishery, by Andrij Z. Horodysky and John E. Graves 108 Cross-shelf and seasonal variation in larval fish assemblages on the southeast United States con- tinental shelf off the coast of Georgia, by Katrin E. Marancik, Lisa M. Clough, and Jonathan A. Hare 130 Year-class formation in Pacific herring (Clupea pal- lasi) estimated from spawning-date distributions of juveniles in San Francisco Bay, California, by Michael R. O'Farrell and Ralph J. Larson 142 Diet of oceanic loggerhead sea turtles (Caretta caretta) in the central North Pacific, by Denise M. Parker, William J. Cooke, and George H. Balazs 153 Indirect validation of the age-reading method for Pacific cod (Gadus macrocephalus) using otoliths from marked and recaptured fish, by Nancy E. Rob- erson, Daniel K. Kimura, Donald R. Gunderson, and Allen M. Shimada 161 Age and growth estimates of the thorny skate {Amblyraja radiata) in the western Gulf of Maine, by James A. Sulikowski, Jeff Rneebone, Scott Elzey, Joe Jurek, Patrick D. Danley, W. Huntting Howell, and Paul C. W Tsang 169 Age-validation, growth modeling, and mortality estimates for striped trumpeter (Latris lineata ) from southeastern Australia: making the most of patchy data, by Sean R. Tracey and Jeremy M. Lyle 183 Larval development of estuary perch (Macquaria colonorum ) and Australian bass (M. novemaculeata ) (Perciformes: Percichthyidae) and comments on their life history, by Thomas Trnksi, Amanda C. Hay, and D. Stewart Fielder 195 Early life history of the Argentine sandperch Pseu- dopercis semifasciata (Pinguipedidae) off northern Patagonia, by Leonardo A. Venerus, Laura Machi- nandiarena. Martin D. Ehrlich, and Ana M. Parma 207 Geographic variation among age-0 walleye pollock (Theragra chalcogramma): evidence of mesoscale variation in nursery quality?, by Matthew T. Wilson, Annette L. Brown, and Kathryn L. Mier 219 Tagging studies on the jumbo squid (Dosidicus gigas) in the Gulf of California, Mexico, by Unai Markaida, Joshua J. C. Rosenthal, and William F Gilly 103(2) 97 Age validation of quillback [Sebastes maliger) using bomb radiocarbon, by Lisa A. Kerr, Allen H. Andrews, Kristen Munk, Kenneth H. Coale, Brian R. Frantz, Gregor M. Cailliet, and Thomas A. Brown 229 Sex change rules, stock dynamics, and the perfor- mance of spawning-per-recruit measure in pro- togynous stocks, by Suzanne H. Alonzo and Marc Mangel List of titles 739 246 Neonatal growth of Steller sea lion (Eumetopias jubatus) pups in Alaska, by Elisif A. A. Brandon, Donald G. Calkins, Thomas R. Loughlin, and Ran- dall W. Davis 380 Maximum likelihood estimation of mortality and growth with individual variability from multiple length-frequency data, by You-Gan Wang and Nick Ellis 258 Reproductive biology of carpenter seabream (Argy- rozona argyrozona) (Pisces: Sparidae) in a marine protected area, by Stephen L. Brouwer and Marc H. Griffiths 270 Decline in sea otter (Enhydra lutris) populations along the Alaska Peninsula, 1986-2001, by Douglas M. Burn and Angela M. Doroff 280 Growth dynamics of the spinner shark (Carcharhi- nus brevipinna I off the United States southeast and Gulf of Mexico coasts: a comparison of methods, by John K. Carlson and Ivy E. Baremore 292 Tracking Pacific bluefin tuna (Thunnus thynnus orientalis) in the northeastern Pacific with an auto- mated algorithm that estimates latitude by match- ing sea-surface-temperature data from satellites with temperature data from tags on fish, by Michael L. Domeier, Dale Kiefer, Nicole Nasby-Lucas, Adam Wagschal, and Frank O'Brien 307 Age, growth, mortality, and radiometric age vali- dation of gray snapper (Lutjanus griseus) from Louisiana, by Andrew J. Fischer, M. Scott Baker Jr., Charles A. Wilson, and David L. Nieland 320 Estimating exploitable stock biomass for the Maine green sea urchin (Strongyloeentrotus droebachien- sis) fishery using a spatial statistics approach, by Robert C. Grabowski, Thomas Windholz, and Yong Chen 331 Abundance and distribution of California sea lions (Zalophus californianus) in central and northern California during 1998 and summer 1999, by Mark S. Lowry and Karin A. Forney 344 Variability in spawning frequency and reproductive development of the narrow-barred Spanish mackerel (Scomberomorus commerson) along the west coast of Australia, by Michael C. Mackie, Paul D. Lewis, Daniel J. Gaughan, and Stephen J. Newman 355 Seasonal marine growth of Bristol Bay sockeye salmon (Oncorhyncus nerka) in relation to competi- tion with Asian pink salmon (O. gorbuscha ) and the 1977 ocean regime shift, by Gregory T Ruggerone, Ed Farley, Jennifer Nielsen, and Peter Hagen 392 Effects of fishing on growth traits: a simulation analysis, by Erik H. Williams and Kyle W Shertzer 404 Preliminary evidence of increased spawning aggre- gations of mutton snapper (Lutjanus analis) at Riley's Hump two years after establishment of the Tortugas South Ecological Reserve, by Michael L. Burton, Kenneth J. Brennan, Roldan C. Munoz, and Richard O. Parker Jr. 411 Feeding habits of European hake (Merluccius merluccius) in the central Mediterranean Sea, by Paolo Carpentieri, Francesco Colloca, Massimiliano Cardinale, Andrea Belluscio, and Giandomenico D. Ardizzone 417 Biology of queen snapper (Etelis oculatus: Lutjani- dae) in the Caribbean, by Bertrand Gobert, Alain Guillou, Peter Murray, Patrick Berthou, Maria D. Oqueli Turcios, Ester Lopez, Pascal Lorance, Jerome Huet, Nicolas Diaz, and Paul Gervain 426 Courtship and spawning behaviors of carangid spe- cies in Belize, by Rachel T. Graham and Daniel W. Castellanos 433 Comparison of two approaches for estimating natu- ral mortality based on longevity, by David A. Hewitt and John M. Hoenig 438 Effects of current speed and turbidity on stationary light-trap catches of larval and juvenile fishes, by David C. Lindquist and Richard F. Shaw 445 Can a change in the spawning pattern of Argentine hake {Merluccius hubbsi) affect its recruitment?, by Gustavo J. Macchi, Marcelo Pajaro, and Adrian Madirolas 453 Feeding habits of the dwarf weakfish (Cynoscion nannus) off the coasts of Jalisco and Colima, Mexico, by Alma R. Raymundo-Huizar, Horacio Perez-Espana. Maite Mascaro, and Xavier Chiappa-Carrara 461 Using bone measurements to estimate the original sizes of bluefish (Pomatomus saltatrix) from digested remains, by Anthony D. Wood 103(3) 371 Distribution, feeding condition, and growth of Japa- nese Spanish mackerel (Scomberomorus niphonius) larvae in the Seto Inland Sea, by Jun Shoji and Masaru Tanaka 469 Using poststratification to improve abundance esti- mates from multispecies surveys: a study of juve- nile flatfishes, by Sherri C. Dressel and Brenda L. Norcross 740 Fishery Bulletin 103(4) 489 Length at maturity in three pelagic sharks (Lamna jiasus, Isurus oxyrinchus, and Prionace glauea ) from New Zealand, by Malcolm P. Francis and Clinton Duffy 601 Sexual differentiation and gonad development in striped mullet (Mugil cephalus L. ) from South Carolina estuaries, by Christopher J. McDonough, William A. Roumillat, and Charles A. Wenner 501 Survey- and fishery-derived estimates of Pacific cod 620 (Gadus macrocephalus) biomass): implications for strategies to reduce interactions between groundfish fisheries and Steller sea lions (Eumetopias jubatus), by Lowell W. Fritz and Eric S. Brown 516 Mitochondrial gene sequences useful for species identification of western North Atlantic Ocean 635 sharks, by Thomas W. Greig, M. Katherine Moore, Cheryl M. Woodley, and Joseph M. Quattro 524 Genetic variation of rougheye rockfish (Sebastes aleutianus) and shortraker rockfish (S. borealis) inferred from allozymes, by Sharon L. Hawkins, Jonathan Heifetz, Christine M. Kondzela, John E. Pohl, Richard L. Wilmot, Oleg N. Katugin, and Vladimir N. Tuponogov 536 The reproductive cycle of the thorny skate (Ambly- raja radiata ) in the western Gulf of Maine, by James A. Sulikowski, Jeff Kneebone, Scott Elzey, Joe Jurek, Patrick D. Danley, W. Huntting Howell, and Paul C. W. Tsang 544 Effect of type of otolith and preparation technique on age estimation of larval and juvenile spot (Leiosto- mus xanthurus), by Dariusz P. Fey, Gretchen E. Bath Martin, James A. Morris, and Jonathan A. Hare 553 Preliminary use of oxygen stable isotopes and the 1983 EI Nino to assess the accuracy of aging black rockfish {Sebastes melanops), by Kevin R. Piner, Melissa A. Haltuch, and John R. Wallace 103(4) 561 Patterns of growth, mortality, and size of the tropi- cal damselfish Acanthochromis polyacanthus across the continental shelf of the Great Barrier Reef, by Michael J. Kingsford and Julian M. Hughes 574 Variation in the distribution of walleye pollock (Theragra chalcogramma) with temperature and implications for seasonal migration, by Stan Kot- wicki, Troy W. Buckley, Taina Honkalehto, and Gary Walters 588 Toward identification of larval sailfish (Istio- phorus platypterus), white marlin (Tetrapturus albidus), and blue marlin (Makaira nigricans) in the western North Atlantic Ocean, by Stacy A. Luthy, Robert K. Cowen, Joseph E. Serafy, and Jan R. McDowell Incidental catch and estimated discards of pelagic sharks from the swordfish and tuna fisheries in the Mediterranean Sea, by Persefoni Megalofonou, Con- stantinos Yannopoulos, Dimitrios Damalas, Gregorio De Metrio, Michel Deflorio, Jose M. de la Serna, and David Macias Reproductive biology of female Rikuzen sole (Dex- istes rikuzenius ), by Yoji Narimatsu, Daiji Kitagawa, Tsutomu Hattori, and Hirobumi Onodera 648 Temporal and spatial distribution and abundance of flathead sole (Hippoglossoides elassodon ) eggs and larvae in the western Gulf of Alaska, by Steven M. Porter 659 Movements and spawning of white marlin (Tetraptu- rus albidus) and blue marlin (Makaira nigricans) off Punta Cana, Dominican Republic, by Eric D. Prince, Robert K. Cowen, Eric S. Orbesen, Stacy A. Luthy, Joel K, Llopiz, David E, Richardson, and Joseph E. Serafy 670 Life history characteristics for silvergray rockfish (Sebastes brevispinis) in British Columbia waters and the implications for stock assessment and management, by Richard D. Stanley and Allen R. Kronlund 685 Impact of the California sea lion (Zalophus califor- nianus) on salmon fisheries in Monterey Bay, Cali- fornia, by Michael J. Weise and James T. Harvey 697 Estimates of growth and comparisons of growth rates determined from length- and age-based models for populations of purple wrasse (Notolabrus fuci- cola ). by Dirk C. Welsford and Jeremy M. Lyle 712 Effects of harvesting methods on sustainability of a bay scallop fishery: dredging uproots seagrass and displaces recruits, by Melanie J. Bishop, Charles H. Peterson, Henry C. Summerson, and David Gaskill 720 Longline-caught blue shark (Prionace glauea): fac- tors affecting the numbers available for live release, by Guillermo A. Diaz and Joseph E. Serafy 725 Length correction for larval and early-juvenile Atlantic menhaden (Brevoortia tyrannus) after pres- ervation in alcohol, by Dariusz P. Fey and Jonathan A. Hare 728 Comparison of average larval fish vertical distribu- tions among species exhibiting different transport pathways on the southeast United States continen- tal shelf, by Jonathan A. Hare and John J. Govoni 741 Fishery Bulletin Index Volume 103(1-4), 2005 List ot authors Alonzo, Suzanne H. 229 Andrews, Allen H. 97 Ardizzone. Giandomenico D. 411 Baker Jr., M. Scott 307 Balazs, George H. 142 Banta, Sarah E. 1 Baremore, Ivy E. 280 Bath Martin, Gretchen E. 544 Belluscio, Andrea 411 Berthou, Patrick 417 Bishop, Melanie J. 712 Bochenek, Eleanor A. 1 Bonner. Allison J. 1 Brandon, Elisif A. A. 246 Brennan, Kenneth J. 404 Brouwer, Stephen L. 258 Brown, Annette L. 207 Brown, Eric S. 501 Brown, Thomas A. 97 Buckley, Troy W. 574 Burn, Douglas M. 270 Burton, Michael L. 404 Cailliet, Gregor M. 97 Calkins, Donald G. 246 Cardinale, Massimiliano 411 Carlson. John K. 280 Carpentieri, Paulo 411 Castellanos, Daniel W. 426 Chen.Yong 320 Chiappa-Carrara, Xavier 453 Clough, Lisa M. 108 Coale, Kenneth H. 97 Colloca, Francesco 411 Cooke, William J. 142 Cooper, Daniel W. 15 Cowen, Robert K. 588,659 Damalas, Dimitrios 620 Danley, Patrick D. 161. 536 Davis, Randall W. 246 De la Serna, Jose M. 620 De Metrio, Gregorio 620 Deflorio, Michel 620 DeMartini, Edward E. 23 Diaz, Guillermo A. 720 Diaz, Nicolas 417 Domeier, Michael L. 292 Doroff, Angela M. 270 Dressel, Sherri C. 469 Duffy, Clinton 489 Ehrlich, Martin D. 195 Ellis, Nick 380 Elzey, Scott 161, 536 Farley, Ed 355 Fey, Dariusz P. 544, 725 Fielder, D. Stewart 183 Fischer. Andrew J. 307 Fisher, Joseph P. 34 Forney, Karin A. 331 Francis, Malcolm P. 489 Frantz, Brian R. 97 Fritz. Lowell W. 501 Gaskill, David 712 Gaughan, Daniel J. 344 Gervain, Paul 417 Gilly, William F. 219 Gobert, Bertrand 417 Govoni, John J. 728 Grabowski, Robert C. 320 Graham, Rachel T 426 Graves, John E. 84 Grieg, Thomas W 516 Griffiths, Marc H. 258 Groeneveld, Johan C. 52 Grusha, Donna S. 63 Guillou, Alain 417 Gunderson, Donald R. 15, 153 Hagen, Peter 355 Haltuch, Melissa A. 553 Hare, Jonathan A. 108, 544, 725, 728 Harvey, Chris J. 71 Harvey, James T. 685 Hattori, Tsutomu 635 Hawkins, Sharon L. 524 Hay, Amanda C. 183 Heifetz, Jonathan 524 Hewitt, David A. 433 Hoenig, John M. 433 Honkalehto, Taina 574 Horodysky, Andrij Z. 84 Howell, W. Huntting 161, 536 Huet, Jerome 417 Hughes, Julian M. 561 Jurek, Joe 161,536 Katugin, Oleg N. 524 Kerr, Lisa A. 97 Khanyile, Jimmy P. 52 Kiefer, Dale 292 Kimura, Daniel K. 153 Kingsford, Michael J. 561 Kitagawa, Daiji 635 Kneebone, Jeff 161,536 Kondzela, Christine M. 524 Kotwicki, Stan 574 Kronlund, Allen R. 670 Larson, Ralph J. 130 Lewis, Paul D. 344 Lindquist, David C. 438 Llopiz, Joel K. 659 Lopez, Ester 417 Lorance, Pascal 417 Loughlin, Thomas R. 246 Lowry, Mark S. 331 Luthy, Stacy A. 588,659 Lyle, Jeremy M. 169, 697 Macchi, Gustavo J. 445 Machinandiarena, Laura 195 Macias, David 620 Mackie, Michael C. 344 Madirolas, Adrian 445 Mangel, Marc 229 Marancik, Katrin E. 108 Markaida, Unai 219 Mascaro, Maite 453 McCracken, Marti L. 23 McDonough, Christopher J. 601 McDowell, Jan R. 588 Megalofonou, Persefoni 620 Mier, Kathryn L. 207 Moffitt, Robert B. 23 Moore, M. Katherine 516 Morris, James A. 544 Munk, Kristen 97 Munoz, Roldan C. 404 Murray, Peter 417 Narimatsu, Yoji 635 Nasby-Lucas, Nicole 292 Newman, Stephen J. 344 Nieland, David L. 307 Nielsen, Jennifer 355 Norcross, Brenda L. 469 O'Brien, Frank 292 OTarrell, Michael R. 130 Onodera, Hirobumi 635 Oqueli Turcios, Maria D. 417 Orbesen, Eric S. 659 Pajaro, Marcelo 445 Parker, Denise M. 142 Parker Jr., Richard O. 404 Parma, Ana M. 195 Patterson, Mark R. 63 Pearcy, William G. 34 742 Fishery Bulletin 103(4) Pearson. Katherine E. 15 Perez-Espana. Horacio 453 Peterson, Charles H. 712 Piner, Kevin R. 553 Pohl, JohnE. 524 Porter, Steven M. 648 Powell, Eric N. 1 Prince, Eric D. 659 Quattro, Joseph M. 516 Raymundo-Huizar, Alma R. Richardson, David E. 659 Roberson, Nancy E. 153 Rosenthal. Joshua J. C. 219 Roumillat, William A. 601 Ruggerone, Gregory T. 355 453 Schoeman, David S. 52 Serafy. Joseph E. 588, 659. 720 Shaw, Richard F. 438 Shertzer, Kyle W. 392 Shimada, Allen M. 153 Shoji.Jun 371 Stanley, Richard D. 670 Sulikowski. James A. 161, 536 Summerson, Henry C. 712 Tanaka, Masaru 371 Tracey, Sean R. 169 Trnski, Thomas 183 Tsang, Paul C. W. 161, 536 Tuponogov, Vladimir N. 524 Venerus, Leonardo A. 195 Wagschal. Adam 292 Wallace, John R. 553 Walters. Gary 574 Wang,You-Gan 380 Weise, Michael J. 685 Welsford, Dirk C. 697 Wenner, Charles A. 601 Wetherall, Jerry A. 23 Williams Erik H. 392 Wilmot, Richard L. 524 Wilson. Charles A. 307 Wilson, Matthew T. 207 Windholz, Thomas 320 Wood, Anthony D. 461 Woodley, Cheryl M. 516 Yannopoulos, Constantinos 620 743 Fishery Bulletin Index Volume 103(1-4), 2005 List ot subjects Abundance Argentine hake 445 sandperch 195 California sea lion 331 flatheadsole 648 sockeye salmon 355 Acan thoch romis polyacan th us 561 Acoustic survey 445 Aerial survey 270, 331 Age and growth damselfish 561 gray snapper 307 shark, spinner 280 silvergray rockfish 670 striped trumpeter 169 thorny skate 161 at maturity 635 determination Rikuzen sole 635 striped mullet 601 estimates accuracy 544 precision 544 validation damselfish 561 gray snapper 307 Pacific cod 153 rockfish black 553 quillback 97 spot 544 striped trumpeter 169 Age-0 207 Aggregation 404 Alaska 97, 207, 247, 270, 355, 469, 501, 524, 553, 574, 648 Alaska Peninsula 270,648 Albacore 620 Aleutian Islands 246, 501, 524 Allozymes 524 Alopias vulpinus 620 Amblyraja radiata 161, 536 ANOVA' 685,712,725 Archival tag 292 Argentina 195 Argopecten irradians 712 Argyrozona argyrozona 258 Atlantic Ocean 516. 536, 659 southwest 445 northwest 161, 280, 553, 588, 720 western 404, 417 Australia 169, 183, 344, 561 Automated algorithm 292 Back-calculation 130. 153, 461 Bahamas 588 Band count, vertebral section 161 Bass Australian 183 black sea 1 Batch fecundity 258 Batch spawner 15 Beaufort Inlet 725 Belize 426 Bering Sea 153, 501, 524, 574 Bioenergetics model 71 Biomass Pacific cod 501 seagrass 712 sea urchin 320 walleye pollock 574 Bluefish 461 Body condition 635 Bogue Sound 712 Bone measurements 461 Bottom trawl fishery 670 nets 574 Brevoortia tyrannus 725, 728 Bristol Bay 355 British Columbia 670 Bycatch mortality 574 Callinectes sapidus 433 California 130, 331, 553, 685 Canonical correspondence analysis (CCA) 108 Canonical variates analysis (CVA) 588 Cape rock lobster 52 Carangidae 426 Carapace base 52 Carcharhinus b?-evipinna 280 Caretta caretta 142 Carinaria cithara 142 Caribbean 420,516 Catch efficiency 438 Catch per unit of effort 438, 469, 501, 620, 670, 685 Central California Valley Index (CVI) 685 Centropristis striata 1 Chesapeake Bay 720 ChiniakBay 469 Chi-square test 620 Chondrophore 142 Circulus spacing 34, Cirripedia 142 Clupea pallasi 130 Cod, Pacific 153, 501 Codends 1 Colima coast 453 Commercial harvest 712 Commercial passenger fishing vessel 685 Commercial traps 52 Commercial troll fishery 685 Conductivity-temperature-depth probe 108 Copepod parasite 670 Coral reef 561 Courtship behavior 426 CPUE 438, 469, 501, 620, 670, 685 Crabs, blue 433 Croaker Atlantic 728 Cross-shelf transport 728 variation 108, 561 Current 438 Cynoscion nannus 453 Damselfish 561 Decapoda 142 Deep snapper resources 417 Demographic assessment 561 Depredation 685 Dexistes tikuzenius 635 Diet, loggerhead turtle 142 Discard 620 mortality 1,720 to-landings ratio 1 Displacement 659 Distribution and abundance Argentine sandperch 195 vertical larval 728 walleye pollock 574 DNA 516, 588 Dosidicus gigas 219 Drag and lift 63 Dredging 712 Egg geographic distribution and abundance 648 mortality 130 production 445 Elasmobranch 536 El Nino 71, 553 Southern Oscillation 685 Endangered Species Act 270 Energetic cost 63 Energy consumption 71 Enhydra lutris 270 744 Fishery Bulletin 103(4) Escapement, from lobster trap 52 Essential fish habitat 659 Estuarine bivalve fisheries 712 Estuarine-dependent species 728 Etelis oculatus 417 Eumetopias jubatus 246, 501 External body metric 23 Fecundity gravimetric estimates 15 Rikuzen sole 635 silvergray rockfish 670 stereological estimates 15 thornyhead 15 thorny skate 536 Feeding habits 411,453 First increment formation 544 Fisheries management 1, 229, 380, 392, 469, 501 Fishery biology 417 Fishery interaction 685 Fishing gear 620,712 Fishing mortality 720 Flatfishes 469 Flounder gulf 728 southern 728 summer 728 Flow, vertically structured 728 G-statistic 728 Gadus macrocephalus 153, 501 Galeorh in us galeus 620 Gametogenesis 601 Gas platforms 438 Gastropoda 142 Gene sequences 516 Genetic identification 516,588 Genetic variation 524 Geographic distribution 648 variation 207 Geolocation 292 Georgia 108 Gompertz model 280 Gonadal maturation 635 Gonad development 601 Gonadosomatic index 536, 635 Grand Banks 720 Gravimetric technique 15 Grapsidae 142 Gray's Reef National Marine Sanctuary 108 Great Barrier Reef 561 Groundfish 469, 501 Growth 380,392 damselfish 561 dimorphism 670 effects of fishing on 392 Pacific herring 130 rate, daily 219 salmon 34 scale 355 seasonal variation 34 Steller sea lion 246 striped trumpeter 169 thorny skate 161 Gulf of Alaska 246, 524, 648 of California 219 of Maine 161,536 of Mexico 280,438,516 Habitat destruction 712 flatfish 469 Hake Argentine 445 European 411 Halibut, Pacific 469 Harvesting, effects of 712 Hepatosomatic index 536 Herring, Pacific 130 Heteropoda 142 Hippoglossoides elassodon 469, 648 Hippoglossus stenolepis 469 Hook type mortality estimates 91 Horizontal transport 728 Hydrodia 142 Ichthyoplankton 108, 195, 371, 648 Identification 516, 588 Incidental catch 620 Increment formation 544 Indirect validation 153 Individual-based model 392 Individual variability 380 Interannual variability 469 Inverse distance-weighted 574 Isochronal spawning fish 610 Istiophoridae 588 Istiophorus platypterus 588 Isurus oxyrinchus 489, 620 Jalisco coast 453 Janthina spp. 142 Japan 373, 635 Jasus lalandii 52 Juvenile Argentine sandperch 195 effects of turbidity on 438 flatfish 469 Pacific herring 130 salmon 34 scallops 712 spot 544 walleye pollock 207 Kamchatka coast 524 Kruskall-Wallis test 620 Kodiak Island 207,648 Lamna nasus 489 Larval fish abundance 195. 648 age estimation 544 assemblages 108 Atlantic menhaden 725 billfish 588 cross-shelf variation 108 development 183, 195, 207 diet 207 distribution 371, 648 effects of turbidity 438 feeding conditions 371 flathead sole 648 geographic variation 207 growth 207,371 mortality 130 seasonal variation 108 survival 130 transport 728 Latris lineata 169 Leiostomus xanthurus 5 44, 728 Length at maturity 489 Length correction 720 Length frequency 380 Lepas spp. 142 Lepidopsetta spp. 469 Life history 183, 195, 229, 536, 670, 392 Light traps 438 Linear regression analysis 725 Linear regression model 433, 588 Lobster Hawaiian spiny 23 slipper 23 South African Cape rock 52 Logistic 280 Longevity 433 Longline 720 Louisiana 307 Lowrie Island rookery 246 Lunar periodicity 426 Lutjanidae 420 Lutjanus analis 404 griseus 307 Mackerel Japanese Spanish 371 narrow-barred Spanish 344 Macquaria colonorum 183 novemaeuleata 183 Makaira nigricans 588, 659 Maine 320 Marginal increment 161 List of subjects 745 Marlin blue 588, 659 white 84, 588, 659 Marine mammal 331 Marine protected areas 258 Maturity lobster 23 pelagic sharks 489 Rikuzen sole 635 silvergray rockfish 670 striped mullet 601 Maximum age 433 likelihood 380 sustainable yield 659 Mediterranean Sea 411, 620 Menhaden, Atlantic 725, 728 Merluccius hubbsi 445 merluccius 411 Mesh size 52 Mesoscale variation 207 Metabolic cost estimation 63 Mexico 219,453 Micropogonias undulatus 728 Mid-Atlantic Bight 1 Migration jumbo squid 219 walleye pollock 574 Mitochondrial DNA 516 Models Bayesian 524 bioenergetic 71 generalized additive model (GAM) 670 general linear model (GLM) 670 growth 169,380,392,670 linear regression 433, 489 Leslie 501 Levenburg-Marquardt 601 mortality 380,433,489 Schnute growth model 670 von Bertalanffy 380,392 Monte Carlo simulation 588 Monterey Bay 685 Morphological-based maturity 23, 601 Morphometries 588 Mortality 380,433 blue shark 720 damselfish 561 gray snapper 307 hook type 84 natural, estimation of 422 release 720 sea turtle 142 striped trumpeter 169 Moss Landing 685 Movement patterns 659 vertical 728 Mugil cephalus 601 Multivariate analysis 108 Natural mortality 433 Neonatal growth 246 Neustonic species 142 New Hampshire 536 New Zealand 489 Nonparametric analysis of variance 620 North Atlantic, western 84 North Carolina 712, 725 North Pacific, central 142 Notolab/'its fucicola 697 Nursery quality 207 Ocean regime shift 355 Oil platforms 438 Oncorhynchus gorbuscha 355 kisutch 34 nerka 355 tshawytscha 685 Ontogenesis 411 Oocyte maturation 635 Oogenesis 601,635 Opportunistic feeders 142 Oregon 34 Original prey size 461 Otariidae 246 Otolith 97, 130, 153, 169, 307, 373, 544, 553, 561, 601, 635, 670 Otolith microchemistry 553 Ovarian atresia 601 Ovary 1, 23, 536 Oxygen isotope 55 Pacific Ocean eastern 71, 453, 685 north 355,635 northeastern 15, 130, 292, 331, 648 Panulirus marginatus 23 Paralichthys albiguta 728 dentatus 728 lethostigma 728 Parasites 524 Patagonia 195 Patagonian stock 445 Pelagic Observers Program, U.S.Atlantic 720 Penaeus semisulcatus 380 Perch, estuary 183 Percichthyidae 183 Permit 426 Phenotypic plasticity 229 Phylogenetics 516 Pigmentation patterns 588 Pinguipedidae 195 Pinniped 331,685 Pleopod measurement 23 Pleuronectes asper 469 Pollock, walleye 207, 574 Pomacentridae 561 Pomotomus saltatrix 461 Population decline 270 dynamics 229 Pop-up satellite archival tags 63, 84, 292, 659 Postrelease survival 84 Postspawning morphology 601 Poststratification 469 Potential energetic costs 63 Prawn, tiger 380 Preservation shrinkage 725 Prey size 461 Prionace glauca 489, 620, 720 Protogynous sex change 229 Pseudopercis semifasciata 195 Punta Cana 659 Pup, sea lion 246 Pyrosomas 142 Radiocarbon 97, 307 Rajidae 536 Ray, cownose 63 Recreational fishery 84 salmon 685 Recruitment Argentine hake 445 gray snapper 307 silvergray rockfish 670 Reef fish 426,369 Reef promontory 426 Regression analysis 34, 142 Release mortality 720 Reproductive development 344, 601 Reproductive maturity 670 Reproduction carpenter seabream 258 marlin 659 pelagic shark 489 Rikuzen sole 635 Spanish mackerel 344 thorny skate 536 striped mullet 601 Restriction fragment length polymorphism analysis 588 Rhinoptera bonasus 63 Riley's Hump 404 Rockfish age validation 97 black 553 bioenergetics model 71 quillback 97 rougheye 524 shortraker 524 silvergray 670 trophic ecology 71 yelloweye 97 746 Fishery Bulletin 103(4) Rookeries 246 Russia 524 Sailfish 588 Salmon Asian pink 355 coho 34 chinook 685 sockeye 355 Sandperch, Argentine 195 San Francisco Bay 130 Sarcotaces arcticus 670 Santa Cruz 685 Scale circuli 34, 355, Scallops, bay 712 Scup 1 Sciaenidae 453 Scomberomorus commerson 344 niphonius 371 Scombridae 371 Scorpaenidae 15, 71 Scyllarides squammosus 23 Sea bass, black 1 Seabream, carpenter 258 Seagrass 712 Sea lion California 331,685 Steller 246, 501 SeaofHiuchi 373 Sea otter, northern 270 Sea-surface temperature 292 Seasonal growth 34, 179, 355 Seasonal migration 219 Seasonal variation 108 Sea turtles loggerhead 142 Sea urchin, Maine green 320 Sebastes spp. 71 aleutianus 524 borealis 524 brevispinis 670 maliger 97 melanops 553 mystinus 74 ruberrimus 97 Sebastolobus alascanus 15 altivelis 15 Selection differentials 392 Selectivity curves 52 Senescence 635 Semidemersal gadid 207 Sequence 516 Seto Inland Sea 371 Sexual differentiation 601 Sexual dimorphism 169, 635 Sharks 516 blue 489,620 574 coastal 280 common thresher 620 pelagic 489, 620 porbeagle 489 shortfin mako 489, 620 spinner 280 tope 620 Size at maturity 258, 601 Skate, thorny 161,536 Snapper gray 307 mutton 404 queen 417 Sole flathead 469 Rikuzen 635 rock 469 yellowfin 469 South Africa 52,258 South America 183 South Carolina 601 Southeast United States continental shelf 108, 728 Sparidae 258 Spatial analysis 320 Spatial distribution 574, 648 Spatial variability 320 Spawning aggregations 404, 426 behavior 426 date distribution 130 flathead sole 648 frequency 258, 344 habitat 659 marlin 659 mutton snapper 404 Pacific herring 130 pattern change 445 per-recruit 229 season 258 Spawning stock biomass per recruit analysis 670 Species identification 516 Spermatogenesis 536 Spot 544,728 Squid, jumbo 219 Starch gel electrophoresis 524 Stenotomus chrysops 1 Stereological techniques 15 Stock assessment 320, 433, 670 dynamics 229 management 670 Straits of Florida 588 Strip transect survey 270, 331 Strongylocen trotus droeboch iensis 320 Student-Newman-Keul test 685, 712 Submerged aquatic vegetation 712 Subtropical front 142 Survival rate 620 Swordfish 620 Tagging jumbo squid 219 pop-up satellite 63, 84, 292, 659 Temperature 561, 574 Temporal distribution 648 Tetrapturus albidus 84, 588, 648 Theragra chalcogramma 207, 574 Thornyhead shortspine 15 longspine 15 Thunnus alalunga 620 thy turns 620 thynnus orientcdis 292 Tortugas South Ecological Reserve 404 Trachinotus falcatus 426 Trap selectivity 52 Trawl survey 445, 501, 574 echo integration 574 groundfish 469 Triangulated Irregular Networks 320 Trophic breadth variation 453 Trumpeter, striped 169 Tsitsikamma National Park 258 TukeyHSDtest 725 Tuna bluefin 620 Pacific bluefin 292 Turbidity 438 U.S. Atlantic Pelagic Observers Program 720 Variation genetic 524 spawning frequency 344 Velella velella 142 Vertebral band analysis 161 Vertical distribution 728 von Bertalanffy 380 damselfish 561 gray snapper 307 Pacific cod 153 spinner shark 280 striped trumpeter 169 thorny skate 161 Washington 34, 524 Water column 728 Wax histology technique 601 Weakfish, dwarf 453 Wrasse, purple 697 Xiphias gladius 620 Year-class strength 130 Zalophus californianus 331, 685 Zoatera marina 712 Fishery Bulletin 103(4) 747 Superintendent of Documents Publications Order Form *5178 I I YES, please send me the following publications: Subscriptions to Fishery Bulletin for $55.00 per year ($68.75 foreign) The total cost of my order is $ . Prices include regular domestic postage and handling and are subject to change. (Company or Personal Name) (Please type or print) (Additional address/attention line! ( Street address ) (City, State, ZIP Code) (Daytime phone including area code) ( Purchase Order No. ) Please Choose Method of Payment: ~] Check Payable to the Superintendent of Documents □ GPO Deposit Account I I I I I I [ I — I I ] VISA or MasterCard Account To fax voui- orders (202) 512-2250 (Credit card expiration date) ( Authorizing Signature ) Mail To: Superintendent of Documents P.O. Box 371954, Pittsburgh, PA 15250-7954 Thank you for your order! This statement is required by the Act of August 12, 1970, Section 3685. Title 39. U.S. Code, showing ownership, management, and circulation of the Fishery Bulletin, publication number 366-370, and was filed on 2 September. 2005. The Bulletin is published quarterly (four issues annually) with an annual subscription price of $55.00 ( sold by the Superintendent of Documents, U.S. Gov- ernment Printing Office. Washington, DC 20402). The complete mailing address of the office of publica- tion is NMFS Scientific Publications Office, NOAA, 7600 Sand Point Way NE, BIN C 15700, Seattle, WA 98115. The complete mailing address of the head- quarters of the publishing agency is National Marine Fisheries Service, NOAA, Department of Commerce, 1335 East- West Highway, Silver Spring. MD 20910. The name of the publisher is Willis Hobart and the managing editor is Sharyn Matriotti; their mailing address is: NMFS Scientific Publications Office, 7600 Sand Point Way NE, BIN 15700, Seattle. WA98115. The owner is the U.S. Department of Commerce, 14th St. N.W, Washington, DC 20230; there are no bondholders, mortgages, or other security hold- ers. The purpose, function, and nonprofit status of the organization (agency) and the exempt status for Federal income tax purposes has not changed during the preceding 12 months. The extent and nature of circulation is as follows: total number of copies (A) (average number of copies of each issue during the preceding 12 months) was 1337 and the actual number of copies of the single issue published nearest to the filing dates was 1240. Paid circulation (B) is handled by the U.S. Govern- ment Printing Office, Washington, DC 20402, and the total number printed for sales (mail subscrip- tions and individual sales) was 444 for the average number of copies each issue during the preceding 12 months and 350 the actual number of copies of the single issue published nearest to the filing date (C). Free distribution (Dtby mail; samples, compli- mentary and other free copies (average number of copies each issue during the preceding 12 months) was 843 and the actual number of copies of the single issue published nearest to the filing date was 840. Free distribution outside the mail by carriers or other means was 0 for both average number of copies and actual number of copies. Total free distribution (F) was 0 for both average number of copies and actual number of copies of the single issue published nearest the filing date. The total distribution (G: sum of D and B) (average number of copies each issue during the preceding 12 months) was 1287 and the actual number of copies of the single issue published nearest to the filing date was 1190. There were 50 copies (avg. annual) not distributed (H). The total (I: sum of G and H ) is equal to the net press run figures shown in Item A: 1337 and 1240 copies, respectively. I certify that the statements made by me above are correct and complete: (Signed) Willis Hobart, Publisher. MBL M'HOI LIBRARY wh nxi o Fishery Bulletin Guidelines for authors Content of manuscripts Contributions published in Fishery Bulletin de- scribe original research in marine fishery sci- ence, fishery engineering and economics, as well as the areas of marine environmental and ecolog- ical sciences (including modeling). Although all conj nbutions are subject to peer review, respon- sibility for the contents of papers rests upon the authors and not upon the editor or publisher. Submission of an article implies that the article is original and is nof being considered for publi- cation elsewhere. Manuscripts should be written in English. Authors whose native language is not English are strongly advised to have their man- uscripts checked by English-speakingcolleagues prior to submission. Articles may range from relatively short contributions ( 10-15 typed and double-spaced pages to extensive contributions (20-30 typed pages). Notes are reports of 5 to 10 pages without an abstract and describe meth- ods or results not supported by a large body of data. Manuscript preparation Title page should include authors' full names and mailing addresses and the senior author's telephone, fax number and e-mail address, and a list of key words to describe the contents of the manuscript. Abstract should be limited to 150 words (one-half page), state the main scope of the research, and emphasize the author's con- clusions and relevant findings. Because ab- stracts are circulated by abstracting agencies, it is important that they represent the research clearly and concisely. Text must be typed in 12 point Times New Roman font throughout. A brief introduction should convey the broad significance of the paper; the remainder of the paper should be divided into the following sections: Materials and methods. Results, Discussion (or Con- clusions), and Acknowledgments. Headings within each section must be short, reflect a logi- cal sequence, and follow the rules of multiple sub- division I i.e., there can be no subdivision without at least two items). The entire text should be intelligible to interdisciplinary readers; there- fore, all acronyms, abbreviations, and technical terms should be written out in full the first time they are used. Include FAO common names for species in the list of keywords and in the open- ing statements. Regional common names may be used throughout the rest of the text if they are different. FAO common names can be found at http://www.fishbase.org/search.html. Follow the U.S. Government Printing Office Style Manual ( 1984 ed. ) and the CBE Style Manual ( 6th ed. ) for editorial style, and the most current issue of the American Fisheries Society's Common and Sci- entific Names of Fishes from the United States and Canada for fish nomenclature. Dates should be written as follows: 11 November 2000. Mea- surements should be expressed in metric units, e.g., 58 metric tons (t); if other units of measure- ment are used, please make this fact explicit to the reader. Write out the numbers zero through nine unless they form part of measurement units (e.g., nine fish but 9 mm). Text footnotes should be inserted in 9-point font at the bottom of the page that displays the first citation of the footnote. Footnotes should be formatted in the same manner as citations. Footnote all personal communications, unpub- lished data, and unpublished manuscripts with full address of the communicator or author, or, as in the case of unpublished data, where the data are on file. Authors are advised to avoid refer- ences to nonstandard (gray) literature (such as internal, project, processed, or administrative reports, ICES Council Minutes, IWC Minutes or Working Papers, any "research" or "working" documents, laboratory or contract reports. Man- agement Council reports, and manuscripts in review) wherever possible. If these references are used, present them as footnotes and list whether they are available from NTIS (National Tech- nical Information Service) or from some other public depository. Cite all software and special equipment or solutions used in the study, not in a footnote but within parentheses in the text (e.g., SAS, vers. 6.03, SAS Inst., Inc., Cary. NC ). 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. If there is a sequence of citations in the text, list chrono- logically: (Smith, 1932; Green, 1947; Smith and Jones, 1985). Abbreviations of serials should conform to abbreviations given in the Serial Sources for the BIOSIS Previews Database. Authors are responsible for the accuracy and completeness of all citations. Literature cita- tion format: Author (last name, followed by first-name initials). Year. Title of report or manuscript. Abbreviated title of the series to which it belongs. Always include number of pages. If there is a sequence of citations by the same first author, list the works alphabetically according to the last name of following authors (e.g.. Smith G. P., L. C. Brown, 1982; Smith, G. P., and T. P. Stuart, 1982 ). If the authorship is identical, list works chronologically. Tables and figures— general format • Zeros should precede all decimal points for values less than one. • Sample size, n, should be italicized. • Capitalize the first letter of the first word in all labels within figures. • Do not use overly large font sizes in maps and for units of measurements along axes in figures. • Do not use bold fonts or bold lines in figures. • Submit photographs on glossy paper. • Do not place outline rules around graphs. • Do not use horizontal lines in graphs to indi- cate measurement units on axes. • Use a comma in numbers of five digits or more (e.g. 13,000 but 3000). • Maps should have a North arrow and degrees latitude-longitude (e.g., 170(E) Tables should not be excessive in size and must be cited in numerical order in the text. Headings should be short hut 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 footnote markers. Use asterisks to indicate probability in statistical data. Do not type table legends on a separate page; place them on the same page as the table data. Figures include line illustrations, photographs (or slides), and computer-generated graphs and must be cited in numerical order in the text. Line illustrations may he submitted as high quality laser prints. We require a hard copy of photo- graphs in addition to an electronic copy. Figures art' to be labeled with author's name and number of figure. Avoid placing labels vertically (except on y ax is j. Figure legends should explain all symbols and abbreviations and should be double-spaced on a separate page at the end of the manuscript, Please note that we do not print graphs in color. FAILURE TO FOLLOW THESE GUIDELINES WILL DELAY PUBLICATION OF A MANUSCRIPT Copyright law does not apply to Fishery Bul- letin, which falls within the public domain. However, if an author reproduces any part of an article from Fishery Bulletin in his or her work, reference to source is considered correct form (e.g.. Source: Fish. Bull 97:105). Reprints are available free of charge to the senior author (50 copies) and to his or her labora- tory (50 copies). Additional copies may be pur- chased in lots of 100 when the author receives page proofs. Submission The Scientific Editorial Office encourages au- thors to submit their manuscripts as a single PDF ( preferred J, Word ( zipped ), or WordPer- fect (zipped) document by e-mail to Fishery. Bulletin@noaa.gov. Please use the subject head- ing "Fishery Bulletin manuscript submission." Do not send encrypted files. For further details on electronic submission, please contact the Scien- tific Editorial Office directly (see address below). Or you may send your manuscript on compact disc in one of the above formats along with four printed copies (one original plus three copies) — clipped, not stapled — to the Scientific Editor, at the address shown below. Send photocopies only of figures with initial submission of manu- script; do not send original figures. Original figures and electronic copies of figures will be requested later when the manuscript has been accepted for publication. Until August 2005 Dr. Norman Bartoo Scientific Editor. Fishery Bulletin NOAA/NMFS/SWFSC 8604 La Jolla Shores Dr. La Jolla.CA 92038 Starting August 2005 Dr. Adam Moles Scientific Editor, Fishery Bulletin 11305 Glacier Hwy Juneau, AK 99801-8626 Once the manuscript has been accepted for pub- lication, you will be asked to submit a final soft- ware copy of your manuscript. When requested, the text and tables should be submitted in Word or Word Rich Text Format. Figures should be sent as PDF files, Windows metafiles, or as EPS files. Send a copy of figures in original software if conversion yields a degraded version.