U.S. Department of Commerce Volume 106 Number 1 January 2008 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 John Oliver Acting Assistant Administrator for Fisheries Scientific Editor Adam Moles, Ph.D. Associate Editor Elizabeth Siddon Ted Stevens Marine Research Institute Auke Bay Laboratories Alaska Fisheries Science Center 17109 Pt. Lena Loop Road Juneau, Alaska 99801 Managing Editor Sharyn Matriotti National Marine Fisheries Service Scientific Publications Office 7600 Sand Point Way NE, BIN C15700 Seattle, Washington 98115-0070 The Fishery Bulletin (ISSN 0090-0656) is published quarterly by the Scientific Publications Office, National Marine Fisheries Service, NOAA, 7600 Sand Point Way NE, BIN C15700, Seattle, WA 98115-0070. Periodicals postage is paid at Seattle, WA. POSTMASTER: Send address changes for subscriptions to Fish- ery Bulletin, Superintendent of Docu- ments, Attn.: Chief, Mail List Branch, Mail Stop SSOM, Washington, 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: $36.00 domestic and $50.40 foreign. Cost per single issue: $21.00 domestic and $29.40 foreign. See back for order form. Editorial Committee Jeffrey M. Leis Thomas Shirley David Somerton Mark Terceiro Australian Museum, Sydney, Australia Texas A&M University 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 106 Number 1 January 2008 Fishery Bulletin Contents Articles 1-11 Rooper, Christopher N. An ecological analysis of rockfish ( Sebastes spp.) assemblages in the North Pacific Ocean along broad-scale environmental gradients 12-23 Brooks, Elizabeth N., Kyle W. Shertzer, Todd Gedamke, and Douglas S. Vaughan Stock assessment of protogynous fish: evaluating measures of spawning biomass used to estimate biological reference points 24-39 Stevenson, Duane E., James W. Orr, Gerald R. Hoff, and John D. McEachran Emerging patterns of species richness, diversity, population density, and distribution in the skates (Rajidae) of Alaska 40-46 The National Marine Fisheries Service (NMFS) does not approve, recommend, or endorse any proprie- tary product or proprietary material mentioned in this publication. No reference shall be made to NMFS, or to this publication furnished by 47—57 NMFS, in any advertising or sales promotion which would indicate or imply that NMFS approves, recom- mends, or endorses any proprietary product or proprietary material mentioned herein, or which has as its purpose an intent to cause 58—71 directly or indirectly the advertised product to be used or purchased because of this NMFS publication. Stiansen, Stian, Anders Ferno, Dag Furevik, Terje Jorgensen, and Svein Lokkeborg Efficiency and catch dynamics of collapsible square and conical crab pots used in the red king crab ( Paralithodes comtschaticus) fishery Latour, Robert J., James Gartland, Christopher F. Bonzek, and RaeMarie A. Johnson The trophic dynamics of summer flounder ( Paralichthys dentatus) in Chesapeake Bay Haddon, Malcolm, Craig Mundy, and David Tarbath Using an inverse-logistic model to describe growth increments of blacklip abalone (Ha Hot is rubra ) in Tasmania Fishery Bulletin 106(1 ) 72-81 Byrd, Barbie L, Aleta A. Hohn, Fentress H. Munden, Gretchen N. Lovewell, and Rachel E. Lo Piccolo Effects of commercial fishing regulations on stranding rates of bottlenose dolphin (Tursiops truncatus) 82-92 Schwenke, Kara. L., and Jeffrey A. Buckel Age, growth, and reproduction of dolphinfish ( Coryphaena hippurus) caught off the coast of North Carolina 93-107 Vasslides, James M., and Kenneth W. Able Importance of shoreface sand ridges as habitat for fishes off the northeast coast of the United States 108-109 Guidelines for authors Abstract — Environmental variabil- ity affects the distributions of most marine fish species. In this analy- sis, assemblages of rockfish ( Sebastes spp.) species were defined on the basis of similarities in their distributions along environmental gradients. Data from 14 bottom trawl surveys of the Gulf of Alaska and Aleutian Islands (n = 6 767) were used. Five distinct assemblages of rockfish were defined by geographical position, depth, and temperature. The 180-m and 275-m depth contours were major divisions between assemblages inhabiting the shelf, shelf break, and lower conti- nental slope. Another noticeable divi- sion was between species centered in southeastern Alaska and those found in the northern Gulf of Alaska and Aleutian Islands. The use of envi- ronmental variables to define the species composition of assemblages is different from the use of tradi- tional methods based on clustering and nonparametric statistics and as such, environmentally based analyses should result in predictable assem- blages of species that are useful for ecosystem-based management. Manuscript submitted 7 February 2007. Manuscript accepted 1 August 2007. Fish. Bull 106:1-11 (2008). The views and opinions expressed or implied in this article are those of the author and do not necessarily reflect the position of the National Marine Fisheries Service, NOAA. An ecological analysis of rockfish ( Sebastes spp.) assemblages in the North Pacific Ocean along broad-scale environmental gradients Christopher N. Rooper Email address: Chns.Rooper@noaa.gov Alaska Fisheries Science Center National Marine Fisheries Service 7600 Sand Point Way NE Seattle, Washington 98115 Ecosystem-based management of marine fish species requires all eco- system components to be accounted for in the management framework (Liv- ingston et ah, 2005). This necessitates knowing the major environmental gra- dients along which important species are organized within marine systems. Rockfish ( Sebastes spp.) comprise an important component of marine ecosys- tems; they are abundant, diverse, and a widely dispersed group found across a wide range of habitats on the Pacific coast of North America. Rockfish management is problematic because of the different habitat requirements of these species at different life his- tory stages (Love et ah, 1991; Rooper et al., 2007), their episodic recruit- ment (Ralston and Howard, 1995; Field and Ralston, 2005), and their susceptibility to overfishing (Parker et al., 2000). Knowledge of the environ- mental gradients upon which rockfish organize themselves should be useful in predicting where different species and age groups co-occur and, thus, the best strategies for managing rock- fish as an important component of the marine ecosystem. Analyses used to define fish as- semblages typically are based on methods that cluster characteristics of the catch, such as the Bray-Curtis dissimilarity index (Weinberg, 1994; Williams and Ralston, 2002) and oth- er measures of co-occurrence (Wein- berg, 1994; Mueter and Norcross, 2002), or that classify stations into categories of similar catch (Methratta and Link, 2006; Zimmermann, 2006). These analyses do not explicitly take into account the distribution of spe- cies across environmental gradients, even though fish species are known to respond to characteristics such as water depth, temperature, salinity, and sediment type (Friedlander and Parrish, 1998; Rooper et al., 2005). According to ecological theory, spe- cies will inhabit a preferred niche of environmental conditions (Hutchin- son, 1957). Application of this prin- ciple can be useful in predicting the biological basis for co-occurrence of rockfish species (Murawski and Finn, 1988), as well as for defining groups of species with similar habitat re- quirements. Additionally, the species comprising assemblages defined by ecological relationships may be ex- pected to respond to environmental changes in a predictable fashion. The objective of this study was to analyze the distribution of rockfish species across two large ecosystems: the Gulf of Alaska and the Aleutian Islands. Relationships based on niche theory were developed for rockfish life history stages, sexes, and species sub- groups with depth, temperature, and geographical position in order to test for important overlaps and to infer common distributions among species. The co-occurrence of species in bottom trawl catches was compared to their environmental overlap to determine whether species with similar distribu- tions were likely to be captured to- gether. Finally, the major gradients along which rockfish species organize themselves were examined in relation to the varying life history stages of the species. These methods should al- low for a more robust analysis of the assemblage of similar species, as well 2 Fishery Bulletin 106(1 ) 160°0'0"E 1 70°0'0"E 180“0'0"E 170°0'0"V\ / 160°0'0"W 150“0'0"W 140“0'0"W 130°0'0"W 120°0'0"W 1 70°0'0"W 1 60°0'0"W 1 50o0'0"W 140°0'0"W Figure 1 Map of Alaska showing the Gulf of Alaska and Aleutian Islands ecosystems (sepa- rated by a dashed line). The geographical position of each bottom trawl haul was calculated as the distance from Hinchinbrook Island (star). The distance of trawl hauls from Hinchinbrook Island along the southeastward arrow was coded as nega- tive, and the distance along the westward arrow was coded as positive. as documentation of the important differences among life-stage and species subgroups along environmental gradients. Materials and methods Study area The data used for these analyses were collected during bottom trawl surveys of the Gulf of Alaska and Aleutian Islands ecosystems (Britt and Martin, 2001; Zenger, 2004). The Aleutian Islands are a chain of volcanic islands stretching from the Alaska Peninsula in south- west Alaska across the North Pacific Ocean and dividing the western Gulf of Alaska from the Bering Sea. The upper continental slope in the Aleutian Islands is narrow and steep. In the Gulf of Alaska, the continental shelf ranges in width from 20 km to greater than 200 km and the continental slope is steep and features periodic gul- lies and submarine canyons extending into the continen- tal shelf. These two ecosystems encompass a large area of the Alaska continental shelf, from Dixon Entrance (133°W) in the southeastern Gulf of Alaska to Stalemate Bank (170°E) at the far western end of the Aleutian Islands (Fig. 1). The Gulf of Alaska bottom trawl survey is conducted from the Islands of Four Mountains (170°W) to Dixon Entrance. The Aleutian Islands bottom trawl survey is conducted along the island chain from 165°W to Stalemate Bank on the Bering Sea side and from the Islands of Four Mountains (170°W) to Stalemate Bank on the Gulf of Alaska side. The Aleutian Islands and Gulf of Alaska ecosystems are connected by oceanic currents over the shelf. The Alaska Coastal Stream and Alaska Coastal Current flow westward (counter-clockwise) around the Gulf of Alaska from Dixon Entrance to the end of the Aleutian Island chain, whereas on the Bering Sea side of the Aleutian Islands the current flows eastward and pro- vides extensive transport through passes in the chain from the Gulf of Alaska to the Bering Sea (Stabeno et al., 1999, 2002). The Islands of Four Mountains area is thought to be an area of change in both oceanographic properties (a higher influence of marine waters to the west) and biological properties (Ladd et ah, 2005; Lo- gerwell et al., 2005). Trawl survey data The National Marine Fisheries Service (NMFS) Alaska Fisheries Science Center (AFSC) has conducted standard bottom trawl surveys in the Gulf of Alaska and the Rooper: An ecological analysis of rockfish assemblages in the North Pacific Ocean 3 Aleutian Islands region since 1980 (Britt and Martin, 2001; Zenger, 2004). Surveys were conducted triennially between 1990 and 2000 and biennially thereafter (Table 1). For this analysis, the AFSC bottom trawl data from 1990 through 2005 were combined across years and survey areas in order to maximize the number of useful data points for each species. A Poly Nor’Eastern high- opening bottom trawl (with 24.2-m roller gear equipped with 36-cm rubber bobbins that are separated by 10-cm rubber disks) was used in these AFSC bottom trawl sur- veys. Trawl hauls were conducted at a speed of 5.6 km/h (3 knots) for 15 or 30 minutes. Bottom contact and net dimensions were recorded throughout each trawl haul by using net mensuration equipment. In a few cases, net width was not recorded (n = 35); therefore the overall average net width (15.84 m, standard error [SE] = 0.01) was used. For these analyses, records were used only if trawl performance was satisfactory and if the distance fished, geographical position, average depth, and bottom water temperature were recorded (/? = 6767). Trawl hauls were deemed satisfactory if the net opening was within a predetermined normal range, the roller gear maintained contact with the seafloor, and the net suffered little or no damage during the tow (Zenger, 2004). All fish captured during a survey tow were sorted by species, counted, measured for total or fork length, and the total weight of each species in the catch was deter- mined. For large catches, the total catch was weighed and subsampled for length data. Catch per unit of ef- fort (CPUE, no of fish/ha) for each rockfish species was calculated by using the area swept computed from the net width for each tow multiplied by the distance towed recorded with geographical positioning systems. The rockfish catch data were transformed by using natural log (CPUE + 1) before analyses, hereafter shortened to CPUE. Environmental variables Three environmental variables were included in the analyses: depth, temperature, and geographic position along the Alaska coastline. Temperature measurements were collected throughout each trawl haul with a cali- brated Branker bathythermograph, either a SeaBird-19 or SeaBird-39 microbathythermograph (Sea-Bird Elec- tronics, Inc., Bellevue, WA) attached to the net. Depth was also recorded during each trawl haul either off the vessel echosounder or from the microbathythermograph attached to the net. The average bottom temperature and average depth from each trawl haul were used in the analyses. The final environmental variable included in the analyses was the relative position of each trawl haul along the Alaska coastline. Because the major axes of the Alaska coastline are from south to north in the southeastern Gulf of Alaska and from east to west in the remainder of the Gulf of Alaska and Aleu- tian Islands (Fig. 1), a reference point was chosen to standardize the spatial patterns in trawl hauls. The reference point was chosen in the central Gulf of Alaska at the eastern side of Hinchinbrook Island, (146.08°W, Table 1 Summary of the number of Alaska Fishery Science Center bottom trawl survey hauls used in the analysis of rockfish (Sebastes spp.) assemblages in the North Pacific Ocean. Data from trawl hauls with no temperature, depth, lati- tude, longitude, and inadequate gear performance were not used in the analyses. The Gulf of Alaska and Aleutian Islands bottom trawl surveys were the data source for each ecosystem, although in some years (i.e., 1998) a full survey of the ecosystem was not conducted. Year Ecosystem Number of trawl hauls 1990 Gulf of Alaska 286 1991 Aleutian Islands 63 1993 Gulf of Alaska 727 1994 Aleutian Islands 389 1996 Gulf of Alaska 716 1997 Aleutian Islands 399 1998 Gulf of Alaska 5 1999 Gulf of Alaska 765 2000 Aleutian Islands 415 2001 Gulf of Alaska 550 2002 Aleutian Islands 417 2003 Gulf of Alaska 795 2004 Aleutian Islands 415 2005 Gulf of Alaska 825 Total 6767 60.37°N). The distance from this point to each trawl haul provided the change in spatial distribution (both by latitude and longitude) of rockfish species and the variable is hereafter referred to as the position of each trawl. Thus, a negative position indicates a trawl haul occurring in southeastern Alaska, and a positive position indicates a trawl haul occurring west of Hinchinbrook Island. Although geographical position is not actually an environmental variable, it was used as a proxy for longitudinal and latitudinal gradients that affect fish distribution and ranges. Data analyses In every bottom trawl haul, the CPUE of each rockfish species was divided into adult and juvenile stages by fish length (Table 2). The juvenile stage was broadly defined to include all subadult fish with lengths less than the size at 50% maturity from literature sources (Paraketsov, 1963; McDermott, 1994; NPFMC, 1998, 2005; Love et al., 2002; Pearson and Gunderson, 2003). The adult rockfish of each species were further divided into male and female CPUE components. Trawl hauls from which no length or sex data were collected were eliminated from the analyses for that species. Species for which catches occurred in fewer than 100 trawl hauls were also excluded from the analyses. For each trawl haul, these divisions resulted in three estimates of CPUE for each species: juveniles, adult females, and adult males. 4 Fishery Bulletin 106(1) Table 2 Species of rockfish captured in the Gulf of Alaska and Aleutian Islands trawl surveys across all years (1990-2005). The number captured in all years combined, the number of trawl hauls with catch, and the length distributions (FL, in mm) of juvenile and adult rockfish captured in the bottom trawl surveys are also provided. Dashed lines indicate either no catch in that category or insufficient length data were collected to distinguish juveniles and adults. Species Common name No. captured No. of trawl hauls with catch Juvenile length distribution Adult length distribution Sebastes alutus Pacific ocean perch 1,651,753 3059 40-250 250-620 Sebastes aleutianus Rougheye rockfish 36,250 1894 50-439 439-820 Sebastes polyspinis Northern rockfish 459,372 1547 50-361 361-500 Sebastolobus alascanus Shortspine thornyhead 171,405 1515 40-215 215-770 Sebastes borealis Shortraker rockfish 14,549 676 50-449 449-1030 Sebastes variabilis Dusky rockfish 22,622 633 130-428 428-560 Sebastes babcocki Redbanded rockfish 2985 509 110-420 420-630 Sebastes variegatus Harlequin rockfish 28,834 462 60-230 230-420 Sebastes zacentrus Sharpchin rockfish 63,343 442 Females 70-265, Males 70-240 Females 265-660, Males 240-660 Sebastes brevispinis Silvergray rockfish 11,279 340 Females 150-415, Males 150-395 Females 415-720, Males 395-720 Sebastes helvomaculatus Rosethorn rockfish 2447 158 100-215 215-360 Sebastes proriger Redstripe rockfish 17,868 151 Females 100-290, Males 100-280 Females 290-480, Males 280-480 Sebastes ruberrimus Yelloweye rockfish 287 125 Females 90-450, Males 90-540 Females 450-740, Males 540-740 Sebastes elongatus Greenstriped rockfish 750 77 Females 100-220, Males 100-240 Females 220-380, Males 240-380 Sebastes ciliatus Dark rockfish 1607 65 210-428 428-500 Sebastolobus altivelis Longspine thornyhead 4463 46 60-190 190-340 Sebastes wilsoni Pygmy rockfish 514 44 — - Sebastes crameri Darkblotched rockfish 122 43 Females 120-390, Males 120-370 Females 390-500, Males 370-500 Sebastes melanops Black rockfish 542 34 230-400 400-590 Sebastes maliger Quillback rockfish 98 30 230-290 290-480 Sebastes reedi Yellowmouth rockfish 690 20 Females 240-380, Males 240-370 Females 380-550, Males 370-550 Sebastes pinniger Canary rockfish 182 20 Females 390—510, Males 390-430 Females 510-630, Males 430-630 Sebastes diploproa Splitnose rockfish 55 20 90-270 270-270 Sebastes flavidus Yellowtail rockfish 423 17 Females 330—405, Males 330-380 Females 405-580, Males 380-580 Sebastes entomelas Widow rockfish 86 14 Females 350-365 Females 365-570, Males 345-570 Sebastes paucispinis Bocaccio 9 8 — 690-720 Adelosebastes latens Aleutian scorpionfish 6 3 — — Sebastes emphaeus Puget Sound rockfish 11 2 — 170-180 Sebastes saxicola Stripetail rockfish 9 2 — 100-270 Sebastolobus macrochir Broadfin thornyhead 4 2 — — Sebastes miniatus Vermilion rockfish 2 1 — — Sebastes nigrocinctus Tiger rockfish 1 1 — — In order to determine the species composition of rock- fish assemblages, the CPUE-weighted distribution of each of the species subgroups was computed for each environmental variable with the formulation given in May (1973) and Murawski and Finn (1988) (Fig. 2). A weighted mean value for each environmental variable Rooper: An ecological analysis of rockfish assemblages in the North Pacific Ocean 5 Figure 2 Theoretical distribution of three species along an environmental gradient ( K ). In this study, K was depth, temperature, or distance. The value w is the standard deviation of the resource utilization curve for a single species and d is distance between the weighted means for two species along a resource continuum. Reproduced from May (1973). (depth, temperature, and position) was computed as Mean = where ft - the CPUE of each rockfish species group in tow i; and xt = the value of the environmental variable at tow i. The weighted standard deviation (SD) was then com- puted as X(/>,) It (I A?) \ M *mean2 j i (i> i-1 These calculations yielded the niche dimensions for each species group defined along each of the three environ- mental gradients. The overlap of each species group across each envi- ronmental gradient (A) can then be calculated as Av = C,j exp where the normalization constant (CL) is calculated by Cv = 2W:W , 1 J 2 , 2 Wj +Wj where d = the distance between means for a pair of species i and j; wt = the standard deviation (SD) for species i; and W: = the SD for species j (Murawski and Finn, 1988). Overlap indices for each variable were calculated for each species and within each species, by the groupings of males, females, and juveniles. The overlaps between males and females were first examined and if the overlap index was greater than 0.9 across all three environmen- tal gradients, these males and females were combined and the means and SDs for each environmental gradi- ent were recalculated. The resulting groupings (adults and juveniles) were then compared and, again, where the indices exceeded 0.9, the catches for the entire spe- cies (all size and sex classes) were combined and the means and SDs for each environmental gradient were recalculated. Finally, the overlap indices were compared to de- termine the amount of separation among both species and the remaining groupings by size and sex. The multinomial intersection in the overlap indices across the three environmental gradients was calculated by multiplying the individual overlap coefficients together. This was used as a measure of the relative similarity in environmental preferences computed for each pair of species subgroups and resulted in a matrix of overlap coefficients for all species-group pairs. The matrix was then clustered into assemblages with similar distribu- tions across environmental gradients by using Primer Analysis software (PRIMER-E Ltd., Plymouth, UK). The combined overlap index was used as a measure of similarity and the average cluster linkage method was used to determine the species composition of rockfish 6 Fishery Bulletin 106(1 ) Table 3 The estimated mean weighted values and niche width (w) across each of the three resource gradients (position, depth, and tem- perature) for the species-subgroups examined in this analysis. Although the species were initially split into male, female, and juvenile components, they were recombined if overlaps within a species (i.e., between sexes or between juveniles and adults) were greater than 0.9 across all three environmental variables. Species Common name Group Position (km) mean w Depth (m) Temperature (°C) mean w mean w Sebastes variabilis Dusky rockfish All 757 780 141.3 57.0 5.4 0.9 Sebastes variegatus Harlequin rockfish All 200 1124 159.8 50.3 5.6 0.7 Sebastes polyspinis Northern rockfish Adults 1265 818 134.1 54.4 5.0 0.9 Juveniles 1819 1180 137.6 56.0 4.7 1.0 Sebastes alutus Pacific ocean perch Adults 1205 1176 211.8 75.9 4.7 0.9 Juveniles 857 1231 164.2 47.9 5.1 0.9 Sebastes babcocki Redbanded rockfish All -443 1532 232.5 78.5 5.4 0.6 Sebastes proriger Redstriped rockfish All -676 1720 193.8 42.3 5.6 0.6 Sebastes helvomaculatus Rosethorn rockfish All -779 1760 215.1 54.7 5.6 0.6 Sebastes aleutianus Rougheye rockfish Adults 1112 1043 315.7 164.6 4.4 1.0 Juveniles 685 987 244.3 112.9 5.0 0.9 Sebastes zacentrus Sharpchin rockfish All -430 1482 195.9 45.9 5.6 0.6 Sebastes borealis Shortraker rockfish All 1325 1124 354.4 194.5 4.2 1.1 Sebastolobus alascanus Shortspine thornyhead All 468 1160 318.4 192.9 4.7 0.9 Sebastes brevispinis Silvergray rockfish Adults -713 1737 202.4 49.1 5.7 0.7 Juveniles -186 1319 153.0 54.2 6.0 1.0 Sebastes ruberrimus Yelloweye rockfish All 66 1101 143.8 45.8 5.7 0.7 assemblages that had similar distributions along the environmental gradients. The combined overlap indi- ces for each species pair were also compared to the frequency of co-occurrence for the species pair in trawl hauls by using linear regression to determine if the distribution of species across environmental variables was directly linked to their co-occurrence in trawl hauls. Results There was little difference in the distributions of the examined rockfish species among sexes. For all species, the overlap indices exceeded 0.9 across all three envi- ronmental gradients between males and females and, thus, the CPUE data were combined across sexes. Some differences between adults and juveniles were observed in their distributions across all three environmental variables (Table 3). Juvenile Pacific ocean perch (POP) ( Sebastes alutus) and silvergray rockfish ( S . brevispinis) were distributed at shallower depths than were adults of the same species. There was also a distinct separation between juvenile and adult rougheye rockfish (S. aleu- tianus) along the temperature gradient; juveniles were found at slightly higher temperatures than were adults. Juvenile northern rockfish ( S . polyspinis) were found farther west (approximately 600 km) along the Alaska Peninsula than were adults. In total, these preliminary analyses indicated that there were four species where juveniles and adults were found to be separate and nine species that did not have different distributions between either sexes or life stages for any of the three environ- mental gradients. As a result, 17 species subgroups were analyzed (Table 3). The cluster analysis resulted in five assemblages of rockfish species subgroups (Fig. 3). There was a rel- atively shallow-water assemblage (Aleutian Islands- shelf) containing northern rockfish, juvenile POP, and dusky rockfish (S. variabilis). These species had mean weighted depths from 134 to 164 m and were widely distributed around the northern Gulf of Alaska and Aleutian Islands (mean weighted distances from 757 to 1819 km). A similar assemblage (central Gulf of Alaska- shelf) had mean weighted depths from 144 to 160 m and was distributed from 200 to -186 km. This assemblage contained three species subgroups: yelloweye rockfish (S. ruberrimus), harlequin rockfish (S. variegatus), and juvenile silvergray rockfish. The third assemblage (southeastern Alaska-break) consisted of a group of spe- cies found predominantly in southeastern Alaska (mean weighted distance -430 to -779) at depths centered around 208 m. These species included adult silvergray rockfish, redbanded rockfish OS. babcocki ), rosethorn rockfish (S. helvomaculatus), sharpchin rockfish (S. zacentrus ), and redstripe rockfish (S. proriger). The Rooper: An ecological analysis of rockfish assemblages in the North Pacific Ocean 7 1.0 Central Gulf of Alaska- shelf Southeastern Alaska- break r Aleutian Islands-shelf Aleutian J Islands-break | Aleutian Islands-slope Juvenile silvergray rockfish Harlequin rockfish — Yelloweye rockfish — Redbanded rockfish — Rosethorn rockfish Adult silvergray rockfish Redstripe rockfish Sharpchin rockfish Juvenile northern rockfish Adult northern rockfish Dusky rockfish Juvenile Pacific ocean perch Adult Pacific ocean perch Juvenile rougheye rockfish Shortspine thornyhead Adult rougheye rockfish Shortraker rockfish — 0.8 Relative similarity 0.6 0.4 0.2 Figure 3 The results of cluster analysis of rockfish showing relative similarity amongst species-subgroups. The x-axis shows the relative similarity amongst species derived from the multinomial overlap indices among species-group pairs along the three environmental gradients (depth, position, and temperature). The dashed line (0.73) is where rockfish species assemblages were defined based on a similarity of 0.9 across the three environmental gradients. fourth assemblage (Aleutian Islands-break) included juvenile rougheye rockfish and adult POP. These spe- cies were distributed at the shelf break at depths of 244 and 212 m, in the north Gulf of Alaska and Aleutian Islands (mean weighted distances 1205 and 685 km). A final assemblage (Aleutian Islands-slope) was com- posed of adult rougheye rockfish, shortraker rockfish (S. borealis ), and shortspine thornyhead ( Sebastolobus alascanus). These species were generally distributed in the north Gulf of Alaska and Aleutian Islands, at depths from 318 to 355 m. Bivariate plots of the three environmental variables indicate that the assemblages had very different distributions across the depth and position variables, and less distinct separation around the temperature variable (Fig. 4). There were clear lines of division along the depth variable between shelf and shelf break assemblages at 180 m, and between shelf break and slope assemblages at 275 m. There were also divisions along the position variable between southeast- ern Alaska species, central Gulf of Alaska species, and the Aleutian Island species at approximately Icy Strait (-350 km) and Seward, Alaska (250 km). According to the trawl survey CPUE, the most abun- dant 10 species subgroups captured were adult POP, shortspine thornyhead, juvenile POP, juvenile northern rockfish, juvenile rougheye rockfish, adult northern 8 Fishery Bulletin 106(1 ) A E Cl (D Q X o o o X □ X X o 400 x Aleutian Islands-shelf X Central GOA-shelf - 350 O Aleutian Islands-break A Southeastern Alaska-break ’ i,MI o Aleutian Islands-slope X ■ 250 A ■200 A ■ 15* 6 ■ 100 • 50 2000 1500 1000 500 0 -500 -1000 Position (km) o Position (km) Figure 4 Plots of mean weighted (by catch per unit of effort) distributions of each rockfish species-group (n = 17) along three environmental variables. Mean weighted distributions of rockfish species-groups are shown for (A) depth versus position, (B) temperature versus position, and (C) temperature versus depth. Position is the distance from Hinchinbrook Island, Alaska; positive values are west of this central point in the trawl surveys and negative values are southeastward. Dashed lines indicate break points in the mean-weighted environmental variables. Symbols indicate the assemblage membership (from Fig. 3) of each species-group as belonging to the Aleutian Islands-shelf, central Gulf of Alaska (GOA shelf), Aleutian Islands-break, southeastern Alaska- break, or Aleutian Islands-slope assemblage. rockfish, adult rougheye rockfish, shortraker rockfish, sharpchin rockfish, and dusky rockfish. For these species the co-occurrence in trawl survey catches had a strong linear relationship to the amount of overlap in their distributions across environmental variables, and over 50% of the variance in co-occur- rence was explained by species group overlap (Fig. 5). When all species subgroups were considered, the relationship was significant, but the overlap index explained only 32% of the variance in co-oc- currence among pairs of species subgroups. This demonstrates that the co-occurrence of rockfish in the trawl survey data is positively cor- related to their overlap in environ- mental preferences. Discussion The method used in this study provided resolution of the species composition of rockfish assemblages across large-scale environmental gradients that influence fish dis- tribution (depth and geographical position). The results of these anal- yses were similar to findings on fish assemblages in other areas. Rock- fish (and other species) on the west coast of North America distribute themselves by depth and latitude into distinct assemblages (Wein- berg, 1994; Williams and Ralston, 2002; Tolimieri and Levin, 2006). In the Gulf of Alaska, an analysis conducted on 72 groundfish species (including rockfish species) with data from five NMFS trawl surveys revealed that the major gradients for variation in species diversity were depth and alongshore distance, whereas temperature and temporal gradients had only minor effects on species composition (Mueter and Norcross, 2002). As in the current study, Mueter and Norcross (2002) and Williams and Ralston (2002) found a peak in groundfish species richness near the shelf break (200 m), in the region of overlap between shallower species and those occur- ring at deeper depths. The method of defining rockfish assemblages described here is very Rooper: An ecological analysis of rockfish assemblages in the North Pacific Ocean 9 0.4 0.5 0.6 Overlap Figure 5 Frequency of co-occurrence of species in trawl hauls versus combined multinomial overlap index. Data are shown for the top 10 species- subgroups in terms of total catch per unit of effort for the Gulf of Alaska and Aleutian Island trawl surveys, and each dot represents a comparison of species-groups 0i = 90). different from more commonly used meth- ods where trawl survey catches or stations with similar components are clustered together. More commonly used approaches define assemblages by comparing patterns of catch by species in trawl hauls (i.e., Weinberg, 1994; Williams and Ralston, 2002; Zimmermann, 2006). Although these methods are highly effective for de- termining patterns in catches, they typi- cally are based on complex analyses such as nonmetric multidimensional scaling, principle components, and cluster analy- ses that can make interpretation difficult. Determining the scaling method to apply to catch data for abundant versus rare species, the patchy distribution of some species, and inherent differences in catch- ability (such as between small and large fish of the same species) are all problems that must be addressed with these meth- ods. By first defining the relationship of a species to environmental variables, and then comparing the parameters of that relationship to parameters for other spe- cies or life history stages, this analysis method may avoid some of those potential pitfalls. For example, the absolute abundance or catchability should not matter in the identification of the correct assemblage member- ship for a species. If the proper distribution of a species along a depth gradient is known from the trawl data; the species will be placed within a group of species with similar depth distributions regardless of its total abundance. The issue of whether trawl survey data can be used to determine the underlying relationship of rockfish spe- cies to large-scale environmental gradients may have some limitations. Less sampling effort was directed at shallow inshore depths (0-50 m) than at greater depths; therefore shallow-water species were likely under-repre- sented in the catches. Although this analysis included three environmental gradients, there are obviously more variables needed to fully describe the niche dimensions for any of the rockfish species. One important feature that was omitted from this analysis was the effect of small-scale habitat features on rockfish distribution. Variability in rockfish species assemblages on a small scale is often related to the local habitat, where higher habitat heterogeneity is correlated with higher diversity and abundance (Matthews, 1990; Stein et al., 1992; Yoklavich et al., 2000). The data used in this analysis was collected only on trawlable ground, and large tracts of untrawlable area were unsampled. Species composi- tion and abundance can be starkly different between trawlable and untrawlable locations (Matthews and Richards, 1991). Even within trawlable areas, habitat characteristics such as presence of epibenthic inverte- brates can be correlated with increased catches of some life history stages of rockfish, such as juveniles that seek out complex habitat features (Rooper and Boldt, 2005; Rooper et al., 2007). Larger-scale phenomena, such as patterns in prey productivity and the effects of local currents on rockfish distribution, were also absent from this analysis and can certainly affect the distribu- tion of those species that prey on planktonic organisms. Geographical position was used as a proxy for these large-scale phenomena and other unknown variables influencing rockfish distributions. By pooling the data, I did not take into account interannual changes in geographical position due to climate events, such as El Nino or the Pacific decadal oscillation shifts. The ef- fects of El Nino events are typically exhibited through changes in water temperature and there was only a weak response of the species to temperature gradients, and major climate shifts were not detected in Alaska during the years examined. Knowledge of the effects of climate shifts and local habitat features would further improve future assemblage analyses. Because these analyses were based on rockfish re- lationships with environmental variables, they should result in predictable species assemblages useful for ecosystem-based management. For example, species that co-occur in trawl catches due to overlapping dis- tributions with environmental variables would be likely to experience similar fishing mortalities. Additionally, based on these assemblage analyses, marine protected areas could be designed for specific depth and geo- graphical areas that would protect portions of rockfish populations. A series of marine protected areas has been suggested for shortraker and rougheye rockfishes in the Gulf of Alaska as a first step towards spatial management (Soh et al., 2000). The spatial and depth separation of juveniles and adults in many of the spe- cies examined here could also provide information to implement spatially based management systems for 10 Fishery Bulletin 106(1) some species, where bycatch of smaller, immature mem- bers of a population would be protected from harvest. Acknowledgments This work would not have been possible without the assistance of numerous scientists and fishermen who have collected data on the trawl surveys of the Gulf of Alaska and Aleutian Islands. J. Boldt provided assis- tance in the use of the PRIMER software in the data analyses. The manuscript was improved with reviews by R. Reuter, D. Somerton, M. Wilkins, M. Zimmermann, and four anonymous reviewers. Literature cited Britt, L. L., and M. H. Martin. 2001. Data report: 1999 Gulf of Alaska bottom trawl survey. NOAA Tech. Memo. NMFS-AFSC-121, 249 P Field, J. C., and S. Ralston. 2005. Spatial variability in rockfish ( Sebastes spp.) re- cruitment events in the California Current System. Can. J. Fish. Aquat. Sci. 62:2199-2210. Friedlander, A. M., and J. D. Parrish. 1998. Habitat characteristics affecting fish assem- blages on a Hawaiian coral reef. J. Exp. Mar. Biol. Ecol. 224:1-30. Hutchinson, G. E. 1957. Concluding remarks. Cold Spring Harbor Symp. Quant. Biol. 22:415-427. Ladd, C., G. L. Hunt Jr., C. Mordy, S. Salo, and P. Stabeno. 2005. Marine environment of the eastern and central Aleutian Islands. Fish. Oceanogr. 14:22-38. Livingston, P. A., K. Aydin, J. Boldt, J. Ianelli, and J. Jurado- Molina. 2005. A framework for ecosystem impacts assess- ment using an indicator approach. ICES J. Mar. Sci. 62:592-597. Logerwell, E. A., K. Aydin, S. Barbeaux, E. Brown, M. E. Con- ners, S. Lowe, J. W. Orr, I. Ortiz, R. Reuter, and P. Spencer. 2005. Geographic patterns in the demersal ichthyofauna of the Aleutian Islands. Fish. Oceanogr. 14:93-112. Love, M. S., M. H. Carr, and L. J. Haldorson. 1991. The ecology of substrate-associated juveniles of the genus Sebastes. Environ. Biol. Fishes 30:225-243. Love, M. S., M. Yoklavich, and L. Thorsteinson. 2002. The rockfishes of the northeast Pacific, 404 p. Univ. California Press, Berkeley, CA, and Los Angeles, CA. Matthews, K. R. 1990. A comparative study of habitat use by young-of-the- year, subadult, and adult rockfishes on four habitat types in central Puget Sound. Fish. Bull. 88:223-239. Matthews, K. R., and L. J. Richards. 1991. Rockfish iScorpaenidae) assemblages of trawlable and untrawlable habitats off Vancouver Island, British Columbia. N. Am. J. Fish. Manag. 11:312-318. May, R. M. 1973. Stability and complexity in model ecosystems, 265 p. Monographs in population biology — 6. Princeton Univ. Press, Princeton, NJ. McDermott, S. F. 1994. Reproductive biology of rougheye and shortraker rockfish, Sebastes aleutianus and Sebastes borealis. M. S. thesis, 76 p. Univ. Washington, Seattle, WA. Methratta, E. T., and J. S. Link. 2006. Associations between surficial sediments and groundfish distributions in the Gulf of Maine-Georges Bank region. N. Am. J. Fish. Manag. 26:473-489. Mueter, F. J., and B. L. Norcross. 2002. Spatial and temporal patterns in the demersal fish community on the shelf and upper slope regions of the Gulf of Alaska. Fish. Bull. 100:559-581. 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. NPFMC (North Pacific Fishery Management Council). 1998. Stock assessment and Fishery evaluation report for the groundfish resources of the Gulf of Alaska, 32 p. NPFMC, 605 W. 4th Ave, Suite 306, Anchorage, AK 99510. 2005. North Pacific groundfish stock assessment and fishery evaluation reports for 2006, 1188 p. NPFMC, 605 W. 4th Ave, Suite 306, Anchorage, AK 99510. Paraketsov, I. A. 1963. On the biology of Sebastodes alutus of the Bering Sea. In Soviet fisheries investigations in the Northeast- ern Pacific, part I (P. A. Moiseev, ed.), p. 319-327. Avail- able from United States Dep. Commer., Natl. Tech. Inf. Serv., Springfield, VA, as TT67-51203. Parker, S. J., S. A. Berkeley, J. T. Golden, D. R. Gunderson, J. Heifetz, M. A. Hixon, R. Larson, B. M. Leaman, M. S. Love, J. A. Musick, V. M. O’Connell, S. Ralston, H. J. Weeks, and M. M. Yoklavich. 2000. Management of Pacific rockfish. Fisheries 25(3):22-30. 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 67:117-136. Ralston, S., and D. F. Howard. 1995. On the development of year-class strength and cohort variability in two northern California rockfishes. Fish. Bull. 93:710-720. Rooper, C. N., and J. L. Boldt. 2005. Distribution of juvenile Pacific ocean perch Se- bastes alutus in the Aleutian Islands in relation to ben- thic habitat. Alaska Fish. Res. Bull. 11:102-112. Rooper, C. N., J. L. Boldt, and M. Zimmermann. 2007. An assessment of Pacific ocean perch ( Sebastes alutus ) habitat use in a deepwater nursery. Estuarine Coastal Shelf Sci. 75:371-380. Rooper, C. N., M. Zimmermann, and P. Spencer. 2005. Distribution of flathead sole ( Hippoglossoides elas- sodon) by habitat in the eastern Bering Sea. Mar. Ecol. Prog. Ser. 290:251-262. Soh, S., D. R. Gunderson, and D. H. Ito. 2000. The potential role of marine reserves in the management of shortraker rockfish ( Sebastes borea- lis) and rougheye rockfish {S. aleutianus ) in the Gulf of Alaska. Fish. Bull. 99:168-179. Stabeno, P. J., J. D. Schumacher, and K. Ohtani. 1999. The physical oceanography of the Bering Sea. In Dynamics of the Bering Sea (T. R. Loughlin, and Rooper: An ecological analysis of rockfish assemblages in the North Pacific Ocean 11 Ohtani, K., eds.), p. 1-59. Univ. Alaska Sea Grant College Program, AK-SG-99-03, Fairbanks, AK. Stabeno, P. J., R. K. Reed, and J. M. Napp. 2002. Transport through Unimak Pass, Alaska. Deep- Sea Res. II 49:5919-5930. Stein, D. L., B. N. Tissot, M. A. Hixon, and W. Barss. 1992. Fish-habitat associations on a deep reef at the edge of the Oregon continental shelf. Fish. Bull. 90:540-551. Tolimieri, N., and P. S. Levin. 2006. Assemblage structure of eastern Pacific ground- fishes on the U.S. continental slope in relation to physi- cal and environmental variables. Trans. Am. Fish. Soc. 135:317-332. Weinberg, K. L. 1994. Rockfish assemblages of the middle shelf and upper slope off Oregon and Washington. Fish. Bull. 92:620-632. Williams, E. H., and S. Ralston. 2002. Distribution and co-occurrence of rockfishes (family: Sebastidae) over trawlable shelf and slope habi- tats of California and southern Oregon. Fish. Bull. 100:836-855. Yoklavich, M. M„ H. G. Greene, G. M. Cailliet, D. E. Sullivan, R. N. Lea, and M. S. Love. 2000. Habitat associations of deep-water rockfishes in a submarine canyon: an example of a natural refuge. Fish. Bull. 98:625-641. Zenger, H. H., Jr. 2004. Data report: 2002 Aleutian Islands bottom trawl survey. NOAA Tech. Memo. NMFS-AFSC-143, 247 p. Zimmer mann, M. 2006. Benthic fish and invertebrate assemblages within the National Marine Fisheries Service U.S. west coast triennial bottom trawl survey. Cont. Shelf Res. 26:1005-1027. 12 Abstract — In stock assessments, recruitment is typically modeled as a function of females only. For pro- togynous stocks, however, dispropor- tionate fishing on males increases the possibility of reduced fertiliza- tion rates. To incorporate the impor- tance of males in protogynous stocks, assessment models have been used to predict recruitment not just from female spawning biomass (SO, but also from that of males (Sm) or both sexes ( Sb ). We conducted a simulation study to evaluate the ability of these three measures to estimate biologi- cal reference points used in fishery management. Of the three, St pro- vides best estimates if the potential for decreased fertilization is weak, whereas S'" is best only if the poten- tial is very strong. In general, Sb esti- mates the true reference points most closely, which indicates that if the potential for decreased fertilization is moderate or unknown, Sb should be used in assessments of protogynous stocks. Moreover, for a broad range of scenarios, relative errors from St and Sb occur in opposite directions, indicating that estimates from these measures could be used to bound uncertainty. Manuscript submitted 25 April 2007. Manuscript accepted 1 August 2007. Fish. Bull 106:12-23 (2008). The views and opinions expressed or implied in this article are those of the author and do not necessarily reflect the position of the National Marine Fisheries Service, NOAA. Stock assessment of protogynous fish: evaluating measures of spawning biomass used to estimate biological reference points Elizabeth N. Brooks (contact author) Email addresss: Liz.Brooks@noaa.gov National Marine Fisheries Service, NOAA Southeast Fisheries Science Center 75 Virginia Beach Drive Miami, Florida 33149 Present address: National Oceanic and Atmospheric Administration National Marine Fisheries Service, Northeast Fisheries Science Center 166 Water Street Woods Hole, Massachusetts 02543 Kyle W. Shertzer National Marine Fisheries Service, NOAA Southeast Fisheries Science Center Center for Coastal Fisheries and Habitat Research 101 Pivers Island Road Beaufort, North Carolina 28516 Todd Gedamke Virginia Institute of Marine Science College of William and Mary PO. Box 1346, Route 1208 Greate Road Gloucester Point, Virginia 23063 Douglas S. Vaughan National Marine Fisheries Service, NOAA Southeast Fisheries Science Center Center for Coastal Fisheries and Habitat Research 101 Pivers Island Road Beaufort, North Carolina 28516 Populations persistence requires that losses from mortality must at least be matched by gains from the production of new individuals (i.e., recruitment). The theory of stock reproduction relates recruitment to total egg pro- duction (Beverton and Holt, 1957). In practice, however, stock assess- ment often relates recruitment to the biomass of mature females rather than to total egg production. The two predictors are functionally similar if egg production of a mature female is highly correlated to body mass (Roth- schild and Fogarty, 1989), as observed or assumed for many stocks. With the use of either predictor — biomass of mature females or total egg production — the proportion of eggs fertilized is assumed to be con- stant. This assumption is believed to be valid for stocks with little fluc- tuation in sex ratio, as in most gono- choristic stocks (fish that remain the same sex throughout life). However, this assumption may be inappropri- ate for protogynous stocks (fish that begin life as female and later become male). Under natural mortality alone, sex ratios of protogynous stocks are expected to be skewed toward fe- males (Allsop and West, 2004). The addition of fishing mortality could skew the ratio even further (Cole- man et al., 1996; McGovern et al., 1998; Armsworth, 2001), particularly Brooks et a!.: Stock assessment of protogynous fish 13 if fishing preferentially removes males by targeting larger (older) individuals, for example through gear selectivity or management regulations. A disproportion- ate reduction of males could lower fertilization rates if not enough males are available to fertilize the eggs of mature females (i.e., the reduction could result in sperm limitation). The possibility of reduced fertilization rates raises the question of whether protogynous stocks are more susceptible than gonochoristic stocks to overexploita- tion. Several studies have concluded that protogynous stocks are more susceptible, based on hypothesized patterns of reproduction, sexual transition, and fish- ing (Huntsman and Schaaf, 1994; Alonzo and Mangel, 2004, 2005). At least one study (Bannerot et al., 1987) indicates that, under some conditions, protogynous stocks are more resilient to exploitation. Either way, management of protogynous stocks merits the consid- eration of unconventional techniques (Shepherd and Idoine, 1993; Armsworth, 2001; Heppell et ah, 2006). In the United States, fishery management under the Magnuson-Stevens Fishery Conservation and Manage- ment Reauthorization Act of 2006 emphasizes the con- cept of maximum sustainable yield (MSY). Internation- ally, the use of MSY as a reference point for evaluating sustainable development is well established (FAO, 1999). Standard MSY-based biological reference points — the benchmarks used to gauge stock status — include fish- ing mortality rate at MSY (FMSY), spawning biomass at MSY ( SMSY ), and MSY itself. All depend fundamentally on the spawner-recruit relationship, which is typically a function of spawning biomass (S). In conventional stock assessments, S is computed from females only (SO, and fertilization rate is im- plicitly assumed to be constant. Some assessments of protogynous stocks have emphasized the importance of males, by computing S from spawning biomass of males alone (Sm) or from the sum of both sexes ( Sb ) (Punt et al., 1993; Vaughan et al., 1995). Early use of Sb was in per-recruit analyses (Vaughan et al., 1992; Punt et al., 1993; Vaughan et al., 1995), and later, in spawner-recruit relationships (Vaughan and Prager, 2001). The measure of spawning biomass — Sf, Sm, or Sb — used in an assessment plays a key role in estimates of biological reference points, and thus in subsequent man- agement advice. For example, in U.S. fishery manage- ment, a stock is considered to be overfished if the most recent estimate of S is sufficiently less than SMSY. (The level associated with “sufficiently” varies by stock, but the criterion to determine that level often takes natural mortality into account.) Declaring a stock overfished triggers development of a rebuilding plan to increase the stock to SMSY. In general, the choice of measure used to represent spawning biomass influences analyses on which management is based, including any esti- mate of stock status. Although various measures are used in assessments, the properties of reference points estimated from Sf, Sm, or Sb have not been examined comprehensively. We use simulations to evaluate the performance of each measure of spawning biomass. To begin, we simu- late a protogynous stock over an array of biological and fishery characteristics and calculate biological reference points for each case. Then we apply an assessment model to estimate those same reference points using each of the three S measures. The estimated reference points are compared to their simulated counterparts to quantify estimation error. These results are intended to help stock assessment biologists identify a robust measure of spawning biomass that is appropriate for the protogynous stock being modeled. Materials and methods Two deterministic models were constructed, both struc- tured by age and sex, to describe a protogynous stock. The first, referred to as the simulation model, was con- sidered a representation of the real world. It was used to compute true values of MSY-based biological refer- ence points (BRPs), which determine stock status. The second, the assessment model, was used to estimate those same reference points. Both models included age- specific values of maturity, mortality, sex ratio, and size. They differed only in computation of recruitment: the simulation model derived recruits directly from fertil- ized eggs, and the assessment model derived recruits indirectly from the spawning biomass of males, females, or both. Thus, with the assessment model the common assumption is that fertilization rates are static. Because that assumption creates the only structural difference between the simulation and assessment models, the source of any estimation error of computed quantities (BRPs) could be isolated and the most robust measure of spawning biomass could be identified. In this sense, estimation error refers to error caused by model mis- specification, rather than from fitting data. To quantify error systematically, BRPs were computed and estimated under many combinations of biological parameters and fishery conditions, as described below. Simulation model This study used an age-structured population model to compute the number of individuals at age (Na), N a N„ -iM+F, a-1 ' [Na_1e~{M+Fa~l) / (1- e~{M+Fa)) 2 < a < 50 a = 50 (1) where Nl represents the number of recruits (described below), and the maximum age (50) was treated as a plus group. The parameter M is natural mortality rate (constant across age), and Fa is fishing mortality rate at age, equal to the product of total fishing mortality rate (F) and selectivity at age (sa). Selectivity was assumed to be knife-edge, that is, sa = 0 for all ages younger than the first vulnerable age class (as) and sa = 1 otherwise. 14 Fishery Bulletin 106(1 ) Length at age (la) was modeled with the von Berta- lanffy equation (von Bertalanffy, 1938), Za=L00(l-e_if(a_4o)), in which Lx is the asymptotic length, K is the growth coefficient, and t0 is the theoretical age at which length is zero (Z0 = 0 assumed arbitrarily). Length at age was converted to weight at age ( wa ) by the allometric rela- tionship Wa = Vlla2’ (2) where u2 and v2 are constants under the assumption of isometric growth. This relationship was also used to model fecundity at age ( ea , eggs per mature female), ea = £iC2> (3) where f7 and S2 are constants. Fecundity often scales nearly linearly with weight, such that £2 = u2 = 3. Transition from female to male was modeled as a logistic function of age, 1 with pa the proportion male at age, j3p the slope of sexual transition, and ap the age at 50:50 sex ratio. The same function was used to model female maturity at age ( ga ), with parameter j3g = the slope, and ag = the age at 50% maturity. All males were considered to be mature on the basis of low numbers of transitional fish observed in the field and the apparent ability to complete sex transition between spawning seasons (Collins et al., 1987). Total egg production ( E ) was determined by the prod- uct of mature females and eggs per female, summed across ages, E = ^Na^-Pa^aea- <5> k , which can range from 0.2 to 1.0 (Fig. 1). A high value of k corresponds to a stock that can maintain its fertilization rate when males are scarce. In terms of life histories, one might expect group spawners to have higher k than pair spawners. The number of fertilized eggs (i/d under fishing rate F was computed as the product of fertilization rate and total egg production (t p-f(xF)E). Recruitment was computed from fertilized eggs (R(ijj)) with the Beverton-Holt spawner-recruit model, Because fertilization may become limited by sperm avail- ability, fertilization rate (/) was modeled as a function of sex ratio, f(xF) = 4k xF (l-Kr) + (5x-l)xF (6) In Equation 6, xF is the ratio of the proportion of males in the population (in numbers) under fishing rate F to the proportion males at the unfished level, a measure of male depletion (x^E [0,1]). The fertilization rate function f is a form of the Beverton-Holt recruitment model scaled to one for xF = 1. It has similar shape to the fertilization function of Heppell et al. (2006) and has the following desirable properties. In the absence of males, / takes its minimum value of 0.0, and at the unfished sex ratio, f takes its maximum value, which is set arbitrarily to 1.0. In between these extrema, fertilization rate depends on the steepness parameter R( (//) = 4hR0y/ R0(p0 ( 1 - h ) + ( 5/z - 1 )y/ (7) In this parameterization (Mace and Doonan, 1988), cp0 is the unfished level of fertilized eggs per recruit, R0 is unfished recruitment, and h is steepness (analogous to k in the fertilization function). Parameter values Based on life-history theory and empirical study of pro- togynous fish, values of several parameters were related to natural mortality rate in order to avoid untenable parameter combinations and to maintain generality of results. Gardner et al. (2005) reported relationships between growth rate and natural mortality (K=0.64M), age at 50% maturity and natural mortality (e/_,=0.96/ M ), and size at 50:50 sex ratio and asymptotic length (L50 = 0.77Loo). The results from Gardner et al. (2005) Brooks et al.: Stock assessment of protogynous fish 15 Table 1 Model parameters. Values in braces are levels used for the primary analysis, where the assessment model did not account for dynamics of fertilization. Parameter Value(s) Description M 10.1,0.2,0.31 L„ 1000 K 0.64M ag 0.96/M cg 10.75, 1.0, 1.251 ag ii Pg 10.2, 0.4, 0.8, 1.61 aP 2.3/M Pp (0.2, 0.4, 0.8, 1.61 Cs 10.75, 1.0, 1.251 as «s=cs«g V1 1x1 0 s v2 3.0 £i 1.0 % 3.0 K 10.2, 0.3, ..., 1.01 h (0.4, 0.6, 0.8} Ro 1x10s Natural mortality rate Asymptotic maximum length Growth coefficient (Gardner et al., 2005) Mean age at 50% maturity (Gardner et al., 2005) Age at 50% maturity relative to the mean Age at 50% maturity Slope of logistic maturity function Age at 50:50 sex ratio (Gardner et al., 2005) Slope of logistic sex-transition function Age at selection relative to maturity Age at selection Weight-at-age coefficient Weight-at-age exponent Fecundity-at-age coefficient Fecundity-at-age exponent Steepness of fertilization function (f) Steepness of spawner-recruit function Unfished recruitment were used to describe K and ag, and to derive the age at 50:50 sex ratio by substituting K and L50 into the von Bertalanffy model and solving for a ( ap=2.3/M ). Remaining parameters were set to values or ranges considered reasonable (Table 1). Note that results will be independent of Ep v,, and L , because these parameters are merely scalars. Biological reference points (BRPs) This study focused on four BRPs: maximum sustainable yield ( MSY ) and the associated fishing mortality rate (F msy )> spawning biomass (SMSY), and spawning poten- tial ratio ( SPRmsy ), defined as fertilized eggs per recruit in relation to that at the unfished level. True values of BRPs were computed numerically from the simulation model by maximizing equilibrium yield computed over a range of F at intervals of 0.01. For each F, equilibrium yield (YF ) was calculated from the Baranov catch equa- tion (Baranov, 1918) YF^^Nawa(l-e-Za). (8) a a The MSY was defined as maximum YF, FMSY as the F resulting in MSY, and SPRmsy as the corresponding spawning potential ratio (SPR). Unlike those three refer- ence points, the value of SMSY is specific to the measure of spawning biomass (f, m, and b) and was therefore computed as such, &MSY ~ 'y, Nq ( 1 Pa)gaWa a = (9) a ®MSY ~ 'y',Nqgawa a where Na = the equilibrium number at age at MSY. Although fertilized eggs, rather than spawning bio- mass, determined recruitment in the simulation model, values of Sj^gY were computed because of the key role that plays in determining whether a stock is over- fished. Equation 9 provided values in units comparable to estimates from the assessment model, where spawn- ing biomass did determine recruitment. Assessment model and estimation of biological reference points The assessment model was structurally identical to the simulation model with the single exception that recruits were computed from a measure of spawning biomass (mature females, males, or both), rather than from fer- tilized eggs. This difference represents a simplifying assumption common to almost all assessment models. Its inclusion allowed examination of how that assumption affects estimates of BRPs and identification of a robust measure of spawning biomass. In the assessment model, recruitment (R) was com- puted from spawning biomass (S=Sf, S'n, or Sb ) by using the same functional form as Equation 7, 16 Fishery Bulletin 106(1 ) R(S) = 4 hR0S R00(l-h) + (5h -1)S ’ (10) where R0 = unfished recruitment; li = steepness; and = BRP'-BRP BRP S\ MSY S\ ■’MSY MSY (11) Brooks et al. : Stock assessment of proiogynous fish 17 Table 3 Summary statistics of relative error (RE) in biological reference points (BRPs) estimated by each measure of spawning biomass. BRPs are maximum sustainable yield ( MSY ) and the corresponding fishing mortality rate (F M SY), spawning biomass (Sj^gy6), and spawning potential ratio ( SPRmsy )■ Statistics are 25th quantile, 50th quantile (median), 75th quantile, distance covered by interquartile range (IQD), proportion of model runs with relative error greater than zero (RE>0), mean, and standard deviation (SD). Bold font designates for each BRP the median error closest to zero, mean error closest to zero, proportion of positive RE closest to 0.5, smallest IQD, and smallest SD. BRP Spawning biomass 25th quantile 50th quantile 75th quantile IQD RE>0 Mean SD MSY female 0.02 0.09 0.25 0.23 0.99 0.19 0.25 male -0.34 -0.23 -0.14 0.20 0.05 -0.24 0.14 both -0.15 -0.07 0.01 0.16 0.27 -0.05 0.16 F MSY female 0.06 0.26 0.68 0.62 0.82 0.50 0.63 male -0.50 -0.37 -0.25 0.25 0.02 -0.36 0.19 both -0.22 -0.10 0.11 0.33 0.3 -0.01 0.33 c f ° MSY female -0.11 -0.05 -0.01 0.10 0.12 -0.07 0.07 Qtn ° MSY male 0.15 0.33 0.66 0.51 0.94 0.54 1.05 qb ° MSY both -0.09 0.01 0.08 0.17 0.54 0.01 0.12 SPRmsy female 0.01 0.06 0.18 0.17 0.93 0.14 0.19 male -0.24 -0.13 -0.07 0.17 0.06 -0.16 0.13 both -0.14 -0.08 -0.03 0.11 0.17 -0.08 0.11 where iE\fm,b) indicates female, male, or both, and BRP represents MSY, FMSY, or SPRmsy. When interpreting relative error, one should be aware that RE has no upper bound but has a lower bound of -1 because the BRPs and estimates are always non- negative. The distribution of relative errors was used to evaluate estimated reference points and thus to provide a general picture of which measure of spawning biomass is most robust. Analysis of variance (ANOVA) of relative errors was conducted as a form of sensitivity analysis. Factors that explained a significant proportion of total varia- tion represent biological or fishery parameters to which estimates were sensitive. Factors found to be important were then examined in greater detail. Results Primary analysis— model misspecification Aggregated across model runs, variability in estimation error, as indicated by distance covered by interquartile ranges and standard deviations of relative errors, was similar among the three measures of spawning biomass (Table 3). Two exceptions occurred: variability was relatively large when FMSY was computed from females only (Sf) and when SMSY was computed from males only (Sm). Estimates of BRPs were closest to the true values (from simulations) when the assessment model counted both males and females ( Sb ), as indicated by mean and median relative error near zero (Table 3). The assess- ment model based on females only tended to overesti- mate Fmsy, MSY, and SPRmsy, and it tended to under- estimate S^sy slightly. The assessment model based on males only showed the opposite pattern; more than 90% of relative errors in F^SY, MSYm, and SPR'^sy were negative, and more than 90% in S'^SY were posi- tive. Relative error in S'^SY could be quite large when fertilization rates were independent of male availability (x=l). In those cases, males could be almost completely removed from the simulation model without detriment to the population’s persistence, but not from the as- sessment model based on males only. Consequently, the computation of relative error of S^sy- (Eq. 11) included a denominator that approached zero, which magnified the relative error to values much greater than one. The interquartile range of relative error from S f and from S'n did not include the value of zero for any reference point, where a relative error of zero would correspond to a perfect estimate (Table 3). These relative errors, with opposite signs, were mediated when both sexes ( Sh ) defined spawning biomass in the assessment model. For all measures of spawning biomass, the steep- ness of the fertilization function (k) explained more of the variation in estimated BRPs than any other model factor (Table 4). The slope of sex transition (/!p) and steepness of the spawner-recruit function (h) explained much of the remaining variation. The remaining factors explained very little. The residual or unexplained error (Table 4) is attributable to interaction terms, which were not included in the ANOVA. Relative errors of estimated BRPs were further exam- ined by levels of k and fip (Fig. 2, A and B). These two parameters were chosen for related reasons; because es- 18 Fishery Bulletin 106(1) Table 4 Sensitivity of relative errors in estimated biological reference points to each model factor in the primary analysis, where the assessment model did not account for dynamics of fertilization. For each reference point ( FMSY , S^Y, MSY , and SPRmsy), the measure of spawning biomass (female, male, or both) with the smallest total model error (total SS) demonstrated the least vari- ability (values in italics). Table cells give the proportion of total SS explained by each factor. Values >0.1 are indicated by bold font and values <0.01, by dashes. The term “Residual” is variation explained by all possible interaction terms. Factors (model parameters) are defined in Table 1. Factor Fmsy of °MSY Female ora ° MSY Male qb ^ MSY Both MSY SPRmsy Female Male Both Female Male Both Female Male Both M 0.03 — 0.02 0.08 — 0.02 0.03 — — K 0.42 0.60 0.56 0.23 0.17 0.46 0.49 0.47 0.64 0.33 0.32 0.54 h — 0.05 0.07 0.07 0.02 0.05 0.09 0.28 0.08 0.23 0.41 0.06 PP 0.16 0.02 0.08 0.16 0.04 0.10 0.11 0.03 0.02 0.08 — — Pg — — — — — — — — — — — — Cg 0.03 0.03 — — 0.02 — — — — — — 0.02 Cs 0.05 0.09 — — 0.03 — — 0.04 - — 0.02 0.02 Residual 0.31 0.22 0.27 0.45 0.72 0.36 0.27 0.15 0.23 0.34 0.23 0.36 Total SS 4624 441 1283 57 12,894 179 728 240 307 423 183 144 timates were sensitive to them (Table 4) and because of their influences on the dynamics of fertilization (Fig. 1). Examining relative error by steepness of fertilization revealed that the most appropriate measure of spawn- ing biomass depended on the level of k. If male deple- tion had little effect on fertilization success (k in the range 0. 8-1.0), the conventional measure, Sf, produced estimates with the least error. However, as fertilization became more limited by male depletion (0.2 < x<0.8), er- ror in estimates from Sf became increasingly more vari- able and further from the true values. At intermediate values of k (~0. 4-0.7), Sb produced the best estimates. Only for the most limiting values of k (0.2, 0.3) did Sm appear to be appropriate. The influence of [3p on fertilization success was per- haps more subtle than that of k. A shallower slope of sex transition (smaller /) ) provided a broader range of age classes where both males and females were pres- ent. This decreased the propensity for fishing-induced male depletion, thereby allowing sex ratio to remain in the range where fertilization rates were relatively high. Conversely, if sex transition occurred across only a few ages (large f} ), disproportionate fishing on males was more likely. The tendency for the depletion of males with a steeper slope of sex transition explains why the assessment model based on S1 performed progressively worse as fi increased (Fig. 2, A and B). In general, our examination of relative error by slope of sex transition revealed that Sb provided the best estimates. A consistent pattern in relative errors was that BRPs based on Sf had the opposite sign from those based on Sm, and in most cases (fc>0.4), from those based on Sb as well (Table 3, Fig. 2, A and B). Specifically, SMSY tended to be underestimated by Sf and overestimated by Sb, and the other three reference points ( MSY , FMSY, and SPRmsy ) tended to show the reverse. This result indicates that, in most cases, estimates from Sf and from Sb could be used to bound uncertainty. Secondary analysis— additional misspecifications When the true age at 50:50 sex ratio (a ) was younger than the age used in the assessment model, Sf provided the best estimates of BRPs; when the true age used in the simulation model was older than the age used in the assessment model, Sb provided the best estimates (Fig. 3). When it was assumed incorrectly with the assessment model that fecundity increased linearly with weight, whether too quickly or too slowly, Sb generally provided the best results (Fig. 3). As was seen in the primary analysis, resilience of fertilization to male depletion (k) explained the most variation in relative errors of estimated BRPs, followed closely by the parameter (xp) defining misspecifica- tion in the age at 50:50 sex ratio (Table 5). Steepness of the spawner-recruit function (h) and slope of sex transition (j3 ) also explained some variation. Neither natural mortality (M) nor the parameter (/-) defining misspecification of the fecundity exponent explained much variation. Discussion We used simulations to investigate the performance of three measures of spawning biomass — females only (Sf), males only (Sm), and both sexes combined ( Sb ) — for their ability to estimate BRPs. Performance was quan- tified in terms of relative errors, which were computed across many sets of values of biological and fishery parameters. In the primary analysis, with misspecifi- cation in the spawner-recruit relationship only, an Brooks et al.: Stock assessment of protogynous fish 19 5 li => v7 N 1 2.0 - 2.0 - 1.5 - 1.5 - 1.0 - 1.0 - 0.5 - 0.5- 0.0 - 1 A 1 1 y1 y1 v- -0.5 - -0.5- 1.0 - -1.0- 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 0.2 04 0.8 1 6 0.4 0.8 1.6 Sloped y 0.5 0.6 0.7 Sleepness (k) Figure 2 Box-percentile plots of relative error (RE) in estimates of biological reference points, shown across levels of steepness in fertilization (k) and slope of sex transition (/} ). Rows corre- spond to the measure of spawningjoiomass (female, male, or both) used in the assessment model. (A) RE in F, MSY (B) RE in SMSY . Values of SMSY were calculated separately for each measure of spawning biomass as in Equation 9. Width at each percentile is proportional to the percent of observations more extreme than that percentile. The 25th, 50th, and 75th percentiles are indicated by horizontal lines within each box-percentile plot. Distributions of RE in MSY and SPRmsy were qualitatively similar to FMSY, but were less variable (typically ±0.5). 20 Fishery Bulletin 106(1 ) 1.0 ■SCO LU CC 0.0 CC <=0. CO HI tr 1.0 0.5 0.0 -0.5 female male both 1.0 - 0.5- vT1 1.0 - 0.5- i.o - 0.5 - . A A 'V v -0.5- -0.5- 0.0 -0.5 - ^ T V' _ -1.0- -1.0- -1.0 female male female male both female male both 0.5 0.0 -0.5 female male both - i.o- i.o- i.o - 1 0.5- A 1 A 0.5 - A . A 0.5 * A A A -0.5 ~| -0.5- • -0.5 - _ -i.o - -i.o- -1.0 female male both Sex transition younger female male both Sex transition older female male both Fecundity lower female male both Fecundity higher Figure 3 Box-percentile plots of relative error (RE) in estimates of biological reference points from each mea- sure of spawning biomass (female, male, and both), computed in secondary analyses, where an incor- rect value of age at 50:50 sex ratio was assumed in the assessment model or where fecundity was incorrectly assumed to scale linearly with weight. Values of SMSY were calculated separately for each measure of spawning biomass as in Equation 9. The first column of panels corresponds to sex transi- tion occurring at a younger age than that assumed in the assessment (xp = 0.75), the second column to sex transition occurring at an older age than that assumed in the assessment (% =1.25), the third column to fecundity at age being lower than that assumed in the assessment (X/=0-75), and the fourth column to fecundity at age being higher than that assumed in the assessment (x^l-25). Width at each percentile is proportional to the percent of observations more extreme than that percentile. The 25th, 50th, and 75th percentiles are indicated by horizontal lines within each box-percentile plot. assessment model using spawning biomass of both sexes generally provided the best results. When we incor- porated additional misspecifications, the assessment model based on both sexes still performed best, with the exception of cases where the age of sex transition in the assessment model was biased towards an older age. Such bias could occur if sex change is adaptive (i.e., if fish alter the timing of sex transition). However, if the age of sex transition is derived from an exploited population, we would expect an estimate used in the assessment to already reflect any adaptation, and thus it seems more likely that any bias in the estimate would be towards a younger age. Of all the parameters in the factorial design, re- silience of fertilization to male depletion, quantified by k, explains the most variation in relative error of estimates. When x>0.8, an assessment model based on females only provides the best estimates of BRPs. This result is logical, because for the largest values of k, the proportion of males can be driven quite low before fer- tilization is limited, and therefore, the number of fertil- ized eggs will be exactly (x=l) or approximately ( k = 0 . 8 or 0.9) proportional to S? (given that the exponents of weight at age and fecundity at age are equal). A value of K— 1 is a limiting case because it implies fertilization can occur even in the absence of males (xF=0). When k is in the range of about 0.4-0. 7, an assessment model based on both sexes provides the best results. For these levels of k, fertilization rates decline moderately with depletion of males — an effect that is captured by the use of Sb. Only at the most limiting values of k (0.2, 0.3), where fertilization rates decline dramatically with depletion of males, did an assessment model based on males provide the best estimates. Brooks et al.: Stock assessment of protogynous fish 21 Table 5 Sensitivity of relative errors in estimated biological reference points to each model factor in secondary analysis, where an incor- rect value of age at 50:50 sex ratio was assumed in the assessment model or where fecundity was incorrectly assumed to scale lin- early with weight. For each reference point (FMSY, S{^Y , MSY, and SPRmsy ), the measure of spawning biomass (female, male, or both) with the smallest total model error (total SS) demonstrated the least variability (values in italics). Table cells give propor- tion of total SS explained by each factor. Values >0.1 are indicated by bold font and values <0.01, by dashes. The term “Residual” is variation explained by all possible interaction terms. Factors (model parameters) are defined in Table 2. Factor fmsy qf °MSY Female ° MSY Male qb ° MSY Both MSY SPRmsy Female Male Both Female Male Both Female Male Both M 0.02 — — — — — 0.02 — K 0.38 0.45 0.41 — 0.19 0.32 0.37 0.38 0.45 0.29 0.26 0.39 h — 0.05 0.07 0.02 0.02 0.12 0.06 0.21 0.05 0.16 0.39 0.08 pP 0.13 0.02 0.05 — 0.05 0.05 0.08 0.03 0.02 0.05 — — xp 0.20 0.24 0.22 0.89 0.34 0.14 0.16 0.17 0.20 0.13 0.12 0.18 Xe — — — — 0.02 — — — — — — Residual 0.25 0.22 0.24 0.08 0.39 0.35 0.30 0.20 0.27 0.34 0.22 0.33 Total SS 1044 148 419 226 2869 64 248 78 116 112 55 48 Sensitivity of results to k may indicate that esti- mates of fertilization success, if obtainable, would be quite valuable. Although k itself may be difficult to es- timate directly, fertilization success could be assessed qualitatively if it shifts, for example from high to low, with a change in sex ratio. Such information would make it possible to infer a likely range for steepness of the fertilization function, and hence, to select the measure of spawning biomass most appropriate for that range. In addition to influencing assessment error, k in- fluences the values of BRPs themselves, and lower k results in higher SMSY and lower FMSY and MSY. Com- paring the BRPs of these simulated protogynous stocks with those of gonochoristic equivalents, we found that, on average, protogynous stocks could support higher F 'msy and MSY when k> 0.5. This finding resulted from the condition that if age structures are equivalent, protogynous stocks are not inherently more vulner- able to exploitation than gonochoristic stocks, at least over moderate ranges of fishing mortality (Bannerot et ah, 1987). It indicates that protogynous stocks are not inherently more vulnerable to exploitation than gonochoristic stocks, at least over moderate ranges of fishing mortality. We caution, however, that higher ^ msy does not imply more resilience to all levels of F. If fertilization rate depends on sex ratio, some level of F>Fmsy is still likely to be more detrimental to a protogynous stock, where that level would depend on characteristics of the stock in question. For example, if sex transition is rapid, and occurs across only a few ages, fishing could more readily deplete males, leading to fertilization failure and thus recruitment failure. Reproductive behavior could also affect a stock’s vul- nerability to exploitation. Spawning aggregations can make a species easier to target but probably better able to adapt to changes in sex ratios. Pair spawners, on the other hand, may be less easy to target, but more susceptible to effects of male depletion. Without any information on fertilization rates, a likely range of k could be postulated from evolutionary con- siderations. We expect that nature would select against values of k near its limits (0.2 and 1.0). At the lower bound of k = 0.2, any decline in the proportion of males would lead to a relatively steep reduction in fertilization success. Individual fitness could be increased by greater sperm production per male, thereby increasing fertiliza- tion success and driving k above 0.2. However, greater sperm production would likely be associated with an energetic cost. Thus, a tradeoff should exist between energy allocated toward sperm production versus other functions, such as somatic maintenance, foraging, or reproductive behavior (Alonzo and Warner, 2000; Scag- giante et al., 2005). The tradeoff may be worth the cost, but only to the extent that an increase in fertilization success improves fitness. At the upper range of k, the marginal gains in fertilization success are only real- ized if males are extremely depleted (i.e., as xF^0 in Fig. 1). Furthermore, the value of k=1.0 implies that a single male can fertilize the eggs of every female in the population, which is obviously not realistic. We therefore hypothesize that moderate values of k should be most prevalent. Theoretical predictions, and several field experiments, indicate that fertilization is less than 100% and may decline as less sperm is released per spawning event (Petersen et al., 1992; Petersen and Warner, 2002). Moderate values of k correspond to the range where BRPs are best estimated from spawning biomass of both sexes, and we therefore recommend a default choice of Sb when the degree of sperm limitation is unknown. The direction and degree of relative error indicate that Sb would produce nearly perfect estimates of SMSY and risk-averse estimates of FMSY, and only a 22 Fishery Bulletin 106(1) small loss in potential MSY (negative relative error was slight). Results of this study are insensitive to the assump- tion that fecundity is linearly related to body weight, most likely because there are few females remaining at ages where curves of fecundity at age and weight at age would diverge if £2*3. This finding provides support for using spawning biomass as a proxy for total egg production, which is reassuring given that this assumption is commonplace in assessments. How- ever, it does not support the conventional proxy (SO unless fertilization rates are nearly constant over a wide range of sex ratios. Furthermore, this finding does not address whether total egg production itself represents reproductive potential adequately. As dis- cussed by Murawski et al. (2001), total egg production does not include potentially important influences such as spawning experience or effects of maternal age and size on offspring quality. For simplicity, the simulations considered only knife- edge selectivity. In some fisheries, selectivity is dome shaped, as a result of regulations (e.g., slot limits), gear type (e.g., traps), or migration patterns (e.g., if larger fish leave the fishing grounds). In protogynous fish, dome-shaped selectivity would reduce fishing pressure mainly on males. If enough fish can survive the ages of full exploitation, dome-shaped selectivity could allow the proportion of males to remain sufficiently high to avoid severe decline in fertilization success. This effect should maintain sex ratio in the range where Sf and Sb would perform comparatively well. Indeed, this expecta- tion was confirmed by additional simulations where we repeated our primary analysis but with dome-shaped rather than knife-edge selectivity. Although this simulation study focuses on protogy- nous fish, we expect the results to hold for any stock that experiences preferential fishing on males. This may occur in gonochoristic stocks, for example, if sexually dimorphic growth or spatial segregation renders males more vulnerable to fishing gear. This investigation was deterministic by design, so that error from model misspecification could be isolated. A useful extension would be to include other sources of er- ror— observation, process, or both. Data sets that incor- porate these additional sources could be generated with the simulation model, and then fitted with the assess- ment model. This type of approach would make it pos- sible to evaluate the effect of additional error sources on estimates of key population parameters (for example, un- fished recruitment [f?0] or steepness [, h ] in the spawner- recruit relationship) and on management advice. Although Sb performs best in general, no measure of spawning biomass is best in all cases. One consistent finding is that the relative errors of Sf and Sb tend to have opposite signs over the range of k that we consider probable; therefore, the use of St and Sb in assessments should bound uncertainty in estimates of BRPs. This pattern of relative errors tending to have opposite signs occurred because Sf never accounts for reduction in fertilization success and Sb always does. As a result, St tends to overestimate the ability of a stock to support exploitation, and Sb tends to provide more conservative reference points. This consistent pattern in the relative errors of Sf and Sb indicates that error in reference points could be re- duced by creating a measure of spawning biomass that counts both sexes, but with a heavier weight on females (the measure Sb counted both sexes equally). Alterna- tively, error could be reduced by combining estimates through model averaging (e.g., Brodziak and Legault, 2005). Either way, estimates from Sf and Sb could be used to bound uncertainty in biological reference points for managing protogynous fish. Acknowledgments This work was supported by Marine Fisheries Initia- tive Program (MARFIN) grant 04MFIH10. The authors are grateful for comments from G. Fitzhugh, C. Porch, J. Powers, M. Prager, G. Scott, A. Stephens, and E. Williams. Literature cited Allsop, D. J., and S. A. West. 2004. Sex-ratio evolution in sex changing animals. Evo- lution 58:1019-1027. Alonzo, S. H., and M. Mangel. 2004. The effects of size-selective fisheries on the stock dynamics of and sperm limitation in sex-changing fish. Fish. Bull. 102:1-13. 2005. Sex-change rules, stock dynamics, and the perfor- mance of spawning-per-recruit measures in protogynous stocks. Fish. Bull. 103:229-245. Alonzo, S. H., and R. R. Warner. 2000. Allocation to mate guarding or increased sperm production in a Mediterranean wrasse. Am. Nat. 156:266-275. Armsworth, P. R. 2001. Effects of fishing on a protogynous hermaphro- dite. Can. J. Fish. Aquat. Sci. 58:568-578. Bannerot, S., W. W. Fox, Jr., and J. E. Powers. 1987. Reproductive strategies and the management of snappers and groupers in the Gulf of Mexico and the Caribbean. In Tropical snappers and groupers: biology and fisheries management (J. J. Polovina, and S. Ralston, eds.), p. 561-603. Westview Press, Boulder, CO. Baranov, F. I. 1918. On the question of the biological basis of fisheries. Nauchn. Issled. Ikhtiologicheskii Inst. Izv. 1:81-128. [Translated from Russian by W. E. Ricker.] Barot, S., M. Heino, L. O'Brien, and U. Dieckmann. 2004. Long-term trend in the maturation reaction norm of two cod stocks. Ecol. Appl. 14:1257—1271. Beverton, R. J. H., and S. J. Holt. 1957. On the dynamics of exploited fish populations, 533 p. Facsimile reprint (1993). Chapman and Hall, London. Brodziak, J., and C. M. Legault. 2005. Model averaging to estimate rebuilding targets for Brooks et al. : Stock assessment of protogynous fish 23 overfished stocks. Can. J. Fish. Aquat. Sci. 62:544- 562. 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, M. R., C. W. Waltz, W. A. Roumillat, and D. L. Stubbs. 1987. Contribution to the life history and reproductive biology of gag, Mycteroperca microlepis (Serranidae) in the South Atlantic Bight. Fish. Bull. 85: 648-653. FAO (Food and Agriculture Organization) Fishery Resources Division. 1999. Indicators for sustainable development of marine capture fisheries. FAO Technical Guidelines for Respon- sible Fisheries. No. 8, 68 p. FAO, Rome. Gardner, A., D. J. Allsop, E. L. Charnov, and S. A. West. 2005. A dimensionless invariant for relative size at sex change in animals: explanation and implications. Am. Nat. 165:551-566. Goodyear, C. P. 1980. Compensation in fish populations. In Biological monitoring of fish (C. H. Hocum, and J. R. Stauffer Jr, eds.), p. 253-280. Lexington Books, D. C. Heath and Co., Lexington, MA. Harris, J. P, 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. Heppell, S. S., S. A. Heppell, F. C. Coleman, and C. C. Koenig. 2006. Models to compare management options for pro- togynous fish. Ecol. Appl. 16:238-249. Huntsman, G. R., and W. E. Schaaf. 1994. Simulation of the impact of fishing on reproduc- tion of a protogynous grouper, the graysby. N. Am. J. Fish. Manag. 14:41-52. Mace, P. M., and I. J. Doonan. 1988. A generalized bioeconomic simulation model for fish population dynamics. New Zealand Fishery Assess- ment, Res. Doc. 88/4, 51 p. McGovern, J. C., D. M. Wyanski, O. Pashuk, C. S. Manooch III, and G. R. Sedberry. 1998. Changes in the sex ratio and size at maturity of gag, Mycteroperca microlepis, from the Atlantic coast of the southeastern United States during 1976-1995. Fish. Bull. 96:797-807. Murawski, S. A., P. J. Rago, and E. A. Trippel. 2001. Impacts of demographic variation in spawn- ing characteristics on reference points for fishery management. ICES J. Mar. Sci. 58:1002-1014. Petersen, C. W., and R. R. Warner. 2002. The ecological context of reproductive behavior. In Coral reef fishes: dynamics and diversity in a com- plex ecosystem (P. F. Sale, ed.), p. 103-118. Academic Press, San Diego, CA. Petersen, C. W., R. R. Warner, S. Cohen, H. C. Hess, and A. T. Sewell. 1992. Variable pelagic fertilization success: Implications for mate choice and spatial patterns of mating. Ecology 73:391-401. Punt, A. E., P. A. Garratt, and A. Govender. 1993. On an approach for applying per-recruit methods to a protogynous hermaphrodite, with an illustration for the slinger Chrisoblephus puniceus (Pisces: Sparidae). S. Afr. J. Mar. Sci. 13:109-119. Rothschild, B. J., and M. J. Fogarty. 1989. Spawning-stock biomass: A source of error in recruitment/stock relationships and management advice. J. Cons. Int. Explor. Mer 45:131-135. Scaggiante, M., M. B. Rasotto, C. Romualdi, and A. Pilastro. 2005. Territorial male gobies respond aggressively to sneakers but do not adjust their sperm expenditure. Behav. Ecol. 16: 1001-1007. Shepherd, G. R., and J. S. Idoine. 1993. Length-based analyses of yield and spawning bio- mass per recruit for black sea bass Centropristis striata, a protogynous hermaphrodite. Fish. Bull. 91:328-337. Vaughan, D. S., M. R. Collins, and D. J. Schmidt. 1995. Population characteristics of the black sea bass Centropristis striata from the Southeastern. U.S. Bull. Mar. Sci. 56:250-267. Vaughan, D. S., G. R. Huntsman, C. S. Manooch III, F. C. Rohde, and G. F. Ulrich. 1992. Population characteristics of the red porgy Pagrus pagrus off the Carolinas. Bull. Mar. Sci. 50:1-20. Vaughan, D. S., and M. H. Prager. 2001. Severe decline in abundance of the red porgy ( Pagrus pagrus) population off the southeastern LTnited States. Fish. Bull. 100:351-375. von Bertalanffy, L. 1938. A quantitative theory of organic growth. Hum. Biol. 10:181-213. 24 Abstract — Six years of bottom- trawl survey data, including over 6000 trawls covering over 200 km2 of bottom area throughout Alaska’s subarctic marine waters, were ana- lyzed for patterns in species richness, diversity, density, and distribution of skates. The Bering Sea continental shelf and slope, Aleutian Islands, and Gulf of Alaska regions were stratified by geographic subregion and depth. Species richness and relative density of skates increased with depth to the shelf break in all regions. The Bering Sea shelf was dominated by the Alaska skate ( Bathyraja parmifera), but species richness and diversity were low. On the Bering Sea slope, richness and diversity were higher in the shallow stratum, and relative density appeared higher in subre- gions dominated by canyons. In the Aleutian Islands and Gulf of Alaska, species richness and relative density were generally highest in the deepest depth strata. The data and distribu- tion maps presented here are based on species-level data collected throughout the marine waters of Alaska, and this article represents the most compre- hensive summary of the skate fauna of the region published to date. Manuscript submitted 16 May 2007. Manuscipt accepted 1 August 2007. Fish. Bull. 106:24-39(2008). The views and opinions expressed or implied in this article are those of the author and do not necessarily reflect the position of the National Marine Fisheries Service, NOAA. Emerging patterns of species richness, diversity, population density, and distribution in the skates (Rajidae) of Alaska Duane E. Stevenson (contact author)1 James W. Orr1 Gerald R. Hoff1 John D. McEachran2 E-mail address for D. E. Stevenson: duane.stevenson@noaa.gov 1 National Oceanic and Atmospheric Administration National Marine Fisheries Service, Alaska Fisheries Science Center Resource Assessment and Conservation Engineering Division 7600 Sand Point Way NE Seattle, Washington 98115 2 Department of Wildlife and Fisheries Sciences Texas A&M University College Station, Texas 77843 Patterns of species richness, diversity, density, and distribution for the spe- cies of skates inhabiting the North Pacific Ocean and Bering Sea are still largely unknown. Earlier stud- ies have been limited because of prob- lems with identification of skates in the field and, to some degree, in the laboratory. Summarizing trawl survey data for commonly encountered spe- cies, Allen and Smith (1988) reported serious problems in the fisheries and survey data reported for even the most common skates throughout Alaska and the eastern North Pacific Ocean because of widespread problems with field identification of skates. Simi- larly, in the only previously published analyses of skate abundance and dis- tribution data for Alaska, Teshima and Wilderbuer (1990) and Raschi et al. (1994) treated their data in the aggregate at the family level because of difficulty with identification of spe- cies. In contrast, Japanese and Rus- sian authors have published several studies including general species-level information on the skate fauna of the western North Pacific Ocean and Sea of Okhotsk (Dudnik and Dolganov, 1992; Nakaya and Shirai, 1992; Dol- ganov, 1999, 2001). Skates present difficulties for iden- tification because they are a mor- phologically conservative group of fishes, and although they represent a large proportion of the diversity of elasmobranchs worldwide, external morphological differences among spe- cies (or even genera) are often subtle. Moreover, the extent of morphologi- cal variation in many species is poor- ly known despite earlier taxonomic work (Ishiyama and Ishihara, 1977; Ishihara and Ishiyama, 1985, 1986), and although molecular methods have shown promise for species identifica- tion in the laboratory (Tinti et al., 2003; Bremer et al., 2005; Spies et al., 2006), skates are often difficult to identify in the field. In Alaska wa- ters, namely the eastern Bering Sea, Aleutian Islands, and Gulf of Alas- ka, this has been particularly true as Allen and Smith (1988), Teshima and Wilderbuer (1990), and Raschi et al. (1994) have noted. More recently, Mecklenburg et al. (2002) stated that the poorly understood taxonomic rela- tionships of the skates in this region complicate the determination of spe- cies distributions. These challenges are compounded by the fact that skates are generally large fishes and are, thus, difficult to collect, preserve, and curate. Therefore, they are poorly represented in museum collections and difficult to study in the laboratory. Because of identification difficulties and a relative lack of commercial im- Stevenson et al.: Patterns of species richness, diversity, population density, and distribution in the skates of Alaska 25 portance, species-specific data on skate populations are often not available, and catch statistics are commonly recorded only at the aggregate (genus or family) level. This lack of data is a concern in that skates may be particularly vulnerable to fishing pressure, even if they are only encountered as bycatch, because of their large size, relatively long life expectancy, and low fecundity, and are considered highly vulnerable to extinction or extirpation due to overfishing or habitat disturbances (Casey and Myers, 1998; Stevens et ah, 2000; Dulvy and Reynolds, 2002). Moreover, apparent stability or increases in aggregate skate catches may mask de- clines in some components of those species aggregates, particularly among larger species (Dulvy et al., 2000). Therefore, species-level management, which can only be achieved through accurate identification, reporting, and monitoring, is essential for maintaining viable skate populations (Stevens et al., 2000). Although skate populations in the eastern North Pa- cific Ocean and Bering Sea have been inadequately studied and inaccurately represented because of ques- tionable field identifications, recent research efforts are improving the resolution and consistency with which skates can be identified by field survey personnel. The taxonomic works of Ishihara and Ishiyama (1985, 1986), Dolganov (1985), and Stevenson et al. (2004) have helped to clarify the taxonomy of North Pacific skates. In addition, range extensions of species previously un- known from Alaskan waters (Stevenson and Orr, 2005), the ongoing development of a comprehensive field guide to the skates of the region (Stevenson et al., 2007), and the establishment of a voucher collection process that allows for laboratory verification or reidentification of significant specimens, have greatly improved field iden- tifications on Alaska Fisheries Science Center (AFSC) resource assessment surveys. Because of advances in the knowledge of the skates of the North Pacific Ocean and Bering Sea, species-level skate identifications made by AFSC personnel begin- ning with the 1999 survey year are reliable and consis- tent. The purpose of this article is to describe species richness, diversity, relative density, and distribution of skates throughout Alaskan waters, based on data from six years of groundfish bottom-trawl surveys in the Bering Sea, Aleutian Islands, and Gulf of Alaska from 1999 through 2004. Materials and methods Specimens were collected aboard a variety of commercial fishing vessels chartered by the AFSC Resource Assess- ment and Conservation Engineering (RACE) Division from May through August in the eastern Bering Sea, Aleutian Islands, and Gulf of Alaska, between 1999 and 2004 (Fig. 1). Survey gear and methods differed sub- stantially among the four resource assessment surveys (Bering Sea shelf, Bering Sea slope, Aleutian Islands, and Gulf of Alaska), and bottom habitat types differ among regions, resulting in differences in the catch- ability of skates. The effects of these differences on estimates of skate species richness, diversity, density, and distribution remain largely unknown; therefore the eastern Bering Sea shelf, Bering Sea slope, Aleutian Islands, and Gulf of Alaska were treated separately in this study. The eastern Bering Sea shelf survey has been con- ducted annually in approximately its present form since 1982, and data from survey hauls during the years 1999-2004 were used in this study. This survey covered the eastern Bering Sea shelf from the Alaska Penin- sula north beyond St. Matthew Island to approximately 62°40'N, from 20 m to 200 m depth, and was conducted with an 83-112 Eastern otter trawl. Hauls were made at previously established survey stations on a 20x20 nautical mile grid, and bottom time for each haul was approximately 30 minutes. For the purposes of this study, the region was divided into three subregions of approximately equal survey effort, each including three depth ranges (<50 m, 51-100 m, and 101-200 m). Sub- region 1 comprised the southeastern part of the eastern Bering Sea shelf, extending from the Alaska Peninsula to the southeastern rim of Pribilof Canyon; subregion 2 comprised the central part of the eastern Bering Sea shelf, from the northwestern boundary of subregion 1 to the southeastern rim of Zhemchug Canyon; and subre- gion 3 comprised the northernmost portion of the survey area, bounded on the northwest by the U.S. -Russian border, with the northernmost hauls at approximately 62°40'N (Fig. 2A). For more information on the design and methods of this survey, see Stauffer (2004) and Lauth and Acuna (2007). The eastern Bering Sea slope survey was conducted in 2000, 2002, and 2004. It covered the eastern Ber- ing Sea upper continental slope (200 m to >1200 m depth) from just north of Unalaska Island north to the U.S. -Russian border, and was conducted with a Poly Nor’eastern bottom trawl equipped with mud-sweep roller gear on the footrope. Haul locations were chosen according to a stratified random sampling design, with the region divided into six subregions and five depth strata, and bottom time for each haul was approximate- ly 30 minutes. For the purposes of this study the same six sub-regions were used. Subregions 1 and 6 consisted of a broad, low-angle slope, and form the southeastern and northwestern edges of the survey area; subregions 2 and 4 consisted of Pribilof and Zhemchug Canyons, respectively; and subregions 3 and 5 were intercanyon faces characterized by a steep-angle slope (Fig. 2A). For this study, depth strata were combined into two depth ranges, representing the upper slope (200-600 m) and the lower slope (>600 m). For more information on the design and methods of this survey, see Hoff and Britt (2005). The Aleutian Islands survey was conducted in 2000, 2002, and 2004. These surveys covered the continental shelf and upper slope (to 500 m) of the entire Aleu- tian Archipelago from Unimak Pass (165°W) to Stale- mate Bank (170°30'E), and were conducted with a Poly Nor’eastern bottom trawl equipped with rubber bobbin 26 Fishery Bulletin 106(1 ) 145°E 150°E 160°E 170°E 180° 170°W 155°W 140°W 125°W 115°W 105°W Figure 1 Map of the area covered in bottom-trawl surveys conducted by the Resource Assessment and Conservation Engineering Division of the Alaska Fisheries Science Center between 1999 and 2004. Surveys of the Bering Sea extended from the Alaska Peninsula to the U.S. -Rus- sian border northwest of St. Matthew Island, surveys of the Aleutian Islands extended from Unimak Pass west to Stalemate Bank, and surveys of the Gulf of Alaska extended from Unimak Pass east to the U.S. -Canada border at Dixon Entrance. Black lines (west and east) denote U.S. -Russian and U.S. -Canadian border. Depth contour = 200 m. roller gear on the footrope. The southern part of the eastern Archipelago from the Islands of Four Moun- tains to Unimak Pass has been assessed as part of the Gulf of Alaska survey but because zoogeographically it is part of the Aleutian Islands and because survey methods are the same (Britt and Martin, 2000), it was included in our regional study of the Aleutian Islands. Haul locations were chosen on the basis of a stratified random sampling design, and the region was divided into 45 area-depth strata, and bottom time for each haul was approximately 15 minutes. For the purposes of this study the region was divided into four subregions, numbered from west to east, bounded by the major deep-water passes and gaps bisecting the archipelago. Subregion 1 comprised Stalemate Bank and the Near Islands, extending to approximately 174°30'E; subregion 2 extended east to Amchitka Pass, at the 180° line; subregion 3 extended from Amchitka Pass to Amukta Pass (approximately 171°30'W); and subregion 4 ex- tended from Amukta Pass to Unimak Pass (Fig. 2B). For this study, depth strata were combined into three depth ranges (<100 m, 101-200 m, and >200 m). For more information on the design and methodology of this survey, see Zenger (2004). The Gulf of Alaska survey was conducted in 1999, 2001, and 2003. It covered the continental shelf and upper slope (to 500 m in 2001, 700 m in 2003, and 1000 m in 1999) of the Gulf of Alaska from the southern part of the eastern Aleutian Islands beginning at the Islands of Four Mountains (included for this study in the Aleutian Islands region) to the U.S. -Canada border at Dixon Entrance, except in 2001 when the survey ended to the east at Hinchinbrook Entrance. Survey design, methods, and gear were the same as those of the Aleutian Islands survey (see Britt and Martin, 2000). For the purposes of the present study the region was divided into five subregions, numbered from west to east, corresponding approximately with National Marine Fisheries Service (NMFS) management areas. Subregion 1 extended from the tip of the Alaska Penin- sula west to 159° W; subregion 2 extended from 159° W to the west side of Kodiak Island; subregion 3 extended from the west side of Kodiak Island to Hinchinbrook Entrance (147°W); subregion 4 extended from Hinchin- brook Entrance to Cross Sound at the northern tip of the Alexander Archipelago; and subregion 5 included the Alexander Archipelago, extending to the U.S. -Can- ada border (Fig. 2C). As in the Aleutian Islands region, Stevenson et al.: Patterns of species richness, diversity, population density, and distribution in the skates of Alaska 27 depth strata were combined into three depth ranges (<100 m, 101-200 m, and >200 m). In all surveys, skate specimens were either examined and discarded at sea or fixed onboard in 10% seawa- ter-buffered formalin or frozen for later study. Many of the specimens were photographed at sea, and the majority of fixed specimens were transferred to 70% ethanol and archived at the University of Washington (UW) fish collection. In compiling a list of skates for Alaska, we followed the comprehensive guide of Meck- lenburg et al. (2002), with the following updates. We included Rathyraja mariposa (butterfly skate), described in Stevenson et al. (2004), and Amblyraja badia (rough- shoulder skate), recently documented in the Bering Sea (Stevenson and Orr, 2005). Species richness was defined as the number of skate species encountered and identified in each haul. Mean species richness values were calculated for each depth- and-subregion stratum as a simple average of the num- ber of skate species encountered in all hauls performed in that stratum, including zero values for hauls in which no skates were encountered. Because mean species rich- ness figures were heavily influenced by zero values in some regions, and because simple species richness gives no indication of evenness, we also calculated a single Shannon’s diversity index ( H ' ) for each depth-and-sub- region stratum. This index was calculated as s H' = ^[(re- / re)ln(re;- / re)], i=l where ni = the number of individuals belonging to the zth of S species in the depth-and-subregion stratum; and n - the total number of individuals captured in the depth-and-sub-region stratum (Ludwig and Reynolds, 1988). Bottom area swept was calculated for each haul by multiplying the distance fished by the mean net spread. Density was calculated as the number of individuals per km2 of bottom area swept. Means are presented ± standard error (SE). Because the distributions of both species richness and density values were heavily skewed, and had a large proportion of zero values, tests for significant differences among means were not per- formed. Species distributions represent a summary of all recorded encounters of each species during standard survey operations in successful survey hauls. Results The total amount of effort represented in this data set consisted of 6096 successful bottom trawls covering over 201 km2 of bottom area (Table 1). The Bering Sea shelf accounts for significantly more trawling effort than any of the other regions, because of differences in survey frequency (annual [Bering Sea shelf] vs. biennial [other regions]) and standard tow duration (30 minutes [Bering Boundaries of sampled subregions covered in bottom-trawl surveys from 1999 through 2004 in (A) the eastern Bering Sea shelf (underlined numbers) and slope (numbers not underlined), ( B ) the Aleutian Islands, and (C) the Gulf of Alaska. Sea shelf and slope] vs. 15 minutes [ Aleutian Islands and Gulf of Alaska]). Thirteen of the 14 described species of skates known from Alaskan waters (all but Rathyraja violacea) were encountered in bottom-trawl surveys from 1999 through 2004. Survey trawls collected 0-7 species per haul (mean=0.93 ±0.01), and at least one species of skate was collected in 62.8% of hauls. The number of 28 Fishery Bulletin 106(1 ) Table t Total number of hauls (n) and bottom area swept (km2) by subregion and depth range for bottom-trawl surveys conducted in the eastern Bering Sea (shelf and slope), Aleutian Islands, and Gulf of Alaska from 1999 through 2004 (see text for subregion boundaries). Subregion 1 2 3 4 5 6 Total Depth range (m) 72 km2 n km2 n km2 n km2 n km2 n km2 n km2 Bering Sea shelf <50 182 7.68 245 10.90 18 0.81 445 19.39 51-100 386 17.86 361 17.18 302 15.03 1049 50.07 101-200 141 7.33 147 7.11 328 16.51 616 30.95 Total 709 32.87 753 35.19 648 32.35 2110 100.40 Bering Sea slope 201-600 112 4.38 31 1.09 32 1.10 32 1.08 15 0.55 53 1.94 275 10.14 601+ 57 2.30 32 1.12 33 1.23 30 1.12 23 0.87 33 1.15 208 7.78 Total 169 6.68 63 2.21 65 2.33 62 2.20 38 1.42 86 3.09 483 17.92 Aleutian Islands <100 34 0.78 83 1.83 67 1.54 171 3.90 355 8.05 101-200 103 2.37 154 3.37 164 3.73 103 2.37 524 11.84 201+ 70 1.63 134 3.05 170 3.99 156 3.72 530 12.38 Total 207 4.78 371 8.25 401 9.26 430 9.99 1409 32.27 Gulf of Alaska <100 247 5.74 174 4.02 280 6.60 50 1.19 15 0.33 766 17.87 101-200 121 2.87 188 4.52 365 8.90 104 2.51 58 1.33 836 20.12 201 + 30 0.72 128 3.17 138 3.64 119 3.68 77 1.79 492 13.00 Total 398 9.33 490 11.71 783 19.14 273 7.38 150 3.45 2094 50.99 individual skates captured in a single haul ranged from 0 to 833, yielding aggregate density (all species combined) estimates for individual survey hauls ranging from 0 to 22,005 individuals per km2 (mean = 160.55 ±5.39). For individual species, maximum density figures ranged over several orders of magnitude, from 26.42 individuals/km2 for the rare and deepwater B. abyssicola to over 19,759 individuals/km2 for an extraordinarily large haul of B. parmifera on the northern Bering Sea slope. Bering Sea shelf The Bering Sea shelf received as much trawling effort as the other three regions combined and skates were common, but skate species richness and diversity were low. Approximately half of the effort was expended in the 51-100 m depth range, and the other half of the effort was almost equally distributed between the other two depth strata (Table 1). The southern, central, and northern subregions were sampled approximately evenly, and the depth distribution of the effort was similar in subregions 1 and 2. However, the depth distribution of the effort was significantly different in subregion 3, where the deepest depth stratum (100-200 m) was most heavily sampled, and very little sampling was con- ducted in the shallow stratum. Skates were encountered throughout the entire geographic and bathymetric range of the survey area in over 87% (1830 of 2110) of the hauls conducted in this region. Skate species richness was highest in subregion 3 and lowest in subregion 1 (1.21 vs. 0.94), increasing with depth in all three subregions (Table 2). In all three subregions, species richness was highest in the deepest strata, and the lowest species richness was encountered in the shallow stratum of sub- region 1. For all subregions combined, species richness increased significantly (PcO.0001) with depth (Fig. 3). Diversity indices were near zero in all three subregions in the shallow and middle depth strata, but were slightly higher in the deepest depth zone (Table 3). Skates were often encountered at moderate to high densities on the Bering Sea shelf. Aggregate skate den- sity on the eastern Bering Sea shelf ranged from 0 to 5103 individuals/km2, and an overall mean of 229.49 ±5.97. The largest mean density values were encoun- tered in the middle depth stratum (51-100 m) of sub- region 2 and the shallow and deep strata of subregion 3, although the variability associated with the mean in the shallow stratum of subregion 3 was extremely high. The smallest mean density was encountered in the shallow stratum of subregion 1 (Table 4). Overall mean density for the entire region was considerably lower in the shallow depth stratum than in either of the two Stevenson et al.: Patterns of species richness, diversity, population density, and distribution in the skates of Alaska 29 Table 2 Mean skate species richness (no. of species/haul, with standard error in parentheses) by subregion and depth range for bottom- trawl surveys conducted in the eastern Bering Sea (shelf and slope), Aleutian Islands, and Gulf of Alaska from 1999 through 2004. Subregion Depth range (m) 1 2 3 4 5 6 Total Bering Sea shelf <50 51-100 101-200 Total 0.51 (0.04) 0.93 (0.03) 1.52 (0.06) 0.94(0.03) 0.79 (0.03) 1.06(0.02) 1.58(0.06) 1.07(0.02) 0.72 (0.11) 0.96 (0.02) 1.47 (0.04) 1.21 (0.02) 0.67 (0.02) 0.98 (0.01) 1.51 (0.03) 1.07 (0.01) Bering Sea slope 201-600 601+ Total 2.43 (0.11) 1.91 (0.16) 2.25 (0.09) 3.16 (0.32) 3.19 (0.24) 3.17 (0.20) 2.22 (0.22) 2.94 (0.18) 2.58 (0.15) 3.59 (0.18) 2.63(0.28) 3.13 (0.17) 2.87 (0.40) 1.52 (0.20) 2.05 (0.22) 3.81 (0.16) 2.27 (0.24) 3.22 (0.16) 2.91 (0.08) 2.39 (0.09) 2.69 (0.06) Aleutian Islands <100 101-200 201+ Total 0.59 (0.10) 0.89 (0.11) 0.69 (0.12) 0.77 (0.07) 0.52 (0.06) 0.59 (0.07) 0.93(0.08) 0.70(0.04) 0.51 (0.08) 0.77 (0.06) 1.04 (0.07) 0.84(0.04) 0.38(0.04) 0.50(0.07) 0.90 (0.07) 0.60 (0.04) 0.46 (0.03) 0.69 (0.04) 0.92 (0.04) 0.72 (0.02) Gulf of Alaska <100 101-200 201+ Total 0.25(0.03) 0.27 (0.05) 0.30 (0.10) 0.26 (0.02) 0.39(0.04) 0.70 (0.06) 1.02(0.10) 0.67(0.04) 0.39 (0.04) 0.74(0.04) 0.65(0.07) 0.60 (0.03) 0.44 (0.09) 0.46 (0.06) 0.47 (0.07) 0.46 (0.04) 0.27 (0.12) 0.26(0.07) 0.40 (0.08) 0.33 (0.05) 0.34 (0.02) 0.60(0.03) 0.64 (0.04) 0.51 (0.02) Table 3 Shannon’s diversity index ( H' ) by subregion and depth range for bottom-trawl surveys conducted in the eastern Bering Sea (shelf and slope), Aleutian Islands, and Gulf of Alaska from 1999 through 2004. Subregion Depth range (m) 1 2 3 4 5 6 Total Bering Sea shelf <50 0.02 0.01 0.00 0.01 51-100 0.15 0.06 0.03 0.09 101-200 0.57 0.62 0.34 0.46 Total 0.28 0.20 0.25 0.24 Bering Sea slope 201-600 1.48 1.44 1.33 1.42 1.67 1.70 1.71 601+ 1.47 1.33 1.30 1.07 1.10 0.99 1.46 Total 1.65 1.42 1.57 1.38 1.85 1.85 1.82 Aleutian Islands <100 0.72 0.46 0.70 1.65 1.27 101-200 0.80 1.20 1.04 1.52 1.13 201+ 1.31 1.22 0.97 0.93 1.10 Total 0.94 1.32 1.18 1.33 1.33 Gulf of Alaska <100 1.01 0.87 0.93 0.52 0.17 0.91 101-200 1.43 1.53 1.32 0.90 0.90 1.42 201+ 0.79 1.51 1.35 0.68 1.56 1.44 Total 1.33 1.64 1.46 1.00 1.44 1.53 30 Fishery Bulletin 106(1 ) deeper strata, yielding a significant trend (PcO.OOOl) of increasing density with increasing depth for combined subregions (Fig. 4). Subregions 2 and 3 exhibited a higher overall mean density than subregion 1. Bathyraja parmifera (Alaska skate) was by far the most common species in this region, appearing in over 86% of hauls and with a mean density an order of magnitude larger than the next most common species, B. interrupta (Ber- ing skate) (Table 5). A total of seven skate species were encountered in this region, two of which (B. maculata and R. rhina) were encountered in only one or two hauls over the entire six-year period. Bering Sea slope The Bering Sea slope received comparatively little trawl- ing effort, but skates were common and species rich- ness and diversity were high. Slightly more effort was expended on the upper slope (201-600 m depth) than on the lower slope (>601 m). Subregion 1, in the south- ern part of the eastern Bering Sea, received the most effort and the remaining five subregions were sampled approximately evenly (Table 1). Skates were encountered throughout the entire geographic and bathymetric range of the survey region, occurring in 95% (460 of 483) of hauls conducted. Species richness was approximately 50% higher in subregions 2, 4, and 6 (canyons and northern gentle slope habitats) than in subregions 1, 3, and 5 (intercanyons and southern gentle slope habi- tats). In four of the six subregions, mean species rich- ness was higher on the upper slope than on the lower slope (Table 2), and the overall trend was for richness to decrease significantly (PcO.OOOl) with increasing depth (Fig. 3). Mean species richness was highest on the upper slope in subregion 6 and lowest on the lower slope in subregion 5. Diversity indices for the three southern subregions were similar on the upper and lower slope, whereas in the three northern subregions the diversity index was notably higher on the upper slope (Table 3). Skates were encountered at high densities on the Ber- ing Sea slope. Aggregate skate density ranged from 0 to 22,005 individuals/km2, and had an overall mean of 545.81 ±53.48. These figures are heavily influenced by one particularly large haul of B. parmifera during the 2002 survey. With this haul removed, the density range and overall mean became 0 to 5350 individuals/km2 and 501.29 ±29.69 individuals, respectively. The larg- est mean density was encountered on the upper slope in the northernmost subregion (subregion 6), although this mean was also influenced by the large haul of B. parmifera. With the aberrant haul removed from the data set, the largest mean was that of the lower slope in subregion 2 (Table 4). The smallest mean density was encountered on the lower slope in subregion 5. Subre- gions 2, 4, and 6 exhibited considerably higher overall mean density values than those of subregions 1, 3, and 5 — a pattern similar to that of species richness in this region. Overall mean density was similar for the two depth ranges, but slightly higher on the lower slope and increasing nonsignificantly (P= 0.128) with increasing depth (Fig. 4). Although B. parmifera exhibited the highest maximum density, B. aleutica produced the highest mean density value in this region, followed by B. lindbergi and B. interrupta (Table 5). A total of ten skate species were encountered in this region, although two of these (A. badia and B. abyssicola ) were encoun- tered in only two hauls. Aleutian Islands The Aleutian Islands received moderate trawling effort, and skates were inconsistently encountered. Trawl- ing effort in this region was distributed approximately evenly between the two deeper depth strata, whereas the shallow stratum received slightly less effort (Table 1). The easternmost subregion (subregion 4) was most heavily sampled, but effort progressively decreased in the more western subregions. Skates were encountered Stevenson et at: Patterns of species richness, diversity, population density, and distribution in the skates of Alaska 31 Table 4 Mean aggregate density (no. of individuals/km2, with standard error in parentheses) of all skates combined by subregion and depth range for bottom-trawl surveys conducted in the eastern Bering Sea (shelf and slope), Aleutian Islands, and Gulf of Alaska from 1999 through 2004. * = calculations for this area of the Bering Sea slope did not include the particularly large catch of B. parmifera on 29 June 2002 (see text for details). Subregion ueptn range (m) 1 2 3 4 5 6 Total Bering Sea shelf <50 51-100 101-200 Total 86.20(10.09) 220.24(16.53) 216.34(38.33) 185.06(12.25) 191.85 (14.24) 330.91 (15.70) 215.94(14.67) 263.22(9.58) 340.92(119.05) 175.03(9.74) 292.13(11.16) 238.91 (8.27) 154.67 (10.46) 245.31 (8.83) 256.60(11.24) 229.49(5.97) Bering Sea slope 201-600 311.32(43.28) 492.60(83.61) 328.23 (64.99) 670.47 (121.44) 277.00(57.20) 797.53 (134.44)* 466.14 (38.02)* 601+ 255.91 (48.04) 1046.67(171.08) 467.24(80.55) 735.72(145.61) 105.51 (18.05) 784.85(111.65) 547.59(47.09) Total 292.63(32.92) 774.03(101.66) 398.81 (52.25) 702.04(93.61) 173.20(28.21) 792.61(92.51)* 501.29(29.69) Aleutian Islands <100 101-200 201+ Total 135.65(77.68) 279.19(55.00) 69.72(30.45) 184.78(32.47) 44.54(7.89) 50.33(7.95) 99.90(14.49) 66.94(6.55) 55.32(15.28) 110.12(29.89) 219.66(32.47) 147.4(18.83) 21.51 (2.80) 45.39(8.73) 151.07(22.86) 74.24(9.06) 44.21 (8.38) 113.06(15.03) 149.39(13.72) 109.38(7.96) Gulf of Alaska <100 12.21 (1.68) 35.60(6.11) 31.92(4.10) 67.41 (26.64) 76.97(65.26) 29.60(3.04) 101-200 12.52(2.36) 43.31 (4.96) 48.66(3.61) 30.45(5.15) 13.96(4.43) 37.56(2.14) 201+ 18.08 (7.61) 66.67(7.29) 50.14(8.04) 41.87(7.17) 22.25 (4.84) 46.10(3.59) Total 12.75(1.38) 46.67(3.5) 42.94(2.66) 42.16(6.13) 24.52(7.16 ) 36.65 (1.64) Table 5 Mean and maximum density (no. of individuals/km2) by species for skates collected in bottom-trawl survey hauls performed in the eastern Bering Sea (shelf and slope), Aleutian Islands, and Gulf of Alaska from 1999 through 2004. * = not collected in this region, n = total number of hauls performed within each region. Species Bering Sea shelf (71=2110) Bering Sea slope (tj = 483) Aleutian Islands (n=1409) Gulf of Alaska (71=2094) All regions (77 = 6096) Mean Max Mean Max Mean Max Mean Max Mean Max Bathyraja parmifera 217.09 5103.83 55.00 19759.77 18.36 1012.26 1.64 201.97 84.30 19759.77 B. aleutica 0.99 575.15 214.16 3981.10 8.52 639.17 5.07 207.53 21.02 3981.10 B. maculata 0.01 19.47 43.07 2350.51 52.37 4414.28 0.12 88.94 15.56 4414.28 B. interrupta 10.95 565.55 60.59 1057.54 1.01 340.40 5.60 238.45 10.75 1057.54 B. taranetzi 0.13 41.67 31.48 1117.97 27.31 1781.36 0.02 40.11 8.86 1781.36 B. lindbergi * * 78.02 1953.68 0.09 46.22 0.04 82.24 6.22 1953.68 B. minispinosa * * 36.26 1094.62 0.11 66.08 0.02 36.02 2.90 1094.62 B. trachura * * 26.79 468.10 0.08 83.66 0.69 502.86 2.38 502.86 B. mariposa * * * * 0.34 103.78 * * 0.08 103.78 B. abyssicola * * 0.10 26.42 * * * * 0.01 26.42 Raja rhina 0.01 44.67 * * 0.10 50.81 13.83 592.97 4.78 592.97 R. binoculata 0.25 197.37 * * 1.08 178.52 9.61 986.51 3.64 986.51 Amblyraja badia * * 0.25 80.00 * * * * 0.02 80.00 All skates 229.43 5103.83 545.81 22,005.20 109.38 4609.02 36.65 986.51 160.52 22,005.20 32 Fishery Bulletin 106(1 ) 6000 -I 5000 - 4000 - 3000 - o o Bering Sea 0 2000 1000 0 0 200 400 600 800 1000 1200 CM E A 03 3 -O > TD C >, <7> c 03 Q 5000 i 4500 - 4000 - 3500 - 3000 - 0 200 400 600 Aleutian Islands 800 1000 1200 Figure 4 Relationship between overall skate density (no. of indi- viduals/km2) and depth for bottom-trawl survey hauls completed in the Bering Sea, Aleutian Islands, and Gulf of Alaska from 1999 through 2004. Solid lines represent linear regressions of estimated skate density values, dashed line in upper panel divides Bering Sea shelf records (open circles) from Bering Sea slope records (closed circles). throughout the survey area, but in only 51% (718 of 1409) of the hauls conducted in this region. Species richness was similar among the four subregions of the Aleutian Islands (Table 2) and increased with depth in all but subregion 1, the westernmost subregion. For all subre- gions combined, species richness increased significantly (P<0.0001) with increasing depth (Fig. 3). Maximum species richness was encountered in the deepest strata of subregions 2, 3, and 4 and the intermediate depth of subregion 1, whereas the minimum species richness was encountered in the shallow stratum of subregion 4. More species of skates were encountered in the Aleutian Islands (11) than in any of the other regions (vs. 10 in the Gulf of Alaska, 9 on the Bering Sea slope, and 7 on the Bering Sea shelf). Diversity indices were lowest in the shallow strata in all but subregion 4, where the shallow stratum exhibited the highest diversity and the deepest stratum exhibited the lowest (Table 3). Skate population densities were variable in the Aleu- tian Islands. Aggregate skate density ranged from 0 to 4609 individuals/km2, and had an overall mean of 109.38 ±7.96. The spatial pattern of mean density was similar to that of mean species richness and diversity, increasing with depth in all but one subregion (Table 4), although in this case it was the easternmost subregion (subregion 1) that contrasted with the other subregions. The minimum mean density was encountered in the shallow stratum of subregion 4. The maximum mean density was encountered in the middle depth stratum of subregion 1 and in the deep stratum of subregion 3, and mean density was not as uniform as mean species richness across the four subregions. Subregions 1 and 3 yielded overall mean density values considerably higher than those of subregions 2 and 4. For all subregions combined, density increased significantly (P<0.0001) with increasing depth (Fig. 4). The three most common skate species in the Aleutian Islands were B. maculata, B. taranetzi, and B. parmifera (Table 5). Bathyraja aleu- tica was also moderately common. Gulf of Alaska The Gulf of Alaska received moderate trawling effort, which was concentrated in the western and central subregions, and skates were comparatively rare in this region. Effort was similar for the shallow and middle depth strata, but the deep stratum was not as heavily sampled (Table 1), and central subregion 3 received the most effort whereas southeastern subregion 5 received by far the least. Although skates were encountered through- out the survey area, they occurred in less than 40% (828 of 2094) of the hauls conducted in the region. Mean spe- cies richness was generally higher in the central Gulf of Alaska (subregions 2, 3, and 4) than in either the western (subregion 1) or southeastern subregions (subregion 5) (Table 2). Although the overall trend in this region was for species richness to be higher at deeper depth strata, and the relationship between richness and depth for all subregions combined was significant (P=0.0047) (Fig. 3), this pattern was clearly evident only in subregion 2. Maximum species richness was encountered in the deep stratum of subregion 2, whereas the minimum species richness was encountered across all strata of subregion 1 and the two shallow strata of subregion 5. Diversity indices were not as clearly related to subregion, and only subregion 5 showed a strong trend toward increasing diversity with increasing depth (Table 3). Skate densities were generally low in the Gulf of Alaska. Aggregate skate density ranged from 0 to 986 individuals/km2, and had an overall mean of 36.65 ±1.64. The highest mean density was encountered in the shallow depth stratum of subregion 5, but the variabil- ity associated with this mean was extremely high. The lowest mean density was encountered in the shallow stratum of subregion 1 (Table 4). Overall mean density generally increased with depth (Fig. 4), although the Stevenson et at: Patterns of species richness, diversity, population density, and distribution in the skates of Alaska 33 Table 6 Depth range (m) for skate species encountered in bottom-trawl survey hauls performed in the eastern Bering Sea (shelf and slope), Aleutian Islands, and Gulf of Alaska from 1999 through 2004. * = not collected in this region. Species Bering Sea shelf Bering Sea slope Aleutian Islands Gulf of Alaska Overall Bathyraja parmifera 19-190 202-392 47-283 62-375 19-392 B. aleutica 77-173 202-1200 47-535 29-882 29-1200 B. maculata 174 206-1200 91-488 185-208 91-1200 B. interrupta 53-190 202-1017 64-369 37-566 37-1017 B. taranetzi 77-153 202-1063 82-488 64 64-1054 B. lindbergi * 342-1200 342-458 917 342-1200 B. minispinosa * 206-1200 223-368 666 206-1200 B. trachura * 221-1200 369-407 345-946 221-1200 B. mariposa * * 95-457 * 95-457 B. abyssicola * 951-1400 * * 951-1400 Raja rhina 70-139 * 87-133 24-601 24-601 R. binoculata 37-135 * 26-192 16-376 16-376 Amblyraja badia * 1508-1556 * * 1508-1556 relationship was not significant (P=0.117). Subregions 4 and 5 seemed to contradict this trend, with consider- ably higher mean density values in the shallow stratum, but the standard errors of these means indicated that catches in these subregions were highly variable. Like species richness, mean density was higher in the central subregions and lower in the western and southeastern subregions. The two Alaskan species of the genus Raja ( R . binoculata and R. rhina) were the most common skates in this region (Table 5). The only two species of Bathyraja that were relatively common in the Gulf of Alaska were B. interrupta and B. aleutica. A total of ten skate species were encountered in this region, but four of these were recorded in five or fewer hauls. Species distributions Of the 14 species of skates currently known from Alaska, 13 were encountered on the surveys included in this study, and 5 of these species were found in all four regions. Two species ( B . abyssicola and Amblyraja badia) were encountered only on the Bering Sea slope, and one {B. mariposa) was encountered only in the Aleutian Islands. Bathyraja violacea, although known from Alas- kan waters, was not encountered in these surveys. Some species of the genus Bathyraja were commonly and widely encountered in all survey regions (Table 6). Bathyraja parmifera, the most common species of skate in Alaskan waters, was found throughout all the survey regions (Fig. 5A). Bathyraja aleutica (Fig. 5B) and B. interrupta (Fig. 5C) were also widespread, both geo- graphically and bathymetrically. Although much more rarely encountered than the aforementioned species, B. trachura (Fig. 5D) was also found throughout all survey regions in deep hauls. Other species were common in some survey regions, but rarely (if ever) encountered in others. Bathyraja maculata (Fig. 6A) and B. taranetzi (Fig. 6B) were com- monly encountered on the Bering Sea slope and in the Aleutian Islands at broad depth ranges (Table 6), but B. maculata was recorded only a few times in the Gulf of Alaska, and B. taranetzi was not recorded east of Unimak Pass. Bathyraja lindbergi (Fig. 6C) and B. minispinosa (Fig. 6D) were common only on the Bering Sea slope, but rarely encountered elsewhere. Raja bin- oculata (Fig. 7A) and R. rhina (Fig. 7B) were common only in the Gulf of Alaska, although R. binoculata was also encountered several times in the Bering Sea and eastern Aleutian Islands. Three skate species were encountered in only a few hauls throughout the 1999-2004 survey period (Fig. 8). Bathyraja abyssicola and A. badia were encountered only in a few deep hauls on the Bering Sea slope at depths greater than 950 m, and Bathyraja mariposa was encountered only in the central Aleutian Islands. Discussion Recent NMFS bottom-trawl surveys provide a wealth of reliable species-specific data on the geographic and bathymetric distributions of the skates of Alaska, as well as insight into relative population densities and regional species assemblages. Although most of the skate spe- cies of Alaska have relatively widespread distributions, each geographic and bathymetric region of Alaska has its own characteristic skate fauna and demographic characteristics. The eastern Bering Sea shelf supports large popu- lations of fishes and invertebrates and serves as an important commercial fishing area. Among skates, B. parmifera is by far the most common species, ac- counting for over 90% of the skate catch of this region. Therefore, even though skates are encountered in the 34 Fishery Bulletin 106(1 ) § Bering See Aleutian Islands Bathyraja parmlfera 0 150 300 600 kilometers y RUSSIA sgfSy 1 ALASKA & i Bering Sea SL f ^ jp Gulf of Alaska ^ Aleutian Islands Bathyraja interrupta Fig Distribution of (A) Bathyraja parmifera (Alaska skate), (B) B. aleutica (Aleutian skate), (C) B. interrupta (Bering skate), and (D) B. trachura (roughtail skate) based on data from bottom-trawl surveys conducted in the eastern Bering Sea, Aleutian Islands, and Gulf of Alaska from 1999 through 2004. Black circles indicate the presence of the species in one or more survey hauls. Depth contour = 200 m. large majority of the hauls, and overall skate num- bers have been increasing over the past 20 years (Hoff, 2006), the mean species richness and skate diversity are quite low in this region. In the deeper waters near the shelf-slope break, the skate fauna is more diverse, and B. aleutica and B. interrupta, two of the most geo- graphically and bathymetrically widespread species in Alaska, are encountered with greater frequency. In addition to widespread distributions in Alaskan waters, both of these species are also found farther south along the North American west coast and have been collected at depths of 20 to over 1000 m. Bathyraja taranetzi and B. maculata also begin to appear near the shelf break in the eastern Bering Sea, although they are much more common on the continental slope. The presence of these species near the shelf break accounts for the high mean species richness values in the deep depth strata in this region. In the southeastern corner of the Bering Sea shelf, near the eastern Aleutian Islands, the two Alaskan species of Raja are occasionally encountered, and their effect on the mean species richness values in subregion 1 are offset by the fact that this area includes the extensive shallow waters of Bristol Bay, where skates are less frequently encountered. There were no clear bathymetric or geographic trends in skate population density on the eastern Bering Sea shelf. Although overall mean density generally increased with increasing depth, the middle and deep depth strata yielded similar overall means. The low overall mean density values for both the shallow depth stratum and the southern subregion (subregion 1) appeared to be heavily influenced by the relatively low skate densities in the shallow waters of Bristol Bay. The mean density for the shallow stratum of subregion 3 was probably not meaningful because of the relatively low sample size and high variability in density data for this stratum. The continental slope of the eastern Bering Sea is a region of high skate species richness and diversity, and skates are encountered in nearly every haul. Several species, including B. aleutica, B. interrupta, B. tara- netzi, and B. maculata, are encountered from the shelf break down to well over 1000 m depth. Another group of species, characterized by a dark ventral surface — B. lindbergi, B. minispinosa, and B. trachura — -begin to Stevenson et al.: Patterns of species richness, diversity, population density, and distribution in the skates of Alaska 35 Figure 6 Distribution of (A) Bathyraja maculata (whiteblotched skate), (B) B. taranetzi (mud skate), (C) B. lindbergi (Commander skate), and (D) B. minispinosa (whitebrow skate) based on data from bottom-trawl surveys con- ducted in the eastern Bering Sea, Aleutian Islands, and Gulf of Alaska from 1999 through 2004. Black circles indicate the presence of the species in one or more survey hauls. Depth contour = 200 m. appear at depths of 300-400 m and are more common in deeper waters. Thus, there is a degree of assemblage separation; one suite of species is found on the lower slope and another suite is found on the upper slope but overall species richness remains relatively consistent throughout the depth range of the survey. A similar skate fauna has been reported from the northern Kuril Islands and southern Kamchatka (Orlov, 2005; Orlov et al., 2007), although trends in species richness in that region are unclear. Two additional species, A. badia and B. abyssicola, have been only rarely encountered in the deep waters of the eastern Bering Sea slope, although these species are widely distributed and have been re- corded from both sides of the North Pacific Ocean (Zorzi and Anderson, 1990; Stevenson and Orr, 2005). Spe- cies richness and density both show a clear geographic pattern in this region; subregions 2, 4, and 6 exhibit higher species richness and density than the other three subregions, although this effect is less pronounced in the diversity indices. These areas of high species rich- ness and density are generally associated with canyon features on the Bering Sea slope; this association may indicate that skates aggregate near these features. Alternatively, it may simply indicate that the skates in these areas cluster in the same habitats that are most suitable for bottom trawling. In the eastern Bering Sea, skate density increased with depth from shallow areas to the shelf break, and remained relatively uniform on the upper continental slope. A similar density pattern has also been observed along the Oregon coast (Pearcy et al., 1982) and in northwestern Australia (Williams et al., 2001). In the subregion encompassing Pribilof Canyon the pattern was slightly different; skate density was highest at greater depths on the lower slope, in the same general deeper depth range (700+ m) as noted by Gordon and Duncan (1985) and Merrett et al. (1991) in the north- east Atlantic. Although several distributional studies published for the western Pacific region have included basic distribution, depth range, and relative abundance information for some of the same species of skates re- ported here (Dudnik and Dolganov, 1992; Nakaya and Shirai, 1992; Dolganov, 1998, 1999; Orlov, 1998, 2003; Orlov et al., 2007), none of these studies have provided 36 Fishery Bulletin 106(1 ) Figure 7 Distribution of (A) Raja binoculata (big skate) and (B) R. rhina (longnose skate) based on data from bottom-trawl surveys conducted in the eastern Bering Sea, Aleutian Islands, and Gulf of Alaska from 1999 through 2004. Black circles indicate the presence of the species in one or more survey hauls. Depth contour = 200 m. data on overall skate density in enough detail for direct comparison with the present work. Similarly, limited sampling effort on the continental slope of the Aleutian Islands and Gulf of Alaska have precluded us from demonstrating a similar pattern in other regions of the eastern North Pacific. More species of skates have been recorded from the Aleutian Islands than from any other survey region, and yet the mean species richness figures are relatively low for this region. This finding is a result of the low encounter rate, because skates were present in only about half of the survey hauls completed in the Aleu- tian Islands. In areas where skates were encountered in the Aleutian Islands, Shannon’s diversity indices were relatively high, which may indicate that many species have similar habitat preferences. The most commonly encountered species of skates in the Aleutian Islands were B. maculata, B. taranetzi , and B. parmifera. Bathy- raja aleutica was also relatively common throughout Figure 8 Distribution of Amblyraja badia (roughshoulder skate: squares), Bathyraja abyssicola (deepsea skate: trian- gles), and B. mariposa (butterfly skate: circles) based on data from bottom-trawl surveys conducted in the east- ern Bering Sea, Aleutian Islands, and Gulf of Alaska from 1999 through 2004. Black circles indicate the presence of the species in one or more survey hauls. Depth contour = 200 m. the archipelago. Although the Aleutians Islands survey did not explore depths greater than 500 m, some of the species commonly found on the Bering Sea slope, such as B. lindbergi , B. minispinosa, and B. trachura, have occasionally been collected in the archipelago, and therefore the skate fauna on the continental slope of the Aleutian Islands is probably similar to that of the eastern Bering Sea slope. The fact that these species were only encountered in the deepest hauls in the Aleu- tian Islands survey explains why mean species richness increases with increasing depth in this region. The skate fauna of the Aleutians also displayed some regional variation. Bathyraja interrupta was relatively common in the eastern Aleutians but only rarely en- countered in the central and western Aleutians, and Raja binoculata was not collected west of Unalaska Island. Bathyraja mariposa appeared to be endemic to the central Aleutian Islands. In the western Aleu- tians B. parmifera is apparently replaced by a similar undescribed species (Stevenson et ah, 2007), treated here as B. parmifera, and the western Pacific species B. violacea is known from one specimen collected by a groundfish fisheries observer near Buldir Island in the western Aleutians. According to Orlov (2005) and Orlov et al. (2007), B. maculata, B. aleutica, and B. vio- lacea are the most abundant skate species farther west in the northern Kuril Islands and along the southern Kamchatka Peninsula. At least two of the skate spe- cies found in the Aleutian Islands, B. parmifera and B. taranetzi, exhibit very different coloration in the Aleutian Islands than in other regions where they are known. In the eastern Bering Sea, both of these species are a relatively uniform brown, often with dark brown or black blotches and occasionally pale yellow markings. Stevenson et al Patterns of species richness, diversity, population density, and distribution in the skates of Alaska 37 However, specimens of both species collected from the Aleutian Islands are much more colorful, usually with vivid yellow or olive-green markings on the disc. Ad- ditional investigation of these regional differences may lead to the recognition of more undescribed species. The deepwater species A. badia and B. abyssicola have not been encountered in the Aleutian Islands in recent trawl surveys, although this may be due to the lack of deep trawling effort in the region because both species are known to inhabit the North Pacific and the eastern Bering Sea (see Stevenson and Orr, 2005), and B. ab- yssicola has been recorded from the Aleutian Islands (Zorzi and Anderson, 1990). Unlike the eastern Bering Sea and Aleutian Islands, the skate fauna of the Gulf of Alaska is dominated by the two Alaskan species of the genus Raja. Moreover, the encounter rate of skates in the Gulf of Alaska is even lower than that of the Aleutians. These two fac- tors combine to produce very low mean species richness values. The depth distributions of the two species of Raja appear to be somewhat complementary, in that R. binoculata generally inhabits shallower waters than R. rhina, but neither species is found frequently below 400 m. Like these two species of Raja, Bathyraja inter- rupta and B. aleutica are both found throughout the Gulf of Alaska, and the ranges of both species extend well south of the Alaska border. Bathyraja parmifera is also found throughout the Gulf of Alaska, but is rare west of Kodiak (i.e., in subregions 3, 4, and 5). Like species richness values for the Aleutian Islands, species richness values are low in the Gulf of Alaska, but Shannon’s diversity indices are not particularly low. Some of the species common in the slope waters of the Bering Sea and Aleutian Islands, such as B. lindbergi (2005 survey data) and B. minispinosa (Love et al., 2005), have recently been recorded in the Gulf of Alaska, as well. Bathyraja trachura, a species relatively abundant off the U.S. west coast from Washington to California (Lauth, 2000) as well as on the eastern Ber- ing Sea slope, is also probably common throughout the deeper waters of the Gulf of Alaska, and it seems prob- able that the deepwater species B. abyssicola and A. badia (both known from around the Pacific Rim; Zorzi and Anderson, 1990; Stevenson and Orr, 2005) would be found in the Gulf of Alaska if additional deep sampling efforts were initiated. Interpretations of these data should be viewed within the context of the sampling limitations. Groundfish assessment surveys conducted in Alaska have been primarily restricted to the shelf and the shallowest por- tion of the continental slope. Although recent surveys on the eastern Bering Sea slope have improved our understanding of the distribution of skates and other deeper dwelling fishes, the lack of deeper samples from the Gulf of Alaska and Aleutian Islands is problem- atic because without these samples we are unable to make meaningful comparisons among regions. Another limitation of this data set concerns the gear used for AFSC bottom-trawl surveys. All surveys included in this study were performed with otter trawls. However, the sea floor substrate and topographic features differ markedly among the regions surveyed. Because of dif- ferences in the suitability of the sea bottom for trawl- ing, many areas within Alaskan waters (particularly in the Aleutians and Gulf of Alaska) have been considered “untrawlable,” and therefore remain unsampled. More- over, the relatively rough trawling conditions typically encountered in some regions necessitate the use of dif- ferent types of trawl gear. The effects that these differ- ences may have on the catchability of skates, and how these catchability differences may affect comparisons among surveys, are unknown (Kotwicki and Weinberg, 2005). Finally, information on any seasonal changes that may affect skate distributions (Dolganov, 1998, 1999) in Alaska is very limited. Virtually all NMFS groundfish assessment surveys in Alaska have been conducted during summer months, and we know little about the distributions of these species during the rest of the year. This study represents a major advance in our knowl- edge of the species richness, diversity, population den- sity, and distribution of the skate fauna of Alaska. It provides a reference strategy for the reliable assess- ment and monitoring of skate diversity and abundance with data from resource assessment surveys. It also provides an example of how advances in taxonomy and field identification tools can enable more detailed and robust assessments of species diversity than were pre- viously possible. These considerations are particularly critical for the skates of this region because of their high species diversity and their vulnerability to fishing pressure and habitat disturbance. We hope this study will serve for comparison with similar studies of other regions or other groups of fishes, as well as provide a baseline for monitoring future changes in the skate fauna of Alaska. Acknowledgments We thank the scientific staffs of the Alaska Fisheries Sci- ence Center groundfish bottom-trawl surveys conducted from 1999 through 2004, as well as the crews of the FV Aldebaran, FV Arcturus, FV Dominator, FV Gladiator, FV Morning Star, FV Northwest Explorer, FV Sea Storm, and FV Vesteraalen. We also thank T. Pietsch and K. Maslenikov (Univ. of Washington Fish Collection) for curation of specimens and access to collections. North Pacific groundfish observer R. Morse collected the only specimen of B. violacea. M. Martin, R. Lauth, and S. Kotwicki provided critical reviews of earlier drafts of this manuscript. Literature cited Allen, M. J., and G. B. Smith. 1988. Atlas and zoogeography of common fishes in the Bering Sea and northeastern Pacific. U.S. Dep. Commer., NOAA Tech. Rep. NMFS 66, 151 p. 38 Fishery Bulletin 106(1 ) Bremer, J. R. A., M. G. Frisk, T. J. Miller, J. Turner, J. Vinas, and K. Kwil. 2005. Genetic identification of cryptic juveniles of little skate and winter skate. J. Fish. Biol. 66:1177-1182. Britt, L. L., and M. H. Martin. 2000. Data report: 1999 Gulf of Alaska bottom trawl survey. U.S. Dep. Commer., NOAA Tech. Memo. NMFS- AFSC-121, 249 p. Casey, J. M., and R. A. Myers. 1998. Near extinction of a large, widely distributed fish. Science 281:690-691. Dolganov, V. N. 1985. New species of skates of the family Rajidae from the northwestern Pacific Ocean. Vopr. Ikhtiol. 25:415— 425. 1998. Distribution and migration of skates of the family Rajidae in Far East Seas of Russia. Izv. Tikhookean. Nauchno-Issled. Inst. Rybn. Khoz. Okeanogr. 124:433— 437. 1999. Geographical and bathymetric distribution of the skates of the Rajidae family in the Far Eastern Seas of Russia and adjacent waters. J. Ichthyol. 39:340- 344. 2001. Origin and distribution of skates of the suborder Rajoidei of the Far East Seas of Russia. J. Ichthyol. 41:355-361. Dudnik, Y. I., and V. N. Dolganov. 1992. Distribution and abundance of fish on the continen- tal slopes of the Sea of Okhotsk and of the Kuril Islands during the summer of 1989. J. Ichthyol. 32:58-76. 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. Conserv. Biol. 14: 283-293. Dulvy, N. K., and J. D. Reynolds. 2002. Predicting extinction vulnerability in skates. Con- serv. Biol. 16:440-450. Gordon, J. D. M., and J. A. R. Duncan. 1985. The ecology of the deepsea benthic and benthope- lagic fish on the slopes of the Rockall Trough, north- eastern Atlantic. Prog. Oceanogr. 15:37-69. Hoff, G. R. 2006. Biodiversity as an index of regime shift in the eastern Bering Sea. Fish. Bull. 104:226-237. Hoff, G. R„ and L. L. Britt. 2005. Results of the 2004 eastern Bering Sea upper continental slope survey of groundfish and invertebrate resources. NOAA Tech. Memo. NMFS-AFSC-156, 276 p. Ishihara, H., and R. Ishiyama. 1985. Two new North Pacific skates (Rajidae) and a revised key to Bathyraja in the area. Jpn. J. Ichthyol. 32:143-179. 1986. Systematics and distribution of the skates of the North Pacific (Chondrichthyes, Rajoidei). In Indo-Pacific fish biology: proceedings of the second international conference on Indo-Pacific fishes (T. Uyeno, R. Aria, T. Taniuchi, and K. Matsuura, eds.), p. 269-280. Ichthyol. Soc. of Japan, Tokyo, Japan. Ishiyama, R., and H. Ishihara. 1977. Five new species of skates in the genus Bathy- raja from the western North Pacific, with reference to their interspecific relationships. Jpn. J. Ichthyol. 24:71-90. Kotwicki, S., and K. L. Weinberg. 2005. Estimating capture probability of a survey bottom trawl for Bering Sea skates ( Bathyraja spp.) and other fish. Alaska Fish. Res. Bull. 11:135-145. Lauth, R. R. 2000. The 2000 Pacific west coast upper continental slope trawl survey of groundfish resources off Wash- ington, Oregon, and California: estimates of distri- bution, abundance, and length composition. U.S. Dep. Commer., NOAA Tech. Memo. NMFS-AFSC-120, 284 p. Lauth, R., and E. Acuna. 2007. Results of the 2006 eastern Bering Sea continental shelf bottom trawl survey of groundfish and invertebrate resources. U.S. Dep. Commer., NOAA Tech. Memo. NMFS-AFSC-176, 175 p. Love, M. S., C. W. Mecklenburg, T. A. Mecklenburg, and L. K. Thorsteinson. 2005. Resource inventory of marine and estuarine fishes of the west coast and Alaska: a checklist of North Pacific and Arctic Ocean species from Baja California to the Alaska-Yukon border, 276 p. U.S. Geological Survey, Seattle, WA. Ludwig, J. A., and J. F. Reynolds. 1988. Statistical ecology: a primer on methods and com- puting, 337 p. John Wiley and Sons, New York, NY. Mecklenburg, C. W., T. A. Mecklenburg, and L. K. Thorsteinson. 2002. The fishes of Alaska, 1037 p. Am. Fish. Soc., Bethesda, MD. Merrett, N. R., J. D. M. Gordon, M. Stehmann, and R. L. Haedrich. 1991. Deep demersal fish assemblage structure in the Porcupine Sea Bight (eastern North Atlantic): slope sampling by three different trawls compared. J. Mar. Biol. Assoc. UK 71:329-358. Nakaya, K., and S. Shirai. 1992. Fauna and zoogeography of deep-benthic chondrich- thyan fishes around the Japanese Archipelago. Jpn. J. Ichthyol. 39:37-48. Orlov, A. 1998. The diets and feeding habits of some deep-water benthic skates (Bathyraja spp., Rajidae) in the Pacific waters off the northern Kuril Islands and southeastern Kamchatka. Alaska Fish. Res. Bull. 5:1-17. 2003. Diets, feeding habits, and trophic relations of six deep-benthic skates (Rajidae) in the western Bering Sea. Aqua J. Ichthyol. Aquat. Biol. 7:45-60. 2005. Bottom trawl-caught fishes and some features of their vertical distribution in the Pacific waters off the north Kuril Islands and south-east Kamchatka, 1993-1999. Aqua J. Ichthyol. Aquat. Biol. 9:139- 160. Orlov, A., A. Tokranov, and R. Fatykhov. 2007. Common deep-benthic skates (Rajidae) of the northwestern Pacific: basic ecological and biological features. Cybium 2007:1-17 Pearcy, W. G., D. L. Stein, and R. S. Carney. 1982. The deep-sea benthic fish fauna of the North- eastern Pacific Ocean on Cascadia and Tufts Abyssal Plains and adjoining continental slopes. Biol. Ocean. 1:375-428. Raschi, W., D. V. Preziosi, and G. E. Walters. 1994. Recent trends in skate (family Rajidae) stocks from the eastern Bering Sea and Aleutian Islands. In Systematics and evolution of Indo-Pacific fishes: pro- ceedings of the fourth Indo-Pacific fish conference (H. L. Pratt, Jr., S. H. Gruber, and T. Taniuchi, eds.), p. 187-199. Faculty of Fisheries, Kasetart University, Bangkok, Thailand. Stevenson et at: Patterns of species richness, diversity, population density, and distribution in the skates of Alaska 39 Spies, I. B., S. Gaichas, D. E. Stevenson, J. W. Orr, and M. F. Canino. 2006. DNA-based identification of Alaska skates ( Ambly - raja, Bathyraja, and Raja: Rajidae) using cytochrome c oxidase subunit I (col) variation. J. Fish. Biol. 69:283-292. Stauffer, G. 2004. NOAA protocols for groundfish bottom trawl surveys of the nation’s fishery resources. U.S. Dep. Commer., NOAA Tech. Memo. NMFS-F/SPO-65, 205 p. Stevens, J. D., R. Bonfil, N. K. Dulvy, and R A. Walker. 2000. The effects of fishing on sharks, rays, and chimae- ras (chondrichthyans), and the implications for marine ecosystems. ICES J. Mar. Sci. 57:476-494. Stevenson, D. E., and J. W. Orr. 2005. Records of two deepwater skate species from the eastern Bering Sea. Northwest. Natur. 86:71-81. Stevenson, D. E., J. W. Orr, G. R. Hoff, and J. D. McEachran. 2004. Bathyraja mariposa: a new species of skate (Rajidae: Arhynchobatinae) from the Aleutian Islands. Copeia 2004:305-314. Stevenson, D. E., J. W. Orr, G. R. Hoff, and J. D. McEachran. 2007. Field guide to sharks, skates, and ratfish of Alaska, 77 p. Alaska Sea Grant, Fairbanks, AK. Teshima, K., and T. K. Wilderbuer. 1990. Distribution and abundance of skates in the eastern Bering Sea, Aleutian Islands region, and the Gulf of Alaska. In Elasmobranchs as living resources: advances in the biology, ecology, systematics, and the status of fisheries (H. L. Pratt Jr., S. H. Gruber, and T. Taniuchi, eds.), p. 257-267. U.S. Dep. Commer., NOAA Tech. Rep. NMFS 90. Tinti, F., N. Ungaro, P. Pasolini, M. DePanfilis, F. Garoia, I. Gur- aniero, B. Sabelli, G. Marano, and C. Piccinetti. 2003. Development of molecular and morphological mark- ers to improve species-specific monitoring and system- atics of Northeast Atlantic and Mediterranean skates (Rajiformes). J. Exp. Mar. Biol. Ecol. 288:149-165. Williams, A., J. A. Koslow, and P. R. Last. 2001. Diversity, density and community structure of the demersal fish fauna of the continental slope off western Australia (20-35°S). Mar. Ecol. Prog. Ser. 212:247-263. Zenger, H. H. 2004. Data report: 2002 Aleutian Islands bottom trawl survey. U.S. Dep. Commer., NOAA Tech. Memo. NMFS- AFSC-143, 247 p. Zorzi, G. D., and M. E. Anderson. 1990. Summary of records of the deep-water skates, Raja ( Amblyraja ) badia Garman, 1899 and Bathyraja abys- sicola (Gilbert, 1896), in the eastern North Pacific. In Elasmobranchs as living resources: advances in the biology, ecology, systematics, and the status of fisher- ies (H. L. Pratt Jr., S. H. Gruber, and T. Taniuchi, eds.), p. 389-390. U.S. Dep. Commer., NOAA Tech. Rep. NMFS 90. 40 Abstract — We compared the capture efficiency and catch dynamics of col- lapsible square and conical pots used in resource assessment and harvest- ing of red king crabs ( Paralithod.es camtschaticus [Tilesius, 1815]) in the Barents Sea. After two days of soak- ing, square pots caught three times as many crabs as conical pots, and their catches consisted of a higher pro- portion of male crabs and male crabs larger than 160 mm carapace length compared to the catches in the coni- cal pots. Catches in the square pots did not increase as soak times were increased beyond two days, which indicates equilibrium between the rate of entries into and the rate of exits from the pots. Catches in conical pots, however, increased with increas- ing soak times up to eight days, the longest soak time examined in this study. These findings demonstrate the higher efficiency of square pots and the importance of understanding catch dynamics when making popu- lation assessments based on catch- per-unit-of-effort data. The favorable catch characteristics and handling properties of the collapsible square pot may make this pot design suitable for other crab fisheries, as well. Manuscript submitted: 16 January 2007. Manuscript accepted 3 August 2007. Fish. Bull. 106:40-46(2008). The views and opinions expressed or implied in this article are those of the author and do not necessarily reflect the position of the National Marine Fisheries Service, NOAA. Efficiency and catch dynamics of collapsible square and conical crab pots used in the red king crab ( Paralithodes camtschaticus ) fishery Stian Stiansen1'2 Anders Ferno1-2-* Dag Furevik1* Terje Jergensen (contact author)1* Svein Lokkeborg1* Email address forT. Jorgensen: ter|ej@imr.no 1 Institute of Marine Research PO Box 1870 Nordnes NO-5817 Bergen, Norway 2 Department of Biology University of Bergen NO-5020 Bergen, Norway The introduction of red king crab (Paralithodes camtschaticus [Tilesius, 1815]) (RKC) in the Barents Sea in the 1960s led to a commercial pot fish- ery in this area. In Norwegian waters, surveys conducted with conical pots in order to estimate population abun- dance on the basis of catch-per-unit-of- effort (CPUE) data began in 1993. In 2002, the first licenses for commercial fishing of RKC were issued. To har- vest this new resource, an efficient, selective, and habitat-friendly fishing gear is required. Only small coastal fishing vessels (< 15 m) are licensed in the Norwegian RKC fishery and these vessels are too small to effec- tively operate large rigid pots. Ini- tially, conical pots were used because they have been used in the Russian and Japanese fishery for king crabs (Ivanov, 2002). Preliminary studies with a modified square, collapsible cod pot showed higher catches, and according to the fishermen, the square pot had better handling properties than the conical pot. The industry therefore requested adoption of this modified square pot design in the RKC fishery. The introduction of new fishing gear in the commercial fishery, and subsequently in population assess- ments, requires a knowledge of its catching properties in order to ensure sustainable harvesting practices and to interpret changes in CPUE data (Hilborn and Walters, 1992; Nizyaev and Bukin, 2002). Thus, before the collapsible square pot was adopted, fishing trials had to be carried out to compare the catching efficiency and selection characteristics of the square pot with those of the conical pot. Fishing with pots without escape- preventing devices (e.g., triggers or soft-eyed entrances) is a dynamic process, in which the ratio of crabs entering the pot to those escaping the pot usually decreases as soak time increases (Zhou and Shirley, 1997a; Watanabe and Yamasaki, 1999). Our fishing trials were therefore carried out with soak time as an explanatory variable. The objectives of the experiment were to compare the two different pot designs with regard to 1) catch effi- ciency, 2) size distribution and sex ratio of the catch, and 3) the degree to which these characteristics (catch efficiency, size distribution, and sex ratio) are affected by soak time. Such * signifies equal authorship. Stiansen et al.: Efficiency and catch dynamics of square and conical crab pots 41 Aluminium frame Float (12 mm diameter) (12 mm diameter) Seen from one end Seen from the side Figure 1 Drawing of the square pot design used to capture red king crab (Paralithodes camtschaticus). The frame is lined with 3-mm braided polyethylene netting with a mesh size of 100 mm. The 20-cm long bait bag is made of polyamide netting and has a mesh size of 22 mm. The pot weighs 37 kg out of water. All pot dimensions are in mm. Figure 2 Drawing of the conical pot design used to capture red king crab ( Paralithodes camtschaticus). The frames are made of 20-mm steel bars and lined with 4-mm polyethylene diamond netting with a mesh size of 150 mm. The 20-cm long bait bag is made of polyamide netting with mesh size of 22 mm. The pot weighs 17 kg in air. All pot dimensions are in mm. information is crucial for both im- proved population estimates and re- source-friendly harvesting practices. Materials and methods The square pot (Fig. 1) had two funnel-shaped entrances with outer openings of 150x120 cm. The entrance area thus made up half of the pot’s horizontal circumference. The netting used to line the frames had a mesh size of 100 mm. The 20- cm long bait bag was made of small- mesh netting (22 mm mesh size) and was placed in the center of the pot between the entrance openings. The weight of the pot out of water was 37 kg. The funnel of the conical pot (Fig. 2) was made of hard plastic, had an outer diameter of 45 cm, and was located at the top of the pot. The mesh size of the netting liner was 150 mm. A bait bag similar to that used in the square pots was located at the center of the pot, 10 cm below the inner funnel opening. The weight of the pot out of water was 17 kg. The fishing trials were carried out between 19 Octo- ber and 6 November 1998 in the Varangerfjord (close to the Russo-Norwegian border) (Fig. 3), which is the area with the highest density of RKC in Norwegian waters. Four commercial coastal fishing vessels of 10-12 m length were chartered for the experiment. The pots were set at depths of 50-250 m on silty and muddy bottom substrates. Each vessel fished with 24 pots, arranged in six strings — each string consisting of two square and two conical pots (Fig. 4). The two types of pots were at- tached alternately to a 12-mm diameter rope at inter- vals of about 30 m, a between-pot distance commonly used by commercial fishermen. Each pot was baited with about 1 kg of chopped (~2 cm pieces), thawed At- lantic herring ( Clupes harengus). The analyses were based on the 112 string settings where there was catch in at least one of the four pots. 42 Fishery Bulletin 106(1) Four different categories of soak times were analyzed. The strings that were given a nomi- nal two days of soak time (as is commonly used in commercial fishing) were physically in the water for 36 to 57 h (IV = 61 string set- tings; mean = 46.7 ±0.5 SE, standard error). The strings that were given three days of soak time were left out for 68 to 78 h (N= 30 string settings; mean=71.6 ±0.5 SE). Strings that were soaked for 90 to 120 h were clas- sified as soaking for four to five days (N=8 string settings; mean=107.4 ±3.6 SE). A soak time of seven to eight days was ascribed to strings that were left to fish for 167 to 192 h (A = 13 string settings; mean=176 ±3.0 SE). During pot hauling, the sex of the crabs was identified and the carapace length (CL) was measured to the nearest millimeter with a caliper gauge (Zhou and Shirley, 1997b). The difference in mesh size between the square (100-mm mesh) and conical pots (150-mm mesh) may have contributed to differences in catchability of the smallest crabs. All crabs below 91 mm CL (a total of 54 crabs in the square pots and 16 in the conical pots) were therefore excluded from the length analysis because crabs with a CL of up to 90 mm are capable of escaping through a 152-mm mesh (Zhou and Shirley, 1997c). The pooled catch of crabs from the two pots of the same type within a string was used as a single ob- servation in the statistical analyses to prevent infla- tion of tests by the use of pseudoreplicates (here: individual pots of the same type in the same string) and to minimize the effects of current (Sinoda and Kobayasi, 1969; Vienneau et al., 1993), and differ- ences in depth, bottom conditions, and patchy distri- Figure 3 Map of the study area where the square and concical pots were used to capture red king crab ( Paralithod.es camtschaticus). Sampling locations are marked with asterisks. bution of the crabs (Wallace et al., 1949; Zhou and Shirley, 1998). The catch and length observations were not nor- mally distributed. Non-parametric statistics were therefore employed in the analyses. Paired compari- sons (Wilcoxon test) of catch of RKC in square and conical pots in each string were carried out for each soak time. The Kruskal-Wallace one-way analysis of variance of ranks (Zar, 1999) was used to test for whether catch rates varied with soak time within each type of pot. The size composition of male and female crabs in the square and conical pots was com- Bouy Polypropylene rope (12 mm diameter) // 30 m Polypropylene pot strap (10 mm diameter) Figure 4 Drawing of the rigging formation used to capture red king crab ( Paralithod.es camtschaticus) in square and conical crab pots. Square and conical pots were arranged alternately along the string. Stiansen et al.: Efficiency and catch dynamics of square and conical crab pots 43 Table 1 Catch data from the comparative fishing experiment with square and conical red king crab ( Paralithodes camtschaticus) pots. Total catch and catch by sex of red king crab (Paralithodes camtschaticus ) at different soak times is shown. Within-string catches of red king crab for square and conical pots were compared by using a Wilcoxon paired comparisons test. np is number of pot set- tings for each type of pot, and N is the number of strings with nonzero catch. SE = standard error. Number of crabs caught Square pots Conical pots Paired comparisons Soak time (days) nP Median Mean SE Median Mean SE N P value Males 2 122 12.0 14.9 0.86 2.0 3.5 0.37 61 <0.001 3 60 14.0 13.9 1.04 5.0 6.2 0.66 30 <0.001 4-5 16 10.0 10.9 1.26 5.0 6.0 0.90 8 <0.05 7-8 26 13.0 14.0 1.68 9.0 9.7 1.43 13 >0.05 Females 2 122 15.0 16.1 1.02 5.0 6.2 0.53 61 <0.001 3 60 18.0 17.4 136 12.0 11.5 1.08 30 <0.001 4-5 16 20.5 19.6 3.57 11.0 10.6 1.10 8 >0.05 7-8 26 20.0 21.2 1.56 18.0 17.8 1.69 13 >0.1 Total catch 2 122 29.0 31.0 1.66 7.5 9.7 0.79 61 <0.001 3 60 32.0 31.3 1.81 15.5 17.6 1.43 30 <0.001 4-5 16 29.0 30.5 3.90 15.5 16.6 1.33 8 <0.05 7-8 26 34.0 35.2 2.83 26.5 27.5 2.54 13 <0.05 pared by means of the Mann-Whitney U- test for each soak time. A Wilcoxon test was used to test for differences be- tween the pot types in thq proportion of male crabs and males that were larger than 160 mm CL (the smallest commercial size in the Norwegian RKC fishery). In or- der to reduce the effect of small sample size (Zar, 1999), proportions equal to zero were replaced by 1/4 n (where n is the denominator in the calculation of the original proportion) and proportions equal to one were replaced by l-(l/4n). Only observed data on strings with catches in both types of pots were included in these analyses. All statistical tests were performed with Statistica, vers. 8.0 (StatSoft Inc., Tulsa, OK). Results The square pots caught significantly more crabs than the conical pots at all soak times (Table 1). At the short- est soak times, the catch ratio of square to conical pots exceeded 3:1, whereas the mean ratio (squarexonical) for the two longest soak times was close to 3:2. Except for the longest soak time, the square pots caught more of both male and female crabs (Table 1, Fig. 5). Total catches and the catches of both male and female crabs in the conical pots increased with increasing soak time (P< 0.001), but catches in the square pots did not increase beyond two days of soak time. The percentage of male crabs in the catch was higher in the square than in the conical pots after the two, three, and four to five days of soak time (Table 2). The difference was greatest for the shortest soak times be- cause the proportion of female crabs in the square pots increased with increasing soak time (P<0.05). No sig- nificant changes were found in the percentage of female crabs taken in conical pots over time. The overall percentages of male crabs larger than 160 mm were 60% (SE = 3) and 50% (SE = 3) in the square and conical pots, respectively. At all soak times, the proportion of male crabs larger than 160 mm was sig- nificantly higher in the square pots (P<0.001, Fig. 5). The median length of males was larger in the square pots for all but the longest soak time tested (Table 3). There were no differences between the types of pots with respect to the size of female crabs caught. Discussion The square pots caught up to three times more red king crab than the conical pots, but the difference decreased with increasing soak time. Provided that the two types of pots attracted equal numbers of crabs (identical bait was used), the differences in catches must be due to dif- ferent rates of entry versus exit. Because escape rates from the conical pot have been shown to be low (Godpy et al., 2003), the low catch rate for the conical pot was 44 Fishery Bulletin 106(1) D 4 / \ / / / * “v' "* 90 110 130 150 170 190 210 90 110 130 150 170 190 210 Carapace length (mm) Figure 5 Size distribution of red king crabs ( Paralithodes camtschaticus) caught in the comparative fishing experiment. The four panels give the size distribution of (A) male crabs in square pots; (B) male crabs in conical square pots, (C) female crabs in square pots, and (D) female crabs in conical pots. The x-axis is divided by size intervals of 10-mm carapace length. Note that the minimum commercial landing size of male crabs is 160 mm, marked with a vertical dashed line. presumably due to low entry rates. RKC and other species of crabs have been shown to restrict their horizontal search movement to the area suffused by the odor plume from the bait (Miller, 1980; Zhou and Shirley, 1997c; Archdale et al., 2003). However, the dif- ferences in the probability of crabs finding their way into the two pots is not explained by the size of the horizontal entrance sector (area available for entry) because the less efficient coni- cal pots had an entrance sector of 360°, whereas only half of the perimeter of square pots led to a funnel opening. Verti- cal search behavior of chemi- cally stimulated RKC has also been observed to be limited to the extent of the odor plume (Stiansen, 2004). In the square pot, the bait and funnel were at the same vertical height, whereas the entrance of the conical pot was located above the plume. Crabs were seldom observed to search for the source of the odor outside the odor plume (Miller, 1978; Vien- neau, 1993; Stiansen, 2004), which they have to do to locate the entrance of conical pots. Although the catches in the conical pots continued to increase beyond two days of soak time, the catch appeared to stabilize in the square pots, indicating that the two pot types were at different phases of the catch cycle even after two days of soak time. An ap- proximately linear increase in catches with time in the conical pots indicates that the ratio of entries to exits was relatively constant throughout the period of obser- vation. The vertical plastic funnel used in the conical pots effectively prevents crabs from escaping (Miller, Table 2 Catch data from the comparative fishing experiment with square and conical red king crab (Paralithodes camtschaticus) pots. Median and mean percentages of female red king crabs in catches taken by square and conical pots at different soak times are shown. Calculations were made on a string basis, i.e., the pooled catch from the two pots of the same type within a string was used as a single observation. The hypothesis of no difference in percentage of females in square and conical pots was tested by a Wilcoxon paired comparisons test. Only string settings with catch in both types of pots were used in the test. N is the number of strings with nonzero catches, SE is the standard error, and Nb is number of strings with catch in both types of pot. Percentages of female crabs caught Statistical test Square pots Conical pots Paired comparisons Soak time (days) N Median Mean SE N Median Mean SE P value 2 57 53.4 51.0 2.17 52 66.7 63.8 2.46 51 <0.001 3 29 56.5 52.0 4.21 27 71.0 66.4 3.76 26 <0.005 4-5 8 65.9 58.2 8.31 8 69.8 63.3 6.82 8 <0.05 7-8 13 62.9 62.2 2.82 13 65.3 67.0 4.30 13 >0.5 Stiansen et al Efficiency and catch dynamics of square and conical crab pots 45 Table 3 Catch data from square and conical pots used to capture red king crab ( Paralithodes camtschaticus). Median and mean carapace length of male and female red king crabs caught by the two types of pot at different soak times are shown. The length distribu- tions were compared by means of a nonparametric Mann-Whitney U test for each soak time, n is the number of crabs caught, SE is the standard error. Carapace length of crabs caught (mm) Square pots Conical pots Soak time U test (days) n Median Mean SE n Median Mean SE P value Males 2 1797 161 154 0.69 412 154 150 1.38 <0.01 3 823 164 154 1.04 369 151 148 1.44 <0.001 4-5 232 174 168 1.46 95 166 159 2.37 <0.01 7-8 364 142 146 1.42 250 142 143 1.65 >0.2 Females 2 1954 131 130 0.39 755 132 131 0.60 >0.1 3 1030 130 130 0.53 683 131 130 0.63 >0.7 4-5 312 139 137 0.85 170 137 136 1.22 >0.6 7-8 548 129 128 0.62 463 131 130 0.67 >0.05 1990; Cyr and Sainte-Marie, 1995), and it is not until catches are large, and crabs (by climbing on top of each other) can reach to the funnel top, that the entry and exit rates reach equilibrium (Zhou and Shirley, 1997a). The linear increase in catches, the relative low catches, and the stable sex ratio (60% females) until seven to eight days of soak time indicated that the exit rate was low for the conical pots. Unlike the catch in the conical pots, the amount of catch taken by the square pots did not increase after two days of soak time as was also observed by Zhou and Kruse (2000) who used a different square-pot design. The initially high rate of entry into pots may lead to the bait being eaten within a short time and to the result that few additional crabs are attracted into the pot. In addition, the likelihood of escape should be higher for square pots that have two horizontally orientated fun- nels, which are placed lower and made of netting that enable crabs to climb up the funnel and escape (Zhou and Shirley, 1997c). The square pots caught a higher proportion of male crabs and male crabs larger than 160 mm CL than did the conical pots. If the largest crabs take longer to es- cape than small crabs, large crabs will accumulate in the square pots over the course of time. Sexual dimor- phism, where the two sexes have different probabilities of leaving the square pot, would result in more males in the catch. Although large crabs accumulated in the square pots over time, the proportion of female crabs increased from 50% to 60% with longer soak time. This increase may also be explained by sex-dependent escape probability. Generally, total body length and height (critical dimensions in entering and escaping) are great- er for females than for males of the same carapace length (Zhou and Shirley, 1997b), and the differences increase with increasing carapace length. Thus, at a carapace length of 100 mm, Wallace et al. (1949) found female crabs to be 6 mm longer and 11 mm taller than male crabs with the same carapace length. Moreover, in the autumn (the season of this study) most female crabs larger than 100 mm CL were carrying external eggs, which would further increase body length and height. Motivational differences related to sex and condition may also have influenced the ability of RKC to climb into and out of pots (Zhou and Shirley, 1997c). The catch characteristics and design (collapsible) of the square pots make them highly suitable for com- mercial fishing. This pot design may also prove to be efficient and practical in other regions, e.g., in the Alas- kan king crab and Tanner crab ( Chionoecetes bairdi) fisheries. The size selection of the square pots resulted in them taking five times as many large male crabs (>160 mm CL, approximately 3.5 kg) as the conical pots after two days of soak time. The selectivity and catch efficiency of the square pots also resulted in a 50% re- duction in the catch of females and males smaller than 137 mm CL (minimum legal landing size). Even though discard mortality should be minimal with correct on- deck handling, there is nevertheless the likelihood that some crabs are injured, e.g., spines, rostrum, and limbs can be damaged during handling (Zhou and Shirley, 1995). RKC without all legs intact are of no commercial value and have to be discarded. This problem would be mitigated with the use of square pots. A further advantage of the square pot over the conical pot is that the higher escapement rate afforded by the square pot reduces crab mortality at sea when pots that are lost at sea continue to fish (ghost fish); the pots used in the 46 Fishery Bulletin 106(1) Norwegian fishery do not at present contain a degrad- able escapement panel. Such panels are mandatory in some regions, such as Alaska. Catch rates, sex ratio, and size distribution were clearly different for square and conical pots. CPUE data from the square pots will provide much higher estimates of the exploitable part of the population (i.e., large male crabs) than data from conical pots. Large differences were also observed in the catch dynamics; the square pot reached equilibrium at much shorter soak times than did the conical pot. Gear saturation may lead to underestimation of population abundance at high densities of crabs. In an area with high crab density, no effects of increasing soak time beyond two days were found (Zhou and Kruse, 2000), and smaller pots underestimated population density to a higher extent than did the larger pots (Nizyaev and Bukin, 2002). The relationship between pot design, catch dy- namics, and selectivity observed in the present study demonstrates the importance of adjusting for and stan- dardizing the duration of soak time when CPUE data from pots are used in population assessments. Most importantly, when using effective pots like the square pot tested in this study, soak time needs to be short in order to prevent underestimation of population size and biased sex and size distributions. Acknowledgments The authors are grateful for the invaluable comments and advice from the referees and the editorial staff. We also thank B. K. H. Ulvestad and A.-B. S. Tysseland for preparing the figures. Literature cited Archdale, M. V., K. Anraku, T. Yamamoto, and N. Higashitani. 2003. Behavior of the Japanese rock crab ‘Ishigani’ Charybdis japonica (A. Milne Edwards) towards two collapsible baited pots: evaluation of capture effectiveness. Fish. Sci. 69:785-791. Cyr, C., and B. Sainte-Marie. 1995. Catch of Japanese crab traps in relation to bait quantity and shielding. Fish. Res. 24:129-139. Godpy, H., D. M. Furevik, and S. Stiansen. 2003. Unaccounted mortality of red king crab ( Para - lithodes camtschaticus) in deliberately lost pots off Northern Norway. Fish. Res. 64:171-177. Hilborn, R., and C. J. Walters. 1992. Quantitative fisheries stock assessment, 570 p. Chapman and Hall, Inc., New York, NY. Ivanov, B. G. 2002. Red king crab {Paralithodes camtschaticus) in the eastern Okhotsk Sea: problems and stock management and research. In Crabs in cold water regions: biology, management, and economics (A. J. Paul, E. G. Dawe, R. Elner, G. S. Jamieson, G. H. Kruse, R. S. Otto, B. Sainte-Marie, T. C. Shirley, and D. Woodby, eds.), p. 651-680. Univ. Alaska Sea Grant College Program, AK-SG-02-01, Fairbanks, AK. Miller, R. J. 1978. Entry of Cancer productus to baited traps. J. Cons. Int. Explor. Mer 38(2):220-225. 1980. Design criteria for crab traps. J. Cons. Int. Explor. Mer 39:140-147. 1990. Effectiveness of crab and lobster traps. Can. J. Fish. Aquat. Sci. 47:1228-1251. Nizyaev, S. A., and S. D Bukin. 2002. Methodological problems associated with assessing crab resources based on trap catch data. In Crabs in cold water regions: biology, management, and economics (A. J. Paul, E. G. Dawe, R. Elner, G. S. Jamieson, G. H. Kruse, R. S. Otto, B. Sainte-Marie, T. C. Shirley, and D. Woodby, eds.), p. 521-536. Univ. Alaska Sea Grant College Program, AK-SG-02-01, Fairbanks, AK. Sinoda, M., and T. Kobayasi. 1969. Studies on the fishery of Zuwai crab in the Japan Sea — VI. Efficiency of the Toyama Kago (a kind of crab trap) in capturing the Beni-zuwai crab. Bull. Jpn. Soc. Sci. Fish. 25(10):948-956. Stiansen, S. 2004. Catch differences between two types of king crab pots — behavior studies in the near field and compara- tive fishing trials. Cand. scient. thesis, 85 p. Univ. Bergen, Norway. [In Norwegian.] Vienneau, R., A. Paulin, and M. Moriyasu. 1993. Evaluation of catch mechanism of conventional conical snow crab ( Chinoecetes opilio) traps by underwa- ter video camera observations. Can. Tech. Rep. Fish. Aqua. Sci. 1903:1-22 p. Wallace, M. M., C. J. Pertuit, and A. H. Nvatum. 1949. Contributions to the biology of the king crab Para- lithodes camtschtica (Tilesius). U.S. Fish and Wildlife Service. Fish. Leafl. 340:1-49. Watanabe, T., and S. Yamasaki. 1999. Catch variation and the soak time of gear in the pot fishery for the red queen crab Chinoecetes japonicus. Nippon Suisan Gakkaishi 65(4):642-649. [In Japanese] Zar, J. H. 1999. Biostatistical analysis, 4th ed., 931 p. Prentice- Hall, Upper Saddle River, NJ. Zhou, S., and G. H. Kruse. 2000. Capture efficiency and size selectivity of two types of pots for red king crabs in the Bering Sea. Alaska Fish. Res. Bull. 6(21:94-103. Zhou, S., and T. C. Shirley. 1995. Effects of handling on feeding, activity and survival of red king crabs, Paralithodes camtschaticus (Tilesius, 1815). J. Shellf. Res. 14:173-177. 1997a. A model expressing the relationship between catch and soak time for trap fisheries. N. Am. J. Fish. Manag. 17:482-487. 1997b. Behavioural responses of red king crab to crab pots. Fish. Res. 30:177-189. 1997c. Performance of two red king crab pot designs. Can. J. Fish. Aquat. Sci. 54:1858-1864. 1998. A submersible study of red king crab and Tanner crab distribution by habitat and depth. J. Shellfish Res. 17:1477-1479. 47 The trophic dynamics of summer flounder ( Paralichthys dentotus ) in Chesapeake Bay Abstract — Data on the trophic dy- namics of fishes are needed for management of ecosystems such as Chesapeake Bay. Summer floun- der (Paralichthys dentatus) are an abundant seasonal resident of the bay and have the potential to impact food- web dynamics. Analyses of diet data for late juvenile and adult summer flounder collected from 2002-2006 in Chesapeake Bay were conducted to characterize the role of this flat- fish in this estuary and to contrib- ute to our understanding of summer flounder trophic dynamics through- out its range. Despite the diversity of prey, nearly half of the diet com- prised mysid shrimp (Neomysis spp.) and bay anchovy ( Anchoa mitchilli). Ontogenetic differences in diet and an increase in diet diversity with increasing fish size were documented. Temporal (inter- and intra-annual) changes were also detected, as well as trends in diet reflecting peaks in abundance and diversity of prey. The preponderance of fishes in the diet of summer flounder indicates that this species is an important piscivorous predator in Chesapeake Bay. Manuscript submitted: 8 May/2007. Manuscript accepted: 28 September 2007. Fish. Bull. 106:47-57 (2008). The views and opinions expressed or implied in this article are those of the author and do not necessarily reflect the position of the National Marine Fisheries Service, NOAA. Robert J. Latour (contact author) James Gartland Christopher F. Bonzek RaeMarie A. Johnson Email address for R. J. Latour: latour@vims.edu Department of Fisheries Science Virginia Institute of Marine Science College of William and Mary Route 1208 Greate Road Gloucester Point, Virginia 23062 Summer flounder ( Paralichthys den- tatus) are found along the eastern seaboard of North America from Nova Scotia to Florida, but are most abundant between Massachusetts and North Carolina (Ginsberg, 1952; Leim and Scott, 1966; Gutherz, 1967). This species supports both commercial and recreational fisheries throughout southern New England and the Mid- Atlantic Bight. The commercial fishery for summer flounder has historically accounted for about 60% of the annual landings and occurs mainly in the offshore waters of the continental shelf during late fall and winter. The majority of the recreational fishery, which on occasions has exceeded the commercial harvest, takes place in state waters (i.e., estuaries and the coastal waters out to 3 nautical miles) during summer and early fall. Both fisheries contribute millions of dollars to economies on local and regional scales (Terceiro, 2002). The trophic dynamics of summer flounder have been fairly well stud- ied (Poole, 1964; Smith and Daiber, 1977; Powell and Schwartz, 1979; Roundtree and Able, 1992; Link et al., 2002; Staudinger, 2006). However, the majority of these investigations have documented the diet of summer flounder in coastal waters or in more northern estuarine environments, rather than in the southern estuaries. The latter ecosystems support a high abundance of summer flounder and provide vital summertime habitats for this species (Desfosse, 1995). The Chesapeake Bay is the larg- est estuary in the summer flounder range. No known studies have been undertaken to document summer flounder diet in these waters, and thus there has been a gap in our un- derstanding of the feeding habits of this species within an important area of its range. Further, there is growing awareness regionally, nationally, and internationally of the importance of ecosystem-based approaches to fish- eries management (EBFM). A neces- sary element in support of EBFM is nontraditional types of fisheries data, including information on the trophic dynamics of fishes. In this article, we present the diet composition of summer flounder col- lected in Chesapeake Bay from 2002 through 2006 to explore ontogenetic, interannual, and intra-annual vari- ability in diet using canonical corre- spondence analysis (CCA). Collective- ly, this information provides insight into the role of summer flounder in the Chesapeake Bay foodweb, and contributes to our understanding of the trophic dynamics of this species throughout its range. Materials and methods Field collections The data presented in this article were collected from the Chesapeake Bay Multispecies Monitoring and Assessment Program (ChesMMAP), 48 Fishery Bulletin 106(1) which is a bottom trawl survey program designed to sample late-juvenile and adult fishes in the mainstem Chesapeake Bay (i.e., nontributary waters). During 2002-2006, a total of 25 ChesMMAP cruises were conducted (March, May, July, September, and Novem- ber annually) and approximately 80 to 90 sites were sampled during each cruise. Sampling locations were chosen according to a stratified random design, and strata were defined by water depth (3-9 m, 9-15 m, and >15 m) within five 30-latitudinal minute regions of the bay. The locations sampled in each stratum of each region were randomly selected and the number of locations was in proportion to the surface area of that stratum. At each sampling location, a 13.7-m 4- seam balloon otter trawl (15.2-cm stretch mesh in the wings and body and 7.6-cm stretch mesh in the cod end) was towed for 20 min at approximately 6.5 km/h. The catch from each tow was sorted and individual lengths (total length, TL) were recorded according to species or size-class if distinct classes within a par- ticular species were evident. Stomachs were removed from a subsample of each species or size-class and immersed in preservative for diet composition analysis after each cruise. Identification of stomach contents The contents of each stomach were removed for identi- fication to the lowest possible taxon. Prey encountered in the esophagus and buccal cavity were included for identification (and assumed not to be the result of net feeding because of a lack of retention of prey in large mesh gear), whereas prey in the intestines were ignored because of the difficulty associated with identifying digested prey items in advanced stages of decomposition. All prey items were sorted, measured (either fork or total length, as appropriate and when possible), and the wet weight (0.001 g) of each was recorded. General diet description To summarize the diet composition of summer flounder in the mainstem of Chesapeake Bay, a measure of per- cent weight was calculated for each prey type (Hyslop, 1980). Because the ChesMMAP trawl collections yielded a cluster of summer flounder at each sampling loca- tion, the aforementioned percentages were calculated by using a cluster sampling estimator (Bogstad et ah, 1995; Buckel et ah, 1999). Therefore, the contribution of each prey type to the diet by weight (%Wk) was n %wk = *100 , (1) « n != 1 where qlk = — 1 — > and where n - the number of trawls containing summer flounder; Mt = the number of summer flounder collected at sampling site i; w( = the total weight of all prey items encoun- tered in the stomachs of summer floun- der collected from sampling location i; and wik = the total weight of prey type k in these stomachs. The variance estimate for %Wk was given by n ^M?(qlk-Wk)2 var (%W,,) = — xl002> (2) k nM 2 n- 1 where M = — is the average number of summer flounder collected at a sampling location. Ontogenetic and temporal changes in diet Canonical correspondence analysis (CCA; ter Braak, 1986), a multivariate direct gradient analysis tech- nique, was used to explore the relationship between summer flounder diet and three factors: fish size (mm), year (2002, 2003, 2004, 2005, 2006), and month (March, May, July, September, November). Spatial variations in summer flounder diet were not explored because the distribution of summer flounder in Chesapeake Bay is restricted primarily to the polyhaline ( > 18 ppt) region of the bay (Fig. 1). The summer flounder collected ranged in size from 148 to 712 mm TL (Fig. 2). To examine the effect of fish size on diet using CCA, we grouped summer floun- der into size categories such that all members of a given category exhibited a relatively consistent diet composition. Summer flounder were grouped into 25- mm size-classes, and diet was calculated for each with Equation 1. After trimming 10% of the observations (i.e., 25-mm size-classes) on account of low probability density in order to minimize outliers, cluster analy- sis (Euclidean distance, average linkage method) was used to group size-classes with similar diet composi- tions into broader categories. A scree plot indicated the presence of four clusters (Fig. 3A) (McGarigal et ah, 2000), corresponding to four broad size-categories: <225 mm TL (small), 225-374 mm TL (small-medium), 375-574 mm TL (large-medium), and >574 mm TL (large) (Fig. 3B). For the CCA, each element of the response matrix was the mean percent weight of a given prey type at a given sampling site in a particular size, month, and year combination. The matrix was log-transformed (ln[;c+l]) to account for the log-normal distribution Latour et al. : The trophic dynamics of Paralichthys dentatus in Chesapeake Bay 49 76°30' 76°00' 75°30' Figure 1 Average catch of summer flounder ( Paralichthys dentatus) in the mainstem (i.e., nontribu- tary waters) of Chesapeake Bay by sampling month (Mar, May, Jul, Sep, Nov) from 2002 through 2006. Horizontal histograms represent raw catch data by 0.1 latitudinal degrees corresponding to the map scale. of the data (Garrison and Link, 2000). Size, month, and year were coded by using ordinal variables. Observations (sampling sites) con- taining fewer than three summer flounder and explanatory variable blocks (size, month, year categories) containing fewer than three obser- vations were excluded to eliminate variance issues related to small sample size. The CCA was used to determine the amount of variability in the summer flounder diet ex- plained by the canonical axes, which are linear combinations of the three explanatory variables correlated to weighted averages of prey within blocks (ter Braak, 1986; Garrison and Link, 2000). The significance of the ontogenetic and temporal factors was determined by using per- mutation tests (ter Braak, 1986). A prey species- explanatory factor biplot was constructed to examine the correlations between the factors and the canonical axes and to explore the di- etary trends associated with these variables. Detailed diet descriptions were then generated for each of the significant factors identified by the CCA. The CCA was performed with the program CANOCO, vers. 4.5 (Microcomputer Power, Ithaca, NY). Total length (mm) Figure 2 Size frequency of summer flounder (Paralichthys dentatus ) sampled in the mainstem (i.e., nontributary waters) of Chesa- peake Bay from 2002 through 2006. 50 Fishery Bulletin 106(1 ) Results General diet description From 2002 through 2006, summer flounder were collected at 877 sampling locations, and at 688 of these locations at least one summer flounder had prey in its stomach. Overall, prey were encountered in 1780 (57.8%) of the 3079 stomachs collected. The total observed diet was composed of 123 prey types, 70 of which were identifiable to the species level (24 fishes and 46 invertebrates). In an effort to pres- ent summer flounder diet composition in the most efficient manner, prey types contributing relatively little to the overall diet were combined at higher taxonomic levels. Mysid shrimp ( Neomysis spp.) and bay an- chovy ( Anchoa mitchilli) were the main prey of the summer flounder, accounting for approxi- mately 42% combined (24.1% and 17.9%, respec- tively, Fig. 4) of the diet by weight, and mantis shrimp (Squilla empusa — 11.2%) and weakfish ( Cynoscion regalis — 11.1%) were of secondary and nearly equal importance. Of the remaining prey types, spot ( Leiostomus xanthurus), Atlantic croaker ( Micropogonias undulatus ), and spotted hake ( Urophycis regia ) were the most important fishes, and sand shrimp (Crangon septemspinosa) was the main invertebrate prey. Each of these species represented between 2% and 7% of the diet. All other identifiable prey types each con- tributed <2% to the diet. Unidentifiable prey items (i.e., unidentifiable fish and unidentifiable material) were prevalent, likely because of the shearing action of the teeth of these predators, and composed 6.0% of the diet by weight. Although many of the unidentifiable items were encountered in stomachs along with identifiable prey and were likely the same spe- cies as the latter, they were, however, classified as unidentifiable so as to provide a conservative diet description. Ontogenetic and temporal changes in diet The CCA indicated that summer flounder dietary changes by fish size, month, and year were sta- tistically significant. Taken together, the afore- mentioned factors explained 6.0% (P=0.001) of the variability in diet; the first and second canonical axes accounted for 51.2% and 34.5% of the explainable variation, respectively. Fish size (r=-0.459; P=0.001) more closely corresponded to the first canonical axis than the second and, of the three variables examined, accounted for the greatest portion of the variation that was explicable. Month (r=-0.481; P=0.001) and year (r=-0.094; P=0.001) were more closely correlated to the second axis (Fig. 5). The amount of fish in the diet of summer floun- der increased with increasing size (Fig. 6A). My- sid shrimp, sand shrimp, and mantis shrimp accounted for approximately 79% of the diet of the summer flounder <225 mm TL. Bay anchovy (9.5%) and weakfish (2.3%) were the main fish prey of these individuals. The diet of summer flounder ranging from 225 to 374 mm TL was al- so dominated by mysid shrimp. The contribution 1.8 -i 1.6 1.4 - 1.2 1.0 -| 0.8 0.6 0.4 0.2 0.0 • • • • “I 1 1 1- “I 1 1 1 0 575-599 - 600-624 - 625-649 - B y9 <24 - y ,40 ' 4 6 8 10 12 14 16 18 Number of clusters Trim observations Average distance between clusters Figure 3 (A) Scree plot depicting average distance between clusters versus the number of clusters which was used to identify the number of clusters into which 25-mm size-classes of summer flounder ( Paralichthys dentatus ) should be grouped (four size groups were selected since the curve leveled out at five or more clusters), (B) cluster diagram representing the relationships among the diet compositions of 25-mm size-classes of summer flounder. Trim observations represent the 25-mm size-classes omitted from the analysis because of low probability density, and average distance represents the coefficient used as a measure of dissimilarity among size-classes. Latour et al.: The trophic dynamics of Paralichthys dentatus in Chesapeake Bay 51 nc = 688 n,= 1780 Figure 4 Percent weight of prey types present in the diet of summer flounder ( Paralichthys dentatus) collected from the mainstem of Chesapeake Bay from 2002 through 2006. The total number of clusters collected is given by nc, and nt represents the total number of specimens included in this study. Standard error estimates, represented by error bars, were calculated from cluster sampling variance estimates and all were less than 0.03%. of sand shrimp to the diet of these fish was approximately the same as in the small- est size-category, whereas that of mantis shrimp increased. Fishes were again of sec- ondary importance and were represented mainly by bay anchovy, weakfish, and At- lantic croaker. Weakfish was the primary prey of the large-medium summer flounder and, although the contribution of bay an- chovy declined, anchovy still represented 15.4% of the diet. The contribution of spot to the diet of summer flounder increased from less than 1% in the small-medium fish to 13% in the 375-574 mm TL size-group. Mantis shrimp was the most important in- vertebrate prey of the large-medium fish. Sciaenids (i.e., spot, weakfish, and Atlan- tic croaker) were the main prey of of the largest summer flounder and accounted for 67.3% of the diet. Our representation of the diet composition of these fish should be viewed as preliminary because of the small cluster sample size (nc= 23). Seasonal changes in summer flounder diet likely mirrored the temporal variabil- ity of prey assemblages in Chesapeake Bay. The contribution of sand shrimp and spot- ted hake peaked in the spring and early summer (Fig. 6B). Atlantic brief squid (Lol- liguncula brevis), Atlantic croaker, mantis shrimp, silver perch ( Bairdiella chrysoura), spot, and weakfish accounted for a greater portion of the diet throughout the sum- mer and autumn. Bay anchovy and mysid shrimp were always two of the top three main prey types in the diet of summer flounder from May to No- vember. The diet of summer flounder was dominated by mantis shrimp and bay anchovy in 2002, whereas mysid shrimp was the main prey from 2003 through 2006 (Fig. 60. Atlantic brief squid, crab, mantis shrimp, and spotted hake generally decreased in importance over this time period, whereas the contribution of mysid shrimp and spot generally increased. Predator-prey size relationships The available data on sizes of whole prey consumed by summer flounder (the primary prey types excluding mysid shrimp) were examined with respect to summer flounder size. For all prey types, the size of the prey con- sumed increased significantly with increasing summer flounder size (P<0.05, Fig. 7). With respect to Atlantic croaker and spot, the majority of the individuals con- sumed were likely young-of-the-year (YOY), and a few of the larger individuals were age-1. However, summer flounder appear to have preyed exclusively on YOY weak- fish. At a given size of summer flounder, the sizes of bay anchovy, mantis, and sand shrimp consumed were more variable than the sizes of the sciaenid prey, and this finding may indicate less probability of a size-modulated predator-prey relationship. Discussion Summer flounder feed on a diverse array of prey in Chesapeake Bay, as evidenced by over 120 prey types encountered in the diet. However, despite this diversity, approximately half of the diet comprised only two prey types, mysid shrimp and bay anchovy. The other half of the diet consisted of a few fishes (sciaenids-weakfish, spot, and Atlantic croaker) and invertebrates (mantis and sand shrimps). Similar results have been reported for other upper trophic level predators in Chesapeake Bay (i.e., striped bass [Morone saxatilis] bluefish [Poma- tomus saltatrix ] and weakfish) — results that further support the notion that although the Chesapeake Bay food web is complex, the number of prey species sup- porting these predators is relatively few ( Hartman and Brandt, 1995). Mysid shrimp dominate the diets of summer flounder in other estuarine and coastal habitats (Smith and Daiber, 1977; Link et al., 2002). Our study shows that mysid shrimp also play an important role in the tro- phic dynamics of summer flounder in Chesapeake Bay. 52 Fishery Bulletin 106(1) CCA axis 1 (variance 51.2%) Figure 5 Canonical correspondence analysis (CCA) biplot for summer flounder (Paralichthys dentatus) diet in the mainstem of Chesapeake Bay from 2002 through 2006. Arrows represent the significant explanatory factors and dots represent prey types. The canonical axes represent linear combina- tions of the three explanatory variables (fish size, month, and year). Moreover, mysid shrimp have dominated the diets of other teleost piscivores in the bay over the past sev- eral years, which indicates that this prey represents a crucial linkage between lower and upper trophic level production. Despite the importance of mysid shrimp in the diets of fishes, very little is known about the population dynamics and abundance of this species (when compared to other prey types, e.g., bay anchovy) in Chesapeake Bay. Data on mysid shrimp abundance would be instrumental to better understanding not only trophic interactions of summer flounder, but those of other top teleost predators in this estuary. Significant ontogenetic changes in the diet were docu- mented; small flounder mainly consumed small inver- tebrates and bay anchovy. The diversity of the diet in terms of numbers and sizes of prey types increased with increasing summer flounder size. Medium-size flounder continued to consume prey types found in the diet of small flounder, but the diet of medium-size flounder ap- peared to be an expansion of rather than a shift from the diet of small flounder. Fishes (primarily sciaenids) were found almost exclusively in the diet of the largest summer flounder, and because bay anchovy and the aforementioned invertebrate prey types were absent in the stomachs of these fish, there appeared to be a diet shift at approximately 575 mm TL. Although similar changes in the diet of summer flounder (>500 mm TL) have been documented in offshore waters (Link et ah, 2002), cephalopods were the primary prey type as op- posed to fishes. This contrast in the diets of the larger summer flounder is likely due to the lack of an abun- dant and comparable large soft-bodied invertebrate prey in Chesapeake Bay. Seasonal trends in summer flounder diet composi- tion were not surprising given the well documented spatiotemporal patterns of summer flounder prey. Sand shrimp and spotted hake abundance generally peaks during late winter and early spring in the mainstem of the lower bay; hence, it follows that they composed ap- preciable fractions of the summer flounder diet during this season (Haefner, 1976; Murdy et ah, 1997). Faunal diversity in Chesapeake Bay reaches a maximum dur- ing late August and September and corresponds with a highest diversity of prey types in the diet of summer Latour et al.: The trophic dynamics of Paralichthys dentatus in Chesapeake Bay 53 Small (<225 mm) nc = 88 n,= 128 — -r-jjl K B t t .j 1 T i ,£f\_ X)'' yy yyy^ <>vs Large-med. (375 mm-574mm) nc = 312 n, = 438 J I I w /»€ 4?# 4^ O “= s # Figure 6 Diet composition (percent weight) of summer flounder (Paralichthys dentatus) collected from the mainstem of Chesapeake Bay, presented by (A) size-category, (B) month, and (C) year. The number of clusters collected in each subcategory is given by nc, and nt represents the total number of speci- mens. Error bars represent standard error of the percent weight values of each of the prey types encountered in the summer flounder diet, which were calculated from cluster sampling variance estimates. flounder. Interannual variations in the diet of summer flounder generally followed fluctuations in the indices of relative abundance for several prey species routinely monitored by the Virginia Institute of Marine Science (VIMS) Juvenile Finfish and Blue Crab Trawl Survey. There was a weak visual correspondence between the trends in relative abundances of bay anchovy and YOY weakfish and their contributions to summer flounder diet throughout the study period. However, the diet of summer flounder more strongly mirrored trends in relative abundance of YOY spot. In general, it is difficult to compare studies of diet composition of the same species because it is often the case that survey design (including gear types), indices reported (e.g., percent weight, %W vs. percent number, %N), and the methods used to calculate these indices (e.g., simple random vs. cluster sampling) vary among studies. Although these differences prohibit direct comparisons among investigations, it is still possible to draw some informative qualitative conclusions. For example, Smith and Daiber (1977), using the percent frequency of occurrence (%F) index, reported that the diet of summer flounder in Delaware Bay was domi- nated by invertebrates; yet their results also indicated that fishes composed an important part of their diet in the estuary. Poole (1964) reported that sand shrimp were the main prey by weight of summer flounder in Great South Bay, NY; however, fishes were also abun- 54 Fishery Bulletin 106(1 ) Figure 6 (continued) dant in the diet. The relative importance of specific fish species in the diet of summer flounder has varied across studies, likely because of spatial variations in prey assemblages and perhaps because of differences in study methods. Nevertheless, these studies in combina- tion with the results of the present study indicate that summer flounder are piscivorous within estuarine en- vironments throughout their range. Additionally, there appears to be appreciable similarity in the invertebrate taxa consumed by summer flounder in estuaries because sand and mysid shrimps have been found in the diet in multiple areas across decades (Poole, 1964; Powell and Schwartz, 1979). Striped bass, weakfish, and bluefish represent the abundant upper trophic level teleost piscivorous preda- tors in Chesapeake Bay (Dovel, 1968; Boynton et ah, Latour et al.: The trophic dynamics of Paralichthys dentatus in Chesapeake Bay 55 1981; Hartman and Brandt, 1995), however, the pre- ponderance of fishes in the diet of summer flounder indicates that this species also fits that characterization (i.e. , fishes represent approximately 50% or more of the diet of summer flounder >225 mm TL). In terms of life history and estuarine dependence, appreciable abun- dances of summer flounder have been consistently pres- ent in our samples over the past several years. Hence, the sheer abundance, protracted use of estuarine habi- tat, and piscivorous diet of summer flounder combine to indicate that the impacts on piscine prey by this species have the potential to match those of the aforementioned three fishes. Piscivory was also documented in several size-classes of summer flounder within offshore habitats along the continental shelf (>10 m depth) from southern New England through the Mid-Atlantic Bight (Link 56 Fishery Bulletin 106(1 ) Summer flounder TL (mm) Figure 7 Relationship of prey size (whole prey items only) consumed by summer flounder ( Paralichthys dentatus) in the mainstem of Chesapeake Bay from 2002 through 2006 versus summer flounder size (TL, mm). All regressions were significant (P< 0.05). et al., 2002; Staudinger, 2006). Hence, fishes repre- sent an important component of summer flounder diet throughout its range implying that this species should be included in analyses designed to quantify pathways of production to piscivorous fishes. Quantitative analyses of foodweb dynamics provide valuable insights into the structure of ecosystems and ultimately support the development of EBFM plans. However, these analyses require several data types, in- cluding information on the ontogenetic and temporal (in- tra- and interannual) changes in the trophic interactions of species within an ecosystem. This study provides fun- damental trophic data for an important fish species in Chesapeake Bay and, taken with previous studies, con- tributes significantly to our understanding of the role of summer flounder as a predator throughout its range. Acknowledgments The authors acknowledge E. A. Brasseur, P. D. Lynch, D. J. Parthree, and M. L. F. Chattin for their efforts Latour et at: The trophic dynamics of Paralichthys dentatus in Chesapeake Bay 57 associated with field collections and sample processing. Captain L. Durand Ward and the Virginia Institute of Marine Science (VIMS) vessels staff deserve thanks for their contributions in the field. T. Miller and two anonymous reviewers provided helpful comments on ear- lier versions of this manuscript. Funding was provided by the Virginia Environmental Endowment, National Oceanic and Atmospheric Administration Chesapeake Bay Office, and the U.S. Fish and Wildlife Service. This article is VIMS contribution no. 2850. Literature cited Bogstad, B., M. Pennington, and J. H. Volstad. 1995. Cost-efficient survey designs for estimating food consumption by fish. Fish. Res. 23:36-47. Boynton, W. R., T. T. Polgar, and H. H. Zion. 1981. Importance of juvenile striped bass food habits in the Potomac estuary. Trans. Am. Fish. Soc. 110:56-63. Buckel, J. A., D. O. Conover, N. D. Steinberg, and K. A. McKown. 1999. Impact of age-0 bluefish ( Pomatomus saltatrix ) predation on age-0 fishes in the Hudson River estu- ary: evidence for density-dependent loss of juvenile striped bass ( Morone saxatilis). Can. J. Fish. Aquat. Sci. 56:275-287. Desfosse., J. C. 1995. Movements and ecology of summer flounder, Para- lichthys dentatus tagged in the southern Mid-Atlantic Bight. Ph.D. diss., 187 p. College of William and Mary, Williamsburg, VA. Dovel, W. L. 1968. Predation by striped bass as a possible influence on population size of the Atlantic croaker. Trans. Am. Fish. Soc. 97:313-319. Garrison, L. P., and J. S. Link. 2000. Diets of five hake species in the northeast United States continental shelf system. Mar. Ecol. Prog. Ser. 204:243-255. Ginsberg, I. 1952. Flounders of the genus Paralicthys and related genera in American waters. Fish. Bull. 52:267-351. Gutherz, E. J. 1967. Field guide to the flatfishes of the family Bothi- dae in the western North Atlantic, 47 p. U.S. Fish and Wildlife Service, Bureau of Commercial Fisheries, Circular 263, Washington, D.C. Haefner, P. A., Jr. 1976. Seasonal distribution and abundance of sand shrimp Crangon septemspinosa in the York River-Chesapeake Bay estuary. Chesapeake Sci. 17:131-134. Hartman, K. J., and S. B. Brandt. 1995. Trophic resource partitioning, diets, and growth of sympatric estuarine predators. Trans. Am. Fish. Soc. 124:520-537. Hyslop, E. J. 1980. Stomach contents analysis-a review of methods and the application. J. Fish Biol. 17:411-429. Leim, A. H., and W. B. Scott. 1966. Fishes of the Atlantic Coast of Canada. Bull. Fish. Res. Board Can. 155:1-485. Link, J. S„ K. Bolles, and C. G. Milhken. 2002. The feeding ecology of flatfish in the Northwest Atlantic. J. North. Atl. Fish. Sci. 30:1-17. McGarigal, K., S. Cushman, and S. Stafford. 2000. Multivariate statistics for wildlife and ecology research, 283 p. Springer, New York, NY. Murdy, E. O., R. S. Birdsong, and J. A. Musick. 1997. Fishes of Chesapeake Bay, 324 p. Smithsonian Institution Press, Washington, D.C. Poole, J. C. 1964. Feeding habits of summer flounder in Great South Bay. NY Fish Game J. 11:28-34. Powell, A. B., and F. J. Schwartz. 1979. Food of Paralichthys dentatus and P. lethostigma (Pisces: Bothidae) in North Carolina estuaries. Es- tuaries 2:276-279. Roundtree, R. A., and K. W. Able. 1992. Foraging habits, growth, and temporal patterns of salt-marsh creek habitat use by young-of-the-year summer flounder in New Jersey. Trans. Am. Fish. Soc. 121:765-776. Smith, R. W., and F. C. Daiber. 1977. Biology of the summer flounder, Paralychthys den- tatus, in Delaware Bay. Fish. Bull. 75:823-830. Staudinger, M. D. 2006. Seasonal and size-based predation on two species of squid by four fish predators on the Northwest Atlantic continental shelf. Fish. Bull. 104:605-615. ter Braak, C. J. F. 1986. Canonical correspondence analysis: a new eigen- vector technique for multivariate direct gradient analysis. Ecology 67:1167-1179. Terceiro, M. 2002. The summer flounder chronicles: science, poli- tics, and litigation, 1975-2000. Rev. Fish Biol. Fish. 11:125-168. 58 Abstract — A new description of growth in blacklip abalone ( Haliotis rubra) with the use of an inverse- logistic model is introduced. The inverse-logistic model avoids the dis- advantageous assumptions of either rapid or slow growth for small and juvenile individuals implied by the von Bertalanffy and Gompertz growth models, respectively, and allows for indeterminate growth where neces- sary. An inverse-logistic model was used to estimate the expected mean growth increment for different black- lip abalone populations around south- ern Tasmania, Australia. Estimates of the time needed for abalone to grow from settlement until recruit- ment (at 138 mm shell length) into the fishery varied from eight to nine years. The variability of the residu- als about the predicted mean growth increments was described with either a second inverse-logistic relationship (standard deviation vs. initial length) or by a power relationship (standard deviation vs. predicted growth incre- ment). The inverse-logistic model can describe linear growth of small and juvenile abalone (as observed in Tasmania), as well as a spectrum of growth possibilities, from determinate to indeterminate growth (a spectrum that would lead to a spread of maxi- mum lengths). Manuscript submitted 9 April 2007. Manuscript accepted 22 October 2007. Fish. Bull. 106:58-71 (2008). The views and opinions expressed or implied in this article are those of the author and do not necessarily reflect the position of the National Marine Fisheries Service, NOAA. Using an inverse-logistic model to describe growth increments of blacklip abalone ( Haliotis rubra) in Tasmania Malcolm Haddon (contact author) Craig Mundy David Tarbath Email address for M. Haddon: Malcolm.Haddon@utas.edu.au Marine Research Laboratory Tasmanian Aquaculture and Fisheries Institute University of Tasmania Nubeena Crescent Taroona TAS 7051, Tasmania, Australia Blacklip abalone (Haliotis rubra ) constitute the most valuable fishery in Tasmania, Australia, yielding approximately 30% (2500 tonnes) of the resource captured in the wild worldwide, worth more than AU$100 million per year. There are signifi- cant difficulties in determining the age of blacklip abalone (McShane and Smith, 1992), making them good candidates for size-structured assess- ment modeling (Sullivan et al, 1990; Punt and Kennedy, 1997). Although used informally in the Tasmanian fishery, size-structured models are used formally to assess blacklip aba- lone stocks elsewhere in Australia (Worthington et al., 1998; Gorfine et al., 2005) and to assess Paua ( H . iris) in New Zealand (Breen et al., 2003). With size-structured models it is important to generate a precise and unbiased mathematical descrip- tion of growth because the adoption of an inappropriate growth model could have significant effects on the outcome of an assessment. Day and Fleming (1992) reviewed a range of models previously used to describe abalone growth. In total, 59 growth studies have been undertak- en on different abalone species. The von Bertalanffy growth curve (von Bertalanffy, 1938) has been used in 42 (71%) of these studies and the Gompertz model (Gompertz, 1825). has been used in four studies (6.78%). The dominance of the von Bertalanffy growth model reflects its almost uni- versal adoption in abalone fisheries assessments and the relative ease with which it can be fitted to growth data taken from tagging experiments (Fabens, 1965; Francis, 1988; Had- don, 2001). In nine studies (15.25%), linear growth was proposed, but the focus of these nine studies was on juvenile and small abalone and that focus could imply that growth alters its character above a particular size, at least in some species. A flexible growth description has also been giv- en by Francis (1995) who generated a size-based analogue to the age-based growth description by Schnute (1981). Francis’s model has been used in New South Wales, Australia (Worthington et al., 1998), and New Zealand as- sessments of abalone (Breen et ah, 2003). Sainsbury (1982a, 1982b) fitted von Bertalanffy growth curves to abalone tagging data from New Zea- land (H. iris). Instead of assuming that the familiar parameters (Lx, the asymptotic maximum size, and K, the growth coefficient in the von Bertalanffy equation) were averages for the population, Sainsbury (1982a) inferred the growth dynamics implied when each individual had its own set of von Bertalanffy parameters. Es- sentially, the growth characteristics of individuals were assumed to be variable and were described by using probability density functions to repre- sent the model parameters. Similarly, a probability density function form of the Gompertz growth model was used in Victoria, Australia (Troynikov and Haddon et al.: Using an inverse-logistic model to describe growth increments of Haliotis rubra 59 Table 1 Data by sites and regional groupings of sites (Fig. 1) in southern Tasmania where blacklip abalone ( Haliotis rubra) were collected to compare growth rates. Time step relates to whether the analysis was for annual or seasonal growth, count is the number of tags recovered, and Min-Max Init. L (in mm) are the minimum and maximum initial length of blacklip abalone at tagging for each site. Time step Longitude (East) Latitude (South) Regional grouping Site name Count n Min-Max Init. L Annual 145.492 -42.969 Southwest Black Island 116 57-171 Annual 145.667 -43.075 Southwest Giblin River 84 83-173 Annual 145.781 -43.226 Southwest Hobbs Island 57 57-181 Annual 146.900 -43.566 Actaeon Gagens Point 154 50-142 Annual 146.972 -43.549 Actaeon Middle Ground 353 47-146 Annual 147.381 -43.366 Bruny Island Fluted Cape 135 83-154 Annual 147.385 -43.111 Bruny Island One Tree Point 162 52-153 Seasonal 146.996 -43.534 Actaeon Actaeon Island 390 61-176 Seasonal 146.990 -43.550 Actaeon Sterile Island 373 48-146 Gorfine, 1998; Troynikov et al., 1998; Bardos, 2005). Like the von Bertalanffy model, the Gompertz equation is deterministic in predicting an asymptotic maximum length. The probabilistic forms of these two models pre- dict a more plausible range of final maximum lengths and some configurations of Francis’s (1995) model can also exhibit a spread of final sizes. However, like the von Bertalanffy and Gompertz models, Francis’s (1995) model fails to exhibit the linear-like early growth of small abalone that has been observed in Tasmania. An important characteristic of growth models is an ability to accurately model growth across a broad size range. Nine of the studies cited in Day and Fleming (1992) indicate that early growth in abalone is effec- tively linear. This linearity contrasts strongly with both the von Bertalanffy curve (which predicts faster early growth) and the Gompertz growth curve (which predicts slower early growth). Neither the von Bertalanffy nor the Gompertz growth models are consistent with ob- servations of effectively linear growth in small blacklip abalone in Tasmania, Australia (Prince et al., 1988; Gurney et al., 2005). Such early linear-like growth would imply constant growth increments in small ani- mals and would require a different structural model to represent such growth dynamics. Given the wide variation in maximum sizes found in natural abalone populations, an alternative approach to using deterministic models (with a known or even prob- abilistic asymptotic length) would be to model growth as indeterminate. Indeterminate growth would imply no specific upper limit and animals would be expected to continue growing, even if very slowly, until they die. This indeterminacy would have the disadvantage in that there would be no simple analytical solution for length- at-age, but nevertheless this strategy could provide for intuitively simple empirical descriptions of growth that avoid the complexities of fitting probabilistic models as proposed by Sainsbury (1982a) and Bardos (2005). Here we present a new empirical description of black- lip abalone growth using an inverse-logistic model for both the mean growth increment and the predicted variation about the mean increment for a given shell length. In contrast to both the von Bertalanffy and Gompertz growth curves, the growth description given here allows for both linear growth of small and juve- nile abalone as well as the option of either determinate growth (with a maximum shell length) or indeterminate growth (with a spread of maximum lengths). Materials and methods Examination of growth patterns To provide an initial empirical indication of growth patterns, the size of abalone at tagging were grouped into 10-mm classes and the mean growth increment in each class was then plotted on top of the raw data from the southwest area of Tasmania (a combination of three sites, Fig. 1; Table 1). Tagging methods and locations Two sets of data were used in the description of the inverse-logistic model. Firstly, to examine annual growth increments, we used tagging data from seven sites around the south of Tasmania (Fig. 1). Data from those sites were limited to tagged-and-recaptured abalone and the data were collected approximately one year apart (between 0.96 and 1.05 years apart). Abalone from some groups of sites were found to exhibit very similar growth patterns and the data from these sites were combined to generate three larger regions (namely, southwest, Actaeon, and Bruny Island) (Fig. 1; Table 1). Secondly, tagging data from two sites were used to examine seasonal growth (Table 1), and for this analysis, 60 Fishery Bulletin 106(1 ) A map of Tasmania indicating the locations (black dots) from which the data on blacklip abalone ( Haliotis rubra) growth increment patterns were collected. Left panel represents the southwest region comprising three sites: BI = Black Island, GR = Giblin River, and HI = Hobbs Island. Right panel represents the Actaeon region (comprising two sites: the GP=Gagens Point and MG=Middle Ground sites) and the Bruny Island region (com- prising two sites: FC = Fluted Cape and OTP=One Tree Point sites). Seasonal data are identified by a combination of a circle and cross for both the Actaeon Island (AI) and Sterile Island (SI) sites (Table 1). tagging and recaptures were scattered throughout the year and the recapture intervals ranged from 0.06 to 1.99 years. We tagged blacklip abalone by inserting a plastic rivet into the open exhalent hole furthest from the shell lip. The rate of tag fouling, and the increased risk of the tag not being found, increased dramatically after two years; therefore, only tags at liberty for less than two years were used in the analyses of seasonal growth. Across all sites, tagged abalone ranged in size from 47 mm to 181 mm shell length, but ranges varied at each site (Table 1). Measurements of maximum shell length were taken to the nearest 1.0 mm. There is some evidence that tagging can negatively affect the growth of tagged animals, producing a tagging shock in affected individuals (Prince et al., 1988). It seems plausible that at least some of the variability in growth observed in abalone tagging experiments derived from different responses to the tagging process. In an attempt to minimize such effects in this study, tagging methods were standardized; animals were kept damp and cool during the tagging process before being returned to their reefs by divers. Blacklip abalone <50 mm shell length were not sampled because they tend to be highly cryptic in Tasmania, initial capture below this size is difficult, and adhesives rather than rivets, must be used to attach tags to young abalone. Growth model The units of growth used here are the growth increments (AL) produced by animals of known starting lengths (Lt) that have been at liberty for varying lengths of time (At). The model structure includes seasonality by default, and the seasonality terms are set to have zero influence to form an annual model. The form of the inverse-logistic curve used is a typical logistic selectivity curve described by Haddon (2001). Growth of blacklip abalone does not differ between the sexes. An inverse-logistic model was used to de- scribe the expected length increment AL for a known initial size Lt: Haddon et al.: Using an inverse-logistic model to describe growth increments of Haliotis rubra 61 MaxAL x AL = AZ + Csin^Tr^-p))- Csin(27r(zr-p) Ln(l9) (**-*&) + £t (1) l + e iL%5-L5no) The fact that L/L19) is used instead of -Ln(19) implies that the logistic is inverse and that the Lgs parameters relates to the 95% point (Ln(15) would equate to the 75% point). The inverse-logistic description of variation is general; however, if the expected length increments always remain greater than zero then the standard deviation of the residuals oL can be defined as the simpler where MaxAL = At = Lt = J m — ^ 50 - T m - •^95 “ c = tR and tT = P = the hypothetical asymptotic maximum growth increment at some initial size of abalone that sets the exponential term to zero; the interval between tagging and recapture (as a fraction of a year); the size when first tagged; the initial length at which the mid- way point between the MaxAL and lowest growth increment is reached; the initial length at which 95% of the difference between the smallest and maximum increment is reached; the amplitude of the seasonality effect for AL; the dates of recapture and tagging, respectively (as fractions of a year, e.g., June 30th = 0.5; tR = tT + At); and the date of maximum growth rate (as a fraction of a year). The error term £Lt is additive and normal, and is assumed to have a mean of zero and standard deviation oL that can be defined either as a function of initial length, Lt, or as a function of the predicted length increment A Lt. If the expected length increments ever attain zero, or go negative, then the standard deviation of the residuals oL can be defined in terms of the initial length Lt : MaxoL x °Lt=- AZ + OrSm^/r^-p))- C^sin \2n[tT -p) Lt-L* Ln( 19)— 50_ l + e 95 50 where MaxoL = the hypothetical asymptotic maximum standard deviation of the residual values at some initial size of abalone that sets the exponential term to zero; L|0 and L|5 = the parameters describing the inverse- logistic for how the variability of resid- uals reduces with increasing Lt; and Ca = the amplitude of the seasonality effect for the o, term. Lt °Lt = a (3) where a and jl are parameters of a power relationship with the expected length increment A Lt and the season- ality is achieved from Equation 1. When seasonality is ignored (when estimating annual growth increments), the C and Ca parameters are set to zero leaving the simple At so that any slight devia- tions from a At of one year are assumed to alter the predicted growth in a linear fashion. Thus, 0.95 of a year permits 95% of the growth increment of that year. With the use of At alone, there is the assumption that a simple linear scaling of growth increment with respect to time elapsed will provide sufficient adjustment for small deviations of At from one year. Using a normal distribution to describe the residuals, we found that there was an excellent match of this dis- tribution to available data. However, if some probability density function other than the normal distribution provided a better fit for some other species or popula- tion, then the equivalent measure of spread about the expectation would need to be implemented. Where the tagging interval is greater than one year, the expected growth increment is estimated in two steps. First, the expected growth increment and stan- dard deviation that would be expected during a year of growth are estimated, and then the growth increment for the fraction of the year remaining from the date of tagging to the date of recapture (after subtracting one year) is estimated by using the initial size plus the estimated yearly growth increment as the starting length for the second installment of growth. Thus, AL is first estimated with Equation 1 with the C and Ca parameters set to zero, and At set to 1.0, and then the fraction of a year remaining from the date of tagging to the date of recapture (after subtracting one year) is used in the full version of Equation 1 and the Lt is set to the original Lt plus the AL predicted from one year of growth. The two sequential AL estimates are added together to obtain the total predicted growth incre- ment. Using Equation 2 to define the variation about the curve, we applied a similar sequential process to the estimation of the standard deviation of the respec- tive residual errors. In this case, the expectation was that the variability would reduce with increasing size so the Ca parameter was expected to be negative rather than positive as was expected for the C parameter. The use of Equation 3 requires that it be applied to 62 Fishery Bulletin 106(1 ) the predicted AL. Given that the first year of growth is always assumed to equal the average increment, there is an increased chance that the overall variability of the residuals will be underestimated. However, in practice, bias appears to be small as long as the data available from greater than a single year overlap the available data from durations less than a year in terms of the initial shell lengths. Alternative growth models To provide a comparison with the inverse-logistic model both the von Bertalanffy (Fabens, 1965) and Gompertz curves (Troynikov et al., 1998) were fitted to the tagging increment data from southwest Tasmania: AL = {L^-Lt)[l-e~K*t\ (4) where L ^ - the asymptotic maximum size; and K = the von Bertalanffy growth rate coefficient; and with A L = L„ h. L exp(-gAf) ~Lt, (5) only on the seasonal changes in variability, thus reduc- ing the number of parameters to seven: °Lt = MaxoL At+Casm{2n(tR-p))- Casin{2n(tT-p )) (7) An alternative approach to implementing this structural change would be to set the L|0 and L|5 parameters in the denominator of Equation 2 to values much larger than the maximum observed initial size. This change in the denominator leads to the exponential term becoming insignificant so that the denominator contracts to one, the division thus has no noticeable effect, and Equation 2 becomes equivalent to Equation 7. When only annual data are available, the seasonality terms could be ignored and thus a six-parameter model could be used. After the six-parameter model was fitted to real data, it became clear that the L|5 value was often close to the maximum size of abalone found in Tasmania; therefore it was possible to generate a five- parameter model by replacing the L|5 parameter with a constant 210 mm (the size of an abalone that was never tagged but sometimes found in nature). In addition, the L|0 value was often close to the L'<£5 value. By replacing the former with the latter it was possible to generate a four-parameter model: where g - the growth rate parameter in the Gompertz equation. Likelihoods °Lt = MaxoL x At Ln(19) — 9a 1 + e 2F0-L- (8) At each geographical site, normal likelihoods with non- constant variances (Eqs. 2 or 3), were used to fit the inverse-logistic model to the n available data points. The negative log-likelihood was minimized to determine the optimum parameter estimates: -veLL = - Ln Lf= l AL-AL 42noL 2a 2 Lf (6) The nonlinear solver in Excel 2003™ (Microsoft, Seattle, WA) was used to fit all models. Alternative model arrangements The full seasonal model has nine parameters, but alter- native model structures are possible that use fewer parameters. The alternative model structures suggested relate to the description of the variability about the expected curve. With the seasonal growth description, instead of using Equation 2 to describe the expected residual structure with the Tasmanian data, an accept- able alternative was to ignore the denominator and focus The three alternative annual models (4, 5, and 6- parameter models) were fitted to the available data from the three regional groups of sites from around southern Tasmania. A comparison of the relative fit of each model was made by using Akaike’s information criterion, AIC = -2 LL + 2k, where LL is the log-likeli- hood and k is the number of parameters. In addition, the Bayesian Information Criterion BIC = -2 LL + kLnin) was also used, where n is the total number of observations (Burnham and Anderson, 2002). For each of these statistics, the model with the smallest value is to be preferred (Quinn and Deriso, 1999). The AIC only includes the log-likelihood and the number of parameters, whereas the BIC also includes the natural log of the sample size. Where the sample size is greater than 7 [Ln(7.389)=2], the BIC penalizes the addition of extra parameters more than the AIC. Thus, with the sample sizes observed in the tagging data (Table 1) the BIC would be expected to recommend more par- simonious models (those with fewer parameters) than would the AIC. In addition to comparing the AIC and BIC values for the different models, likelihood ratio tests were con- ducted to compare the alternative model fits by using different numbers of parameters (Quinn and Deriso, 1999; Haddon, 2001). Given the log-likelihood for each model fit, the likelihood ratio test is Haddon et al.: Using an inverse-logistic model to describe growth increments of Haliotis rubra 63 Xdf ~ -2 (LLr - LLf (9) where LL, LL, - the log-likelihood for the model with fewest parameters; and - the log-likelihood of the model with most param- eters. In the chi-squared statistic, x2df> df is the difference in number of parameters (in this case comparing 4 with 6 and 5 with 6 means df takes the value of either 2 or 1, respectively). Defining the growth transition matrix The data from all sites includes instances of negative increments (Fig. 2) and hence it is not surprising when the predicted increments also feature nega- tive values. This implies that abalone can decrease in size during a time step. However, abalone tend not to exhibit negative growth; instead these negative increments are assumed to be the result of measurement error. The transition probabilities are simply the cumulative normal distribution to the upper size limit of each size class minus the cumulative normal distribution to the lower size limit of each size class in turn: 65 75 85 95 105 115 125 135 145 155 165 175 185 Initial shell length (mm) Figure 2 Plot of growth increment (mm) after approximately one year against ini- tial length (mm) for blacklip abalone ( Haliotis rubra) from the southwest region of Tasmania. The vertical dashed lines represent the boundaries of the 10-mm size classes and the curved line and black squares represent the trend and mean growth increments, respectively, for each class of initial sizes. Mean values are only shown for initial size classes repre- senting more than one observation. Negative increments were included in the mean estimates. G-r | 2 U-L. LJ) j2noJL . dL L. = L, OJ , Li L, = the standard deviation of the normal distri- bution of growth increments for the initial size class j. = the expected average final size for initial size class j, which equals + A L( •, where A L; • is the average expected growth increment for initial size class j. Summing the smallest size class to -oo and the largest size class to +°° effectively makes both of these size classes plus-groups that ensure that the transition prob- abilities for all n size classes sum to one. L.+LW G,'J J Li-L. hj) LW^< G,-J J rl- Lj-L. , 1 i, dL LMin < L- < LMax , ( 10 ) 2 al Li- 2 LW dL L, = L. where G- • = the transition probability of an abalone growing from size class j into size class i; L( = the mid-size of size class i; LW - the size class width; and Length at age Because of the potentially indeterminate nature of the growth description, there is no analytical version of the growth equation that can provide a length for a given age. Instead, growth needs to be simulated to estimate length at age. That is, an initial length is assumed and then the predicted growth increment in a given time interval (seasonally short or annual) is estimated; this is then added to the initial length and the process is repeated to generate a predicted length at age. The simulated growth increments may include stochasticity (guided by the non-constant variance with initial length) and lead to a scatter of predicted sizes. Alternatively, growth can be simulated by using the growth transi- tion matrix from the model fit. For annual growth only one transition matrix is required, however, to describe seasonal growth there would need to be an array of 64 Fishery Bulletin 106(1 ) von Bertalanffy Gompertz Inverse-logistic 25 50 75 100 125 150 Initial size Lf (mm) 175 Figure 3 Visual comparison of the von Bertalanffy, Gompertz, and inverse-logistic curves fitted to the annual data from the southwest region of Tasmania. The expected growth increments for smaller blacklip abalone ( Haliotis rubra) for each curve are also illustrated as extensions of the lines (to the left) and these demonstrate major differences between the curves. The horizontal line at zero represents the point of no growth. growth transition matrices describing the expected growth for different parts of the year. Comparison of productivity between areas With the inverse-logistic description of growth the parameter combinations do not provide an intuitively obvious indication of the productivity of different areas. A large MaxAL does not necessarily mean an area has high productivity if the large growth increments occur only for relatively small animals. An index of relative productivity can be obtained by applying the growth transition matrix derived for an area to a standard initial vector of numbers at size. The areas considered here were compared by generating an annual transition matrix for each area with 31 five-mm size classes from 60 mm up to 210 mm. This transi- tion matrix was multiplied with an initial numbers-at-size vector containing 1000 individuals in the smallest size class. This multiplication was then repeated itera- tively for 10 years of growth (i.e., without mortality). The final numbers at size were converted to mass by using the standard Weight = a Lengthb where, in southern Tasmania, a = 5.669.E-05 and b = 3.1792, to provide a comparable index of relative productivity between areas in kilograms. Results Initial summary of growth patterns The general growth pattern apparent in the southwest Tasmania (Fig. 2) was also found at other sites around Tasmania, although its full expression was sometimes obscured because the range of available data was lim- ited or truncated by intense size-selective fishing on the larger abalone or because of difficulty in finding cryptic smaller abalone. The growth pattern begins with a relatively constant growth increment (implying linear-like growth) in the smaller-size abalone. This early linear-like growth is followed by a steady decline in growth increment, possibly approaching some mini- mum annual increment in what would be an asymptotic fashion. Not only do the growth increments follow this decreasing pattern, but a similar pattern is exhibited by the variability of the observed growth increments around the mean trend, although the decrease in varia- tion only occurs at larger sizes (Fig. 2). This pattern of growth differed markedly from the expectation of both the von Bertalanffy and the Gompertz growth models, even when these models were implemented with prob- ability density functions instead of constant parameters (Sainsbury, 1982a; Troynikov et ah, 1998; Bardos, 2005). Instead of these standard curves, the growth pattern observed indicated that some kind of inverse-logistic curve might describe the observations well across the range of available data. If the minimum predicted mean growth increment was greater than zero, it would indi- cate indeterminate growth in which the dynamics would permit growth to continue (possibly very slowly) until each animal died. An alternative way of looking at indeterminate growth is to note that there may be an upper size limit, at which the growth increment becomes zero, but it is so high that individuals never reach it before death. Comparison of the inverse-logistic, von Bertalanffy, and Gompertz models For the southwest region, the three different growth curves all predicted or described the expected growth of blacklip abalone reasonably well . The overlap between the von Bertalanffy and Gompertz curves was especially close over this range (Fig. 3). However, at smaller and larger sizes all three curves diverged significantly. The von Bertalanffy and Gompertz curves both predicted negative growth increments beyond the Lx (although using a probabilistic version of these curves could pre- vent this problem). The major difference between the three curves was therefore found in what was predicted for smaller abalone. The von Bertalanffy curve predicted linearly decreasing growth increments (Fig. 3) as initial size increased. The Gompertz curve predicted initial exponential growth, starting from very small increments Haddon et al.: Using an inverse-logistic model to describe growth increments of Hahotis rubra 65 for the smallest abalone and reaching a maximum and tailing off as initial length increases. Finally, the inverse- logistic curve predicted constant increments for smaller abalone (initial linear growth) until the growth incre- ments began to decrease with increasing initial length. The minimum AIC and BIC were produced by the 4-parameter inverse-logistic curve and not by the 3- parameter von Bertalanffy and Gompertz curves. The log-likelihoods were -726.1 for the von Bertalanffy, -712.2 for the Gompertz, and -681.1 for the inverse- logistic model. With only one more parameter than the other two models, a likelihood ratio test implies that the inverse-logistic curve was a significant improvement over the other two curves. Annual growth descriptions Within each of the three regional groups of sites, the predicted mean growth increment for given initial shell lengths for the 4-, 5-, and 6-parameter models was very similar (Fig. 4). In the case of the southwest and Actaeon regions, the predicted lines were visually coincident, whereas for Bruny Island there were only very slight differences in the three curves (Table 2). For both the southwest and the Actaeon regions, the 4-parameter model was deemed the optimum model configuration by both the AIC and BIC. For the Bruny Island region, the BIC indicated that the 4-parameter model was optimal and the AIC indicated the 6-param- 66 Fishery Bulletin 106(1 ) Table 2 Alternative annual model structures, with their parameters, for the three regional groups of collection sites. The number after each regional name denotes the number of free parameters fitted to the available data. For site locations see Figure 1 and Table 1. MaxAL is the hypothetical asymptotic maximum growth increment, L'g0 is the initial length at which the midway point between the MaxAL and lowest growth increment is reached, and L'g5 denotes the initial length at which 95% of the difference between the smallest and maximum increment is reached. MaxoL is the hypothetical asymptotic maximum standard deviation, Ls50 and L‘ |5 are the inverse-logistic parameters describing how the variability of residuals decreases with increasing Lt, and Prod Kg is the relative productivity in kilograms derived from the respective transition matrix. See Equations 1 and 2. Model MaxAL, ^”50 J m U 95 MaxaL L%0 ^‘95 Prod Kg Southwest 6 20.393 130.648 164.824 4.461 163.736 214.934 473.5 Southwest 5 20.381 130.669 164.768 4.396 163.735 210 473.1 Southwest 4 20.364 130.688 164.551 4.346 T m U 95 210 472.1 Actaeons 6 23.922 106.084 144.431 4.311 138.679 175.809 308.4 Actaeons 5 23.889 106.136 144.243 4.623 142.934 210 308.6 Actaeons 4 23.873 106.165 144.152 4.551 T m U 95 210 308.1 Bruny Island 6 28.612 119.736 160.142 4.267 151.438 160.876 459.2 Bruny Island 5 28.386 119.893 159.105 4.297 173.763 210 453.3 Bruny Island 4 28.916 119.421 161.336 4.603 T m U 95 210 467.2 Table 3 For each model and parameter combination, the Bayesian information criterion (BIC), Akaike information criterion (AIC), nega- tive log-likelihood (-veLL), and total number of observations n are given. The italicized cells denote the minimum for each criterion and region. The columns labeled “Model 5” and “Model 4” denote the likelihood ratio test values compared to the models in the Model column. The comparisons in Model 5 column had one degree of freedom and for the 5-parameter model to be better than the 4-parameter, it had to be greater than x\ =3.84, and the comparisons in the Model 4 column had two degrees of free- dom and had to be greater than x% =5.99 for the models with more parameters to be significantly better than the 4-parameter model. Model BIC AIC -veLL n Model 5 Model 4 Southwest 6 1395.1 1373.9 680.96 252 0.14 0.18 Southwest 5 1389.7 1372.1 681.03 252 0.04 Southwest 4 1384.2 1370.1 681.05 252 Actaeons 6 2716.3 2690.9 1339.5 500 3.6 3.8 Actaeons 5 2713.7 2692.7 1341.3 500 0.2 Actaeons 4 2707.6 2690.7 1341.4 500 Bruny Island 6 1747.6 1725.5 856.7 295 2.4 4.8 Bruny Island 5 1744.2 1725.8 857.9 295 2.4 Bruny Island 4 1740.9 1726.2 859.1 295 eter model was optimal (Table 3). At the same time, the likelihood ratio test indicated in all cases that the 4-parameter model was not significantly worse than any other model (Table 3; Fig. 3). The key differences that occur between the fitted models relate to how well they describe the trends in variation along the curves. The data for the southwest had the widest size range and the similarity of the fitted L|0 parameter to the Z/§5 parameter was clear. At the same time, the L|5 parameter was only slightly bigger than 210 mm (Table 2). Thus, the substitutions to create the 4-parameter model had little effect on the outcome if this model was used to predict the likely distribution of growth increments (Fig. 5). The main effect of reducing the number of parameters used was to slightly reduce the variation beyond about 170 mm initial shell length. The 6-parameter model has sufficient flexibility in that it can accurately describe both the mean growth trend and the pattern of variation around the expected mean growth increments. This was not necessarily an advantage when the high level of fishing mortality ap- plied to legal-size abalone means that the availability of larger size abalone in the tag return data can be Haddon et at: Using an inverse-logistic model to describe growth increments of Haliotis rubra 67 4-Parameters 6-Parameters Figure 5 Plots of the mean predicted growth increments (solid black line), and a set of simulated data derived from the optimum model in each case (fine dots). The left hand panel represents the 4-parameter model; the right hand panel represents the 6-parameter models. All predicted nega- tive increments were omitted. For each area the curves were fitted to the same data as those illustrated in Figure 4. limited and only provided a biased perception of the growth of larger abalone. With the southwest data the difference between the 4- and 6-parameter models was slight; however, larger differences were found when the 4- and 6-parameter models were applied to the other two regions, which were less well represented in the larger initial sizes (Fig. 4). Most obviously, at Bruny Island (Fig. 5), the 6-parameter model so closely de- scribed the available data that beyond about 160 mm, the model predicted essentially no variation around the predicted mean growth increments. Although the 6-parameter model accurately described the available data, the 4-parameter model provided a more realistic representation of the spread of predicted growth incre- ments for initial sizes for which there were few or no observed data (Fig. 5). Of the three regions considered, the southwest and Bruny Island were similar in terms of relative produc- tivity and the Actaeon regional sites were the least productive (Table 2). The parameter combinations for the southwest and Bruny Island regions were rather different, but although the MaxAL was 8 mm larger at Bruny Island than in the southwest, it was offset in the southwest by the higher value for Ln^0. The two sites 68 Fishery Bulletin 106(1 ) constituting the Actaeon region samples were within relatively low productivity areas that may be consid- ered to be atypically low for the Actaeon area (see the seasonal analysis). Table 4 Seasonal growth-description parameter estimates. All parameter definitions are given with Equation 1 under the heading “Growth model” in the Materials and meth- ods section. In the Actaeon Island estimates L|0 and Ls95 reached the limits placed on the parameters. At Sterile Island, the parameter estimates obtained did not change when the L' |0 and L |5 estimates were replaced with 209 and 210, respectively. In all cases the logistic reduction in the variance was negligible. The -veLL is the nega- tive log-likelihood. Productivity, in kilograms, is the rela- tive productivity derived from the respective transition matrix. Parameter Actaeon Island Sterile Island MaxAL 21.8464 20.0111 ^50 129.9702 121.7953 T m ^ 95 162.1723 161.4016 MaxaL 6.7129 6.8766 ^50 209 186.9362 210 192.3000 C 0.1123 0.0907 P 0.1263 0.1561 C, -0.0821 -0.0489 Productivity Kg 484.2 419.4 -veLL 1249.673 944.564 Oct- Oct- Oct- Oct- Oct- Oct- Oct- Oct- Oct- Oct- Oct- 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 Figure 6 A comparison of the implied seasonal growth curves at the Actaeon and Sterile Islands in southern Tasmania. The fine horizontal dashed line at 138 mm is the current legal minimum length. The curves remain approximately linear until about 100 mm shell length. The slowest growth periods line up approximately with October each year. Seasonal growth descriptions Seasonal growth descriptions were fitted to data from two sites in southeast Tasmania: Actaeon Island and Sterile Island (Table 1; Fig. 2). In each case, when the inverse-logistic curve describing the standard deviation of the residual errors was fitted, the estimates of L|0 and Ls95 were both much larger than the maximum size observed. In effect, this result implied that the vari- ability was best described with Equation 7 rather than Equation 3. For this reason, the 7-parameter model was preferred to the 9-parameter model, even though the optimum fit in each case lead to the same results in terms of parameter estimates and log-likelihood. The two sites are very close together geographically (~2 km apart) and, in this case, the parameter estimates were similar between the two sites (Table 4). By assuming a starting size of 0.25 mm on the 25th November (a typical size and date of settle- ment for a newly settled blacklip abalone) in an arbitrary year, and by calculating the expected size increment in a series of 8-day steps forward in time, adding those increments to the initial size and then repeating the process, it was possible to visualize the seasonal growth of animals (Fig. 6). The main difference in the parameters is in the L^0, which was 8.0 mm larger at Actaeon Island (Table 4). This difference led to the abalone at Actaeon Island reaching the current minimum legal length of 138 mm in approximately eight years, whereas the difference led to the current minimum legal length in nine years, on average, at Sterile Island (Fig. 6). The Actaeon Island site was far more productive than Gagens Point and the Middle Ground, both in the Ac- taeon region and, in fact, was as productive as both the Bruny Island and southwest regions (Tables 2 and 4). The Sterile Island site had a productivity that was only 89% of that of the Actaeon Island site (431 kg vs. 485 kg; Table 4), this was also reflected in a one year difference in time to legal minimum size (Fig. 6). The pattern of seasonality was very similar be- tween the two sampled sites; linear-like initial growth proceeded at least until 100 mm shell length. The phase parameter, p, value of 0.118 implies that the fastest period of growth occurred on 11th Febru- ary at the Actaeon Islands, whereas at Sterile Island the estimate of 0.156 indicates the fastest growth occurred two weeks later on 25th February. Discussion Comparison of inverse-logistic, von Bertalanffy, and Gompertz models The major difference between the inverse-logistic growth description, the von Bertalanffy growth curve, and the Gompertz growth curve is seen in the model fit at the extremes of the growth trajectory. Without Haddon et al.: Using an inverse-logistic model to describe growth increments of Haliotis rubra 69 a probabilistic interpretation of the parameters, the von Bertalanffy and Gompertz curves predict negative growth increments at initial lengths greater than Lx, and the inverse-logistic predicts ever decreasing growth increments as the initial length at tagging increases. Thus, a realistic representation of the final size distribu- tion of larger abalone is provided by the inverse-logistic model without the complexity of a probabilistic inter- pretation of the parameters. In Tasmania, the growth of small abalone, at least above 10 mm, appears to be linear-like and to increase in relatively constant growth increments through time (Prince et al., 1988; Gurney et al., 2005). Although the von Bertalanffy and Gomp- ertz equations can approximate linear-like growth over these small sizes, linear growth limits their capacity to describe accurately the growth of larger animals at the same time. The von Bertalanffy curve predicts a linear relationship between growth increment and initial shell length. The Gompertz equation, on the other hand, pre- dicts that small abalone would have very small growth increments that initially increase with initial length and then decline again. Neither of these alternatives is consistent with observations in Tasmania of linear-like early growth. Growth pattern The tagging data on growth increments of blacklip aba- lone from various sites around the south of Tasmania were able to be grouped according to similarity of growth pattern. All regions exhibited a similar pattern of mean growth increments that were well described by a sym- metric inverse-logistic curve. Negative growth increments observed in the tagging data were not taken to be evidence of negative growth, but were rather taken to be a reflection of measurement or recording errors, a possible chipping of shell edges during collection, or an increased chance of shell ero- sion in disturbed animals (or a combination of these possibilities). Because of this, when simulating growth, negative increments were not included. The tagging data were, in some cases, truncated ei- ther in the smaller or the larger sizes. The fishing mor- tality rate on legal-size abalone is high and numbers of animals much larger than the minimum legal length are significantly reduced. In addition, the cryptic nature of undersize abalone means that obtaining representa- tive data across the whole size range can be difficult. It is also possible that the tagging process could influence the subsequent rate of growth. Intuitively, if there were an impact, it would probably be a negative bias on the growth increments that would increase the variation ob- served (by extending growth into smaller increments). Despite the limitations of the data, the proposed sys- tem of two linked inverse-logistic curves proved capable of fitting and simulating data from three sites in south- ern Tasmania. The inverse-logistic model was fully capable of producing transition matrices with predicted values across the full range of size classes required (60 mm to 210 mm). The truncation of the available data by high levels of fishing pressure did have effects, however. Surprisingly, the more complex 6-parameter model did not always provide the most workable de- scription of growth because the fit with six parameters could over-emphasize missing data; that is, the absence of data could influence the fitted curve, especially when the flexibility of the 6-parameter curve was used. It can be argued that the simpler 4-parameter model provides a more useful description of growth because it is less likely to be influenced by peculiarities or limitations of the available data. For example, at the legal minimum size limit (136 mm shell length at the time of data collection), abalone from the Bruny Island region were growing an average of 8 mm per annum (with a range from 2 mm to 15 mm). However, fishing mortality rates at Bruny Island were very high and few legal-size ani- mals remained for long at this site with the result that the tagging growth increment data were sparse above 140 mm. The 6-parameter model describes the specific pattern of growth in the data from Bruny Island, trun- cating any growth beyond the maximum size available, whereas the 4-parameter model extrapolates the growth pattern beyond the maximum size available in the data and, in this case, provides a much more plausible solu- tion. For stock assessment purposes, the 4-parameter model would be more useful in practice. The symmetry of the inverse-logistic curve enables the 4-parameter model to project the growth dynamics into size classes for which there are few or no samples. Seasonal growth The independent samples from Actaeon Island and Sterile Island, which are close together geographically, generated very similar estimates of the timing of the seasonal changes in growth rates. These samples were so similar that the mean curves remained close until the abalone reached about 80 mm shell length. The abalone at Actaeon Island, however, continued growing rapidly for longer than the animals at Sterile Island; therefore these curves diverged. An implication of this difference in productivity is that instead of taking about 8 years to reach the legal minimum length, as at Actaeon Island, it takes 9 years at Sterile Island. The blacklip abalone at the Middle Ground and Ga- gens Point sites were selected by local abalone divers as having notoriously slow growth. Compared to Actaeon Island, these two sites in the Actaeon Island region did indeed have relatively low productivity compared to the seasonal sample from Actaeon Island (only about 308 kg relative to 484 kg); and this occurred despite the MaxAL being about 23 mm for the Middle Ground and Gagens Point sites but only 21 mm at Actaeon Island. Productivity was strongly and positively correlated with the Ln^0 and the L'g5 parameter values, which relate to how long the linear-like growth phase continues. The different sites in the Actaeon region are all rela- tively close together geographically and yet variation in the parameter estimates and consequent productivity among sites were high. This result is consistent with 70 Fishery Bulletin 106(1 ) the idea that abalone growth is likely to be determined by local site-specific influences in addition to regional scale influences (McShane and Naylor, 1995; Naylor et ah, 2006). Thus, although sites within the Actaeon region were variable, similarities were evident between sites located in distant regions (e.g., in the southwest and Bruny Island). This variability in growth has ob- vious implications for the confidence with which it is possible to conduct stock assessments for abalone over large areas. The description of growth in size-structured models is so influential that the interpretation of any model outputs would need to be made with great atten- tion paid to any potential biases brought about by us- ing an under- or over-productive description of growth. Estimates of productivity derived from the inverse- logistic description of growth would be expected to lie somewhere between that predicted by the von Berta- lanffy curve and the Gompertz curve. The von Berta- lanffy curve predicts very rapid early growth and so, all other things being equal, would predict the highest productivity levels, whereas the Gompertz curve pre- dicts very slow early growth and thus would predict the lowest productivity. These differences are why the selection of the most appropriate model of growth is critical for stock assessments. For blacklip abalone in Tasmania the inverse-logistic model provides the most realistic representation of the dynamics of growth. Acknowledgments The authors thank the array of divers who contributed to the field work, especially S. Dickson, J. Bridley, T. Karlov, C. Jarvis, and M. Porteus. Some of the diving was undertaken from off the RV Challenger, crewed by M. Francis and J. Gibson. We also thank F. Helidoniotis for assistance when preparing the map. Literature cited Bardos, D. C. 2005. Probabilistic Gompertz model of irreversible growth. Bull. Math. Biol. 67:529-545. Breen, P. A., S. W. Kim, and N. L. Andrew. 2003. A length-based Bayesian stock assessment model for the New Zealand abalone Haliotis iris. Mar. Freshw. Res. 54:619-634. Burnham, K. P., and D. R. Anderson. 2002. Model selection and multimodel inference: A practical information-theoretic approach, 2nd ed., 488 p. Springer, New York, NY. Day, R. W., and A. E. Fleming. 1992. The determinants and measurement of abalone growth. In Abalone of the world. Biology, fisheries and culture; proceedings of the 1st international symposium on abalone (S. A. Shepherd, M. J. Tegner, and S. A. Guzman Del Proo, eds.) p. 141-168. Fishing News Books, Carlton, Victoria, Australia. Fabens, A. J. 1965. Properties and fitting of the von Bertalanffy growth curve. Growth 29:265-289. Francis, R. I. C. C. 1988. Maximum likelihood estimation of growth and growth variability from tagging data. NZ. J. Mar. Freshw. Res. 22:42-51. Francis, R. I. C. C. 1995. An alternative mark-recapture analogue of Sch- nute’s growth model. Fish. Res. 23:93-111. Gompertz, B. 1825. On the nature of the function expressive of the law of human mortality, and on a new model of determining the value of life contingencies. Philos. Trans. R. Soc. Lond. Ser. B.. Biol. Sci 115:513-585. Gorfine, H., B. Taylor, M. Cleland, M. Haddon, A. Punt, D. Worthington, and I. Montgomery. 2005. Development of a spatially-structured model for stock assessment and TAG decision analysis for Austral- ian abalone fisheries, 42 p. Fisheries Research and Development Corporation. Primary Industries Research Victoria, Queenscliff, Victoria. Gurney, L. J., C. Mundy, and M. C. Porteus 2005. Determining age and growth of abalone using stable oxygen isotopes: a tool for fisheries management. Fish. Res. 72:353-360. Haddon, M. 2001. Modelling and quantitative methods in fisheries, 406 p. CRC Press, Chapman and Hall, Boca Raton, FL. McShane, P. E., and M. G. Smith. 1992. Shell growth checks are unreliable indicators of age of the abalone Haliotis rubra (Mollusca, Gastropoda). Aust. J. Mar. Freshw. Res. 43:1215-1219. McShane, P. E., and J. R. Naylor. 1995. Small-scale spatial variation in growth, size at maturity, and yield- and egg-per-recruit relations in the New Zealand abalone Haliotis iris. NZ. J. Mar. Freshw. Res. 29:603-612. Naylor, J. R., N. L. Andrews, and S. W. Kim. 2006. Demographic variation in the New Zealand abalone Haliotis iris. Mar. Freshw. Res. 57:215-224. Prince J. D., T. L. Sellers, W. B. Ford, and S. R. Talbot. 1988. Recruitment, growth, mortality and population structure in a southern Australian population of Haliotis rubra (Mollusca, Gastropoda). Mar. Biol. 100:75-82. Punt, A. E., and R. B. Kennedy. 1997. Population modelling of Tasmanian rock lobster, Jasus edwardsii , resources. Mar. Freshw. Res. 48: 967-980. Quinn, T. J., and R. B. Deriso. 1999. Quantitative fish dynamics, 542 p. Oxford Univ. Press, Oxford. Sainsbury, K. J. 1982a. Population dynamics and fishery management of the paua, Haliotis iris I. Population structure, growth, reproduction, and mortality. NZ. J. Mar. Freshw. Res. 16:147-161. Sainsbury, K. J. 1982b. Population dynamics and fishery management of the paua, Haliotis iris II. Dynamics and management as examined using a size class population model. NZ. J. Mar. Freshw. Res. 16:163-173. Schnute, J. 1981. A versatile growth model with statistically stable parameters. Can. J. Fish. Aquat. Sci. 38:1128-1140. Sullivan, P. J., L. Han-Lin, and V. F. Gallucci. 1990. A catch-at-length analysis that incorporates a Haddon et al.: Using an inverse-logistic model to describe growth increments of Haliotis rubra 71 stochastic model of growth. Can. J. Fish. Aquat. Sci. 47:184-198. Troynikov, V. S., and H. K. Gorfine. 1998. Alternative approach for establishing legal mini- mum lengths for abalone based on stochastic growth models for length increment data. J. Shell. Res. 17:827-831. Troynikov, V. S., R. W. Day, and A. M. Leorke. 1998. Estimation of seasonal growth parameters using a stochastic Gompertz model for tagging data. J. Shell- fish Res. 17:833-838. von Bertalanffy, L. 1938. A quantitative theory of organic growth (Inquiries on growth laws. II). Human Biol. 10:181-213. Worthington, D. C., R. C. Chick, C. Blount, P. A. Brett, and P. T. Gibson. 1998. A final assessment of the NSW abalone fishery in 1997. Fish. Res. Assess. Ser. 5:1-33. 72 Abstract — Fisheries management actions taken to protect one species can have unintended, and sometimes positive, consequences on other spe- cies. For example, regulatory mea- sures to reduce fishing effort in the winter gillnet fishery for spiny dog- fish ( Squalus acanthias) off North Carolina (NC) also led to decreases in the number of bycaught bottlenose dolphins (Tursiops truncatus). This study found that a marked decrease in fishing effort for spiny dogfish in NC also corresponded with a marked decrease in winter stranding rates of bottlenose dolphins with entanglement lesions (P=0.002). Furthermore, from 1997 through 2002, there was a sig- nificant positive correlation (r2 = 0.79; P=0.0003) between seasonal bycatch estimates of bottlenose dolphins in gill nets and rates of stranded dol- phins with entanglement lesions. With this information, stranding thresholds were developed that would enable the detection of those increases in bycatch in near real-time. This approach is valuable because updated bycatch estimates from observer data usu- ally have a time-lag of two or more years. Threshold values could be used to detect increases in stranding rates, triggering managers immediately to direct observer effort to areas of potentially high bycatch or to institute mitigation measures. Thus, observer coverage and stranding investigations can be used in concert for more effec- tive fishery management. Manuscript submitted 20 April 2007. Manuscript accepted 30 October 2007. Fish. Bull. 106:72-81 (2008). The views and opinions expressed or implied in this article are those of the author and do not necessarily reflect the position of the National Marine Fisheries Service, NOAA. Effects of commercial fishing regulations on stranding rates of bottlenose dolphin (Tursiops truncatus ) Barbie L. Byrd (contact author) Aleta A. Hohn Email address for B. L. Byrd: Barbie.Byrd@noaa.gov National Marine Fisheries Service Southeast Fisheries Science Center National Oceanic and Atmospheric Administration, Beaufort Laboratory 101 Pivers Island Road Beaufort, North Carolina 28516 Fentress H. Munden North Carolina Division of Marine Fisheries 3441 Arendell Street Morehead City, North Carolina 28557 Gretchen N. Lovewell Rachel E. Lo Piccolo National Marine Fisheries Service Southeast Fisheries Science Center National Oceanic and Atmospheric Administration, Beaufort Laboratory 101 Pivers Island Road Beaufort, North Carolina 28516 The occurrence of beach-cast or stranded marine animals has been used to indicate fishery-induced (i.e., bycatch) mortality of marine birds (Salzman, 1989), turtles (Caillouet et ah, 1991; Epperly et ah, 1996), and mammals (Forney et al., 2001; Fried- laender et al., 2001). Direct documen- tation of bycatch mortality is obtained by placing trained observers on com- mercial fishing vessels (Edwards and Perrin, 1993; Epperly et al., 1995), but limited resources allow for observa- tion of only a small proportion of fish- ing trips and a few types of fisheries. Additionally, updated bycatch esti- mates can take years to become avail- able, preventing real-time responses to significant changes in bycatch rates. Strandings of marine animals, there- fore, can serve as the primary, and sometimes the only, evidence of cur- rent bycatch mortality. Gear is rarely present on stranded animals; however, entanglement lesions on the epidermis of cetaceans can help identify animals that have been captured incidentally by fishing gear (Kuiken et ah, 1994; Read and Murray, 2000). Early indications of bottlenose dol- phin ( Tursiops truncatus) bycatch mortality off North Carolina (NC) came from stranding data. From 1993 through 1996, 29% of the 230 stranded bottlenose dolphins recov- ered in NC exhibited signs of en- tanglement in fishing gear (War- ing et ah, 1997). Early observer data (1993-96) were inconsistent with stranding data because only one entanglement was documented in the observer program (Waring et ah, 1997). As a result, observer cov- erage was expanded in 1997 to in- clude more of the various ocean-side gillnet fisheries (Waring et ah, 1999). The annual estimated bycatch mortality in ocean gill nets from No- vember 1995 through October 2000 confirmed high levels of mortality of bottlenose dolphins off NC. All but one observed entanglement was that of the coastal morphotype, which is morphologically and genetically dis- Byrd et al.: Effects of commercial fishing regulations on stranding rates of bottlenose dolphin ( Tursiops truncatus) 73 79°0'0"W 78°0'0"W 77°0'0"W 76°0'0"W 75“0'0"W Figure 1 The coastal bottlenose dolphin ( Tursiops truncatus ) is divided into seasonal manage- ment units (MUs). During summer ( May-October), two of the management units (MUs) occur off North Carolina ( NC ; ) the northern NC MU and the southern NC MU. During winter (November-April), two summer MUs overlap with a third MU, the northern migratory MU, which occurs north of the Virginia-NC border during the summer. These three MUs are referred to collectively as the winter mixed MU (Waring et al., 2006). Solid horizontal lines represent latitudinal boundaries of MUs and does not imply offshore (i.e., longitudinal) distribution. The dashed horizontal line represents the northern boundary of the NC portion of the winter mixed MU. tinct from the offshore morphotype (Mead and Pot- ter, 1995; Hoelzel et al., 1998; Waring et al., 2002). The bycatch estimates for the coastal morphotype were stratified according to current stock structure of coastal bottlenose dolphins, which consists of seven seasonal management units (MUs) (Waring et al., 2002). Three of the MUs are seasonal off NC: the summer (May-Oc- tober) northern NC MU, the summer southern NC MU, and the NC portion of the winter (November-April) mixed MU (see Fig. 1 for delineations of the units). Bycatch exceeded the potential biological removal (PBR) level (i.e., the sustainable anthropogenic mortality level) (MMPA 16 U.S.C. 1362 [20] ; Barlow et ah, 1995) for one of the two summer MUs and for the winter MU (Waring et al., 2002, 2006). During the summer, the annual esti- mated bycatch for the northern NC MU was 23 animals, exceeding the PBR level (20), and the annual estimated bycatch for the southern NC MU was zero, not exceed- ing the PBR level (10). For the winter mixed MU (NC and VA submanagement units), the annual estimated bycatch was 180 animals, more than twice the PBR lev- el (68). The majority of this bycatch (146 out of 180 ani- mals) was attributed to the NC submanagement units (Rossman and Palka1). The spiny dogfish (FAO common name: picked dogfish) fishery was the primary contribu- tor to the bycatch mortality in the winter mixed MU. In 2005, new annual bycatch estimates, based on observer data from ocean gill nets from November 2000 through October 2002, became available (Rossman and Palka1; Waring et al., 2006). The new bycatch estimate for the summer northern NC MU decreased to eight animals per year and the new estimate for the NC win- 1 Rossman, M. C., and D. L. Palka. 2005. A review of coastal bottlenose dolphin bycatch mortality estimates in relation to the potential effectiveness of the proposed BDTRP. Bottle- nose Dolphin Take Reduction Team Document No. 1-13-05F, 9 p. National Marine Fisheries Service, Northeast Fisheries Science Center, 166 Water Street, Woods Hole, MA 02543. 74 Fishery Bulletin 106(1 ) ter mixed MU decreased to 19 animals per year; both estimates were below their corresponding PBR level. Reductions in bycatch estimates were attributed to a reduction in fishing effort, as measured in landings. In particular, fishing effort was drastically reduced for spiny dogfish, which was listed as overfished by the National Marine Fisheries Service (NMFS) in 1998 (Federal Register, 1998). Fishery Management Plans (FMPs) were implemented by NMFS for federal waters (Federal Register, 2000a, 2000b), and by state agencies for state waters (ASMFC2), to reduce fishing effort. The purpose of this study was to conduct a post-hoc analysis of bottlenose dolphin strandings in NC in re- lation to fisheries bycatch estimates and spiny dogfish landings. First, the frequency of stranded dolphins ex- hibiting signs of fishery entanglement was examined to determine if this frequency reflected corresponding levels of estimated dolphin bycatch. Second, it was hy- pothesized that the frequency of those strandings would decrease concomitant with a reduction in spiny dogfish landings, but that the frequency of stranded dolphins without signs of entanglement would not change. Lastly, two methods for establishing stranding threshold lev- els were evaluated to determine if they could be used in real-time to detect increases in fisheries bycatch before revised bycatch estimates are available or when observer programs do not exist. Materials and methods Fishing-effort data Monthly landings data on spiny dogfish caught in com- mercial ocean gill nets off NC from November 1997 through April 2005 were obtained through the Trip Ticket Program of the North Carolina Division of Marine Fisheries (NCDMF). These data were used to determine the timing and magnitude of effort reduction in the spiny dogfish fishery for comparison with the frequency of bottlenose dolphin strandings. Stranding data Data were derived from ocean-side bottlenose dolphin strandings in NC between November 1997 and April 2005 (n = 580) and were stratified by season: winter (November-April) and summer (May-October). These seasons reflect both the seasonal definition for bottlenose dolphin MUs and the two commercial fishing seasons for spiny dogfish as defined by the FMP November 1997 was chosen as the beginning of the winter season because this month marked the beginning of the first spiny 2 ASMFC (Atlantic States Marine Fisheries Commis- sion). 2006. Review of the Atlantic States Marine Fish- eries Commission’s interstate fishery management plan for spiny dogfish ( Squalus acanthias) May 2004-April 2005 fishing year, 18 p. Prepared by the Spiny Dogfish Plan Review Team, ASMFC, 1444 Eye Street, NW, 6th Floor, Washington, DC 20005. dogfish season in NC (November-April) for which there was consistent coast-wide coverage of the NC shore for strandings. April 2005 was the end of the last season for which landings data were available for this study. All reported stranded bottlenose dolphins were eval- uated for signs of human interaction (HI) and then classified as Hl-yes (i.e., with signs of HI), Hl-no (i.e., no signs of HI), or HI-CBD (could not be determined) (Kuiken et al., 1994; Read and Murray, 2000). Stranded dolphins categorized as Hl-yes were further stratified as fishery interaction (HI-FI) (e.g., entanglement lesions or gear present) or Hi-other (e.g., mutilation, propeller wounds evident). All stranded dolphins classified as Hi-other in our data set (n=12) were mutilated but too decomposed to determine if entanglement lesions were also present; therefore, they were treated separately. Animals were categorized as HI-CBD when it could not be determined whether or not the animal exhibited signs of HI because of factors such as decomposition, significant damage by scavengers, or lack of experience on the part of the stranding responder. Several criteria were established for the stranding re- cords used in this study. Animals genetically confirmed as being the offshore morphotype (?i = 6) were excluded so that comparisons could be made to bycatch data of coastal bottlenose dolphins. Animals <119 cm in total length (>i = 109), presumed to be neonates (Fernandez and Hohn, 1998), were also excluded to prevent a bias from the high natural mortality rates of neonates dur- ing the spring and fall birthing seasons (Hohn, 1980; Thayer et al., 2003). Unless they were classified as adults, stranded dolphins for which no total length was recorded were excluded (n= 21). Dolphins removed from gear other than a gill net (e.g., trawlers, crab pots, hook-and-line gear) were also excluded (n- 5). Stranding rates through time were examined in rela- tion to bycatch estimates and changes in fishing effort in the spiny dogfish fishery. Regression analyses (SAS, vers. 9.1, SAS Inst., Inc., Cary, NC) were used to com- pare the number of HI-FI strandings per season (winter and summer) per year to the corresponding bycatch estimates for ocean gill nets provided in Rossman and Palka.1 Because stranding rates can never be less than zero, the regression line was forced through the origin. Rank-sum tests for each HI category were used to deter- mine if the mean number of bottlenose dolphin strand- ings per month was different between the first time pe- riod (TP1: November 1997-October 2000), when bycatch estimates were greater than the PBR levels, and the second time period (TP2: November 2000-April 2005), when bycatch estimates were either less than PBR levels or were unknown. For the rank-sum tests, only winter data (November-April) were used after a preliminary investigation of spiny dogfish landings revealed that the fishery operates only off NC during those months. Stranding thresholds Two methods were used to calculate stranding thresh- olds. One calculation emulated a method currently Byrd et at: Effects of commercial fishing regulations on stranding rates of bottlenose dolphin ( Tursiops truncatus ) 75 8 CO Figure 2 Spiny dogfish ( Squalus acanthias) landings in metric tons (denoted by bars) and numbers of bottlenose dolphin ( Tursiops truncatus) strandings (denoted by a line) classified as HI-FI (i.e. , having evidence of fishery interaction) in North Carolina from November 1997 through April 2005. Shaded areas represent winter months (November-April). Asterisks (*) represent months when minimal landings occurred. No bars or asterisks represent months with no landings. Monthly strandings in winter decreased between time period (TP) 1 and TP2, delineated by a dashed vertical line. used to help detect unusual mortality events (UME) for overall strandings by the Marine Mammal Health and Stranding Response Program (Wilkinson, 1996) and was termed the “UME threshold method.” It was calculated as the mean number of strandings (in this case HI-FI) per month plus two standard deviations (SD). Stranding thresholds were calculated for each bottlenose dolphin MU in NC. For the NC winter mixed MU, data collected in TP2 were used. The stranding threshold was then compared to monthly HI-FI strandings during TP1 and TP2 to determine whether it serves as an adequate indi- cator of relative bycatch levels. For the summer northern NC MU, the stranding threshold also was calculated with data collected in TP2. For the summer southern NC MU, the stranding threshold was calculated with data from TP1 and TP2 because estimated bycatch levels never exceeded PBR levels in TP1. The second method that was investigated to establish stranding threshold levels was based on the regression analysis of seasonal HI-FI strandings and estimated bycatch. This method used the maximum likelihood es- timates to calculate the predicted values of bycatch and the 68% confidence intervals (CIs) and 95% CIs. The Cl values of predicted bycatch rates were then evaluated to determine if they would be appropriate for identifying periods of elevated bycatch. Results From November 1997 through April 2005, NC gillnetters landed 6310 t (metric tons) of spiny dogfish. Landings occurred almost entirely in winter (November-April), and less than 0.1% occurred in other months (Fig. 2). More than 96% of all landings occurred before Novem- ber 2000 (TP1). During winter, mean landings were 2020 t (SD = 561) per fishing season during TP1 and 49 t (SD = 104) during TP2. After November 2000, 96% of these landings occurred during the 2003-04 fishing year. During the same time period (November 1997 through April 2005), 439 bottlenose dolphin strandings met the criteria for inclusion in this study. Overall, more strand- ings occurred during winter than summer in each HI category (Table 1). For all years, HI-CBD strandings comprised 60% of winter (range: 45—69%) and 52% of summer (range: 25-65%) totals (Hl-yes, Hl-no, and HI-CBD). HI-FI strandings comprised 22% of winter (range: 11-35%) and 21% of summer (range: 11-33%) totals for all years. However, of strandings for which it was possible to determine whether an interaction occurred (HI-FI, Hi-other, and Hl-no), HI-FI strand- ings comprised 56% (range: 27-75%) of winter and 44% (range: 20-57%) of summer totals for all years. Rates of HI-FI strandings had a similar pattern to that of bycatch estimates and effort in the spiny dogfish fishery. There was a significant positive relationship be- tween the number of HI-FI strandings and the bycatch estimate per season (r2 = 0.79, P=0.0003). Additional- ly, the mean number of winter HI-FI strandings per month was significantly greater during TP1 than TP2 (P=0.001) (Table 2). There was no significant difference in winter Hl-no or HI-CBD strandings between TP1 and TP2. HI-FI strandings showed a monthly periodicity similar to that for fishing effort during TP1 (Fig. 2); four to six animals were recovered per month during the height of the fishery, compared to generally two or 76 Fishery Bulletin 106(1 ) Table 1 Numbers of bottlenose dolphin ( Tursiops truncatus) recovered ocean-side in North Carolina between November 1997 and April 2005. Strandings are listed by winter (W) (November-April), summer (S) ( May-October), and all months (T), and categorized according to the human interaction (HI) classification: Hl-yes (evidence of human interaction including HI-FI [evidence of fishery interaction], and Hi-other [evidence of mutilation, propeller wounds]), Hl-no (no signs of HI), and HI-CBD (human interaction could not be determined). For this study, data were not available (n/a) for the 2005 summer season (May-October) and thus totals for 2005 are for a partial year, denoted by an asterisk. HI -yes HI-FI Hi-other Hl-no HI-CBD Total W S T W S T W S T W S T W S T 1998 (Nov 97-Oct 98) 11 8 19 4 0 4 7 6 13 18 14 32 40 28 68 1999 (Nov 98-Oct 99) 18 4 22 0 0 0 6 5 11 27 17 44 51 26 77 2000 (Nov 99-Oct 00) 14 2 16 2 3 5 7 5 12 36 8 44 59 18 77 2001 (Nov 00-Oct 01) 3 4 7 1 0 1 7 3 10 17 7 24 28 14 42 2002 (Nov 01-Oct 02) 9 3 12 0 0 0 6 4 10 32 10 42 47 17 64 2003 (Nov 02-Oct 03) 5 4 9 1 1 2 6 4 10 24 3 27 36 12 48 2004 (Nov 03-Oct 04) 6 3 9 0 0 0 2 4 6 18 9 27 26 16 42 2005 (Nov 04-Apr 05)* 3 n/a n/a 0 n/a n/a 6 n/a n/a 12 n/a n/a 21 n/a n/a Total 69 28 97 8 4 12 47 31 78 184 68 252 308 131 439 less during TP2. One notable exception was in October 1998, when seven strandings occurred but minimal spiny dogfish landings were reported. Monthly Hl-no and Hi-other strandings showed no seasonal pattern across months and years (Fig. 3). Monthly HI-CBD strandings, however, showed a similar pattern to HI- FI strandings during TP1 and continued to show some periodicity afterwards, with increases primarily during winter months. With the UME threshold method, the stranding threshold for HI-FI strandings for the NC winter mixed MU was 2.95. This threshold was exceeded 7 out of 18 months of the winter spiny dogfish fishing seasons dur- ing TP1, or 39% of the time (Fig. 4). During TP2, the threshold was exceeded as a result of bycatch in other fisheries in three months, or only 10% of the time: No- vember 2001 from strandings south of Cape Lookout; April 2003 from strandings north of Cape Hatteras, and November 2004 from strandings south of Cape Lookout. During summer, the stranding threshold produced by the UME threshold method for the northern NC MU was 1.77, and the strand- ing threshold for the south- ern NC MU was 2.19. The stranding threshold was exceeded three times for the northern NC MU, once during TP1 in May 1999 (6% of the time; 1 out of 18 months) and twice during TP2 in October 2001 and October 2004 (8% of the time; 2 out of 24 months) (Fig. 4). For the southern NC MU, the stranding threshold was exceeded only once, in October 1998 dur- ing TP1 (2% of the time; 1 out of 42 months) (Fig. 4). With the use of the re- gression equation, the 68% Cl and 95% Cl values had wide bounds around the predicted bycatch estimates Table 2 Mean (standard deviation [SD]) of monthly bottlenose dolphin (Tursiops truncatus) strandings by human interaction (HI) categories in the winter (November-April) during time period (TP) 1 (November 1997-October 2000) (n = 18) and TP2 (November 2000- April 2005) (n = 30). The HI categories are as follows: HI-FI (evidence of fishery interac- tion), and Hi-other (evidence of mutilation, propeller wounds), Hl-no (no signs of HI), and HI-CBD (human interaction could not be determined). For this study, data were not available (n/a) for the 2005 summer season (May-October) and thus totals for 2005 are for a partial year, denoted by an asterisk. HI category Time period Mean (SD) per month P-value HI-FI TP1 2.39 (1.72) *0.001 HI-FI TP2 0.87 (1.04) Hi-other TP1 0.33 (0.69) 0.11 Hi-other TP2 0.07 (0.25) Hl-no TP1 1.11 (1.02) 0.54 Hl-no TP2 0.90 (0.80) HI-CBD TP1 4.50 (3.31) 0.16 HI-CBD TP2 3.43 (2.45) Byrd et al.: Effects of commercial fishing regulations on stranding rates of bottlenose dolphin ( Tursiops truncatus) 77 Figure 3 Number (n), mean (x), and standard deviation (SD) of stranded bottlenose dolphins ( Tur- siops truncatus) recovered per month in North Carolina from November 1997 through April 2005 for each human interaction (HI) category: HI-FI (i.e., evidence of fishery interaction) or Hi-other (e.g., evidence of mutilation, propeller wounds), Hl-no (i.e., no signs of HI), and HI-CBD (human interaction could not be determined). Shaded areas represent winter months (November-April). The vertical dashed line delineates the two time periods (TPs), TP1 and TP2. Monthly HI-FI strandings increased during winters of TP1, but were more diffuse during TP2. Numbers of stranded dolphins classified as Hi-other and Hl-no showed little variability among months and years. Numbers of monthly HI-CBD strandings were variable; increased rates were evident generally during winter months. 78 Fishery Bulletin 106(1) (Fig. 5). For example, the predicted bycatch for two HI- FI strandings per season was 14 animals, but may have equaled between -23 to 51 animals (68% Cl) or -65 to 93 animals (95% Cl). The lower Cl bounds are nega- tive statistically; however, in reality they cannot be less than the number of HI-FI strandings recovered. (/) CD C TD C CO CO X o 6 <$> c£> <£> $£ 5^* 5^ C?' & & & v-4 ^ ^ ys' ^ ^ Figure 4 Number of bottlenose dolphin (Tursiops truncatus) strandings classified as HI-FI (i.e., having evidence of fishery interaction) per month and the stranding threshold (mean + 2 standard deviations; horizontal dashed lines) by seasonal management unit (MU) in North Carolina (NC). During winter (November-April), the thresh- old was exceeded seven out of 18 months during time period (TP) 1 and only three out of 30 months during TP2. During summer (May-October), the threshold was exceeded once during TP1 and twice during TP2 for the northern NC MU, and once during TP1 for the southern NC MU. The vertical dashed lines indicate separation between TP1 and TP2. Discussion This study provides a unique situation in which three con- current data sets can be used to test the model of using strandings as an indicator of fishery bycatch. It was dem- onstrated that fisheries reg- ulations can affect the level of dolphin bycatch mortality and that increases in bycatch mortality can be detected in near real-time by monitoring changes in stranding rates. There was a significant posi- tive correlation between sea- sonal HI-FI strandings and bycatch estimates of bottle- nose dolphins in gill nets. That correlation was mir- rored by a marked decrease in winter stranding rates of bottlenose dolphins with entanglement lesions coinci- dent with a marked decrease in the fishing effort for spiny dogfish off NC. Many factors can influence the rate of deposition of dead dolphins on beaches. For ex- ample, the overall increases in the number of strandings during winter compared to summer are likely due, in part, to an increase in local bottlenose dolphin abundance when three MUs overlap off the NC coast (Waring et ah, 2006). Within winter, strand- ing rates of HI-FI strandings were further influenced by changes in fishing effort for spiny dogfish between TP1 and TP2 (Table 2) rather than to changes in environ- mental factors such as wind direction and currents. This finding was consistent with the reduction in bycatch esti- mates (Rossman and Palka1) and the nonsignificant dif- ference for Hl-no strandings between the two time peri- ods. There were reductions, Byrd et al.: Effects of commercial fishing regulations on stranding rates of bottlenose dolphin ( Tursiops truncatus) 79 albeit insignificant, in the mean strandings per month for the Hi-other and HI-CBD categories between TP1 and TP2. However, stranding rates for these categories could have been influenced by bycatch reductions. All stranded dolphins categorized as Hi-other were found either with missing appendages, cuts on the abdomen, or both, but they were too decomposed for a determi- nation of whether or not entanglement lesions were present. Fishermen occasionally cut appendages from a marine mammal to aid in the removal of the animal from their nets, or they slit the abdomen to aid in the sinking of the carcass, or do both (Kuiken et al., 1994; Read and Murray, 2000). It is likely, therefore, that a portion of the Hi-other stranded dolphins were indeed entangled in fishing gear because of the mutilations they exhibited. Of the HI-CBD strandings, an unknown proportion was likely caused by fishing interaction, but decomposition obscured evidence of entanglement lesions. Reductions in bycatch would decrease rates of strandings categorized as Hi-other and HI-CBD be- cause of the portion of them that were really HI-FI but could not be identified as such. Because the rate of HI-FI strandings is proportional to the number of bycaught animals, stranding rates may be used as a proxy to detect increases in bycatch mortality and to determine a threshold for triggering a management response. The Cl values calculated with the regression analyses were too broad for this method to be useful for setting threshold values even though the r-squared (coefficient of determination) value was high (0.79). The UME threshold method proved more useful; the stranding threshold value for the NC win- ter mixed MU adequately identified elevated stranding levels during months when the spiny dogfish fishery was most active and bycatch levels were highest. Fur- thermore, with the UME threshold method we were able to identify other periods of elevated strandings during winter and summer. The stranding thresholds calculated by the UME threshold method, while informative, have some limita- tions. First, stranding data between October 2002 and April 2005 were included in the calculations, but the corresponding bycatch estimates were not yet avail- able. If new bycatch estimates for that time period exceeded PBR levels, then stranding thresholds would need to be recalculated after eliminating the corre- sponding stranding data. New thresholds may result in the identification of other months that exceed the re- vised stranding threshold because the current thresh- old would be biased upward. Another limitation of a stranding threshold is the level at which increased bycatch is apparent as HI-FI strandings. The bycatch estimate for years with an active spiny dogfish fishery was more than 7.5 times greater than bycatch esti- mates for years after an active fishery (Rossman and Palka1); our method of calculating a threshold may be too conservative in that it may not detect increases in strandings soon enough. Alternative methods to deter- mine thresholds may be more sensitive and could be investigated by using this or a similar data set that c No. of HI-FI strandinas oer season Figure 5 Predicted bycatch (solid line) of bottlenose dolphin ( Tur- siops truncatus) in ocean gill nets, with 95% confidence intervals (CIs; dashed lines) and 68% CIs (dotted lines) by using HI-FI strandings (i.e., having evidence of fish- ery interaction) per season (winter: November-April, summer: May-October) and their corresponding bycatch estimates from observer data (dots). has periods of estimated bycatch that are greater and lesser than the PBR level. Once an appropriate threshold is established and ex- ceeded, a series of response actions can be triggered, as is done for UMEs (Wilkinson, 1996). Most importantly, active fisheries in the area would need to be identi- fied. The possibilities for response actions then would vary from immediately increasing observer coverage in these fisheries to implementing emergency fish- ing regulations to reduce mortality (MMPA 16 U.S.C. 1387 [ 1 18] ) such as gear modifications, time and area closures, or limited soak durations. The advantage of an increase in observer coverage is to not only increase the precision of the bycatch estimate but to also docu- ment fishing practices and determine if practices have changed in a way that may be affecting the level of bycatch. Stranding data provide valuable information about fisheries bycatch if there is consistent, thorough de- termination of human interaction and comprehensive coverage of shorelines to establish baseline data and to detect changes. For example, stranding data have served as indicators of bycatch in fisheries that do not have federal observer coverage, such as crab-pot, stop- net, pound-net, and inshore gillnet fisheries (Steve et al., 2001; Waring et al., 2002). Additionally, stranding data provide additional in- formation about bycatch in gillnet fisheries that have low observer coverage. Although observers have not documented a bottlenose dolphin entanglement in the gillnet fishery for spot (FAO name: spot croaker; Leios- tomus xanthurus), stranding data indicated that bycatch had occurred. In 1997 and 1998, more than 50% of 80 Fishery Bulletin 106(1 ) stranded bottlenose dolphins that were found spatially and temporally concurrent with the spot fishery exhib- ited entanglement lesions (Friedlaender et al., 2001). In the current study, stranding thresholds were exceeded twice coincident in time (October and November) and place (south of Cape Lookout) with the typical gillnet season for spot (Steve et al., 2001), spanning both MU seasons. The disparity between bycatch and stranding data likely is due, in part, to low observer effort in nearshore waters of southern NC where the spot fishery is most active (Rossman and Palka1). In 2006, NMFS implemented an Alternative Platform Observer Program in NC whereby observers use an independent vessel to find and observe gill nets fished from small boats, which are commonly used in nearshore waters (Kolkmeyer et al., 2007) but difficult for traditional observers to get onboard. There is no observer program, however, for recreational gill nets, which are not regulated by the MMPA (MMPA 16 U.S.C. 1362 [20] ) but are commonly used in NC to target spot (NCMFC, 2003); thus, it is not known if or at what level bycatch occurs in the rec- reational fishery. Given the rate of HI-FI strandings, it is reasonable to assume that the PBR level was ap- proached or exceeded because of mortality in the spot fishery during some years. However, one importance of a threshold value is to represent periods when fishery- related mortality does not exceed levels the population can sustain (i.e., PBR levels). In the case of the spot fishery, it is likely that the threshold values used in the present study are too high. Establishing threshold values is an iterative process, whereby values are ad- justed according to changes in either PBR or bycatch estimates. Management actions for the spiny dogfish fishery had unintentional but beneficial consequences on the by- catch of bottlenose dolphins. State and federal regula- tions severely decreased fishing effort off the NC coast (Federal Register, 2000b; ASMFC2), essentially closing the NC fishery in November 2000. For the 2003-04 season, NC was allowed a 227-t quota of spiny dogfish from state waters, about 7% of the average annual landings in NC before November 2000. Fishing effort occurred almost exclusively in January and February of 2004. Only two HI-FI stranded dolphins occurred in these months, one of which was wrapped in a large- mesh gill net (20.3-cm stretch mesh) more indicative of the striped bass ( Morone saxatilis ) fishery than spiny dogfish fishery (Steve et al., 2001). No bycatch was reported by federal observers on fishing vessels dur- ing the 2003-04 spiny dogfish fishery off NC (Ross- man and Palka1). The soak times were shorter in the 2003-04 season than during previous years (Rossman and Palka1) due to trip limits imposed by the NCDMF and these likely contributed to a lower bycatch rate. Quota shares were not allocated on a state-by-state basis for the next fishing year (May 2004-April 2005) (ASMFC2) and, as a result, landings of spiny dogfish in NC were almost nonexistent. Independently, managers enacted fisheries regulations for spiny dogfish that inadvertently decreased bottle- nose dolphin bycatch. The opposite situation conceivably could occur; that is, fisheries regulations could alter fishing practices or effort in a manner that could in- crease dolphin bycatch. Gillnetters in NC are dynamic, altering their fishing practices in response to a vari- ety of factors including changes in fishery regulations (Steve et al., 2001). Thus, researchers and managers need to be proactive, working towards managing spe- cies as an interrelated community and considering how regulations for one species may affect others. These analyses indicate that, at least in some situa- tions, strandings can serve as a near real-time indica- tor of fishery bycatch. Absolute estimates of bycatch mortality must be obtained using observer data, but the multi-year time lag associated with obtaining those es- timates prevents real-time mitigation of that mortality. Near real-time detection of increased bycatch can also be used to direct observer effort to areas of potentially high bycatch. Thus, observer coverage and stranding investigations can be used in concert for more effective management. Acknowledgments We are grateful for the dedicated participants in the North Carolina Marine Mammal Stranding Network. We also thank the North Carolina Division of Marine Fisheries, Statistics Division, for providing NC fisheries data and P. Rosel (National Marine Fisheries Service [NMFS], Southeast Fisheries Science Center [SEFSC], Layfayette, LA) for providing results for the stranding database on the six strandings genetically confirmed as the offshore morphotype. L. Hansen, M. Prager, and D. Vaughan (NMFS/SEFSC, Beaufort, NC), M. Rossman (NMFS/Northeast Fisheries Science Center, Woods Hole, MA), M. Scott (Inter-American Tropical Tuna Commis- sion, La Jolla, CA), and three anonymous reviewers provided thoughtful comments on the manuscript. E. Griffith (contract employee with NMFS/SEFSC) assisted with statistical analyses. Literature cited Barlow, J., S. L. Swartz, T. C. Eagle, and P. R. Wade. 1995. U.S. Marine Mammal Stock Assessments: Guide- lines for Preparation, Background, and a Summary of the 1995 Assessments. NOAA Tech. Memo. NMFS- OPR-95-6, 73 p. Caillouet, C. W., Jr., M. J. Duronslet, A. M. Landry Jr., D. B. Revera, D. J. Shaver, K. M. Stanley, R. W. Heinly, and E. K. Stabenau. 1991. Sea turtle strandings and shrimp fishing effort in the northwestern Gulf of Mexico, 1986-89. Fish. Bull. 89:712-718. Edwards, E. F., and C. Perrin. 1993. Effects of dolphin group type, percent coverage, and fleet size on estimates of annual dolphin mortality derived from 1987 U.S. tuna-vessel observer data. Fish. Bull. 91:628-640. Byrd et al. : Effects of commercial fishing regulations on stranding rates of bottlenose dolphin ( Tursiops truncatus) 81 Epperly, S. P., J. Braun, A. J. Chester, F. A. Cross, J. V. Merriner, and P. A. Tester. 1995. Winter distribution of sea turtles in the vicinity of Cape Hatteras and their interactions with the summer flounder trawl fishery. Bull. Mar. Sci. 56:547-567. Epperly, S. P., J. Braun, A. J. Chester, F. A. Cross, J. V. Merriner, P. A. Tester, and J. H. Churchill. 1996. Beach strandings as an indicator of at-sea mortal- ity of sea turtles. Bull. Mar. Sci. 59:289-297. Federal Register. 1998. Magnuson-Stevens Act provisions; overfished fishery for spiny dogfish. Notification of an overfished fishery. Federal Register 63(691:17820-17821. Office of the Federal Register, National Archives and Records Administration (NARA), College Park, MD. 2000a. Fisheries of the northeastern United States; spiny dogfish fishery management plan. Final rule. Federal Register 65(7):1557-1571. Office of the Federal Reg- ister, NARA, College Park, MD. 2000b. Fisheries of the northeastern United States; spiny dogfish fishery. 2000 specifications. Interim final rule. Federal Register 65(871:25887-25891. Office of the Federal Register, NARA, College Park, MD. Fernandez, S., and A. A. Hohn. 1998. Age, growth, and calving season of bottlenose dolphins, Tursiops truncatus, off coastal Texas. Fish. Bull. 96:357-365. Forney, K. A., S. R. Benson, and G. A. Cameron. 2001. Central California gillnet effort and bycatch of sensitive species, 1990-1998. In Seabird bycatch: trends, roadblocks, and solutions (E. F. Melvin, and J. K. Parrish, eds.), p. 141-160. Univ. Alaska Sea Grant College Program, Fairbanks, AK. Friedlaender, A. S., W. A. McLellan, and D. A. Pabst. 2001. Characterising an interaction between coastal bottlenose dolphins (Tursiops truncatus) and the spot gillnet fishery in southeastern North Carolina, USA. J. Cetacean Res. Manag. 3:293-303. Hoelzel, A. R., C. W. Potter, and P. B. Best. 1998. Genetic differentiation between parapatric ‘near- shore’ and ‘offshore’ populations of the bottlenose dolphin. Proc. R. Soc. Lond. 265:1177-1183. Hohn, A. A. 1980. Age determination and age related factors in the teeth of western north Atlantic bottlenose dolphins. Sci. Rep. Whales Res. Inst. 32:39-66. Kolkmeyer, T., B. Guthrie, B. L. Byrd, and A. A. Hohn. 2007. Report on the alternative platform observer program in North Carolina: March 2006 to March 2007. NOAA Tech. Memo. NMFS-SEFSC-558, 20 p. Kuiken, T., V. R. Simpson, C. R. Allchin, P. M. Bennett, G. A. Codd, E. A. Harris, G. H. Howes, S. Kennedy, J. K. Kirkwood, R. J. Law, N. R. Merrett, and S. Phillips. 1994. Mass mortality of common dolphins (Delphinus delphis) in south west England due to incidental capture in fishing gear. Vet. Rec. 134:81-89. Mead, J. G., and C. W. Potter. 1995. Recognizing two populations of the bottle- nose dolphin (Tursiops truncatus) off the Atlantic coast of North America: Morphological and ecologi- cal considerations. Int. Mar. Biolog. Res. Inst. Rep. 5:31-44. National Museum of Natural History, Smith- sonian Institution, Washington, DC. NCMFC (North Carolina Marine Fisheries Commission). 2003. North Carolina fisheries rules for coastal waters, 297 p. North Carolina Department of Environment and Natural Resources, North Carolina Division of Marine Fisheries, Morehead City, NC. Read, A. J., and K. T. Murray. 2000. Gross evidence of human-induced mortality in small cetaceans. NOAA Tech. Memo. NMFS-OPR-15, 21 p. Salzman, J. E. 1989. Scientists as advocates: the Point Reyes Bird Obser- vatory and gill netting in central California. Conserv. Biol. 3:170-180. Steve, C., J. Gearhart, D. Borggaard, L. Sabo, and A. A. Hohn. 2001. Characterization of North Carolina commercial fisheries with occasional interactions with marine mammals. NOAA Tech. Memo. NMFS-SEFSC-458, 57 p. Thayer, V. G., A. J. Read, A. S. Friedlaender, D. R. Colby, A. A. Hohn, W. A. McLellan, D. A. Pabst, J. L. Dearolf, N. I. Bowles, J. R. Russell, and K. A. Rittmaster. 2003. Reproductive seasonality of western Atlantic bottle- nose dolphins off North Carolina, USA. Mar. Mamm. Sci. 19:617-629. Waring, G. T., E. Josephson, C. P. Fairfield, andK. Maze-Foley (eds.). 2006. Bottlenose dolphin ( Tursiops truncatus): western North Atlantic coastal stock. In U.S. Atlantic and Gulf of Mexico marine mammal stock assessments — 2005. NOAA Tech. Memo. NMFS-NE-194, p. 89-99. Waring, G. T., D. L. Palka, P. J. Clapham, S. Swartz, M. C. Ross- man, T. V. N. Cole, L. J. Hansen, K. D. Bisack, K. D. Mullin, R. S. Wells, D. K. Odell, and N. B. Barros. 1999. Bottlenose dolphin ( Tursiops truncatus): western North Atlantic coastal stock. In U.S. Atlantic and Gulf of Mexico marine mammal stock assessments — 1999, p. 128-135. NOAA Tech. Memo. NMFS-NE-153, 187 p. Waring, G. T., D. L. Palka, K. D. Mullin, J. H. W. Hain, L. J. Hansen, and K. D. Bisack. 1997. Bottlenose dolphin ( Tursiops truncatus): western North Atlantic coastal stock. In U.S. Atlantic and Gulf of Mexico marine mammal stock assessments — 1996. NOAA Tech. Memo. NMFS-NE-114, p. 128-133. Waring, G. T., J. M. Quintal, and C. P. Fairfield (eds.). 2002. Bottlenose dolphin ( Tursiops truncatus): western North Atlantic coastal stock. In U.S. Atlantic and Gulf of Mexico marine mammal stock assessments — 2002. NOAA Tech. Memo. NMFS-NE-169, p. 169-180. Wilkinson, D. M. 1996. National contingency plan for response to unusual marine mammal mortality events. NOAA Tech. Memo. NMFS-OPR-9, 188 p. 82 Age, growth, and reproduction of dolphinfish ( Coryphaena hippums ) caught off the coast of North Carolina Email address for K. L. Schwenke: Kara.Schwenke@noaa.gov Department of Zoology Center for Marine Sciences and Technology North Carolina State University 303 College Circle Morehead City, North Carolina 28557 Present address (for K. L. Schwenke): National Ocean Service (NOS) National Oceanic and Atmospheric Administration (NOAA) 1305 East West Highway Silver Spring, Maryland 20910 Abstract— Age, growth, and repro- ductive data were obtained from dolphinfish (Coryphaena hippurus, size range: 89 to 1451 mm fork length [FL]) collected between May 2002 and May 2004 off North Carolina. Annual increments from scales (« = 541) and daily increments from sagittal otoliths (n = 107) were examined; estimated von Bertalanffy parameters were L„ (asymptotic length) = 1299 mm FL and k (growth coefficient) = l ,08/yr. Daily growth increments reduced much of the residual error in length-at-age estimates for age-0 dolphinfish; the estimated average growth rate was 3.78 mm/day during the first six months. Size at 50% maturity was slightly smaller for female (460 mm FL) than male (475 mm FL) dolphin- fish. Based on monthly length-adjusted gonad weights, peak spawning occurs from April through July off North Carolina; back-calculated hatching dates from age-0 dolphinfish and prior reproductive studies on the east coast of Florida indicate that dolphinfish spawning occurs year round off the LT.S. east coast and highest levels range from January through June. No major changes in length-at-age or size-at-maturity have occurred since the early 1960s, even after substan- tial increases in fishery landings. Manuscript submitted 21 February 2007. Manuscript accepted 6 November 2007. Fish. Bull. 106:82-92(2008). The views and opinions expressed or implied in this article are those of the author and do not necessarily reflect the position of the National Marine Fisheries Service, NOAA. Kara L. Schwenke (contact author) Jeffrey A. Buckel The dolphinfish ( Coryphaena hippurus ) is a highly migratory oceanic pelagic fish found worldwide in tropical and subtropical waters. The distribution range for dolphinfish in the western Atlantic Ocean is from Nova Scotia (Vladykov and McKenzie, 1935; Tibbo, 1962) to Brazil ( Shcherbachev, 1973). However, this species is most common from North Carolina, throughout the Gulf of Mexico and Caribbean, to the northeastern coast of Brazil where it is seasonally abundant (Oxenford, 1999). Dolphinfish support economi- cally important recreational and com- mercial fisheries in the United States, Caribbean, and Brazil, and is thus a shared resource among multiple coun- tries. Previous reviews of the scientific literature on dolphinfish biology in the western Atlantic were completed by Palko et al. (1982) and Oxenford (1999). Landings of dolphinfish from the At- lantic, Caribbean, and Gulf of Mexico have increased. According to the Na- tional Marine Fisheries Service land- ings statistics, recreational landings in the Atlantic Ocean have increased gradually, whereas commercial land- ings in the Atlantic have increased dramatically from approximately 20 metric tons (t) in the 1980s to over 620 t in the 1990s. Although dol- phinfish are fast growing and mature early, concern has been raised about this trend in landings and the poten- tial for localized depletion of stocks. Intense harvesting may select for traits such as slow growth (Conover and Munch, 2002) or early maturity (Trippel, 1995); it is important to up- date growth and reproductive data to test for changes in these data and to provide current information for stock assessments. Unfortunately, the most recent estimates of these parameters for dolphinfish in the southeast Unit- ed States were based on data from the 1960s (Beardsley, 1967; Rose and Hassler, 1968). Here, we update the age and growth relationship and collect re- productive data on dolphinfish cap- tured in North Carolina from recre- ational and commercial sources and fishery-independent collections. Our specific objectives were 1) to deter- mine daily ages of age-0 dolphinfish and determine age-0 dolphinfish growth rates, 2) to identify the best method of aging >age-0 dolphinfish (either by otolith or scale annual marks) and, with the method deter- mined to be the best, to determine the annual ages of >age-0 dolphin- fish, 3) to validate annual marks, and 4) to estimate time of spawning and size-at-maturity. Schwenke and Buckel: Age, growth, and reproduction of Coryphaena hippurus 83 Materials and methods Collections Dolphinfish from recreational fishery sources were obtained every month between May 2002 and May 2004 (except December 2002 and 2003, January 2003, and February 2004) from fishing ports in North Carolina. Recreational anglers typically fished for dolphinfish in waters associated with the western wall of the Gulf Stream. In the summers of 2002-03, samples of large fish were provided through various sportfishing tourna- ments held in these same areas. To supplement length- at-maturity data once peak spawning was identified, maturity staging was conducted on male and female dolphinfish from April 2005 through July 2005. Sampling of commercial dolphinfish catches was done in addition to recreational dolphinfish sampling and was primarily conducted in the winter months to in- crease the sample sizes available for this time period. Small dolphinfish were not readily available through recreational and commercial sampling; therefore sample sizes were augmented by two different methods. First, a total of four fishery-independent trips were made in August 2003 and July 2004. During these trips, the distance traveled offshore averaged 20 km, and small lures were trolled, as opposed to large dead bait or large lures as is done in the recreational and com- mercial fishery. Second, small and intact dolphinfish were obtained from stomachs of larger dolphinfish and yellowfin tuna (Thunnus albacares) caught by anglers from recreational charter boats. Dolphinfish were measured to the nearest mm for fork length (FL) and total length (TL), sex was deter- mined (through macroscopic examination of the gonads), and the fish were weighed (to the nearest 0.1 kg) and tagged. Date and location of port sampled were recorded for each dolphinfish. Scale samples were collected be- fore the fish were filleted according to methods estab- lished by Beardsley (1967). In some instances, filleted dolphinfish carcasses were only available; therefore scale samples were not obtained on all sampled fish. All tagged carcasses were brought to the laboratory for extraction of otoliths and gonads. Age and growth To determine if daily rings were present on sagittal otoliths of age-0 dolphinfish, the otoliths were removed, cleaned, and stored dry until mounted in epoxy resin. To avoid interotolith variability, only the left otolith was used for reading. Otoliths were prepared for read- ing following methods described for transverse sections in Secor et al. (1992). Reading was done with a light microscope equipped with a digital camera. The image from the camera was transmitted to a computer and examined by using Image-Pro Plus software (Image- Pro, vers. 4.5, Media Cybernetics, Silver Spring, MD). Growth increments were counted from the core, begin- ning at the first clearly defined mark that encircled the primordium (Massuti et al., 1999), towards either the dorsal or ventral edge, depending on ease of counting. To determine the precision of the readings of juvenile dolphinfish ages, blind readings of daily growth incre- ments were conducted twice by the same investigator. Error greater than 10% in reading precision for an indi- vidual otolith caused that otolith to be rejected. If error in reading precision was less than or equal to 10%, then the average between the first and second readings was taken as the final age. The deposition of increments in dolphinfish otoliths begins on the hatching date, and rings are laid down daily (Uchiyama et al., 1986; Massuti et al., 1999). Thus, no adjustment was required to estimate age from incremental counts of sagittae, and it was assumed that rings were formed daily. Previous studies on the mi- crostructure of sagittal otoliths of dolphinfish from the western Mediterranean Sea had found that the daily ages from larger dolphinfish (>650 mm FL) appeared to be underestimated (Massuti et al., 1999). Furthermore, daily ages of dolphinfish have been validated to a size of 554 mm FL (Uchiyama et al., 1986). Therefore, our analysis was restricted to dolphinfish less than or equal to 650 mm FL. To determine individual dolphinfish growth rates, the fork length at capture was divided by the daily age. The annual age of dolphinfish was estimated with scales. Eight to ten scales were mounted, sculptured side down, on sheets of cellulose acetate 0.5 mm thick, and then placed on a scale press to make impressions. Scale impressions were examined with a microfiche reader at 32 x magnification to permit detection of cir- culi, annual marks, and other features of the scale. Age groups were classified according to the number of annual marks present (see Beardsley, 1967, for a figure of an annual mark on a dolphinfish scale). To determine the precision of dolphinfish age esti- mates, blind readings of annual marks on scales were conducted twice by the same investigator. If agreement between the first and second reading was not 100%, then a scale was reread a third time and was only used in the analyses if the third reading agreed with either the first or second reading. Additionally, blind read- ings of a subsample (n = 50) of dolphinfish scales were conducted by an independent reader who was trained to identify annual marks on dolphinfish scales. To validate annual marks in dolphinfish scales, an in- direct validation based on marginal increment analysis was used. Marginal increment widths were determined by measuring the distance from the outer edge of the scale to the closest annual mark. Marginal increment width was measured only on dolphinfish with one an- nual mark in order to standardize the method, and because the majority of dolphinfish aged with annual marks were age-1. Measurements (mm) were taken from the magnified (32x) scale image on a microfiche reader along a straight line from the lateral edge of the scale to the outermost annual mark by using a digital caliper. Marginal increment widths were analyzed by analysis of variance (ANOVA) to test for an effect of month. 84 Fishery Bulletin 106(1 ) Marginal increment widths were only analyzed for the months of March through November because of the low sample sizes of dolphinfish scales from the winter months (December through February). To differenti- ate between changes in the marginal increment width attributed to potentially sampling different cohorts of age-1 dolphinfish, we calculated the monthly mean fork length of all age-1 dolphinfish whose scales were measured for a marginal increment width. A subsample (n = 50) of dolphinfish that was deter- mined to be >age-0 by using scales was further exam- ined for the presence of annual marks by using otoliths. We prepared transverse cross sections of sagittal oto- liths using methods described above. These sections were viewed under the light microscope (first at 100 x, then 400 x) to determine if annual marks could be de- tected in these structures. The von Bertalanffy growth curve was fitted to two dolphinfish age-length data sets: 1) daily ages from age-0 dolphinfish with a fork length less than 650 mm and annual marks on scales from >age-0 dolphinfish by using absolute ages, and 2) daily ages from age-0 dolphinfish with a fork length less than 650 mm and relative scale ages. Relative scale ages were assigned by adding the number of days after the fixed birth date of 15 April (middle of estimated southeastern U.S. spawning season) when the dolphinfish was caught to the absolute annual age determined from scales. The 15 April birth date was chosen according to the trends in gonadosomatic indices in Florida and North Carolina and back-calculated hatching dates (Beardsley, 1967; this study). The von Bertalanffy growth parameters were es- timated separately by nonlinear regression for male and female dolphinfish and were compared by using the likelihood-ratio test (Kimura, 1980; Cerrato, 1990; Haddon, 2001). To detect if any significant changes in growth had occurred since the last dolphinfish aging study in North Carolina, the mean size-at-age values from Rose and Hassler (1968) were plotted with the von Bertalanffy growth curve fit (relative age data set) and compared qualitatively. Additionally, von Ber- talanffy growth functions estimated from past stud- ies within different regions were plotted together for comparison. Reproduction Gonadosomatic indices and back-calculated hatching dates were used to determine timing of spawning, and maturity staging was used to determine length-at- maturity. When available, intact gonads were removed, weighed to the nearest 0.1 g, and assigned a maturity stage determined by gross examination of the gonads. Maturity stages for both male and female dolphinfish have been described (Beardsley, 1967; Oxenford, 1985). Female dolphinfish were considered mature or immature on the basis of the criteria developed by Beardsley (1967). Male dolphinfish were classified as mature on the basis of the presence or absence of milt in their gonads. A gonadosomatic index (GSI) was calculated as go- nad weight /(body weight - gonad weight ) separately for male and female dolphinfish pooled for 2002-04. Because dolphinfish body weight and length are cor- related with GSI values (Chatterji and Ansari, 1982) and dolphinfish size differed significantly by month (see below), ANCOVA was used to compare In ( gonad weight) by month with In ( fork length ) as the covariate for males and females separately. Log transformations were used to meet assumptions of ANCOVA. To deter- mine which months had significantly different gonad weights, the length-adjusted mean In ( gonad weight ) value was compared among months for both male and female dolphinfish by using ANCOVA univariate test of significance for planned comparisons. Significance levels were adjusted by the standard Bonferroni technique to account for multiple comparisons. Hatching dates were determined by subtracting age in days (determined from age-0 otoliths) from the catch date. Because the daily deposition of increments in dolphinfish sagittal otoliths begins on the hatching date (Uchiyama et ah, 1986; Massuti et al., 1999), and because ripe eggs hatch within 50-60 hours after fer- tilization (Palko et al., 1982), back-calculated hatching dates provide an estimate of spawning dates for surviv- ing offspring. The length at which 50% of the fish had become ma- ture was determined for both sexes by using a logistic model. The model was fitted by using nonlinear regres- sion analysis based on the following equation: % Maturity - 1/(1 + e (~^ * (L~L50^), where Q = model parameter; L = fork length (mm); and L50 - fork length (mm) at 50% maturity. Results Collections Dolphinfish were collected mostly from the recreational charter fishery (ti = 611, 76%), but also from the com- mercial fishery {n = 45, 6%), sportfishing tournaments (rc = 130, 16%), and from four fishery-independent trips (77 = 16, 3%). There was a seasonal trend in the total amount (number) of dolphinfish collected by month, with nearly half (n- 364, 45%) of all dolphinfish col- lected in the months of June, July, and August. Only 17 fish (2%) were obtained in the months of November, December, and January (Table 1). The majority of the dolphinfish were sampled from catches in Morehead City, NC (ti = 676, 84%). The size range for the pooled sample of dolphinfish was 89 to 1451 mm FL (mean=736 mm FL, standard error (SE) = 9.3). Males (n- 257) ranged in length from 310 to 1451 mm FL (mean length and weight of all males sampled was 855 mm FL [SE = 16.0] and 6.44 kg [SE = 0.4]) and females (n = 422) ranged in length Schwenke and Buckel: Age, growth, and reproduction of Coryphaena hippurus 85 Table 1 Monthly mean (±SE) fork length (FL; mm), FL range, and sample size (n) for male and female dolphinfish (Coryphaena hippurus) collected from May 2002 through May 2004. Sex was unable to be determined for six dolphinfish and their size information is not shown here. No dolphinfish were caught in December. SE = standard error. Male Female Mean FL (SE) FL range n Mean FL (SE) FL range n January 682 (81.1) 559-835 3 708(70.3) 575-966 5 February — — 0 699(62.4) 545-850 4 March 709 (61.3) 453-915 6 730 (17.9) 560-889 20 April 820(26.1) 550-1130 24 773(12.1) 608-1020 50 May 935 (17.4) 582-1315 92 765 (14.2) 485-1275 101 June 1123 (25.4) 552-1395 72 688 (31.8) 295-1145 43 July 705(50.9) 395-1451 31 607 (25.5) 205-980 48 August 805(41.5) 310-1333 46 650(24.1) 295-1205 101 September 689 (51.0) 462-1280 16 510(19.8) 278-800 40 October 594(25.6) 432-798 17 556(26.1) 410-1435 41 November 570(50.0) 520-620 2 628(60.7) 460-905 7 from 205 to 1435 mm (mean length and weight of all females sampled was 655 mm FL [SE = 9.0] and 3.13 kg [SE = 0.2]; Table 1). There were significant differences in male dolphinfish mean weight (Kruskal-Wallis ANOVA: X2=80.6, df=9, P<0.001) and fork length (%2=98.9, df=9, P<0.001) by month (pooled over 2002-04). There were also significant differences in the mean weight of female dolphinfish (Kruskal-Wallis ANOVA: %2 = 85.1, df=9, PcO.OOl) and fork length (x2=140.0, df=10, P<0.001) by month (pooled over 2002-04). Age and growth Because of the small size and complex structure of dolphinfish sagittae, counts were typically made on the dorsal side of the otolith as that region was the easiest to follow a clear increment sequence (Massuti et al., 1999). Alternating light and dark bands, assumed to be daily increments (see Methods section), varied in width; tightly packed increments were located more toward the core and outer edge of the sagittae, and wider increments were located more in the center of the dorsal wing (see Massuti et al., 1999, for a picture of growth increments). A total of 181 dolphinfish otoliths were examined (u = 131 age-0 otoliths, n = 50 >age-0 otoliths). Annual marks could not be detected in transverse cross-sec- tions of sagittal otoliths of >age-0 dolphinfish. Daily increment counts were possible for a total of 107 (82%) otoliths from age-0 dolphinfish (designated age-0 be- cause of a lack of annual marks on scales [see be- low]). Of these, 62 were from female dolphinfish (mean FL = 509; range: 278-650 mm) and 39 were from males (mean FL = 538; range: 310-650 mm). Sex could not be determined for five of the smallest dolphinfish whose sagittae were examined (mean FL=152; range: 89-285 mm) and was not recorded for one of the larger dol- phinfish aged from daily growth increments (FL = 575 mm); however, these dolphinfish were still used in the von Bertalanffy analyses. Four of the 131 otoliths from age-0 dolphinfish were rejected because percent agree- ment between the first and second count exceeded 10%, and 20 age-0 otoliths were unreadable because of prob- lems with cross-sectioning or polishing. Minimum and maximum age estimates ranged from 31 to 204 days. Average growth rates based on daily ring counts were 3.78 mm FL/day for all age-0 fish less than 650 mm FL. Scales were collected from 560 fish; 14 of the result- ing scale impressions were unreadable and five more were discarded because of uncertainty in the determi- nation of age (i.e., the three counts did not agree with each other). A total of 234 scales were classified as age- 1 or older (84 females, 150 males) and the remaining scales (n = 307) were estimated to be age-0. The >age-0 dolphinfish were classified as follows: 175 age-1 dol- phinfish ranging from 575 to 1435 mm FL (mean=938 mm, SE = 9.8), 46 age-2 dolphinfish ranging from 925 to 1451 mm FL (mean = 1197 mm, SE = 17.3), and 13 age-3 dolphinfish ranging from 1095 to 1334 mm FL (mean=1249 mm, SE = 17.9). Final agreement between readings by the same investigator was 99%; an inde- pendent reader who was trained to identify annual marks on dolphinfish scales examined a subsample of 50 >age-0 dolphinfish scales, and agreement between the independent reader’s reading and the first reader’s final age was 69% (Schwenke, 2004). Marginal increment widths from age-1 dolphinfish (rc = 182) were greatest in May, June, and July, dropped slightly in August, then dropped considerably during 86 Fishery Bulletin 106(1 ) Table 2 Von Bertalanffy growth parameters calculated for male, female, and combined sexes (including individuals whose sex could not be determined) of dolphinfish ( Coryphaena hippurus) from nonlinear regression model fits. Data are presented for (A) daily ages of otoliths (from dolphinfish <650 mm) along with annual ages of >age-0 dolphinfish presented as absolute ages, and (B) daily aged otoliths (from dolphinfish <650 mm) along with annual ages of >age-0 dolphinfish presented as relative ages, assuming a 15 April hatching date. The standard errors of each parameter are shown in parentheses. n = sample size, Lco=asymptotic length, £=growth coefficient, and t0=theoretical age at zero length. Method n ^oo (mm) k (1/yr) *0 age-0 dolphinfish. However, the sizes of age-0 dolphinfish (where age was es- timated by using annual marks on scales) at their relative age (Fig. 2B) did not show good agreement with size-at-age based on daily ages determined from otoliths. Thus, a combination of otolith-based daily ages for age-0 dolphinfish and scale-based relative ages for >age-0 dolphinfish were used when fitting a second von Bertalanffy growth model. The second von Bertalanffy growth func- tion showed that males grow faster and reach a larger maximum size than females (likelihood ra- tio test: x2=10.14, df=3, P=0.02) (Fig. 2C; Table 2). By using a biological hatching date, the combined sexes model fit was improved from an r2 of 0.67 to an r2 of 0.73. The mean length-at-age values for dolphinfish collected in June, July, and August Schwenke and Buckel: Age, growth, and reproduction of Coryphaena hippurus 87 Figure 2 Length-at-age data for male (filled circles) and female (open circles) dolphinfish ( Coryphaena hippurus) from (A) annual marks on scales from >age-0 dolphinfish, (B) annual marks on scales from age-0 and >age-0 dolphinfish (with the assump- tion of a 15 April hatching date, and (C) annual marks on scales from >age-0 dolphinfish (with the assumption of a 15 April hatching date). Length-at-age data (A and C) from daily otolith increments for age-0 dolphinfish with fork length <650 mm (male, female, and sex undetermined; open triangles) are presented. Functions A and C are presented for von Bertalanffy model fits to male (solid line), female (gray dashed line), and combined sexes (dark dashed line) length-at-age data (otolith and scale data combined for model fitting). Mean size-at-age data for dolphinfish from Rose and Hassler (1968) are plotted (C; open squares) for comparison with 2002-04 length-at-age values; values from Rose and Hassler (1968) were not used in fitting the von Bertalanffy growth function. VBGF=von Bertalanffy growth function. 1961-62 are shown in Figure 2C (where a mean capture date of 15 July was assumed for all plots; Rose and Hassler, 1968), and are similar to length-at-age values from the present study. Length-at-age data for dolphinfish from past studies from different regions show an apparent trend in re- gional groupings (Fig. 3). The von Bertalanffy growth functions calculated for the Gulf of Mexico and Carib- bean all display faster growth rates than those for Florida, North Carolina, and the Mediterranean, and an average longevity of less than one year. Dolphinfish col- lected from Florida, North Carolina, and the Mediter- ranean Sea all displayed similar first-year growth rates and a maximum age of 3 or 4 years (Fig. 3). However, Mediterranean dolphinfish have a slightly smaller size at age 2 and 3 compared to size for these ages of Florida and North Carolina dolphinfish. Reproduction Males reached 50% maturity at 476 mm FL and 100% maturity was reached at 645 mm FL (Table 3). Females reached 50% maturity at a slightly smaller size than males, although confidence limits for this parameter overlapped with those of males. At 458 mm FL, 50% of female dolphinfish were mature, and 100% were mature at about 560 mm FL. 88 Fishery Bulletin 106(1 ) TabSe 3 Length at 50% maturity for male and female dolphinfish ( Coryphaena hippurus). Lengths were estimated by fitting a logistic model (see text) with nonlinear regression analysis. The standard errors (SE) of each parameter are shown in parentheses. n = sample size, Q = model parameter, L50 = fork length (mm) at 50% maturity, and Cl = 95% confidence interval for L50 Sex n Q(SE) L50(SE) Cl Males 74 0.05(0.02) 476.13(6.24) 460.9-494.7 Females 154 0.08(0.02) 457.58(2.54) 453.1-462.5 The highest median GSI values occurred in May for both male (Fig. 4A) and female (Fig. 4B) dolphinfish; however, these values were not corrected for differences in body size. Length-adjusted mean gonad weights were significantly different by month (ANCOVA: P<0.001) for both male and female dolphinfish (Fig. 4C). Length- adjusted mean gonad weights were highest in the late spring and summer and then decreased from midsum- mer into the fall. Gonad weights from November to February were not included because of the low sample size (n = 9). September gonad weights were significantly lower than May and June gonad weights in males (Fig. 4C); all other male comparisons were nonsignificant. There were significant differences in the length-adjusted gonad weights of females between almost every month, but most differences were found for October, when go- nad weights were significantly lower than in May, June, July, and August (P<0.001 for all) (Fig. 4C). For both 2002 and 2003, hatching dates of dolphin- fish occurred for all months, but the bulk occurred from January to June (Fig. 5). In 2002, the majority of age-0 dolphinfish sampled (83%) had back-calculated hatching dates in the months of January through June. Similarly, in 2003, 76% of age-0 dolphinfish had back- calculated hatching dates for this same period. Discussion Age and growth This study is the first to use transverse cross-sections of sagittal otoliths to determine daily ages of dolphinfish; whole otoliths (Oxenford and Hunte, 1983; Uchiyama et al., 1986; Rivera and Appledoorn, 2000) or exposed sagittal planes (Massuti et al., 1999; Morales-Nin et al., 1999) were used in prior studies. Our estimated birth dates are in good agreement with known spawning dates (Beardsley, 1967; this study); a similar indepen- dent comparison indirectly validated daily age data for dolphinfish in the Mediterranean Sea (Massuti et al., 1999; Morales-Nin et al., 1999). Future work is needed to compare the multiple techniques that have been used to prepare age-0 dolphinfish oto- liths in order to determine which technique is most efficient. The daily growth rates for dolphinfish are faster than those of many species, but are a common characteristic of pelagic pi- scivores (Brothers et al., 1983; La Mesa et al., 2005). Our estimate of daily growth rate (3.78 mm FL/day) is similar to that of 550-1325 mm FL dolphinfish from Puerto Rico (3.59 mm/day; Rivera and Appledoorn, 2000) and 200-600 mm FL dolphinfish from the western Mediterranean Sea (~3.50 mm/day; Massuti et al., 1999). In Barba- dos, the average growth rate of dolphin- fish of 174-1100 mm SL is estimated at 4.71 mm standard length per day (Oxenford and Hunte, 1983). Based on daily growth increments in sagittal otoliths of dolphin- fish from the Gulf of Mexico, average first- year growth rate is 4.15 mm FL/day for fish in the size range of 250-1200 mm SL (Bentivoglio, 1988). Annual marks are not detectable on sagittal otoliths of >age-0 dolphinfish with 0 1 2 3 4 5 Age (years) Figure 3 Calculated von Bertalanffy growth functions (VBGFs) for dolphinfish ( Coryphaena hippurus) from various locations in the North Atlantic. GOM = Gulf of Mexico, FL = Florida, NC = North Carolina. Schwenke and Buckel: Age, growth, and reproduction of Coryphaena hippurus 89 4.5 4.0 • o Females • Males £ 3.5 .o> Q) T3 CT3 C o O) Oab ^AB . 3.0 Qa 2.5 ■ II AB Qa |B || AB Qa 2.0 ^ ^ ^ ^ ^ s0<5 o0' Figure 4 Box plots of gonadosomatic indices (GSI) for (A) males and (B) female dolphinfish ( Coryphaena hippurus) collected from January 2002 through December 2004, and (C) mean (±SE) In (gonad weight) of male (closed circles) and female (open circle) dolphinfish adjusted to a common length for March through October of 2002-04. Like letters for each sex in C indicate no significant difference between months as determined with ANCOVA. The 25th percentile of the GSI data is represented by the boundary of the box closest to zero, in relation to the y-axis, and the 75th percentile is represented by the boundary of the box farthest from zero. The line within the box is the median value. Whiskers (error bars) above and below the box indicate the 90th and 10th percentiles, respectively. Outlying values for both upper and lower ranges are represented by closed circles (in A and B). SE = standard error. Sample sizes are given inside the boxes (in A and B). methods used to date. Massuti et al. (1999) using sagit- tal plane sections did not observe annual marks on dol- phinfish otoliths from the Mediterranean Sea, although the authors speculated that detection of annual marks on the outer edges of adult otoliths may have been hin- dered by otolith preparation. A transverse cross-section approach was used in our study in an attempt to obtain a view of the outer edges, but the technique used in otolith preparation or the complex structure of >age-0 dolphinfish otoliths may have prevented detection of any annual marks. Alternatively, annual marks may not exist on sagittal otoliths of dolphinfish. Validation of scale annuli has been attempted in only a few studies of dolphinfish. Although the annual marks on scales of >age-0 dolphinfish were relatively easy to interpret and within-laboratory agreement of age assignments was good in our study, these features do not establish that the ages are correct. In general, the monthly pattern in marginal increment widths in our study was similar to prior work in Florida (Beardsley, 1967). After measuring the distance between the last annulus and the margin of the scale for all dolphin- fish with one or more year marks, Beardsley (1967) considered November to be the period of annulus for- mation because of an abrupt decrease in width of the increments from October to November. The smallest mean marginal increment in our study occurred dur- ing winter; this finding supports the hypothesis that dolphinfish lay a new annulus in winter as a result of decreased water temperature. The temperature of the 90 Fishery Bulletin 106(1 ) 20 (/) 0 1 10 o> c 'sz o CO f 0 'o 0 CT -10 £ LL -20 ^ ^ ^ age-0 dolphinfish in this study reduces much of the variability associated with length-at-age seen in the von Bertalanffy growth curve where a biological hatching date is not used. Some variability in length- at-age still exists, however, and may be a result of other environmental factors experienced by an individual dolphinfish (i.e., water temperature, differences in prey consumption and prey quality). However, the protracted spawning season is likely the most important factor responsible for variability in length-at-age. Overall, there has been little evidence of changes in size-at-maturity in dolphinfish off the U.S. east coast from the 1960s to the time of our study. Males first begin to mature at a fork length of about 435 mm, which is in agreement with Beardsley’s (1967) observation of first maturity in males at a fork length of 427 mm. A previous estimate of the length at 50% maturity for female dolphinfish caught off Florida in the 1960s (Beardsley, 1967) is nearly identical to our estimate (450 [Beardsley, 1967] vs. 457 mm FL [our study]); however, Beardsley (1967) found earlier first maturity in females (-350 mm FL) compared to our study (-430 mm FL). Summary and implications Age, growth, and reproduction data for dolphinfish caught off the coast of North Carolina are provided. Using scale annual marks and daily growth increments from otoliths, we determined an updated age-length function. Furthermore, comprehensive seasonal esti- mates of gonad weights and marginal increment widths, as well as back-calculated hatching dates and daily growth-rate estimates, are the first for dolphinfish in North Carolina waters. Because this species is highly migratory, a much broader study encompassing the U.S. east coast or western North Atlantic may be needed in order to truly characterize dolphinfish reproduction and marginal increment widths. Direct validation through mark and recapture studies could also confirm annual marks on scales and provide good estimates of growth rates for tagged dolphinfish that remain at large through suspected periods of annulus deposition. Intense positive size-selective mortality can lead to changes in life history parameters (Pitcher and Hart, 1982; Conover and Munch, 2002). However, there have been no changes in size-at-age or size-at-maturity of dolphinfish; therefore, the increased harvests in the 1980s and 1990s have not influenced these life history parameters to date. Because of their fast growth rates and small size-at-maturity, dolphinfish appear an ideal fishery resource species capable of withstanding high rates of fishing mortality. Acknowledgments We thank Captain J. Gay of the FV Old Smokey for his field assistance, R. Gregory and L. Daniel from North Carolina Division of Marine Fisheries for dis- cussions and use of equipment, and J. Arnott and J. Edwards from Center for Marine Science and Technol- ogy (CMAST), North Carolina State University, for extensive help with dolphinfish sampling. We thank P. Rudershausen for his assistance with scale read- ings. P. Rudershausen, J. Hightower, and L. Stefanski reviewed an earlier draft of the manuscript. We also thank B. Morales-Nin, E. Massutf, and J. Moranta from the Mediterranean Institute for Advanced Stud- ies (IMEDEA) in Mallorca, Spain, D. Ahrenholz from the Center for Coastal Fisheries and Habitat Research, National Oceanic and Atmospheric Administration, in Beaufort, North Carolina, and S. Searcy from CMAST for their assistance with our otolith preparation tech- nique. This project was funded by a North Carolina Fishery Resource Grant no. 02-EP-01. Literature cited Beardsley, G. L., Jr. 1967. Age, growth, and reproduction of the dolphin, Cory- phaena hippurus, in the Straits of Florida. Copeia 1967:441-451. 92 Fishery Bulletin 106(1 ) Bentivoglio, A. A. 1988. Investigations into the growth, maturity, mortality rates and occurrence of the dolphin ( Coryphaena hip- purus, Linnaeus) in the Gulf of Mexico. M.S. thesis, 37 p. Univ. College of North Wales, Bangor, UK. Brothers, E. B., E. D. Prince, and D. W. Lee. 1983. Age and growth of young-of-the-year bluefin tuna, Thunnus thynnus, from otolith microstructure. In Pro- ceedings of the international workshop on age determina- tion of oceanic pelagic fishes: tunas, billfish, and sharks (E. D. Prince, and L. M. Pulos, eds.), p. 49-59. NOAA Tech. Rept. NMFS 8. Cerrato, R. M. 1990. Interpretable statistical tests for growth com- parisons using parameters in the von Bertalanffy equation. Can. J. Fish. Aquat. Sci. 47:1416-1426. Chatterji, A., and Z. A. Ansari. 1982. Fecundity of dolphinfish, Coryphaena hippurus L. Mahasagar 15(2):129-133. Conover, D. O., and S. B. Munch. 2002. Sustaining fisheries yields over evolutionary time scales. Science 297:94-96. Haddon, M. 2001. Modelling and quantitative methods in fisheries, 424 p. Chapman and Hall, CRC, Boca Raton, FL. Kimura, D. K. 1980. Likelihood methods for the von Bertalanffy growth curve. Fish. Bull. 77:765-776. La Mesa, M., M. Sinopoli, and F.Andaloro. 2005. Age and growth rate of juvenile bluefin tuna Thunnus thynnus from the Mediterranean Sea (Sicily, Italy). Sci. Mar. 69 (21:241-249. Massuti, E., B. Morales-Nin, and J. Moranta. 1999. Otolith microstructure, age, and growth pat- terns of dolphin, Coryphaena hippurus, in the western Mediterranean. Fish. Bull. 97:891-899. Morales-Nin, B., M. Di Stefano, A. Potoschi, E. Massuti, P. Rizzo, and S. Gancitano. 1999. Differences between the sagitta, lapillus and ver- tebra in estimating age and growth in juvenile Mediter- ranean dolphinfish (Coryphaena hippurus). Sci. Mar. 63(3-41:327-336. Oxenford, H. A., and W. Hunte. 1983. Age and growth of dolphin, Coryphaena hippurus, as determined by growth rings in otoliths. Fish. Bull. 81:906-909. Oxenford, H. A. 1985. Biology of the dolphin Coryphaena hippurus and its implications for the Barbadian fishery. Ph.D. diss., 366 p. Univ. West Indies, Cave Hill, Barbados. Oxenford, H. A. 1999. Biology of the dolphinfish (Coryphaena hippurus ) in the western central Atlantic: a review. Sci. Mar. 63 (3-41:277-301. Palko, B. J., G. L. Beardsley, and W. J. Richards. 1982. Synopsis of the biological data on dolphin fishes, Coryphaena hippurus Linnaeus and Coryphaena equise- lis Linnaeus. U.S. Dept. Commer., NOAA Tech. Rept. NMFS Circ. 443, 28 p. Pitcher, T. J., and P. J. B. Hart. 1982. Fisheries ecology, 414 p. Croom Helm, London. Rivera, G. A., and R. S. Appeldoorn. 2000. Age and growth of dolphinfish, Coryphaena hip- purus, off Puerto Rico. Fish. Bull. 98:345-352. Rose, C. D. 1966. The biology and catch distribution of the dolphin, Coryphaena hippurus (Linnaeus), in North Carolina waters. Ph.D. diss., 153 p. North Carolina State Univ., Raleigh, NC. Rose, C. D., and W. W. Hassler. 1968. Age and growth of the dolphin, Coryphaena hip- purus (Linnaeus) in North Carolina waters. Trans. Am. Fish. Soc. 97:271-276. Schuck, H. A. 1951. Notes on the dolphin (Coryphaena hippurus) in North Carolina waters. Copeia 1951:35-39. Schwenke, K. L. 2004. Age, growth and reproduction of dolphin ( Cory- phaena hippurus) caught off the coast of North Carolina. M.S. thesis, 62 p. North Carolina State Univ., Raleigh. NC. Secor, D. H., J. M., Dean, and E. H. Laban. 1992. Otolith removal and preparation for microstructural examination. In Otolith microstructure examination and analysis (D. Stephenson, and S. Campana, eds.), p. 17-59. Special Publication, Can. J. Fish. Aquat. Sci. 117. Shcherbachev, Y. N. 1973. The biology and distribution of the dolphins (Pisces, Coryphaenidae). J. Ichthyol. 13:182-191. Tibbo, S. N. 1962. New records for occurrence of the white-tip shark, Pterolamiops longimanus (Poey), and the dolphinfish, Coryphaena hippurus L., in the northwest Atlantic. J. Fish. Res. Board Can. 9:517-518. Trippel, E. A. 1995. Age at maturity as a stress indicator in fish- eries. Bioscience 45:759-771. Uchiyama, J. H., R. K., Burch, and S. A. Kraul Jr. 1986. Growth of dolphins, Coryphaena hippurus and C. Equiselis, in Hawaiian waters as determined by daily increments on Otoliths. Fish. Bull. 84:186-191. Vladykov, V. D., and R. A. McKenzie. 1935. The marine fishes of Nova Scotia. Proc. Nova Scotian Inst. Sci. 19:17-113 93 Abstract — To determine if shoreface sand ridges provide unique habitats for fish on the inner continental shelf, two cross-shelf trawl surveys (23 km in length) were conducted in south- ern New Jersey (July and September 1991-95 with a beam trawl and July and September 1997-06 with an otter trawl) to assess whether species abun- dance, richness, and assemblages dif- fered on and away from the ridge. The dominant species collected with both gears were from the families Parali- chthyidae, Triglidae, Gobiidae, Ser- ranidae, Engraulidae, Stromateidae, and Sciaenidae. Overall abundance (n = 41,451 individuals) and species richness (n = 61 species) were distrib- uted bimodally across the nearshore to offshore transect, and the highest values were found on either side of the sand ridge regardless of gear type. Canonical correspondence analysis revealed three species assemblages: inshore (<5 meters depth), near-ridge (9-14 meters depth), and offshore (>14 meters depth), and variation in spe- cies composition between gear types. Environmental factors that corre- sponded with the assemblage changes included depth, temperature, distance from the top of the ridge, and habi- tat complexity. The most abundant near-ridge assemblages were distinct and included economically important species. Sand ridges of the inner con- tinental shelf appear to be important habitat for a number of fish species and therefore may not be a suitable area for sand and gravel mining. Manuscript submitted 20 July 2007. Manuscript accepted 19 November 2007. Fish. Bull. 106:93-107 (2008). The views and opinions expressed or implied in this article are those of the author and do not necessarily reflect the position of the National Marine Fisheries Service, NOAA. Importance of shoreface sand ridges as habitat for fishes off the northeast coast of the United States James M. Vasslides (contact author) Email address: vasslide@marine.rutgers.edu Marine Field Station Institute of Marine and Coastal Sciences Rutgers University 800 do 132 Great Bay Blvd. Tuckerton, New Jersey 08087-2004 Kenneth W. Abie Marine Field Station, Institute of Marine and Coastal Sciences Rutgers University 800 c/o 132 Great Bay Blvd., Tuckerton, New Jersey 08087-2004 Shoreface sand ridges are common features within the inner continental shelf of the northeast Atlantic coast, although the importance of these sand ridges as fish habitat has received little attention. These topographic fea- tures consist of unconsolidated fine- to medium-grain sand, typically have vertical relief up to 10 meters, and are generally oriented obliquely to the adjacent shoreline (Stahl et ah, 1974; McBride and Moslow, 1991). Over 200 shoreface sand ridges have been iden- tified from Montauk Point, New York, to Miami Beach, Florida, and over 71 are found along the coast of New Jersey (McBride and Moslow, 1991). These sand ridges may be impor- tant bathymetric features for com- mercial and recreational fishing areas if they provide important fish habi- tat; however, there is little evidence to refute or support that possibility. Adults, settled juveniles, and larvae of a number of fish species have been documented on sand ridges and in the immediate vicinity of sand ridges, in- dicating that these features are used by multiple fish at various life history stages (Able et al., 2006). Although sand ridges may provide habitat for important fish species, sand ridges from Massachusetts to North Caro- lina, including ridges off New Jersey (Byrnes et al., 2004), have gained attention as potential locations from which to extract sand and gravel for ongoing beach nourishment projects and to provide construction materials (Drucker et al., 2004). Extraction of sand and gravel reduces the complex- ity of a ridge, rendering it similar to the surrounding bottom. Recent research on the effects of sand mining at ridges has focused on physical oceanographic processes (Nairn et al., 2004), and others with a focus on benthic invertebrates and their role in providing trophic sup- port to fishes have provided limited evaluation of the biological response to sand mining (Diaz et al., 2004; Nairn et al., 2004). Although season- al and spatial patterns of aquatic or- ganisms near sand ridges have been examined for the presence of decapod crustaceans (Viscido et ah, 1997) on and near sand ridges off New Jersey and for juvenile fish off Delaware and Maryland (Diaz et ah, 2003), there have been no evaluations of corre- spondingly varying spatial patterns in the fish community or of their causal relationships with sand ridges. If sand ridges provide unique habi- tat within inner continental shelf 94 Fishery Bulletin 106(1 ) Figure t Station locations for the 1991-95 beam trawl surveys and 2001-06 otter trawl surveys off Little Egg Inlet off southern New Jersey. Locations sampled only by beam trawl are denoted by (•), those sampled only by otter trawl, by (A), and those sampled with both gears, by (□). The thick black line delineates the top contour of Beach Haven Ridge. ecosystems, then changes to fish assemblages asso- ciated with changes in the structure of sand ridges could be excellent indicators of the effects of sand and gravel mining or other habitat alterations. Additionally, sand ridges may not be the optimal choice for sand and gravel mining if it is shown that sand ridges act as strategic ecological features (whether in their influ- ence on increased fish abundance or species richness) or provide essential fish habitat (EFH) for economically important species. The purpose of this study was to determine how use of habitat by fish varies between shoreface sand ridges and the surrounding inner con- tinental shelf through an analysis of abundance and species assemblage patterns. The specific objectives were 1) to ascertain if there is a difference in fish abun- dance and species richness between the sand ridge and adjacent areas; 2) to determine if there are spatial or temporal patterns in species assemblages that are dif- ferent between the sand ridge and adjacent areas; and 3) to describe any relationships between the species as- semblages and environmental factors. Collectively, this information can provide resource managers a better understanding of the potential impacts of the mining of sand ridges on fishes. Materials and methods Study area The study area encompassed the inner continental shelf waters off southern New Jersey between Barnegat Inlet and Brigantine Inlet (Fig. 1). Sampling was conducted along a 23-km transect across a shoreface sand ridge, Beach Haven Ridge (BHR). Beach Haven Ridge extends northeastward from the ebb tidal delta of Little Egg Inlet; it has a maximum relief of 8 m between the ridge crest and the trough on the seaward side, and the relief on the shoreward side is 4-5 m (Stahl et al., 1974). The substrate on top of the ridge is composed primarily of coarse sand (Craghan, 1995). The seaward side of the ridge has two major substrate types: 1) coarse sand with shells of the surf clam ( Spisula solidissima) and 2) areas with a mixture of semilithified clay and sand. The landward side of the ridge is characterized as having two major substrate types: 1) areas of sand and clay mixture and 2) patches of semilithified clay and sand mixture. Mounds composed of tubes of the polychaete worm Asa- bellides occulata can also be found landward of the ridge, but are temporally variable. Bedforms (ripples) are Vasslides and Able: Importance of shoreface sand ridges as habitat for fishes 95 Table 1 Physical attributes of individual sampling stations along the Beach Haven Ridge (BHR) transect. Distance to ridge top is the distance (m) from that station to the station located on top of the ridge, plus signs indicate distances seaward of the ridge top, negative signs indicate distances landward of the ridge top. Habitat complexity index is an index of complexity from 1 to 3 and is based on the amount and type of substrate and macroalgae or other structural components present. See Figure 1 for station locations. Distance to Distance Water Habitat Sampling ridge top from shore depth Trawl Number of complexity Station (m) (m) (m) type samples Type of habitat index BHR-1 -6000 0 2.8 Otter 67 Bare sand 1 BHR-2 -4600 300 3.1 Beam 4 Bare sand 1 BHR-3 -3700 750 5.1 Beam 8 Bare sand 1 BHR-4 -3200 1250 4.7 Beam 8 Bare sand 1 BHR-5 -1400 3050 9.4 Beam 10 Sand with patches of macroalgae and Diopatra tubes; clay+silt 3 Otter 66 BHR-6 -500 3950 11.5 Beam 22 Clay and sand with Diopatra tubes; 3 clay+silt Otter 64 BHRTOP 0 4450 10.3 Beam 16 Bare sand 1 Otter 18 BHR-7 1100 5550 13.6 Beam 27 Sand with shell hash and Diopatra tubes; clay/sand 3 Otter 61 BHR-9 5300 9750 16.3 Beam 2 Shell hash 2 Otter 62 BHR-10 10000 14450 18.0 Otter 29 Shell hash 2 BHR-11 19000 23450 19.9 Otter 22 Shell hash 2 consistently largest on the crest and flanks of the ridge. The crests are often bare, but the troughs are filled with varying amounts of shell valves and shell hash, which are frequently buried and uncovered. Although patches of Diopatra cuprea (plumed worm) tubes were found along the flanks and base of the ridge, they were never identified on the crest of the ridge. Field sampling Data from two independent sampling surveys, one with a beam trawl (1991-95), and the other with an otter trawl (1997-2006), were analyzed. Although the data sets were not collected concurrently, they did overlap both temporally (sampling months) and spatially (four sampling stations). This overlap between gear types provided the opportunity to observe temporal and spatial variation in species abundance and richness both within and between gears. Beam-trawl sampling Sampling for fishes was con- ducted at eight stations along a transect from Little Egg Inlet across Beach Haven Ridge with a 2 meter (3-mm bar mesh) beam trawl (Fig. 1; Table 1). Both midsummer (July or the first few days in August) and late summer (September) samples were collected. The number of tows conducted at each station varied from 2 to 22 (Table 1). Tow speed was approximately 2. 8-3. 7 km/h and each tow was one minute in duration in an attempt to sample from discrete habitat types. All fish captured were identified to species where possible and measured to the nearest millimeter. Surface and bottom water samples were obtained with a Nansen bottle. Tempera- ture and salinity were obtained from the water samples by using a stem thermometer and hand-held refractom- eter, whereas oxygen concentrations were determined by using Winkler titration for samples collected from 1991 through 1995. Otter trawl sampling Eight stations on and within the vicinity of Beach Haven Ridge were sampled in mid- summer (July or the first few days in August) and late summer (September) (Fig. 1; Table 1). Samples were collected with an otter trawl (4.9-m head rope, 19-mm mesh wings, 6-mm-mesh codend) in four replicate tows at the inlet station (BHR-1) with various small boats (4-7 meters) and in three replicate tows at each of the deeper stations with RV Arabella (15 meters). Sam- pling at BHR-10 and BHR-11 did not commence until September 2001. Tow speed varied depending on the 96 Fishery Bulletin 106(1 ) prevailing ocean conditions, but duration of tows never exceeded two minutes in an attempt to ensure that fish were collected from discrete habitat types. For each tow, a random selection of up to 20 individuals of each species were measured to the nearest millimeter fork length (FL) or total length (TL), and the remainder were counted. Beginning in 2001, bottom and surface salinity, temperature, dissolved oxygen, and pH were measured with a YSI model 85 handheld dissolved oxygen and conductivity instrument (Yellow Springs Instruments, Yellow Springs, OH) and recorded after the first tow. Depth and bottom topography were determined with a Furuno model 256 video depth recorder (Furuno, Hyogo, Japan). Water transparency was measured with a Secchi disk at each station. Habitat characteristics To assess the importance of habitat complexity to the structure of fish assemblages, the substrate was char- acterized qualitatively. Because Beach Haven Ridge and its vicinity have been intensively studied in the past, it was possible to identify habitat characteristics for each station from previous research where SCUBA, submers- ible, and remotely operated vehicle, or sidescan sonar were used, or where structural samples of the habitat were collected during trawl sampling. Additionally, any benthic material (clay and silt clods, sea stars, sand dollars, algae, shell hash, Diopatra tubes, Asabellides mounds) retained by the 2006 otter trawl sampling was categorized and quantified to evaluate if any changes in substrate had occurred over time. Stations were assigned an index of complexity of 1, 2, or 3 based upon the type and amount of substrate and macroalgae or other structural components present, with 1 being the least complex (bare sand) and 3 being the most complex (one or more substrate types present with multiple biogenic components). Habitat complexity was lowest inshore, peaked on the sides of the ridge, and was of an interme- diate value offshore (Table 1). Stations BHR-5, BHR-6, and BHR-7 had multiple substrate types and structural components and were given an index value of 3. Station BHRTOP and the inlet stations were all dominated by bare sand substrates and had no structures and were assigned a value of 1. The offshore stations varied in their substrates but had some complexity owing to bio- genic features and were assigned a value of 2. In addi- tion, the distance from a sampling location to the station located at the top of the ridge (BHRTOP), as well as the distance from a sampling location to the shore, were determined by using ArcGIS (Environmental Systems Research Institute, Inc., Redlands, CA). Data analyses A number of univariate and multivariate techniques were used to calculate population measures and assem- blage structure. Catch per unit of effort (CPUE), or the number of fish captured per tow, was determined for each species at each station. Significant differences in patterns of CPUE among stations were tested by using ANOVA procedures. CPUE data were log transformed before analysis to correct for heterogeneity of variance (Underwood, 1997). Mean species richness per unit of effort (RPUE), or the number of different species caught in a tow, was also calculated for each sampling sta- tion. ANOVA and Tukey multiple comparison tests were used to assess the differences in mean species richness (RPUE) between stations. Frequency of occurrence was calculated for each species across tows at each sampling station. All univariate statistics were performed with SAS (vers. 9.1, SAS Inst., Inc., Cary, NC). To determine the structure of the assemblages in the different habitats, two related ordination techniques were employed. Canonical correspondence analysis (CCA) is a constrained ordination technique in which the sample scores are constrained to be linear combina- tions of the explanatory variables (Van den Brink and Ter Braak, 1999). This technique is one of the most widely used gradient analysis tools in ecology because of its capacity to handle highly skewed species distribu- tions, high noise levels, complex sampling designs, and the fact that it does not create an artificial arch effect (Palmer, 1993). Canonical correspondence analysis was performed on a subset of the overall data matrix for which environmental information was available for all stations. For the beam trawl data this “subset” was the entire data set, whereas for the otter trawl samples, the subset was limited to Beach Haven Ridge from 2001 to 2006. The data were arranged in a species-by-sample matrix, where the samples were the combination of all tows for a given location and date, and the CPUE rep- resented the value fields. The data were log (CPUE+ 1) transformed to reduce the influence of abundant spe- cies. Any species whose abundance did not exceed five percent from at least one station was removed from the matrix. Because CCA orders species only along gradients of the measured environmental variables, it may not, depending on the environmental variables available, be representative of the assemblages encountered. Therefore, the CCA was checked for bias by using cor- respondence analysis (CA), an unconstrained ordination method (McGarigal et ah, 2000). The species-by-sample matrices were treated in the same manner as in the CCA. Results Environmental gradients Summer temperature tended to decrease with increas- ing distance from the shoreline, and depth, salinity, and water transparency generally increased with increasing distance (Fig. 2). Average station depth increased with increasing distance from the shoreline (range: 2.8-19.9 m), with the exception of the station located on the top of the ridge (BHRTOP) (Table 1). Within the Beach Haven Ridge transect there was a change in depth from BHR-4 Vasslides and Able: Importance of shoreface sand ridges as habitat for fishes 97 to BHR-5 of 4.7 m, indicating the transition from the Little Egg Inlet to nearshore coastal waters. Although a number of previous studies incor- porating a variety of techniques were used to de- termine habitat homogeneity, the similarity in descriptions when the same station was sampled by multiple sources provided confidence in the ac- curacy of the habitat descriptions, as well as the stability of the habitat over time. Species abundance and richness The specific patterns in species abundance and richness varied with sampling gear and habitat. The locations near the ridge typically had the largest number of species and individuals and the inshore, top of ridge, and locations offshore of the ridge had the least, regardless of sampling gear (Fig. 3). Fish abundance was higher in late summer than mid-summer for both gears, but the spatial patterns in species abundance were simi- lar between seasons. There was no difference in species richness between seasons for either gear. Beam trawl catches were dominated by demersal fish families including Triglidae, Gobiidae, and Serranidae, whereas the dominant families found in the otter trawl consisted of Engraulidae, Stro- mateidae, Sciaenidae, Triglidae, and Bothidae (Tables 2-4). In beam-trawl tows during 1991-95 (n- 97), 2049 individuals were collected (primarily de- mersal but also some pelagic fishes) belonging to 34 species. Fish abundance (CPUE) was lowest inshore, increased slightly towards BHR-5, in- creased significantly (df=94, PcO.OOOl) at the sta- tions on either side of the ridge (BHR-6 BHR-7>BHR-6) and significantly lower (df=381, PcO.OOOl) at all other stations (Fig. 3). Fish abundance at the top of the ridge was not significantly different from that at the inshore station or the offshore stations. The beam trawl typically captured smaller individuals (mean = 38.8 mm, standard error [ SE ] = 1 . 2 ) than did the otter trawl (mean=104.1 mm, [SE =1.1] Tables 2-4; Fig. 4). Species richness was highest at the near-ridge sta- tions and decreased offshore across both gear types (Fig. 3). The mean species richness per tow (RPUE) for the beam trawl was significantly higher (df=89, PcO.OOOl) at the sides of the ridge and offshore than at the other beam-trawl stations. The RPUE for the otter trawl was significantly higher (df=312, PcO.OOOl) 32 31 w 30 - Q. 29 Salinity 28 28 24 Q 20 16 12 8.0 76 sd 7-2 U) E 68 64 60 81 8 0 • 7 9 • Q_ 7,8 • 7 7 * 7.6 • ™ 12 CD c 10 « 8 “ E 6 4 8 2 CO 0 Temperature Dissolved oxygen pH Water transparency BHR-t SMR-2 8 HR-3 BBR-4 8HR-5 BHfi-6 BHRTOP BHfi-7 8HR-9 8HR-10 BHR-tl Sampling location Figure 2 Environmental data (mean bottom values) for the 1991-95 beam trawl samples (O) and 2001-06 otter trawl samples (•). Vertical lines represent standard error. Sampling with a beam trawl at BHR-2 was conducted in midsummer 1995 only and at BHR-9 in midsummer 1993 only. Otter-trawl data at BHRTOP was collected in 2005 and 2006 only. Dissolved oxygen, pH, and a Secchi reading were not recorded at BHR-2, BHR-3, or BHR-4 during sampling with a beam trawl. See Figure 1 for station locations. at the near-ridge stations than at all other remaining stations. Fish-assemblage structure based on beam-trawl samples Canonical correspondence analysis revealed no dis- cernible pattern in fish-assemblage structure at the stations in midsummer (Fig. 5A), which is in contrast to late summer when there were two distinct, discrete assemblages: the inshore locations (BHR-2, BHR-3, BHR-4, BHR-5) and the near-ridge+offshore stations (BHR-6, BHRTOP, BHR-7, BHR-9) (Fig. 5B). However, within the midsummer samples there was variability in the species associated with each station between years (Fig. 5C). In late summer the near-ridge+offshore 98 Fishery Bulletin 106(1 ) stations shared a number of dominant species, whereas the assemblage of inshore stations varied along both axes. The species composition at each inshore station was not only different from the near-ridge+offshore stations but also from each other (Table 2, Fig. 5D). Within the inshore group, BHR-3 was characterized by American sand lance (Ammodytes americanus ) and gobies (Gobiosoma spp.), and BHR-5 was differentiated as the only station with weakfish ( Cynoscion regalis). The remaining inshore stations were differentiated from other stations, and to some degree each other, by the abundance of Atlantic croaker ( Micropogonias undulatus ) and kingfish (Menticirrhus spp.). Less than half (39%) of the species identified were captured in both assemblages (Table 2), and no species were shared among the top five species loading scores for each assemblage. BHR1 BMR 2 BMR 3 BHR 4 BHR 5 BHR 6 BHRTOP BMR 7 BHR 9 BHR ID BUR 15 Sampling locations Figure 3 Mean abundance per tow (CPUE) and mean number of species per tow (RPUE) by sampling location for the otter trawl surveys (A and C) in 2001-06 and beam-trawl surveys ( B and D ) in 1991-95. Vertical lines represent standard error. Sampling locations with different superscripts are significantly different from each other at the alpha=0.05 level. See Figure 1 for station locations. Over 60% of the variance in the species-environment interaction was reflected in both the mid- and late sum- mer ordinations (Table 5). Temperature and habitat complexity were the dominant environmental variables shaping the late summer ordination (Table 6). The ar- rangement of the species assemblages in relation to the station assemblages was similar to that produced with correspondence analysis (CA), providing confidence that constrained ordination gave a satisfactory picture of realized distribution. Fish-assembiage structure based on otter-trawl samples Three assemblages were apparent in the otter-trawl data regardless of season: inshore (BHR-1), near-ridge (BHR-5, BHR-6, BHRTOP, and BHR-7), and offshore (BHR-9, BHR-10, and BHR-11) (Fig. 6). Seasonally, the near-ridge and offshore assemblages overlapped little in midsummer (Fig. 6A) and were discrete in late summer (Fig. 6B). In midsummer, the near- ridge assemblage was spread along both axes, indicating differences between samples, and the offshore assemblage was more compact, indicat- ing a greater similarity in samples (Fig. 6A). In late summer the assemblage configurations were reversed (Fig. 6B). All three assemblages shared a majority of species, although there were some dif- ferences in the abundance of each species (Tables 3 and 4; Fig. 6, C and D). The seasonal difference in assemblages result- ed from a change in both the number and iden- tity of species present in the study area (Tables 3 and 4; Fig. 6, C and D). In midsummer the inshore assemblage was composed predominately of northern pipefish (Syngnathus fuscus) and other species were present in lesser numbers (Ta- ble 3; Fig. 6C). The near-ridge assemblage was dominated by butterfish ( Peprilus triacanthus) and bay anchovy ( Anchoa mitchilli ), and only striped anchovy (Anchoa hepsetus) and spotted hake ( Urophycis regia) were present with a mean catch per tow greater than one (Table 3). Al- though butterfish was also the dominant species in the offshore assemblage, its abundance was one third of what was found in near-ridge trawls (Table 3). Of particular interest in the midsum- mer analysis was a group of species (weakfish, bluefish [ Pomatomus saltatrix], northern puff- er [Sphoeroides maculatus], bay anchovy, and striped anchovy) separated along the primary axis from the rest of the species centroids in the near-ridge assemblage. These species were as- sociated with samples from near-ridge stations taken in 2001 and 2005 (Fig. 6A). During late summer, the abundance of bay anchovy increased substantially within the inshore assemblage, as did the abundance of northern pipefish (Table 4, Fig. 6C). At that time, bay anchovy was the most abundant species in the near-ridge assemblage by nearly two orders of magnitude, and Atlantic Vasslides and Able: Importance of shoreface sand ridges as habitat for fishes 99 Table 2 Catch per unit of effort (CPUE, with standard error in parentheses) and mean size (mm, with standard error in parentheses) for species captured during late summer beam trawl sampling from 1991 through 1995. Station assemblages were determined by using canonical correspondence analysis as shown in Figure 5B. Taxa were measured to total length ( t), fork length (*), or body width (f). A dash indicates that no data were available. Species Station assemblage Inshore Near-ri dge+offshore CPUE Size CPUE Size Syngnathus fuscus t 0.61 (0.27) 108(13) 0.37 (0.11) 128 (11) Prionotus carolinus t 0.17 (0.09) 55 (9) 4.16 (0.95) 24(2) Scophthalmus aquosos t 0.28(0.18) 114 (14) 0.23 (0.09) 233(7) Sphoeroides maculatus t 0.11 (0.08) 105 (5) 0.40 (0.12) 47 (9) Etropus microstomus t 0.5 (0.25) 45 (10) 17.79 (3.39) 27 (1) Micropogonias undulatus t 1.06 (0.61) 15 (2) 0.81 (0.54) 25(1) Prionotus evolans f 0.17 (0.12) 48 (14) 0.53 (0.24) 74(18) Hippocampus erectus t 0.11 (0.08) 68 (17) 0.05 (0.03) 59 (4) Centropristis striatus t 0.11 (0.11) 52 (8) 4.09 (1.58) 38(1) Cynoscion regalis t 0.22 (0.17) 75 (4) — — Ammodytes americanus * 0.06 (0.06) 76 (-) — — Ammodytes spp. * 0.06 (0.06) 81 (-) — — Menticirrhus sp. t 2.83 (1.2) 39(3) — — Gobiosoma sp. t 0.39 (0.33) 34 (2) — — Urophycis regia t — — 0.35(0.09) 201 (15) JJrophycis chuss t — — 1.3 (0.40) 30(1) Gobiosoma ginsburgi f — — 5.28 (1.71) 30 (1) Ophidion marginatum t — — 0.16(0.09) 189 (12) Paralichthys dentatus t — — 0.16 (0.09) 165 (51) Raja erinacea t — — 0.09 (0.04) 251 (25) Bothus sp. | — — 0.02 (0.02) 22 (-) Raja eglanteria $ — — 0.02 (0.02) 420 (-) Tautogolabrus adspersus t — — 0.02 (0.02) 29 (-) croaker, weakfish, and silver perch (Bairdiella chrysoura) were all present at relatively high abundances (Table 4). Bay anchovy and butter- fish were the dominant species in the offshore assemblage (Table 4). The preponderance of spe- cies found near the ridge compared to offshore was reflected in the ordination by the number of species centroids associated with the near-ridge assemblage. However, some species were found at similar abundances in more than one assemblage (i.e., scup t Stenotomus chrysops ]), and their cen- troids may have been located closer to, but not within, any one assemblage (Fig. 6D). The percentage of variance of the species-envi- ronment interaction reflected in the midsummer (54%) and late summer (67%) ordinations was substantial across both time periods (Table 5). Temperature, depth, and dissolved oxygen were significant environmental variables that ex- plained the species assemblages in midsummer; whereas distance from the ridge and depth were the primary factors in late summer (Table 6). 100 Fishery Bulletin 106(1 ) Table 3 Catch per unit of effort (CPUE, with standard error in parentheses) and mean size (mm, with standard error in parentheses) for species captured during mid-summer otter trawl sampling for 2001-06. Station assemblages were determined with canonical correspondence analysis as shown in Figure 6A. Taxa were measured to total length (f) or fork length (*). A dash indicates that no data were available. Assemblage Species Inshore Near-ridge Offshore CPUE Size CPUE Size CPUE Size Syngnathus fuscus t 0.64 (0.20) 116(9) 0.03 (0.03) 127 (6) — — Prionotus carolinus t 0.18 (0.18) 47(2) 0.38(0.13) 189(13) 0.75(0.26) 224 (8) Scophthalmus aquosos t 0.18 (0.12) 62 (19) 0.25 (0.07) 129 (12) 0.08 (0.04) 270(7) Sphoeroides maculatus t 0.18 (0.18) 38(23) 0.61 (0.28) 71(3) — — Anchoa mitchilli * 0.09 (0.09) 30 (-) 16.11 (9.15) 68(1) — — Etropus microstomus t 0.09 (0.09) 95 (-) 0.88 (0.24) 93(3) 0.38(0.10) 94(10) Menidia menidia * 0.09(0.09) 55 (-) — — — — Anchoa hepsetus * — — 1.92(0.98) 65(2) — — Pomatomus saltatrix * — — 0.03 (0.02) 123 (7) — — Peprilus triacanthus * - — 26.55 (15.8) 40(1) 9.8 (3.2) 37(1) Urophycis regia t — — 1.2 (0.35) 173 (4) 0.08 (0.04) 157 (32) Stenotomus chrysops * — — 0.47 (0.14) 111 (5) 0.05(0.05) 106(3) Cynoscion regalis t — — 0.12 (0.14) 131 (11) — — Prionotus evolans t — — 0.28(0.12) 179(13) 0.6 (0.41) 215(6) Menticirrhus saxatilis t — — 0.03 (0.02) 279(25) — — Centropristis striatus t - — 0.02 (0.02) 114 (-) 0.03 (0.03) 265 (-) Paralichthys oblongus t — — 0.05(0.03) 29(2) 0.03 (0.03) 181 (-) Urophycis chuss t — — 0.02 (0.02) 70 (-) 0.03(0.03) 92 (-) Hippocampus erectus t — — — — 0.03 (0.03) 53 (-) Merluccius bilinearis t — — — — 0.05 (0.03) 175 (±21) Citharichthys arctifrons t — — — — 0.03(0.03) 68 (— ) Discussion Species abundance and richness The dominant fish families (Engraulidae, Paralichthy- idae, Gadidae, Triglidae, Serranidae, Sciaenidae, and Stromateidae) and to some degree species (butterfish, spotted hake, northern searobin [Prionotus carolinus], black sea bass [ Centi'opristis striatus ], weakfish) captured by the two gears in the study area were similar to those previously found in inner continental shelf waters off of the northeast United States (Colvocoresses and Musick, 1984; Mahon et ah, 1998) and southeast United States (Walsh et al., 2006). Previous comparisons between beam and otter trawls similar in size to those used in our study have shown that otter trawls collect more species and more individuals and that beam trawls catch smaller fish (Vasslides, 2007). This difference is reflected in the length-frequency histogram for all species (Fig. 4), as well as for the dominant species collected in both gears. Overall species abundance (CPUE) and richness (RPUE) displayed a hi modal distribution across the inlet to the offshore transects, and the highest values occurred on either side of the ridge regardless of gear type (Fig. 3). This bimodal pattern has been previously suggested for fish (Martino and Able, 2003) and deca- pod crustaceans (Viscido et ah, 1997) at Beach Haven Ridge but is in contrast to the findings from a number of studies of larger scale cross-shelf transects, where abundance has been shown to decrease linearly with depth in the Mid-Atlantic Bight (Colvocoresses and Musick, 1984), Chukchi Sea (Barber et al., 1997), and Mediterranean Sea (Colloca et al., 2003), but not in the Bering Sea (Mueter and Norcross, 1999). The species composition and richness, and thus as- semblage structure, varied between sampling gears. This was expected because beam trawls sample fishes on the bottom better than otter trawls (Wennhage et al., 1997) and thus capture more demersal species, in- cluding small, recently settled fish and small juveniles. This selectivity may explain the differences at certain sampling stations in regard to both abundance and spe- cies richness between the two trawl types. In the beam trawl, both abundance per tow and richness per tow at Vasslides and Able: Importance of shoreface sand ridges as habitat for fishes 101 Table 4 Catch per unit of effort (CPUE, with standard error in parentheses) and mean size (mm, with standard error in parentheses) for species captured during late summer otter trawl sampling during 2001-06. Station assemblages were determined using canoni- cal correspondence analysis as shown in Figure 6B. Taxa were measured to total length (t ) or fork length (*). A dash indicates that no data were available. Assemblage Inshore Near-ridge Offshore Species CPUE Size CPUE Size CPUE Size Syngnathus fuscus t 1.55 (0.62) 133 (3) 0.16 (0.07) 132 (7) — — Prionotus carolinus t 0.07 (0.07) 47 (2) 0.23 (0.08) 230(9) 0.12 (0.06) 131 (45) Scophthalmus aquosos t 0.07 (0.04) 62 (19) 0.05 (0.03) 247 (33) — — Sphoeroides maculatus t 0.21 (0.10) 89 (25) 0.08(0.03) 98(12) 0.03(0.03) 90 (-) Anchoa mitchilli * 4.69 (2.9) 51 (2) 372.15(134.37) 58(1) 11.76(9.49) 65 (1) Etropus microstomus t 0.03 (0.03) 95 (-) 0.05 (0.03) 71(4) — — Menidia menidia * 0.41 (0.24) 71 (3) — — — — Anchoa hepsetus * 0.24 (0.18) 76 (3) 2.02 (1.04) 86 (1) — — Pomatomus saltatrix * 0.21 (0.13) 141 (17) 0.14 (0.06) 98 (2) 0.03 (0.03) 100 (-) Peprilus triacanthus * — — 0.68 (0.17) 70(5) 2.52 (0.80) 55 (3) Urophycis regia t — — 0.08 (0.04) 217 (19) 0.06 (0.04) 187 (68) Stenotomus chrysops * — — 0.52 (0.14) 152 (6) 0.35 (0.13) 126 (23) Cynoscion regalis t — — 4.26(0.66) 190 (4) — — Prionotus evolans t - — 0.06 (0.03) 110 (13) — — Menticirrhus saxatilis'f — — 0.05 (0.03) 182 (54) — — Centropristis striatus t — — 0.03 (0.02) 58(24) 0.03 (0.03) 35 (-) Micropogonias undulatus t — — 5.58(2.29) 209 (4) 0.35 (0.21) 309 (9) Bairdiella chrysoura t — — 1.09(0.62) 131 (2) — — Hippocampus erectus t — — 0.02(0.02) 63 (-) — — Merluccius bilinearis f — — 0.02 (0.02) 79 (-) — — stations BHRTOP and BHR-9 were of intermediate and high values, respectively, whereas in the otter trawl these stations accounted for the lowest and intermedi- ate values, respectively. Over 70% of the individuals captured at stations BHRTOP and BHR-9 during the beam trawl were young of the year (YOY) smallmouth flounder ( Etropus microstomus ), and the majority of these averaged 27 mm and 35 mm TL, respectively. During previous direct comparisons between beam and otter trawls it was found that CPUE and frequency of occurrence for smallmouth flounder were greater in beam trawls and that mean length was substantially less. Because this single species accounted for such a large proportion of the abundance at BHRTOP and BHR-9 as compared to the abundance at other stations, it appears that the selectivity of beam trawls plays a large role in estimating fish abundance and thus the fish assemblages in general. The seeming discrepancies in RPUE between gear types at particular locations can be partially explained by the limited numbers of tows. At station BHR-9, data were collected from only two beam-trawl tows. Although ten species were collected, five species were represented by one individual and two species were represented by two individuals. Given this apparent patchiness, there is a possibility that additional sampling at this station would yield a RPUE that would more closely reflect the pattern observed in the otter-trawl tows. Even though a limited number of samples were taken with an otter trawl at BHRTOP (?? = 6), our previous opportunistic surveys with the same gear at other locations on the top of Beach Haven Ridge yielded low numbers of spe- cies per tow. However, twice the number of species were caught during midsummer sampling in 2006 at this station, but abundance of each new species was five individuals or less. Fish-assemblage structure and environmental relationships The number of assemblages and their constituent mem- bers varied by both gear and season. Two general groups were identified in the beam trawl (inshore and near- ridge+offshore) and three in the otter trawl (inshore, 102 Fishery Bulletin 106(1 ) A B Figure 5 Canonical correspondence Analysis (CCA) ordinations of the beam trawl survey data (1991-95) displaying the station assemblages for midsummer only (A), late summer only (B), and the species assemblages for midsummer only (C) and late summer only (D). Solid lines within each figure box (A-D) enclose the boundaries of the identified assemblages. In A and B, each sampling station is identified by a different symbol and the arrows depict the gradient of each environmental variable. In C and D, species that occupy the same area of the graph are grouped by short lines and arrows denote their true locations. near-ridge, and offshore). Within the Beach Haven Ridge beam trawl samples, 55% of the fish used in the analysis were found in both species assemblages, leaving nearly half of the species captured to be found in only one assemblage or the other. This is in stark contrast to the 2001-06 Beach Haven Ridge otter-trawl data subset, where 82% of the species were found in at least two of the three assemblages and only 18% of the species (silver perch, Atlantic silverside [Menidia menidia ], northern kingfish [ Menticirrhus saxatilis], and Gulf stream floun- der t Citharichthys arctifrons ]) were present in only one assemblage. Thus it appears that there is a definite gradient along the transect, represented by changes in species present in the beam trawl and by the relative abundances of shared species in the otter trawl. Although cross-shelf gradients in demersal fish as- semblages have been identified along the northeast United States (Sullivan et ah, 2000), northwest United States (Mueter and Norcross, 1999), southwest United States (Johnson et ah, 2001), and worldwide (Gray and Vasslides and Able: Importance of shoreface sand ridges as habitat for fishes 103 Table 5 Eigenvalues and cumulative percentage variance for each axis analyzed by season for the 1991-95 beam trawl data and 2001-06 otter trawl data by using canonical correspondence Analysis (CCA). The eigenvalues are a relative measure of the importance of each axis. 1 CCA axis 2 3 4 1991-95 beam trawl data Midsummer Eigenvalue Cumulative percentage variance 0.296 0.176 0.145 0.117 of species data 21.3 34.0 44.4 52.8 of species environment 39.1 62.3 81.4 96.8 Late summer Eigenvalue Cumulative percentage variance 0.459 0.209 0.153 0.132 of species data 17.2 25.0 30.8 35.7 of species environment 46.8 68.1 83.8 97.2 2001-06 otter trawl data Midsummer Eigenvalue Cumulative percentage variance 0.572 0.356 0.277 0.176 of species data 10.5 17.0 22.0 25.3 of species environment 33.0 53.6 69.6 79.8 Late summer Eigenvalue Cumulative percentage variance 0.343 0.236 0.128 0.067 of species data 10.7 18.1 22.1 24.3 of species environment 39.8 67.2 82.0 89.9 Table 6 The P values from Monte Carlo permutation tests on the significance of the environmental variables in each data set. Significant P values (P< 0.05) are shown in bold. A dash indicates that the variable was not measured. Variable 1991-95 beam trawl 2001- -06 otter trawl Midsummer Late summer Midsummer Late summer Salinity 0.036 0.874 0.200 0.996 Temperature 0.212 0.002 0.002 0.360 Depth 0.120 0.266 0.004 0.090 Distance from ridge 0.062 0.094 0.406 0.002 Habitat complexity 0.796 0.028 0.132 0.112 Dissolved oxygen — — 0.022 0.654 Water clarity — — 0.218 0.824 pH — — 0.212 0.172 Otway, 1994), these gradients were all at substantially larger spatial scales. Few studies of either juveniles or adults have been conducted at a resolution similar to that of our study in inner continental shelf waters. Juvenile fish assemblages of the nearshore (<40 km) coast of Georgia exhibited a cross-shelf pattern in win- ter and spring, and a shallow group (8 m in depth) was separated from the other stations (12-18 m in depth) (Walsh et ah, 2006). Fish assemblages in northern Ar- gentina in the spring were identified as either those of the inner, central, or middle regions of the coastal shelf (Jaureguizar et al., 2006). Analyses of fish assemblages across continental shelves often point to depth as the primary environ- mental variable correlated with the changes in fish-as- semblage structure (Mahon et al., 1998; Walsh et ah, 2006), whereas studies focused on shorter distances indicate a combination of environmental and physi- 104 Fishery Bulletin 106(1) c D Late summer only Inshore ‘nidia menidia Offshore Sphoeroides mqculatus Syngnathus fuscus PomalomuS sa/ta^rr Anchok hepdptu: Cynoscion recfalis Hippocampus erpctus Bairdiella chrysgura Merluccius bitnearis Near-ridge Raja eglanteria Peprilus triacanthus Prionotus carolinus Stenotomus chrysops Urophycis regia Prionotus evolans Centropristis striatus Scophthatmus aquosos Micropogonias undulatus Etropus microstomus -0.6 Eigenvalue= 0.343 1.0 Figure 6 Canonical correspondence analysis (CCA) ordinations of the otter trawl survey data (2001-06) dis- playing the station assemblages for midsummer only (A), late summer only (B), and the species assemblages for midsummer only (C) and late summer only (D). Solid lines within each figure box (A-D) enclose the boundaries of the identified assemblages. In A and B, each sampling station is identified by a different symbol and the arrows depict the gradient of each environmental variable. In C and D, species that occupy the same area of the graph are grouped by short lines and arrows denote their true locations. cal variables (Martino and Able, 2003; Jaureguizar et al., 2006). The results of our study point to the latter case. Temperature and distance from the top of the ridge were often as important explanatory factors as depth, and habitat complexity and dissolved oxygen were also correlated with the distribution of fish along Vasslides and Able: Importance of shoreface sand ridges as habitat for fishes 105 the transect. Temperature has played an important role in regulating fish distribution in temperate waters (e.g., Colvocoresses and Musick, 1984; Able et al., 2006), and it may explain the variation in the species assemblages between mid- and late summer for both the beam and otter trawl. Furthermore, the temperature differences along a transect can also shape the species assemblages within a season, as seen in the significance levels for temperature between seasons in the otter trawl CCA (Table 6). In midsummer, when the CCA identified tem- perature as a statistically important environmental variable, there was a large temperature gradient from BHR-1 to BHR-5 and a slightly smaller change from BHR-7 to BHR-9. These temperature gradients are coincident with the three assemblages identified for the otter trawl midsummer samples in the CCA (Fig. 6A). In contrast, the late summer temperatures were fairly constant across the transect, and this constancy was reflected in the nonsignificant P-value for temperature in the CCA. Although we examined seasonal and annual temporal scales, episodic events can also have a dramatic effect on species assemblages. The study area is well known as a region of up welling during the summer months; it often has up to five upwelling events each year, typi- cally lasting a week or more (Glenn et ah, 2004). These upwelling events bring cooler water generally found at the offshore sampling locations onto the near-ridge stations. When sampling occurred during an upwell- ing event, similar species were captured at all stations along the transect (excluding BHR-1). However, in years where the bottom temperatures during sampling were higher than the average study temperature (1997, 1998, 2001, and 2005), the near-ridge species assemblage included weakfish and northern puffer, species more commonly associated with late summer when tempera- tures are warmer (Fig. 6C). It is interesting to note that upwelling events of 2001 were some of the most intense recorded during a 9-year study (Glenn et ah, 2004). However, when the samples of 2001 were collected, the water temperatures had returned to the seasonal norm, thus illustrating the rapidity with which upwelling events break down and fish assemblages can change. Although dissolved oxygen appears to be a signifi- cant factor in the arrangement of species assemblages along the sampling transect, its importance may be confounded by its relationship with temperature and depth. As expected, the highest mean dissolved oxygen levels were found at the stations with the coldest mean temperatures, which were also the deepest stations. However, the lowest mean dissolved oxygen value was found at a station in the same assemblage that had the highest value; thus the importance of dissolved oxygen remains unclear. The trend toward less well-defined species assem- blages when the environmental gradients were less pronounced lends some support to the idea that cross shelf patterns in species distributions are attributable to environmental gradients (Jaureguizar et al., 2006; Walsh et al., 2006). However, the importance of habitat complexity in the analyses of assemblages from both gear types (Table 6) indicates selection of specific habi- tats by some species within a large-scale environmental gradient (see Stoner and Abookire, 2002). This has been shown in many laboratory and field experiments for flat- fishes (Neuman and Able, 1998; Stoner and Abookire, 2002) and other demersal fishes (Sullivan et al., 2000; Diaz et ah, 2003). As suggested by Mueter and Norcross (1999), this difference may be due to differences in how juvenile or small fishes use benthic habitat compared to larger adult fishes. The selection of habitat within the study area changed with ontogenetic stage; this is particularly true for the sandy substrate found on the top of the ridge. The beam trawl samples, which contained smaller, presumably younger juveniles, had greater species richness and abundance values on the ridge top than did the otter trawl, which captured larger juveniles and adults. The sandy substrate on the top of the ridge provided im- portant habitat for species that bury themselves, such as northern stargazer (Asti'oscopus guttatus) (Able and Fahay, 1998) and snakefish ( Trachinocephalus myops) (Sulak, 1990). These species were found only on the top of Beach Haven Ridge, although admittedly in small numbers. Sand lances, another group that buries itself in sand, was also found predominantly in the sandy substrates. In a paired video sled and beam trawl sur- vey (where a video camera sled was towed along the bottom and then a beam trawl was dragged along the same area) on sand ridges off the coast of Maryland and Delaware, a substantially larger number of sand lances were captured in the video sled than in the beam trawl (Diaz et al., 2003), indicating that sand lance may be more important to the assemblage at Beach Haven Ridge than expected from the trawl results. Time of day also affects the abundance, species rich- ness, and identity of species captured in various habi- tats. A study of sand ridges offshore of Maryland and Delaware found that when complex habitats were lo- cated in proximity to simple habitats, fish abundance was twice as great in the complex habitats during the day, and the pattern was reversed at night (Diaz et al., 2003) . This pattern is most likely due to 1) changes in foraging behavior over a diel period and to 2) smaller demersal fish selecting refuge from predators in complex habitats. The fact that all of the trawls in this study were conducted during daytime may explain why the stations located on either side of the top of the ridge, which had more complex habitats, had the highest val- ues for abundance and richness. There are a number of other possible explanations for the patterns in species abundance and assemblages identified herein that were not explored as part of this study. Investigations into the abundance and distribu- tion of planktonic larvae around Beach Haven Ridge have revealed physical processes as important mecha- nisms in concentrating mollusk larvae on either side of the sand ridge (Ma et al., 2006). These same processes may be causing the increased abundance of pelagic fish species seen on the flanks of the sand ridges. The 106 Fishery Bulletin 106(1 ) availability of preferred prey items, such as pelagic fishes like bay anchovy and butterfish, may also be af- fecting the abundance and distribution of fish species (Vasslides, 2007). In summary, shoreface sand ridges may have a dis- tinct influence on fish abundance and assemblages. The near-ridge habitats have higher species abun- dances and richness compared to the surrounding inner continental shelf and also possess a distinct species assemblage, including both recreationally and commercially important species. Additionally, the fish found at the top of the ridge were typical prey spe- cies (sand lances, anchovies, smallmouth flounder) favored by both resident and transient piscivores in the Mid-Atlantic Bight (Chao and Musick, 1977; Chase, 2002; Walter et al., 2003; Gartland et al., 2006) and thus sand ridges may influence the distribution of these economically important piscivores. As such, sand ridges appear to be important features of the inner continental shelf and may not be suitable areas for resource extraction activities. Acknowledgments This study owes a great deal to the post-doctoral stu- dents, graduate students, and technicians of the Rutgers University Marine Field Station who collected over 15 years worth of data used in this research. S. Hagan and T. Grothues provided assistance with the data analysis and C. Van Pelt provided editorial support. This research was supported by a National Oceanic and Atmospheric Administration Cooperative Marine Education and Research Grant and the Rutgers University Marine Field Station. This is contribution number 2007-12 of The Institute of Marine and Coastal Sciences, Rutgers University. Literature cited Able, K. W„ and M. P. Fahay. 1998. The first year in the life of estuarine fishes in the Middle Atlantic Bight, 342 p. Rutgers Univ. Press, New Brunswick, NJ. Able, K. W., M. P. Fahay, D. A. Witting, R. S. McBride, and S. M. Hagan. 2006. Fish settlement in the ocean vs. estuary: Compari- son of pelagic larval and settled juvenile composition and abundance from southern New Jersey, LT.S.A. Estuar. Coast. Shelf Sci. 66:280-290. Barber, W. E., R. L. Smith, M. Vallarino, and R. M. Meyer. 1997. Demersal fish assemblages of the northeastern Chukchi Sea, Alaska. Fish. Bull. 95:195-208. Byrnes, M. R., R. H. Hammer, T. D. Thibaut, and D. B. Snyder. 2004. Effects of sand mining on physical processes and biological communities offshore New Jersey, U.S.A. J. Coastal Res. 20(l):25-43. Chao, L ,N., and J. A., Musick, 1977. Life history, feeding habits, and functional mor- phology of juvenile sciaenid fishes in the York River estuary, Virginia. Fish. Bull. 75:657-702. 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. Colloca, F., M. Cardinale, A. Belluscio, and G. Ardizzone. 2003. Pattern of distribution and diversity of demer- sal assemblages in the central Mediterranean Sea. Estuar. Coast. Shelf Sci. 56:469-480. Colvocoresses, J. A., and J. A. Musick. 1984. Species associations and community composition of Middle Atlantic Bight continental shelf demersal fishes. Fish. Bull. 82:295-313. Craghan, M. 1995. Topographic changes and sediment character- istics at a shoreface sand ridge-Beach Haven Ridge, New Jersey. M.S. thesis, 112 p. Rutgers Univ., New Brunswick, NJ. Diaz, R. J., G. R. Cutter Jr., and K. W. Able. 2003. The importance of physical and biogenic struc- ture to juvenile fishes on the shallow inner continental shelf. Estuaries 26(11:12-20. Diaz, R. J., G. R. Cutter Jr., and C. H. Hobbs III. 2004. Potential impacts of sand mining offshore of Mary- land and Delaware: Part 2 — biological considerations. J. Coastal Res. 20(1):61— 69. Drucker, B. S., W. Waskes, and M. R. Byrnes. 2004. The U.S. Minerals Management Service outer continental shelf sand and gravel program: Environ- mental studies to assess the potential effects of offshore dredging operations in federal waters. J. Coastal Res. 20(1):1— 5. Gartland, J., R. J. Latour, A. D. Halvorson, and H. M. Austin. 2006. Diet composition of young-of-the-year bluefish in the lower Chesapeake Bay and coastal ocean of Virginia. Trans. Am. Fish. Soc. 135:371-378. Glenn, S'., R. Arnone, T. Bergmann, W. P. Bissett, M. Crowley, J. Cullen, J. Gryzmski, D. Haidvogel, J. Kohut, M. Molline, M. Oliver, C. Orrico, R. Sherrell, T. Song, A. Weidemann, R. Chant, and O. Schofield. 2004. Biogeochemical impacts of summertime coastal upwelling on the New Jersey Shelf. J. Geophys. Res. 109, C12S02, doi : 10. 1029/2003 JC002265, 15 p. Gray, C. A., and N. M. Otway. 1994. Spatial and temporal differences in assemblages of demersal fishes on the inner continental shelf off Sydney, south-eastern Australia. Aust. J. Mar. Freshw. Res. 45:665-676. Jaureguizar, A. J., R. Menni, C. Lasta, and R. Guerrero. 2006. Fish assemblages of the northern Argentine coastal system: spatial patterns and their temporal variations. Fish. Oceanogr. 15(4):326-344. Johnson, K. A., M. M. Yoklavich, and G. M. Cailliet. 2001. Recruitment of three species of juvenile rockfish ( Sebastes spp.) on soft benthic habitat in Monterey Bay, California. Calif. Coop. Oceanic Fish. Invest. Rep. 42:53-166. Ma, H., J. P. Grassle, and R. J. Chant. 2006. Vertical distribution of bivalve larvae along a cross-shelf transect during summer upwelling and downwelling. Mar. Biol. 149:1123-1138. Mahon, R, S. K. Brown, K. C. T. Zwanenburg, D. B. Atkinson, K. R. Buja, L. Claflin, G. D. Howell, M. E. Monaco, R. N. O’Boyle, and M. Sinclair. 1998. Assemblages and biogeography of demersal fishes of the east coast of North America. Can. J. Fish. Aquat. Sci. 55:1704-1738. Vasslides and Able: Importance of shoreface sand ridges as habitat for fishes 107 Martino, E., and K. W. Able. 2003. Fish assemblages across the marine to low salinity transition zone of a temperate estuary. Estuar. Coast. Shelf Sci. 56(5-61:969-987. McBride, R. A., and T. F. Moslow. 1991. Origin, evolution, and distribution of shoreface sand ridges, Atlantic inner shelf, USA. Mar. Geol. 97:57-85. McGarigal, K., S. Cushman, and S. Stafford. 2000. Multivariate statistics for wildlife and ecology research, p. 20-73. Springer-Verlag Inc., 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. Nairn, R., J. A. Johnson, D. Hardin, and J. Michel. 2004. A biological and physical monitoring program to evaluate long-term impacts from sand dredging opera- tions in the United States outer continental shelf. J. Coastal Res. 20(11:126-137. Neuman, M. J., and K. W. Able. 1998. Experimental evidence of sediment preference by early life history stages of windowpane flounder ( Scoph - thalmus aquosos 1. J. Sea Res. 40:33-41. Palmer, M. W. 1993. Putting things in even better order: the advan- tages of Canonical Correspondence Analysis. Ecology 74(81:2215-2230. Stahl, L., J. Koczan, and D. Swift. 1974. Anatomy of a shoreface-connected sand ridge on the New Jersey shelf: implications for the genesis of the shelf surficial sand sheet. Geology 2:117-120. Stoner, A. W., and A. A. Abookire. 2002. Sediment preferences and size-specific distribu- tion of young-of-the-year Pacific halibut in an Alaska nursery. J. Fish Biol. 61:540-559. Sulak, K. J. 1990. Synodontidae. In Check-list of the fishes of the eastern tropical Atlantic (CLOFETA) (J. C. Quero, J. C. Hureau, C. Karrer, A. Post, and L. Saldanha, eds.1, vol. 1, p. 365-370. Junta Nacional de Investigagao Cientifica e Tecnologica, Lisbon; European Ichthyologi- cal Union, Paris; and UNESCO, Paris. Sullivan, M. C., R. K. Cowen, K. W. Able, and M. P. Fahay. 2000. Spatial scaling of recruitment in four continental shelf fishes. Mar. Ecol. Prog. Ser. 207:141-154. Underwood, A. J. 1997. Experiments in ecology: their logical design and interpretation using analysis of variance, 504 p. Cam- bridge Univ. Press, Cambridge, England. Van den Brink, P. J., and C. J. F. Ter Braak. 1999. Principal response curves: analysis of time-depen- dent multivariate responses of biological community to stress. Environ. Toxicol. Chem. 18(21:138-148. Vasslides, J. M. 2007. Fish assemblages and habitat use across a shore- face sand ridge in southern New Jersey. M.S. thesis, 106 p. Rutgers Univ., New Brunswick, NJ. Viscido, S. V., D. E. Stearns, and K. W. Able. 1997. Seasonal and spatial patterns of an epibenthic decapod crustacean assemblage in north-west Atlan- tic continental shelf waters. Estuar. Coast. Shelf Sci. 45:377-392. Walsh, H. J., K. E. Marancik, and J. A. Hare. 2006. Juvenile fish assemblages collected on unconsoli- dated sediments of the southeast United States conti- nental shelf. Fish. Bull. 104:256-277. Walter, J. F. Ill, A. S. Overton, K. H. Ferry, and M. E. Mather. 2003. Atlantic coast feeding habits of striped bass: a synthesis supporting a coast-wide understanding of trophic biology. Fish. Manag. Ecol. 10:349-360. Wennhage, H., R. N. Gibson, and L. Robb. 1997. The use of drop traps to estimate the efficiency of two beam trawls commonly used for sampling juvenile flatfishes. J. Fish Biol. 51:441-445. 108 Fishery Bulletin 106(1) Fishery Bulletin Guidelines for authors Content of manuscripts Contributions published in Fishery Bulletin describe original research in marine fishery science, fishery engineering and economics, as well as the areas of marine environmental and ecological sciences (including modeling). Preference will be given to manuscripts that examine processes and underlying patterns. Descriptive reports, surveys, and observational papers may occa- sionally be published but should appeal to an audience outside the locale in which the study was conducted. Although all contributions are subject to peer review, responsibility for the contents of papers rests upon the authors and not on the editor or publisher. Submission of an article implies that the article is origin al and is not being considered for publication elsewhere. Articles may range from relatively short contributions (10-15 typed, double-spaced pages, tables and figures not included) to extensive contributions (20-30 typed pages). Notes are reports of 5 to 10 pages without an abstract and describe methods or results not supported by a large body of data. Manuscripts must be written in English; authors whose native language is not English are strongly advised to have their manuscripts checked by English-speaking colleagues before submission. Manuscript Preparation Title page should include authors’ full names and mail- ing 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 200 words (one-half typed page), state the main scope of the research, and emphasize the author’s conclusions and relevant findings. Do not review the methods of the study or list the contents of the paper. Because abstracts are circulated by abstract- ing 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 Conclusions), and Acknowl- edgments. Headings within each section must be short, reflect a logical sequence, and follow the rules of multiple subdivision (i.e., there can be no subdivision without at least two items). The entire text should be intelligible to interdisciplinary readers; therefore, all acronyms, abbreviations, and technical terms should be written out in full the first time they are mentioned. Include FAO common names for species in the list of keywords and in the introduction. Regional common names may be used throughout the rest of the text if they are dif- ferent from FAO common names which 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; for fish nomenclature follow the most current issue of the American Fisheries Society’s Common and Scientific Names of Fishes from the United States and Canada. Dates should be written as follows: 11 November 2000. Measurements should be expressed in metric units, e.g., 58 metric tons (t); if other units of measurement 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). Refrain from using the shorthand slash (/), an ambigu- ous symbol, in the general text Literature cited comprises published works and those accepted for publication in peer-reviewed literature (in press). Follow the name and year system for citation format in the “Literature cited” section (that is say, citations should be listed alphabetically by the authors’ last names, and then by year if there is more than one citation with the same authorship). If there is a sequence of citations in the text, list chronologically: (Smith, 1932; Green, 1947; Smith and Jones, 1985). Abbrevia- tions of serials should conform to abbreviations given in the Serial Sources for the BIOSIS Previews Database. Authors are responsible for the accuracy and complete- ness of all citations. Literature citation 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. Cite all software and special equipment or chemical 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). 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. • Do not place outline rules around graphs. • Do not use horizontal lines through graphs to indicate measurement units. • Use a comma in numbers of five digits or more (e.g. 13,000 but 3000). • Maps require a North arrow and degrees latitude- longitude (e.g., 170°E). 109 Tables are often overused in scientific papers; it is seldom necessary or even desirable to present all the data associated with a study. Tables should not be excessive in size and must be cited in numerical order in the text. Headings should be short but ample enough to allow the table to be intelligible on its own. All unusual symbols must be explained in the table legend. Other incidental comments may be footnoted with italic numeral footnote markers. Use asterisks to indicate probability in statisti- cal data. Do not type table legends on a separate page; place them above the table data. Do not submit tables in photo mode. Figures include line illustrations, photographs (or slides), and computer-generated graphs and must be cited in numerical order in the text. Graphics should aid in the comprehension of the text, but they should be limited to presenting patterns rather than raw data. Figures should not exceed one figure for every four pages of text. Figures must be labeled with author’s name and number of the figure. Avoid placing labels vertically (except of y axis). 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 and failure to correspond with editors in a timely manner will delay publication of a manuscript. Copyright law does not apply to Fishery Bulletin, 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 consid- ered correct form (e.g., Source: Fish. Bull 97:105). Submission The Scientific Editorial Office encourages authors to submit their manuscripts as a single PDF (preferred) or Word (zipped) document by e-mail to Fishery. Bulletin@noaa.gov. Please use the subject heading, “Fishery Bulletin manuscript submission”. Do not send encrypted files. For further details on electronic sub- mission, please contact the Scientific Editorial Office directly (see address below). Or you may send your manuscript on a compact disc in one of the above formats along with four printed copies (one original plus three copies [stapled]) to the Scientific Editor, at the address shown below. Dr. Adam Moles Scientific Editor, Fishery Bulletin NOAA/NMFS/AFSC 17109 Point Lena Loop Road Juneau, AK 99801-8626 Once the manuscript has been accepted for publication, you will be asked to submit a final electronic 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, tiff files, or EPS files. Send a copy of figures in the origi- nal software if conversion to any of these formats yields a degraded version. Questions? If you have questions regarding these guidelines, please contact the Managing Editor, Sharyn Matriotti, at Sharyn.Matriotti@noaa.gov Questions regarding manuscripts under review should be addressed to Adam Moles, Scientific Editor, at Adam. Moles@noaa.gov Fishery Bulletin Subscription form Superintendent of Documents Publications Order Form *5178 I 1 YES, please send me the following publications: Subscriptions to Fishery Bulletin for $36.00 per year ($50.40 foreign) The total cost of my order is $ . Prices include regular domestic postage and handling and are subject to change. (Company or Personal Name) (Please type or print) (Additional address/attention line) (Street address) (City, State, ZIP Code) (Daytime phone including area code) (Purchase Order No.) Charge your order. ITS EASY! Please Choose Method of Payment: | Check Payable to the Superintendent of Documents | | GPO Deposit Account -□ ] VISA or MasterCard Account To fax your orders (202) 512-2250 (Credit card expiration date) (Authorizing Signature) Mail To: Superintendent of Documents P.O. Box 371954, Pittsburgh, PA 15250-7954 Thank you for your order! Also available online at http://bookstore.gpo.gov/collections/fishery-bulletin [ /. F53 U.S. Department of Commerce Volume 106 Number 2 April 2008 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 John Oliver Acting Assistant Administrator for Fisheries Scientific Editor Adam Moles, Ph.D. Associate Editor Elizabeth Siddon Ted Stevens Marine Research Institute Auke Bay Laboratories Alaska Fisheries Science Center 17109 Pt. Lena Loop Road Juneau, Alaska 99801 ^T°FC„ Managing Editor Sharyn Matriotti National Marine Fisheries Service Scientific Publications Office 7600 Sand Point Way NE, BIN C15700 Seattle, Washington 981 15-0070 The Fishery Bulletin (ISSN 0090-0656) is published quarterly by the Scientific Publications Office, National Marine Fisheries Service, NOAA, 7600 Sand Point Way NE, BIN C15700, Seattle, WA 98115-0070. Periodicals postage is paid at Seattle, WA. POSTMASTER: Send address changes for subscriptions to Fish- ery Bulletin, Superintendent of Docu- ments, Attn.: Chief, Mail List Branch, Mail Stop SSOM, Washington, 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: $36.00 domestic and $50.40 foreign. Cost per single issue: $21.00 domestic and $29.40 foreign. See back for order form. Editorial Committee Jeffrey M. Leis Thomas Shirley David Somerton Mark Terceiro Australian Museum, Sydney, Australia Texas A&M University 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 106 Number 2 April 2008 Fishery Bulletin Contents Articles 111-134 Orr, James W., and Sharon Hawkins Species of the rougheye rockfish complex: resurrection of Sebastes melanostictus (Matsubara, 1934) and a redescription of Sebastes aleutianus (Jordan and Evermann, 1898) (Teleostei: Scorpaeniformes) 135-142 Kingsford, Michael J., Heather M. Patterson, and Matthew J. Flood The influence of elemental chemistry on the widths of otolith increments in the neon damselfish ( Pomacentrus coe/estis) 143-151 Aalbers, Scott A. Seasonal, diel, and lunar spawning periodicities and associated sound production of white seabass ( Atractoscion nobilis) The National Marine Fisheries Service (NMFS) does not approve, recommend, or endorse any proprie- tary product or proprietary material mentioned in this publication. No reference shall be made to NMFS, or to this publication furnished by NMFS, in any advertising or sales promotion which would indicate or imply that NMFS approves, recom- mends, or endorses any proprietary product or proprietary 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. 152-160 Clardy, Todd R., William F. Patterson III, Douglas A. DeVries, and Christopher Palmer Spatial and temporal variability in the relative contribution of king mackerel ( Scomberomorus cavalla) stocks to winter mixed fisheries off South Florida 161-173 Parnel, Maria M., Robert L. Emmett, and Richard D. Brodeur Ichthyoplankton community in the Columbia River plume off Oregon: effects of fluctuating oceanographic conditions 174-182 Overton, Anthony S., Charles S. Manooch III, Joseph W. Smith, and Kenneth Brennan Interactions between adult migratory striped bass (Morone saxatilis) and their prey during winter off the Virginia and North Carolina Atlantic coast from 1994 through 2007 II Fishery Bulletin 106(2) 183-193 Vandersea, Mark W., R. Wayne Litaker, Katrin E. Marancik, Jonathan A. Hare, Harvey J. Walsh, Siya Lem, Melissa A. West, David M. Wyanski, Elisabeth H. Laban, and Patricia A. Tester Identification of larval sea basses (Centropristis spp.) using ribosomal DNA-specific molecular assays 194-203 Rhodes, Kevin L., and Mark. H. Tupper The vulnerability of reproductively active squaretail coralgrouper ( Plectropomus areolatus ) to fishing Notes 204-212 Garcfa-Rodrfguez, Francisco J., German Ponce-Dfaz, Isabel Munoz-Garcfa, Rogelio Gonzalez-Armas, and Ricardo Perez-Enriquez Mitochondrial DNA markers to identify commercial spiny lobster species (Panulirus spp.) from the Pacific coast of Mexico: an application on phyllosoma larvae 213-221 Nolan, Cormac J., and Bret S. Danilowicz Advantages of using crest nets to sample presettlement larvae of reef fishes in the Caribbean Sea 222-223 Guidelines for authors Ill Species of the rougheye rockfish complex: resurrection of Sebastes melanostictus (Matsubara, 1934) and a redescription of Sebastes aleutianus (Jordan and Evermann, 1898) (Teleostei: Scorpaeniformes) Email address: James.Orr@noaa.gov Resource Assessment and Conservation Engineering Division Alaska Fisheries Science Center National Marine Fisheries Service, NOAA 7600 Sand Point Way NE Seattle, Washington 98115-0070 Abstract — The widespread and com- mercially important rougheye rock- fish, Sebastes aleutianus (Jordan and Evermann, 1898), has been con- sidered a single variable species, with light- and dark-colored forms, found on the outer continental shelf and upper slope of the North Pacific Ocean. Genetic analysis of 124 speci- mens verified the presence of two spe- cies in new specimens collected from Alaska to Oregon, and the two species were analyzed for distinguishing color patterns and morphological charac- ters. Characters distinguishing the two were extended to an analysis of 215 additional formalin-fixed speci- mens representing their geographic ranges. Sebastes aleutianus is pale, often has dark mottling on the dorsum in diffuse bands, and does not have distinct dark spots on the spinous dorsal fin; it ranges from the eastern Aleutian Islands and southeastern Bering Sea to California. Sebastes melanostictus (Matsubara, 1934), the blackspotted rockfish, ranges from central Japan, through the Aleutian Islands and Bering Sea, to southern California. It is darker overall and spotting is nearly always present on the spinous dorsal fin. Sebastes swifti (Evermann and Goldsborough, 1907) is a synonym of S. aleutianus', S. kawaradae (Matsubara, 1934) is a synonym of S. melanostictus. The subgenus Zalopyr is restricted to S. aleutianus and S. melanostictus. Nomenclatural synonymies, diagno- ses, descriptions, and distributions are provided for each species. Manuscript submitted 20 September 2007. Manuscript accepted 20 November 2007. Fish. Bull. 106:111-134 (2008) The views and opinions expressed or implied in this article are those of the author and do not necessarily reflect the position of the National Marine Fisheries Service, NOAA. James W. Orr (contact author) Sharon Hawkins Auke Bay Laboratories Alaska Fisheries Science Center National Marine Fisheries Service, NOAA Ted Stevens Marine Research Institute 17109 Point Lena Loop Road Juneau, Alaska 99801-8626 Of the more than 110 species of Sebastes recognized worldwide, by far the greatest number of species are found in the North Pacific Ocean, where about 100 species are currently considered valid (Kai and Nakabo, 2002; Orr and Blackburn, 2004; Hyde and Vetter, 2007). Although very important commercially, mem- bers of the genus have a reputation for being difficult to identify, because they are similar in body shape and share other morphological characters. Live coloration is also an extremely important character used in diagnosis and identification. Molecular charac- ters serve to identify most of these species and generally lend support to traditional morphological identifica- tions and concepts of species limits. In addition, the use of genetic data has uncovered previously unrecog- nized species-level variation in what had been considered single species, as in the case of the rougheye rock- fish complex, now known to comprise the two species Sebastes aleutianus (Jordan and Evermann, 1898) and S. melanostictus (Matsubara, 1934). The powerful combination of genetic and morphological analyses provides a means to identify the important characters useful in distinguishing new specimens, as well as archived material that, in some cases, has been preserved for over a century. Ranging around the rim of the North Pacific Ocean and Bering Sea, from Japan to southern California, the species presently recognized as Sebastes aleutianus was originally described by Jordan and Evermann (1898) from four specimens taken in waters around Kodiak Island, Alaska. Evermann and Goldsborough (1907) later described S. swifti, a synonym of Sebastes aleutianus, from the north- eastern Gulf of Alaska. In the western Pacific, Matsubara (1934) described two species similar to Sebastes aleutianus, S. melanost- ictus and S. kawaradae. Barsukov (1964, 1970) determined that the names of both of these species were synonyms of S. aleutianus, apparent- ly without examining type material. Kanayama and Kitagawa (1982) later examined the types and comparative material used by Matsubara (1934, 1943) as well as new material of S. aleutianus from the Bering Sea and Japanese Pacific coast. They also con- 112 Fishery Bulletin 106(2) eluded that both S. melanostictus and S. kawaradae are synonymous with S. aleutianus. These decisions were subsequently followed by Sheiko and Fedorov (2000), Love et al. (2002), Mecklenburg et al. (2002), and Nak- abo (2002). The confused history of the nomenclature of Se- bastes aleutianus has been a result of the perceived wide variation in body coloration and close similarity in morphological features of several other species of Sebastes, including S. borealis (Barsukov, 1970), short- raker rockfish, and S. melanostomus (Eigenmann and Eigenmann, 1890), blackgill rockfish, in the eastern Pacific. Gilbert (1896) misidentified S. aleutianus , and probably S. borealis, material from the Bering Sea as that of S. introniger (Gilbert, 1890), which was origi- nally described from southern California and is now considered a synonym of S. melanostomus (Phillips, 1957; Tsuyuki and Westrheim, 1970). In the same pub- lication as their original description of S. aleutianus, Jordan and Evermann (1898) provided a species account of S. introniger that was primarily based on Gilbert’s earlier (1896) redescription (Tsuyuki and Westrheim, 1970), again listing its range as including the Bering Sea. Until clarified by Barsukov (1970) and Tsuyuki and Westrheim (1970) with their original descriptions of the new species S. borealis Barsukov and its synonym S. caenaematicus Tsuyuki and Westrheim, these reports led to persistent records in the literature of S. melano- stomus having been taken in the Bering Sea (Jordan and Evermann, 1898; Evermann and Goldsborough, 1907; Jordan et al., 1930; Barsukov, 1964; Allen and Smith, 1988), although it is a species that may range north only to extreme southeastern Alaska (Butler and Love, 2002; Mecklenburg et al., 2002; Love et al., 2002, 2005; Kramer and O’Connell, 2003). The early work of Tsuyuki and coauthors (Tsuyuki et al., 1965, 1968; Tsuyuki and Westrheim, 1970) and Seeb (1986) on hemoglobin and allozymes of eastern Pacific S. aleutianus provided evidence for species-level differ- ences among morphological and color forms — differences that were correlated with genetic variants. Using larger sample sizes and broader geographic ranges, Hawkins et al. (2005) with allozyme data and Gharrett et al. (2005) with DNA markers provided conclusive evidence of species-level differences among individuals in Alas- kan waters, and the reality of the presence of two spe- cies within S. aleutianus was confirmed. In this revision of the complex presently identified as Sebastes aleutianus, we recognize two species: Sebastes aleutianus, the rougheye rockfish, restricted to the east- ern Pacific; and Sebastes melanostictus, the blackspotted rockfish, ranging across the North Pacific from Japan to California. Sebastes kawaradae is considered a synonym of S. melanostictus, and S. swifti is a synonym of S. aleutianus. We expand upon the efforts of Gharrett et al. (2006), who distinguished dark and light forms in the Gulf of Alaska, extending our morphological exami- nation across almost the entire range of both species, clarifying the nomenclature, and providing diagnoses and redescriptions of both species. Methods and materials Counts, measurements, and statistical analyses fol- lowed Orr and Blackburn (2004), except as noted below. The structure of the swimbladder was examined after dissection following the method of Hallacher (1974). Unless indicated otherwise, standard length (SL) is used throughout, always measured from the tip of the snout. Measurements and counts are presented in spe- cies descriptions as the range for all material examined followed by the value for the holotype or lectotype in parentheses, when intraspecific variation is indicated. Collection depths are noted for each cataloged lot in the Material examined section (see Appendix) when known. In the analysis of depth distributions, only those lots with a single depth reported were used; when a range of depths was reported for a lot, it was not included. All depths reported are bottom depths unless otherwise stated. Institutional abbreviations follow Eschmeyer (1998). Survey records for “rougheye rockfish” ( Sebastes aleutianus or S. melanostictus) were taken from the Resource Assessment and Conserva- tion Engineering Division, Alaska Fisheries Science Center (AFSC), database, which included catch data from groundfish surveys conducted from the Bering Sea and Aleutian Islands to southern California from 1961 to 2005. During initial genetic analyses, 124 individuals were identified as S. aleutianus or S. melanostictus (=S. sp. cf. aleutianus of Hawkins et al., 2005) by using allozyme data following Hawkins et al. (2005) and are marked with an asterisk in the Material examined section (see Appendix). These specimens were collected primarily in the northern Gulf of Alaska and off Washington; two specimens, however, were collected from the eastern Aleutian Islands and northern California. Based on examination of these genetically identified individuals, significant differences were found in body color, specifi- cally spotting on the spinous dorsal fin or an overall darker color (or both) in S. melanostictus (see Species descriptions below), that were then used to identify preserved specimens for which tissue was not avail- able for genetic analysis. Allozyme characteristics and body color were used to group individuals for univari- ate tests, as well as for labeling individuals in graphs of principal components analysis scores. Nomenclature of allozyme protein variants and microsatellite alleles follows conventions of the American Fisheries Society (Shaklee et al., 1990). Statistical analyses were per- formed with Statgraphics Plus 4.1 (Manugistics, Rock- ville, MD), Splus 6.2 (Insightful Corp., Seattle, WA), and SPSS 11.5.1 (SPSS Inc., Chicago, IL). Differences were considered significant at P < 0.05. Out of a total of 329 specimens examined morphologi- cally, many had broken spines or other missing charac- ters; thus, the data set was reduced to 137 specimens with complete data, including 40 of those genetically identified, for which both univariate and multivariate analyses could be conducted. A suite of 29 morphometric and 5 meristic characters was analyzed (Table 1). All Orr and Hawkins: Species of the rougheye rockfish complex 113 Table 1 Proportional morphometries and meristics of Sebastes aleutianus (rougheye rockfish) and Sebastes melanostictus (blackspotted rockfish) for specimens in which all characters were used for multivariate analyses. Morphometric data are given in percent SL (standard length) and presented as the range, followed by the mean ± standard deviation (SD). P values are reported for sta- tistically significant differences at a level of 0.05, as evaluated by ANCOVA or Kruskal-Wallis tests when appropriate; ns = not statistically significant at an a level of 0.05. Sebastes aleutianus Sebastes melanostictus Range (mean ±SD) Range (mean ±SD) P n 57 80 Standard length (mm) 77.1-470 101.6-445 Meristics Dorsal-fin rays 12-14 (13.4 ±0.5) 13-15 (13.8 ±0.5) <0.0001 Anal-fin rays 6-8(7 ±0.3) 6-8(7 ±0.2) ns Pectoral-fin rays (left) 17-19 (18.1 ±0.3) 17-19 (18.1 ±0.3) ns Lateral-line pores (left) 30-34 (31.2 ±0.9) 30-36(32 ±1.2) <0.0001 Gill rakers 29-33 (31.1 ±1.0) 30-35 (33 ±1.1) <0.0001 Morphometries Head length 34.6-42.5 (37 ±1.6) 34.6-41.4 (37.6 ±1.5) Orbit length 8.8-13.5 (10.7 ±1.1) 8.2-13.3 (10.4 ±1.0) ns Snout length 5. 8-9. 7 (7.5 ±0.7) 6. 6-9. 5 (7.9 ±0.6) 0.0486 Interorbital width 5.5-8. 9 (7.1 ±0.7) 5. 6-8. 8 (7.4 ±0.7) Suborbital depth 1.6-2. 9 (2.3 ±0.3) 1.5-3. 3 (2.3 ±0.4) 0.0019 Upper jaw length 16.9-21.9 (19.2 ±0.9) 17.2-21.5 (19.2 ±0.7) Lower jaw length 20.4-27.2(23.4 ±1.2) 20.7-26.8(24.5 ±1.1) Gill raker length 3. 4-6.1 (4.8 ±0.6) 3. 9-6. 8 (5.6 ±0.5) <0.0001 Depth at pelvic-fin base 31.6-42.6(36 ±1.9) 31.3-39.7 (34.9 ±1.8) Depth at anal-fin origin 23.9-31.9(28 ±1.6) 23.5-29.4 (26.2 ±1.5) <0.0001 Depth at anal-fin insertion 11.9-15.3 (13.9 ±0.7) 11.3-15.3 (13.3 ±1.0) <0.0001 Dorsal-fin spine I length 4. 6-7. 2 (5.8 ±0.6) 64-9.6(7.9 ±0.8) <0.0001 Dorsal-fin spine IV length 10.6-15 (13.2 ±0.9) 12.3-18.9 (14.8 ±1.3) <0.0001 Pectoral-fin base depth 8.4-10.4(9.4 ±0.4) 8.2-10.4 (9.5 ±0.4) ns Pectoral-fin ray length 25-31 (27.4 ±1.3) 22-30 (26.3 ±1.4) 0.0013 Pelvic-fin ray length 19.3-25.1 (22.3 ±1.1) 19.4-25 (21.5 ±1.2) 0.0298 Pelvic-fin spine length 11.1-15.5 (13.5 ±1.2) 11.4-16.9 (13.8 ±1.2) <0.0001 Anal-fin spine I length 4. 7-9. 4 (6.6 ±1.0) 4.7-10.1 (7.1 ±1.2) <0.0001 Anal-fin spine II length 11-18.4(14.4 ±1.8) 11.8-20.9 (15.2 ±2.2) <0.0001 Anal-fin spine III length 14.2-19.8 (16.5 ±1.3) 12.4-20.9 (16.6 ±1.7) <0.0001 Caudal peduncle depth 7.8-10.7 (9.7 ±0.5) 8.2-10.6 (9.2 ±0.5) Caudal peduncle dorsal length 10.5-15.3 (12.2 ±0.9) 11.3-15 (13 ±0.8) Caudal peduncle ventral length 17.8-21.9 (19.6 ±0.9) 17.4-21.7 (20.1 ±0.9) Preanal length 67.1-77.6 (71.8 ±2.5) 63.2-78.5 (70 ±2.6) <0.0001 Predorsal length 31.2-39.5 (34.3 ±1.9) 31.6-38.4 (35 ±1.4) Spinous dorsal-fin-base length 31.9-41.7 (35.5 ±2.3) 32.4-40.9 (36.3 ±2.0) ns Soft dorsal-fin-base length 20.7-25.6 (22.9 ±1.2) 17.7-24.7 (21.5 ±1.4) <0.0001 Anal-fin-base length 13.3-18.2 (15.3 ±0.9) 12.4-17 (14.2 ±1.0) ns Pre-pelvic-fin length 37.8-53.1 (44.4 ±4.2) 36.6-50.5 (42.4 ±2.5) morphometric characters were also tested for sexual dimorphism, where gender was treated as a categorical factor and the gender-species interaction was tested. Arcsine-transformed morphometric ratios (with SL or head length [HL] as denominator) were tested to meet the assumptions of normality and equality of variance required for ANCOVA. The following characters exhib- ited normal distributions and did not differ significantly in variance between species and thus were subjected to ANCOVA: orbit length, snout length, suborbital depth, gill-raker length, body depth at anal-fin origin, body depth at anal-fin insertion, dorsal-fin spine 1 length, dorsal-fin spine 4 length, pectoral-fin base depth, pec- toral-fin ray length, pelvic-fin ray length, pelvic-fin 114 Fishery Bulletin 106(2) spine length, anal-fin spine 1 length, anal-fin spine 3 length, preanal length, spinous-dorsal-fin base length, soft-dorsal-fin base length, and anal-fin base length. Counts of meristic characters were tested by using the Kruskal-Wallis test. To aid in distinguishing the two species, a standard principal component analysis (PCA) was conducted on both morphometric and meristic characters. The analy- ses were conducted on the covariance matrix of log- transformed raw morphometric data and the correlation matrix of raw meristic data. Differences between spe- cies were illustrated by plotting principal component (PC) 2 against PC3 of the morphometric analysis, PCI against PC2 of the meristic analysis, and morphometric PC2 against meristic PCI. Following the PCAs, a stepwise discriminant func- tion analysis (DFA) was conducted by using morpho- metric and meristic data to establish the relative sig- nificance of those characters in distinguishing between the species. Morphometric data were standardized by dividing by standard length. Only characters meeting assumptions of multivariate normality and that exhib- ited statistically significant differences were analyzed. The robustness of the DFA was tested by conducting a leave-one-out cross-validation procedure (SPSS Inc., Chicago, IL). For the data set containing all 329 specimens, only univariate analyses were conducted. The following char- acters, in addition to all those except orbit length in the reduced dataset, exhibited normal distributions, did not differ significantly in variance between species, and were subjected to ANCOVA: head length, lower-jaw length, body depth at pelvic-fin base, anal-fin spine 2 length, caudal-peduncle depth, caudal-peduncle dor- sal length, caudal-peduncle ventral length, predorsal length, and pelvic-fin base to anal-fin origin length. Differences in counts of meristic characters were tested using the nonparametric Kruskal-Wallis test. With the exception of unbranched pectoral-fin rays, meristic data from specimens of all sizes were tested. In juveniles of less than 100 mm, all pectoral-fin rays were simple and counts for these specimens were not included in tests or presented in tables of lower pectoral-fin rays. Meristic data is presented in tables of frequency distributions by region. Four general regions were identified: 1) west- ern Pacific Ocean, for material taken in Japanese and Russian waters; 2) Bering Sea and Aleutian Islands, for material from the eastern Bering Sea and Aleutian Islands to Unimak Pass; 3) Gulf of Alaska, from Uni- mak Pass to the Alaska-British Columbia border; and 4) the Lower West Coast, from Canada, Washington, Oregon, and California. Results Color pattern Both in life and in preservation, body color differs con- sistently between S. aleutianus and S. melanostictus (Figs. 1-4 ; see detailed description below). The spinous dorsal fin in S. melanostictus is almost invariably dis- cretely spotted. In most specimens, many small spots are scattered across the spinous dorsal fin, often continuing onto the soft dorsal fin; in some individuals, only two or three spots are present on the spinous dorsal fin. The presence or absence of these discrete spots was used in initial statistical analyses as the basis for identifying preserved specimens lacking tissue for genetic analysis. A few dark individuals without spots, or with spots apparently obscured by dark blotching, were aligned morphologically with other spotted individuals, and these dark individuals were also eventually identified as S. melanostictus. In contrast, all S. aleutianus examined were pale overall and lacked discrete spots, although often having diffuse mottling and blotches extending from the body onto the bases of the spinous and soft dorsal fins. Morphometric and meristic characters Specimens of S. melanostictus generally had longer dorsal-fin spines (Fig. 5, A and B), other spines, and gill rakers (Fig. 5C), whereas specimens of S. aleutianus had a deeper, more robust head and body and longer pelvic- and pectoral-fin rays. Among morphometric char- acters of specimens with all characters, snout length, suborbital depth, gill raker length, body depth at anal- fin origin, body depth at anal-fin insertion, dorsal-fin spine 1 length, dorsal-fin spine 4 length, pectoral-fin ray length, pelvic-fin ray length, pelvic-fin spine length, anal-fin spine 1 length, anal-fin spine 2 length, anal-fin spine 3 length, preanal length, and soft-dorsal-fin base length differed significantly between S. aleutianus and S. melanostictus (Table 1). In addition, among specimens of the larger data set, head length, lower jaw length, depth at pelvic-fin base, caudal-peduncle depth, caudal- peduncle dorsal length, and caudal-peduncle ventral length also differed significantly, and anal-fin spine 1 and 3 lengths became nonsignificant (Table 2). Plots of PCA scores revealed prominent differences between the two species among morphometric char- acters, as well as slight differences among meristic characters. In the morphometric PCA, all characters were positively loaded on PCI (the size component; Table 3), which explained 96.8% of the total variation. Among the principal shape components, PC2 explained 1.3% of variation and was heavily loaded on dorsal- fin spine 1 length, gill-raker length, anal-fin spine 1 length, anal-fin spine 2 length, and dorsal-fin spine 4 length (Table 3); PCS explained 0.3% of variation and was heavily loaded on anal-fin spine 1 length, suborbital depth, gill-raker length, prepelvic length, and orbit length (Table 3). In the plot of PC2 versus PC3, the two clusters representing S. aleutianus and S. melanostictus overlapped narrowly; 16 (8 S. aleutia- nus, 8 S. melanostictus) of 137 individuals examined were in this area of overlap (Fig. 6A). Of 40 genetically identified individuals, three S. aleutianus and three S. melanostictus were included in the overlap area. One Orr and Hawkins: Species of the rougheye rockfish complex 115 specimen with an unknown genotype, a possible hybrid, was also included. Although meristic characters broadly over- lapped in ranges, several characters were in- dicative of a species-level difference between S. aleutianus and S. melanostictus. Numbers of gill rakers, lateral-line pores, and dorsal-fin rays dif- fered significantly (Tables 1 and 2). No regional or clinal differences were evident in frequency data (Tables 4 and 5). Significant differences were also expressed in the meristic PCA as heavy loadings on PCI (Table 6), which, however, revealed only a slight separation between broadly overlapping clusters of the two species when scores were plotted for PCI and PC2 (Fig. 6B). Plotting the scores of PCI of the meristic anal- ysis versus PC2 of the morphometric analysis resulted in slightly better resolution between the two species clusters (Fig. 60. In contrast with the morphometrics-only plot that had 16 indi- viduals in the area of overlap and the meristics- only plot in which the clusters nearly completely overlapped, only nine individuals were found in the area of overlap when the morphometric PC2 was plotted against the meristic PCI. As in the morphometrics-only plot, six of these individuals were genetically identified. Figure 2 Underwater photo of Sebastes aleutianus (rougheye rockfish) taken south of Kodiak Island, Alaska, ca. 56°N, 153°N, July 2005. Photo by D. Hanselman. 116 Fishery Bulletin 106(2) Figure 3 Sebastes melanostictus (blackspotted rockfish), UW 48464, 249.5 mm, male, Aleutian Islands, north of Islands of Four Mountains, 53.2207°N, 169.7395°W, 328 m depth, 2 June 2002. Photo by J. W. Orr. Discriminant function analysis Although the PCA produced overlapping species clusters of individuals identified and labeled separately from the analysis, the discriminant function analysis verified the a priori identification of nearly all individuals. The single linear discriminant function equation produced was highly significant (Wilks’s A=0.197, x2=222.772, 8 df, P<0.0001): D = 101 . 557 (c? 1 ) + 52.453(s/iZ) + 0.294(gr) + 5l.92(grl) + 0.564(cZr) - 38.604(p2rZ) - 22.601(d2Z>) - 10.203(paZ) - 10.445, where D = the discriminant score of an individual; dl = length of dorsal-fin spine 1 divided by SL; snl = snout length divided by SL; gr = number of gill rakers; grl = length of gill rakers divided by SL; Figure 4 Underwater photo of Sebastes melanostictus (blackspotted rockfish) taken south of Kodiak Island, Alaska, ca. 56°N, 153°N, July 2005. Photo by D. Hanselman. dr = number of dorsal-fin rays; p2rl - length of pelvic-fin rays divided by SL; d2b - length of soft-dorsal-fin base; and pal - preanal length divided by SL. All but two individuals with negative scores had been previ- ously identified as S. aleutianus, whereas all but one individual with positive scores were S. melanostictus. The discrimi- nant function equation, there- fore, correctly classified 97.8% of individuals. The possible hybrid was classified correctly as S. aleutianus. The cross-validation procedure correctly classified 97.1% of individuals. In addition to the three individuals above, one S. aleutianus was also mis- classified as S. melanostictus in cross validation. When using this equation for future species Orr and Hawkins: Species of the rougheye rockfish complex 117 o Sebastes aleutianus m Sebaste melanostictus a ft ° '! * o 7° Figure 5 Plots of diagnostic morphometric characters versus standard length of Sebastes aleutianus (rough- eye rockfish, open diamond) and S. melanostictus (blackspotted rockfish, closed square). (A) Dorsal-fin spine I length, (B) dorsal-fin spine IV length, and (C) gill raker length. The first and fourth dorsal-fin spines and gill rakers are longer in most specimens of S. melanostictus than in S. aleutianus. separation, one would identify individuals with a score above 0 as S. melanostictus and those below 0 as S. aleutianus. Systematics Sebastes aleutianus (Jordan and Evermann, 1898) Rougheye rockfish Figures 1, 2, 5—8; Tables 1, 2, 4, 5 Sebastodes aleutianus Jordan and Evermann, 1898:1795, pi. 16, figs. 1-4 (original description, -0.2 -0.25 -0.2 -0.15 -0.1 -0.05 0 0.05 0.1 0.15 0.2 0.25 PC2 PC2 PC2 (morphometric) Figure 6 Plots of principal component (PC) scores for morphometric and meristic characters for Sebastes aleutianus (rougheye rockfish, open diamond) and S. melanostictus (blackspotted rockfish, closed square). (A) Morphometric characters only, (B) meristic characters only, and (C) morphometric (PC2) versus meristic characters (PCI). Differences among specimens along the major axis of dispersion in the morphometric analysis (PC2) were primarily due to longer dorsal-fin spines, anal-fin spines, and gill rakers in S. melanostictus. The slight separation among specimens along the major axis in the meristic analysis (PC2) was primarily due to higher counts of gill rakers, lateral-line pores, and dorsal-fin rays in S. melanostictus. four specimens: lectotype herein designated USNM 48800, male, 374.2 mm; paralectotypes, three speci- mens 370-470 mm, SU 12928 , Albatross station 3676, Shelikof Strait, Alaska, off Karluk, Kodiak Island, 223 m depth, 20 July 1897). 118 Fishery Bulletin 106(2) Table 2 Proportional morphometries and meristics of Sebastes aleutianus (rougheye rockfish) and Sebastes melanostictus (blackspotted rockfish) for all specimens examined. Morphometric data are given in percent SL (standard length) and presented as the range, followed by the mean ± standard deviation (SD). P values are reported for statistically significant differences at 0.05 level, as evaluated by ANCOVA or Kruskal-Wallis tests when appropriate; ns = not statistically significant at an « level of 0.05. Sebastes aleutianus Sebastes melanostictus n Range (mean ±SD) n Range (mean ±SD) P Standard length (mm) 117 63.4-555.2 193 95.5-539 Meristics Dorsal-fin spines 114 12-14 (13.0 ±0.2) 182 12-14 (13.0 ±0.2) ns Dorsal-fin rays 109 12-14 (13.5 ±0.5) 184 13-15 (13.7 ±0.5) <0.001 Anal-fin rays 108 6-8 (7.1 ±0.3) 182 6-8 (7.1 ±0.3) ns Pectoral-fin rays 106 17-19(18.1 ±0.3) 170 17-19 (18.1 ±0.4) ns Unbranched pectoral-fin rays 99 4-9 (7.7 ±0.9) 177 5-10(7.9 ±0.7) ns Lateral-line pores 105 30-34(31.4 ±1.0) 184 30-36(32.0 ±1.2) <0.001 Gill rakers 116 29-34 (31.2 ±1.0) 191 30-36 (33.0 ±1.2) <0.001 Gill rakers of upper arch 98 8-10(9.1 ±0.5) 165 8-11 (9.6 ±0.6) <0.001 Gill rakers of lower arch 98 21-24(22.1 ±0.8) 165 21-26 (23.4 ±0.8) <0.001 Morphometries Head length 106 33.5-41.4 (36.8 ±1.4) 179 34-41.4 (37.6 ±1.4) 0.006 Orbit length 93 8.7-13.5 (10.7 ±1.1) 145 8.2-13.3 (10.2 ±0.9) Snout length 93 5. 7-9. 7 (7.5 ±0.7) 143 6. 2-9. 6 (8 ±0.6) <0.001 Interorbital width 93 5. 5-8. 9 (7.2 ±0.7) 141 5.6-9. 2 (7.5 ±0.7) Suborbital depth 90 1.6-2. 9 (2.3 ±0.3) 141 1.5-3. 3 (2.3 ±0.3) ns Upper jaw length 89 16.9-21.1(19.1 ±0.7) 141 17.2-21.5 (19.2 ±0.8) Lower jaw length 91 20.4-27.2 (23.6 ±1.2) 138 21.6-27.1 (24.6 ±1.0) <0.001 Gill raker length 106 3.4— 6.3 (4.9 ±0.6) 178 3. 9-7.1 (5.6 ±0.6) <0.001 Depth at pelvic-fin base 89 31.6-42.6 (36.1 ±1.9) 143 31.3-40.1 (35.3 ±1.9) 0.005 Depth at anal-fin origin 97 23.5-31.9 (27.9 ±1.6) 154 22.8-30.5 (26.6 ±1.6) <0.001 Depth at anal-fin insertion 90 11.4-15.6 (13.8 ±0.8) 144 11.3-15.6(13.5 ±0.9) 0.003 Dorsal-fin spine I length 91 4. 3-7.2 (5.8 ±0.6) 135 5. 9-9. 6 (7.8 ±0.7) <0.001 Dorsal-fin spine IV length 97 9.8-15.5 (13.2 ±1.0) 151 12-18.4 (14.8 ±1.2) <0.001 Dorsal-fin ray length 82 13.5-20(16.6 ±1.3) 135 10.4-18.8 (15.5 ±1.3) Pectoral-fin base depth 88 8.4-10.4 (9.4 ±0.4) 142 8.2-10.4(9.5 ±0.4) ns Pectoral-fin ray length 90 24.4-31 (27.3 ±1.3) 140 22-30 (26.3 ±1.4) <0.001 Pelvic-fin ray length 92 19.3-25.1 (22.1 ±1.1) 159 18.8-25 (21.4 ±1.2) <0.001 Pelvic-fin spine length 103 9.5-16.6 (13.4 ±1.3) 160 9.6-16.9(13.7 ±1.2) ns Anal-fin spine I length 94 4. 5-9.7 (6.8 ±1.1) 156 4.5-10.1 (7 ±1.1) ns Anal-fin spine II length 94 11-18.4(14.5 ±1.9) 147 11.8-19.6 (14.9 ±1.9) ns Anal-fin spine III length 95 13-20.2 (16.6 ±1.5) 153 12.2-21.3 (16.5 ±1.7) ns Anal-fin ray 1 length 63 16-24.2 (19.5 ±1.6) 119 15.3-25 (18.9 ±1.8) Anal-fin ray 2 length 62 17.4-26.4(21.1 ±1.8) 115 16.7-26.9(19.9 ±2.0) Last anal-fin ray length 62 6-12.6 (10.4 ±1.1) 113 8.3-13.7 (10.4 ±0.8) Caudal peduncle depth 90 7.8-10.7 (9.6 ±0.5) 141 7.9-10.6(9.2 ±0.5) <0.001 Caudal peduncle dorsal length 87 10.5-13.9 (12 ±0.8) 141 11.1-15 (12.9 ±0.7) <0.001 Caudal peduncle ventral length 89 17.7-21.9(19.5 ±0.9) 141 17.4-22.5 (20.1 ±1.0) <0.001 Preanal length 89 67.1-77.6 (71.8 ±2.3) 141 63.2-78.5 (70.2 ±2.6) <0.001 Predorsal length 90 30.3-39.3 (34.3 ±1.7) 141 31.6-38.4 (34.9 ±1.4) ns Pelvic-fin base to anal-fin origin length 87 25.9-37 (30.8 ±2.6) 139 23.8-38.6 (30.9 ±2.9) ns Spinous dorsal-fin-base length 91 30.8-41.7 (35.7 ±2.3) 141 32.4-41.2 (36.4 ±2.0) ns Soft dorsal-fin-base length 90 20.2-25.8 (22.8 ±1.2) 141 17.3-24.7 (21.4 ±1.5) <0.001 Anal-fin base length 89 13.1-18.2 (15.2 ±0.9) 141 12.4-17 (14.2 ±1.0) Pre-pelvic-fin length 87 37.8-53.1 (44.2 ±3.8) 139 36.6-52.6 (42.6 ±2.8) Caudal-fin length 80 20-28.1 (24 ±1.5) 128 19.2-25.2 (22 ±1.2) Orr and Hawkins: Species of the rougheye rockfish complex 119 Sebastodes swifti Evermann and Goldsborough, 1907:285, fig. 36 (original description, two specimens: holotype, USNM 57821, sex unknown, 128.2 mm , Albatross sta- tion 4234, Yes Bay, Alaska; paratype, USNM 126656, sex unknown, 75.2 mm , Albatross station 4246, Kasaan Bay, Alaska, 185-225 m depth, 11 July 1903). Zalopyr aleutianus: Jordan et al., 1930:365 (new com- bination). Sebastes aleutianus'. Barsukov, 1970:997 (new combina- tion; comparison with S. borealis). Diagnosis This species of Sebastes is distinguished from all other species except S. melanostictus in having eight pairs of head spines and two or more infraorbital spines. It is distinguished from S. melanostictus by the following combination of color patterns and morphological and genetic characteristics: body pale, distinct spots absent on the membranes of the spinous dorsal fin; dorsal-fin spine 1 shorter, 1. 5-3.0 times into orbit length, 4. 3-7.2% SL (vs. 1.0-1. 8 times into orbit length and 5. 9-9. 6% SL); dorsal-fin spine 4 shorter, 9.8-15.5% SL (vs. 12.0-18.4% SL); gill rakers shorter, 3. 4-6. 3% SL (vs. 3. 9-7.1% SL), and fewer, 29-34 (vs. 30-36) (Table 2). Genetically diag- nosed by a combination of the presence of protein vari- ants ACP*100, either IDDH*100 or * 500 or both, either PGM-2*100 or *83 or both, and XO*100 (Hawkins et al., 2005); homozygosity at microsatellite allele pSma6*183 (Gharrett et al., 2005, 2006); and mitochondrial haplo- type B of Gharrett et al. (2005, 2006). Description D XII-XIV (XIII), 12-14 (14); A III, 6-8 (8); PI 17-19 (18), 4-9 (6) simple; lateral-line pores 30-34 (31), scales 42-57; infraorbital spines 2-10 (9); gill rakers 29-34 (8-10 + 21-24) (31:9+22); vertebrae 27 (10+17); pyloric caeca 9-12. Meristic frequency and statistical data are presented in Tables 1, 2, 4, and 5. Table 3 Factor loadings for principal component (PC) analysis of morphometric characters of Sebastes aleutianus (rougheye rockfish) and Sebastes melanostictus (blackspotted rockfish), in which all characters were used for multivariate analyses. PCI PC2 PC3 Head length 0.1911 -0.0190 -0.1584 Orbit length 0.1538 0.0215 -0.2248 Snout length 0.2020 -0.0121 -0.2133 Interorbital width 0.2261 -0.0922 -0.0498 Suborbital depth 0.2288 -0.2145 0.3787 Upper jaw length 0.1936 -0.0528 -0.0943 Lower jaw length 0.2026 0.0024 -0.1023 Gill raker length 0.1753 0.3399 -0.3728 Depth at pelvic-fin base 0.1960 -0.1374 -0.0036 Depth at anal-fin origin 0.1944 -0.2055 0.0864 Depth at anal-fin insertion 0.1929 -0.1621 0.0306 Dorsal-fin spine I length 0.1982 0.6188 0.1775 Dorsal-fin spine IV length 0.1755 0.2379 -0.0909 Pectoral-fin base width 0.1905 -0.0489 0.0642 Pectoral-fin ray length 0.1741 -0.0983 -0.0211 Pelvic-fin ray length 0.1746 -0.0690 -0.0030 Pelvic-fin spine length 0.1654 0.0766 0.0457 Anal-fin spine I length 0.1417 0.3270 0.4728 Anal-fin spine II length 0.1369 0.2472 0.1530 Anal-fin spine III length 0.1586 0.1029 0.1761 Caudal peduncle depth 0.1913 -0.1736 0.0632 Caudal peduncle dorsal length 0.1975 0.0578 -0.1842 Caudal peduncle ventral length 0.1869 0.0142 -0.1128 Preanal length 0.1881 -0.0926 0.0627 Predorsal length 0.1876 -0.0057 -0.2195 Spinous dorsal-fin-base length 0.1957 -0.0223 -0.2150 Soft dorsal-fin-base length 0.1804 -0.1481 0.0548 Anal-fin-base length 0.1654 -0.1322 0.1176 Pre-pelvic-fin length 0.1874 -0.1256 0.2757 120 Fishery Bulletin 106(2) Table 4 Counts of dorsal-fin spines, soft-dorsal-fin rays, anal-fin rays, and total and unbranched pectoral-fin rays in Sebastes aleutianus (rougheye rockfish) and Sebastes melanostictus (blackspotted rockfish). Lower West Coast = waters off the coast of British Colum- bia, Washington, Oregon, and California. Western Pacific Ocean = waters off the coast of the Kuril Islands and Japan. Dorsal-fin spines Total left pectoral-fin rays 12 13 14 n 17 18 19 n Sebastes aleutianus Sebastes aleutianus Lower West Coast 29 1 Lower West Coast 30 Gulf of Alaska 1 63 2 Gulf of Alaska 2 55 7 Bering Sea and Aleutian Is. 18 Bering Sea and Aleutian Is. 9 3 Total 1 110 3 114 Total 2 94 10106 Sebastes melanostictus Sebastes melanostictus Lower West Coast 17 1 Lower West Coast 16 1 Gulf of Alaska 1 80 4 Gulf of Alaska 2 72 12 Bering Sea and Aleutian Is. 2 53 Bering Sea and Aleutian Is. 4 50 4 Western Pacific Ocean 24 Western Pacific Ocean 20 Total 3 174 5 182 Total 6 158 17181 Dorsal-fin rays Unbranched left pectoral-fin rays 12 13 14 15 n 4 5 6 7 8 9 10 n Sebastes aleutianus Sebastes aleutianus Lower West Coast 11 19 Lower West Coast 1 3 20 4 Gulf of Alaska 1 36 30 Gulf of Alaska 1 1 5 9 39 4 Bering Sea and Aleutian Is. 5 13 Bering Sea and 1 5 4 2 Total 1 52 62 115 Aleutian Is. Sebastes melanostictus Total 1 1 7 17 63 10 99 Lower West Coast 5 13 Sebastes melanostictus Gulf of Alaska 32 53 Lower West Coast 1 5 9 1 Bering Sea and Aleutian Is. 17 37 1 Gulf of Alaska 10 58 13 Western Pacific Ocean 6 14 4 Bering Sea and 1 9 38 10 Total 60 117 5 182 Aleutian Is. Western Pacific Ocean 8 10 2 Anal-fin rays Total 2 32 115 25 1 175 6 7 8 n Sebastes aleutianus Lower West Coast 1 29 Gulf of Alaska 1 59 6 Bering Sea and Aleutian Is. 10 2 Total 2 98 8 108 Sebastes melanostictus Lower West Coast 17 1 Gulf of Alaska 2 75 7 Bering Sea and Aleutian Is. 53 5 Western Pacific Ocean 20 Total 2 165 13 180 Body robust, depth at pelvic-fin base 31.6-42.6 (38.1) % SL; relatively deep at anal-fin origin 23.5-31.9 (28.0) % SL; profile of dorsal margin of head gently sloping from dorsal-fin origin to snout, dorsal rim of orbit in- cluded in lateral margin of frontals. Mouth large, with posterior end of maxilla extending between pupil and posterior rim of orbit to just beyond the posterior rim of the orbit; maxilla length 16.9-21.1 (19.9) % SL; sym- physeal knob moderately pronounced with blunt tip, lower-jaw length 20.4-27.2 (25.0) % SL; mandibular pores large to moderate in size. Cranial spines strong, often rough. Nasal, preocular, supraocular, postocular, tympanic, coronal, parietal, and supratemporal spines invariably present; all major head spines strong, ex- cept for coronal spines, which are weak; supraocular and postocular spines and parietal ridge often rough; Qrr and Hawkins: Species of the rougheye rockfish complex 121 Table 5 Counts of gill rakers, lateral-line pores, and left infraorbital spines in Sebastes aleutianus (rougheye rockfish) and Sebastes melanostictus (blackspotted rockfish). Lower West Coast = waters off the coast of British Columbia, Washington, Oregon, and California. Western Pacific Ocean = waters off the coast of the Kuril Islands and Japan. Gill rakers 29 30 31 32 33 34 35 36 n Sebastes aleutianus Lower West Coast 1 8 15 6 3 Gulf of Alaska 5 8 28 19 6 2 Bering Sea and Aleutian Is. 5 4 6 Total 6 16 43 25 9 2 101 Sebastes melanostictus Lower West Coast 1 3 3 4 7 1 Gulf of Alaska 1 9 25 22 26 10 Bering Sea and Aleutian Is. 1 7 12 24 10 3 2 Western Pacific Ocean 1 5 6 5 3 Total 4 19 45 56 48 17 2 191 Gill rakers of the upper arch 8 9 10 11 n Sebastes aleutianus Lower West Coast 4 21 4 Gulf of Alaska 3 39 13 Bering Sea and Aleutian Is. 4 10 Total 11 70 17 98 Sebastes melanostictus Lower West Coast 2 12 1 Gulf of Alaska 2 32 40 6 Bering Sea and Aleutian Is. 3 21 23 1 Western Pacific Ocean 9 9 2 Total 5 64 84 10 163 Gill rakers of the lower arch 21 22 23 24 25 26 n Sebastes aleutianus Lower West Coast 7 14 7 1 Gulf of Alaska 10 31 10 4 Bering Sea and Aleutian Is. 2 6 6 Total 19 51 23 5 98 Sebastes melanostictus Lower West Coast 2 5 8 Gulf of Alaska 1 9 25 40 5 Bering Sea and Aleutian Is. 1 6 18 18 4 1 Western Pacific Ocean 1 3 8 7 1 Total 3 20 56 73 10 1 163 continued sphenotic spines often prominent, occasionally moder- ate or small in size. Interorbital region wide, 5. 5-8. 9 (8.7) % SL, flat to slightly concave; frontal ridges be- tween orbits moderate to strong; parietal ridges strong, with area between ridges slightly convex or flat. Pre- opercular spines 5, directed posteroventrally; opercular spines 2, upper spine directed posteriorly, lower spine directed posteriorly and slightly ventrally. Posttempo- 122 Fishery Bulletin 106(2) Table 5 (continued) 30 31 32 Laterabline pores 33 34 35 36 n Sebastes aleutianus Lower West Coast 2 17 9 1 1 Gulf of Alaska 8 32 15 5 2 Bering Sea and Aleutian Is. 2 3 4 2 2 Total 12 52 28 8 5 105 Sebastes melanostictus Lower West Coast 1 6 1 6 2 Gulf of Alaska 7 29 37 13 2 1 Bering Sea and Aleutian Is. 3 12 24 12 5 1 Western Pacific Ocean 5 6 3 3 1 2 Total 11 52 68 34 12 2 3 182 Left infraorbital spines 2 3 4 5 6 7 8 9 10 11 12 n Sebastes aleutianus Lower West Coast 1 9 8 6 4 3 1 Gulf of Alaska 2 10 13 10 6 8 3 2 2 Bering Sea and Aleutian Is. 1 2 6 3 1 Total 3 20 23 22 13 12 4 2 2 101 Sebastes melanostictus Lower West Coast 2 1 5 5 1 1 1 Gulf of Alaska 4 9 9 12 17 5 11 2 3 3 Bering Sea and Aleutian Is. 5 11 4 11 6 5 2 1 1 Western Pacific Ocean 1 3 4 1 1 1 Total 9 22 15 31 32 11 15 4 4 4 1 148 Left infraorbital 1 spines 0 1 2 3 4 n Sebastes aleutianus Lower West Coast 1 22 8 Gulf of Alaska 5 28 15 7 Bering Sea and Aleutian Is. 3 8 2 Total 9 58 25 7 99 Sebastes melanostictus Lower West Coast 1 2 9 4 Gulf of Alaska 2 29 29 9 6 Bering Sea and Aleutian Is. 2 24 15 4 Western Pacific Ocean 1 9 1 Total 5 56 62 18 6 147 continued. ral and supracleithral spines present. Ventral margin dorsal margin of opercle nearly horizontal; lower mar- of lachrymal typically with two rounded lobes, both gin of gill cover with 1 or 2 small spines, produced by anterior and posterior often with a single small pos- posteroventral tip of interopercle and anteroventral tip teriorly directed spine, posterior occasionally with as of subopercle. many as 4 or 5 small spines forming a serrate margin; Dorsal-fin origin above anterodorsal portion of gill infraorbital spines small, 2- 10 (9) 0-3 (3) on first slit; dorsal fin continuous, gradually increasing in infraorbital, 0-5 (5) on second, and 1-3 (1) on third; height from spine I (4. 3-7. 2% SL, 1. 5-3.0 times into Orr and Hawkins: Species of the rougheye rockfish complex 123 Table 5 (continued) Left infraorbital 2 spines 0 1 2 3 4 5 6 n Sebastes aleutianus Lower West Coast 11 15 4 1 Gulf of Alaska 1 14 25 10 2 3 Bering Sea and Aleutian Is. 1 9 2 1 Total 1 26 49 16 3 4 99 Sebastes melanostictus Lower West Coast 1 5 6 2 1 1 Gulf of Alaska 4 14 31 13 9 3 1 Bering Sea and Aleutian Is. 3 16 20 4 1 1 Western Pacific Ocean 1 7 2 1 Total 8 36 64 21 12 3 3 147 Left infraorbital 3 spines 0 1 2 3 4 5 6 7 n Sebastes aleutianus Lower West Coast 18 12 1 Gulf of Alaska 26 25 4 Bering Sea and Aleutian Is. 2 10 1 Total 46 47 6 99 Sebastes melanostictus Lower West Coast 7 8 1 Gulf of Alaska 3 27 38 6 1 Bering Sea and Aleutian Is. 1 18 22 2 1 1 Western Pacific Ocean 3 7 1 Total 4 55 75 10 2 0 0 1 147 orbit length; Fig. 5A) to spine IV (9.8-15.5% SL; Fig. 5B; tips of dorsal-fin spines broken in lectotype) and decreasing in height to spine XII; spine XIII much lon- ger, forming anterior support of dorsal-fin rays; mem- branes of spinous dorsal fin moderately incised, less so posteriorly; soft dorsal fin with anterior rays longest, posterior rays gradually shortening. Anal-fin spine II shorter than or equal to III (11.0-18.4 vs. 13.0-20.2% SL; tip of anal-fin spine II broken in lectotype), rela- tively longer in juveniles, soft rayed portion of anal fin with anterior rays longest, posterior rays gradually shortening, posterior margin perpendicular to body axis or with slight posterior slant, anterior ray tips directly ventral to or forward of posterior tips, anterior tip of anal fin slightly rounded. Pectoral fins with ray 10 or 11 longest, extending to or slightly anterior to vent, fin-ray length 24.4-31.0 (25.5) % SL; fin-base width 8.4-10.4 (9.1) % SL; dorsalmost ray simple, ventral 5-10 simple, others branched in adults; all rays simple in three juveniles less than 90 mm (63.4-77.1 mm), the smallest examined. Pelvic fins extending about 50-90% of distance from pelvic-fin base to anal-fin origin, fall- ing well short of vent, ray length 19.3-25.1 (22.8) % SL, spine length 51.7-76.0 (59.2) % ray length. All fin Table 6 Factor loadings for principal component (PC) analysis of meristic characters of Sebastes aleutianus (rougheye rockfish) and Sebastes melanostictus (blackspotted rock- fish) in which all characters were used for multivariate analyses. PCI PC2 PC3 Dorsal-fin rays 0.5620 0.2917 -0.0630 Anal-fin rays 0.2956 0.5901 -0.5291 Pectoral-fin rays (left) 0.1390 -0.6710 -0.5828 Lateral-line pores (left) 0.4905 -0.1233 0.6118 Gill rakers 0.5804 -0.3181 -0.0470 spines exhibit allometric growth, and juveniles have relatively longer spines. Caudal fin shallowly emargin- ate, length 20.0-28.1% SL (fin damaged in lectotype). Vent positioned below dorsal-fin spine 10-11, 7-21 (9.4) % HL from anal-fin origin. Lateral body scales often with many (ca. 5-7) acces- sory scales in posterior field. Maxilla and underside of 124 Fishery Bulletin 106(2) 140°E 150°E 160°E 170'E 180" 170"W 160"W 150”W 140°W 130’W 120"W Distribution of Sebastes aleutianus (rougheye rockfish, open circle) and Sebastes mela- nostictus (blackspotted rockfish, closed circle) based on material examined. Closed circles superimposed on open circles represent collections of both species at the same locality. Each symbol may represent more than one specimen. mandible completely scaled; suborbital region scaled; branchiostegal rays scaled. Gill rakers relatively short, 3. 4-6. 3 (4.5) % SL (Fig. 5C), and slender on first arch; longest raker in joint between epi- and ceratobranchials; length of preceding rakers on upper arch and succeeding rakers on lower arch progressively shorter; rudiments absent. Extrinsic swimbladder muscle is Type I a-z of Hallacher (1974). Body color in life pink, red, or reddish orange and darker brownish-red mottling is present in faint bands at and above the lateral line, often extending onto dor- sal-fin base. Head light, with irregular shaped dark blotch at posterodorsal corner, other blotches often pres- ent on operculum between orbit and lower posterior margin. Orobuccal membranes pink to red, often with dark blotches; jaw membranes light, occasionally dusky. Spinous and soft dorsal fins uniformly pink to red, usu- ally dark along fin margins, occasionally with small diffuse blotches near base of fin. Anal and caudal fins uniformly pink to red, dark along fin margin. Paired fins red, rays often with dark tips. Peritoneum gray to dusky, rarely white or black; stomach, pyloric caeca, and intestines pale. See Figures 1 and 2 and color figures in S. aleutianus species accounts of Love (2002; “juvenile” upper left), Kramer and O’Connell (1986, 1988, 1995, 2003; “juvenile”). Juveniles in life similar to adults in general body color, often with more distinct dark red to brown mottling. After preservation, reddish background color fading to light gray, yellowing with age. Dark areas remaining dark brown to black. No sexual dimorphism is evident in morphometric or meristic characters. Largest specimen examined 555 mm (726 mm total length [TL], 710 mm fork length [FL]; FAKU 119293). Distribution The range of Sebastes aleutianus, based on material examined, extends from the eastern Aleutian Islands off Unalaska Island and the eastern Bering Sea at Pribilof Canyon at 55.7°N, south to southern Oregon at 43.9°N (Fig. 7). This distribution is nearly identical to that reported in the analyses of Hawkins et al. (2005, as S. aleutianus) and Gharrett et al. (2005, 2006, as “Type II”). Our material was collected at depths of 45 m to at least 439 m — a range that overlapped the depth distri- bution of S. melanostictus but which was typically shal- lower than the depth range of 0.05; pm: F14=0.06, P>0.05) or by treatment (ANO- VA: am: F14=2.61, P>0.05; pm: F14=5.13, P>0.05). Fish were maintained for a total of nine days in each experi- ment (i.e., nine days per year) after which three fish Kingsford et al.: The influence of elemental chemistry on the widths of otolith increments in Pomacentrus coelestis 137 from each tank were randomly selected and frozen (-20°C) for microstructural analysis of otoliths. The SL of the P. coelestis used in the experiment (mean=15.2 mm ±0.07 SE) did not differ between treatment groups for any year of the experiment (ANOVA: 2000: P4> 16=3.31, P>0.05; 2001: F1 16=1.43, P>0.05; 2002: Fx 16 = 2.4, P>0.05) or among years (ANOVA: F2 51=0.24, P>0.05). The otoliths (the left or right sagittae was chosen randomly), were rinsed three times in Milli-Q water (Millipore, Billerica, MA). and were allowed to dry. Otoliths were polished as described above (see Wild fish section. Following elemental analysis of the otoliths by means of la- ser ablation inductively coupled mass spectrometry (for a complete description of the elemental analysis see Patterson et al., 2004a), the daily increments were counted and measured in the same manner as described earlier. Note that all increments were counted and measured, not just those pertaining to the experiment. Settlement marks, based on the criteria defined by Wilson and McCormick (1999), were also noted. Crystallography The orientation of crystals was examined with SEM near the sulcus of sectioned sagittae. The surface of sectioned otoliths was lightly etched (EDTA 0.1 to 0.01M for 3 minutes) to observe the orientation of the crystal-lattice. Although etching altered the edge of crystals, orientation of the long-axis of the crystal-lattice was still easily seen (e.g., Linkowski et al., 1993). The crystal-lattice was observed for four fish from each treat- ment (ocean vs. lagoon) for the year 2000 experiment. The crystal-lattice of experimental fish was compared with that of 18 control fish. These fish were collected in the same light traps as fish used in the experiments, but were not subjected to any experimental conditions. Statistical analysis Data were tested for homogeneity of variances with a Cochran’s C-test and were found to be homogeneous. Wild fish otolith increments were analyzed with a nested ANOVA design with the factors habitat (ocean vs. lagoon) and site nested within habitat. However, because site nested within habitat was found to be nonsignificant at the P = 0.25 level, sites were pooled with the residual to increase degrees of freedom and the ANOVA was undertaken again with only habitat as a factor (Under- wood, 1997). To analyze otolith increments corresponding to both the pre-experimental (nine days before the start of the experiment) and the experimental phase (nine days of the experiment), fully nested ANOVAs were used for each year of the experiment. This design used two factors (treatment and tank nested within treatment). Treatment was a fixed factor and tank was a random factor. One-factor ANOVAs and Tukey’s HSD post hoc tests were used to examine differences among years A Sites B Figure 1 Mean ( + 1 standard error) increment width (pm) in oto- liths of nonexperimental neon damselfish ( Pomacentrus coelestis) (n = 7 per site) collected from two sites on the reef slope in ocean water (Ocean 1 and 2) and within the lagoon (Lagoon 1 and 2) at One Tree Island, Great Bar- rier Reef, Australia. Tukey’s HSD contrast groupings are indicated as letters (i.e. , A, B) where overall differences were significant (P<0.05). within treatment groups. Paired /-tests were also used to examine the relationships of increment widths before and during the experiment for each experimental treat- ment. Based on the findings of Patterson et al. (2004a), correlation analyses were used to examine the relation- ships between incremental widths and elemental ratios (i.e., Ba/Ca, Sr/Ca). Results Neon damselfish in natural habitats displayed signifi- cant variation in increment widths; increments were nar- rower in otoliths of fish collected in the lagoon (ANOVA: P4 26=14.76, P<0.001; Fig. 1). For experimental fish, which all originated in ocean water, no difference in incremental spacing was found before the start of the experiment by treatment or by tank nested within treat- ment for any year of the experiment (ANOVA: 2000: P4 4 = 1.34, P>0. 05; 2001: P4 4 = 0.17, P>0.05; 2002: Fx 4= 0.04, P>0.05; Fig. 2). In addition, no difference in Ba/Ca ratios was found before the experiment (Fig. 3; Patterson et al., 2004a). Fish in ocean water in 2000 and 2002 had wider increments than those in the lagoon treatment groups (ANOVA: 2000: P4 4=28.53, P <0.01; 2001: P, 4 = 1.62, P>0.05; 2002: P, 4=26.42, P<0.01), as well as a significant difference in Ba/Ca ratios (Figs. 2 and 3; Patterson et al., 2004a). In addition, there was a signifi- cant difference in increment widths among tanks within treatments for 2002. A Tukey’s HSD test (P<0.05) indi- cated this was due to a tank of lagoon water where fish 138 Fishery Bulletin 106(2) 6.0 5.8 5.6 5.4 5.2 5.0 ^ elssss Ocean r — i Lagoon 2000 2001 3.5 3.0 2.5 2.0 B i 2001 Year I 2002 Figure 2 Mean (+1 standard error) increment width (pm) in oto- liths of neon damselfish ( Pomacentrus coelestis) (n = 9 per treatment per year) for all years from ocean and lagoon treatments (A) before the start of the experi- ment and (B) during the experiment for each year of the experiment. Tukey’s HSD contrast groupings are indicated as letters (i.e., A, B) where overall differences were significant (P<0.05). Note the difference of the y-axis scale in A and B. 3 0 A Ocean r — -i Lagoon 2.5 ' 2000 2001 2002 A B A B Year Figure 3 Ba/Ca ratios from otoliths of neon damselfish ( Pomacen- trus coelestis) (n = 9 per treatment per year) for all years from ocean and lagoon treatments (A) before the start of the experiment and (B) during the experiment. Tukey’s HSD contrast groupings are indicated as letters (i.e., A, B) where overall differences were significant (P<0.05). Note the difference of the y-axis scale in A and B. had a higher mean increment width than the fish in the other two tanks (2.8 ,um vs. 2.4 pm and 2.4 am), but this did not obscure major differences between treatments. No significant difference in increment widths between treatment groups was detected for 2001. Finally, a sig- nificant difference in increment widths was detected among years for the ocean treatment group, but not for the lagoon treatment group (ANOVA: ocean: F2 24=16.56, P<0.05; lagoon: F2 24=2.28, P>0.05). Fish in the 2002 ocean treatment group had the widest increments of any fish in the experiments. Individual paired /-tests indi- cated that for both ocean and lagoon treatment groups there was a significant difference in increment width between the pre-experimental and the experimental increments (paired /-test: ocean: T24,,=90.45, P <0.05; lagoon: T242= 3 2 2.05, P<0.05). There were significant relationships between mean increment widths and Ba/Ca ratios for two of the three experimental years (2000: Pearson correlation coef- ficient, r-0.644, P<0.01; 2001: r=0.052, P>0.05; 2002: r=0.671, P<0.01; Fig. 4). There was no significant cor- relation between increment widths and Sr/Ca ratios for any of the experimental years (2000: Pearson coef- ficient, r=-0.417, P>0.05; 2001: r=0.033, P> 0.05; 2002: r=0.266, P>0.05; Fig. 4). SEM indicated that the aragonite crystal-lattice with- in experimental otoliths was orientated perpendicularly to daily increments (Fig. 5). This pattern was consistent for ocean and lagoon treatments (angle from daily incre- ments: ocean: 90° ±0°; lagoon: 90° ±0°) and control fish (90° ±0°). Variation in increment width could only be explained by variation in the length of the intra-incre- mental lattice rather than by a change in orientation or crystal packing. Discussion The pattern of wider increments observed for experimen- tal P. coelestis held in ocean waters was consistent with Kingsford et al.: The influence of elemental chemistry on the widths of otolith increments in Pomacentrus coelest/s 139 3.6 3.4 3.2 3.0 2.8 2.6 2.4 2.2 2.0 1.8 * * 2000 Ocean 2000 Lagoon 2001 Ocean 2001 Lagoon 2002 Ocean 2002 Lagoon 2 3 Ba/Ca (pmol/mol) Figure 4 Relationship (mean +1 standard error) between increment width (pm) in otoliths of neon damselfish ( Pomacentrus coelestis ) and elemental ratios for all experimental years for each treat- ment group (n = 9 per treatment per year) for (A) Ba/Ca and (B) Sr/Ca. the pattern observed for wild fish. Additionally, these experiments demonstrated that increment widths could be altered in the absence of food and temperature differences. There are a number of factors that can potentially influence the incre- ment widths of otoliths, such as feeding regime (McCormick and Molony, 1992), lipid reserves (Paperno et al., 1997; Molony and Sheaves, 1998), the onset of metamorphosis (Wilson and McCor- mick, 1999), water temperature (Powell et al., 2004), and metabolic rate (Mosegaard et al., 1988; Wright et al., 2001). Temperature and food, the most well-studied factors, were not relevant here because both were held constant during the experiments. Additionally, metabolic rates were expected to be consistent among individuals, given the similar size, age, and the constant tem- perature and feeding regimes of these individu- als. It is also unlikely that metamorphosis played a role; experimental fish immediately “settled” in the tanks. Settlement marks were noted to form with narrower post-settlement increments, consis- tent with what is known for this species (Wilson and McCormick, 1999). However, these narrower increments were unlikely to have affected the results as they occurred in all fish in the experi- mental treatments over all years. It is possible that the differences noted in in- crement widths were related to initial size or the condition of the fish during the presettlement phase. Although the standard length of the fish was measured after the study and no differences were found, it was not possible to measure fish before the study (see Methods) and more explicit measures of condition (e.g., Fulton’s K\ Hoey and McCormick, 2004) were not applied before the experiment. Therefore, pre-existing differences in size or condition may have influenced the re- sults to some degree (e.g., some fish may have started the experiment with higher lipid reserves which could have influenced their growth rate and increment widths irrespective of experimen- tal conditions). The randomized allocation of fish between treatments on three separate occasions, however, makes this scenario highly unlikely. In addition, any size or conditional differences were more likely to have introduced variation within the treatment groups rather than the between treatment differences observed in this study. Water chemistry may have influenced increment width in an indirect manner. Late-stage reef fish have well-developed olfactory systems and are capable of detecting chemical differences of ocean and lagoon wa- ters found around OTI (Atema et al., 2002). Some taxa prefer lagoon water (e.g., apogonids), whereas it has been shown that P. coelestis actively avoid settling in the lagoon and prefer the coral rubble habitat found on reef slopes (Doherty et al., 1996). Indeed, P. coelestis were relatively rare in the lagoon at OTI, and this finding confirms the previous report of Doherty et al. (1996). Although not well-studied in reef fish (but see Arvedlund et al., 1999 and Gerlach et al., 2007), natal imprinting and homing has been documented in fish species, most notably in the olfactory-driven homing of salmonids (Dittman et al., 1996). Because P. coelestis are likely spawned and undergo their presettlement life outside of lagoon habitats, they may have imprinted on nonlagoon waters. Lagoon water, therefore, may be an indication of substandard habitat that P. coelestis actively avoid. Holding fish in lagoon water may have triggered a physiological response (e.g., stress), result- ing in reduced growth and narrower otolith increment widths for fish held in nonlagoon waters (e.g., Marchand et al., 2003). However, this seems unlikely to have been 140 Fishery Bulletin 106(2) the primary mechanism driving our results. Although it was not possible to measure fish before each experi- ment, there was no difference in standard length be- tween treatment groups for any year, indicating that fish in both groups had similar growth rates. Alternatively, it is possible that lagoon water lacks some physiologically important trace element which limits growth. Again, this seems unlikely to fully ac- count for our results because P. coelestis do inhabit and spawn in the lagoon, albeit not in great num- bers, as do other pomacentrid species. This would not be expected if lagoonal waters were lacking an ele- ment of such physiological importance. Furthermore, Ba is not this limiting element because it has been clearly shown that Ba concentrations in otoliths are indicative of Ba concentrations in water (Bath et al., 2000; Walther and Thorrold, 2006) and that Ba is not influenced by the physiology of the fish (Campana, 1999). Because there was a significant relationship between increment widths and Ba/Ca, it seems likely that another mechanism was involved in producing the results observed. A related study, in which the otolith chemistry of the same otoliths used here was examined, indicated sig- nificant differences in the otolith chemistry of the two treatment groups for 2000 and 2002, primarily driven by differences in Ba/Ca and Sr/Ca (Patterson et al., 2004a). Similar to the results of the present study, no difference in otolith chemistry between the treatment groups was detected for 2001. The differences in oto- lith chemistry between the treatment groups probably reflected differences in the composition of the water masses (Bath et al., 2000). Oceanographic features that may affect water chemistry (i.e. , upwelling, phy- toplankton blooms; Patterson et al., 2004a, 2004b) are not temporally stable. This may explain why signifi- cant differences in otolith chemistry and microstructure were found only for fish from the 2000 and 2002 experiments. In addition, there was a significant relation- ship between Ba/Ca ratios and increment widths, and annual variation in increment widths reflected differences for Ba. No sig- nificant correlation was found between Sr/ Ca ratios and incremental widths, which concurs with previous studies (Kalish, 1989; Gallahar and Kingsford, 1992). The mechanisms by which trace ele- ments are deposited into otoliths are not entirely understood at this time. There- fore, the exact mechanisms by which oto- lith chemistry may directly influence increment widths remain unknown. Oto- liths are composed primarily of calcium carbonate in the form of aragonite. Di- valent metal ions of similar size to Ca, such as those of Ba and Sr, can substi- tute directly for Ca in the crystal-lat- tice (Campana, 1999). However, other elements such as Zn and Mn may occupy different positions in the otolith matrix (i.e., in the interstitial spaces or in as- sociation with the proteinaceous matrix; Campana, 1999). Although it is known that otolith crystal orientation is pro- tein-mediated (Lowenstam, 1981), it is not known how physiological stress can affect protein deposition, or how trace el- ement inclusion may influence increment width. The orientation of the crystal-lat- tice did not change in fish treated with ocean and lagoon waters and crystal orientation was similar to other fishes (e.g., Linkowski et al., 1993). Elemental availability in the location of the trace metal inclusions may be capable of cre- ating differences in increment width by increasing the length, or number, of Figure 5 Scanning electron microscope images of the crystal lattice in otoliths of neon damselfish (Pomacentrus coelestis). (A) Image of an otolith of an experimental (Expt) fish. “Expt” indicates the area (above the white line) where increments were laid down in the otolith of a fish exposed to ocean waters. (B) Detail of crystal-lattice from a control fish; (C) Detail of the crytalline lattice of a postsettlement control fish; the white line indicates the settlement mark. Large white arrows indicate the orientation of the crystal-lattice. Scale: increments near the edge of the otolith are separated by approximately 1.5 pm. Kingsford et al.: The influence of elemental chemistry on the widths of otolith increments in Pomacentrus coelestis 141 crystals that are perpendicular to incremental zones, or by altering the way in which crystals are bundled (tightly packed vs. chaotic bundles). The SEM used in this study did not have the resolution to distinguish between these alternatives. In summary, this study has provided convincing evidence that water chemistry can affect the incre- ment widths of reef fish otoliths by either direct or indirect mechanisms. It is also possible that multiple mechanisms have interacted to produce the results observed. Although the results presented here are pre- liminary, they indicate that interpreting increment widths may be more complex than has previously been noted and, at least under some circumstances, differ- ences in water chemistry may confound patterns pre- viously attributed only to growth and condition. Fur- ther research on this topic is therefore warranted and should include controlled experiments with temporally comparable measurements of water chemistry, otolith chemistry, and increment widths to tease apart the complex mechanisms at work. Lagoons are also found on other reefs (Atema et al., 2002), and further studies using other water masses are needed to determine if the results presented here are prevalent in coral reef fish or are specific to the system at One Tree Island. Finally, continuing research on natal imprinting may elucidate to what extent reef fish imprint on specific water masses, and the physiological consequences of that imprinting. Acknowledgments We thank J. Hughes, Mark O’Callaghan, and Kevin Blake for assistance with SEM and J. Eagle and J. Brown for their assistance in the field. Comments by J. Ackerman and B. Curley improved the manuscript. An Australian Research Council grant to M. J. Kingsford and a Doctoral Merit Research Scholarship from James Cook University, Cooperative Research Centre Reef Research Centre Grant, and Great Barrier Reef Marine Park Authority Augmentative Grant to H. M. Patterson supported this project. This study is a contribution from One Tree Island Research Station. Literature cited Arvedlund, M., M. I. McCormick, D. G. Fautin, and M. Bildsoe. 1999. Host recognition and possible imprinting in the anemonefish Amphiprion melanopus (Pisces : Pomacentridae). Mar. Ecol. Prog. Ser. 188:207-218. Atema, J., M. J. Kingsford, and G. Gerlach. 2002. Larval reef fish could use odour for detection, retention and orientation to reefs. Mar. Ecol. Prog. Ser. 241:151-160. Bath, 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 aragonite oto- liths of marine fish. Geochim. Cosmochim. Acta 64:1705-1714. Bergenius, M. A. J., M. G. Meekan, D. R. Robertson, and M. I. McCormick. 2002. Larval growth predicts the recruitment success of a coral reef fish. Oecologia 131:521-525. Campana, S. E. 1999. Chemistry and composition of fish otoliths: path- ways, mechanisms and applications. Mar. Ecol. Prog. Ser. 188:263-297. Dittman, A. H., T. P. Quinn, and G. A. Nevitt. 1996. Timing of imprinting to natural and artificial odors by coho salmon ( Oncorhynchus kisutch). Can. J. Fish. Aquat. Sci. 53:434-442. Doherty, P. J., M. J. Kingsford, D. Booth, and J. Carleton. 1996. Habitat selection before settlement by Pomacentrus coelestis. Mar. Freshw. Res. 47:391-399. Flood, M. J. 2000. Otolith microstructure in pre- and post-settlement Pomacentrus coelestis (Pisces: Pomacentridae). Honours thesis, 112 p. Univ. Sydney, Sydney, Australia. Gallahar, N. K., and M. J. Kingsford. 1992. Patterns of increment width and strontium:cal- cium ratios in otoliths of juvenile rock blackfish, Girella elevata. J. Fish Biol. 41:749-763. Gerlach, G., J. Atema, M. J. Kingsford, K. P. Black, and V. Miller-Sims. 2007. Smelling home can prevent dispersal of reef fish larvae. Proc. Natl. Acad. Sci. 104:858-863. Hoey, A. S., and M. I. McCormick. 2004. Selective predation for low body condition at the larval-juvenile transition of a coral reef fish. Oecologia 139:23-29. Hussy, K., H. Mosegaard, H. H. Hinrichsen, and U. Bottcher. 2003. Factors determining variations in otolith micro- increment width of demersal juvenile Baltic cod Gadus morhua. Mar. Ecol. Prog. Ser. 258:243-251. Kalish, J. M. 1989. Otolith microchemistry: validation of the ef- fects of physiology, age and environment on otolith composition. J. Exp. Mar. Biol. Ecol. 132:151-178. Kingsford, M. J., J. M. Leis, A. Shanks, K. C. Lindeman, S. G. Morgan, and J. Pineda. 2002. Sensory environments, larval abilities and local self-recruitment. Bull. Mar. Sci. 70:309-340. Linkowski, T. B., R. L. Radtke, and P. H. Lenz. 1993. Otolith microstructure, age and growth of two species of Ceratoscopelus (Osteichthyes: Myctophidae) from the eastern North Atlantic. J. Exp. Mar. Biol. Ecol. 167:237-260. Lowenstam, H. A. 1981. Minerals formed by organisms. Science 211:1126- 1131. Ludington, C. A. 1979. Tidal modifications and associated circulation in a platform reef lagoon. Aust. J. Mar. Freshw. Res. 30:425-430. Marchand, J., A. Tanguy, J. Laroche, L. Quiniou, and D. Moraga. 2003. Responses of European flounder Platichthys flesus populations to contamination in different estuaries along the Atlantic coast of France. Mar. Ecol. Prog. Ser. 260:273-284. McCormick, M. I. 1998. Condition and growth of reef fish at settlement: Is it important? Aust. J. Ecol. 23:258-264. McCormick, M. I., and B. W. Molony. 1992. Effects of feeding history on the growth charac- 142 Fishery Bulletin 106(2) teristics of a reef fish at settlement. Mar. Biol. 11: 165-175. Meekan, M. G., J. H. Carleton, A. D. McKinnon, K. Flynn., and M. Furnas. 2003. What determines the growth of tropical reef fish larvae in the plankton: Food or temperature? Mar. Ecol. Prog. Ser. 256:193-204. Moksness, E. 1992. Differences in otolith microstructure and body growth rate of North Sea herring ( Clupea harengus L.) larvae in the period 1987-1989. ICES J. Mar. Sci. 49:223-230. Molony, B. W. 1996. Episodes of starvation are recorded in the otoliths of juvenile Ambassis vachelli (Chandidae), a tropical estuarine fish. Mar. Biol. 125: 439-446. Molony, B. W., and M. J. Sheaves. 1998. Otolith increment widths and lipid contents dur- ing starvation and recovery in adult Ambassis vach- elli (Richardson). J. Exp. Mar. Biol. Ecol. 221:257- 267. Mosegaard, H., H. Svedang, and K. Taberman. 1988. Uncoupling of somatic and otolith growth rates in Arctic char ( Salvelinus alpinus) as an effect of differ- ences in temperature response. Can. J. Fish. Aquat. Sci. 56:231-241. Paperno, R., T. E. Targett, and P. A. Grecay. 1997. Daily growth increments in otoliths of juvenile weakfish, Cynoscion regalis: experimental assessment of changes in increment width with changes in feeding rate, growth rate, and condition factor. Fish. Bull. 95:521-529. Patterson, H. M., M. J. Kingsford, and M. T. McCulloch. 2004a. The influence of oceanic and lagoonal plume waters on otolith chemistry. Can. J. Fish. Aquat. Sci. 61:898-904. 2004b. Elemental signatures of Pomacenirus coelestis otoliths at multiple spatial scales on the Great Barrier Reef, Australia. Mar. Ecol. Prog. Ser. 270:229-239. Powell, A. B., R. T. Cheshire, E. H. Laban, J. Colvocoresses, P. O’Donnell, and M. Davidian. 2004. Growth, mortality, and hatch date distributions of larval and juvenile spotted seatrout (Cynoscion nebulo- sus) in Florida Bay, Everglades National Park. Fish. Bull. 102:142-155. Sponaugle, S., and D. R. Pinkard. 2004. Impact of variable pelagic environments on natural larval growth and recruitment of the reef fish Thalas- soma bifasciatum. J. Fish Biol. 64:34-54. Underwood, A. J. 1997. Experiments in ecology: their logical design and interpretation using analysis of variance, 522 p. Cam- bridge Univ. Press, Cambridge, UK. Walther, B. D., and S. R. Thorrold. 2006. Water, not food, contributes the majority of stron- tium and barium deposited in the otoliths of a marine fish. Mar. Ecol. Prog. Ser. 311:125-130. Wilson, D. T., and M. I. McCormick. 1999. Microstructure of settlement-marks in the otoliths of tropical reef fishes. Mar. Biol. 134:29-41. Wright, P. J., P. Fallon- Cousins, and J. D. Armstrong. 2001. The relationship between otolith accretion and rest- ing metabolic rate in juvenile Atlantic salmon during a change in temperature. J. Fish Biol. 59:657-666. 143 Seasonal, diel, and lunar spawning periodicities and associated sound production of white seahass ( Atractoscion nohiBis ) Scott A. Aalbers Email address: scott@pier.org Hubbs-SeaWorld Research Institute 2595 Ingraham Street San Diego, California 92109 and California State University, Fullerton 800 North State College Blvd. Fullerton, California 92634-9480 Present address: Pfleger Institute of Environmental Research 315 N Clementine Oceanside, California 92054 Abstract — Spawning periodicities of white seabass (Atractoscion nobilis) were evaluated by observing spawning behavior, by collecting eggs, and moni- toring recognizable sounds produced during the release of gametes. A total of 297 spawning events were docu- mented from 15 male and 47 female white seabass contained within the seminatural confines of a 526-m3 net pen located in Catalina Harbor, Santa Catalina Island, California. Consis- tent spawning occurred from March through July 2001-03, and peaked in May at a photoperiod of 14 hours. Most spawning occurred within the 2-hour period following sunset or from 19:00-20:00 hours Pacific Standard Time. White seabass spawned at every phase of the lunar cycle; but an increase in successive spawning events followed the new moon. Most spawning occurred in water tempera- tures from 15 to 18°C, and there was no apparent correlation with tidal cycles. Seasonal and diel spawning periods were directly correlated with increases in the rate, intensity, and variety of white seabass sounds; this correlation may indicate that sounds function to enhance reproductive suc- cess. These findings can be extended to further develop seasonal fishery regulations and to better comprehend the role of sound in the reproduction of sound-producing fishes. Manuscript submitted 13 September 2007. Manuscript accepted 10 January 2008. Fish. Bull 106:143-151 (2008). The views and opinions expressed or implied in this article are those of the author and do not necessarily reflect the position of the National Marine Fisheries Service, NOAA. The white seabass (Atractoscion nobi- lis), also known as white weakfish (Eschmeyer et ah, 1983), is an ovipa- rous group spawner; multiple males fertilize the eggs of a gravid female as gametes are broadcast into the water column (Aalbers and Drawbridge, in press). As the largest member of the family Sciaenidae (croakers and drums) that inhabits California coastal waters (Miller and Lea, 1972), white seabass produce relatively large eggs, averaging 1.27 mm in diameter (Moser et ah, 1983). Fertilized eggs are buoyant and drift with the ocean surface currents for two days before hatching into planktonic larvae that disperse for approximately 30 days and settle out nearshore at a size of 7-10 mm (Moser et ah, 1983; Allen and Franklin, 1992). A previous study, in which the go- nads of commercially caught white seabass were examined, indicated that spawning occurs from April through August and peaks in May and June (Skogsberg, 1925). The ma- jority of white seabass have tradition- ally been landed in the late spring and early summer, when spawning aggregations develop nearshore and around coastal islands (Skogsberg, 1939; Thomas, 1968). Following a sharp decline in California landings, the commercial harvest of white sea- bass was restricted from March 15 through June 15 to protect spawning stocks, while recreational bag lim- its were reduced from three to one fish during this three-month period. Although current restrictions offer white seabass spawning aggregations some refuge, a better understanding of spawning periods will allow man- agers to further develop seasonal fish- ery regulations. Reproduction has been coupled with sound production in other spe- cies of Sciaenidae, including the red drum ( Sciaenops ocellata) (Guest and Lasswell, 1978), spotted seatrout (Cy- noscion nebulosis) (Mok and Gilmore, 1983), weakfish (Cynoscion regalis ) (Connaughton and Taylor, 1995), and orangemouth corvina (Cynoscion xan- thulus) (Fish and Cummings, 1972). Typical of most sciaenids, male white seabass possess sonic musculature that resonates pulsed sounds from the adjacent gas bladder (Tavolga, 1964). Although the sonic muscula- ture and gas bladder structures vary considerably among the 270 sciaenid species (Chao, 1986; Ramcharitar et ah, 2006), the sonic structures that Smith (1905) and Tower (1908) de- scribed a century ago in male weak- fish closely resemble those of white seabass. The importance of sound to this diverse family has been well documented during periods of repro- ductive activity (Ramcharitar et ah, 2006); however, recognizable sounds during the actual release of gametes 144 Fishery Bulletin 106(2) have only been described in the white seabass (Aalbers and Drawbridge, in press). Acoustically monitoring the sounds produced by spawning fish can reveal the periodicity and critical habitat of spawning aggregations and help managers in the development of fishery regulations and in the better placement of marine protected areas (Luczkov- ich et ah, 1999). However, patterns of spawning-related sound production must be established in order to yield meaningful results for such application. The objective of this investigation was to determine the spawning periodicities and associated sound production patterns of white seabass under seminatural environmental conditions. This study presents a comprehensive as- sessment of sciaenid sound production as it relates to known spawning activity and establishes the foun- dation from which passive acoustic surveys can be conducted. Materials and methods Sampling locations and procedures Research was conducted from an aquaculture platform consisting of four 9x9x6. 5-m net pens moored in Cata- lina Harbor, Santa Catalina Island, CA. Forty-one adult white seabass were transported live to an individual net pen (526 m3) from March through June 2000. An additional 21 fish, collected in May 2001, were excluded from the 2001 spawning data while being acclimated in a separate net pen. In October 2001, all fish were mea- sured to the nearest mm, weighed to the nearest 10 g, examined to determine sex, tagged with a passive inte- grated transponder (PIT; Avid, Greeley, CO), and com- bined into a single net pen. Forty-seven female and 15 male fish ranged in initial size from 83 to 126 cm total length (TL) and 4.9 to 19.5 kg in weight. Fish were only manipulated when remeasured in February 2003 and no mortality occurred during the study. Captive white seabass were fed variable rations of freshly caught squid (Loligo opalescens), chub mackerel (Scomber japonicus), and jack mackerel (Trachurus symmetricus ) as they became available locally. Documentation of spawning Fish spawning periodicity was assessed from 2001 to 2003 by direct observation from an elevated deck, by sampling the net pen for freshly spawned eggs with a 500-pm-mesh dip net, and by acoustically monitoring fish sounds. Successive spawning events were difficult to differentiate from egg sampling alone; therefore mul- tiple spawning events were documented only if visually observed or acoustically detected and accompanied by increased egg abundance in dip-net samples in compari- son to previous samples. Gravid females and courting males were occasionally scanned for individual identi- fication with a PIT (passive intergrated transponder) reader when fish approached the surface. Analyses of spawning data Date and time of day, photoperiod, lunar phase, and tidal cycle were noted upon detection of a spawning event. Data were grouped by month, hour, and phase of the synodic month to determine seasonal, diel, and lunar spawning periodicities. Hourly spawning periodic- ity was analyzed in relation to sunset and to the 24-h clock after correcting to Pacific Standard Time (PST), without adjusting for daylight savings time. Water temperature was recorded to the nearest 0.15°C every hour at a depth of 3 m with an Optic Stowaway tem- perature logger (Onset Computer Corp., Pocasset, MA) suspended from the net pen. Continuous temperature data for the 2002 spawning season were obtained from a temperature logger at an adjacent net-pen site (300 m away) within Catalina Harbor. Additional temperature and salinity measurements were made upon detection of eggs with a YSI 85 instrument (YSI Inc., Yellow Springs, OH). Acoustic equipment Sounds were received with an omnidirectional hydro- phone (International Transducer Corporation 6050-C, Santa Barbara, CA) with an essentially flat frequency response between 50 Hz and 35 kHz. The hydrophone was powered by a custom 24-V power supply and ampli- fier with an incorporated reference tone (Frank Aubrey, Hubbs-SeaWorld Research Institute, San Diego, CA). Sounds were recorded with a PCM-M1 digital audio tape recorder (Sony Corp., Tokyo, Japan) set to a con- stant gain. Audio recordings Three-minute audio recordings were made at regular intervals with the hydrophone suspended at a depth of 3 m in the center of the net pen. Evening sound recordings, taken on 17 sampling dates from March through July 2003, were focused around periods of spawning activity, and individual recordings were made every 30 minutes from 1 hour before to 3 hours after sunset. Additional recordings were made every hour for 24 hours, once a month, on randomly selected dates from April 2003 through March 2004 to document seasonal and diel sound patterns. No recordings occurred in February 2004 because of extreme weather conditions. Acoustic analyses Sound recordings (n = 404) were digitally transferred to a PC hard drive for spectral analysis (SpectraPlus, vers. 2.32.04, Pioneer Hill, Campbell, CA). Based on spectral characteristics, the following categories of white seabass sounds were noted: single pulse-trains, multiple pulse- trains, drum-rolls, thuds, abbreviated chants, spawning chants, or hydrodynamic booms as described in Aalbers and Drawbridge (in press). Sound production rates for each sound type were determined from each recording. Aalbers: Spawning periodicities of Atractosaon nobihs 145 35% 0% February March April May June July August Figure 1 Monthly frequency distribution and photoperiods (♦) of spawning events for 297 white seabass (Atractoscion nobilis) spawning events from 2001 through 2003. The most intense sounds, as recorded, of each type were analyzed against a 1-kHz reference tone on each 3-min recording. Sound inten- sities were scaled with values consistent to one another, but not as absolute sound pres- sure level (dB re IpPa) because the entire hydrophone system had not been calibrated recently. The terms “level” and “intensity” are used synonymously to convey the relative data presented here. Statistical analyses A one-way analysis of variance (ANOVA) was used to examine whether diel spawning times were consistent throughout the spawning season both in relation to sunset and the 24-h clock. The numbers of spawning events observed at each phase of the lunar cycle were analyzed against expected occurrences by using a chi-square test. Sound rates from 24-h recordings during the spawning season (March- July) were compared to those when there was no spawning (August-January) using a two- sample £-test. Two-sample f-tests were also used to compare the rates and relative levels of sounds from evening recordings made on days that spawning did and did not occur. Results sunset. The inverse seasonal trend was observed when spawning times were analyzed in relation to sunrise. No equivalent seasonal shift in the time of spawning occurred when events were analyzed in relation to the 24-hour clock (ANOVA, F=2.41, df=4, P= 0.06). Spawn- ing peaked between 19:00 and 20:00 h PST in all three seasons and was not confirmed before 17:30 or after 22:00 h (Fig. 2B). Seasonal spawning Over the three-year study, 297 spawning events were documented; 53, 146, and 98 events were recognized in 2001, 2002, and 2003, respectively. The longest spawn- ing season extended between 22 February and 15 August 2002, whereas the other spawning seasons ranged from 17 March to 27 June 2001 and from 3 March until 16 July 2003. Of the documented spawning events, 76% occurred from April through June, and 98% from March through July over the three-year period. Spawn- ing peaked in May as photoperiod increased to 14 hours (Fig. 1). Diel spawning Spawning occurred from two hours before sunset until 4.5 hours after sunset. Sixty-five percent of spawning activity occurred over the two-hour period following sunset and 91% occurred within the four-hour period spanning from one hour before sunset to 3 hours after sunset (Fig. 2A). A significant shift in the diel spawn- ing periodicity in relation to sunset as the spawning season progressed occurred in all three years (ANOVA, F= 8.78, df=4, P>0.001). For instance, in March 2002 the average spawning time was between two and three hours after sunset; however, by July 2002 the average spawning event occurred within the first hour after Lunar spawning Spawning occurred at every phase of the lunar cycle, but not uniformly (Fig. 3, %2 = 49.57, df=29, P=0.01). The greatest amount of spawning occurred from the new moon until four days after the new moon; 63% of documented mass spawning events (dates with at least four successive spawning events) took place during this five-day period. Successive spawning events were common and as many as eight events occurred in a single night, pre- sumably from different females. The time between suc- cessive spawning on 21 sampling dates ranged from 1 to 98 minutes (mean = 31 min, n = 66). A female individual was identified with the PIT reader before spawning on 12 June 2002 and again on 16 June 2002. Environmental factors Ambient water temperatures ranged seasonally from 11.3° to 21.7°C over the study period. Spawning occurred between 12.5° and 20.8°C, and bimodal peaks in the percent occurrence of spawning were apparent at 15° and 18°C. Seventy-six percent of spawning occurred in water temperatures from 15° to 18°C. Salinity at the study site ranged from 34.2° to 34.8 ppt. Spawn- ing occurred throughout the range of tidal cycles with no apparent tendency towards ebb or flow. The ratio of 146 Fishery Bulletin 106(2) spawning that occurred around high tide was equiva- lent to the ratio of high tides that occurred within the fundamental diel spawning period (1 hour before to 3 hours after sunset). Sound production The sound production rate of white seabass peaked in May, averaged 29.2 sounds/min, and reached a minimum average of 1.5 sounds/min (primarily hydrodynamic booms) in December (Fig. 4). Sound production rates were significantly greater on 24-h recordings made during the spawning season (mean=12.7 sounds/min) than during months without spawning (mean = 4.5 sounds/min) (/=4.88, df=118, P>0.001). Within the spawning season, there was a significant increase in the rate of sound production on spawning than on non- spawning days (/= 9.23, df=203, P>0.001), and peak rates occurred during spawning events (Fig. 5). Similarly, a significant increase in sound intensity was observed for sounds recorded on spawning days than on nonspawning days (f=1.89, df=438, P=0.030). The basic sound types that have been identified for the white seabass are single pulse-trains, multiple pulse-trains, drum- rolls, thuds, and hydrodynamic booms (Fig. 6). A rapid succession of overlapping drum-roll and thud sounds resulted in recognizable spawning chants during the release of gametes. All sound types oc- curred on days of spawning; however, only hydrodynamic booms along with single and multiple pulse-trains of reduced rate and intensity occurred on days with no spawning (Fig. 7, A and B). Single pulse-trains increased in rate and intensity on audio recordings that corresponded with spawning. The rate and intensity of pulse-trains remained high between successive spawning events and throughout the 1-h period after spawning. Multiple pulse-trains occurred as a se- ries of two to six pulse-trains, typically in a rhythmic pattern every 4-6 seconds. Weak multiple pulse-trains were detected throughout the day during the spawn- ing season, but sound intensity greatly increased during the 90-min period after spawning events (Fig. 7B). Post-spawn- ing multiple pulse-trains were of higher intensity than any other type of white seabass sounds. Drum-rolls and thuds increased in rate and intensity during the initiation of spawning behavior and were not detect- ed outside of the 4-h period surrounding spawning. Drum-rolls and thuds occurred in repetitive trains and were categorized as abbreviated chants or spawning chants when overlapping sounds became audibly indistinguishable. Abbreviated chants consisted of rap- id sequences of drum-rolls and thuds that terminated abruptly after 3-5 sec of spawning behavior and there was no release of gametes. Abbreviated chants were shorter in duration and lower in in- tensity than spawning chants and were 35% ■ 30% 25% 20% ■ 1 5% ■ 10% 0=174 5% 0% -2 -1 Sunset +1 +2 +3 Hours in relation to sunset +4 +5 45% i g 40% - 35% - n=155 30% - 25% - 20% - 15% - 10% - 5% - 0% 17:00 18:00 19:00 20:00 21:00 22:00 Pacific standard time (h) 23:00 Figure 2 The percent occurrence of white seabass (Atractoscion nobilis) spawning events illustrating diel spawning periodicities (A) in relation to sunset and (B) according to Pacific Standard Time for 2001-03. Sample sizes correspond to the number of spawning events that could be assigned to an hourly interval. (A) Dashed line depicts the time of sunset and shading represents diminishing light levels throughout the evening. There is no shading in (B) because of seasonal changes in the time of civil twilight. Aalbers: Spawning periodicities of Atractosaon nob/lis 147 most commonly detected within 30 minutes before a spawning or in between successive spawning events. Spawning chants were detected only during the re- lease of gametes and consisted of an elongate series of intense, overlapping drum-roll and thud sounds at an average rate of 7 sounds/sec that continued for up to 55 seconds. The intensity of spawning chants was second only to multiple pulse-trains recorded after spawning. Hydrodynamic booms were caused by sudden pressure waves generated underwater by rapid body movement and occurred throughout the day and night over the entire year. Their average rate and intensity remained relatively constant throughout the day and throughout the year, but increased considerably during feeding or in the presence of a predator around the net pens. Discussion Seasonal spawning White seabass maintain a protracted 5-month spawning season from March through July, and 98% of spawning events are documented during this period. Allen and Franklin (1992) identified a peak in the catch per unit of effort of newly settled young-of-the-year white seabass off Southern California in July 1988 and June 1989, supporting the seasonal spawning data presented here. A March-July spawning season is a month earlier than the previously estimated April-August spawning season for white seabass (Skogsberg, 1939). Annual California Cooperative Oceanic Fisheries Investigation (CalCOFI) surveys revealed that white seabass larvae were most abundant off central Baja California from May through August and abundance peaked in July (Moser et al., 1983), which is later than the May peak in spawning described here. Lati- tudinal differences in photoperiod and water temperature throughout the southern extent of the population range may explain this dis- crepancy. One-third of white seabass spawning activity occurred outside the current March 15 -June 15 restricted commercial fishing period and time of reduced catch allowable under recreational bag limits. Under present regulations, spawn- ing aggregations off California are exposed to heightened recreational and commercial fishing effort for approximately 60 days each season. Delaying the opening of commercial fishing season in addition to maintaining a one-fish recreational bag limit throughout the spawn- ing season would considerably benefit white seabass spawning stocks. Diel spawning A dusk and nighttime spawning period, as documented here for white seabass is common in other temperate sciaenids (Holt et al., 1985; 3% 3% 3% Figure 3 Lunar spawning periodicity of white seabass iAtrac- toscion nobilis) for 297 spawning events documented from 2001 through 2003. Values correspond to the % occurrence of spawning events on each day of a synodic month. Increased shading represents the lunar cycle as it moves from full (light) to new (dark). Spawning activity was not uniform over the lunar cycle (x2 = 49.57, df=29, P=0.01). Figure 4 Mean monthly rate (±1 standard error) of all white seabass ( Atrac - toscion nobilis) sounds for audio recordings (n = 366) made from March 2003 to January 2004. A significant increase in sound production occurred during the spawning season (t=4.88, df=118, P>0.001). Hydrodynamic booms, which are generated by rapid movement through the water, accounted for a majority of the sounds recorded during the nonspawning season. 148 Fishery Bulletin 106(2) Average total sound production rate (sounds/min) Figure 5 Mean sound production rate (±1 standard error) of white seabass ( Atractoscion nobilis) for all combined sound types in relation to the release of gametes during 3-min evening recordings (n = 102) when spawning was documented from March to July 2003. The term “Inter” refers to the period between successive spawning events. The time between successive spawning was 1 to 98 min (mean = 31 min). 1500 1000 500 300 200 100 50 20 D 0.0 Hydrodynamic boom Time (s) Figure 6 Three-second sonograms of five basic sound types (A, B, D) gen- erated by white seabass ( Atractoscion nobilis), in addition to an identifiable spawning chant (C) consisting of overlapping drum-roll and thud sounds generated during the release of gametes. Modified from Aalbers and Drawbridge (in press). Taylor and Villoso, 1994) and in many fish species worldwide (Johannes, 1978). Spawning during low-light conditions may occur to reduce predation rates on both freshly spawned eggs and spawning adults (Lobel, 1978; Holt et ah, 1985). Broadcast spawners may further benefit from syn- chronizing spawning activity by temporally concentrating reproductive effort and maxi- mizing time allocated for diurnal movements and feeding. Lunar spawning White seabass spawned throughout the lunar cycle; although more individuals may have achieved reproductive readiness following the new moon, because 63% of mass spawn- ing events occurred from the new moon until four days after the new moon. The major- ity of known lunar spawning periodicities occur around the new or full moon, possibly to reduce egg predation through increased dis- persal around larger tidal flows or to ensure that males and females collectively achieve reproductive readiness (Johannes, 1978). Environmental conditions The majority of spawning occurred as water temperature and photoperiod increased during the spring and early summer; these factors appear to be important in stimulat- ing white seabass spawning activity. Most spawning occurred at water temperatures between 15 and 18°C; this range is consistent with that observed under hatchery protocols to artificially induce spawning in broodstock tanks. Although more spawning was docu- mented around high tide, this effect can be attributed to the increased occurrence of eve- ning high tides during summer within the mixed semidiurnal tide regime of southern California (Flick, 2000). Experimental variables Although variations in the onset and dura- tion of spawning seasons and the number of spawning events per season may represent biological responses to shifting environmen- tal conditions (e.g., temperature changes), differences may be partly attributed to research procedure. The captive population was 34% lower in 2001 because 21 white seabass were not combined with the origi- nal 41 fish until after the 2001 spawn- ing season. Additionally, increases in fish size and acclimation time over the 3-year period may have been responsible for the Aalbers: Spawning periodicities of Atractoscion nobilis 149 30 T 25 - ?E =s o m in -& a) ' > 15 0) 20 - 15 10 5 - Boom P-T Mult. P-T Thud Sound type Drumroll A. chant Chant Figure 7 (A) The mean rates (±1 standard error) and (B) mean relative sound pres- sure levels (dB rms [root mean square] values do not represent absolute sound pressure level) of white seabass ( Atractoscion nobilis) sounds for all 3-min evening recordings (n = 151) made from March through July 2003. Recordings are grouped by period of occurrence in relation to the time of spawning or categorized as “no spawning” on evenings that spawning did not occur. Prespawning intervals spanned the 2-h period before initial spawn- ing; postspawning intervals spanned the 2.5-h period after final spawning events; interspawning intervals spanned the period between successive spawning events, which occurred 1 to 98 minutes apart (mean=31 min). (P-T=pulse-train, Mult. P-T=multiple pulse-trains, A. chant = abbreviated chant, Chant = spawning chant). 46% increase in observed spawning activity in 2003 compared to 2001. Improved recognition of fish behav- iors and patterns likely enhanced the detection of spawning events in 2002 and 2003. Difficulty in differentiat- ing successive spawning events solely on the presence of eggs considerably reduced the total number of docu- mented events, along with an inabil- ity to visually detect spawning after dark. Variations in white seabass feed rations may have influenced interan- nual spawning activity because con- dition factor influences fecundity in weakfish ( Lowerre-Barbieri et al., 1996), haddock ( Melanogrammus aeglefinus) (Hislop et al., 1978), and Atlantic cod ( Gadus morhua) (Kjesbu and Klungsoyr, 1991). Spawning frequency Although data were not extensive enough to accurately determine indi- vidual white seabass spawning fre- quency or fecundity, it was apparent that females are serial spawners. The average number of observed spawning events per female in 2001, 2002, and 2003 was 2.0, 3.1, and 2.1, respectively; however, these values are underesti- mated because of experimental limita- tions. The same female spawned on 12 June 2002 between 2010 h and 2040 h and again on 16 June 2002 at 2038 h, indicating a capability for four-day spawning intervals. The observed shift in the diel spawning periodicity in relation to sunrise and sunset as the spawn- ing season progressed indicates that final oocyte maturation is not trig- gered solely by light intensity or the time of sunrise and sunset. White seabass exhibited a more consistent 24-h diel spawning pattern throughout the season, indicating that gamete development may occur with a circadian rhythm that could be modified through a combination of environmental cues (Taylor, 1984). Spawning rhythms and gamete development fluctu- ate synchronously with plasma estradiol-17/3 levels in female spotted seatrout and with testosterone and llketotestosterone concentrations in males (Brown- Peterson, 2003). In addition to regulating seasonal and diel spawning rhythms, increases in the levels of plasma androgens are correlated with seasonal hyper- trophy of sonic musculature as male weakfish come into spawning condition (Connaughton and Taylor, 1994; Connaughton et al., 1997). Sound patterns White seabass may experience alterations in repro- ductive endocrine hormones and physiological features that are similar to those observed in other sciaenids, because peak diel and seasonal spawning rhythms were directly correlated with increases in the rate, intensity, and variety of sound production. Sound production in orangemouth corvina and weakfish also culminates seasonally during peak spawning in May and diminishes in late July as spawning subsides (Fish and Cummings, 1972; Connaughton and Taylor, 1995). Additionally, diel increases in the sounds of weakfish and spotted seatrout have been documented after sunset during hours of 150 Fishery Bulletin 106(2) reproductive activity (Connaughton and Taylor, 1995; Gilmore, 2003). The rate of white seabass sound production in- creased 20-fold during the release of gametes, estab- lishing the highest rate and second highest intensity for all sounds. Measurable increases in the intensity of white seabass spawning chants and postspawning multiple pulse-trains were consistent with the audible detection of these sounds through the net-pen pipes and the hull of an adjacent boat. A chorus of orange- mouth corvina sounds increased ambient noise within the Salton Sea by 50 dB during the spawning season (Fish and Cummings, 1972). All of the distinct white seabass sound types were present on audio record- ings during spawning. In spotted seatrout, four major sound types were reported, all of which occurred dur- ing courtship and spawning (Gilmore, 2003). Sound function Direct correlations between spawning activity and sound production substantiate the hypothesis that white seabass sounds function to enhance reproduc- tive success. Weak single and multiple pulse-trains that were audible throughout the spawning season may help white seabass maintain aggregations. Single pulse-trains before spawning may augment courtship behaviors or communicate reproductive readiness. Intense multiple pulse-trains after spawning may serve to attract surrounding females into spawning aggregations. Identifiable spawning chants recorded during actual spawning likely function to enhance reproductive success by synchronizing the release of gametes. Conclusion This study provides a strong correlation between veri- fied white seabass spawning activity and sound pro- duction patterns. The noninvasive techniques used in this study can be extended to examine reproductive characteristics of other sound-producing fish species and alleviate difficulties associated with documenting spawning after dark. The essential fisheries information provided on key reproductive characteristics will help fisheries managers in designing strategies to sustain this economically important species and reduce the likelihood of another severe population decline. Baseline findings from this work can be extended to acoustically monitor white seabass spawning aggregations in order to investigate critical spawning habitat and to help determine better placement of marine protected areas throughout California. Acknowledgments Project support was provided by Hubbs-SeaWorld Research Institute, the Catalina Seabass Fund, Cali- fornia State University Fullerton, University of Southern California Wrigley Institute of Environmental Research, and Two Harbors Enterprises. I thank M. H. Horn, S. N. Murray, C. A. Sepulveda, R. G. Gilmore Jr., and anony- mous reviewers for constructive suggestions, and A. E. Bowles, K. M. McClune, P. E. Gardiner, K. C. Lafferty, G. M. Stutzer, K. A. Miller, and E. Forsman for project assistance. I appreciate substantial technical support and editorial comments provided by W. C. Cummings, K. A. Dickson, and M. A. Drawbridge. I am grateful to P. H. Offield and T. Pfleger for their continued dedication to preserving our marine resources. Literature cited Aalbers, S. A., and M. A. Drawbridge. In press. White seabass spawning behavior and sound production. Trans. Am. Fish. Soc. Allen, L. G., and M. P. Franklin. 1992. Abundance, distribution, and settlement of young- of-the-year white seabass Atractoscion nobilis in the Southern California Bight, 1988-89. Fish. Bull. 90:633-641. Brown-Peterson, N. 2003. The reproductive biology of spotted seatrout. In Biology of the spotted seatrout (S. A. Bortone, ed.), p. 99-133. CRC Press, Boca Raton, FL. Chao, L. N. 1986. A synopsis on zoogeography of the Sciaenidae. In Indo-Pacific fish biology: proceedings of the second international conference of Indo-Pacific fishes (T. Uyeno, R. Arai, T. Taniuuchi, and K. Masuura, eds.), p. 570-589. Ichthyol. Soc. Japan, Tokyo, Japan. Connaughton, M., and M. H. Taylor. 1994. Seasonal cycles in the sonic muscles of the weak- fish, Cynoscion regalis. Fish. Bull. 92:697-703. 1995. Seasonal and daily cycles in sound production associated with spawning in the weakfish, Cynoscion regalis. Environ. Biol. Fishes 42:233-240. Connaughton, M. A., M. L. Fine, and M. H. Taylor. 1997. The effects of seasonal hypertrophy and atrophy on fiber morphology, metabolic substrate concentration and sound characteristics of the weakfish sonic muscle. J. Exp. Biol. 200:2449-2457. Eschmeyer, W. N., E. S Herald, and H. Hanmann. 1983. A field guide to Pacific coast fishes of North Amer- ica, 386 p. Houghton Mifflin Co., Boston, MA. Fish, J. F., and W. C. Cummings. 1972. A 50-dB increase in sustained ambient noise from fish, Cynoscion xanthulus. J. Acoust. Soc. Am. 52:1262-1270. Flick, R. E. 2000. Time-of-day of peak tides in a mixed-tide regime. Shore and Beach 68:15-17. Gilmore, R. G., Jr. 2003. Sound production and communication in the spotted seatrout. In Biology of the spotted seatrout (S. A. Bor- tone, ed.), p. 177-195. CRC Press, Boca Raton, FL. Guest, W. C., and J. L. Lasswell. 1978. Notes on courtship behavior and sound production of red drum. Copeia 1978:337-338. Hislop, J. R., A. P. Robb, and J. A. Gauld. 1978. Observations on effects of feeding level on growth Aalbers: Spawning periodicities of Atractoscion nobilis 151 and reproduction in haddock, Melanogrmmus aeglefinus in captivity. J. Fish Biol. 13:85-98. Holt, G. J., S. A. Holt, and C. R. Arnold. 1985. Diel periodicity of spawning in sciaenids. Mar. Ecol. Prog. Ser. 27:1-7. Johannes, R. E. 1978. Reproductive strategies of coastal marine fishes in the tropics. Environ. Biol. Fishes 3:65-84. Kjesbu, O. S., and J. Klungsoyr. 1991. Fecundity, atresia and egg size of captive Atlantic cod Gadus morhua in relation to proxi- mate body composition. Can. J. Fish. Aquat. Sci. 48:2333-2343. Lobel, P. S. 1978. Diel, lunar, and seasonal periodicity in the repro- ductive behavior of the pomacanthid fish, Centopyge potteri, and some other reef fishes in Hawaii. Pac. Sci. 32:193-207. Lowerre-Barbieri, S. K., M. E. Chittenden, and L. R. Barbieri. 1996. Variable spawning activity and annual fecundity of weakfish in Chesapeake Bay. Trans. Am. Fish. Soc. 125:532-545. Luczkovich, J. L., M. W. Sprague, S. E. Johnson, and R. C. Pullinger. 1999. Delimiting spawning areas of weakfish Cynoscion regalis (Family Sciaenidae) in Pamlico Sound, North Carolina using passive hydroacoustic surveys. Bio- acoustics 10:143-160. Miller, D. J., and R. N. Lea. 1972. Guide to the coastal marine fishes of California. Ca- lif. Dep. Fish Game, Fish Bull. 157:1-249. Mok, H., and R. G. Gilmore. 1983. Analysis of sound production in estuarine aggre- gations of Pogonias cromis, Bairdiella chrysouea, and Cynoscion nebulosis (Sciaenidae). Bull. Inst. Zool., Acad. Sin. (Taipei) 22:157-186. Moser, H. G., D. A. Ambrose, M. S. Busby, J. L. Butler, E. M. Sandknop, B. Y. Sumida, and E. G. Stevens. 1983. Description of the early stages of white seabass, Atractoscion nobilis , with notes on distribution. Calif. Coop. Oceanic Fish. Invest. Rep. 24:182-193. Ramcharitar, J., D. P. Gannon, and A. N. Popper. 2006. Bioacoustics of fishes of the family Sciaenidae (croakers and drums). Trans. Am. Fish. Soc. 135: 1409-1431. Skogsberg, T. 1925. Preliminary investigations of the purse seine indus- try of southern California: white seabass. Calif. Div. Fish Game, Fish Bull. 9:53-63. 1939. The fishes of the family Sciaenidae (croakers) of California. Calif. Div. Fish Game, Fish Bull. 54:1- 62. Smith, H. M. 1905. The drumming of the drum-fishes. Science 22: 376-378. Tavolga, W. N. 1964. Sonic characteristics and mechanisms in marine fishes. In Marine bioacoustics (W. N. Tavolga, ed.), p. 195-209. Permagon Press Inc., New York, NY. Taylor, M. H. 1984. Lunar synchronization of fish reproduction. Trans. Am. Fish. Soc. 113:484-493. Taylor, M. H., and E. P. Villoso. 1994. Daily ovarian and spawning cycles in weak- fish. Trans. Am. Fish. Soc. 123:9-14. Thomas, J. C. 1968. Management of the white seabass Cynoscion nobi- lis in California waters. Calif. Dep. Fish Game, Fish Bull. 142:1-33. Tower, R. W. 1908. The production of sound in the drumfishes, the sea-robin and the toadfish. Ann. New York Acad. Sci. 18:149-180. 152 Abstract — King mackerel (Scomb- eromorus cavalla) are ecologically and economically important scom- brids that inhabit U.S. waters of the Gulf of Mexico (GOM) and Atlantic Ocean (Atlantic). Separate migra- tory groups, or stocks, migrate from eastern GOM and southeastern U.S. Atlantic to south Florida waters where the stocks mix during winter. Cur- rently, all winter landings from a management-defined south Florida mixing zone are attributed to the GOM stock. In this study, the stock composition of winter landings across three south Florida sampling zones was estimated by using stock-specific otolith morphological variables and Fourier harmonics. The mean accura- cies of the jackknifed classifications from stepwise linear discriminant function analysis of otolith shape variables ranged from 66-76% for sex-specific models. Estimates of the contribution of the Atlantic stock to winter landings, derived from maxi- mum likelihood stock mixing models, indicated the contribution was highest off southeastern Florida (as high as 82.8% for females in winter 2001-02) and lowest off southwestern Florida (as low as 14.5% for females in winter 2002-03). Overall, results provided evidence that the Atlantic stock contributes a certain, and perhaps a significant (i.e., >50%), percentage of landings taken in the management- defined winter mixing zone off south Florida, and the practice of assigning all winter mixing zone landings to the GOM stock should be reevaluated. Manuscript submitted 10 September 2007. Manuscript accepted 18 January 2008. Fish. Bull 106:152-160 (2008). The views and opinions expressed or implied in this article are those of the author and do not necessarily reflect the position of the National Marine Fisheries Service, NOAA. Spatial and temporal variability in the relative contribution of king mackerel ( Scomberomorus cavalla) stocks to winter mixed fisheries off South Florida Todd R. Clardy Email address: tclardy@bvaenviro.com Department of Marine Sciences University of South Alabama Mobile, Alabama 36688 Present address: Barry A. Vittor and Associates, Inc. 8060 Cottage Hill Road Mobile, Alabama 36695 William F. Patterson III University of West Florida 11000 University Parkway Pensacola, Florida 32514 Douglas A. DeVries Christopher Palmer Southeast Fisheries Science Center National Marine Fisheries Service, NOAA 3500 Delwood Beach Road Panama City, Florida 32408 King mackerel (Scomberomorus cavalla) are large coastal pelagic scombrids distributed from Massachu- setts to Brazil in the western Atlantic Ocean, including the Caribbean Sea (McEachran and Fechhelm, 2005). They support important commercial and recreational fisheries through- out the U.S. Gulf of Mexico (GOM) and in the Atlantic Ocean (Atlantic) off the southeastern United States. King mackerel currently are man- aged in U.S. waters as two migratory groups, one resident in the GOM and one off the southeast U.S. coast. The two-stock migratory group or two- stock model of population structure was adopted in the early 1980s based on tagging data indicating fish from the respective regions had distinct migratory pathways (Sutter et al., 1991). Subsequent studies demon- strated growth differences (DeVries and Grimes, 1997) and genetic dis- tinctiveness (Gold et al., 1997, 2002) between the stocks. Assessment and management of U.S. king mackerel stocks is compli- cated because of the seasonal mix- ing between GOM and Atlantic fish. Mark-recapture (Sutter et al., 1991) and catch-per-unit-of-effort studies (Trent et al., 1987) have indicated that winter migrations of king mack- erel occur from both the eastern GOM and Atlantic to south Florida where the mixed stock is targeted by a winter fishery. Thus, an area that stretches from the Volusia-Fla- gler county line in northeast Florida to the Monroe-Collier county line in southwest Florida was defined in the early 1980s by the Gulf of Mexico and South Atlantic Fishery Manage- ment Councils as a mixing zone be- tween the two stocks (GMFMC and SAFMC, 1985; Fig. 1). From April to October, all king mackerel landings in the mixing zone are attributed to the Atlantic stock, and landings from November through March are attributed to the GOM stock. This Clardy et at: Relative contribution of Scomberomorus cavalla stocks to winter fisheries off South Florida 153 85°W 80°W Figure 1 Winter mixing zone established for king mackerel ( Scomberomorus cavalla) off south Florida. The zone exists throughout the U.S. exclusive economic zone (EEZ) but fish are mostly found over the shelf (200-m isobath). All landings taken from the zone from November through March are attributed to the Gulf of Mexico stock. During the rest of the year, landings are attributed to the Atlantic stock. somewhat subjective stock assignment system was implemented in an effort to protect the GOM migratory group, which was estimated to be overfished. However, assessment and manage- ment are complicated by the presence of both Atlantic and GOM fish in the mixing zone during winter. Accurate estimation of the contribu- tion of each stock to winter landings is necessary for effective management and conservation. Several different techniques to distinguish these two groups have been explored in vari- ous studies. Tagging studies support the current two-stock management approach but have not resolved win- ter inter-stock mixing proportions. Likewise, although studies of popu- lation genetics have confirmed that genetically distinct Atlantic and GOM stocks exist, genetic divergence be- tween the two stocks is weak; thus differences are not robust enough to distinguish winter landings effec- tively (Broughton et al., 2002; Gold et al., 2002). Analysis of otolith shape has prov- en to be a useful technique for stock discrimination in several marine tele- osts (e.g., Atlantic cod, Gadus moi'hua [Campana and Casselman, 1993]; At- lantic salmon, Salmo salar [Friedland and Reddin, 1994]; and haddock, Melcinogrammus aeglefinus [Begg et al., 2001]). DeVries et al. (2002) demonstrated that otolith-shape parameters effectively distinguish Atlantic and GOM female king mackerel; classification accuracies from linear discriminant func- tion models ranged from 65.8% to 85.7%. They applied otolith-shape variables as natural markers to estimate the stock identity of female king mackerel landed be- tween Cape Canaveral and West Palm Beach, Flori- da, in winter 1996-97. A maximum likelihood model parameterized with stock-specific otolith-shape data revealed that Atlantic fish represented 99.8% of the composition of winter mixed fishery landings, thus cast- ing doubt on the management practice of attributing all winter mixing zone landings to the GOM stock. The objective of this study is to employ otolith-shape analysis to examine temporal and spatial variability in the Atlantic and GOM stock contribution to king mackerel landings around during winter around the southern tip of Florida. We build on the earlier suc- cess of DeVries et al. (2002) by examining sex-specific differences in otolith shape and by estimating the contribution of both Atlantic males and females to landings in the winter mixing zone. Temporal and spatial variability in stock mixing also is examined by estimating the Atlantic stock contribution to land- ings in three south Florida sampling zones distributed across the winter mixing area during two consecutive winters. Materials and methods King mackerel were sampled from recreational land- ings caught in eastern GOM and U.S. south Atlantic waters from April to October 2001 and 2002 when stock distributions did not overlap (Fig. 2); nearly all samples came from summer (June through September) months. Fish were measured to the nearest cm fork length (FL) and sex was determined by macroscopic examination of gonads. When possible, both sagittal otoliths were removed from fish, but for some samples only one sagitta was available. Once extracted, otoliths were cleaned of adhering tissue and placed in plastic vials for storage. Age of fish was estimated by examining whole otoliths for fish less than 80 cm FL and thin sections were prepared for aging larger fish (DeVries and Grimes, 1997). Stratified random sampling was employed once all samples were aged to select up to 15 fish from each of five age categories: ages 2, 3, 4, 5, and 6 years. This age range was selected because winter landings typically are of small, young fish. King mackerel also were sampled from commercial and recreational landings from three different zones 154 Fishery Bulletin 106(2) off south Florida from December 2001 to March 2002 and December 2002 to March 2003 (Fig. 2). Zone 1 represented southwest Florida and primarily consisted of samples from the commercial gillnet fishery near and to the east of the Dry Tortugas. Zone 2 represented south central Florida and consisted of samples from the recreational charter boat fishery operating south of Islamorada in the Florida Keys. Zone 3 represented southeast Florida and primarily consisted of samples from the commercial troll fishery from Sebastian Inlet to south of West Palm Beach, Florida. Collection and aging procedures for winter fish otoliths followed the same protocol as summer sampling. Left sagittal otoliths were digitized sulcus side down with an image analysis system running Image-Pro image analysis software (vers. 4.5, Media Cybernetics Inc., Bethesda, MD). Otolith samples were magnified by 13 x with a dissecting microscope before their im- ages were captured with the image analysis system. When left otoliths were damaged or unavailable, right otoliths were digitized and their mirror images were used for shape analysis (DeVries et al., 2002). The auto- trace feature in Image Pro then was used to trace the posterior surface of the otolith. Otolith tracing began at the tip of the antirostrum, was directed manually across the base of the rostrum, and then the software traced the posterior portion of the otolith. Thus, rostra were excluded from otolith-shape analysis because the anterior rostrum is fragile and often was broken during otolith collection (DeVries et ah, 2002). Fourier coefficients were computed with an algo- rithm within Image-Pro, and we used the mathemati- cally determined centroid as the center of an otolith. The Image-Pro algorithm used 128 vectors at equally spaced polar angles to create an accurate picture of the otolith outline. The amplitudes of the first 20 Fourier harmonics were calculated for analysis because each additional harmonic provides increasingly finer detail of the otolith outline. For example, 97-99% of otolith- shape variability in haddock ( Melanogrammus aeglefi- nus) is contained in the first ten harmonics (Begg and Brown, 2000). Fourier amplitudes were standardized to remove the effect of otolith size by dividing each amplitude by the mean radial length of the otolith. In addition to the first 20 standardized Fourier harmon- ics, the Image-Pro software calculated otolith area, perimeter, rectangularity, circularity, and roundness for a total of 25 shape variables. All variables were tested for univariate normality with the Shapiro-Wilks statistic and for homogeneity of variance with an Fmax test. Transformations were necessary for perimeter (natural log) and Fourier harmonics 13-16 (square- root) in order to meet parametric statisti- cal analysis assumptions of normality and homogeneity of variances. Ontogenetic effects on otolith shape were tested by computing the correlations of shape variables with fish length. Ontoge- netic effects were removed from each shape variable that was significantly correlated with fish length by subtracting the slope of the least squares linear relationship be- tween length and a given variable. Slope- corrected data were used in all subsequent analyses. Multivariate analysis of variance (MAN- OVA) was performed to test for potential shape differences between sides in a subset of 50 left and right sagittal otolith pairs (SAS, vers. 6.11, SAS Inst., Inc., Cary, NC). A second MANOVA also was per- formed to test for stock-specific differences in summer samples. The effect of other factors, including sex, age, and sampling year, on otolith shape parameters also was tested within this second MANOVA. Stepwise linear discriminant function (LDF) analysis was performed separately for sexes and years on otolith-shape vari- ables from summer sampled fish with the PROC STEPDISC procedure in SAS. The LDF procedure selected variables that were effective predictors of stock iden- tity. Jackknife cross-validation was used to evaluate the performance of resultant discriminant functions. Classification suc- 90°W 80°W Figure 2 Map of sampling locations for king mackerel ( Scomberomorous cavalla) in summers 2001 and 2002 in U.S. Atlantic Ocean waters (squares) and the Gulf of Mexico (circles). The map also shows the three winter sampling zones around southern Florida from which fish were sampled in winter 2001-02 and 2002-03 for estimates of the Atlantic stock contribution to winter landings. Clardy et al.: Relative contribution of Scomberomorus cavalla stocks to winter fisheries off South Florida 155 cess was estimated as the per- centage of individuals correctly classified to stock. The contribution of the At- lantic stock to winter fishery landings in each winter sam- pling zone was estimated by using the maximum likelihood (ML) modeling approach de- scribed in DeVries et al. (2002). Mixing estimates were calcu- lated for males and females separately by sample year. Otolith-shape variables were used in a two-step expectation- maximization (EM) algorithm written for the S-Plus statisti- cal package (Insightful Corp., Seattle, WA) (Millar, 1987; DeVries et al., 2002). Sex- and year-specific ML models first were parameterized with oto- lith-shape data from summer- sampled fish. Then, the EM algorithm computed estimates of the percentage of landings within a given winter sam- pling zone that were members of the Atlantic stock based on their otolith shape parameters. A bootstrap procedure (n = 500 bootstraps) was used to compute bias-corrected ninety percent confidence intervals around the maximum likelihood estimate (MLE) of Atlantic stock contribution. Results Summer sample sizes differed somewhat between stocks, sexes, and sampling years. One hundred twenty-six king mackerel (60 females, 66 males) were sampled in summer 2001, and 110 fish (56 females, 54 males) were sampled in summer 2002 from Atlantic waters. Seventy-three fish (37 females, 36 males) were sampled in summer 2001, and 120 fish (71 females, 49 males) were sampled in summer 2002 from the GOM. The age distributions of summer-sampled king mackerel gener- ally were similar between sexes, migratory groups, and years (Fig. 3). Sex-specific sample sizes were more variable from south Florida sampling zones during winter than dur- ing summer. In winter 2001-02, 153 fish (85 females, 68 males) were sampled in zone 1, 50 fish (44 females, 6 males) were collected in zone 2, and 142 fish (67 females, 75 males) were sampled in zone 3. In win- ter 2002-03, 158 fish (85 females, 73 males) were collected in zone 1, 72 fish (50 females, 22 males) were collected in zone 2, and 153 fish (86 females, 67 males) were collected in zone 3. The age distribu- tions of winter-sampled king mackerel were skewed toward younger fish in comparison to summer samples (Fig. 4). Correlation analysis indicated some ontogentic ef- fects on otolith shape may have been present. Several shape variables were significantly correlated with fish length (area, perimeter, roundness, rectangular- ity, circularity, Fourier harmonics 1-9, 11-14, 17, 19, and 20); the method described above was applied to remove the correlation of those variables with re- spect to fish length. MANOVA results indicated there were no significant differences in otolith shape be- tween left otoliths and right otoliths (MANOVA, P<0.601). Morphological features of otoliths proved to be different between stocks, but several other factors also significantly affected otolith shape. Sex and age, as well as stock, significantly affected otolith shape (MANOVA, PcO.001), but sampling year did not (MANOVA, P=0.964). Six of 25 shape variables were significantly different between sexes (ANOVA, P<0.05). Most of the shape differences were in vari- ables that described gross otolith morphological fea- tures (area, perimeter, roundness, circularity, and rectangularity), and only one of the significantly dif- ferent variables was a Fourier harmonic. Twelve of 25 shape variables were significantly different among ages (ANOVA, P<0.05), and most of the differences were in Fourier harmonics. Nine of 25 shape variables were significantly different between stocks (ANOVA, P<0.05). Most of the stock-specific shape differences 156 Fishery Bulletin 106(2) were in gross otolith morphological features or low- order Fourier harmonics. Sex and year-specific linear discriminant functions yielded a range of shape variables selected, and the mean accuracy of classifications ranged from 65.8% to 76.4% among models (Table 1). Discriminant functions included between five and seven variables. The highest classification accuracies from a given model were 71.1% for GOM females and 81.7% for Atlantic females in 2001 (mean accuracy 76.4%). The lowest classification accuracies were 61.2% for GOM males and 70.4% for Atlantic males in 2002 (mean accuracy 65.8%). Clas- sification accuracies were slightly higher for Atlantic fish (67.9-81.7%) than for GOM fish (61.2-71.1%) for most models. Atlantic stock king mackerel contributed to landings in all three winter sampling zones. Maximum likelihood models estimated that the contribution of Atlantic fish to winter landings ranged from 14.5% for females in zone 1 in 2002 to 99.9% for males in zone 2 in 2001 (Table 2). Bias-corrected bootstrapped 90% confidence intervals varied among zones and between years but generally were on the order of point estimates ±20%. Bootstrap cumulative frequency distributions demon- strated that although the majority of bootstraps fell near point estimates, wide confidence intervals resulted from long upper and lower distribution tails (Figs. 5 and 6). The estimated contribution of the Atlantic stock to 2001-02 winter landings was similar between males and females among all three winter sampling zones, except for zone 2 where few males were sampled (Table 2). In winter 2002-03, Atlantic females contributed less than males and also had lower contribution es- timates than females in 2001-02. Atlantic males had similar contribution estimates during both sampling years. Overall, a gradient in contribution estimates was observed; there were higher Atlantic stock percent- ages in southeast Florida (zone 3) and declining Atlantic stock presence in southwest Florida landings (zone 1). Discussion Classification accuracies from stepwise linear discriminant function analysis confirm the feasability of using otolith-shape parameters to distinguish king mackerel stocks but also dem- onstrate that stock-specific oto- lith-shape parameters provide natural tags that are far from perfect (i.e., <=100% stock dis- crimination success). The clas- sification success that we report (61.2% to 81.7%) is similar to the range reported in shape-based stock or population discrimina- tion for other fishes (e.g., 54.9% to 89.3% for lake whitefish, Core- gonus clupeaformis [Casselman et al., 1981]; 63.9% to 87.5% for Atlantic salmon [Friedland and Reddin, 1994]; 61% to 83% for haddock [Begg et al., 2001]; and, 63.6% to 83.3% for coral trout, Plectropomus leopardus [Bergenius et al., 2006], as well as that previously reported by DeVries et al. [2002] for female king mackerel [65.8% to 85.7%]). However, the lack of more dis- tinct differences in otolith shape between stocks likely contrib- uted significantly to the wide confidence intervals estimated 60 50 40 30 20 10 0 B Females Males Figure 4 Age distribution of king mackerel (Scomberomorus c avalla) in samples collectged in winters 2001-02 and 2002-03. (A) = 2001 zone 1 ; (B) = 2002 Zone 1 ; (C) = 2001 zone 2 ; (D) = 2002 zone 2 ; (E) = 2001 zone 3 ; and (F) = 2002 zone 3. Clardy et al.: Relative contribution of Scomberomorus cavalla stocks to winter fisheries off South Florida 157 Table 1 Accuracies of jackknifed classifications from stepwise linear discriminant function models computed with otolith shape param- eters to estimate summer king mackerel (Scomberomorus cavalla) stock identity. The model column identifies which sex- and year-specific models are examined. Numbers in the parameters column represent Fourier harmonics; Ro = Roundness, Re = Rectangularity, and P = Perimeter. Remaining columns indicate the percentage of fish correctly classified to the Atlantic and Gulf of Mexico (GOM) stocks with the jackknife algorithm. Model Parameters Accuracy (%) for Atlantic region Accuracy (%) for GOM region Mean accuracy (%) Females 2001 Ro, Re, 3, 7, 20 81.7 71.1 76.4 Males 2001 3, 5, 6, 8, 9, 10 69.7 67.6 67.8 Females 2002 P,Ro, 2,9, 13, 15, 16 67.9 70.8 69.4 Males 2002 P, Re, 2, 8, 11, 13 70.4 61.2 65.8 Table 2 Maximum likelihood estimates ( MLE ) of the contribution (%) of Atlantic stock king mackerel ( Scomberomorus cavalla ) to winter landings in each of three south Florida winter sampling zones by sex and year, with 90% bias-corrected confidence intervals (Cl) provided. The model column indicates which zone and year was estimated. Model MLE females 90% Cl MLE males 90% Cl Zone 1, 2001-02 61.0 32.2-82.7 61.0 40.2-73.9 Zone 2, 2001-02 48.6 20.1-67.2 99.9 60.9-100.0 Zone 3, 2001-02 82.8 62.8-99.8 76.0 57.0-99.7 Zone 1, 2002-03 14.5 0.0-28.9 45.0 21.2-70.0 Zone 2, 2002-03 41.3 20.9-68.9 83.1 49.4-100.0 Zone 3, 2002-03 40.4 24.2-59.5 71.9 51.5-99.4 from bootstrapped MLEs of Atlantic stock contribu- tion to south Florida winter king mackerel landings. Imprecision in those estimates prohibits more definitive conclusions about the relative contribution of GOM and Atlantic stocks to winter fisheries off south Florida. Nonetheless, it is possible to infer from our results that the Atlantic stock contributes substantially more than the zero percent of winter south Florida landings that is currently assumed by fishery management groups. Most of the otolith-shape differences between king mackerel stocks were observed in gross morphological variables and low-order Fourier harmonics. Low-order Fourier harmonics are related to general otolith shape, whereas high-order Fourier harmonics are related to increasingly fine-scale variation (Bird et al., 1986). DeVries et al. (2002) reported that gross otolith mor- phological parameters and low-order Fourier harmonics are significant contributors to otolith-shape variability in female king mackerel in southwest Florida, but they also reported many high-order Fourier harmonics to be significant as well. Sex effects on king mackerel otolith shape were sig- nificant for every gross morphological variable but for only one Fourier harmonic; this results indicates that sex-specific shape differences exist at a general level. Sex effects are not surprising given that sexually dimor- phic growth occurs in king mackerel; females achieve higher growth rates than males (Johnson et al., 1983; Manooch et al., 1987; Sturm and Salter, 1989; DeVries and Grimes, 1997). DeVries et al. (2002) examined only female king mackerel as a precaution against potential sex effects due to sexually dimorphic growth observed in this species. Most otolith shape studies that have tested for sex effects have found no significant differ- ences between males and females (Bird et al., 1986; Castonguay et al., 1991; Bolles and Begg, 2000; Begg et al., 2001). In studies where sex effects were significant, other factors were deemed more influential (Campana and Casselman, 1993). Otolith-shape variables in the models that best clas- sified king mackerel migratory groups were not con- sistent between sampling years. This result indicates that new shape-analysis models should be developed each summer and used only to estimate the migratory group composition of landings of the next winter. It is unclear why parameters in a discriminant function model may be important one year but of little value in distinguishing stocks the next year. However, interan- nual variability in growth rates between stocks may explain why LDFs do not perform well from one year to 158 Fishery Bulletin 106(2) Estimated contribution {%) of the Atlantic stock Figure 5 Cumulative probability distributions of bootstrapped estimates of the contribu- tion of Atlantic stock female king mackerel (Scomberomorus cavalla) to landings at three south Florida sampling zones. Drop-lines indicate the 5th, 50th, and 95th percentiles of bootstrap distributions. the next (Campana and Casselman, 1993). For example, cohort-specific discriminant function models computed for coral trout sampled on the Great Barrier Reef did a poor job distinguishing fish from another cohort to sampling region (34.3% to 39.7% classification success), a percentage that Bergenius et al. (2006) attributed to differences in growth rates ultimately caused by vari- ability in oceanographic conditions. Maximum likelihood estimates indicated that some percentage of winter landings in all three zones orig- inated from the Atlantic stock in both study years. However, bootstrapped confidence intervals indicated considerable imprecision around point estimates. Cu- mulative probability distributions of bootstraps (rc = 500) were broad for females and males in both study years. However, even at the lower end of the confidence inter- vals, Atlantic fish were estimated to have contributed greater than 20% of landings in all three zones, except for females sampled in zone 1 during winter 2002-03. Results potentially indicate that a distribution gradi- ent may exist; more Atlantic king mackerel may contrib- ute to landings from the Atlantic side (zone 3) and fewer Atlantic stock king mackerel contribute toward the GOM (zone 1). Mixing estimates for zone 2 are some- where in the middle, with the exception of zone 2 males in 2001-02. However, the sample size of king mackerel in zone 2 in 2001-02 generally was low, particularly for males, and this shortage could account for the higher estimate for the Atlantic stock contribution. Atlantic male and female king mackerel appear to have had similar contributions across all three south Florida sampling zones in winter 2001-02, but this Clardy et at: Relative contribution of Scomberomorus cava/la stocks to winter fisheries off South Florida 159 was not the case in winter 2002-03. Zone 1 and zone 3 in particular showed reductions of 35% and 45%, respectively, in the contribution of Atlantic females in 2002-03. It is unclear why Atlantic females were esti- mated not to have contributed as significantly to land- ings in these zones. Differences in classification accura- cies between summer 2001 and summer 2002 females may have affected landings contribution estimates, but discriminant function classification accuracies differed by only 7% between years. The reduced contribution of Atlantic females in winter 2002-03 most likely reflects temporal variability in stock mixing. Overall, results of this study provide further evi- dence that the U.S. Atlantic king mackerel stock con- tributes a certain, and perhaps a significant, per- centage of landings taken in the management-defined winter mixing zone off south Florida. Based on our results, fisheries managers should consider adopting some form of a gradient approach in attributing south Florida winter landings to GOM and Atlantic stocks. An alternative, and perhaps more easily defended, management approach may be to assign 50% of win- ter mixing zone landings to the Atlantic stock in the absence of annual estimates of stock-specific landing contributions. Acknowledgments This study was supported by the Marine Fisheries Initiative (MARFIN) Program (contract number: NA17FF2013). The authors thank C. Newton, J. Lehrter, 160 Fishery Bulletin 106(2) J. Jackson, E. Little, M. Gamby, and National Marine Fisheries Service and North Carolina Division of Marine Fisheries port agents for assistance with sample collec- tion. The authors also thank the many recreational and commercial fishermen who allowed us to sample their catches, as well as marina owners, seafood dealers, and charterboat captains who permitted us to sample at their facilities. Literature cited Begg, G. A., and R. W. Brown. 2000. Stock identification of haddock, Melanogrammus aeglefinus, on Georges Bank based on otolith shape analysis. Trans. Am. Fish. Soc. 129:935-945. Begg, G. A., W. J. Overholtz, and N. J. Munroe. 2001. The use of internal otolith morphometries for iden- tification of haddock (Melanogrammus aeglefinus) stocks on Georges Bank. Fish. Bull. 99:1-14. Bergenius, M. A. J., G. A. Begg, and B. D. Mapstone. 2006. The use of otolith morphology to indicate the stock structure of common coral trout ( Plectropomus leopar- dus) on the Great Barrier Reef, Australia. Fish. Bull. 104:498-511. Bird, J. L., D. T. Eppler, and D. M. Checkley Jr. 1986. Comparisons of herring otoliths using Fourier series shape analysis. Can. J. Fish. Aquat. Sci. 43:1228-1234. Bolles, K. L., and G. A. Begg. 2000. Distinction between silver hake (Merluccius bilin- earis) stocks in U.S. waters of the northwest Atlantic based on whole otolith morphometries. Fish. Bull. 98:451-462. Broughton, R. E., L. B. Stewart, and J. R. Gold. 2002. Microsatellite variation suggests substantial gene flow between king mackerel ( Scomberomorous cavalla ) in the western Atlantic Ocean and Gulf of Mexico. Fish. Res. 54:305-316. Campana, S. E., and J. M. Casselman. 1993. Stock discrimination using otolith shape analysis. Can. J. Fish. Aquat. Sci. 50:1062-1083. Casselman, J. M., J. J. Collins, E. J. Crossman, P. E. Ihssen, and G. R. Spangler. 1981. Lake whitefish (Coregonus clupeaformis) stocks of the Ontario waters of Lake Huron. Can. J. Fish. Aquat. Sci. 38:1772-1789. Castonguay, M., P. Simard, and P. Gagnon. 1991. Usefulness of Fourier analysis of otolith shape for Atlantic mackerel ( Scomber scombrus) stock discrimination. Can. J. Fish. Aquat. Sci. 48: 296-302. DeVries, D. A., and C. B. Grimes. 1997. Spatial and temporal variation in age and growth of king mackerel, Scomberomorus cavalla, 1977-1992. Fish. Bull. 95:694-708. DeVries, D. A., C. B. Grimes, and M. H. Prager. 2002. Using otolith shape analysis to distinguish east- ern Gulf of Mexico and Atlantic Ocean stocks of king mackerel. Fish. Res. 57:51-62. Friedland, K. D., and D. G. Reddin. 1994. Use of otolith morphology in stock discriminations of Atlantic salmon ( Salmo salar). Can. J. Fish. Aquat. Sci. 51:91-98. Gold, J. R., A. Y. Kristmunddsdottir, and L. R. Richardson. 1997. Mitochondrial DNA variation in king mackerel (Scomberomorus cavalla) from the western Atlantic Ocean and Gulf of Mexico. Mar. Biol. 129:221-232. Gold, J. R., E. Pak, and D. A. DeVries. 2002. Population structure of king mackerel ( Scomb- eromorus cavalla) around peninsular Florida, as revealed by microsatellite DNA. Fish. Bull. 100:491-509. GMGMC (Gulf of Mexico Fishery Management Council) and SAFMC (South Atlantic Fishery Management Council). 1985. Final amendment 1, Fishery management plan and environmental impact statement for coastal migra- tory pelagic resources (mackerels) in the Gulf of Mexico and South Atlantic region, 202 p. [Available from GMFMC, 2203 N. Lois Ave., Suite 1100, Tampa, FL 33607 and SAFMC, 4055 Faber Place Drive, Suite 201, North Charleston, SC 29405.] Johnson, A. G., W. A. Fable, Jr., M. L. Williams, and L. E. Barger. 1983. Age, growth, and mortality of king mackerel, Scomberomorus cavalla, from the southeastern United States. Fish. Bull. 81:97-106. Manooch, C. S. Ill, S. P. Naughton, C. B. Grimes, and L. Trent. 1987. Age and growth of king mackerel, Scomberomorus cavalla, from the U.S. Gulf of Mexico. Mar. Fish. Rev. 49:102-108. McEachran, J. D., and J. D. Fechhelm 2005. Fishes of the Gulf of Mexico, vol. 2, 1004 p. Univ. Texas Press, Austin, TX. Millar, R. B. 1987. Maximum likelihood estimation of mixed stock fishery composition. Can. J. Fish. Aquat. Sci. 44: 583-590. Sturm, M. G., and P. Salter. 1989. Age, growth, and reproduction of the king mack- erel, Scomberomorus cavalla, (Cuvier) in Trinidad waters. Fish. Bull. 88:361-370. Sutter, F. C. Ill, R. O. Williams, and M. F. Godcharles. 1991. Movement patterns and stock affinities of king mackerel in the southeastern United States. Fish. Bull. 89:315-324. Trent, L., B. J. Palko, M. L. Williams, and H. A. Brusher. 1987. Abundance of king mackerel, Scomberomorus cavalla, in the southeastern United States based on CPUE data from charterboats, 1982-1985. Mar. Fish. Rev. 49:78-90. 161 Abstract — Ichthyoplankton samples were collected at approximately 2-week intervals, primarily during spring and summer 1999-2004, from two stations located 20 and 30 km from shore near the Columbia River, Oregon. Northern anchovy ( Engraulis mordax) was the most abundant species collected, and was the primary species associated with summer upwelling conditions, but it showed significant interannual and seasonal fluctuations in abundance and occurrence. Other abundant taxa included sanddabs (Citharich- thys spp. ), English sole ( Parophrys vetulus), and blacksmelts (Bathy- lagidae). Two-way cluster analysis revealed strong species associations based primarily on season (before or after the spring transition date). Ich- thyoplankton abundances were com- pared to biological and environmental data, and egg and larvae abundances were found to be most correlated with sea surface temperature. The Pacific Decadal Oscillation changed sign (from negative to positive) in late 2002 and indicated overall warmer conditions in the North Pacific Ocean. Climate change is expected to alter ocean upwelling, temperatures, and Columbia River flows, and conse- quently fish eggs and larvae dis- tributions and survival. Long-term research is needed to identify how ichthyoplankton and fish recruitment are affected by regional and large- scale oceanographic processes. Manuscript submitted 23 August 2007. Manuscript accepted 28 January 2008. Fish. Bull. 106:161-173 (2008). The views and opinions expressed or implied in this article are those of the author and do not necessarily reflect the position of the National Marine Fisheries Service, NOAA. Ichthyoplankton community in the Columbia River plume off Oregon: effects of fluctuating oceanographic conditions Maria M. Parnel Robert L. Emmett (contact author) Richard D. Brodeur Email address for Robert L Emmett: Robert.Emmett@noaa.gov National Oceanic and Atmospheric Administration National Marine Fisheries Service Hatfield Marine Science Center 2030 S Marine Science Drive Newport, Oregon The Columbia River and its plume form a biologically rich environment. It is the largest river flowing into the Pacific Ocean in North America (Sher- wood et al., 1990), has relatively large salmon runs, a healthy white sturgeon (Acipenser transmontanus ) population, and probably the largest American shad ( Alosa sapidissima) run in the world. The Columbia River estuary is also one of the largest estuaries on the west coast (Emmett et al., 2000) and supports abundant piscivorous marine mammal and bird populations (NMFS, 1997; Collis et al., 2002). Important marine fisheries located in the Columbia River plume include those for the following species: Dunge- ness crab ( Cancer magister ), salmon ( Oncorhynchus spp.), and Pacific sar- dine ( Sardinops sagax) (Emmett et al., 2005, 2006). River plumes and their associ- ated fronts are important habitats for fishes during their larval stages (Grimes and Funucane, 1991; Govo- ni and Grimes, 1992; Govoni, 1997). Large freshwater discharges from the Columbia River create abrupt ocean- ographic fronts that may be impor- tant for salmon ( Schabetsburger et al., 2003; DeRobertis et al., 2005) and other fish species, and they con- centrate some species of zooplankton (Morgan et al., 2005). The Columbia River plume has also been shown to be an important spawning and rear- ing habitat for some fishes (Richard- son, 1981; Doyle et al., 1993, 2002; Doyle, 1995; Emmett et al., 1997), particularly for the northern stock of northern anchovy ( Engraulis mordax) (Richardson, 1981). This region also has regular upwelling, which is a dominant feature of the California Current system and in the Columbia River plume region, and it ultimately determines the amount of primary, and probably secondary, production within this system (Hickey and Ba- nas, 2003). Ichthyoplankton surveys have been conducted to augment available abun- dance data on many commercially important fish species and are often used to determine spawning biomass (Lasker, 1985; Hunter and Lo, 1993) as well as to understand recruitment processes (Bailey, 1981; Cowan and Shaw, 2002). Data sets generated by such ichthyoplankton investigations provide an opportunity to determine the role that oceanographic processes and climate change play in recruit- ment variability of commercially valu- able species. There have been relatively few ich- thyoplankton surveys conducted off the Pacific Northwest. In previous ichthyoplankton surveys conducted off Oregon, Richardson and Pearcy (1977), Brodeur et al. (1985), and Auth and Brodeur (2006) identified the ichthyofaunal community, but these surveys were relatively short in duration and the area near the Co- lumbia River was not sampled. Oth- er coastal ichthyoplankton surveys 162 Fishery Bulletin 106(2) have been conducted off the Columbia River, but have been limited to one year (Waldron, 1972) and thus it was not possible to identify any relationship between fluctuations in ocean conditions and changes in the ichthyofaunal community. However, Doyle (1995) did extensive sampling during spring (April-May), and some during other months from 1980-85 and 1987, and found distinct changes in the Northwest ichthyo- plankton community resulting from the 1983 El Nino. Later, Doyle et al. (2002) examined ichthyoplankton in the Northeast Pacific, including the Pacific Northwest and the Columbia River plume, and identified regional, onshore, and offshore differences in the ichthyoplankton communities. Northern anchovy was a dominant spe- cies off the Pacific Northwest during all these ichthyo- plankton surveys. Northern anchovy eggs and larvae have been found to be abundant around the Columbia River plume (Richardson, 1981; Emmett et ah, 1997). However, detailed information on the temporal and annual variability of the dominant ichthyoplankton components is lacking. This is noteworthy, given the importance of larval and postlarval fishes in the diets of salmon in the plume region (Schabetsberger et al., 2003; DeRobertis et al., 2005) and the important role that forage fishes, particular northern anchovy, Pacific herring ( Clupea pallasi), and smelt, play in the diet of piscivorous fishes, mammals, and birds in this region. Our study was initiated to provide data to address this need, and to further elucidate the early life histories of fish species that use the Columbia River plume under a variety of environmental conditions. Repeated plankton sampling at two stations just south of the mouth of the Columbia River over a 6-year period provided us an opportunity to study interan- nual and seasonal changes in the diversity and species assemblages in the nearshore ichthyoplankton com- munity of this region. We also identified the effects of fluctuating river flows and nearshore oceanic environ- mental conditions on the ichthyofaunal community. We were provided this opportunity because of the fortuitous overlap of our research with the regime change in the Pacific Decadal Oscillation (Mantua et al., 1997) in late 2002 (Goericke et al., 2005). Materials and methods Study site The Columbia River has a diverse drainage basin that covers approximately 670,000 km2 and has average flows of approximately 10,000 m3/s the latter of which create a surface lens, or plume, of low-salinity water in the adjacent ocean that can extend up to 400 km offshore (Barnes et al., 1972). During summer, the plume flows outward and southward, but during winter it is confined to a narrow band along the Washington coast (Anderson, 1972; Hickey and Banas, 2003). In summer months the plume is characterized by lower salinities (<32 psu), higher turbidities, and higher temperatures than the correspond- ing variables for surrounding waters and is an important habitat for many species of inver- tebrates and fishes (DeRobertis et al., 2005; Morgan et al., 2005; Emmett et al., 2006). Data collection Plankton samples were collected at two sta- tions approximately every two weeks during the spring and summer, as well as occasionally during fall and winter in 1999-2004 (Table 1). The first station, buoy 1, is located 30 km from the mouth of the river and has a bottom depth of 98 m. The second station, buoy 2, is located 20 km from the mouth of the river and has a bottom depth of 47 m (Fig. 1). Oblique plankton tows to a depth of approximately 40 m were conducted with a 1-m diameter net with 335- fim mesh and a centrally mounted flow meter. The net was washed with seawater after each tow, and the contents were preserved in a 10% buffered formalin in seawater solution. Any large scyphomedusas captured in the net were rinsed and removed from the sample. A Niskin Figure 1 Map of the study area showing locations of two buoys off the Columbia River, Oregon, where ichthyoplankton was collected during survey cruises of the National Oceanic and Atmo- spheric Administration, National Marine Fisheries Service, 1999-2004. Also shown are depth contours (50-500 m) and the location of Astoria Canyon. Parnel et al.: Effects of oceanic conditions on ichthyoplankton communities in the Columbia River plume 163 Table 1 Total number (by year and month) of ichthyoplankton samples collected off the Columbia River, Oregon, during survey cruised conducted by the National Oceanic and Atmospheric Administration, National Marine Fisheries Service. 1999 2000 2001 2002 2003 2004 Total January 2 0 0 0 0 0 2 February 0 0 1 0 0 0 1 March 2 2 0 0 0 0 4 April 3 1 0 2 2 2 10 May 4 3 4 1 6 6 24 June 2 2 6 3 6 4 23 July 2 2 5 2 4 4 19 August 0 0 0 0 0 0 0 September 1 0 0 0 0 0 1 October 0 0 0 0 0 0 0 November 0 0 0 0 0 0 0 December 1 0 0 0 0 0 1 Total 17 10 16 8 18 16 85 bottle was used to collect a water sample for chlorophyll analysis at the 3-m depth at each station. A conductiv- ity-temperature-depth (CTD) cast was also conducted to within 5 m of the bottom at each station. Before they were sorted, ichthyoplankton samples were rinsed in fresh water with a 332-pm mesh sieve. The samples were then poured into a large Pyrex dish on a light box. All fish larvae and eggs were removed from the samples. Identifications were made to the low- est possible taxa by using a dissecting microscope and taxonomic information, primarily from Matarese et al. (1989) and Moser (1996). Zooplankton volume was determined by allowing the sample to settle in a gradu- ated cylinder overnight. Zooplankton volumes and ich- thyoplankton densities were standardized by adjusting these numbers by the volume of water filtered by the net. Data analysis Species diversity was calculated with the Shannon- Weaver index of diversity, H' (Shannon and Weaver, 1949; Krebs, 1989), where higher values denote greater species diversity. Species evenness was calculated with Simpson’s index of evenness. A, which ranges between 0 and 1, and where a value of 1 indicates that all taxa have the same density within a sample (Krebs, 1989). To eliminate the effect of rare and uncommon taxa, data from each station were filtered to remove those taxa that occurred in less than 5% of the samples, thus leaving 12 egg taxa and 12 larval taxa for further analysis of the community structure. Agglomerative hi- erarchical two-way cluster analyses were conducted to identify taxa and sample assemblages. For this analy- sis, densities were averaged from both stations from a single collection date. Samples with no ichthyoplankton were excluded. Catch distributions were highly skewed and thus were log10(cafc/i/1000 m3+l) transformed to de-emphasize the few high catches. We constructed separate two-way clusters for eggs and larvae data with the Bray-Curtis distance measure and a flexible beta (/3=-0.25) clustering algorithm, using PC-ORD software (vers. 5, MJM Software Design, Gleneden Beach, OR). A nonparametric multiresponse permutation proce- dure (MRPP) (Kruskal, 1964; Mather, 1976) was used to test the hypothesis of no difference between two or more groups. Factors tested were station (buoy 1 and 2), year (1999, 2000, 2001, 2002, 2003, 2004), and season (downwelling and upwelling), before and after the spring transition date as defined by Logerwell et al. (2003). The date of the spring transition is the day when the ocean conditions switch from downwelling to upwelling, as identified by analyzing Bakun upwell- ing indices and sea level conditions along the Pacific Northwest (Logerwell et al., 2003). Identification of the primary taxa associated with each grouping was done by using an indicator species analysis (ISA). The ISA measures the fidelity of a taxon within a particular group in relation to its abundance in all groups and assigns an indicator value for that taxon per group. A statistical significance was then calculated by a Monte Carlo resampling method with 1000 runs. These tests were also run with PC-ORD V5 software. Relationships between environmental conditions and the ichthyoplank- ton community were investigated by using the BVSTEP procedure in the PRIMER software package (vers. 5, PRIMER-E Ltd, Lutton Ivybridge, UK). This analysis is analogous to forward stepwise multiple regression and compares nonparametrically paired species and environ- mental similarity matrices. Species similarity matrices were calculated by using Bray-Curtis similarities on fourth-root transformed data. Environmental matrices 164 Fishery Bulletin 106(2) were calculated by using the normalized Euclidean dis- tance on fourth-root transformed data. Environmental variables in the analysis included sea surface tempera- ture, salinity, chlorophyll a at 3 m, the settled volume of zooplankton, and Columbia River flow (measured by U.S. Geological Survey at Beaver, OR). A paired f-test was used to determine if environ- mental conditions and zooplankton volumes at the two stations were significantly different. To account for interannual and seasonal variations in sampling, only those samples collected within a similar season were considered. All statistical tests were considered signifi- cant at the a <0.05 level. Results In total, 85 samples were collected during this study, comprising a total of 155,302 fish eggs and 3565 fish 100 100 1999 2000 2001 2002 Year 2003 2004 Figure 2 The top five egg (A) and larval (B) taxa collected at two stations off the mouth of the Columbia River, Oregon, as a percentage of total catch for each year from 1999 through 2004. larvae, representing 34 taxa and 17 families. However, a majority (n- 78) of these samples were taken from April to August and formed the primary data set that was analyzed (Table 2). The family Pleuronectidae had the greatest number of taxa (9), followed by the family Cottidae (6). The actual number of fish species collected was probably higher than the number we report because some eggs and larvae could be identified only to genus or family, namely eggs and and larvae of Osmeridae, Bathylagidae, Citharichthys spp., Sebastes spp., and Sebastolobus spp. Nine taxa were found only once during our sampling. The most abundant fish eggs were those of northern anchovy, sanddab ( Citharichthys spp.), blacksmelt ( Bath - ylagus spp.), jack mackerel (Trachurus symmetricus), and Pleuronectidae, in descending order of abundance (Fig. 2A). The most abundant larvae, in descending order, were northern anchovy, English sole ( Paroph - rys vetulus), sand sole ( Psettichthys melanostictus), rex sole ( Glyptocephalus zachirus), and Pacific tomcod ( Microgadus proximus) (Fig. 2B). The unidentified Pleuronectidae eggs were of four species (butter sole [Isopsetta isolepis], English sole, starry floun- der [Platichthys stellatus], and sand sole) which cannot be identified to species at early egg stages. Unusual egg occurrences included those of Pacific viperfish ( Chauliodus macouni ), opah (Lampris guttatus ), and jack mackerel. These species are considered either offshore (viperfish and opah) or southern, warm-water spawners (jack mackerel). Jack mackerel eggs were observed in 1999-2002, but not in 2003 or 2004 (Fig. 2A). Peak jack mack- erel spawning occurred in July 1999 and June 2000-2002, although some eggs were also found from May to September during most years. Northern anchovy accounted for 76% of all eggs collected, and were found from May through July (Fig. 3). Overall, northern anchovy were pres- ent in 53 of the total 85 samples collected. Peak northern anchovy egg abundance varied interan- nually, occurring in June of 1999, 2001, and 2002; July of 2000; and in May of 2003. The highest annual density of northern anchovy eggs was ob- served in 2003, when eggs were almost four times as abundant as those observed in 2002 (Fig. 3). Northern anchovy eggs dominated the catch in all years except 2001, when blacksmelt and sanddab contributed about equally to the catch (Fig. 2). Northern anchovy larvae were the most abun- dant fish larvae, and represented 68% of all larvae collected. In 1999, 2001, and 2002, peak numbers of these larvae occurred in June, whereas in 2000, northern anchovy larval numbers peaked in July. In 2003 and 2004, northern anchovy larval num- bers peaked in May and represented the majority of the larval fish catch (Fig. 2B). Nine samples contained only one egg or larval fish species (and therefore a diversity and even- ness of zero), and three samples contained no ich- thyoplankton. Overall sample diversity ranged Parnel et al.: Effects of oceanic conditions on ichthyoplankton communities in the Columbia River plume 165 Table 2 Taxonomic and stage composition of ichthyoplankton collected during survey cruises conducted by the National Oceanic and Atmospheric Administration, National Marine Fisheries Service approximately every two weeks from late April to August 1999-2004 from two stations located off the Columbia River, Oregon. Occurrence is the number of samples in which the life stage of a taxon was found. Occurrence Density Number Stage Species Common name (rc=78) (no. /1000 nr3) captured collected Clupeidae Sardinops sagax Pacific sardine 4 0.00899 86 Eggs Engraulidae Engraulis mordax Northern anchovy 52 7.09502 118,089 Eggs 23 0.18015 2435 Larvae Bathylagidae Blacksmelts 18 0.81678 11,027 Eggs Chauliodontidae Chauliodus macouni Pacific viperfish 2 0.00012 2 Eggs Gadidae Microgadus proximus Pacific tomcod 13 0.00278 32 Larvae Lampridae Lampris guttatus Opah 1 0.00011 1 Eggs Gasterosteidae Stickleback 1 0.0001 1 Larvae Scorpaenidae Sebastes spp. Rockfish 2 0.00021 2 Larvae Sebastolobus spp. Thornyheads 5 0.0004 6 Eggs Hexagrammidae Ophiodon elongatus Lingcod 2 0.00186 2 Larvae Cottidae Artedius spp.7 1 0.00007 1 Larvae Artedius harringtoni Scalyhead sculpin 2 0.00031 5 Larvae Clinocottus acuticeps Sharpnose sculpin 1 0.00012 1 Larvae Leptocottus armatus Staghorn sculpin 1 0.00026 2 Larvae Ruscarius meanyi Puget Sound sculpin 1 0.00008 1 Larvae Liparidae Liparis spp. Unid. snailfish 2 0.00018 3 Larvae Liparis fucensis Slipskin snailfish 1 0.00008 1 Larvae Liparis pulchellus Showy snailfish 2 0.00025 3 Larvae Carangidae Trachurus symmetrieus Jack mackerel 15 0.19986 2178 Eggs Bathymasteridae Ronquilis jordani Northern ronquil 1 0.00095 12 Larvae Pholidae Pholis spp. Unid. gunnel 1 0.00004 1 Larvae Icosteidae Icosteus aenigmaticus Ragfish 2 0.00034 4 Eggs Ammodytidae Ammodytes hexapterus Pacific sandlance 2 0.00187 2 Larvae Paralichthyidae Citharichthys spp. Sanddab 50 1.66568 22,186 Eggs 1 0.00008 1 Larvae Pleuronectidae Unid. flounder 20 0.04348 673 Eggs 2 0.00029 3 Larvae Glyptocephalus zachirus Rex sole 6 0.00667 34 Eggs 3 0.00546 5 Larvae Isopsetta isolepis Butter sole 6 0.00072 9 Larvae Lepidopsetta bilineata Southern rock sole 1 0.00029 4 Larvae continued 166 Fishery Bulletin 106(2) Table 2 (continued) Species Common name Occurrence (n =78) Density (no. /1000 m3) Number captured Stage collected Pleuronectidae (cont.) Lyopsetta exilis Slender sole 4 0.00082 9 Eggs 2 0.00016 2 Larvae Microstomus pacificus Dover sole 8 0.00566 65 Eggs 4 0.00123 24 Larvae Parophrys vetulus English sole 6 0.00101 11 Larvae Platichthys stellatus Starry flounder 1 0.00016 2 Larvae Pleuronichthys decurrens Curlfin sole 4 0.00504 5 Eggs 1 0.00015 2 Larvae Psetticthys melanostictus Sand sole 7 0.00920 112 Eggs 17 0.01202 164 Larvae 1 Either Artedius corallinus or A. notospilotus. Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Figure 3 Densities of northern anchovy ( Engraulis mordax) eggs sampled off the mouth of the Columbia River, Oregon, in relation to sampling date for 1999 through 2004 (note log scale ony-axis). An X represents dates when sampling occurred but no northern anchovy eggs were collected. from 0 to 1.7 and evenness from 0 to 1. Over all years, diversity was highest in April for eggs at both stations and for larvae at buoy 1, whereas at buoy 2, peak di- versity for larvae occurred in March. Lowest diversities occurred in late fall and especially in December. Even- ness was highest in April and lowest in December. More than half of all species included in the egg- stage cluster analysis were flatfish from the families Pleuronectidae and Paralichthyidae (Fig. 4) . The most closely grouped species were the northern anchovy and sanddab. Other taxa with a similar abundance pattern included the blacksmelt (Bathylagidae) and Pleuro- nectidae. Three sampling groups emerged in the egg cluster analysis (Fig. 4). The first group or cluster (A) consisted mainly of samples collected in winter-spring (December-early May), before the spring transition, and was dominated by rex sole and other flatfishes. Cluster B was composed of mainly late spring (April-June) samples and was dominated by jack mackerel and sanddab eggs. In contrast, Cluster C was predomi- Parnel et al.: Effects of oceanic conditions on ichthyoplankton communities in the Columbia River plume 167 nately a spring-summer (May-July) group dominated by northern anchovy, sanddab, and blacksmelt (Fig. 4). Among the larvae, Pacific sandlance ( Ammodytes hexapterus), rex sole, and curlfin sole ( Pleuronichthys decurrens) were tightly clustered, as were northern anchovy and thornyheads ( Sebastolo - bus spp.) (Fig. 5). The remaining larger clusters consisted of mainly flatfishes and Pacific tomcod. Sampling clusters were less defined for larval fish than for eggs (Fig. 5). Cluster A consisted of samples collected from multiple years and months, and most were collected before the spring transition. No spe- cies dominated these collections. In contrast, Cluster B was dominated by Pacific tomcod and Dover sole (Micros- tomus pacificus). Cluster C occurred entirely during May and June and was distinguished by the high abundance of northern anchovy (Fig. 5). There were very limited differences in the biological or physical oceano- graphic conditions between the two sampling stations. Only salinity was significantly different between the two stations U-test, P<0.05). Temperature, chlorophyll a, zooplankton densities, and densities of the five most abun- dant ichthyoplankton species were not significantly different between stations. However, there were significant annual differences in temperatures (ANOVA, P<0.05), but not salinities or chloro- phyll a at these stations during the study period (Table 3). There were also large annual differences in the date of the spring transition (Table 3). The BVSTEP analysis revealed that tem- perature explained most of the inter- annual variation in the ichthyoplank- ton community (Spearman p=0.345), (p ranges from -1 to 1, where values farthest from zero indicated a stron- ger correlation) followed by salinity (p = 0.194) and Columbia River flow (p=0.145). Chlorophyll a and overall ichthyoplankton densities did not pro- vide any explanatory power and did not relate to variations in ichthyoplankton species composition. The egg and larval fish data showed very few differences when these data were grouped and compared by year, season, and station. MRPP analysis revealed few significant differences be- tween groups (Table 4). The taxonomic composition of the dominant egg or Eggs Information remaining (%) 25 50 75 1 h 100 §12 ■ <=-c -C ^ N gj ^ . a> . . OUjmo_CD^i^CLC/)CLCLK' 1-5-99 □□□□□□ ■ ■ 3-11-99 111 m mm 12-6-99 □□□□□□ m 4-23-99 5-5-99 ■■ ■ m □□□□i m -3--21-IXL 7-14-99 Li . . ■ ■ m os 7-31-01 am 7-23-02 mo 2-13-01 ■ T ][ . 5-14-99 S3 H 5-28-99 sc ■ ■ 4-24-02 H ■ 7-28-99 B MSB 5-11-00 ■ m a 5-14-02 H H Si 9-16-99 E3 ■a 5-18-01 □□□□□□ s a 4-30-00 1BIIH m mm _6-29_-QQ - ms 6-18-01 m 6-15-02 ■ a 4-14-99 is m 7-11-01 s 5-9-04 mam 6-29-01 E H a 6-14-04 a m 7-7-04 m ■■ 7-20-01 mm □□□□□□ 7-10-02 Si ■ 7-18-04 )□( ■ 6-26-99 :iC 6-8-01 HI! ■ 5-19-00 ■ ■ EI El 6-3-03 m 7-11-00 Bun ■■ BBH 5-21-03 Sill 5-19-04 Ei 5-8-01 ■■□□□□ 4-24-04 ■ ■■■ IB 6-14-03 ■■■■Ul □□□□□□ 6-27-02 ■■ 5-13-03 mm □□□□□□ 5-30-04 ■i ss 5-4-03 mm 7-17-03 9IBB 6-24-04 ■ ■■■ Figure 4 Two-way cluster analysis of egg densities by sampling date. The taxa clusters are oriented vertically at the top of the figure and sampling date clusters are oriented horizontally at the left side of the figure. The dashed lines delineate sampling dates that contained similar assemblages of eggs. The color of the boxes denotes relative abundance of each taxon during each sampling period as shown in the matrix coding legend. The taxa names are abbreviated as: Cith. spp. ( Cithariclithys spp.), E. mord. ( Engraulis mordax ), Bathyl (Bathylagidae), Pleuron. ( Pleuronectidae ), G. zach. (Glyptocephalus zachirus), L. exil. (Lyopsetta exilis), M. pad. (Microstomus pacificus), P. vetu. ( Paroplirys vetulus), Seb. spp. ( Sebas - tolobus spp.), P. decu. (Pleuronichthys decurrens), P. rnela. (Psetticthys melanostictus), and T. symm. (Tracliurus symmetricus). “Information remaining” is a scaled measure of the amount of information that is left after stations are grouped. 168 Fishery Bulletin 106(2) Larvae Matrix coding o L_ Information remaining (%) 25 50 75 I I 1 1 1 100 A B C 1 -5-99 ■ ■ ■□□□□□[ 4-30-00 ]□□■□□□[ 5-21-03 mum 4-24-04 ■ ■■ 5-1-99 4-24-03 ■ ■ ■ 2-13-01 z 5-8-01 ■ 5-5-99 ■■■ 7-1 0-02 a 12-6-99 ■ 5-1 1 -00 ■ 7-11-01 □□□■□□□□□□□a 7-31-01 □□□■: : ! 1! )□□□□ 7-17-03 m m 3-11-99 ■■ m 5-28-99 ■ * 3-21-00 7-11-00 ■ ■ 4-14-99 ■ !□■□■■■□ 5-29-01 ]□■□□□□□ 6-3-03 ■1 • ■ 7-23-02 S 6-25-03 ■ 7-9-03 ■ 6-26-99 m :jb 6-27-02 ■ 5-9-04 ■ 5-4-03 ■ m 6-14-03 m mum 5-19-04 9 SSI 6-8-01 6-15-02 □□□■□[ 5-13-03 ■ ■□□□□□ 6-18-01 ■ B B ■ 5-30-04 BB 6-14-04 ■ m 6-24-04 m m Figure 5 Two-way cluster analysis of larval densities by sampling date. The taxa clusters are oriented vertically at the top of the figure and sampling dates are oriented horizontally at the left side of the figure. The dashed lines delineate sampling dates that contained similar assemblages of larvae. The color of the boxes denotes relative abundance of each taxon during each sampling period as shown in the matrix coding legend. The taxa names are abbreviated as: A. hexa. ( Ammodytes hexapterus), G. zach. (Glyptocephalus zachirus), P. decu. (Pleuronichtliys decurrens), E. mord. ( Engraulis rrwrdax), Seb. spp. ( Sebastes spp.), M. pad. ( Microstomus paci- ficus), P. mela. ( Psetticthys melanostictus) , P. vetu. ( Parophrys vetulus), I. isol. ( Isopsetta isolepis), M. prox. (Microgadus proximus), Pleuron. (Pleu- ronectidae), and L. exil. ( Lyopsetta exilis ). “Information remaining” is a scaled measure of the amount of information that is left after stations are grouped.” larval taxa did not differ between the two stations, and no indicator species were identified for either station. Year of sampling was a significant factor for eggs; 2003 and 2004 showed a differ- ent composition from most other years. However, eggs of only 3 of the 12 spe- cies (curlfin sole and jack mackerel for 2000; northern anchovy for 2003) were determined to be useful indicator spe- cies. Season was not a significant fac- tor overall for eggs, or for the two taxa considered indicators of downwelling and upwelling conditions. For larvae, year was a marginally insignificant factor overall, although there were pairs of years that were significantly different. Season was a significant fac- tor for larvae: Dover sole and English sole larvae were indicative of winter downwelling, and northern anchovy were indicative of summer upwelling seasons. Station was not a significant factor for eggs or larvae. Discussion The timing of upwelling (spring transi- tion), and its associated ocean tempera- tures, appears to be very important in determining the structure of the ich- thyofaunal community off the Colum- bia River and probably within the California Current. Upwelling is what brings nitrates into the Columbia River plume (Lohan and Bruland, 2006) and the California Current and ultimately determines primary and probably sec- ondary production (Hickey and Banas, 2003). As such, it is not surprising that fishes within this region have evolved life histories adapted to its presence (Bowen and Grant, 1997; Ware and Thomson, 1991). However, although upwelling drives the nutrient enrich- ment processes it also can disrupt the abilities of larval coastal fishes to suc- cessfully recruit if they are unable to take advantage of transport and reten- tion processes (Bakun, 1996). Several factors account for the high abundance of northern anchovy eggs and larvae captured during our study. The Columbia River plume is an impor- tant spawning habitat for northern an- chovy (Richardson, 1981; Emmett et al., 1997) and most samples were collected during the typical northern anchovy spawning season (May- August). Fur- Parnel et al.: Effects of oceanic conditions on ichthyoplankton communities in the Columbia River plume 169 Table 3 Average annual 3-m salinity, temperature, and chlorophyll a values at two stations off the Columbia River, Oregon. Also shown is average spring (April-June) Columbia River flows and day of the spring transition. The day of the spring transition (when upwelling begins) was identified with the methods of Loggerwell et ah, 2003. Only temperature and Columbia River flows were found to have significant annual differences (ANOVA, P<0.05). Year Salinity Temperature (°C) Chlorophyll a (pg/L) Columbia River flows (m3/s) Day of spring transition 1999 28.53 11.51 4.81 9999 1 Apr 2000 29.92 12.61 2.34 8132 12 Mar 2001 29.19 13.28 2.66 4506 2 Mar 2002 27.93 12.61 6.12 8640 21 Mar 2003 26.68 12.93 5.61 8316 22 Apr 2004 27.50 14.09 5.07 6777 19 Apr Table 4 Results of the multi-response permutation procedure (MRPP) and indicator species analysis (ISA) for annual, seasonal (before and after the spring transition [Logerwell et al., 2003]), and station differences in composition of the dominant egg and larval ichthyoplankton taxa. Shown are the MRPP A-statistic, overall significance value, and significant pair-wise comparisons that emerged. The significant indicator species are listed with the year or season with which each species is associated in parentheses. Stage Variable MRPP A-statistic P value Significant differences between levels Significant indicator species Eggs Year 0.1253 <0.001 2003 * all years but 2004 2004 * all years but 2002, 2003 Pleuronichthys decurrens, Trachurus symmetricus (2000), Engraulis mordax (2003) Season 0.0231 0.087 None Psetticthys melanostictus, Microstomus pacificus (downwelling), E. mordax, Citharichthys spp. (upwelling) Station 0.0019 0.631 None None Larvae Year 0.1027 0.068 2003 * 1999 and 2004 2002 * 2004 Parophrys vetulus, M. pacificus (downwelling), E. mordax (2004) Season 0.0787 0.018 downwelling * upwelling E. mordax (upwelling) Station 0.0003 0.412 None None thermore, northern anchovy eggs are positively buoyant and appear to concentrate in plume fronts (Morgan et ah, 2005), which often occurred in our sampling region. The observed annual increase in northern anchovy egg densities appear to be strongly linked to the increase in the abundance of adult northern anchovy population in the study area (Emmett et ah, 2006). Anchovy egg densities also may have been influenced by Columbia River spring flows. For example, densities of northern anchovy egg and larvae were lower in 2001, a drought year during which Columbia River flows were very low and the plume area was relatively small (Brodeur et ah, 2005). However, numbers of northern anchovy eggs were also down in 2002, which was not a drought year, but was the year that the Pacific Decadal Oscillation (PDO) changed from negative to positive (Goericke et ah, 2005). Richardson (1981) found that the peak of northern anchovy spawning in 1975 and 1976 occurred in July, whereas we found that the peak of spawning occurred in May during five of the six study years. However, this difference could have been due to Richardson (1981) not sampling in May and June. Brodeur et al. (1985) found northern anchovy spawning during April 1983, but 1983 was a strong El Nino year and therefore there were unusually warm ocean conditions. Jones et al. (1990) found northern anchovy eggs and larvae in the Columbia River estuary from April to September 1980, 170 Fishery Bulletin 106(2) with a peak in June. We are not certain whether the timing of northern anchovy spawning has shifted to earlier months, whether this early spawning was missed by previous surveys, or whether we have two different age-groups spawning at different times. Nevertheless, our results indicate that timing of northern anchovy egg surveys is an important parameter to consider during surveys to estimate spawning biomass (Emmett et al., 1997) and feeding success relative to plankton produc- tion cycles (Bollens et al., 1992). Jack mackerel, which generally spawn off southern California (MacCall and Stauffer, 1983), are not com- monly collected as juveniles very far north in the Cali- fornia Current. Because jack mackerel eggs hatch in 4.3 days at 12°C (Farris, 1961) and because we observed jack mackerel larvae, it appeared that jack mackerel spawned in the general vicinity of our study area. Jack mackerel spawning has been reported off Washington State and southern Oregon (Ahlstrom, 1956), but not for many years. Ahlstrom (1956) noted that there is a northward progression of the spawning season in ar- eas north of Point Conception, California. During our study period, it was possible that adult jack mackerel moved north and into cooler waters inshore to feed and spawn. The recent capture of age-0 jack mackerel during pelagic fish surveys off Oregon and Washington (Brodeur et al., 2006) indicates that jack mackerel have recently spawned and recruited successfully off the Pacific Northwest coast. The occurrence of rare taxa, such as opah, can be attributed to the long incubation time (up to 3 weeks) exhibited by most lampriform fish eggs (Olney, 1984). In this time, an egg spawned in the open ocean, the normal habitat of this species, could drift inshore to our sampling area. Similarly, Pacific viperfish eggs, released in the open ocean, can drift inshore. Little is known about their egg incubation period. The occurrence of several offshore ichthyoplankton species in our study area may be related to the lo- cal topography. The Astoria Canyon lies directly sea- ward from the mouth of the Columbia River. During the upwelling season, the canyon causes currents to flow landward (Hickey, 1997), thus carrying normally offshore organisms closer to shore. Bosley et al. (2004) concluded that currents over the Astoria Canyon concen- trate oceanic organisms and transport them shoreward and these actions may explain the occurrences of opah. Pacific viperfish, and jack mackerel eggs, and other offshore species (Richardson and Pearcy, 1977) in our coastal sampling. Egg and larval composition at the two sampling sta- tions did not differ significantly, probably because they were relatively close and thus had similar environmen- tal conditions. Out of a total of five physical variables tested, only salinity was found to be significantly differ- ent between the two stations. Surface salinity is highly variable in this region because of the proximity of the Columbia River plume. While the plankton tow depths at each station were the same (40 m), we would have missed eggs and or larvae occurring below 40 m at the station farther offshore (buoy 1), which was in substan- tially deeper water. Boehlert et al. (1985) and Auth and Brodeur (2006) found fish larvae at depths greater than 40 m off the coast of Oregon, although the majority of fish larvae were in surface waters. The cluster dendrograms were dominated by flatfish species because of the high number of pleuronectid species found along the Oregon coast, as well as the tendency of most of these species to have pelagic eggs. Overall patterns in the egg and larval clusters changed seasonally, and many of the samples collected during the downwelling or upwelling seasons were clustered together. MRPP analysis verified that season was an significant contributor to these cluster groups. The lar- val indicator species for downwelling conditions were English sole and Dover sole, which are typically win- ter spawners (although some spawning may occur all year). The indicator species for the upwelling season was northern anchovy, which spawns primarily in the spring and summer. In general, benthic and nearshore species spawn in winter, when larvae are less likely to be transported offshore. In 2003, the PDO shifted from a cool (negative) to a warm (positive) phase. This change resulted in southern or offshore taxa becoming more abundant in 2003-04. However, ichthyoplankton taxa in 2003 and 2004 were also significantly different mainly because of substantial increases in eggs and larvae of northern anchovy, a taxon that is not southern affiliated. The increase in northern anchovy eggs and larvae appeared linked to an increase in the abundance of the adult population. Large-scale climate variability (as observed dur- ing our study period) can cause large changes in fish populations (Cushing, 1982; Beamish, 1993; Francis et al., 1998; Chavez et al., 2003). Pearcy (2002) and Brodeur et al. (2003, 2006) found that abnormal ocean conditions alter the ichthyofauna in the California Cur- rent region. Changing climate conditions can alter currents that advect fish eggs and larvae to or away from nursery areas, thus affecting recruitment. Al- tered currents may also lower the density and change the species composition of planktonic food organisms (Peterson and Schwing, 2003) and thus inhibit lar- val fish from finding adequate prey resources. In late 2002, the PDO (Mantua et al., 1997) became positive (with warm conditions) after four years of being nega- tive (with cold conditions) (Goericke et al., 2005) and caused warmer ocean temperatures, increased diversity of warm-water copepods, and decreased cold-water co- pepods in the California Current (Hooff and Peterson, 2006). Changes in ocean temperatures can also cause shifts in the locations or timing of spawning and affect developmental durations through the early life stages of fishes (Sabates et al., 2006; Phillips et al., 2007). Faster development through the egg and yolk-sac stages may help to minimize the time eggs are vulnerable to invertebrate predation (Bailey, 1981). In the California Current region, upwelling is the dominant feature that influences primary production, and therefore any cli- mate-induced change that affects upwelling or Califor- Parnel et al.: Effects of oceanic conditions on ichthyoplankton communities in the Columbia River plume 171 nia Current circulation patterns will alter zooplankton prey that larval fish feed on. Climate-induced change in upwelling patterns, sea surface temperatures, and perhaps Columbia River flows will have large effects on the larval fish within our study area and in the broader California Current region. We examined interannual variation in the abundance of fish eggs and larvae in the Columbia River plume and have shown that the ichthyoplankton community changes in relation to the dynamic oceanographic pro- cesses of this region. However, research over a greater area and on a longer time scale is needed to determine the effects of these processes on the ichthyoplankton community, as well as to determine how recruitment of marine fish is being affected by regional and larger- scale variability in the marine environment. Acknowledgments We thank P. Bentley and G. Krutzikowsky for field assistance, T. Auth and M. Busby for assistance in identifying eggs and larvae, and T. Auth, S. Bollens, and M. Busby for comments on an earlier version of the manuscript. Funding was provided by the Bonneville Power Administration and National Oceanic Atmospheric Administration’s (NOAA) Fisheries and the Environment Initiative. The lead author was supported in part by an appointment to the NOAA’s research participation program administered by the Oak Ridge Institute for Science and Education. Literature cited Ahlstrom, E. H. 1956. Eggs and larvae of anchovy, jack mackerel, and Pacific mackerel. Fish. Bull. 5:33-42. Anderson, G. C. 1972. Aspects of marine phytoplankton studies near the Columbia River, with special reference to a subsurface chlorophyll maximum. In The Columbia River estu- ary and adjacent ocean waters (A. T. Pruter, and D. L. Alverson, eds.), p. 219-240. Univ. Washington Press, Seattle, WA. Auth, T. D., and R. D. Brodeur. 2006. Distribution and community structure of ichthyo- plankton off the Oregon Coast in 2000 and 2002. Mar. Ecol. Prog. Ser. 319:199-213. Bailey, K. M. 1981. Larval transport and recruitment of Pacific hake Merluccius productus. Mar. Ecol. Prog. Ser. 6:1-9. Bakun, A. 1996. Patterns in the ocean: Ocean processes and marine population dynamics, 323 p. California Sea Grant, San Diego, CA. Barnes, C. A., A. C. Duxbury, and B. A. Morse. 1972. Circulation and selected properties of the Columbia River effluent at sea. In The Columbia River estu- ary and adjacent ocean waters (A. T. Pruter, and D. L. Alverson, eds.), p. 41-80. Univ. Washington Press, Seattle, WA. Beamish, R. J., ed. 1993. Climate and exceptional fish production off the West Coast of North America. Can. J. Fish. Aquat. Sci. 50:2270-2291. Boehlert, G. W., D. M. Gadomski, and B. C. Mundy. 1985. Vertical distribution of ichthyoplankton off the Oregon coast in spring and summer months. Fish. Bull. 83:611-620. Bollens, S. M., B. W. Frost, H. R. Schwaninger, C. S. Davis, K. L. 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. Bosley, K. L„ J. W. Lavelle, R. D. Brodeur, W. W. Wakefield, R. L. Emmett, E. T. Baker, and K. M. Rehmke. 2004. Biological and physical processes in and around Astoria submarine canyon, Oregon, USA. J. Mar. Syst. 50:21-37. Bowen, B. W., and W. S. Grant. 1997. Phylogeography of the sardines ( Sardinops spp.): Assessing biogeographic models and population histories in temperate upwelling zones. Evolution 51:1601-1610. Brodeur, R. D., D. M. Gadomski, W. G. Pearcy, H. P. Batchelder, and C. B. Miller. 1985. Abundance and distribution of ichthyoplankton in the upwelling zone off Oregon during anomalous El Nino conditions. Estuarine Coastal Shelf Sci. 21:365-378. Brodeur, R. D., J. P. Fisher, R. L. Emmett, C. A. Morgan, and E. Casillas. 2005. Species composition and community structure of pelagic nekton off Oregon and Washington under vari- able oceanographic conditions. Mar. Ecol. Prog. Ser. 298:41-57. Brodeur, R. D., W. G. Pearcy, and S. Ralston. 2003. Abundance and distribution patterns of nekton and micronekton in the northern California Current transition zone. J. Oceanogr. 59:515-535. Brodeur, R. D., S. Ralston, R. L. Emmett, M. Trudel, T. D. Auth, and A. J. Phillips. 2006. Recent trends and anomalies in pelagic nekton abundance, distribution, and apparent recruitment in the Northeast Pacific Ocean. Geophys. Res. Let. 33, L22S08, doi:10.1029/2006GL026614. Chavez, F. P, J. Ryan, S. E. Lluch-Cota, and M. C. Niquen. 2003. From anchovies to sardines and back: multidecadal change in the Pacific Ocean. Science 299:217-221. Collis, K., D. D. Roby, D. P. Craig, S. Adamany, J. Y Adkins, and D. E. Lyons. 2002. Colony size and diet composition of piscivorous waterbirds on the lower Columbia River: Implications for losses of juvenile salmonids to avian predation. Trans. Am. Fish. Soc. 131:537-550. Cowan, J. H., Jr., and R. F. Shaw. 2002. Recruitment. In Fishery science: The unique contributions of early life stages (L. A. Fuiman, and R. G. Werner, eds.) p. 88-111. Blackwell Publishing, Oxford, England. Cushing, D. H. 1982. Climate and fisheries, 373 p. Academic Press, London, U.K. DeRobertis, A., C. A. Morgan, R. A. Schabetsberger, R. Zabel, R. D. Brodeur, R. L. Emmett, C. M. Knight, G. K. Krut- zikowsky, and E. Casillas. 2005. Frontal regions of the Columbia River plume: II. 172 Fishery Bulletin 106(2) Distribution, abundance and feeding ecology of juvenile salmon. Mar. Ecol. Prog. Ser. 299:33-44. Doyle, M. J. 1995. The El Nino of 1983 as reflected in the ich- thyoplankton off Washington, Oregon, and north- ern California. Can. Spec. Publ. Fish. Aquat. Sci. 121:161-180. Doyle, M. J., K. L. Mier, M. S. Busby, and R. D. Brodeur. 2002. Regional variations in springtime ichthyoplankton assemblages in the Northeast Pacific Ocean. Prog. Oceanogr. 53:271-281. Doyle, M. J., W. W. Morse, and A. W. Kendall Jr. 1993. A comparison of larval fish assemblages in the temperate zone of the Northeast Pacific and Northwest Atlantic Oceans. Bull. Mar. Sci. 53:588-644. Emmett, R. L., P. J. Bentley, and M. H. Schiewe. 1997. Abundance and distribution of northern anchovy eggs and larvae ( Engraulis mordax) off the Oregon coast, mid 1970’s vs. 1994 and 1995. In Forage fishes in marine ecosystems, p. 505-508. Alaska Sea Grant College Program, Fairbanks, AK. Emmett, R. L., R. D. Brodeur, T. W. Miller, S. S. Pool, P. J. Bent- ley, G. K. Krutzikowsky, and J. McCrae. 2005. Pacific sardine ( Sardinops sagax) abundance, distribution and ecological relationships in the Pacific Northwest. Calif. Coop. Oceanic Fish. Invest. Rep. 46:122-143. Emmett, R. L., G. K. Krutzikowsky, and P. J. Bentley. 2006. Abundance and distribution of pelagic piscivorous fishes in the Columbia River plume during spring/early summer 1998-2003: relationship to oceanographic con- ditions, forage fishes, and juvenile salmonids. Prog. Oceanogr. 68:1-26. Emmett, R. L., R. Llanso, J. Newton, R. Thom, M. Hornberger, C. Morgan, C. Levings, A. Copping, and P. Fishman. 2000. Geographic signatures of North American west coast estuaries. Estuaries 26(61:765-792. Farris, D. A. 1961. Abundance and distribution of eggs and larvae and survival of jack mackerel (Trachurus symmetricus). Fish. Bull. 61:242-279. Francis, R. C., S. R. Hare, A. B. Hollowed, and W. S. Wooster. 1998. Effects of interdecadal climate variability on the ocean ecosystem of the NE Pacific. Fish. Oceanogr. 7:1-21. Goericke, R. E., E. Venrick, A. Mantyla, S. J. Bograd, F. B. Schwing, A. Huyer, R. L. Smith, P. A. Wheeler, R. Hooff, W. T. Peterson, F. Chavez, C. Collins, B. Marinovic, N. Lo, G. Gaxi- ola-Castro, R. Durazo, K. D. Hyrenbach, and W. J. Sydeman. 2005. The state of the California Current, 2004-2005: still cool? Calif. Coop. Oceanic Fish. Invest. Rep. 46:32-71. Govoni, J. J. 1997. The association of the population recruitment of gulf menhaden, Brevoortia patronus, with Mississippi River discharge. J. Mar. Syst. 12:101-108. Govoni, J. J., and C. B. Grimes. 1992. The surface accumulation of larval fishes by hydrodynamic convergence within the Mississippi plume front. Cont. Shelf Res. 122:1265-1276. Grimes, C. B., and J. H. Funucane. 1991. Spatial distribution and abundance of larval and juvenile fish, chlorophyll and macrozooplankton around the Mississippi River discharge plume, and the role of the plume in fish recruitment. Mar. Ecol. Prog. Ser. 75:109-119. Hickey, B. M. 1997. The response of a steep-sided, narrow canyon to time-variable wind forcing. J. Phys. Oceanogr. 27:697-726. Hickey, B. M., and N. S. Banas. 2003. Oceanography of the U.S. Northwest Pacific coast and estuaries with application to coastal ecology. Es- tuaries 26:1010-1031. Hooff, R. C., and W. T. Peterson. 2006. Copepod biodiversity as an indicator of changes in ocean and climate conditions of the northern Cal- ifornia Current ecosystem. Limnol. Oceanogr. 51: 2607-2620. Hunter, J. R., and N. C.-H. Lo. 1993. Ichthyoplankton methods for estimating fish bio- mass: Introduction and terminology. Bull. Mar. Sci. 53:723-727. Jones, K. K., C. A. Simenstad, D. L. Higley, and D. L. Bottom. 1990. Community structure, distribution, and standing stock of benthos, epibenthos, and plankton in the Colum- bia River estuary. Prog. Oceanogr. 25:211-241. Krebs, C. J. 1989. Ecological methodology, 654 p. Harper Collins Pubis., New York, NY. Kruskal, J. B. 1964. Nonmetric multidimensional scaling: a numerical method. Psychomet. 29:1-27. Lasker, R., ed. 1985. An egg production method for estimating spawn- ing biomass of pelagic fish: application to the northern anchovy ( Engraulis mordax ). NOAA Tech. Rep. NMFS 36, 99 p. Logerwell, E. A., N. Mantua, P. Lawson, R. C. Francis, and V. Agostini. 2003. Tracking environmental processes in the coastal zone for understanding and predicting Oregon coho (Oncorhynchus kisutch ) marine survival. Fish. Ocean- ogr. 12:554-568. Lohan, M. C., and K. W. Bruland. 2006. Importance of vertical mixing for additional sources of nitrate and iron to surface waters of the Columbia River plume: Implications for biology. Mar. Chem. 98:260-273. MacCall, A. ■).. and G. D. Stauffer. 1983. Biology and fishery potential of jack mackerel (Trachurus symmetricus ). Calif. Coop. Oceanic Fish. Invest. Rep. 24:46-56. Mantua, N. J., S. R. Hare, Y. Zhang, J. M. Wallace, and R. C. Francis. 1997. A Pacific interdecadal climate oscillation with impacts on salmon production. J. Am. Meterol. Soc. 78:1069-1079. 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. Mather, P. M. 1976. Computational methods of multivariate analy- sis in physical geography, 532 p. J. Wiley and Sons, London, U.K. Morgan, C. A., A. DeRobertis, and R. W. Zabel. 2005. Columbia River plume fronts. I. Hydrography, zoo- plankton distribution, and community composition. Mar. Ecol. Prog. Ser. 299:19-31. Moser, H. G., ed. 1996. The early stages of fishes in the California cur- Parnel et al. : Effects of oceanic conditions on ichthyoplankton communities in the Columbia River plume 173 rent region. Calif. Coop. Oceanic Fish. Invest. Atlas no. 33, 1505 p. NMFS (National Marine Fisheries Service). 1997. Investigation of scientific information on the impacts of California sea lions and Pacific harbor seals on salmonids and on the coastal ecosystems of Wash- ington, Oregon, and California. NOAA Tech. Memo. NMFS-NWFSC-28, 172 p. Olney, J. E. 1984. Lampriformes: development and relationships. In 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), p. 368-379. Am. Soc. Ichthyol. Herpetol. Pearcy, W. G. 2002. Marine nekton off Oregon and the 1997-98 El Nino. Prog. Oceanogr. 54:398-403. Peterson, W. T., and F. B. Schwing. 2003. A new climate regime in Northeast Pacific ecosystems. Geophys. Res. Let. 30(17), 1896, doi:10.1029/2003GL0157528. Phillips, A. J., S. Ralston, R. D. Brodeur, T. D. Auth, R. L. Emmett, C. Johnson, and V. G. Wespestad. 2007. Recent pre-recruit Pacific hake ( Merluccius productus ) occurrences in the northern California Current suggest a northward expansion of their spawn- ing area. Calif. Coop. Oceanic Fish. Invest. Rep. 48:215-229. Richardson, S. L. 1981. Spawning biomass and early life of northern anchovy, Engraulis mordax, in the northern sub- population off Oregon and Washington. Fish. Bull. 78:855-876. Richardson, S. L., and W. G. Pearcy. 1977. Coastal and oceanic fish larvae in an area of upwelling off Yaquina Bay, Oregon. Fish. Bull. 75: 125-145. Sabates, A., P. Martin, J. Lloret, and V. Raya. 2006. Sea warming and fish distribution: the case of the small pelagic fish, Sardinella aurita, in the western Mediterranean. Global Change Biol. 12:2209-2219. Schabetsberger, R., C. A. Morgan, R. D. Brodeur, C. L. Potts, W. T. Peterson, and R. L. Emmett. 2003. Prey selectivity and diel feeding chronology of juvenile Chinook (Oncorhynchus tshawytscha) and coho (O. kisutch) salmon in the Columbia River plume. Fish. Oceanogr. 12:523-540. Shannon, C. E., and W. Weaver. 1949. The mathematical theory of communication, 117 p. Univ. Illinois Press, Urbana, IL. Sherwood, C. R., D. A. Jay, R. B. Harvey, P. Hamilton, and C. A. Simenstad. 1990. Historical changes in the Columbia River estuary. Prog. Oceanogr. 25:299-352. Waldron, K. D. 1972. Fish larvae collected from the northeastern Pacific Ocean and Puget Sound during April and May of 1967. NOAA Tech. Rep. NMFS SSRF-663, 16 p. Ware, D. M., and R. E. Thomson. 1991. Link between long-term variability in upwelling and fish production in the northeast Pacific Ocean. Can. J. Fish. Aquat. Sci. 48:2296-2306. 174 Abstract — -The migratory population of striped bass ( Morone saxatilis ) (>400 mm total length [TL]) spends winter in the Atlantic Ocean off the Virginia and North Carolina coasts of the United States. Information on tro- phic dynamics for these large adults during winter is limited. Feeding habits and prey were described from stomach contents of 1154 striped bass ranging from 373 to 1250 mm TL, collected from trawls during winters of 1994-96, 2000, and 2002-03, and from the recreational fishery during 2005-07. Nineteen prey species were present in the diet. Overall, Atlan- tic menhaden ( Brevoortia tyrannus) and bay anchovy ( Anchoa mitchilli) dominated the diet by biomass (67.9%) and numerically (68.6%). The per- cent biomass of Atlantic menhaden consumed increased from 50.3% during 1994-2003 to 87.0% during 2005-07. Demersal fish species such as Atlantic croaker ( Micropogonias undulatus) and spot ( Leiostomus xan- thurus) represented <15% of the diet biomass, whereas alosines (Alosa spp.) were rarely observed. Invertebrates were least important, contributing <1.0% by biomass and numerically. Striped bass are capable of feeding on a wide range of prey sizes (2% to 43% of their total length). This study outlines the importance of clupeoid fishes to striped bass winter produc- tion and also shows that predation may be exerting pressure on one of their dominant prey, the Atlantic menhaden. Manuscript submitted 3 December 2007. Manuscript accepted 6 February 2008. Fish. Bull. 106:174-182 (2008). The views and opinions expressed or implied in this article are those of the author and do not necessarily reflect the position of the National Marine Fisheries Service, NOAA. Interactions between adult migratory striped bass ( Morone saxatilis) and their prey during winter off the Virginia and North Carolina Atlantic coast from 1994 through 2007 Anthony S. Overton (contact author) Department of Biology-Howell Science Complex Harriot College of Arts and Science East Carolina University Greenville, North Carolina 27858 Email address: overtona@ecu.edu Charles S. Manooch III 2900 Dogwood Lane Morehead City North Carolina 28557 Joseph W. Smith Kenneth Brennan Center for Coastal Fisheries and Habitat Research National Marine Fisheries Service, NOAA 101 Pivers Island Road Beaufort, North Carolina 28516 Striped bass ( Morone saxatilis) has a well documented history along the U.S. east coast, from its dramatic population declines during the 1980s to its subsequent recovery by the early 1990s (Field, 1997; Richards and Rago, 1999). Successful management efforts have resulted in a greater than tenfold increase in striped bass abun- dance between the 1980s and 1990s and a subsequent increase in popu- lation-level prey consumption (Hart- man, 2003), and therefore a concern for coastal populations of prey spe- cies (Hartman, 2003; Overton, 2003; Uphoff, 2003). Under current man- agement regimes, it may be difficult to maintain high population levels of striped bass and their prey (Hartman, 2003; Uphoff, 2003). Typically, striped bass along the U.S. east coast spend their first years maturing in their natal estuaries and then emigrate to the Atlantic Ocean. Most striped bass along the Atlan- tic coast migrate northward during spring and summer to waters off the northeast coast of the U.S. During fall and winter, they return south to overwinter off the coasts of Virginia and North Carolina (Boreman and Lewis, 1997). These migratory fish are generally large (>500 total length [TL] mm) and feed prodigiously dur- ing their migrations. Theoretically, these large piscivores are capable of structuring prey fish populations through predation and prey selection (Bax, 1998; Harvey et al., 2003), and in turn they can potentially influence the recruitment success of prey spe- cies. Predators such as striped bass are capable of consuming prey that are a wide range of sizes (Hartman, 2000); therefore to understand tro- phic relationships it is important to examine their dietary habits. Multi- species fisheries and ecosystem man- agement approaches require dietary information for upper-level predators such as striped bass (Latour et al., 2003). The literature on diets and feeding habits of striped bass (see Walter et al., 2003) is voluminous. However, information on feeding habits during Overton et al.: Interactions between Morone saxatilis and their prey during winter off the North Carolin coast 175 81 WW 80WW 79"0'0"W 78WW 77*0'0"W 76"0'0"W 75WW Figure t Map of U.S. east coast and the general sampling area (shaded oval shape) for striped bass (Morone saxatilis) collected in trawl and rec- reational catch samples from 1994 through 2007. their oceanic migrations in winter is limited. Walter et al. (2003) iden- tified the paucity of information on the foraging habits of striped bass along the Atlantic coast during their winter residency as one of the ma- jor gaps in the life history of this species. This paucity of information about feeding habits during winter is especially acute, given the impor- tance of predator-prey interactions and their relation to the population base in the area. The objective of this paper was to gather and synthesize detailed information on annual feed- ing habits of striped bass during winter off the coasts of Virginia and North Carolina from 1994 through 2007. Therefore, we determined the important prey types and the prey- size spectrum of striped bass during winter. Materials and methods Striped bass were collected by two methods: trawls and dockside sam- pling of the catch of recreational fish- ermen. Beginning in 1988, a number of fisheries management agencies (National Marine Fisheries Service, United States Fish and Wildlife Ser- vice, North Carolina Department of Marine Fisheries, Maryland Depart- ment of Natural Resources, and Vir- ginia Marine Resource Commission) organized a trawl survey for striped bass from federal research vessels during winter off the coast of Virginia and North Carolina (Fig. 1). The primary objective of the survey was to tag and release striped bass to assess annual mortality of the coastal migratory stock. The trawl specification and trawl duration varied over the years. Generally, trawl sampling occurred around the clock during mid-January. Once the trawl was recovered, most striped bass were tagged and released; however, dead or moribund fish not selected for tagging were sac- rificed and processed for aging or food habit analyses. Few fish (n<19) were examined for stomach contents from 1997-99, 2001, and 2004-07; therefore data from these years were excluded from our analyses. From December to March 2004-07, striped bass were collected weekly at the Oregon Inlet Fishing Center (OIFC) in Manteo, North Carolina. Fish were sampled from recreational fishermen who brought their daily catches to the fish cleaning station at the OIFC. We randomly selected fish once they were cleaned (gener- ally filleted), measured the carcasses for total length (±1.0 mm), and identified the sex of each individual. Stomachs were removed and frozen for later analysis. Additional samples were obtained from Virginia Beach Fishing Center, Virginia Beach, Virginia and processed in the same manner as that used to process fish from the OIFC. All stomach samples presumably came from fish captured in the Atlantic Ocean 0-4.8 km from the shoreline because no fishing for striped bass is permit- ted in the U.S. Exclusive Economic Zone (EEZ), beyond 4.8 km (3 miles) from shore. In the laboratory, stomach contents were thawed and all prey items removed, sorted, identified (to the lowest taxon possible, usually to species for fish and decapod crustaceans and family for other invertebrates), enu- merated, weighed to the nearest 0.1 g, and measured to the nearest mm (standard, carapace, or total length). The percentage of prey by number and percent composi- tion by weight (wet weight-biomass) were determined. A quantitative assessment of number and weight of each prey item was used, as well as the respective percentage values for each (Markle and Grant, 1970; Macdonald and Green, 1983). Percent weight is a mea- sure of the nutritional value of the prey (Macdonald and Green, 1983) and is calculated as the total weight of each prey category divided by the total weight of all prey categories. Frequency distributions of prey 176 Fishery Bulletin 106(2) total length to predator total length (prey-to-predator ratios; PPR) were examined. We used one-way analysis of variance (ANOVA) to test for differences in mean length (Log10 transformed) between the fish from the trawl and recreational samples (P< 0.05). We also fitted a least squares linear regression of prey total length and striped bass total length. Results We collected 263 stomachs from striped bass in the trawl samples from 1994 through 2003 (Table 1); specimens ranged from 373 to 955 mm TL (mean=662.2 ±129.1 standard deviation (SD); Fig. 2). The percentage of stomachs that contained food ranged from 73.5% to 100% (mean = 84.6%). From the recreational samples (2005-07), 891 fish were examined (Table 1). The striped bass size ranged from 509 to 1250 mm TL (mean=918.9 ±93.8 SD; Fig. 2). The size of fish collected from the recreational catch were significantly larger (ANOVA, P=<0.0001; df =1) than those collected by trawl. The percentage of stomachs containing food items was more variable for striped bass caught by recreational anglers (23.6-80.7%), than for the fish caught by trawl (73.5-100.0%). In 2005, 23.6% of the stomachs con- tained food; increased to 24.4% in 2006, and to 80.7% by 2007. Collectively, 19 fish and invertebrate species constituted the diet of striped bass (Table 2), and fish predominated. Feeding habits (trawl samples 1994—2003) Atlantic menhaden and bay anchovy were the most abundant species present in striped bass stomachs in all years sampled; they also dominated the diet in biomass and numerically. Atlantic menhaden accounted 9.5% of the diet numerically and 50.3% by biomass (Fig. 3). The biomass of Atlantic menhaden was constant (40%) from 1994 to 2000; this contribution nearly doubled to 73.8% in 2002 and 72.4% by 2003. Atlantic menhaden showed no consistent pattern numerically and was gen- erally <15%. Concurrent with the increase in the biomass of At- lantic menhaden consumed was a decline in the percent biomass of bay anchovy found in the diet of striped bass. Bay anchovy accounted for 16.5% of the biomass to 29.9% of the diet numerically throughout the study period (Fig. 3). Between 1994 to 2000 mean percent biomass for bay anchovy was 43.4% and they repre- sented 71.3% of the diet numerically. However, by 2002, the percent biomass declined to 16.7% and represented <6.6% in 2003. Bay anchovy dominated the diet by number, representing 94.5% in 1995 and remained >80% from 1996 through 2002 (Fig. 3). Sciaenids and alosine species were minor contribu- tors to the diet of striped bass. Weakfish (Cynoscion regalis ) was absent from the diet before 2002 and rep- resented <0.5% of the diet during the study (Table 2). From 1994 to 2003, Atlantic croaker ( Micropogonias undulatus ) was absent, except in 1995 and 2003 when it represented 10.1% and 14.7% of the diet biomass, respectively. Alosines (American shad [A/osa sapidissima]; blueback herring [A. aestivalis]', and hickory shad [A. mediocris ]) were a minor part of the diet of striped bass, and they occurred only during 1994 and 1996. American shad were found once in 1996 (4.5% bio- mass, 6.8% numerically). Blue- back herring were present in 1994 and 1996 and represented 3.1% and 8.1% of the diet bio- mass, respectively. Invertebrates were a minor portion of the diet of large striped bass generally contributing <1.0% to the diet (Table 2). Feeding habits (determined from recreational catch samples 2005—07) In the recreational catches, At- lantic menhaden and bay anchovy dominated striped bass diet both by biomass (88.9%) and numeri- cally (93.6%). Biomass of Atlantic menhaden remained consistent 25 -i 20 S' 15 10 - 5 - I I Trawl ■■■ Recreational catch JZL 300 400 500 600 700 800 900 1000 1100 1200 1300 Total length (mm) Figure 2 Length-frequency histogram for striped bass ( Morone saxatilis) collected during the winter off the coasts of Virginia and North Carolina during 1994-2007. Overton et at: Interactions between Morone saxatiiis and their prey during winter off the North Carolm coast 177 Table 1 Mean size (total length |TL] mm ±standard deviation [SD ] ), size range, and number of striped bass ( Morone saxatiiis) with food in their stomachs collected off the coasts of North Carolina and Virginia from trawl and hook-and-line samples during 1994-2007. Collection year and Mean TL Size range Number of fish examined gear type (mm, ±SD) (TL mm) (% with food in the stomach) 1994, trawl 613.1 (72.3) 425-765 73(99) 1995, trawl 639.7 (57.0) 525-718 19 (100) 1996, trawl 805.9 (75.6) 666-955 34(74) 2000, trawl 561.3 (90.5) 465-770 50(84) 2002, trawl 616.2(180.0) 373-941 60(77) 2003, trawl 836.6(80.6) 745-953 19 (84) 2005, hook-and-line 881.2 (94.5) 509-1150 253 (23) 2006, hook-and-line 914.6(87.9) 720-1200 450(28) 2007, hook-and-line 994.2 (99.0) 760-1250 140(81) Table 2 Diet summary of prey contributions (%B=biomass, %N=number) and mean prey size (total length [TL] ±standard deviation [SD] ) of striped bass ( Morone saxatiiis ) during the winter off the coast of North Carolina from 1994 through 2007. Asterisk rep- resents only one prey item. Year 1994 1995 1996 2000 2002 Fish Scientific name %B %N %B %N %B %N %B %N %B %N American eel Anguilla rostrata American shad Alosa sapidissima Atlantic croaker Micropogonias undulatus Atlantic herring Clupea harengus Atlantic menhaden Brevoortia tyrannus 38.8 10.6 Bay anchovy Anchoa mitchilli 34.3 55.2 Blueback herring Alosa aestivalis 3.1 6.1 Butterfish Peprilus triacanthus Hickory shad Alosa mediocris Round herring Etrumeus teres Sciaenid sp. 14.1 13.7 Silver perch Bairdiella chrysoura Spot Leiostomus xanthurus Tonguefishes Symphurus sp. Unknown clupeid 1.6 2.1 Fish remains 1.5 2.7 Weakfish Cynoscion regalis Menticirrhus spp. Invertebrates Gastropod shell Gastropod Decapods Decapoda 2.1 2.2 Sand shrimp Crangon septimspinosa 2.8 4.7 Mud crab Panopeus herbstii Longfin squid Loligo pealeii 10.1 0.9 4.1 0.4 29.9 4.5 30.5 2.2 31.8 13.7 73.8 7.3 60.1 94.5 40.4 85.2 58.9 85.8 16.7 79.4 8.1 0.5 1 0.4 16.2 4.8 0.3 0.6 9.3 0.5 0.3 0.7 1.4 0.2 0.3 0.2 2.4 11.4 continued 178 Fishery Bulletin 106(2) Table 2 (continued) Fish Scientific name 2003 2005 Year 2006 2007 Overall Mean TL mm (±SD) %B %N %B %N %B %N %B %N %B %N American eel Anguilla rostrata 0.3 0.7 0.8 0.1 1.4 <0.1 334(62) American shad Alosa sapidissima 0.4 0.5 154(42) Atlantic croaker Micropogonias undulatus 14.7 1.5 0.8 1.4 5 1.3 1.7 0.6 3.4 0.4 126 (51) Atlantic herring Clupea harengus 7.3 10.4 1.4 0.2 217 (24) Atlantic menhaden Brevoortia tyrannus 72.4 17.8 82.8 60.4 71.1 4.9 94.5 81.8 67.9 17.3 183(73) Bay anchovy Anchoa mitchilli 6.7 63.4 0.9 11.8 11.9 92.1 0.2 12.6 16.5 68.6 55.9(13) Blueback herring Alosa aestivalis 1.1 1.7 99.1(67) Butterfish Peprilus triacanthus 3.1 2.8 0.5 3.41 0.3 0.4 179(35) Hickory shad Alosa mediocris 5.4 0.1 0.3 <0.1 442* Round herring Etrumeus teres 3.5 0.5 0.5 0.5 135(9) Sciaenid sp. 1.9 3.7 85.9 (28) Silver perch Bairdiella chrysoura 0.6 2.1 0 0 0.1 0.1 112* Spot Leiostomus xanthurus 0.5 0.7 2 0.6 <0.1 0.1 0.1 0.3 122(14) Tonguefishes Symphurus sp. 5.9 14.9 0.5 0.4 Unknown clupeid 1.1 0.9 391(1.4) Fish remains 0.4 2.5 0.1 2.1 0.2 0.9 1 0.9 1.5 1.2 Weakfish Cynoscion regalis 3.6 6.3 0.4 0 1.1 0.3 0.7 0.2 159(15) Menticirrhus spp 0.4 0.1 0.3 0.1 0.2 0.1 137 (8.3) Invertebrates Gastropod shell Gastropod 0.2 0.7 Decapods Decapoda 0.3 0.6 Sand shrimp Crangon septimspinosa 0.4 1.3 Mud crab Panopeus herbstii <0.1 0.7 0.1 0.1 <0.1 <0.1 Longfin squid Loligo pealeii <0.1 <0.1 <0.1 <0.1 (>70%) from 2005 through 2007 and was highest in 2007 (94.5%). Contribution of Atlantic menhaden by number, however, was quite variable: in 2005 they contributed 60.4%, declining to <5.0% in 2006, and increasing to 81.1% in 2007. Percent by number of bay anchovy was variable during 2005-07. In 2005 and 2007 their contribution by number was <15%. However, in 2006 they contributed >90% to the diet numerically. Weakfish and Atlantic croaker were present in stom- achs of recreationally caught striped bass for all years (Table 2). There were no clear patterns to their contri- butions, and collectively represented a small portion <5.0% (biomass) of the diet. Only one alosine, a hickory shad (2006), was found in the samples (2.7% biomass, 0.1% number). Other clupeoids, including Atlantic her- ring (7.3% of the biomass in 2005) and round herring ( Etrumeus teres) (3.5% of the biomass in 2006), were found sporadically in the diet. Invertebrates were un- important to the diet of large striped bass, generally representing <1.0% by weight and by number in the diet (Table 2). Predator-prey size relationships Prey consumed by striped bass ranged from 35 to 423 mm TL (mean=102.5 mm ±79.3 SD), although most prey items (86.7%) were less than 125.0 mm long. Mean length of Atlantic menhaden consumed was 204.2 TL mm (±76.2 mm SD). Prey length showed a significantly positive relationship with striped bass total length (P<0.001, r2=0.31) (Fig. 4). The distribution of prey-to- predator ratios (PPR) ranged from 0.02 to 0.43 but had a skewed distribution toward the lower end of the range (75% of PPRs <0.15) (Fig. 5). This PPR distribution had a bimodal pattern with peaks at 0.07 and 0.14. Mean PPR for all prey was 0.12 (±0.07 SD) but the mean PPR for Atlantic menhaden was 0.19 and for bay anchovy, 0.06. Discussion This study concentrated on migratory adult striped bass that reside in nearshore waters of the Atlantic Ocean Overton et al: Interactions between Morone saxatilis and their prey during winter off the North Carolin coast 179 100 80 in (8 60 E o ^ 40 a? 20 0 20 ® 40 E ^ 60 a? 80 100 i 1 1 1 1 1 1 1 r Tf‘«y''OOrNrOLr)'Dt~" O'C'O'OOOOOO O'C'O'OOOOOO — — - M N N (N N M Atlantic menhaden Bay anchovy .—I r— | — — □ u n 1 1 1 1 1 1 1 r~ O'C'O'OOOOOO O'O'O'OOOOOO — — — (N M ri 400 mm TL, during their ocean residency in winter. The predominance of fish in the diet of striped bass in this study agrees with the findings of other published studies (Manooch, 1973; Overton, 2003; Walter et al., 2003; Walter and Austin, 2003). Several species of clupeoid fishes (e.g., Atlantic menhaden, Atlantic herring, and bay anchovy) dominated diet biomass of striped bass. This dependency on clupeoids, particularly Atlantic menhaden, has been well-documented throughout the range of striped bass (Walter et ah, 2003). The only other study to address the diet of striped bass >500 mm TL was conducted in Chesapeake Bay by Walter and Austin (2003). They showed that Atlantic menhaden contributed 58% of the diet biomass. In the present study, Atlantic menhaden represented a higher biomass (67.9%) of the striped bass diet, indicating a greater dependency on Atlantic men- haden during the period of ocean residence in winter. Anadromous species, particularly alosines, contribute substantially to the diet of striped bass (Nelson et al., 2003; Walter et al., 2003; Savoy and Crecco, 2004). In our study, there were less than five occurrences of alosines, which would indicate that anadromous alo- sines contribute little to the production of striped bass during their ocean residency in winter. Striped bass share similar migration patterns of other anadromous species (Walter et al., 2003) and we commonly observed alosines in the same trawls in which striped bass were collected. Invertebrates were seemingly unimportant to large striped bass winter production because they contributed little to the diet. However, throughout their range, large striped bass routinely feed on a wide variety inverte- brate prey. In New England waters during summer and fall, striped bass consumed large amounts of in- vertebrate prey such as sand shrimp ( Crangon septem- spitiosa), rock crabs ( Cancer irroratus), and American lobster (Homarus americanus ) (Nelson et al., 2003). Large striped bass in Chesapeake Bay routinely fed on invertebrate prey, primarily blue crab (Callinectes sapi- dus), in summer (Walter and Austin, 2003). Presum- ably, these differences among studies due to differences in prey availability. The percentage of stomachs with food varied among years but ranged from 23% in 2005 to 100% in 1995. In Chesapeake Bay, the percentage of large striped bass with food in their stomach during fall and early winter (November and December) was greater than 75% 180 Fishery Bulletin 106(2) (Walter and Austin, 2003). Similarly, the percentage of striped bass stomachs with food during winter in our study was generally greater than 75%. Less than 30% of stomachs sampled during 2005-06 from the recreational catch contained prey, but 80% did during 2007. Striped bass likely expel some stomach contents while being reeled to the surface or while in the codend of a trawl. We did not determine the factors that influence regurgitation with respect to capture method. The alimentary canal musculature is stronger in larger fish and would result in lower regurgitation rates (Staniland et al., 2001). Regurgitation of stomach contents from striped bass collected by hook and line generally consisted of slurry (Overton, 2003). Regurgi- tation rates for adult striped bass captured in gillnets was 8.3% (Sutton et al., 2004). For 2007, the percentage of stomachs with food was greater than 80%. The high frequency of nonempty stomachs in this study may in- dicate that the winter feeding period for the migratory stock may play an important role in providing energy for growth and gonadal development. Striped bass consumed small prey and the mean size of prey consumed was 12% of their total length and ranged from 2% to 43%. This mean percentage was lower than the predicted optimal size of prey (21%) predicted for striped bass (Overton, 2003), but was within the range of the predicted minimum profitable prey lengths (7%), peak profitable (12%) and maximum (40%) for striped bass (Hartman, 2000). In Albemarle Sound, North Carolina, striped bass consumed prey up to 60% of their body length, although mean prey size consumed was 20% of body length (Manooch, 1973). In a more recent study, age 1-3 striped bass in Albe- marle Sound, North Carolina, on average consumed prey about 21% of their body length (Rudershausen et al., 2005). Piscivores generally select for smaller- size prey (Juanes and Conover, 1994). The differences among the studies indicate that larger striped bass include smaller prey in their diet. It may also indicate that there are fewer large prey available to striped bass during the winter. However, we observed that the fin- fish bycatch during the striped bass survey comprised prey larger than what was observed in the stomachs of striped bass. There was a significant positive relationship between prey size and predator size which suggests that larger striped bass consumed larger prey. Nevertheless, the fit of the regression was weak (r2=0.31), indicating a wide variation of prey size was included in the diet. The invertebrate prey in the stomachs was generally <5.0% of the predator total length. About 75% of all prey con- sumed were less than 15% of the total predator length. These percentages were primarily driven by the large number of bay anchovy consumed by striped bass. The average size of Atlantic menhaden that were found in the stomach of striped bass was 204 mm TL. Atlantic menhaden undergo an extensive coastal migration south- ward around the Virginia and North Carolina capes in fall and winter (Reintjes and Pacheco, 1966). All ages in the population participate in this migration; how- ever, younger fish tend to be found within a few miles of the shoreline, while older individuals may be found farther offshore (Reintjes and Pacheco, 1966). Thus, the age-specific distribution of Atlantic menhaden probably influences prey-size availabil- ity to striped bass in nearshore ocean waters. In turn, striped bass potentially have significant impacts through the reduction of age-0 fish on the spawning stock of Atlantic menhaden. The frequency of Atlantic menhaden in the diet has in- creased from 1997 through 2007 and likely represents an increase in the competition be- tween other predators and the existing commercial fishery (Up- hoff, 2003). Given that Atlantic menhaden provide up to 60% of the diet for age 3+ striped bass in Chesapeake Bay (Hartman and Brandt, 1995) and 69% of the diet for striped bass in this study, and given the increased population levels of striped bass, it is likely that striped bass pre- dation represents a large part of the natural mortality for Atlan- tic menhaden (Hartman, 2003; Hartman and Margraf, 2003; Uphoff, 2003). Striped bass (TL, mm) Figure 4 Relationship between striped bass ( Morone saxatilis) total length (TL, mm) and prey length (prey TL mm=-12.07 + 2.84 (striped bass TL); P=<0.0001, r2=0.31) Overton et al.: Interactions between Morone saxatilis and their prey during winter off the North Carolin coast 181 itlUXL J k 0.25 0.30 0.35 Prey-predator ratio (PPR) 0.40 0.45 0.50 Figure 5 Prey-to-predator ratio (PPR) frequency distributions determined from the diet of striped bass ( Morone saxatilis ) collected in trawl and recreational catch samples during winter off the coasts of Virginia and North Carolina during 1994-2007. Coastwide population-level con- sumption of Atlantic menhaden by striped bass in the Atlantic Ocean increased from 50xl03 t in 1982 to over 250xl03 t in 2000 (Overton, 2003). Striped bass are capable of exerting considerable pressure on prey populations through predation (Hartman, 2003; Grout, 2006). With concerns over Atlantic menhaden re- cruitment, it is essential to quantify its role as a prey fish and its major sources of mortality. Our diet data were collected by using two different methods during two separate time periods; therefore we were unable to test the effects of collection methods on the diet com- position. Nevertheless, we feel that the two collection methods comple- ment each other. For example, the trawl samples (1994-2003) show an increasing trend in the amount of Atlantic menhaden consumed; these data are supported by the recreational catch data. Simultane- ously, the trawl data show a decline in the consumption of bay anchovy diet, which is also supported by the recreational catch data. These re- sults indicate that the recreational catch data provide a reasonable representation of the diet of striped bass during the winter off the coasts of North Carolina and Virginia. To further understand the predator-prey interac- tions of striped bass, we suggest a continued low-fre- quency monitoring of predator diets along the Atlantic coast. Low-frequency monitoring approaches have been used to estimate the consumption of commercially important fish by predatory fish in the western North Atlantic and can provide important insights regard- ing the importance of prey types (Overholtz et ah, 2000). These data can be used to calibrate different predator-prey, bioenergetic, and multispecies models for different management systems. This information could provide data that would add significantly to knowledge of trophic interactions of striped bass and other predators. This analysis of the foraging behavior of large migra- tory striped bass during their winter residency in the Atlantic Ocean contributes to the increasing literature on the foraging dynamics of predatory fishes. Whether the patterns observed during our study period were the result of prey dynamics or predator function is unclear. However, striped bass feed on a large number of prey during winter and are also capable of feeding on a wide range of prey sizes. This work outlines the importance of clupeoid fishes to striped bass winter production and also shows that predation may be ex- erting prey pressure on Atlantic menhaden stocks. Acknowledgments We thank W. Laney (United States Fish and Wildlife Service) and H. King (Maryland Department of Natural Resources) for financial support. We especially thank J. Price of the Chesapeake Bay Ecological Foundation for continued support both financially and in the field and K. Riley for providing comments that greatly improved this manuscript. We also thank personnel at Oregon Inlet Fishing Center, particularly M. Swain, for use of their facilities. We acknowledge J. Clermont, M. Butler, C. Lee, and N. Jones for assisting in the field collections. Literature cited Bax, N. J. 1998. The significance and prediction of predation in marine fisheries. ICES J. Mar. Sci. 55:997-1030. Boreman, J., and R. R. Lewis. 1997. Atlantic coastal migration of striped bass. Am. Fish. Soc. Syrnp. 1:331-339. Field, J. D. 1997. Atlantic striped bass management: Where did we go right? Fisheries 22:69. Grout, D. E. 2006. Interactions between striped bass ( Morone saxatilis ) rebuilding programmes and the conservation of Atlantic salmon ( Salmo salar ) and other anadromous fish species in the USA. ICES J. Mar. Sci. 63:1346-1352. 182 Fishery Bulletin 106(2) Hartman, K. J. 2000. The influence of size on striped bass foraging. Mar. Ecol. Prog. Ser. 194:263-268. 2003. Population-level consumption by Atlantic coastal striped bass and the influence of population recovery upon prey communities. Fish. Manag. Ecol. 10:281- 288. Hartman, K. J., and S. B. Brandt. 1995. Trophic resource partitioning, diets, and growth of sympatric estuarine predators. Trans. Am. Fish. Soc. 124:520-537. Hartman, K. J., and F. J. Margraf. 2003. US Atlantic coast striped bass: issues with a recov- ered population. Fish. Manag. Ecol. 10:309-312. Harvey, C. J., S. P. Cox, T. E. Essington, S. Hansson, and J. F. Kitchell. 2003. An ecosystem model of food web and fisheries inter- actions in the Baltic Sea. J. Mar. Sci. 60:939-950. Juanes, F., and D. O. Conover. 1994. Piscivory and prey size selection in young-of-the- year bluefish: predator preference or size dependent capture success? Mar. Ecol. Prog. Ser. 114:59-69 Latour, R. J., M. J. Brush, and C. F. Bonzek. 2003. Toward ecosystem-based fisheries management: strategies for multispecies modeling and associated data requirements. Fisheries 28:10-22. Macdonald, J. S., and R. H. Green. 1983. Redundancy of variables used to describe impor- tance of prey species in fish diets. Can. J. Fish. Aquat. Sci. 40:635-637. Manooch, C. S. III. 1973. Food habits of yearling and adult striped bass, (Morone saxatilis, Walbaum) from Albemarle Sound, North Carolina. Ches. Sci. 14 73-86. Markle, D., and G. Grant. 1970. The summer food habits of young-of-the-year striped bass in three Virginia Rivers. Ches. Sci. 11:50-54. Nelson, G. A., B. C. Chase, and J. Stockwell. 2003. Food habits of striped bass (Morone saxatilis ) in coastal water of Massachusetts. J. Northeast Atl. Fish. Sci. 32:1-25. Overholtz, W. J., J. S. Link, and L. E., Suslowicz. 2000. Consumption of important pelagic fish and squid by predatory fish in the northeastern USA shelf with some fishery comparisons. J. Mar. Sci. 57:1147-1159. Overton, A. S. 2003. Striped bass predator-prey interactions in Chesa- peake Bay and along the Atlantic coast. Ph.D. diss., 226 p. Univ. Maryland Eastern Shore, Princess Anne, MD. Reintjes, J. W., and A. Pacheco. 1966. The relation of menhaden to estuaries. Am. Fish. Soc. Spec. Publ. 3:50-58. Richards, R. A., and P. J. Rago. 1999. A case history of effective fishery management: Chesapeake Bay striped bass. N. Am. J. Fish. Manag. 19:365-375. Rudershausen, P. J., J. E. Tuomikoski, J. A. Buckel, and J. E. Hightower. 2005. Prey selectivity and diet of striped bass in western Albemarle Sound, North Carolina. Trans. Am. Fish. Soc. 134:1059-1074 Savoy, T. F., and V. A. Crecco. 2004. Factors affecting the recent decline of blueback her- ring and American shad in the Connecticut River. Am. Fish. Soc. Mono. 9:361-377. Staniland, I. J., P. J. B. Hart, and P. J. Bromley. 2001. The regurgitation of stomach contents in trawl caught whiting, evidence of a predator size effect. J. Fish. Biol. 59:1430-1432. Sutton, T. M., M. J. Cyterski, J. J. Ney, and M. C. Duval. 2004. Determination of factors influencing stomach con- tent retention by striped bass captured using gillnets. J. Fish. Biol. 64:903-910. Uphoff, J. H. 2003. Predator-prey analysis of striped bass and Atlantic menhaden in upper Chesapeake Bay. Fish. Manag. Ecol. 10:313-322. Walter J. F. . and H. M. Austin. 2003. Diet composition of large striped bass (Morone saxa- tilis) in Chesapeake Bay. Fish. Bull. 104:414-423. Walter, J. F. Ill, A. S. Overton, K. H. Ferry, and M. E. Mather. 2003. Atlantic coast feeding habits of striped bass: a synthesis of data supporting a comprehensive coast-wide understanding of the trophic biology. Fish. Manag. Ecol. 10:349-360. 183 Identification of larval sea basses ( Centropristis spp.) using ribosomal DNA-specific molecular assays* Abstract — The identification of sea bass ( Centropristis ) larvae to spe- cies is difficult because of similar morphological characters, spawning times, and overlapping species ranges. Black sea bass ( Centropristis striata) is an important fishery species and is currently considered to be over- fished south of Cape Hatteras, North Carolina. We describe methods for identifying three species of sea bass larvae using polymerase chain reac- tion (PCR) and restriction fragment length polymorphism (RFLP) assays based on species-specific amplifica- tion of rDNA internal transcribed spacer regions. The assays were tested against DNA of ten other co- occurring reef fish species to ensure the assay’s specificity. Centropristis larvae were collected on three cruises during cross-shelf transects and were used to validate the assays. Seventy- six Centropristis larvae were assayed and 69 (91%) were identified success- fully. DNA was not amplified from 5% of the larvae and identification was inconclusive for 3% of the larvae. These assays can be used to identify sea bass eggs and larvae and will help to assess spawning locations, spawn- ing times, and larval dispersal. Manuscript submitted on 10 October 2007. Manuscript accepted on 6 February 2008. Fish. Bull. 106:183-193 (2008). The views and opinions expressed or implied in this article are those of the author and do not necessarily reflect the position of the National Marine Fisheries Service, NOAA. Mark W. Vandersea (contact author)1 R. Wayne Litaker1 Katrin E. Marancik2 Jonathan A. Hare2 Harvey J. Walsh3 Email address for Mark W. Vandersea: Mark.W. Vandersea @ noaa.gov 1 National Ocean Service, NOAA Center for Coastal Fisheries and Habitat Research 101 Pivers Island Road Beaufort, North Carolina 28516 2 National Marine Fisheries Service, NOAA 28 Tarzwell Drive Narragansett Laboratory Narragansett, Rhode Island 02882 3 Woods Hole Oceanographic Institution Marine Research Facility 217, MS#50 Woods Hole, Massachusetts 02543 A fundamental requirement of early life stage studies is the ability to iden- tify individuals to species. Consider- able knowledge of the eggs and larvae of most fish species in temperate and boreal ecosystems has been achieved (Kendall and Matarese, 1994; Berrien and Sibunka, 1999). In contrast, in subtropical and tropical ecosystems, the larvae of many species cannot be identified (Kendall and Matarese, 1994), even for groups of important fishery species. For example, of the 73 species in the snapper-grouper complex that range within U.S. federal waters, larvae of only one-half can be identi- fied to species and eggs of most spe- cies are undescribed (Richards, 2006). As anthropogenic stress on marine ecosystems continues to increase, new approaches to early life stage identi- fication are needed to obtain critical information about exploited species and to improve the scientific basis for conservation. For example, im- munological methods have been used successfully to identify invertebrate and fish larvae (Miller et ah, 1991; Paugam et ah, 2000; Garland and Zimmer, 2002; Taylor, 2004). Simi- Siya Lem4 Melissa A. West5 David M. Wyanski6 Elisabeth H. Laban1 Patricia A. Tester1 4 Duke University Medical Center PO. Box 2628 Durham, North Carolina 27710 5 David Clark Laboratory North Carolina State University PO Box 7617 Raleigh, North Carolina 27695-7617 6 Marine Resources Research Institute South Carolina Department of Natural Resources PO. Box 12559 Charleston, South Carolina 29422-2559 larly, mitochondrial DNA sequences (Hare et al., 1994; Pegg et ah, 2006) and restriction fragment length poly- morphisms (RFLP) have been used to identify fish eggs, larvae, and adults (Daniel and Graves, 1994; Aranishi et al., 2005; Hyde et al., 2005). Though highly accurate, these methods have yet to be used widely as part of stan- dard protocols for identifying fish larvae collected during research and monitoring surveys. We examined the application of molecular techniques for identifying species within the genus Centropristis (sea bass). Five species of Centropris- tis are currently recognized in the western North Atlantic Ocean. In ad- dition to black sea bass (C. striata), there are the following: bank sea bass (C. ocyurus), rock sea bass (C. phila- delphica), twospot sea bass (C. fuscu- la), and C. rufus. Black sea bass is an important commercial and recreation- al species on the northeast and south- east U.S. continental shelf. Along * Contribution number 635 of the South Car- olina Marine Resources Center, Charles- ton, South Carolina 29422-2559. 184 Fishery Bulletin 106(2) A B (A) Five species of Centropristis (sea bass) are currently recognized in the western North Atlantic Ocean. Black sea bass ( Centropristis striata ) ranges along the northeast and southeast shelves of the U.S. coast and inhabits the Gulf of Mexico on the northwest coast of Florida. Bank sea bass (C. ocyurus ) and rock sea bass (C. philadelphica) are found along the southeast U.S. coast and in the Gulf of Mexico. Centropristis rufus is reported off Martinique in the southeastern Caribbean Sea. Twospot sea bass ( C . fuscula) is reported from Cuba. (B) Sampling stations are marked with an x where Centropristis spp. larvae were collected during three cruises ranging from Chesapeake Bay to southern Georgia during September 2000, November 2000, and February through March 2001. During each cruise, transects running cross-shelf from 10 m to > 1000 m water depth were sampled. The dashed line approximates the 200-m depth contour. the southeast U.S. shelf and the Gulf of Mexico where C. striata, C. ocyurus, and C. philadelphica cohabit (Fig. 1), larvae less than 12 mm cannot be reliably iden- tified to species because fin ray formation is not com- plete. To identify larvae of these species, we developed species-specific polymerase chain reaction (PCR) and RFLP assays based on the internal transcribed spacer (ITS) regions of the ribosomal gene complex. The ITS regions are noncoding spacer regions situated between the 18S and 5.8S rRNA genes (ITS1) and between the 5.8S and 28S rRNA genes (ITS2) (Fig. 2A). The ITS regions work well for species identification because in most eukaryotic organisms these regions diverge rapidly during speciation, enabling even closely related species to be distinguished. The ITS sequences have been used to successfully identify species in groups as diverse as fungi (Lu et al., 2002), plants (Baldwin, 1992), insects (Rafferty et al., 2002), and dinoflagellates (Connell, 2001; Litaker et al., 2003). Other ITS-based assays have been developed to help address conservation and management problems in the trade of shark products (Chapman et al., 2003; Shivji et al., 2005) and for iden- tifying wildlife tissues in forensic applications (Sweijd et al., 2000; Ambercrombie et al., 2005). PCR and RFLP assays offer rapid methods for iden- tification of fish larvae and eggs. The PCR assays can be performed in a one-step process after DNA extrac- tion and they enable species identifications to be made even when meristic characters and other identifying features have been damaged or are underdeveloped. Alternatively, the RFLP assays provide a cost efficient, two-step, PCR-based species identification method. In the first step, PCR is used to amplify the ITS regions. In the second step, restriction enzyme digestion of the PCR product is conducted and species-specific frag- ments are analyzed. This study describes the develop- ment, optimization, and validation of PCR and RFLP assays for three species of Centropristis. Materials and methods Collection and preservation of adult fish Tissue samples were obtained from adult and juvenile fish and either stored in 95% ethanol or frozen at -80°C. Centropi'istis striata were collected from the Atlantic coast of the southeast United States, and C. ocyurus and C. philadelphica were collected from both the southeast- ern U.S Atlantic coast and the Gulf of Mexico. At least three individuals from each location were analyzed. Ten other co-occurring southeast U.S. reef fishes were also collected for sequencing of the ITS regions and served as negative controls for the PCR and RFLP assays. Five of these species were from the subfamily Serraninae: sand perch ( Diplectrum formosum), butter hamlet ( Hypoplec - Vandersea et at: Identification of larval Centropristis spp using ribosomal DNA-specific molecular assasy 185 A F1SH5.8SR CentropFWl FISH5 8SF2 CentropREVl NSF17877F FISH5.8SF FISHLSU5'REV Cphil-fTSR5 Cocy-ITSR7 Cstri-ITSR4 Figure 2 Diagram of the small subunit (SSU), internal transcribed spacer regions (ITS), 5.8S, and large subunit (LSU) genes. (A) Location of the universal polymerase chain reaction (PCR) primers used to amplify the ITS spacer regions, the genus-specific restriction fragment length polymorphism (RFLP) primers, and the internal sequencing primers. (B) Location in the ITS1 region of the Centropristis PCR assay primers. trus unicolor), pygmy sea bass (Serraniculus pumilio), tattler ( Serranus phoebe), and belted sandfish {Serranus subligarius). Four species were from the family Ser- ranidae, red grouper {Epinephelus rnorio ), rock hind ( Epinephelus adscensionis), speckled hind ( Epinephelus drummondhayi) , gag grouper (Mycteroperca microlepis), and one from a related perciform family, Sparidae, the spottail pinfish ( Diplodus holbrooki). Collection and preservation of larvae Three cruises were conducted from Chesapeake Bay to southern Georgia during early fall (September 2000), late fall (November 2000), and late winter (February and March 2001). During each cruise, transects run- ning cross-shelf from 10 m to >1000 m water depth were sampled (Fig. 1). The deepest stations were sampled to 50 m. Ichthyoplankton were collected with a 1-m Tucker trawl fitted with 333-pm mesh nets. The Tucker trawl was fitted with three nets. The first net sampled from the surface to the deepest deployment depth. For stations in <50 m water depth, the lower and upper halves of the water column were sampled discretely by using the second and third nets during the retrieval of the Tucker trawl. For stations in >50 m water depth, the 50-25 m depth was sampled discretely with the second net and the 25-0 m depth interval was sampled discretely with the third net. Depth was determined from the wire angle of the gear and length of wire deployed. Tow speed was between 1.5 and 2 knots. Samples were initially preserved in 95% ethanol. All larvae were sorted from these samples, transferred to 70% ethanol, and identified to the lowest taxonomic level possible. Larvae were identified to the genus Centropris- tis based on characteristic body shape, pigmentation, and when available, dorsal and anal fin meristics (Ken- dall, 1972). Seventy-six larvae were chosen for genetic analyses. These specimens were selected in an attempt to analyze equal numbers of larvae collected from north and south of Cape Hatteras, North Carolina. In this respect, the larvae were not a random sample of the Centropristis larvae identified from the collections. The larvae ranged in size from 1.5-11 mm notochord length (or standard length). Each individual to be used for PCR analyses was rinsed three times with clean 95% ethanol, digitally imaged, and stored in a separate sterile vial with 95% ethanol. To prevent DNA cross-contamination, the forceps used for handling larval fish were decon- taminated between specimens with 10% sodium hypo- chlorite, rinsed with clean deionized water, and dried with a clean Kimwipe (Kimberly-Clark, Roswell, GA). Determination of ITS region sequences To identify species-specific primer sites it was first necessary to amplify and sequence the ITS regions from adult Centropristis and other reef fish. To accom- plish this, universal PCR primers were designed for the amplification of the 3' end of the small subunit (SSU), the internal, transcribed spacer region ( ITS 1 ) , 5.8S subunit (5.8S), the second internal transcribed spacer region (ITS2), and the first ~48 base pairs of the large subunit (LSU) rRNA genes (~1300-bp product). The uni- versal forward primer NSF1787F (Fig. 2A, Table 1) was designed with the reverse complement of the universal reverse primer, NSR1787/18, listed on the ribosomal RNA database website. The universal reverse primer FishLSU5’Rev (Fig. 2A, Table 1) was designed by align- ing (CLUSTAL X algorithm; Thompson et ah, 1997) the large subunit rDNA sequences of the following fish spe- cies that were available in GenBank: European perch ( Perea fluviatilis Z18686), rainbow trout ( Oncorhynchus 186 Fishery Bulletin 106(2) mykiss OMU34341, Z18683), tub gurnard ( Trigla lucerna Z18768), Atlantic mackerel ( Scomber scombrus Z18693), zebrafish ( Brachydanio rerio AJ306603), Pimelodella cristata (AJ306596), Atlantic herring ( Clupea harengus Z18764), dab ( Limanda limanda Z18681), brown bullhead ( Ictalurus nebulosus Z18678), spotted green pufferfish ( Tetraodon nigroviridis AJ270038, AJ270037), common carp (Cyprinus carpio AF133089), black ghost ( Apterono - tus albifrons AJ306595), Eurasian minnow ( Phoxinus phoxinus AJ306604), Synodontis clarias (AJ306597), Orinocodoras eigenmanni (AJ306606), angler ( Lophius piscatorius Z18765), European pilchard {Sardina pilchar- dus Z18767), coelacanth ( Latimeria chalumnae U34336), ghost shark ( Callorhinchus milii AY049812), Florida gar ( Lepisosteus platyrhynchus Z18680), arawana ( Osteo - glossum sp. Z18684), rabbit fish ( Chimaera monstrosa Z18674), European eel ( Anguilla anguilla Z18673), bowfin ( Amia calva Z18672), and browneye skate ( Raja schmidti AF278683). DNA purification For adult specimens, approximately 50-100 mg of muscle tissue or a clipping from the fin was collected from either frozen or ethanol-preserved specimens and DNA was extracted with a Roche High Pure PCR Template Preparation Kit (Roche Diagnostics GmbH, Mannheim, Germany) following the manufacturer’s protocol for isolation of nucleic acids from mammalian tissue. The one exception to the manufacturer’s protocol was that we used a volume of 50 pL to elute the DNA in the final step of the procedure. For ethanol-preserved specimens, the tissues were rehydrated in three changes of sterile tris ethylenediaminetetraacetic acid (TE) buffer for 30 minutes each. Tissues were then transferred to the cell lysis solution, and DNA was purified following the manufacture’s protocol. DNA extraction from larval fish was accomplished by first placing them in separate sterile 55 mm Pe- tri dishes and allowing them to rehydrate in sterile TE buffer [lOmM Tris HC1 (pH 7.4), ImM disodium ethylenediaminetetraacetate (EDTA) (pH 8.0)] for 30 minutes. Next, an eye was microdissected with fine forceps and the DNA was purified as described above. This limited the physical damage to the larvae, allow- ing them to be used for future meristic studies. The forceps used for microdissection were decontaminated between specimens as described above. PCR amplification, cloning, and sequencing procedures for ITS regions of adult fish The PCR amplification reaction mixture used to amplify adult fish ITS regions is listed in Table 2 . DNA was PCR amplified in a MJ Research MiniCycler (MJ Research, Waltham MA) under the following touchdown cycling conditions: 2 min. at 95°C, 35 cycles each consisting of 40 sec. denaturation at 95°C, 40 sec. initial annealing temperature at 64°C which decreased by 0.5°C per cycle for six cycles and 61°C thereafter and an extension of 1.5 min. at 72°C. This procedure was followed by a final extension of 5 min. at 72°C. A 4-pL aliquot of each PCR reaction was checked for the presence of a specific amplification product by agarose gel electrophoresis (2% agarose, tris acetate EDTA fTAE gel], 50 volts) and ethidium bromide staining. PCR reactions containing specific products were cloned into the plasmid vector pCR2.1 by using the Topo TA Cloning Kit and by follow- ing the manufacturer’s protocol (Invitrogen, Carlsbad, CA). Plasmids were isolated and purified with a QIAprep Spin Miniprep Kit (Qiagen, Valencia, CA) and sequenced with an ABI377 DNA sequencer employing the Deoxy Terminator Cycle sequencing kit (Applied Biosystems- ABI, Foster City, CA). In addition to the ten local reef fishes, three adult fish of each Centropristis species were PCR amplified and cloned. DNA templates were sequenced completely in both directions with the M13F, M13R, Fish5.8SF, Fish5.8SF2, and Fish5.8SR prim- ers (Fig. 2A, Table 1). The resulting 3' SSU to 5' LSU sequence for each species was assembled by using the Vector NTI program (Informax Inc., Bethesda, MD) and submitted to GenBank (EF472464-EF472500). Phylogenetic analysis labile 1 Primers used in this study to amplify and sequence the 3' end of the small sub- unit gene, internal transcribed spacer 1, 5.8S gene, internal transcribed spacer 2, and the first approximately 48-bp of the large subunit gene Primer name Sequence (5'— 3') Forward sequencing M13F (plasmid vector) NSF1787F (universal forward) Fish5.8SF Fish5.8SF2 GTAAAACGACGGCCAG CCGTAGGTGAACCTGCGG AGCTGCGAGAACTAATGTGAA TGCTCTGCTCGGGCTGTAGCG Reverse sequencing M13R (plasmid vector) FishLSU5’Rev (universal reverse) Fish5.8SR CAGGAAACAGCTATGAC CTTAAATTCAGCGGGTTGTCT TTCACATTAGTTCTCGCAGCT The ITS rDNA gene sequence from C. striata, C. ocyurus, C. philadel- phica , and Diplectrum formosum (for an outgroup) were aligned using the CLUSTAL-X algorithm (Thompson et ah, 1997). A Bayesian phylogenetic analysis of the aligned sequences was performed with the MrBayes 3.1 program (Huelsenbeck et ah, 2001). Posterior probabilities were calcu- lated by using a Metropolis-coupled Markovian Chain Monte Carlo approach and with sampling con- ducted according to the Metropolis- Vandersea et al Identification of larval Centropristis spp using ribosomal DNA-specific molecular assasy 187 Table 2 Polymerase chain reaction (PCR) reagent concentrations used amplify fish internal transcribed spacer regions and conduct PCR and restriction fragment length polymorphism (RFLP) assays for Centropristis striata (black sea bass), C. ocyurus (bank sea bass), and C. philadelphica (rock sea bass). Reagent Fish ITS PCR reaction mix C. striata assay C. ocyurus assay C. philadelphica assay RFLP PCR reaction mix Tris-HCl 100 mM, pH 8.8 100 mM, pH 8.8 200 mM, pH 8.4 100 mM, pH 8.8 100 mM, pH 8.3 MgCl2 15 mM 15 mM 30 mM 15 mM 15 mM KC1 750 mM 750 mM 500 mM 750 mM 750 mM Primer I 25 pmoles 25 pmoles 25 pmoles 25 pmoles 25 pmoles Primer II 25 pmoles 25 pmoles 25 pmoles 25 pmoles 25 pmoles dNTP mix 10 mM 10 mM 10 mM 10 mM 10 mM DMSO 5% 5% 5% 5% 5% Taq DNA polymerase 0.25 units 0.25 units 0.25 units 0.25 units 0.25 units DNA template 20-50 ng 20-50 ng 20-50 ng 20-50 ng 20-50 ng Reaction volume 50 pL 50 pL 50 pL 50 pL 50 pL Table 3 Primer pairs used to conduct the Centropristis (sea bass) polymerase chain reaction (PCR) and restriction fragment length poly- morphism (RFLP) assays. bp=base pairs. Species Primer Primer sequence 5'— 3' PCR product size C. striata assay Cstri-ITSF2 (forward) Cstri-ITSR4 (reverse) TGGACCGGCTTTCCTCCCG CAATGAGGGTTGGAGAAAGGG 230 bp C. ocyurus assay Cocy-ITSF5 (forward) Cocy-ITSR7 (reverse) CTCGTCCTCCTTGCGGTGG GGAGGTTTTGGTTTGTGTAGG 158 bp C. philadelphica assay Cphil-ITSF5 (forward) Cphil-ITSR5 (reverse) CACTGCCACTGCCTCCAC ACGGAGCCAGCTTTCACC 238 bp Centropristis RFLP assay CentropFWl (forward) CentropREVl (reverse) GATCATTACCGGTCGGTTGC GTAGTCGAAAAGTGGAGGCAG -1200 bp Hastings algorithm. The phylogenetic analysis employed two concurrent analyses of four chains each. In each analysis, one cold and three incrementally heated chains were used where the heat of the tth chain is B = l/[l+(i-l)T] and T = 0.2. Initial phylogenetic trees for each chain were random and used default starting values of MrBayes. A single run consisting of 100,000 genera- tions was conducted. The run was sampled at every 100th tree. Only trees sampled after a stable burn-in of 1000 generations were used. The results were plotted as a rooted phylogram with Diplectrum formosum as the outgroup. An outgroup is any species occurring outside a particular branch or clade, i.e., a species that is further towards the root of a phylogenetic tree. Development of species-specific PCR assays for identifying larval fish The 3' SSU — 5' LSU sequences for Centropristis striata ( EF472473-EF472481), C. philadelphica (EF472482- EF472490), C. ocyurus (EF472491-EF472499), Diplectrum formosum (EF472472), Diplodus holbrooki (EF472471), Epinephelus morio (EF472468), Epinephelus adscensionis (EF472470), Epinephelus drummondhayi (EF472469), Hypoplectrus unicolor (EF472467), Mycteroperca micro- lepis (EF472466), Serraniculus pumilio (EF472465), Serranus phoebe (EF472464), and Serranus subligarius (EF472500) were aligned by using the CLUSTAL-X algo- rithm. The alignments were used to identify unique ITS sequences and to develop species-specific PCR assays. For all three Centropristis assays, we used forward and reverse primers located within the ITS1 region (Fig. 2B). The PCR primers used for each assay are listed in Table 3. The assays were conducted under touchdown PCR cycling conditions with an MJ Research PTC-150 MiniCycler. The PCR reaction profile for the C. striata assay was as follows: 2 min. at 95°C, 35 cycles each consisting of 30 sec. denaturation at 95°C, 40 sec. ini- tial annealing temperature at 66°C which decreased by 0.5°C per cycle for four cycles and 64°C thereafter, and 188 Fishery Bulletin 106(2) an extension of 45 sec. at 72°C. This was followed by a final extension of 5 min. at 72°C. The cycling conditions for the C. ocyurus assay were the following: 2 min. at 95°C, 35 cycles each consisting of 30 sec. denaturation at 95°C, 40 sec. initial annealing temperature at 64°C which decreased by 0.5°C per cycle for four cycles and 62°C thereafter and an extension of 45 sec. at 72°C. This was followed by a final extension of 5 min. at 72°C. Lastly, the cycling conditions for the C. philadelphica assay were: 2 min. at 95°C, 35 cycles each consisting of 30 sec. denaturation at 95°C, 40 sec. initial annealing temperature at 60°C which decreased by 0.5°C per cycle for four cycles and 58°C thereafter and an extension of 45 sec. at 72°C. This was followed by a final extension of 5 min. at 72°C. An aliquot (4 pL) from each amplifi- cation was analyzed by agarose gel electrophoresis (3% agarose TAE gel, 75V). The sizes of the PCR products were estimated with either a 123-bp or 50-bp ladder (Promega, Madison, WI). Validation of the PCR assays for identifying larval fish Following assay development, the primer pairs were tested for cross-reactivity with a panel of DNAs that included the 10 reef fish species listed previously (20-50 ng of DNA per PCR reaction). DNA from 76 Centropi'istis larvae was extracted using the methods described above. The larvae were assayed by using 20-50 ng of DNA from each specimen. Each PCR assay included a posi- tive control, a negative control, a blank DNA extraction control, and two PCR inhibition controls. The positive control incorporated 20 ng of appropriate Centropris- tis DNA in the reaction mixture. The negative control substituted lx PCR buffer for DNA to confirm that the reagents were not contaminated with target DNA. The blank extraction control was included to assess pos- sible cross-contamination during the extraction proce- dures. The inhibition control consisted of spiking 20 ng of appropriate Centropristis DNA into two arbitrarily chosen larval fish DNA samples. The PCR inhibition controls confirmed that negative PCR reactions were due to the absence of appropriate Centropristis DNA and not from PCR inhibition. The C. striata and C. ocyurus assays were tested on all of the larvae in the collection. The C. philadelphica assay was applied by process of elimination to larvae that did not yield results from the C. striata or C. ocyurus assays. Development of RFLP assays for identifying larval fish Forward and reverse genus-specific primers for three species of Centropristis were designed from the 3' SSU- 5' LSU sequence alignment described above. The forward primer, CentropFWl (Table 3), overlapped the boundary of the 3' end of the SSU rRNA gene and the 5' end of the ITS1 spacer (Fig. 2A). The reverse primer, CentropRevl (Table 3), overlapped the boundary of the 3' end of the ITS2 spacer and the 5' end of the LSU rRNA gene (Fig. 2A). The primer pair yielded an approximate 1200-bp PCR product and was tested for cross-reactivity against the 10 non -Centropristis reef fishes listed above. The RFLP-PCR reaction mix is listed in Table 2. Twenty to 50 ng of genomic DNA was used per PCR reaction. PCR amplifications were conducted with an MJ Research PTC-150 MiniCycler under the following cycling condi- tions: 2 min. at 95°C, 40 cycles each consisting of 30 sec. denaturation at 95°C, 40 sec. initial annealing temperature at 63°C which decreased by 0.5°C per cycle for six cycles and 60°C thereafter, and an extension of 1 min. at 72°C. This was followed by a final extension of 5 min. at 72°C. A 4-pL aliquot of each PCR reaction was checked for the presence of a specific amplification product by agarose gel electrophoresis (2% agarose TAE gel, 50 V) and ethidium bromide staining. The sizes of the PCR products were estimated by using the DNA molecular weight marker IX (Roche Diagnostics, India- napolis, IN). RFLP analysis of Centropristis PCR products was simulated with the RFLP analysis tool in the Vector NTI program (Informax Inc., Bethesda, MD) to identify species-specific RFLP patterns. The restriction enzyme Alu I (New England Biolabs, Beverly, MA) was selected on the basis of the distinct restriction patterns pre- dicted for each Centropristis species. Restriction digests were performed in 30-pL reactions containing 3 uL of New England Biolabs lOx reaction buffer #3, 25 pL of PCR product, 2000 units of Alu I enzyme, and were incubated at 37°C for 2 hours. Restriction fragments were separated by gel electrophoresis on 3% agarose, Tris-Borate EDTA NuSieve 3:1 gels (Cambrex Bio Sci- ence Rockland, Inc., Rockland, ME) and fragments sizes were estimated by using a 50-bp DNA ladder (Promega, Madison, WI). Results Species-specific molecular assays based on differences in the ITS rDNA regions were developed successfully for the identification of Centropristis larvae. The approach we employed in developing the molecular assays was 1) to identify universal PCR primer sites that would amplify the ITS regions of many fish species; 2) to sequence the ITS regions from three Centropristis species and other co-occurring reef species; 3) to develop species-specific PCR and RFLP assays for the Centropristis species based on alignments of the resulting sequences; and 4) to use the assays to identify field-collected larval Centropristis. The universal PCR primers (Table 1) were successful in amplifying the ITS regions of many reef fish species. The primers generated an approximate 1200-1300 base pair product that varied with the species. A phyloge- netic analysis of the Centropristis ITS sequence align- ments showed that between-species sequence divergence in this region was much greater than within-species divergence (Fig. 3). This variation made it possible to develop species-specific PCR assays for C. striata, C. ocyurus, and C. philadelphica. The primers were tested for cross-reactivity against ten other related reef fish Vandersea et at: Identification of larval Centropristis spp. using ribosomal DNA-specific molecular assasy 189 1.00 1.00 1.00 Diplectrum formosum Centropristis philadelphica Centropristis philadelphica Centropristis philadelphica Centropristis philadelphica Centropristis philadelphica Centropristis philadelphica Centropristis philadelphica Centropristis philadelphica Centropristis philadelphica Centropristis ocyurus Centropristis ocyurus Centropristis ocyurus Centropristis ocyurus Centropristis ocyurus Centropristis ocyurus Centropristis ocyurus Centropristis ocyurus Centropristis ocyurus Centropristis striata Centropristis striata Centropristis striata 1 00 |L Centropristis striata Centropristis striata Centropristis striata Centropristis striata Centropristis striata Centropristis striata 0.1 Substitutions per site Figure 3 Phylogeny showing the relationship between black sea bass ( Centropristis striata), bank sea bass ( C . ocyurus), and rock sea bass ( C . philadelphica) based on ITS1, 5.8S, and ITS2 rDNA sequences. Sand perch ( Diplectrum formosum) served as the outgroup. Branch support for each species was 100% (1.0). The scale at the bottom of the phylogeny is proportional to the branch lengths in the phylogenetic tree that correspond to a diver- gence of 0.1 nucleotide substitutions per base pair in the DNA sequences. The within-species sequence variation was consistently less than that observed between species. species and they did not amplify non- target species (Fig. 4). The accuracy of the species-specific assays was checked by PCR assaying 15 adult and juvenile sea basses (5 of each species). The PCR assays were successful in identifying the target species and did not cross-react among the three Centropristis species. We validated the PCR assays by using them to assess the identity of 76 Centro- pristis larvae (Fig. 5). Sixty-nine of the 76 larvae (91%) were identified success- fully: 33 C. striata, 32 C. ocyurus, and four C. philadelphica. DNA from four of the larvae failed to PCR amplify and in three cases the PCR assay results were inconclusive. The four DNA samples that failed to amplify were tested for PCR inhibition by spiking a small amount of target Centropristis plasmid DNA into PCR reaction mixes containing DNA from the larvae. These spiked controls all amplified, indicating that PCR in- hibition was not a factor. DNA was re- extracted from the larvae and attempts were made to re-amplify it with the spe- cies-specific primers and the universal primers. Neither primer sets produced a PCR product, indicating that DNA deg- radation had occurred to the extent that amplifiable DNA could not be isolated from these four specimens. The three inconclusive results were symptomatic of DNA cross-contamina- tion between larval samples. Both the C. striata and C. ocyurus specific primer sets produced correct size PCR products for these samples. DNA from these lar- vae was re-extracted in an attempt to eliminate the cross-contamination, but multiple PCR assay attempts yielded the same ambiguous results. The issues of cross-contamination and methods to avoid it will be discussed below. RFLP assays were also developed for each species as an alternative to the PCR assays. Genus-specific PCR primers (Table 3) that flanked the ITS regions were used to amplify DNA from identi- fied adult sea basses. The genus-specific primer sites were located very near the universal prim- er sites but were designed to amplify only Centropris- tis DNA. The primers were tested for cross-reactivity against the same related reef species that were used in the PCR assays. Cross-reactivity did not occur (Fig. 6A). The PCR products were digested with the Alu I restriction enzyme. Electrophoresis yielded unique banding patterns (Fig. 6B). Digests of C. striata yielded fragments of 446, 418, 218, 125, and 4 base pairs. The restriction fragment sizes for C. ocyurus were 449, 362, 245, 120, 51 bp, and 4 base pairs. For C. philadelphica the fragments were 481, 261, 187, 185, 121, 58, and 4 base pairs. The advantages and disadvantages of this diagnostic approach are discussed below. Discussion Ribosomal DNA-specific assays for Centropristis larvae have proven to be robust and accurate. The difficul- 190 Fishery Bulletin 106(2) ties associated with identifying Centropristis larvae prompted this work. The ITS region was chosen for assay development because it diverges rapidly during speciation and contains unique primer sites for dis- tinguishing species (Baldwin, 1992; Lu et al., 2002; Litaker et al., 2003). The divergence between Centro- pristis species was supported by the ITS phylogenetic analysis, which showed that C. striata, C. ocyurus, and C. philadelphiea were genetically distinct from each other and from another related serranid species, Diplectrum formosum (Fig. 3). The analysis further showed that C. ocyurus and C. philadelphiea are more closely related when compared to C. striata, which is consistent with the morphological characteristics of the adult fish. One of the goals of this study was to conduct the larval assays in a manner that would cause minimal damage to the preserved specimens. The method used in our study, where DNA was recovered from a single eye, made it possible to perform morphometric analyses on genetically identified fish and to evaluate whether species-specific characteristics exist that can be used to identify the larvae. The PCR assays were successfully validated by us- ing larvae that were collected on cruises ranging from Chesapeake Bay to southern Georgia during September 2000, November 2000, and February through March 2001 (Fig. 1). The peak spawning season for C. striata occurs between June and September in the Mid-Atlantic Bight (Able et al., 1995) and between March and May in the South Atlantic Bight (Wenner et al., 1986; Mercer, 1989; McGovern et al., 2002). Data for spawning of C. philadelphiea comes from Miller (1959) who concluded that spawning occurs during May and June. Ma- nooch (1984) indicated that ripe females of C. ocyurus were collected in the South Atlantic Bight during March and April and that young fish appear in late April, suggesting that spawning occurs offshore in spring. Results from our study confirmed that spawning of C. ocyurus overlapped with that of C. striata and occurred in September and as early as February to March. The data from our study for C. philadelphiea are limited, making it difficult to determine whether the low abun- dance was due to spawning period, spawn- ing locations, or spawning stock biomass. Centropristis philadelphiea larvae may have been less abundant in the collections because most of the sampling was conducted in waters deeper than 10 m. Trawling studies off the southeastern United States indicate that C. philadelphiea is a year-round resident of shal- low coastal waters (<10 m) with sandy-mud substrates (Wenner and Sedberry, 1989) and is not present at greater depths over sandy- mud or hard bottoms (Wenner et al., 1979; Sedberry and Van Dolah, 1984). Furthermore, it is likely that the sampling of larvae did not occur during the peak spawning season for C. philadelphiea (Miller, 1959). Because PCR is a highly sensitive molecu- lar technique, care must be taken to avoid contamination. The data indicated that DNA cross-contamination occurred in three of the larval samples. We hypothesize that the DNA contamination occurred while the Centropris- tis larvae were stored collectively or during the sorting of the larvae. Field-collected lar- vae should therefore be preserved individually and utensils or tools used to process larvae should be decontaminated between larvae during the sorting process with a reagent that destroys DNA. a 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 A jc 246 bp mm 123bp mm c 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 Figure 4 Results of cross-reactivity tests for (A) black sea bass (Centropristis striata), (B) bank sea bass (C. ocyurus ), and (C) rock sea bass (C. philadelphiea) polymerase chain reaction (PCR) assays. In each panel, the gels were loaded in the following order: lane 1, 123-bp ladder; lane 2, positive controls for A) C. striata, B) C. ocyurus, and C) C. philadelphiea ; lane 3, A) C. ocyurus, B) C. striata, C) C. philadelphiea ; lane 4, A) C. philadelphiea, B) C. striata, C) C. ocyurus\ lanes 5-14, sand perch ( Diplectrum formosum), spottail pinfish (Diplodus holbrooki), red grouper (Epinephelus morio), rock hind (Epinephelus adscensionis), speckled hind (Epineph- elus drummondhayi), butter hamlet (Hypoplectrus unicolor), gag grouper ( Mycteroperca microlepis), pygmy sea bass (Serraniculus pumilio), tattler ( Serranus phoebe), and belted sandfish (Serranus subligarius); lane 15, no DNA control. Vandersea et al.: Identification of larval Centropristis spp using ribosomal DNA-specific molecular assasy 191 A 12345678 9 10 11 12 13 14 15 16 17 B 123 4 5 678 9 10 11 12 13 14 15 16 17 246 bp. 123 bp« C 1 2 3 4 5 6 7 8 9 10 11 12131415 16 17 _ . _ , 246 bp 4 123 bp Figure 5 Results of species-specific polymerase chain reaction (PCR) assays for sea bass (Centroprisitis) larvae (A) black sea bass ( Centropristis striata ), (B) bank sea bass (C. ocyurus), and (C) rock sea bass (C. philadelphica). In each panel, the gels were loaded in the follow- ing manner: lane 1, 123-bp ladder; lane 2, positive controls for A (C. striata), B ( C . ocyurus), and C (C. philadelphica)-, lanes 3-13, Centropristis larvae, lane 14, negative control; lane 15-16, PCR inhibition controls; lane 17, no DNA control. PCR versus RFLP assays The species-specific PCR and RFLP assays can be used to accurately identify larvae. The decision of which assay to use could be based on labor and reagent costs. The PCR assays are attractive because they are rapid single step assays, whereas the RFLP assays employ PCR in the first step and restriction enzyme diges- tion of the PCR products in the second step. In most cases, PCR reagents are more expensive than those needed for restriction enzyme analy- sis. However, PCR assays can be cost effective if they are multiplexed. Excellent examples of suc- cessful multiplex PCR assays are described by Chapman et al. (2003) and Hyde et al. (2005). Unfortunately our attempts at multiplexing the assays were unsuccessful because of the produc- tion of nonspecific PCR products. RFLP assays are cost effective because they require less expensive reagents than PCR as- says. However, the RFLP assays require an extra hour or two to process and involve ad- ditional labor costs. The RFLP assays devel- oped here employed genus-specific primers rather than species-specific primers to amplify Centropristis genomic DNA from adult fish (Table 3). Using genus-specific primers, we simplified the PCR reaction mixtures for all three species and ensured that only Centro- pristis DNA would generate PCR products. The resulting PCR products were subsequent- ly digested with Alu I enzyme to produce spe- cies-specific fragments that were easily distin- guished by their unique banding patterns once electrophoretically separated on an agrose gel. The unique sizes of these fragments were due to species-specific differences in the locations of the Alu I restriction sites within the ITS1, 5.8S gene, and ITS2 regions (Fig. 6B). Shipboard operational molecular assays The PCR and RFLP assays could be conducted at sea with a small thermocycler and the other minor equip- ment necessary for DNA extraction. If onboard con- ditions were favorable, the assays could be analyzed immediately after PCR thermocycling by gel electro- phoresis and with the use of a digital imaging camera. Similar methods described by Hyde et al., (2005) were used at sea to identify the eggs and larvae of blue marlin ( Makaira nigricans), shortbill spearfish ( Tetrapturus angustirostris), and wahoo [Acanthocybium solandri). Their procedures incorporated a boiling method to quickly prepare amplifiable DNA from fresh tissue. If this would not be practical, then PCR products could be frozen and analyzed at an onshore laboratory. Results for 30 larval assays could be attained in four hours or less at a cost of ~$5 per sample, excluding the cost of equipment. Alternatively, larvae could be collected and sorted; all Centropristis larvae would be separated and assayed or stored individually in 95% ethanol for pro- cessing at a shore-based laboratory. DNA could then be extracted from a small tissue sample (e.g., an eye) from each specimen. The intact larvae would be returned to storage in 95% ethanol in case any additional morpho- logical or molecular evaluation was required. If analysis of a large number of samples was desired, the individual PCR assays could be adapted to a SYBR green quantita- tive PCR format. Because this approach eliminates the need for gel electrophoresis, it could be used to process hundreds of samples provided that sorting could be expedited. Conclusion Both the species-specific PCR and RFLP assays described in this article can successfully identify C. striata, C. ocyurus, and C. philadelphica. The assays are rapid, cost effective, simple to perform, and highly accurate. 192 Fishery Bulletin 106(2) Furthermore, these techniques can be integrated with environmental and oceanographic data to allow almost immediate investigation of spawning times, spawning locations, and larval dispersal. The ability to identify Centropristis larvae to species throughout their range will provide a foundation for future studies where early life history stages of fishes are used to investigate questions related to fisheries management. This study also provides a model for the future development of species-specific assays for other commercially important fish species. 1011 1213141516 B 1 2 3 4 5 6 7 8 1353 bp«» * ^ 1078 bpr - 603 bp ■ 310 bp . rr ;**..*«$ I -. 'V : 500 bp */450 bp 1X400 bp ^350 bp — -300 bp 250 bp 200 bp 150 bp 100 bp 50 bp Figure 6 (A) Cross-reactivity test of genus-specific polymerase chain reaction (PCR) primers (CentropFWl and CentropRevl) used for restriction fragment length polymorphism analysis (RFLP) of black sea bass (Centropristis striata ), bank sea bass (C. ocyurus), and rock sea bass (C. philadelphica). The gel was loaded in the following order: lanes 1 and 16, DNA molecular weight marker III (Roche Diagnostics, Indianapolis, IN); lane 2, C. striata; lane 3, C. ocyurus; lane 4, C. philadelphica; lanes 5-14, sand perch ( Diplectrum formosum ), spottail pinfish ( Diplodus holbrooki), red grouper ( Epinephelus morio ), rock hind ( Epinephelus adscensio- nis), speckled hind (Epinephelus drummondhayi), butter hamlet ( Hypoplectrus unicolor), gag grouper (Mycteroperca microlepis), pygmy sea bass ( Serraniculus pumilio), tattler (Serranus phoebe), and belted sandfish ( Serranus subligarius); lane 15, no DNA con- trol. (B) PCR-RFLP analysis of Centropristis ITS1, 5.8S, and ITS2 rDNA amplified with genus-specific primers described in the text. Each PCR product was digested with the restriction enzyme Alu I. Lane 1, DNA molecular weight marker IX (Roche Diagnostics, Indianapolis, IN); lanes 2-4 PCR products for C. striata, C. ocyurus, and C. philadelphica; lanes 5-7, Alu I digests of PCR products for C. striata, C. ocyurus, and C. philadelphica; lane 8, 50-bp ladder. Acknowledgments We thank the crew of the RV Cape Hatteras for assis- tance at sea and the crew of the SEAMAP-SA Coastal Survey based at South Carolina Department of Natural Resources in Charleston, SC, for providing juvenile-adult specimens of C. philadelphica. Collec- tion of larvae at sea was supported by funding from the National Science Foundation through OCE 9876565 to C. Jones, S. Thorrold, A. Valle-Levinson, and J. Hare. Additional funding for this project was provided by Office of National Marine Sanc- tuaries and by Grays Reef National Marine Sanctuary. Literature cited Able, K. W„ M. P. Fahay, and G. R. Shepard 1995. Early life history of black sea bass, Cen- tropristis striata, in the mid-Atlantic Bight and a New Jersey estuary. Fish. Bull. 93: 429-445. Abercrombie, D. L., S. C. Clarke, and M. S. Shivji. 2005. Global-scale genetic identification of hammerhead sharks: Application to assess- ment of the international fin trade and law enforcement. Conserv. Genet. 6:775-788. Aranishi, F., T. Okimoto, and I. Shotaro 2005. Identification of gadoid species (Pisces, Gadidae) by PCR-RFLP analysis. J. App! Genet. 46:69-73. Baldwin, B. G. 1992. Phylogenetic utility of the internal tran- scribed spacers of nuclear ribosomal DNA in plants: an example from the compositae. Mol. Phylogenet. Evol. 1:3-16. Berrien, P, and J. Sibunka 1999. Distribution patterns of fish eggs in the U.S. northeast continental shelf ecosys- tem, 1977-1987. NOAA Tech. Rep. NMFS 145, 310 p. Chapman, D. D., D. L. Abercrombie, C. J. Douady, E. K. Pikitch, M. J. Stanhop, and M. S. Shivji. 2003. A streamlined, bi-organelle, multiplex PCR approach to species identification: Appli- cation to global conservation and trade moni- toring of the great white shark, Carcharodon carcharias. Conserv. Genet. 4:415-525. Connell, L. 2001. Rapid identification of marine algae (Raphidophyceae) using three-primer PCR amplification of nuclear internal transcribed spacer (ITS) regions from fresh and archived material. Phycologia 41:15-21. Daniel, L. B., and J. E. Graves 1994. Morphometric and genetic identification of eggs of spring-spawning sciaenids in lower Chesapeake Bay. Fish. Bull. 92:254-261. Garland, E. D., and C. A. Zimmer 2002. Techniques for the identification of bivalve larvae. Mar. Ecol. Prog. Ser. 225: 299-310. Vandersea et al.: Identification of larval Centropristis spp. using ribosomal DNA-specific molecular assasy 193 Hare, J. A., R. K. Cowen, J. P. Zehr, F. Juanes, and K. H. Day 1994. Biological and oceanographic insights from larval labrid (Pisces: Labridae) identification using mtDNA sequences. Mar. Biol. 118:17-24. Huelsenbeck, J. P., F. Ronquist, R. Nielsen, and J. P. Bollback 2001. Bayesian inference of phylogeny and its impact on evolutionary biology. Science 294:2310-2314. Hyde, J. R., E. Lynn, R. Humphreys Jr., M. Musyl, A. P. West, and R. Vetter 2005. Shipboard identification of fish eggs and larvae by multiplex PCR, and description of fertilized eggs of blue marlin, shortbill spearfish, and wahoo. Mar. Ecol. Prog. Ser. 286:269-277. Kendall, A. W. 1972. Description of black sea bass, Centropristis stri- ata (Linnaeus), larvae and their occurrences north of Cape Lookout, North Carolina, in 1966. Fish. Bull. 70:1243-1259 Kendall, A. W., and A. C. Matarese 1994. Status of early life history descriptions of marine teleosts. Fish. Bull. 92:725-736. Litaker, R. W., M. W. Vandersea, S. R. Kibler, K. S. Reece, N. A Stokes, A. K. Steidinger, D. F. Millie, B. J. Bendis, R. J. Pigg, and P. A. Tester 2003. Identification of Pfiesteria piscicida (Dinophyceae) and Pfiesteria- like organisms using internal transcribed spacer-specific PCR assays. J. Phycol. 39:754-761. Lu, L., J. Li, and Y. Cang 2002. PCR-based sensitive detection of medicinal fungi Hericium species from ribosomal internal tran- scribed spacer (ITS) sequences. Biol. Pharm. Bull. 25:975-980. McGovern, J. C., M. R. Collins, O. Pashuk, and H. S. Meister 2002. Temporal and spatial differences in life history parameters of black sea bass in the southeastern United States. N. Am. J. Fish. Manag. 22:1151-1163. Manooch, C. S. Ill 1984. Fisherman’s guide, fishes of the southeastern United States, 362 p. North Carolina State Museum of Natural History, Raleigh, NC. Mercer, L. P. 1989. Species profiles: life histories and environmental requirements of coastal fishes and invertebrates (South Atlantic) — black sea bass, 16 p. U.S. Fish Wildl. Serv. Biol. Rep. 82(11.99). U.S. Army Corps of Engineers, TR EL-82-4. Miller, K. M., P. Jones, and J. Roughgarden 1991. Monoclonal antibodies as species-specific probes in oceanographic research: examples with intertidal barnacles. Mol. Mar. Biol. Biotech. 1:35-47. Miller, R. J. 1959. A review of the seabasses of the genus Centropristes (Serranidae). Tulane Stud. Zool. 7:33-68. Paugam, A., M. Le Pennec, and A.-F. Genevieve 2000. Immunological recognition of marine bivalve larvae from plankton samples. J. Shellfish Res. 19: 325-331. Pegg, G. G., B. Sinclair, L. Briskey, and W. J. Aspden 2006. MtDNA barcode identification of fish larvae in the southern Great Barrier Reef, Australia. Sci. Mar. (Bare.) 70:7-12. Rafferty, C. S., S. R. Campbell, R. A. Wirtz, and M. Q. Benedict 2002. Polymerase chain reaction-based identification and genotyping of Anopheles mosquitoes with a 96-pin bacte- rial replicator. Am. J. Trop. Med. Hyg. 66:234-237. Richards, W. J. (ed) 2006. Early stages of Atlantic fishes: an identification guide for the western central Atlantic, 2640 p. CRC Press, Boca Raton, FL. Sedberry, G. R., and R. F. Van Dolah 1984. Demersal fish assemblages associated with hard bottom habitat in the South Atlantic Bight of the U.S. A. Environ. Biol. Fishes 11:241-258. Shivji, J. S., D. D. Chapman, E. K. Pikitch, and P. W. Raymond. 2005. Genetic profiling reveals illegal international trade in fins of the great white shark, Carcharodon carcharias. Conserv. Genet. 6:1035-1039. Sweijd, N. A., R. C. K. Bowiel, B. S. Evans, and A. L. Lopata. 2000. Molecular genetics and the management and conservation of marine organisms. Hydrobiologia 420:153-164 Taylor, D. L. 2004. Immunological detection of winter flounder ( Pseu - dopleuronectes americanus) eggs and juveniles in the stomach contents of crustacean predators. J. Exp. Mar. Biol. Ecol. 300:55-73. Thompson, J. D., T. J. Gibson, F. Plewniak, F. Jeanmougin, and D. G. Higgins 1997. The ClustalX windows interface: flexible strate- gies for multiple sequence alignment aided by quality analysis tools. Nucleic Acids Res. 24:4876-4882. Wenner, C. A., C. A. Barans, B. W. Stender, and F. H. Berry 1979. Results of Marmap otter trawl investigations in the South Atlantic Bight, I. Fall, 1973. S. Carolina Mar. Resources Center, Tech. Rep. no. 33, 79 p. Wenner, C. A., W. A. Roumillat, and C. W. Waltz 1986. Contributions to the life history of Black Sea bass, Centropristis striata, off the southeastern United States. Fish. Bull. 84:723-741. Wenner, C. A., and G. R. Sedberry. 1989. Species composition, distribution, and relative abun- dance of fishes in the coastal habitat off the southeastern United States. NOAA Tech. Rep. NMFS 79, 49 p. Abstract — Squaretail coralgrouper (Plectropomus areolatus) were cap- tured and tagged at a fish spawning aggregation (FSA) site with conven- tional and acoustic tags to assess their vulnerability to fishing and spatial dynamics during reproductive periods. Males outnumbered females in catch and, on average, were larger than females. Findings revealed a high vulnerability to fishing, par- ticularly during reproductive peri- ods, and most fish were recaptured within the 5-month spawning season and within 10-12 km of the aggrega- tion site. Individual and sex-specific variability in movement to, and resi- dency times at, the FSA site indicates that individual monthly spawning aggregations represent subsets of the total reproductive population. Some individuals appeared to move along a common migratory corridor to reach the FSA site. Sex-specific behavioral differences, particularly longer resi- dency times, appear to increase the vulnerability of reproductively active males to fishing, particularly within a FSA, which could reduce reproduc- tive output. Both fishery-dependent and fishery-independent data indi- cate that only males were present within the first month of aggrega- tion. The combined results indicate that reproductively active P. areolatus are highly vulnerable to fishing and that FSAs and migratory corridors of reproductively active fish should be incorporated into marine protected areas. The capture of P. areolatus during reproductive periods should be restricted as part of a comprehen- sive management strategy. Manuscript submitted 1 October 2007. Manuscript accepted 8 February 2008. Fish. Bull. 106:194-203 (2008). The views and opinions expressed or implied in this article are those of the author and do not necessarily reflect the position of the National Marine Fisheries Service, NOAA. The vulnerability of reproductively active squaretail coralgrouper ( Plectropomus areolatus ) to fishing Kevin L. Rhodes (contact author)1 Mark H. Tupper 2 Email address for Kevin L. Rhodes: klrhodes_grouper@sbcglobal.net 1 The University of Hawaii at Hilo, College of Agriculture Forestry and Natural Resource Management 200 W. Kawili St. Hilo, Hawaii 96720 2 WorldFish Center PO Box 500 GPO 10670 Penang, Malaysia Epinepheline serranids (groupers, hinds, lyretails) that form fish spawn- ing aggregations (FSAs) are among the most vulnerable coral reef fishes to overexploitation (Coleman et ah, 2000; Sadovy and Domeier, 2005). Indeed, 20 of 162 epinepheline ser- ranids (hereafter, serranids), includ- ing species that form FSAs, are now considered vulnerable or endangered according to IUCN Red List criteria. Among those listed as vulnerable is the squaretail coralgrouper ( Plectro- pomus areolatus ) (Riippell, 1830) that forms temporally and spatially pre- dictable FSAs, often numbering in the 100s to 1000s of individuals. These traits contribute to the targeting of P. areolatus by local and foreign commer- cial fisheries, including the Southeast Asia-based live reef food fish fishery (Sadovy et al., 2003). When both FSAs and reproductively active individu- als are targeted enroute to spawning sites, reproductive populations may experience population-level changes that affect reproductive output, includ- ing skewed aggregation sex ratios fol- lowing sexual selection (e.g., Koenig et al., 1996), size reductions (e.g., Beets and Friedlander, 1998), changes in genetic diversity (Chapman et al., 1999), and loss or decline in FSA (e.g., Sadovy and Domeier, 2005). With few exceptions, FSA fishing is unsustain- able under anything more than light fishing pressure, such as that char- acteristic of limited subsistence-level fishing (Sadovy and Domeier, 2005). Reports of grouper FSA loss continue to increase worldwide (e.g., Matos- Caraballo et al., 2006; Aguilar-Perera, 2007) and increased fishing mortality at FSAs is implicated in local fishery declines (Sadovy and Domeier, 2005). Among measures recently touted to protect FSAs and reproductive popu- lations from overfishing are marine protected areas (MPAs). However, few reports show the effectiveness of MPAs in limiting or preventing FSA loss, or in improving population-level abundance (Nemeth, 2005). Moreover, most existing FSA-based MPAs are small and likely leave reproductively active individuals vulnerable to fish- ing, such as some reproductively active serranids that appear to use common migratory corridors to reach FSAs, where they can be fished before they spawn (Nemeth et al., 2007; Starr et al., 2007). Migratory corridors and the dangers of overexploitation when corridors are left unprotected are recorded for a number of fishes, such as gadids (e.g., Rose, 1993). To design effective management for reproduc- tive populations of serranids such as squaretail coralgrouper, areas of high fishing vulnerability must be identi- fied, including FSA sites, migratory corridors, and other areas where fish may concentrate during reproductive periods. Within the tropical Pacific, only Pa- lau and Pohnpei have developed long- term management protocols at the national level specifically to protect Rhodes and Topper: The vulnerability of Plectropomus areolatus to fishing 195 commercially important serranids ( P . areolatus , Epi- nephelus fuscoguttatus [brown-marbled grouper], and E. polyphekadion [camouflage grouper]) during repro- ductive periods (Rhodes and Sadovy, 2002). Temporary community-based MPAs are used elsewhere, such as at the Solomon Islands and Papua New Guinea, whereby fishing is allowed at FSAs after periods of build-up within the aggregating population. For Pohnpei, where the current study was focused, management protocols include a small-scale MPA that protects P. areolatus, E. polyphekadion, and E. fuscoguttatus at a common FSA site (Kehpara Marine Sanctuary [KMS]) and a seasonal serranid sales ban that coincides with peak reproductive periods for these species (March-April). However, these measures fall short of fully protecting these and other FSA-forming serranids because 1) some unprotected FSAs continue to be fished (e.g., at nearby Ant Atoll and Palikir Pass); 2) sales bans do not cover entire spawning seasons for the aforementioned species; and 3) some serranids appear to reproduce outside the sales ban period (Rhodes and Tupper, 2007). Migratory corridors have not been investigated locally, including around the KMS, and thus reproductively active indi- viduals possibly exist outside current MPA boundaries and are, therefore, vulnerable to exploitation. Finally, subsistence catch for aggregating serranids, shown to lead to overexploitation elsewhere (e.g., Olsen and La- Place, 1979), remains open year round, including during the sales ban period. In Pohnpei, P. areolatus is a favored target of lo- cal subsistence fishermen and small-scale commercial fishermen, representing approximately 12% of the com- bined-gear commercial coral reef grouper fishery that includes 24 species (Rhodes and Tupper, 2007). The species is taken year-round largely as juveniles and small adults from inner reef areas during nighttime spearfishing activities, but is also known to be targeted at FSA sites, including at least three known unpro- tected FSA sites. Anecdotal evidence indicates that one local P. areolatus FSA may have been extirpated in the 1990s after continued targeting of an annual re- productive migration that ceased to form around 1995. Gravid P. areolatus are commonly found in markets during reproductive months outside the March-April serranid sales ban period, occasionally in the hundreds of individuals, and there is concern that reproductively active P. areolatus from the KMS may be targeted along migratory corridors or other areas where fish congre- gate (i.e., staging areas) for noncommercial use during sales-ban periods. Using both conventional and acoustic tag-recapture techniques, we examined the vulnerability of squaretail coralgrouper to fishing by examining spatial and tem- poral trends of tagged individuals at a single protected spawning aggregation site. The study objectives were 1) to determine potential sexual differences in behavior (within reproductive periods) that may facilitate capture and impact reproduction; 2) to examine the potential catchment area(s) (determined from recaptures) from which reproductive populations of P. areolatus at KMS may be drawn; 3) to assess the direction and distance of movement of fish in relation to the FSA to improve MPA design; 4) to identify potential migratory corridors of spawners in relation to KMS where fishermen may concentrate fishing activities; 5) to determine if the ex- isting MPA at KMS adequately protects reproductively active individuals; and 6) to assess whether current fisheries management could be improved to reduce the capture of reproductive fishes during spawning periods, particularly at the FSA site. Estimates of overall fish- ing mortality and fisheries sustainability for squaretail coralgrouper in Pohnpei are outside the scope of the current study because information on stock character- istics and the fishery (particularly subsistence) remain incomplete and, therefore, statements concerning these indices would be premature. Material and methods Site description The tagging study was conducted at the Kehpara Marine Sanctuary, a known protected FSA site in Pohnpei, Fed- erated States of Micronesia (6°55'N, 158°15'E), where P. areolatus, E. polyphekadion, and E. fuscoguttatus aggre- gate seasonally to spawn (Fig. 1). Plectropomus areolatus forms aggregations annually from ca. January-May at KMS, seaward of the barrier reef along approximately 0.5 km of the reef flat and wall at -10-30 m depth (first author, unpubl. data). During the aggregation period, individuals seek shelter and remain in relatively close proximity (<5 m) to the low-relief coral that covers much of the site. During daytime, individuals can be observed meandering among large high-relief stands of coral in shallower portions of the reef flat and over the inter- mittent sand patches and rubble piles that are found throughout the area. Although P. areolatus aggrega- tions have been monitored at the site over a 7-yr period, spawning has never been observed within aggregation months. Histological analyses of gonads have recently confirmed a seasonal and lunar periodicity of spawning for this species (first author, unpubl. data). Fish capture, catch per unit of effort, and tagging To capture fish for tagging, two locally hired fishermen used live bait ( Myripristis sp.), hook-and-line, and mask and snorkel to target P. areolatus from the surface within the FSA site. Fishing was conducted daily over 5 days in January and 7 days monthly from February through May just before and including the full moon, for a total of 33 fishing days. Target depths for fishing ranged from -15-30 m where fish aggregate within the FSA site. To estimate catch-per-unit-of-effort (CPUE), daily soak times and catch volumes were recorded from February to May. No CPUE estimates were made in Jan- uary, although fishing methods and times were similar among months. Following capture and before processing, fish were brought onboard and anesthetized in a 0.75 g/L 196 Fishery Bulletin 106(2) 158° 10' E 158°20'E N Pacific 10 kilometers Ocean I 1 L 158°06’50'E Receiver leoend ® Center of FSA ® KMS boundary O Others Figure 1 (A) Map of the Pohnpei Island showing the general catch locales (circles) and number (ital- ics) of recaptured conventionally tagged squaretail coralgrouper (Plectropomus areolatus). Catch locales are (counterclockwise from left) Dawak, Peleng Channel, Nalap, Liap, Penieu, and Temmen. Catch probabilities (in parentheses), or the number of recaptures anticipated within a municipal reef area, are based on the total number of recaptures from this study (January 2005-February 2006) divided by the percent fishing effort allocated to a municipal reef area. Effort estimates were derived from a 2006 market survey that included interviews of 1123 commercial reef fishermen (Rhodes et al., 2007). Estimates of catch probability exclude recaptures with unreported catch locales (n = 5) and fish captured by researchers inside the Kehpara Marine Sanctuary (KMS) (n= 20). Municipalities (Nett, Sokehs, Kitti, Madelonimw, Uh) are separated by black diagonal lines. (B) Inset map shows the direc- tion of movement and number of individuals detected by Vemco VR2 acoustic receivers outside the KMS during the study period. The Kehpara Marine Sanctuary is represented by the area outlined in gray. FSA=fish spawning aggregation. Others = other receivers. tricane methanesulfonate-seawater solution until fish lost equilibrium (—3—5 minutes). Following anesthesia and air bladder deflation, all individuals were weighed (nearest g body weight), measured (nearest mm total length [TL] and standard length [SL] ), and sex was determined macroscopically by using a 1-mm bore nylon cannula (Rhodes and Sadovy, 2002). To determine the potential distance of movement, times at liberty and catchment areas (defined as the area from which spawning individuals are drawn [Sa- dovy and Domeier, 2005]), all captured fish were tagged with a uniquely numbered Floy FT-1-94 conventional tag (Floy Tag, Seattle, WA) that provided contact and reward information. Of these Floy-tagged specimens, 40 fish (20 males and 20 females) were surgically im- planted with Vemco V16 acoustic transmitters (Vemco AMIRIX Systems, Halifax, Nova Scotia) in January and February. For tag implantation, abdominal incisions (—3.5 cm) were made just anterior to the vent and were closed with ConMed Reflex One® 35 Wide surgical skin staples (ConMed Endosurgery, Utica, NY) (Tupper and Able, 2000). After surgery, fish were allowed to recover 10-20 min onboard in fresh aerated seawater before release into shallow (2-5 m) water near the reef crest. Most acoustic-tagged and some Floy-tagged fish were observed from the surface by snorkel divers to follow initial recovery and to monitor potential predation. Acoustic tracking and tag recovery For acoustic tracking of fish distance and direction of movement, and to determine residency times within the FSA, a total of 7 Vemco VR2 receivers were moored in January 2005 at the following locations: center of the FSA, at north and south KMS boundaries, within the Kehpara and Peleng channels adjacent to the FSA, and seaward of Dawak and Nalap islands that are north and south, respectively, of channels adjacent to the FSA Rhodes and Tupper: The vulnerability of Plectropomus areolatus to fishing 197 Table 1 Summary table of sex-specific catch of squaretail coralgrouper (Plectropomus areolatus) at the aggregation site during tagging in 2005. N = total number of individuals collected during the reproductive season; n=monthly sample size; FDC = first day of catch before the full moon; na = data not available. Month Sex n January February March Apri i May FDC n FDC n FDC n FDC n FDC N Male 71 4 125 4 96 6 130 9 89 8 511 Female 0 na 32 3 38 5 37 5 22 5 129 Unknown 4 3 0 na 0 na 0 na 3 8 7 Total 75 157 134 167 114 647 (Fig. 1). All receivers were tethered above the barrier reef at 10-25 m depth at seaward-facing promontories to maximize detection range. All receivers were removed after 17 months that covered two full consecutive spawn- ing seasons. Prior range testing in Palau under similar physical and oceanographic conditions was performed for 13 receivers along four cardinal directions with a line- attached transmitter. The transmitter was towed behind the boat and away from the receivers, and times and distances from the receiver were recorded by GPS. Dis- tance and times were later matched to receiver-logged transmissions to determine minimum and maximum detection distances. Testing confirmed an average maxi- mum detection distance of 442 m along the reef parallel to the barrier reef crest and 975 m from the receiver seaward. Within the channel (here, Ulong Channel), the receiver detected tags at an average of 746 m parallel to the channel, while distance perpendicular to the channel was constrained by the channel width, similar to the constraints for detection distances from the receiver to the reef crest (perpendicular to the barrier reef). To enhance the potential for tag recovery, a reward scheme (US$5 plus fish market value) was broadcast island-wide on local AM radio and by flyer postings. Re- cords for catch location and gear type were taken from local fishermen at the time of reward. Recaptured fish were reweighed and measured, and gonads and otoliths were extracted for subsequent life history analyses. Analysis of catch probability A separate creel survey was conducted in 2006 to deter- mine the distribution of fishing effort within Pohnpei’s small-scale commercial coral reef fishery (Rhodes et al., 2007). The 2006 effort data were used to estimate the number of likely recaptures for P. areolatus under a random distribution (postspawning dispersal) scenario (chi-square analysis, after a square-root transformation) for each municipality. For this analysis, those fish cap- tured by researchers inside the KMS were excluded, as were an additional five individuals for which there was no recorded recapture location. Results Initial capture numbers, CPUE, and size distribution of fish Between January and May 2005, 647 P. areolatus were captured and tagged at the Kehpara Marine Sanctu- ary over 33 fishing days and 170 fishing hours, with an estimated CPUE of 3.8 fish per hour per fisherman, to highlight the vulnerability of reproductive P. areolatus to hook-and-line fishing at the aggregation site. Catch included 511 males, 129 females, and 7 individuals that could not be identified to sex with macroscopic methods. Males typically preceded females in catch by 1-4 days, except in January when no females were taken (Table 1). The average sex ratio for catch (February-May) was 3.4 males: 1 female, indicating either an absence of females within the January FSA or highly variable sex- specific differences in catch between January and other months. Late-stage vitellogenic oocytes were observed in all cannulated females, and no females were captured with hydrated oocytes (to indicate imminent spawn- ing), although subsequent histological examination of gonads from recaptured individuals confirmed evidence of spawning activity. All males were ripe at the time of capture during all months of tagging. Males captured at the FSA were, on average, larg- er than females. The size frequency range was 378-542 mm TL for mature females (mean ±standard error [SE] = 458.0 ±2.8 mm TL], and 450-660 mm TL for mature males (mean=541.4 ±1.3 mm TL) (Fig. 2). Only males were present in size classes above 550 mm TL, and only females were observed in size classes below 440 mm TL. Individuals of undetermined sex ranged from 480 to 575 mm TL (mean=523.3 ±14.6 mm TL) and thus overlapped in size with mature males. The mean size of captured males declined slightly during the survey, while captured female mean size increased. During tagging, similar behavior, including courtship and territorial displays, was observed between tagged and untagged fish to indicate minimal or no effects from tagging on at least some behavior commonly as- sociated with reproduction. 198 Fishery Bulletin 106(2) Conventional tagging The majority of conventional tag recaptures occurred within the January-May spawning season and thus indicated that P. areolatus are most vulnerable during reproductive periods, whereas the potential for sexual selection to reduce reproductive output was reflected in the male-dominated capture and recapture sex ratio and the high spawning-site fidelity shown by males within the reproductive season. Of the 647 P. areolatus tagged, 59 individuals (9.1% of the total) were recaptured (Table 2), including 39 individuals recaptured by commercial fishermen and 20 individuals recaptured at the KMS by researchers during tagging. All individuals recaptured at the KMS were males. Among the males recaptured at KMS, 50% were taken one month after tagging, 15% were taken after two months, 20% after three months, and 5% after 4 months. At the KMS, one male was recaptured twice in the same month and another male was taken three times over three separate months — a finding that demonstrates that at least some individuals return repeatedly within the spawning season. Out- side the KMS, fishermen recaptured 27 males, eight females, and one fish of unknown sex (at the time of tagging). Three additional recaptures by the fishery could not be identified either to sex (owing to illegibil- ity of the information on the tag) or to recapture loca- tion (which was unreported) or both sex and recapture location were not identified. The fishery recapture sex ratio (individuals of known sex taken outside the KMS) was 3.4:1 male:female, which matched the sex ratio of fish captured during initial tagging within the KMS (Table 1). Recaptured fish remained at liberty from 1 Size class (mm TL) Figure 2 Size frequency distribution (mm total length, TL) of square- tail coralgrouper ( Plectropomus areolatus , n = 647) taken by hook and line from the Kehpara Marine Sanctuary during the initial tagging exercise in 2005. White bars=females; black bars=males. to 328 days, averaging 71 days for those taken by the fishery and 51 days for fish retaken within the KMS by researchers. The maximum straight-line distance trav- eled by a recaptured fish was ~27 km between the KMS and Temmen (Fig. 1A). Squaretail coralgrouper appear highly vulnerable to fishing within and in areas immediately near the FSA site. A total of 97.1% of all fish were recaptured within 12 km of the FSA site during the 17-month survey period, and 88% of the recaptures with reported loca- tions were taken from areas north of the FSA. The fish- ery made 61.5% of all recaptures within the 5-month spawning season, and 25.6% of all recaptures occurred during the March-April sales ban period, which dem- onstrates the ineffectiveness of the ban in protecting reproductively active fish (Table 2). Reported recaptures by the fishery were primarily recaptured fish from the inner reef, including a substantial number from areas immediately adjacent to Peleng Channel (42% of all in- dividuals) that included 88% of all females and 64% of male recaptured. The distribution of fishery recaptures significantly deviated from expectations (%2=10.911, 0.255 days before a full moon, whereas males were taken as early as 8-9 days before a full moon, indicating sex-specific variations in feeding or catchability. Virtually all males acoustically tagged in January remained at the aggregation site until spawning was concluded in February, further enhancing the potential for sexual selection before actual spawning within the aggregation period. Few females appeared to remain at the FSA site after spawning during subsequent months. Only three individuals were present in all possible spawning months, and two males and three females returned to or resided at the KMS during nonreproduc- tive periods (Table 3). Discussion The vulnerability of FSAs to overfishing is widely recog- nized (Sadovy and Domeier, 2005) and the present study shows the conditions under which FSA loss and pos- sible population decline can occur. For example, under a hyperstable fishing scenario (no CPUE reduction in relation to abundance) and using CPUE estimates from this study (3.83 fish per hr per fisherman) and peak FSA abundance estimates for squaretail coralgrouper at KMS, only 250 fishing days at 6 h of fishing per day would be required to deplete the entire FSA, equivalent to only 36 fishermen fishing over 7 days. Such a sce- nario was borne out in 1998 when an estimated 4000 reproductively active fish were taken from the adjacent camouflage grouper FSA in just over 7 days (Rhodes and Sadovy, 2002). During that event, up to 20 boats with 2-3 fishermen/boat fished the aggregation daily with no observed change in CPUE, before the aggregation was included into the KMS by executive order in 1999. Underwater visual census (UVC) counts in later years showed that only several hundred camouflage grouper remained at the camouflage grouper FSA site and that only minor increases in abundance were apparent after 7 years of monitoring (2001-07). Similarly heavy aggrega- tion fishing has resulted in either loss, near-extirpation, or a substantial reduction in FSA abundance elsewhere (e.g., Craig, 1969; Sadovy and Eklund, 1999; Sala et al„ 2001). Migratory corridors used by reproductive fish increase the vulnerability of P. areolatus (and other fishes) to overfishing by concentrating reproductively active or resting fish within confined areas similar to the FSA in our study. Although it is not clear whether all P. areolatus (at KMS or elsewhere) use these migratory corridors, several small groups of squaretail coralgrou- per (5-10 individuals, presumably males) were observed moving southward along the reef between Peleng Chan- nel and KMS and groups of up to 100 females (identi- fication based on size and color) were observed moving in deeper water (20-30 m) toward the FSA during the spawning season. Groups of roving female squaretail coralgrouper have also been reported near FSA in the Solomon Islands during reproductive periods. Other ser- ranids reported or suggested to use migratory corridors to reach FSAs include Epinephelus guttatus (red hind) in the U.S. Virgin Islands (Nemeth, 2005), P. areolatus in Palau, and Epinephelus striatus (Nassau grouper) in Belize (Starr et al., 2007). In each case, fishermen apparently target the unprotected areas where fish con- gregate, thereby reducing the effectiveness of existing MPAs to protect reproductive individuals, as observed in Pohnpei. The apparent targeting of these migratory corridors and the ease by which fishermen can remove individuals from these and other areas where fish con- gregate highlights the need to identify and incorporate those areas into MPAs. The observed sex-specific differences in the behavior of P. areolatus within the reproductive season have the potential to promote selective fishing and to impact the spawning sex ratio and reproductive output (e.g., Beets and Friedlander, 1998). In Pohnpei, sex-specific behavioral differences were manifested as 1) a greater abundance of males in FSA-catch both seasonal and monthly, particularly in January when no females were captured; 2) the persistence of males at the FSA over several weeks during the initial aggregation period (January-February); and 3) longer male residency times at the FSA seasonally. Similar to the current study, previous studies of FSA-forming serranids have also shown that males often precede females to the FSA and, therefore, have longer residency times (Samoilys, 1997; Rhodes and Sadovy, 2002; Starr et al., 2007). For squaretail coralgrouper, longer residency times appear to increase the catchability of males. Alternatively, males may outnumber females at the FSA site in all months and be reflected in the catch sex ratio. Further fisheries-independent assessments of aggregation (and population) sex ratio are needed to confirm whether males indeed outnumber females. Acoustic data also indicate that in addition to taking bait over more days than females, male squaretail coralgrouper frequent the FSA over more months — a finding that is similar to those from acoustic surveys of Nassau grouper in Belize (Starr et al., 2007) and UVC monitoring of freeze-brand tagged leopard coralgrouper ( P . leopardus) along the Great Barrier Reef (Zeller, 1998). Fisheries-dependent size data from FSA catch have also indicated protracted residency times for male red hind in the U.S. Virgin Islands (Nemeth, 2005). Protracted male residency time may increase the potential for sexual selection, leading to adverse affects on reproduction, such as sperm limi- Rhodes and Tupper: The vulnerability of Plectropomus areolatus to fishing 201 tation (Coleman et al., 1999). Similarly, the removal of larger individuals (males in protogynous species) may lead to reductions in the mean size of both males and females (from compensated sex change) to reduce overall fecundity (Vincent and Sadovy, 1998) and, in the case of extreme operational sex ratios, may adversely affect re- productive behavior. At the KMS FSA, males appeared to be more vulnerable to line fishing than were females, comprising 79% of the catch during tagging operations and representing all recaptures within the KMS. Males also dominated recaptures by the fishery among spear- caught individuals. Interestingly, females and juvenile squaretail coralgrouper dominated marketed catch of squaretail coralgrouper overall, and 97% of the mar- keted catch in 2006 was smaller than the mean size of sexually mature males (Rhodes and Tupper, 2007). Thus, in Pohnpei, the potential currently exists for both growth overfishing and recruitment overfishing and there is an urgent need for a thorough examination of island-wide stock levels and sustainability within the squaretail coralgrouper fishery, as well as the sustain- ability of both the subsistence and other fisheries tar- geting FSAs. One possible explanation for the paucity of females in catch is that recaptured females went unreported or that fishermen targeted larger individu- als (i.e., males) to benefit from the combined higher fish value and reward. An improvement to our study would be to perform real-time observations of the fishery at heavily fished locations, such as Peleng Channel, to verify the sex ratio of the catch. Among acoustically tagged individuals, the data showed a high degree of variation in the seasonal pat- tern and duration of stay for individuals that frequent the FSA site. Although these results could be explained in part by mortality (natural, fishing, and tag-induced, particularly acoustic) or a failure of receivers to detect all fish present, similar variability has been shown elsewhere for serranids with the use of these and other tagging methods (e.g., Zeller, 1998; Starr et al., 2007). In Pohnpei, a gradual reduction in the number of acous- tically tagged individuals returning to the FSA was observed within a reproductive year; only 50% of tagged individuals returned in the month after initial tagging. Similar findings were found for acoustically tagged Nassau grouper in Belize, where no more than 50% of tagged E. striatus males and 52% of tagged females were present during any reproductive month following tagging (Starr et al., 2007). Nassau grouper returning to the FSA after having been tagged showed a similar sequential decrease in numbers. On the Great Barrier Reef, only 31% of tagged mature leopard coralgrouper participated in spawning aggregation activities within a reproductive year; this percentage may indicate that individuals do not spawn annually (Zeller, 1998). Al- though Zeller (1998) suggested that tagged individuals may reproduce outside FSAs, spawning outside aggre- gations for FSA-forming species has not been shown. Likewise, high site fidelity has been shown for a num- ber of serranids, with individuals observed moving past an active conspecific FSA to return to the site where they previously spawned. Sequential reductions in the number of conventionally tagged red hind have also been shown in the U.S. Virgin Islands (Nemeth et al., 2007). Although it is possible that the observed find- ings could be explained by natural or fisheries-related mortality, it seems more likely that some adults do not participate in all potential spawning months within the reproductive season and that some may not spawn in consecutive years. Although more investigations are needed, if these implications are correct, individual (monthly) FSAs represent subsets of the reproductive population, and these subsets complicate monitoring programs that solely use FSAs to estimate reproductive population abundance and changes therein. Regardless, these results indicate that the relationship between a FSA and the adult population is complex and additional work is needed on serranid reproductive dynamics for designing effective monitoring and management strate- gies for adult populations and in understanding impacts on these populations through FSA-targeted fishing. The assessment of catchment areas for aggregating serranids by using tagging methods is relatively new, yet necessary to complete our understanding of repro- ductive population dynamics and to determine sustain- ability when fishing efforts are unevenly distributed. In Pohnpei, support for a catchment area of -200-300 km2 for the KMS-based squaretail coralgrouper FSA is pro- vided from recapture data that indicates short-scale (10-12 km) movement of individuals in relation to the FSA after spawning. The observed recapture patterns do not appear to be strictly tied to area-specific fishing effort because Kitti municipality receives only 50% of fishing efforts statewide, yet >95% of recaptures oc- curred within a relatively small area of the municipal- ity (Rhodes et al., 2007). Although it is possible that fishing effort during spawning months (and full moon periods) is more concentrated in Kitti than at other municipalities, such a pattern was not borne out from over 1000 interviews (2006) of fishermen participating in commercial fishing activities. Thus, for squaretail coralgrouper in Pohnpei, the reproductive population in relation to the KMS spawning aggregation appears to be highly localized and thus management could be ap- plied to individual aggregations, in addition to combined aggregations within Pohnpei. Similar to sex-specific residency patterns, other studies of FSA-forming ser- ranids also indicate that adults are concentrated within relatively small areas in relation to their respective FSA. For example, evidence exists for confined catch- ment areas for red hind determined from two FSAs in the U.S. Virgin Islands: 1) 500 km2 for the St. Thomas FSA and 2) 90 km2 for the St. Croix site FSA (Nemeth et al., 2007). The catchment area for Nassau grouper at the Glover’s Reef Atoll FSA (Belize) appears to be no more than 384 km2 (Starr et al., 2007). For P. leopardus along the Great Barrier Reef, Zeller (1998) estimated a catchment (migration) area of ca. 80 km2. For these spe- cies, data clearly indicate limited movement by (most) individuals after spawning and highlight the potential for localized extinction and associated impacts to fish 202 Fishery Bulletin 106(2) community structure and fisheries from FSA fishing (or unsustainable fishing of individuals at earlier life his- tory stages). The data also provide additional support to suggest that, while possible, long distance movement by reproductively active serranids may be more rare than previously assumed and that many aggregating grouper are resident to areas near the spawning site. The effectiveness of the no-take year-round KMS to protect P. areolatus FSAs is evident, because without this protection, thousands of groupers could be removed within the spawning season and there would be the potential for rapid FSA loss and localized population ex- tinctions. Regardless of the effectiveness of the KMS in protecting spawners at the spawning site, the potential for the fishery to impact the reproductive individuals still exists. Spawners are being taken along migratory corridors or other areas where fish congregate outside the KMS, with the potential to meet or exceed the level of fishing mortality experienced under aggregation fish- ing. In Pohnpei, problems with the existing sales-ban strategy are now evident. Firstly, 36% of all recaptured fish are taken during serranid sales ban months. Sec- ondly, the serranid sales ban was recently shown to place added pressure on other equally vulnerable com- mercial species not currently managed, such as scarids (Rhodes et al., 2007). The increase in catch volume of scarids, for example, implies that fishermen have greater concern for maintaining catch volume than for the sur- vival of individual species and indicates that efforts should be concentrated on measures to reduce overall catch volume. Similar reductions may be needed in catch volume of juveniles and small adults to stem potential recruitment overfishing. Thus, effective protection of reproductively active serranids will require innovative solutions that provide protection for both reproductive and nonreproductive individuals, while alleviating the potential negative effects of individual species manage- ment. The results from this and related studies highlight the complexities of reproductive population dynamics for aggregating serranids and support total protection through no-take bans during reproductive periods and no-take areas that support aggregations and associated areas where reproductive individuals congregate. Future studies should be designed to fill the gaps in our current understanding of serranid stocks in Pohnpei to enable a determination of sustainable harvest levels. Numerous examples globally indicate that the current harvest of reproductive adults and targeting of juveniles is cur- rently unsustainable and that population declines and aggregation losses are imminent without a rapid move toward improved management. Acknowledgments The Conservation Society of Pohnpei and Pohnpei Department of Lands and Natural Resources provided logistic support and in-kind contributions. P. Dixon, S. Malakai, D. Mathias, P. Moses, and C. B. Wichilmel pro- vided dive and tag assistance. D. Paul provided histori- cal fishing accounts. Funding was provided by a National Oceanographic and Atmospheric Administration Coral Reef Conservation Grant (NA04NMF4630341). Critical reviewer comments greatly improved the manuscript. This article is dedicated to the memory of D. Paul. Literature cited Aguilar-Perera, A. 2007. Disappearance of a Nassau grouper spawning aggregation off the southern Mexican coast. Mar. Ecol. Prog. Ser. 327:289-296. Beets, J., and A. Friedlander. 1998. Evaluation of a conservation strategy: a spawning aggregation closure for red hind, Epinephelus gutta- tus, in the U.S. Virgin Islands. Environ. Biol. Fishes 55:91-98. Chapman, R. W., G. R. Sedberry, C. C. Koenig, and B. E. Eleby. 1999. Stock identification of gag, Mycteroperca microlepis, along the southeast coast of the United States. Mar. Biotechnol. 1:137-146. Coleman, F. C., C. C. Koenig, A.-M. Eklund, and C. B. Grimes. 1999. Management and conservation of temperate reef fishes in the grouper-snapper complex of the southeast- ern United States. In Life in the fast lane: ecology and conservation of long-lived marine animals (J. A. Musick, ed.), p. 233-242. Am. Fish. Soc., Bethesda, MD. Coleman, F. C., C. C. Koenig, G. R. Huntsman, J. A. Musick, A. M. Eklund, J. C. McGovern, R. W. Chapman, G. R. Sed- berry, and C. B. Grimes. 2000. Long-lived reef fishes: the grouper-snapper complex. Fisheries 25:14-20. Craig, A. K. 1969. The grouper fishery of Cay Glory, British Honduras. Ann. Assoc. Am. Geogr. 59:252-263. Matos-Caraballo, D., J. M. Posada, and B. E. Luckhurst. 2006. Fishery-dependent evaluation of a spawning aggre- gation of tiger grouper (Mycteroperca tigris) at Vieques Island, Puerto Rico. Bull. Mar. Sci. 79(1):1-16. Nemeth, R. 2005. Population characteristics of a recovering US Vir- gin Islands red hind spawning aggregation following protection. Mar. Ecol. Prog. Ser. 286:81-97. Nemeth, R. S., J. Blondeau, S. Herzlieb, and E. Kadison. 2007. Spatial and temporal patterns of movement and migration at spawning aggregations of red hind, Epi- nephelus guttatus, in the U.S. Virgin Islands. Environ. Biol. Fishes 78:365-381. Olsen, D. A., and J. A. LaPlace. 1979. A study of a Virgin Islands grouper fishery based on a breeding aggregation. Proc. Gulf Caribb. Fish. Inst. 31:130-144. Rhodes, K. L., and Y. Sadovy. 2002. Temporal and spatial trends in spawning aggrega- tions of camouflage grouper, Epinephelus polyphekadion (Bleeker 1849) in Pohnpei, Micronesia. Environ. Biol. Fishes 63:27-39. Rhodes, K. L., and M. H. Tupper. 2007. A preliminary market-based analysis of the Pohn- pei, Micronesia, grouper (Serranidae: Epinepheline) fishery reveals unsustainable fishing practices. Coral Reefs 26:335-344. Rhodes, K. L., M. H. Tupper and C. B. Wichilmel. 2007. Characterization and management of the com- Rhodes and Tupper: The vulnerability of Plectropomus areolatus to fishing 203 mercial sector of the Pohnpei coral reef fishery, Micronesia. Coral Reefs, DOI (digital object identi- fier) 10.1007/x00338-007-0331. Rose, G. A. 1993. Cod spawning on a migration highway in the north- west Atlantic. Nature 366:458-461. Sadovy, Y. J., and M. Domeier. 2005. Are aggregation-fisheries sustainable? Reef fish fisheries as a case study. Coral Reefs 24:254-262. Sadovy, Y., and A. M. Eklund. 1999. Synopsis of biological data on the Nassau grouper, Epinephelus striatus (Bloch, 1792), and the jewfish, E. itajara (Lichtenstein, 1822). NOAA Tech. Rep. NMFS 146, 68 p. Sadovy, Y. J., T. J. Donaldson, T. R. Graham, F. McGilvray, G. J. Muldoon, M. J. Phillips, M. A. Rimmer, A. Smith, and B. Yeeting. 2003. While stocks last: the live reef food fish trade, 147 p. Asian Development Bank Pubis., Manila, Phil- ippines. Sala, E., E. Ballasteros, and R. M. Starr. 2001 . Rapid decline of Nassau grouper spawning aggrega- tions in Belize: Fishery management and conservation needs. Fisheries 26:23-30. Samoilys, M. A. 1997. Movement in a large predatory fish: coral trout, Plectropomus leopardus (Pisces: Serranidae), on Heron Reef, Australia. Coral Reefs 16:151-159. Starr, R. M., E. Sala, E. Ballesteros, and M. Zabala. 2007. Spatial dynamics of the Nassau grouper Epineph- elus striatus in a Caribbean atoll. Mar. Ecol. Prog. Ser. 343:239-249. Tupper, M., and K. Able. 2000. Habitat use, movements, and food habits of striped bass ( Morone saxatilis ) in Delaware Bay (USA): comparisons between a restored and a reference salt marsh. Mar. Biol. 137:1049-1058. Vincent, A. C. J., and Y. Sadovy. 1998. Reproductive ecology in the conservation and management of fishes. In Behavioural ecology and conservation biology (T. Caro, ed.), p. 209-245. Oxford Univ. Press, New York, NY. Zeller, D. C. 1998. Spawning aggregations: patterns of movement of the coral trout Plectropomus leopardus (Serranidae) as determined by ultrasonic telemetry. Mar. Ecol. Prog. Ser. 162:253-26. 204 Mitochondrial DNA markers to identify commercial spiny lobster species ( Panulirus spp.) from the Pacific coast of Mexico: an application on phyllosoma larvae Francisco J. Garcia-Rodriguez1-3 German Ponce-Diaz1 3 Isabel Munoz-Garcia2 Rogelio Gonzalez-Armas3 Ricardo Perez-Enriquez1 (contact author) Email address for R. Perez-Enriquez: rperez@cibnor.mx 1 Centro de Investigaciones Biologicas del Noroeste (CIBNOR) Mar Bermeio 195, Col. Playa Palo de Santa Rita La Paz, Baha California Sur 23090, Mexico 2 Facultad de Ciencias del Mar, Universidad Autonoma de Sinaloa Apdo. Postal 610 Mazatlan, Sinaloa 82000, Mexico 3 Centro Interdisciplinary de Ciencias Marinas-lnstituto Politecnico Nacional (CICIMAR-IPN) Apdo Postal 592 La Paz, Baha California Sur 23000, Mexico Molecular markers based on mitochon- drial DNA (mtDNA) are extensively used to study genetic relationships. mtDNA has been used in phylogenetic studies to understand the evolutionary history of species because it is mater- nally inherited and is not subject to genetic recombination (Gyllensten et al., 1991). The high mutation rate of mtDNA makes it a useful tool for dif- ferentiating between closely related species (Brown et ah, 1979) — a tool that is especially important when significant variations occur between species, but not within species (Hill et al., 2001; Blair et al., 2006; Chow et al., 2006a). Species-level identification, based on molecular markers, can be very useful when morphological features alone do not provide sufficient dif- ferentiation or when only part of an organism is recovered. In fact, a few authors have successfully applied ge- netic markers for species identifica- tion based on remains recovered from fecal samples, parts of specimens, processed products, or larval and ju- venile forms (Chan et al., 2003; Pur- cell et al., 2004; Hsieh et ah, 2007). Three commercial lobster species inhabit the Pacific coast of Mexico: California spiny lobster ( Panulirus interruptus, west coast of California and the Baja California Peninsula); blue spiny lobster ( P . inflatus, south- ern Baja California Peninsula to the State of Oaxaca, Mexico); and green spiny lobster (P. gracilis, a tropical species from southern Baja California Peninsula to Peru) (Hendrickx, 1995). Taxonomic identification of adult lob- sters is easily done by morphological features (Hendrickx, 1995); however, alternative techniques are required for identification of larvae when mor- phological features are unable to pro- vide the means of identifying early life stages of spiny lobster species (Johnson, 1971; Munoz-Garcia et al., 2004). Furthermore, discrimination between larvae of P. inflatus and P. gracilis could be especially difficult because of their overlapping distri- bution. The species-specific identification of larvae of other Panulirus species has also been difficult. Chow et al. (2006b) found intraspecific and in- tra-individual variation in appendage structures (the subexopodal spines) in the phyllosoma of P. ornatus (or- nate rock lobster) and P. versicolor (painted spiny lobster). Therefore, su- bexopodal spine arrangements may not be a useful diagnostic for distin- guishing between these two species. Because of these potential problems, several authors have suggested the use of molecular markers to identify spiny lobster larvae (Silberman and Walsh, 1992; Chow et al., 2006a, 2006b; Konishi et al., 2006). In this study, the nucleotide varia- tions of the mitochondrial DNA (mtD- NA) in adult lobsters were investigat- ed to obtain genetic markers useful in identifying P. interruptus, P. in- flatus, and P. gracilis through either the polymerase chain reaction-restric- tion fragment length polymorphism (PCR-RFLP) analysis or species-spe- cific primers that amplify different size fragments in a multiplex reac- tion. These techniques were used to identify phyllosoma larvae col- lected outside the Gulf of California. Materials and methods PCR-RFLP analysis DNA from adult specimens of the three species was taken from the following sites on the west coast of Mexico: P. interruptus from Baja Cali- fornia (n= 2) and Baja California Sur (n- 2); P. inflatus from Baja Califor- nia Sur (n = 3), Sinaloa (n- 4), Nayarit (n- 2), and Jalisco (n=l); and P. graci- lis from Baja California Sur (??= 2), Sinaloa (n = 3), and Nayarit (n=l) (see Table 1). A fragment of the 16S rRNA gene was amplified with primers 16Sar-L (5'-CGCCTGTTTATCAAAAACAT) and 16Sbr-H (5'-CCGGTCTGAACT- CAGATCACGT) (Palumbi, 1996). Manuscript submitted 7 June 2007. Manuscript accepted 11 December 2007. Fish. Bull. 106:204-212 (2008). The views and opinions expressed or implied in this article are those of the author and do not necessarily reflect the position of the National Marine Fisheries Service, NOAA. NOTE Garcia-Rodriguez et al.: Mitochondrial DNA markers to identify Panulirus spp. 205 Table 1 Sampling sites and dates for each lobster species along the Pacific coast of California and Mexico. The number of samples in the Seql-16S column refers to the sequences used to find diagnostic restriction sites to discriminate lobster species; RFLP and Multiplex columns are the number of larvae analyzed by the RFLP and multiplex analysis, respectively; Seq2-16S column are the sequences used to evaluate consistency of restriction patterns. Number of samples Sampling RFLP Multiplex Sampling sites Location date Seql-16S 16S 12S-CR Seq2-16S Panulirus interruptus (California spiny lobster) Catalina Island, CA, USA 33°26', 118°29' Mar 2003 1 Ensenada, B.C., Mexico 31°50', 116°38' Aug 1999 1 3 1 Isla Guadalupe, B.C., Mexico 29°00', 118°10' Dec 2002 1 3 1 Punta Eugenia, B.C.S., Mexico 27°49', 115°06' Jun 1999 2 1 Punta Abreojos, B.C.S., Mexico 24°41', 113°34' Jun 1999 2 1 San Juanico, B.C.S., Mexico 24°14', 112°27' Oct 1999 2 1 Bahia Magdalena, B.C.S., Mexico 24°46', 112°06' Dec 1999 2 2 1 Panulirus inflatus (blue spiny lobster) Bahia Magdalena, B.C.S., Mexico 24°46', 112°06' Nov 2001 3 3 1 21 Mazatlan, Sin., Mexico 23°13', 106°26' Aug 2002 4 3 1 14 Las Penitas-Sayulita, Nay., Mexico 21°00', 105°23' Aug 2002 2 3 1 24 Barra de Navidad, Jal., Mexico 19°12', 104°42' Jan 2005 1 3 1 15 Zihuatanejo, Gue., Mexico 17°37', 101°33' May 2005 2 1 29 Puerto Angel, Oax., Mexico 15°39', 96°29' Nov 2002 1 23 Panulirus gracilis (green spiny lobster) Bahia Magdalena, B.C.S., Mexico 24°46', 112°06’ Nov 2001 2 Mazatlan, Sin., Mexico 23°13’, 106°26' Aug 2002 3 4 2 21 Las Penitas-Sayulita, Nay., Mexico 21°00', 105°23' Aug 2002 1 5 1 23 Punta Maldonado, Gue., Mexico 16°20', 98°34' May 2005 5 2 25 Total 20 42 18 195 PCR was performed in a total volume of 25 ,uL (Invitro- gene lx PCR buffer, 0.2 mM dNTP mix, 0.48 pM of each primer, 4.0 mM MgCl2, 1.25 U Taq DNA polymerase), with an iCycler thermal cycler (Bio-Rad Laboratories, Hercules, CA). The PCR program consisted of a dena- turation step at 94°C for 2 min, followed by 30 cycles of 1 min at 94°C, 1 min at 60°C, and 2 min at 72°C, followed by a final extension step of 4 min at 72°C. PCR products were confirmed using electrophoresis on 1% agarose gel, along with a molecular weight marker to estimate the fragment size. The gel was stained with SybrGold (Molecular Probes, Eugene, OR). All ampli- fied products were sequenced with primers 16Sar-L and 16Sbr-H (Macrogen, Korea) and deposited in GenBank (accession numbers EF546597 through EF546616). The 16S rRNA gene sequences were aligned and ed- ited with the program Sequencher, vers. 4.5. (Gene Codes Corporation, Ann Harbor, MI). A neighbor-joining phylogram of haplotypes, based on the Kimura-2 param- eter model, was constructed in Molecular Evolutionary Genetics Analysis (MEGA) software vers. 3.0 (Kumar et al., 2004). Three sequences (one for each species) were used for the identification of diagnostic restriction enzyme recognition sites in ChromasPro software, vers. 1.33 (Technelysium Pty Ltd, Tewantin, Queensland, Australia). Even though several restriction enzymes revealed various cut sites in the sequences, we selected only those that were easily seen on agarose gel. To check for consistency of the restriction enzyme cutting pattern in each species, the analysis was done with 17 additional sequences. The RFLP pattern of the selected enzymes of another 42 adult lobster specimens of the three species was tested for specimens collected at different locations (Table 1). Each digestion reaction occurred in a final 7pL volume with 0.7 unit/pL of the selected enzyme, 3.5 pL PCR product, and according to the manufactur- er’s instructions (New England Biolabs, Ipswich, MA). Reaction products for each enzyme were incubated for eight hours at the optimal temperature suggested by the manufacturer. For electrophoresis, restricted prod- ucts were run on 2% agarose gels and stained with SybrGold. A molecular weight marker was added to the agarose gel to estimate fragment size. To check con- sistency of intra- and interspecific nucleotide variation in addition to the 20 sequences, 195 sequences (a total 206 Fishery Bulletin 106(2) of 215) from different geographical sites of P. inflatus (n- 126) and P. gracilis (n = 69) were analyzed. This extension was important because the larvae of these species are found in the same area and there is a higher possibility that misidentification will occur when using only morphological criteria (Table 1). Panulirus penicillatus (pronghorn spiny lobster) is found along the Pacific coast of Mexico and even though it is found in small numbers, it could be confused with the larvae of the other species. The restriction pattern of a 16S rRNA gene sequence reported in the GenBank (accession number AF337974) was compared to those of the other three lobster species. Multiplex PCR analysis Species-specific primers for P. interruptus have been described in a previous study (Garcia-Rodriguez and Perez-Enriquez, 2006; GenBank accession number EF565146). Based on this sequence, a reverse primer (5'- TGGTGTGATCCCGTTACTTG) was designed to amplify a -1250 base-pair (bp) mtDNA fragment containing the 12S rRNA gene and the control region in P. inflatus and P. gracilis by means of Weider et al.’s (1996) forward primer (srRNA: 5 -CAGGGTATCTAATCCTGGTT). PCR thermal cycling consisted of an initial denaturation of 2 min at 94°C, 35 cycles of 1 min at 94°C, 1 min at 58°C (for P. inflatus, but 52°C for P. gracilis), 2 min at 72°C, and a final cycle of 4 min at 72°C. PCR products from adult specimens of each species were sequenced by using the srRNA primer. The Primer3 program (Rozen and Skaletsky, 2000) was used to design specific prim- ers for each species; these primers would have similar melting temperatures but products of different sizes (Fig. 1). Sequences obtained with specific primers were deposited in GenBank (accession numbers EF565144 and EF565145). The specificity and reliability of the multiplex PCR reaction with 18 previously identified adult lobsters collected in different regions were tested. For P. inter- ruptus ( n-1 ), specimens were obtained from California, Isla Guadalupe, Baja California, and Baja California Sur; for P. inflatus (n- 6), specimens were collected from Baja California Sur, Sinaloa, Nayarit, Jalisco, Guerrero, and Oaxaca; for P. gracilis, specimens (n = 5) came from Sinaloa, Nayarit, and Guerrero (Table 1). Multiplex PCR reactions were carried out in a total volume of 12.5 pL, mixing 0.48 pM of each primer (one common primer: srRNA, and the three species-spe- cific primers: LanCR-R ( 5 -AAAAATTCAGGCTAAT- GGA), PinRCl-b (5'-GATGGCCCATTACCGAACTA), and PgraRCl-b (5 -TTGTGAAACGTCTGTTTACATT- TATTT)), Invitrogene lx PCR buffer, 0.2 mM dNTP mix, 4.0 mM MgCl2, and 0.625 U Taq DNA polymerase. PCR thermal cycling consisted of an initial denatur- ation of 2 min at 94°C, followed by 30 cycles of 1 min at 94°C, 1 min at 59°C, 2 min at 72°C, and a final cycle of 4 min at 72°C. The amplified products from each species were distinguished by electrophoresis on 1.0 % agarose gels. Application of molecular markers Phyllosoma larvae were collected during an oceano- graphic cruise outside the Gulf of California in November 2004. Plankton sampling consisted of horizontal surface tows of a neuston collection net at 3.5 knots (6.4 km/h) for 5 min (see Gonzalez-Armas et ah, 1999). Sampling gear consisted of a rectangular plankton net of Nytex with 505-pm mesh; it had a 30x50 cm mouth area and the net was 3 m long. Phyllosoma larvae were sorted by hand after each tow and fixed with 70% ethanol. A fragment of the pereiopods, antenna, or eyes was obtained from each lobster larva for DNA isolation by lysing the tissue in 15 uL lysis buffer (10 mM Tris- HCL, pH 8.3, 50 mM KC1, 0.5% Tween-20), and 1.87 pL (4 pg/pL) proteinase K and incubated overnight at 55°C. Each reaction was then maintained at 95°C for approximately 10 min and stored at 4°C until analy- sis. Steps of PCR-RFLP and PCR multiplex for larval lobsters were the same as those carried out for adult specimens. Before genetic analysis, 46 lobster larvae collected from the Pacific Ocean were categorized into four pos- sible groups based on morphological criteria (Johnson and Knight, 1966; Johnson, 1971; Baez, 1983): group 1) Panulirus inflatus- like (n=17); group 2) Panulirus graci- Zis-like {n = 8); group 3) Panulirus inflatus-gracilis-like (n =12); and group 4) Panulirus- like (n= 9). Results PCR-RFLPs The 16S rRNA gene fragment was correctly amplified in the three lobster species. The size of the 16S rRNA gene fragment amplified in the three spiny lobster spe- cies was estimated at 563 bp and did not differ in size among species. There was wide inter- and intraspecific variation in nucleotide sequences among the lobster spe- cies, indicating the potential for species discrimination. A haplotype tree showed the aggregation of haplotypes according to each species (Fig. 2). From the analysis of 20 sequences, two restriction enzymes (BsmAl, GTCTCN' and Hinfl, GANTC) were selected that allowed discrimination among the three species (Fig. 3). The restriction products of the BsmAl digests were two fragments (-401 and -162 bp) in both P. interruptus and P. inflatus (named “haplotype A”) and three fragments (-401, -115, and -47 bp) in P. gracilis (named “haplotype B”). Restriction digests that used Hinfl produced two fragments (-440 and - 123 bp) for P. interruptus and most P. gracilis speci- mens (named “haplotype A”), but did not cut P. infla- tus (named “haplotype B”). One P. gracilis sequence appeared to be haplotype B due to the absence of the Hinfl site. The composite haplotypes were constructed by the combination of the haplotype names of each enzyme, resulting in AA for P. interruptus, AB for P. inflatus, and either BA (98.7%) or BB (1.3%) for P. grac- NOTE Garcia-Rodriguez et aL: Mitochondrial DNA markers to identify Panulirus spp. 207 P. ine-ENSl P. inf-JALl P.gra-GUE3 1 TACAGTTACTTGATTTACTAACACTTGCCAATATACAAATTTCACCTCTTATTGTTTCCAGTACTGAATATAAATG-AACTTAA--TTGTCTAACTTTATAA ... T ........ . ....A A A.A.TA.AC..AT...C.TT..TT CGGT . CA . T . . . . C . . TAC . CT . GCGC . C . A . .TT. . .-.T. . . . . .A T . A A.A.TG. .C. .AT. . .-. .T. .TC CGTT . CA . T . . . . C . . TAT . P . ine-ENSl P. inf- JAL1 P.gra-GUE3 103 CTTACTCTTTTTCACTGGGCCTTCAAGTCTAACCGCGGATGCTGGCACAAGTTTTAACCTGACCTAAATTGTTTTTACTTAATCCAATCTTTATTGATT-AT . . . .T. .ACC A. .G C G.G.A.GC. . .G CA. . .ATC T. . . . . .TC.ACC A. .A. A. .G C A.GC. . .G CAT. .AT T. . P. ine-ENSl P. inf-JALl P.gra-GUE3 205 TAAATTTTTAGTACTGCTCATGTAAATTATAGCGTTTAGCTTATTATTCGA CTCACTTTAATATTGAATTAAAGTGCCTATATTTTAATGTACT A A.C.C... .AAT.GTA. . . C . CAC . -CC.A. . . CGTCGCACC . C . CTC . . CG . . . . T . . CG . . CAGTG . .G. .AA. . . -CAA. . . A A.T.TC. . . AAT . . TA . .C.TTAC. . CC . . . TT . - - CACACC . ATTT . CCCG . . . CT . . CG . .C.GTG.C. . .AA. . .-CAA. . . P. ine-ENSl P. inf-JALl P. gra-GUE3 307 AATATTTGCATGTACTATTCAAACCAAATAGCGAAAAGAAATAGCAAGAATCAAACTATAGGACCCTACCAACTTATATTTTTATACTTTAATATTGATTTT ..A T.T G....C. .AT- . . . G . T TA.ATG. .CAA. . . - . G . CC . C . TCGACCC . . . - . .AAA ..A T.T G AT-...G.T CT . TACA . . CAAGG . - . A . AC . A . T . AATCA . . .T. . .AA P. ine-ENSl P. inf -JAL1 P. gra-GUE3 409 AATTTTA-ATAGTAGTTTTTTAAGAACTAACACTAATAGATGACCTTTTATACCTATTA- -CCGGCTAAAAGTTTAGTAAAAATACAATCTACTGATATTAT .T.A.A.C AC. . .A T.TTTTT. .T. . A . ATGTT A.C. . . A . AA . ACT . ATT . TTC . ACGCCCC . . GTTAC . . . A . GTT . AGCGACTA . CCAAATTG . T . . .T.AA.G TATTTTTACT . .A. AT. AT. .AA. . TTTA . AAGATTTAAT . .C--A. . ATCC . GCTTAC . . .A.GCT.A.CG. . GA P. ine-ENSl P. inf-JALl P. gra-GUE3 511 TCCCACAGACTTAGAGGACCAAAAAATGGAA GGAAATTTAAAATGCGCCC-ATTAACTAGTATTCGGTTAACAAGTTATTTACCCATTATATATACAT CT.G.T.TCG.A.C. . - .AA.TCCG. .TA. . GCCCTA . T . ACC . . - CAATT . AAG . C . . . . CGAC . . . . CAA . TTTTTAC . T . . . -T.T C. .G.T.CT. . .CT. . . .AA.TTC. .ATT. . --CCAA.T TCATT . TAAG . C . G . CCTGT . . . CAACC . ATTAC . T . . . - . TTTG P. ine-ENSl P. inf-JALl P . gra-GUE3 613 TTAAAGAAAATGTAATTAGACCGTTTCACAATAAGTACAGGGCTT- -ATTTTTTCAACCAATTAAAAATATAAAAAATA-AAGGCTTATACTAATTTAAACA . . . . TT . T . - - TA. .A. . . . TCAAG . . . .A. A. . .AA.TAC A. . .T CTG.A.TG. . . CT . .TGA.C.TTG . . AAATAAA - TGTAAACAGAC - GTTTCACAA . .TA. .A. .A. TCAAG. . . .A. A. . .AA.TAC T. . . .CC- . . A . TA . . . CT . .TG. .T. . TT P. ine-ENSl P. inf-JALl P. gra-GUE3 715 CCATATCTAAAGAGTAACAATTATAGCATGTATTAAACAAGTACGATATGGCTTCCTTTCTAAAAAAGGTTAAATTGGAGATAGATTAGAGCCGATTTTGTA TT. . . AGCTCC . . . - A .G. . C . GTGTAGTTCGGTAATGGG - CCATCG - .C. . TTC . C . -G. . . ,T. . .AA. . .C. . . - . . .TG A . G . ACTGTTT . . GT . .GC.AT.GG- . .C.C.C. . . . TTC . C . .G. . .T GAC . .T 817 P . ine- ENS1 CTTTACCCCCAGGAAGTACTGTAAAATGCGGCTCTAATCTATCCACATATACTAATGCAAGCCTTTATAAACCATTTCAGCGAATATAACTTATTATTAGAT P. inf-JALl P . gra-GUE3 919 P. ine-ENSl TATACACCTTATTGTATGTTTAATAAAACTTAAATGTATATATATCTAGAGATATCCATTAGCCTGAATTTTTAGCCCCTCCCCCCCCTTTCTCATCCTCTA P. inf-JALl P. gra-GUE3 1021 P . ine- ENS 1 TCTGAGTTTATTTCTTAAATTAATACAATTTTATATTTGGCTGGCGTGGTTAGTTTATATTTAATTAAATATTTTGCCATTAAACATGATGCCTGATAAAAG P. inf-JALl P. gra-GUE3 1123 P. ine-ENSl GTTTGTTTTGATAGGACAAGTAATGCAGATAAAACTGCTCATGTTACAAGTAACGGGATCACACCAATTCCTAGAATATCAAAAATCTTATGCTC P. inf-JALl P. gra -GUE3 Figure 1 Alignment of control region sequences of California spiny lobster (Panulirus interruptus [ P. ine ]), blue spiny lobster (P. inflatus [ P.inf]), and green spiny lobster (P. gracilis [ P.gra ]) used to find specific primers from one specimen of each lobster species. Underlined regions indicate species-specific primers. Label after the species name indicates specimen’s voucher code. Numbers 1, 103, 205,..., 1123, indicate positions in the sequence of the first nucleotide in the row. ilis (Table 2). Thus the composite haplotypes are able to delineate among these species. All electrophoretic patterns of PCR-RFLP products of adult specimens were congruent with those obtained from sequences analysis. Finally, the sequence of P. penicillatus had restriction sites that would produce different patterns from the other species (Fig. 3). Multiplex PCR A high annealing temperature (59°C) was adequate for successful multiplex PCR reactions containing the DNA of the three species. There were no additional fragments to those expected that could prevent identification of lobster species in all tested adult specimens. The size of 208 Fishery Bulletin 106(2) Table 2 Mitochondrial DNA (mtDNA) 12S rRNA-control region (563 base pairs [bp]) fragment sizes resulting from the BsmAl and Hinll- recognizing sites during nucleotide sequence analyses of three Panulirus species: California spiny lobster ( Panulirus interrup- tus), blue spiny lobster (P. inflatus), and green spiny lobster (P. gracilis). Hap = haplotype nomenclature (see text for explanation); % = percentage of haplotypes in a sample of 215 sequences. Composite haplotypes were obtained by combining the haplotype of each enzyme. Species Enzyme Fragment size (bp) Hap % Composite haplotype P. interruptus BsmAl 162,401 A 100.0 AA Hindi 440,123 A 100.0 P. inflatus BsmAl 162, 401 A 100.0 AB Hindi 563 B 100.0 P. gracilis BsmAl 115,47,401 B 100.0 BA/BB Hinfl 440,123 A 98.7 563 B 1.3 the amplified product delineated all three species in a 1% agarose gel. The fragment size ofP. interruptus (with the use of LanCR-R primer) was -1000 bp, the fragment size of P. inflatus (with the use of PinRCl-b) was -800 bp, and the fragment size of P. gracilis (with PgraRCl- b) was -700 bp (Fig. 4). Even though two P. interruptus specimens produced unspecific amplifications, the frag- ments stained weakly compared to the 1000-bp fragment and did not interfere with identification. Because the specimens analyzed by multiplex PCR were collected at 99 p Hap 4 Hap 7 Hap 5 Hap 8 ■— Hap 6 , r; 99 100 0.02 Genetic distance L different sites, these data show that there is no apparent intraspecific variation, which indicates that the multi- plex primer set provides a method that can be used to identify the three lobster species. Larval identification Species identification based on PCR-RFLP did not sup- port the identification based on morphological crite- ria. The fragment patterns produced by digestion with BsmAl and Hinfl provided a means for identi- fying P. inflatus and P. gracilis. Two specimens in each of groups 1 and 2 were identified as P. gracilis. The 42 remaining larvae belonging to groups 1, 2, 3, and 4 were identified as P. inflatus; none of the specimens were identi- fied as P. interruptus. Multiplex PCR analysis confirmed the results. - Hap 10 Hap 9 - Hap 1 1 r Hap 2 Hap 1 Hap 3 B Figure 2 Neighbor-joining phylogenetic tree based on the mitochondrial 16S rRNA gene haplotypes (Hap 1-11) from a total of 20 adult lob- sters of blue spiny lobster (Panulirus inflatus), green spiny lobster (P. gracilis), and California spiny lobster (P. interruptus). Haplo- type groups identified as A, B, and C represent blue, green, and California spiny lobster, respectively. Tree reconstruction was based on Kimura’s two-parameter distance (K2P) with 1000 replications. Numbers on the nodes are bootstrap values. The branch length is measured as the number of nucleotide substitutions. Discussion Several methods with molecular markers have been carried out for identification at the species level. Even though PCR-RFLP and multiplex- PCR are widely used as tools in distinguishing species of different taxonomic groups (Moore et al., 2003; Chow et al., 2006a), efforts have to be made to obtain the proper amplification primers when universal primers do not give consistent results in the studied species. According to our analysis, identification of spiny lobster species can be successfully done by combining different strategies. First, although insufficient for recognizing spiny lobsters larvae, morphological criteria should be used. Then, one or both of the genetic tech- niques can be applied to definitively support the morphological results. The simultaneous use of the PCR-RFLP and multiplex PCR NOTE Garcia-Rodiiguez et al.: Mitochondrial DNA markers to identify Panulirus spp. 209 210 Fishery Bulletin 106(2) Panulirus interruptus Panulirus inflatus Panulirus gracilis 12 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 Figure 4 Polymerase chain reaction (PCR) products with the multiplex-PCR method for each of the three lobster species: California spiny lob- ster ( Panulirus interruptus), blue spiny lobster ( P . inflatus), and green spiny lobster {P. gracilis). Lanes 8, 15, and 21 are 100-bp ladder molecular markers. techniques is recommended because there are instances for which unsuccessful amplifications are obtained by one of the methods. This is a consequence of mutations in the annealing sites, rather than a failure of PCR techniques (Ray et al., 2002). The 20 sequences that we evaluated for differences in restriction patterns revealed species-specific restric- tion sites for BsmAl and Hinfl. Amplified products of previously identified adult lobsters were concordant with the restriction patterns found in these sequences. The PCR-RFLP technique is successful for separating Panulirus larvae because 16S rRNA is relatively easy to amplify at intraspecific conservative sites and inter- specific variable sites. The PCR-RFLP has also been applied to the cy- tochrome oxidase I (COI) gene to identify 10 spiny lobster species of the genus Panulirus found in the northwestern Pacific Ocean (Chow et al., 2006a). In general, Chow et al. (2006a) found more than two hap- lotypes per species; however, no composite haplotype was shared by these species. Although Chow et al. (2006a) suggested increasing efforts in searching for intraspecific variation on larger samples, the probabil- ity of misidentification is very low because substantial divergence has been observed among species of the genus Panulirus (Ptacek et al., 2001). Our results sup- port the finding of high divergence among the three lobster species in Mexico on the basis of nucleotide intraspecific variation (Fig. 2). Multiplex-PCR fragments are clearly capable of dis- criminating among the three lobster species. This meth- od is fast, simple, and relatively inexpensive because species identification can be performed by using just PCR amplifications with no digestion and with small reaction volumes. The multiplex-PCR method used in this study is successful because it is based on spe- cies-specific primers and a sufficiently high anneal- ing temperature (Tm) of 59°C is used that avoids the amplification of unspecific PCR products. Also, a high Tm reduces the possibility of amplifying homologous sequences from other species. Identification of lobster larvae is consistent with the restriction patterns found in adult lobsters and identifications carried out by PCR-RFLP can be cor- rectly confirmed by using multiplex PCR. Molecular analyses showed that the previous larval classifica- tion (based on morphological characters) was probably incorrect because very small or injured specimens were used. Alternatively, misidentification in lob- ster larvae could also be a consequence of morpho- logical criteria that are not diagnostic characters for discriminating between lobster larvae, as it has been reported in other Panulirus species, such as the Japanese spiny lobster (P. japonicus) and species from the Atlantic Ocean (i.e. , the Caribbean spiny lobster, P. argus) (Silberman and Walsh, 1992; Chow et al., 2006a, 2006b). Morphological criteria currently used for the identification of phyllosoma larvae of P. inter- ruptus, P. inflatus, and P. gracilis are insufficient as well. Further descriptions of larvae specimens should be conducted to search for consistent morphological differences between species. Other possible applications of PCR-RFLPs and mul- tiplex PCR are used in fishery forensics, when lobster NOTE Garcfa-Rodriguez et al.: Mitochondrial DNA markers to identify Panulirus spp. 211 products must be identified. This is especially the case when lobster products of unknown origin have to be analyzed if there is suspicion that the lobster was ob- tained by illegal fishing, as in the case of important protected species, such as sea turtles (Moore et al., 2003). Conclusions We used the nucleotide variation of two mtDNA frag- ments from adult spiny lobster samples to find molecu- lar marker applications for species discrimination. The RFLP and multiplex-PCR protocols developed in this study allow for correct discrimination of three commer- cial lobster species inhabiting the Pacific coast of Mexico. Application of both types of molecular markers in the identification of lobster larvae showed concordance and the potential to discriminate between phyllosoma species during early stages. Molecular identification of larvae was inaccurate with our previous assignment based on morphological criteria. Thus, the use of morphological characteristics in phyllosoma larvae can be misleading for identification to the species level because anatomical parts can easily be damaged during collection. This new information on genetic identification of species at the larval stage is of wide interest because studies focused on taxonomic and ecological revisions require accurate species identification. Acknowledgments We thank E. Espino and M. Iacchei for providing adult lobster samples from Jalisco, Mexico and California, USA, respectively. Research was supported by proj- ect SEMARNAT-2004-C01-153 to G. Ponce-Dfaz. Three anonymous reviewers helped to improve the manuscript. Literature cited Baez, R. 1983. Phyllosoma and puerulus larvae of the green spiny lobster Panulirus gracilis Streets 1871 from the Costa Rica Expedition, 1973 (Crustacea: Decapoda: Palinuridae). Rev. Biol. Mar. Valparaiso 19:79-111. [In Spanish.] Blair, D., M. Waycott, L. Byrne, G. Dunshea, C. Smith-Keune, and K. M. Neil. 2006. Molecular discrimination of Perna (Mollusca: Bivalvia) species using the polymerase chain reaction and species-specific mitochondrial primers. Mar. Bio- technol. 8:380-385. Brown, W. M., M. George, and A. C. Wilson. 1979. Rapid evolution of animal mitochondrial DNA. Proc. Natl. Acad. Sci. 76:1967-1971. Chan, R. W. K., P. I. Dixon, J. G. Peppered, and D. D. Reid. 2003. Application of DNA-based techniques for the identi- fication of whale sharks ( Carcharhinus spp.) in protective beach meshing and by recreational fisheries off the coast of New South Wales. Fish. Bull. 101:910-914. Chow, S., N. Suzuki, H. Imai, and T. Yoshimura. 2006a. Molecular species identification of spiny lobster phyllosoma larvae of the genus Panulirus from the North- western Pacific. Mar. Biotechnol. 8:260-267. Chow, S., H. Yamada, and N. Suzuki. 2006b. Identification of mid- to final stage phyllosoma larvae of the genus Panulirus White, 1847 collected in the Ryukyu Archipelago. Crustaceana 79:745-764. Garcia-Rodriguez, F. J., and R. Perez-Enriquez. 2006. Genetic differentiation of the California spiny lob- ster Panulirus interruptus (Randall, 1840) along the west coast of the Baja California Peninsula, Mexico. Mar. Biol. 148:621-629. Gonzalez-Armas, R., O. Sosa, R. Funes, and V. A. Levy. 1999. Confirmation of the spawning area of the striped marlin, Tretrapturus audax, in the so-called core area of the eastern tropical Pacific off Mexico. Fish. Ocean- ogr. 8:238-242. Gyllensten, U., D. Wharton, A. Josefsson, and A. C. Wilson. 1991. Paternal inheritance of mitochondrial DNA in mice. Nature 352:255-257. Hendrickx, M. E. 1995. Lobster (spiny lobster. Pacific lobsterette, fan lobster, and mud lobster, etc.). In FAO Guide for the identification of fisheries species. Central-Eastern Pacific. Vol. I: Plants and invertebrates (W. Fisher, F. Krupp, W. Schneider, C. Sommer, K. E. Carpenter, and V. H. Niem, eds.), p. 385-415. FAO, Rome. [In Spanish], Hill, R. S., L. D. Allen, and A. Bucklin. 2001. Multiplexed species-specific PCR protocol to dis- criminate four N. Atlantic Calanus species, with an mtCOI gene tree from ten Calanus species. Mar. Biol. 139:279-287. Hsieh, H. S., T. Chai, and D. F. Hwang. 2007. Using the PCR-RFLP method to identify the species of different processed products of billfish meats. Food Control 18:369-374. Johnson, M. W. 1971. The palinurid and scyllarid lobster larvae of the Tropical Eastern Pacific, and their distribution as related to the prevailing hydrography. Bull. Scripps. Inst. Ocean. 19:639-647. Johnson, M. W., and M. Knight. 1966. The phyllosoma larvae of the spiny lobster Panu- lirus inflatus (Bouvier). Crustaceana 10:31-47. Konishi, K., N. Suzuki, and S. Chow. 2006. A late stage phyllosoma larva of the spiny lobster Panulirus echinatus Smith, 1869 (Crustacea: Palinu- ridae) identified by DNA analysis. J. Plankton Res. 28:841-845. Kumar, S., K. Tamura, and M. Nei. 2004. MEGA 3: Integrated software for molecular evolu- tionary genetics analysis and sequence alignment. Brief. Bioinform. 5:150-163. Moore, M. K., J. A. Bemss, S. M. Rice, J. M. Quattro, and C. M. Woodley. 2003. Use of restriction fragment length polymor- phism to identify sea turtle eggs and cooked meats to species. Conserv. Genet. 4:95-103. Munoz-Garcia, I., A. Nunez-Pasten, R. Perez-Gonzalez, M. I. Borrego, and L. M. Valadez. 2004. Current status on the phyllosoma larva knowledge of Panulirus inflatus and P. gracilis from Pacific Mexican coasts and Gulf of California. In Contribution to the 212 Fishery Bulletin 106(2) present status of the Eastern Pacific crustaceans 3 (M. E. Hendrickx, ed.), p. 213-234. Instituto de Ciencias del Mar y Limnologia, Universidad Nacional Autonoma de Mexico, Mexico. [In Spanish.] Palumbi, S. R. 1996. PCR and molecular systematics. In Molecular systematics (D. M. Hillis, C. Moritz, and B. K. Mable, eds.), p. 205-247. Sinnauer Assoc., Sunderland, MA. Ptacek, M. B., S. K. Sarver, M. J. Childress, and W. F. Herrnkind. 2001. Molecular phylogeny of the spiny lobster genus Panulirus (Decapoda: Palinuridae). Mar. Fresh. Res. 52:1037-1047. Purcell, M., G. Mackey, E. LaHood, H. Huber, and L. Park. 2004. Molecular methods for the genetic identification of salmonid prey from Pacific harbor seal ( Phoca vitulina richardsi) scat. Fish. Bull. 102:213-220. Ray, D. K., D. W. Freshwater, and W. D. Webster. 2002. Multiplex PCR identification of eastern and allegh- eny woodrats in zones of contact. J. Wildl. Manag. 66:86-90. Rozen, S., and H. Skaletsky. 2000. Primer3 on the WWW for general users and for biologist programmers. In Bioinformatics methods and protocols: methods in molecular biology (S. Krawetz and S. Misener, eds.), p. 365-386. Humana Press, Totowa, NJ. Silberman, J. D., and P. J. Walsh. 1992. Species identification of spiny lobster phyllosome larvae via ribosomal DNA analysis. Mol. Mar. Biol. Biotech. 1:195-205. Weider, L. J., A. Hobaek, T. Crease, and H. Stibor. 1996. Molecular characterization of clonal population structure and biogeography of arctic apomictic Daphnia from Greenland and Iceland. Mol. Ecol. 5:107-118. 213 Advantages of using crest nets to sample presettlement larvae of reef fishes in the Caribbean Sea Cormac J. Nolan (contact author) Email address: cormac. nolan@ucd.ie Marine Biodiversity, Ecology and Evolution School of Biology and Environmental Science University College Dublin Belfield, Dublin 4, Ireland Bret S. Danilowicz Allen E. Paulson College of Science and Technology Georgia Southern University Statesboro, Georgia 30460-8044 Identifying the spatial and tempo- ral patterns of larval fish supply and settlement is a key step in under- standing the connectivity of meta-pop- ulations (Sale et ah, 2005). Because of the potentially dispersive nature of the pelagic larval phase of most reef fishes, tracking cohorts from hatching to settlement is extremely difficult (but see Jones et ah, 1999). However, for many studies it is sufficient to sample larvae immediately before set- tlement. Many coral reef fish species use mangrove and seagrass beds as nursery habitats (Nagelkerken et al., 2001; Mumby et al., 2004) and larvae of these species must pass over the reef crest in order to arrive at their preferred settlement habitats. The ability to sample this new cohort of larval fishes provides opportunities for researchers to explore the intri- cacies of the transition from larva to juvenile (Searcy and Sponaugle, 2001). Quantifying the potential set- tlers also provides valuable informa- tion about the spatial and temporal supply of presettlement larvae (Victor, 1986). Therefore a number of larval sampling methods were developed, one of which is the use of crest nets (Dufour and Galzin, 1993). Crest nets are rigid-frame ta- pering nets that are fixed to the substrate in shallow water imme- diately behind the crest of the reef (see Doherty and Mcllwain, 1996 for full description). The top of the crest net is above the surface of the water and currents and wave action force larvae into the mouth of the net. Because of the turbulence of the water coming over the reef crest and the fact that the whole water column is filtered, net avoidance by larval fishes is estimated to be minimal. Channel nets (Shenker et al., 1993) and light traps (Doherty, 1987), on the other hand, remain the domi- nant methods for sampling settle- ment-stage larval fishes on western Atlantic reefs. Surface channel nets are floating nets that are free to swivel with the prevailing current. Where crest nets are positioned in the shallow back reef, channel nets are positioned in deeper channels between mangroves, further away from the reef. Crest nets have been widely used in the Pacific Ocean to quantify the larval abundance of coral reef fishes immediately before settlement (Leis et al., 1998; Du- four et al., 2002; Leis et al., 2003; Mcllwain, 2003; Lecchini et al., 2004). Despite the apparent success of sampling reef fishes in the Pa- cific Ocean with crest nets, there are currently no reports of crest nets being employed for sampling reef fishes in the Caribbean Sea. The first objective of this study was to simultaneously deploy crest and channel nets to compare the abun- dance and species richness of larval fishes sampled. It was hypothesized that crest nets would capture more larvae by sampling the whole water column on the reef crest as opposed to channel nets that sample only surface waters. Larval reef fish possess impres- sive swimming capabilities (Leis and Carson-Ewart, 1997) and have the ability to detect reefs at a distance (Myrberg and Fuiman, 2002) and can therefore influence their own disper- sal. However, many other abiotic fac- tors can still influence their growth, survival, transport, and eventual ar- rival at a suitable settlement habitat. The abundance of larvae present is related to lunar period in some ar- eas (Robertson et al., 1988, Thorrold et al., 1994; Sponaugle and Cowen, 1996), but this abundance is not fully correlated with peaks in abun- dance in other areas (Kingsford and Finn, 1997). Larval growth rates and swimming ability vary with water temperature in some species (Green and Fisher, 2004) and winds can al- ter the strength and direction of sup- plying currents. The second objective of this study was to explore correla- tions between certain abiotic factors (lunar phase, water temperature, and prevailing wind) and the number of species and individuals collected by each net type. Materials and methods Study site Fieldwork was conducted at Turneffe Atoll, Belize (17°16'5"N, 87°48'57"W, Fig. 1A). Turneffe Atoll is part of the Meso-American Barrier Reef System (MBRS) that runs along southern Mexico through the waters of Belize, Guatemala, and Honduras. The MBRS is the world’s second largest coral reef system after the Great Bar- Manuscript submitted 1 June 2007. Manuscript accepted 10 January 2008. Fish. Bull. 106:213-221 (2008). The views and opinions expressed or implied in this article are those of the author and do not necessarily reflect the position of the National Marine Fisheries Service, NOAA. 214 Fishery Bulletin 106(2) Figure 1 Maps indicating the position of Turneffe Atoll, Belize (A) and the positions of crest and channel nets around Calabash Caye (B). The inset in map A indicates the position of Belize in Central America and the rectangle around Turneffe Atoll specifies the area enlarged in map B. rier Reef in Australia. Turneffe Atoll is a large offshore ring of islands bordered by coral reefs. It has a large central lagoon that contains many mangrove islands and channels. The atoll is located outside the coastal barrier reef, approximately 46 km west of mainland Belize (Fig. 1A). Larval collection and identification The definnition of “larva” will follow that of Leis (2006): the posthatching pelagic life history stage of demersal fishes (which is equivalent to the presettlement stage of Kingsford and Milicich, 1987). Larvae were sampled with crest nets and channel nets from 6 July to 26 August 2005, 24 January to 4 March 2006, and 17 May to 28 July 2006. One crest net was positioned in shallow water directly behind the reef crest in each of three sites approximately 1 km apart (Fig. IB). The crest nets had a mean width of 5.85 m, a mesh size of 2 mm, and were situated in 65-90 cm of water at each site. One surface channel net (Shenker et al., 1993) was placed in each of three separate mangrove channels leading to the central lagoon, each net with a square mouth (2 mx 1 m) with 1.6-mm mesh. It was not our intent to optimize the per- formance of either net. Therefore, although there were differences in net cross-sectional area, mesh sizes, and placement locations between crest and channel nets, these differences represent how each net has been typi- cally deployed. In preliminary sampling at Turneffe Atoll, near zero or zero catches occurred during daylight hours, which was consistent with the findings of Shenker et al. (1993). Therefore, collections were made only at night. Both types of nets were deployed nightly and the catch was retrieved and identified each morning. All indi- viduals of all species of larval reef fishes were counted each day. Where species could not be determined, the lowest taxonomic category that could be unambigu- ously determined was used. Larvae were examined live and identified (Humann and DeLoach, 2002; Richards, 2005). Over the course of the study a number of speci- mens of all species were preserved in 95% ethanol for later validation. Environmental variables Mechanical flowmeters (model 2030R6, General Ocean- ics, Inc., Miami, FL) were deployed with each net. These NOTE Nolan and Danllowicz: Use of crest nets for sampling presettlement larvae of reef fishes in the Caribbean Sea 215 meters are equipped with a high-resolution rotor for low-speed flow and had a minimum threshold of approxi- mately 6 cm/sec. The mean nightly measurement of flow was used to calculate the total volume of water filtered by each net. Underwater temperature loggers (Hobo Pendant Temperature Logger, Onset Computer Corp., Bourne, MA) provided a fine-scale record of the tem- perature of water being sampled (temperature data were not available for 2005). Wind speed and wind direction data were obtained from an automated weather station at Belize City International airport (17°53'N, 88°30'W). These wind reports provided a reasonable record of prevailing conditions at Turneffe Atoll because of the proximity and lack of geographic obstacles between the two points. The mean nightly wind direction was given a positive value for an on-shore wind and a negative value for an off-shore wind. Finally, a variable incorporating both the nocturnal illumination and tidal periodicity of the lunar cycle was calculated (see DAlessandro et al., 2007). The hours of nocturnal flood tides were calculated for each sampling night with tide prediction software (JTide, vers. 5.1, P. Lutus, freeware software available online) and this number was multiplied by the percent- age of the moon that was visible (full moon=100%). Statistical analyses Species-environment ordinations (CANOCO, vers. 4.5, Microcomputer Power, Ithaca, NY ) were used to estab- lish the relative importance of individual environmental factors (sampling season, wind, water temperature, and nocturnal flood tides) in explaining the overall vari- ance in larval abundance and species richness in the catch. The species and environmental data were found to be linear and were examined by redundancy analysis (RDA). An RDA plot shows the best fit of multivariate data in a two-dimensional ordination. The temporal supply of fish larvae was investigated by using correlation plots and circular statistics (Ray- leigh z; Zar, 1984). Cross-correlation plots were used to compare the timing of the capture of larvae in the two different environments, namely behind the reef crest where crest nets were used and the mangrove channels where channel nets were used. Once both net types were shown to collect larvae synchronously (see Results), the data for both nets were combined into a single time series. Auto-correlations were then plotted to examine the temporal periodicity of the catch. To achieve this, all three sampling periods were concat- enated into a continuous time series to ensure that more than 2.5 continuous lunar cycles were included (the minimum necessary for auto-correlation analysis for an examination of lunar periodicity). Each day was assigned a number corresponding to its point in the lunar cycle (lunar days 1-29, l=new moon). To ensure that the cycles were continuous, any overlapping lunar days between the sampling periods were deleted (from the middle period, spring 2006). The final time-series had 164 days, from which 14 overlapping days were deleted. Figure 2 Redundancy analysis plot of flow-corrected data. The angle between two variables repre- sents the correlation between them (0°=positive correlation, 90°=no correlation, 180°=nega- tive correlation), and the length of the arrow represents the magnitude, i.e., the longer the arrow, the greater the correlation coefficient. Environmental variables are shown as labeled arrows. Water = water temperature, Dark = hours of moonless, nocturnal flood tide, and Wind = average speed (km/h) of onshore wind. Categorical variables are shown as triangles. Species are shown as unlabelled grey arrows; individual species names have been omitted. Abundance and species richness are shown as black arrows labeled as Abund. and Species, respectively. Results A total of 53,579 larval reef fishes were caught that represented 33 families and 59 identified species (Table 1). On an average night, a crest net trapped 166.3 larvae (standard deviation [SD] = 407.4) and 8.5 species (SD = 5.8), whereas a channel net trapped 4.1 larvae (SD=12.2) and 0.9 species (SD=1.5). See Table 1 for list of families and species sampled by both net types. Ordinations There was a strong distinction between the species assemblages caught in the two net types (Fig. 2). Only data for 2006 sampling periods are presented in Figure 2, as no water temperatures were available for 2005 216 Fishery Bulletin 106(2) Table t Total number of fish larvae sampled with crest and channel nets at Turneffe Atoll, Belize, during the three sampling periods (summer 2005, spring 2006, and summer 2006). Barred lutjanids refers to Lutjanus apodus, L. analis, L. cyanopterus, L. griseus, and L.jocu. Striped Stegastes refers to Stegastes diencaeus, S. leucostictus, and S. variabilis. Family Genus Species Summer 2005 Crest Channel net net Spring 2006 Crest Channel net net Summer 2006 Crest Channel net net Total Acanthuridae Acanthurus bahianus 1 0 0 0 8 0 9 Acanthurus chirurgus 5 0 0 0 1 0 6 Acanthurus coeruleus 10 0 2 0 8 0 20 Achiridae Achirus sp. 0 0 0 0 3 0 3 Antennariidae Histrio histrio 5 1 0 0 2 0 8 All others 7 0 1 0 8 0 16 Apogonidae Apogon maculatus 302 26 53 6 464 10 861 Apogon quadrisquamatus 207 1 15 0 63 0 286 Astrapogon puncticulatus 114 5 261 0 212 1 593 Aulostomidae Aulostomus maculatus 0 0 2 0 1 0 3 Labrisomidae Starksia spp. 534 0 20 0 270 0 824 Bothidae Bothus spp. 43 18 8 4 28 5 106 Callionymidae Paradiplogrammus bairdi 123 0 25 0 924 1 1073 Carangidae All species 20 2 9 1 13 0 45 Chaetodontidae Chaetodon capistratus 25 1 0 0 7 0 33 Chaetodon ocellatus 5 0 0 0 0 0 5 Cynoglossidae Symphurus spp. 129 0 29 0 6 0 164 Diodontidae Chilomycterus antennatus 5 0 0 0 0 0 5 Elopomorpha All species 471 221 627 251 1759 173 3502 Gerreidae Eucinostomus spp. 13,450 296 1557 21 10,592 17 25,933 Gobiesocidae Areas rubiginosus 0 0 0 0 3 0 3 Gobiidae Bathygobius curacao 1 0 0 0 177 0 178 Ctenogobius saepepallens 0 0 97 1 13 0 111 Gnatholepis tliompsoni 2043 0 882 0 2623 12 5560 Priolepis spp. 23 0 5 0 226 0 254 Unknown spp. 3503 1 329 0 1108 0 4941 Labridae Halichoeres spp. 296 0 23 2 184 0 505 Thalassoma bifasciatum 21 0 3 0 31 0 55 Xyrichtys spp. 83 0 232 1 30 1 347 Lutjanidae Barred lutjanids All 150 12 8 1 18 2 191 Lutjanus synagris 0 1 4 0 1 0 6 Lutjanus mahogoni 4 0 0 0 0 0 4 Ocyurus chrysurus 2 0 6 0 0 0 8 Microdesmidae All species 39 1 52 7 63 4 166 Monacanthidae Cantherines sp. 1 0 4 0 1 0 6 Monacanthus ciliatus 184 2 0 0 23 4 213 Monacanthus tuckeri 113 8 22 1 105 9 258 Ogcocephalidae Ogcocephalus nasutus 3 0 0 0 6 0 9 Halieutichthys aculeatus 4 0 0 0 2 0 6 Ophidiidae All species 5 0 5 0 21 0 31 Ostraciidae Lactophrys spp. 77 1 32 0 3 0 113 Paralichthyidae Syacium spp. 0 0 11 0 4 8 23 Pomacanthidae Pomacanthus spp. 2 0 3 0 2 0 7 Pomacentridae Abudefduf saxatilis 13 0 6 0 3 2 24 Microspathadon chrysurus 0 0 0 0 1 0 1 continued NOTE Nolan and Danilowicz: Use of crest nets for sampling presettlement larvae of reef fishes in the Caribbean Sea 217 Table 1 (continued) Summer Spring Summer 2005 2006 2006 Family Genus Crest Channel Species net net Crest Channel net net Crest Channel net net Total Pomacentridae Stegastes adustus 3 0 0 0 1 0 4 (continued) Stegastes part it us 2 1 0 0 23 1 27 Striped Stegastes All 171 2 0 1 258 12 439 Scaridae Sparisoma spp. 329 3 359 3 838 0 1532 Scorpaenidae Scorpaena spp. 44 2 202 1 64 2 315 Serranidae Diplectrum spp. 45 5 0 0 5 0 55 Pseudogramma gregoryi 95 0 38 0 117 0 250 Rypticus sp. 2 0 0 0 5 0 7 Hypoplectrus spp. 0 0 0 0 7 0 7 Sphyraenidae Sphyraena barracuda 125 13 6 0 39 6 189 Syngnathidae Cosmocampus spp. 341 1 111 0 95 0 548 Tetraodontidae Sphoeroides spp. 0 0 7 3 16 0 26 Canthigaster spp. 435 0 24 7 26 0 492 (when 2005 data were analyzed separately, a very simi- lar plot was obtained). Most species were captured in greater abundance with crest nets and rarely, if ever, caught in the channel nets. For example, the families Acanthuridae, Ogcocephalidae, and Pomacanthidae were only caught in crest nets and there were no species or families that were exclusively caught in channel nets. The summer and spring sampling periods were extremely different (Fig. 2). However, when the three sampling periods were plotted separately, very similar ordinations with respect to environmental factors were obtained. The difference between summer and spring in the combined ordination of Figure 2 could be due to the lower numbers of larvae captured in spring 2006; however, there were notable absences of families in that sampling period, e.g., no Chaetodontidae or Ogcocephali- dae and only a single representative of Pomacentridae. Of the environmental variables (Fig. 2), the onshore wind was positively correlated with abundance and spe- cies richness of larval reef fishes sampled in crest nets. The combined factor (nocturnal illumination and tidal periodicity) was important but did not align strongly with the other explanatory or species variables. Higher water temperatures at the net sites corresponded with fewer larvae caught because water temperature was negatively correlated with the presence of the vast ma- jority of species. Time series analyses Peaks and lows in the supply of fish larvae appeared on the same nights in reef crest nets and channel nets in the mangroves (Fig. 3). The cross-correlation plots between net types revealed that catches (both in terms of abundance and species richness) were significantly correlated at a lag of zero (data sets were aligned for correlation on the same day at a lag of zero, one data set leads the other by one day for correlation at a lag of +1, etc.). For abundance, the greatest correlation between net types was at a lag of zero days (Fig. 3A). A lesser correlation at a lag of plus three days indicates that some groups of larvae took three days to pass from the reef crest to the mangrove channels. The other sig- nificant correlations at lags of -4, -3, and -1 days are more difficult to explain. There seems to be no biologi- cal reason that cohorts of reef fish larvae should arrive in the mangrove channels up to four days before they arrive at the reef crest. This finding may be a result of pooling abundances of all species and could possibly be resolved with further analysis by splitting abundances into families or species (where possible). Species rich- ness was also correlated at a lag of zero days; however, the other significant correlation, at a lag of -4 days, was greater than that at day zero (Fig. 3B). As with abundance, there seems to be no biological explanation for this correlation and more detailed analysis may prove advantageous. The auto-correlation plot for abundance (Fig. 4A) il- lustrates that there was no periodicity in the flow-cor- rected data and that the catch on any one night was not correlated with that on the preceding or following nights. However, the plot for species richness (Fig. 4B) shows a lunar periodicity in the numbers of species caught. The significant negative correlation at a lag of 16 days (at just over half the lunar cycle) shows that greater numbers of species caught in new-moon periods 218 Fishery Bulletin 106(2) A Abundance o o n i i i — i — i — r -7 -6 -5 -4 -3 -2 -1 0 1 Lag (days) LL o o B Species richness 0.4— 0 2- -0.2- -7 -6 -5 -4 -3 -2 -1 0 1 2 3 4 5 6 7 Lag (days) Figure 3 Cross-correlation plots of the average nightly abundance (A) and species richness (B) for larvae sampled in crest and channel nets for flow-correccted data. These plots identify any significant delay between catch in the crest versus chan- nel nets. Lag refers to the number of days by which one of the data sets is offset from the other when the correlation is calculated; data sets are aligned for correlation on the same day at a lag of zero, and the crest net data leads the channel net data by one day for correlation at a lag of +1, etc. The cross-correlation function (CCF, correlation coefficient between the two data sets at each lag) is on the ordinate. Values of the CCF above and below zero represent positive and nega- tive correlations between net types; the horizontal lines above and below the abscissa indicate the upper and lower 95% confidence limits, respectively. corresponded to fewer numbers of species caught in full-moon periods. Discussion Crest nets caught greater numbers of individuals and species per deployment than channel nets and would therefore be an advantageous sampling tool to use in studies attempting to maximize the chance of catch- ing greater numbers of a certain species. However, the difference between net types was not solely due to the design of the net. The two net types were deployed at two different habitats. All larvae passing over the top of a small width of the reef crest were sampled as the reef slope forced them into a constrained water column. In contrast, in the mangrove channels, only the top meter of the water column was sampled and larvae were free to pass underneath the floating channel net. A compari- son of the suites of larvae caught in each habitat would provide information about their settlement preferences. Such a comparison could not be made in the present study because the difference in the amount of the water column sampled was not controlled. However, Shenker et al. (1993) reported poor catches in subsurface deployed channel nets, and this finding indicates that most larvae that are still in the water column as they pass through the mangrove channels behind the reef crest remain near the surface of the water. Lunar periodicity of arriving settlers has been well documented in some reef fish species; greatest recruit- ment usually occurs at the darkest phase of the moon (Victor, 1986; Thorrold et al., 1994; D’Alessandro et al., 2007). Rayleigh 2 tests on non-flow-corrected data showed that significantly more larvae were caught at the new moon in the present study. When the catch was standardized by volume of water filtered however, all lunar periods had similar numbers of individuals per unit of water volume and no periodicity existed. This finding indicates that water flow was greater during the dark moon periods (new and last quarter) than during bright moon periods (first quarter and full), and the greater water flow removed the correlation between the quantity of larvae caught and the lunar period. It appears there was approximately the same number of larval fish per unit of water volume throughout the lunar cycle; the increased flow around the new moon simply carried more of them into the nets. Alterna- NOTE Nolan and Danilowicz: Use of crest nets for sampling presettlement larvae of reef fishes in the Caribbean Sea 219 A Abundance (no. of individuals per unit of water volume) o < rinn 1 Ini Ln I In 01 Lag (days) Figure 4 Auto-correlation plots of the average nightly abundance (A) and species richness (B) of larvae caught in both crest and channel nets for flow-corrected data. These plots identify any significant periodicity in the combined catch of crest versus channel nets. Auto-correlation plots are similar to cross-correlation plots but, unlike cross-correlation (which provides a comparison of two data sets), auto-correlation allows a comparison of one data set to itself. Lag refers to the number of days by which one copy of the data set is offset from the other when the correlation is calculated. Data sets are aligned for correlation on the same day at a lag of zero; one data set leads the other by one day for correlation at a lag of +1, etc. The auto-correlation function (ACF, correlation coefficient between the data sets at each lag) is on the ordinate. Values of the ACF above and below zero represent positive and negative correlations, respectively; the horizontal lines above and below the abscissa indicate the upper and lower 95% confidence limits, respectively. tively, the larvae used this increased flow to facilitate their movement to the reef and the darker conditions to improve predator avoidance. Given that larval fish near the time of settlement possess impressive swimming and sensory abilities, the effect of flow could simply be viewed as an interesting variable that masks true larval abundance in the water column. As reported previously (Shenker et al., 1993; Thor- rold et al., 1994; Kingsford and Finn, 1997), rather than deploying a net continuously, deploying a net around the new moon with an onshore wind would optimize collection efforts. The measurements of wind speed and direction at the international airport on mainland Belize were positively correlated with abun- dance and species richness of fish larvae at Turneffe Atoll. Because water temperature was found to be negatively correlated with the capture of almost all species, it is possible that the emptying of warm water from the lagoon negatively affects the arrival of larvae. All of these factors (lunar period, water temperature, and prevailing wind) may be further considered when trying to optimize the collection of fish larvae in sta- tionary nets. In assessing the effort required to install, maintain, and deploy the codend of each type of net, we found that channel nets were far easier to work with. Because of the position of crest nets, they are subject to high wave energy and strong currents. Therefore more effort is required to anchor the frame to the substrate and more time is needed to repair the unavoidable wear and tear. Channel nets, on the other hand, are quick to retrieve in the case of a storm and require very little ongoing maintenance. Researchers need to be aware of the additional effort required to set and maintain crest nets in comparison to other types of nets. The importance of flow has also been highlighted, and great care should be taken to evaluate this variable when making comparisons of larval catch among times and locations. Environmen- tal factors which alter this rate of flow seem to have the greatest influence on the catch of both stationary net types. Given the results of this study, there are no obvious obstacles to the use of crest nets in other parts of the Caribbean Sea where appropriate sites ex- ist, i.e., shallow reef crest with mainly unidirectional water flow. Given the greater water flow through the 220 Fishery Bulletin 106(2) environment in which they are deployed, they are likely to collect more larvae and hence better meet the needs of researchers working on settlement-stage reef fishes. Acknowledgments This work results from research funded partially by the Connectivity Working Group of the Coral Reef Targeted Research (CRTR) Program, a Global Environment Facil- ity-World Bank-University of Queensland international program. C. Nolan was supported by the Irish Research Council for Science, Engineering, and Technology. We thank S. Planes and J. Grignon (University of Perpi- gnon), S. Thorrold, H. Walsh, and L. Houghton (Woods Hole Oceanographic Institute), and the Institute of Marine Studies, University of Belize. Invaluable field assistance field was provided by P. H. Harbin, J. D. Hogan, R. Fisher, and the entire P. F. Sale laboratory. B. Victor provided essential identifications of larval fish, for both material specimens and for specimens viewed online at the Coral Reef Fish website. Literature cited D Alessandro, E., S. Sponaugle, and L. Thomas. 2007. Patterns and processes of larval fish supply to the coral reefs of the upper Florida Keys. Mar. Ecol. Prog. Ser. 331:85-100. Doherty, P. J. 1987. Light traps: Selective but useful devices for quan- tifying the distributions and abundances of larval fishes. Bull. Mar. Sci. 41(2):423-431. Doherty, P. J., and J. L. Mcllwain. 1996. Monitoring larval fluxes through the surf zones of Australian coral reefs. Mar. Freshw. Res. (47):383-390. Dufour, V., and R. Galzin. 1993. Colonization patterns of reef fish larvae to the lagoon at Moorea Island, French Polynesia. Mar. Ecol. Prog. Ser. 102(1-2):143-152. Dufour, V., G. Lecaillon, and P. Romans. 2002. Colonization of coral reefs by fish larvae. Oceanis 26(3):523-541. Green, B. S., and R. Fisher. 2004. Temperature influences swimming speed, growth and larval duration in coral reef fish larvae. J. Exp. Mar. Biol. Ecol. 299 ( 1 ) : 1 1 5 - 132 . Humann, P., and N. DeLoach. 2002. Reef fish identification: Florida Caribbean Baha- mas, 512 p. New World Publications, Jacksonville, FL. Jones, G. P., M. J. Milicich, M. J. Emslie, and C. Lunow. 1999. Self-recruitment in a coral reef fish population. Na- ture 402:802-804. Kingsford, M., and M. Finn. 1997. The influence of phase of the moon and physical processes on the input of presettlement fishes to coral reefs. J. Fish Biol. 51 (suppl. A):176-205. Kingsford, M. J., and M. J. Milicich. 1987. Presettlement phase of Purika scaber (Pisces: Monacanthidae): A temperate reef fish. Mar. Ecol. Prog. Ser. 36:65-79. Lecchini, D., V. Dufour, J. Carleton, S. Strand, and R. Galzin. 2004. Estimating the patch size of larval fishes during colonization on coral reefs. J. Fish Biol. 65(4): 1142-1146. Leis, J. M. 2006. Are larvae of demersal fishes plankton or nek- ton? Adv. Mar. Biol. 51:57-141. 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. Leis, J. M., T. Trnski, P. J. Doherty, and V. Dufour. 1998. Replenishment of fish populations in the enclosed lagoon of Taiaro Atoll: (Tuamotu Archipelago, French Polynesia) evidence from eggs and larvae. Coral Reefs 17 ( 1 ) : 1— 8 . Leis, J. M., T. Trnski, V. Dufour, M. Harmelin-Vivien, J. P. Renon, and R. Galzin. 2003. Local completion of the pelagic larval stage of coastal fishes in coral-reef lagoons of the Society and Tuamotu Islands. Coral Reefs 22(3):271-290. Mcllwain, J. L. 2003. Fine-scale temporal and spatial patterns of larval supply to a fringing reef in Western Australia. Mar. Ecol. Prog. Ser. 252:207-222. Mumby, P. J., A. J. Edwards, J. E. Arias-Gonzalez, P. G. Lin- deman, K. G. Blackwell, A. Gall, M. I. Gorczynska, A. R. Harborne, C. L. Pescod, H. Renken, C. C. C. Wabnitz, and G. Llewellyn. 2004. Mangroves enhance the biomass of coral reef fish communities in the Caribbean. Nature 427:533- 536. Myrberg, A. A., Jr., and L. A. Fuiman. 2002. The sensory world of coral reef fishes. In Coral reef fishes: dynamics and diversity in a complex eco- system (P. F. Sale, ed.), p. 123-148. Academic Press, San Diego, CA. Nagelkerken, I., S. Kleijnen, T. Klop, R. A. C. J. van den Brand, E. Cocheret de la Moriniere, and G. van der Velde. 2001. Dependence of Caribbean reef fishes on mangroves and seagrass beds as nursery habitats: a comparison of fish faunas between bays with and without mangroves/ seagrass beds. Mar. Ecol. Prog. Ser. 214:225-235. Richards, W. J. 2005. Early stages of Atlantic fishes: An identification guide for the western central north Atlantic, 2640 p. CRC Press, Taylor and Francis Group, Boca Raton, FL. Robertson, D. R., D. G. Green, and B. C. Victor. 1988. Temporal coupling of production and recruit- ment of larvae of a Caribbean reef fish. Ecology 69(2):370-381. Sale, P. F., R. K. Cowen, B. S. Danilowicz, G. P. Jones, J. P. Kritzer, K. C. Lindeman, S. Planes, N. V. Polunin, G. R. Russ, and Y. J. Sadovy. 2005. Critical science gaps impede use of no-take fishery reserves. Trends Ecol. Evol. 20(2):74-80. Searcy, S. P., and S. Sponaugle. 2001. Selective mortality during the larval-juve- nile transition in two coral reef fishes. Ecology 82(9):2452-2470. Shenker, J. M., E. D. Maddox, E. Wishinski, A. Pearl, S. R. Thor- rold, and N. Smith. 1993. Onshore transport of settlement-stage Nassau NOTE Nolan and Danilowicz: Use of crest nets for sampling presettlement larvae of reef fishes in the Caribbean Sea 221 grouper ( Epinephelus striatus) and other fishes in Exuma Sound, Bahamas. Mar. Ecol. Prog. Ser. 98(1— 2):31— 43. Sponaugle, S., and R. K. Cowen. 1996. Nearshore patterns of coral reef fish larval supply to Barbados, West Indies. Mar. Ecol. Prog. Ser. 133(1— 3):13— 28. Thorrold, S. R., J. M. Shenker, R. Mojica, Jr., E. D. Maddox, and E. Wishinski. 1994. Temporal patterns in the larval supply of sum- mer-recruitment reef fishes to Lee Stocking Is- land, Bahamas. Mar. Ecol. Prog. Ser. 112(1— 2):75— 86. Victor, B. C. 1986. Larval settlement and juvenile mortality in a recruitment limited coral reef fish population. Ecol. Monogr. 56(2):145-160. Zar, J. H. 1984. Biostatistical analysis, 2nd ed., 718 p. Prentice- Hall, Inc., Englewood Cliffs, NJ. 222 Fishery Bulletin 106(2) Fishery Bulletin Guidelines for authors Content of manuscripts Contributions published in Fishery Bulletin describe original research in marine fishery science, fishery engineering and economics, as well as the areas of marine environmental and ecological sciences (including modeling). Preference will be given to manuscripts that examine processes and underlying patterns. Descriptive reports, surveys, and observational papers may occa- sionally be published but should appeal to an audience outside the locale in which the study was conducted. Although all contributions are subject to peer review, responsibility for the contents of papers rests upon the authors and not on the editor or publisher. Submission of an article implies that the article is original and is not being considered for publication elsewhere. Articles may range from relatively short contributions (10-15 typed, double-spaced pages, tables and figures not included) to extensive contributions (20-30 typed pages). Notes are reports of 5 to 10 pages without an abstract and describe methods or results not supported by a large body of data. Manuscripts must be written in English; authors whose native language is not English are strongly advised to have their manuscripts checked by English-speaking colleagues before submission. Manuscript Preparation Title page should include authors’ full names and mail- ing 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 200 words (one-half typed page), state the main scope of the research, and emphasize the author’s conclusions and relevant findings. Do not review the methods of the study or list the contents of the paper. Because abstracts are circulated by abstract- ing 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 Conclusions), and Acknowl- edgments. Headings within each section must be short, reflect a logical sequence, and follow the rules of multiple subdivision (i.e., there can be no subdivision without at least two items). The entire text should be intelligible to interdisciplinary readers; therefore, all acronyms, abbreviations, and technical terms should be written out in full the first time they are mentioned. Include FAO common names for species in the list of keywords and in the introduction. Regional common names may be used throughout the rest of the text if they are dif- ferent from FAO common names which 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; for fish nomenclature follow the most current issue of the American Fisheries Society’s Common and Scientific Names of Fishes from the United States and Canada. Dates should be written as follows: 11 November 2000. Measurements should be expressed in metric units, e.g., 58 metric tons (t); if other units of measurement 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). Refrain from using the shorthand slash (/), an ambigu- ous symbol, in the general text Literature cited comprises published works and those accepted for publication in peer-reviewed literature (in press). Follow the name and year system for citation format in the “Literature cited” section (that is say, citations should be listed alphabetically by the authors’ last names, and then by year if there is more than one citation with the same authorship). If there is a sequence of citations in the text, list chronologically: (Smith, 1932; Green, 1947; Smith and Jones, 1985). Abbrevia- tions of serials should conform to abbreviations given in the Serial Sources for the BIOSIS Previews Database. Authors are responsible for the accuracy and complete- ness of all citations. Literature citation 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. Cite all software and special equipment or chemical 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). 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. • Do not place outline rules around graphs. • Do not use horizontal lines through graphs to indicate measurement units. • Use a comma in numbers of five digits or more (e.g. 13,000 but 3000). • Maps require a North arrow and degrees latitude- longitude (e.g., 170°E). 223 Tables are often overused in scientific papers; it is seldom necessary or even desirable to present all the data associated with a study. Tables should not be excessive in size and must be cited in numerical order in the text. Headings should be short but ample enough to allow the table to be intelligible on its own. All unusual symbols must be explained in the table legend. Other incidental comments may be footnoted with italic numeral footnote markers. Use asterisks to indicate probability in statisti- cal data. Do not type table legends on a separate page; place them above the table data. Do not submit tables in photo mode. Figures include line illustrations, photographs (or slides), and computer-generated graphs and must be cited in numerical order in the text. Graphics should aid in the comprehension of the text, but they should be limited to presenting patterns rather than raw data. Figures should not exceed one figure for every four pages of text. Figures must be labeled with author’s name and number of the figure. Avoid placing labels vertically (except of y axis). 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 and failure to correspond with editors in a timely manner will delay publication of a manuscript. Copyright law does not apply to Fishery Bulletin, 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 consid- ered correct form (e.g., Source: Fish. Bull 97:105). Submission The Scientific Editorial Office encourages authors to submit their manuscripts as a single PDF (preferred) or Word (zipped) document by e-mail to Fishery. Bulletin@noaa.gov. Please use the subject heading, “Fishery Bulletin manuscript submission”. Do not send encrypted files. For further details on electronic sub- mission, please contact the Scientific Editorial Office directly (see address below). Or you may send your manuscript on a compact disc in one of the above formats along with four printed copies (one original plus three copies [stapled]) to the Scientific Editor, at the address shown below. Dr. Adam Moles Scientific Editor, Fishery Bulletin NOAA/NMFS/AFSC 17109 Point Lena Loop Road Juneau, AK 99801-8626 Once the manuscript has been accepted for publication, you will be asked to submit a final electronic 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, tiff files, or EPS files. Send a copy of figures in the origi- nal software if conversion to any of these formats yields a degraded version. Questions? If you have questions regarding these guidelines, please contact the Managing Editor, Sharyn Matriotti, at Sharyn.Matriotti@noaa.gov Questions regarding manuscripts under review should be addressed to Adam Moles, Scientific Editor, at Adam. Moles@noaa.gov Fishery Bulletin Subscription form Superintendent of Documents Publications Order Form *5178 I I YES, please send me the following publications: Subscriptions to Fishery Bulletin for $36.00 per year ($50.40 foreign) The total cost of my order is $ . Prices include regular domestic postage and handling and are subject to change. (Company or Personal Name) (Please type or print) (Additional address/attention line) (Street address) (City, State, ZIP Code) (Daytime phone including area code) (Purchase Order No.) Charge your order. IT’S EASY! Please Choose Method of Payment: I | Check Payable to the Superintendent of Documents ] GPO Deposit Account -□ ] VISA or MasterCard Account (Credit card expiration date) To fax your orders (202) 512-2250 (Authorizing Signature) Mail To: Superintendent of Documents P.O. Box 371954, Pittsburgh, PA 15250-7954 Thank you for your order! Also available online at http://bookstore.gpo.gov/collections/fishery-bulletin 3 9088 01433 9972 02 U.S. Department of Commerce Volume 106 Number 3 July 2008 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 James W. Balsiger, Ph.D. Acting Assistant Administrator for Fisheries Scientific Editor Adam Moles, Ph.D. Associate Editor Elizabeth Siddon Ted Stevens Marine Research Institute Auke Bay Laboratories Alaska Fisheries Science Center 17109 Pt. Lena Loop Road Juneau, Alaska 99801 Managing Editor Sharyn Matriotti National Marine Fisheries Service Scientific Publications Office 7600 Sand Point Way NE, BIN C 15700 Seattle, Washington 98115-0070 The Fishery Bulletin (ISSN 0090-0656) is published quarterly by the Scientific Publications Office, National Marine Fisheries Service, NOAA, 7600 Sand Point Way NE, BIN C15700, Seattle, WA 98115-0070. Periodicals postage is paid at Seattle, WA. POSTMASTER: Send address changes for subscriptions to Fish- ery Bulletin, Superintendent of Docu- ments, Attn.: Chief, Mail List Branch, Mail Stop SSOM, Washington, 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: $36.00 domestic and $50.40 foreign. Cost per single issue: $21.00 domestic and $29.40 foreign. See back for order form. Editorial Committee Jeffrey M. Leis Thomas Shirley David Somerton Mark Terceiro Australian Museum, Sydney, Australia Texas A&M University 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 106 Number 3 July 2008 Fishery Bulletin Contents Articles 225-232 Shertzer, Kyle W., Michael H. Prager, and Erik H. Williams A probability-based approach to setting annual catch levels 233-244 Hoff, Gerald R. A nursery site of the Alaska skate ( Bathyrajo parmifera ) in the eastern Bering Sea 245-256 Beacham, Terry D., Nataly V. Varnavskaya, Khai D. Le, and Michael H. Wetklo Determination of population structure and stock composition of chum salmon ( Oncorhynchus keta) in Russia, determined with microsateilites The National Marine Fisheries Service (NMFS) does not approve, recommend, or endorse any proprie- tary product or proprietary material mentioned in this publication. No reference shall be made to NMFS, or to this publication furnished by NMFS, in any advertising or sales promotion which would indicate or imply that NMFS approves, recom- mends, or endorses any proprietary product or proprietary 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. 257-269 Shertzer, Kyle W., and Erik H. Williams Fish assemblages and indicator species: reef fishes off the southeastern United States 270-280 Nichols, Owen C., Robert D. Kenney, and Moira W. Brown Spatial and temporal distribution of North Atlantic right whales ( Eubalaena glocialis ) in Cape Cod Bay, and implications for management 281-292 Conners, M. Elizabeth, and Peter Munro Effects of commercial fishing on local abundance of Pacific cod ( Gadus macrocephalus) in the Bering Sea Fishery Bulletin 106(3) 293-304 Zimmermann, Mark, and Christopher Rooper Comparison of echogram measurements against data expectations and assumptions for distinguishing seafloor substrates 305-316 Maloney, Nancy E., and Michael F. Sigler Age-specific movement patterns of sablefish ( Anoplopoma fimbria) in Alaska Notes 317-320 Kilgour, Morgan J., and Thomas C. Shirley Distribution of red deepsea crab ( Chaceon quinquedens) by size and sex in the Gulf of Mexico 321-327 Erzini, Karim, Lufs Bentes, Rui Coelho, Pedro G. Lino, Pedro Monteiro, Joaquim Ribeiro, and Jorge M. S. Concalves Catches in ghost-fishing octopus and fish traps in the northeastern Atlantic Ocean (Algarve, Portugal) 328-333 Stark, James W. Age- and length-at-maturity of female arrowtooth flounder (Atheresthes stomias) in the Gulf of Alaska 334 Guidelines for authors 225 A probability-based approach to setting annual catch levels Email address for Kyle W. Shertzer: Kyle.Shertzer@noaa.gov Abstract — The requirement of set- ting annual catch limits to prevent overfishing has been added to the Magnuson-Stevens Fishery Conser- vation and Management Reauthoriza- tion Act of 2006 (MSRA). Because this requirement is new, a body of applied scientific practice for deriving annual catch limits and accompanying tar- gets does not yet exist. This article demonstrates an approach to setting levels of catch that is intended to keep the probability of future overfishing at a preset low level. The proposed framework is based on stochastic pro- jection with uncertainty in population dynamics. The framework extends common projection methodology by including uncertainty in the limit reference point and in management implementation, and by making explicit the risk of overfishing that managers consider acceptable. The approach is illustrated with applica- tion to gag ( Mycteroperca microlepis), a grouper that inhabits the waters off the southeastern United States. Although devised to satisfy new leg- islation of the MSRA, the framework has potential application to any fish- ery where the management goal is to limit the risk of overfishing by controlling catch. Manuscript submittted 28 September 2007. Manuscript accepted 15 February 2008. Fish. Bull. 106:225-232 (2008). The views and opinions expressed or implied in this article are those of the author and do not necessarily reflect the position of the National Marine Fisheries Service, NOAA. Kyle W. Shertzer (contact author) Michael H. Prager Erik H. Williams National Oceanic and Atmospheric Administration Southeast Fisheries Science Center Center for Coastal Fisheries and Habitat Research 101 Pivers Island Road Beaufort, North Carolina 28516 The Magnuson-Stevens Fishery Con- servation and Management Reautho- rization Act of 2006 (MSRA) requires that each Fishery Management Plan in the United States “establish a mechanism for specifying annual catch limits ... at a level such that overfishing does not occur in the fish- ery ...” (MSRA, 2006). This require- ment, which reflects an increased emphasis on conservation, is new in the sense that prevention of overfish- ing is mandated to be through annual catch limits (ACLs), rather than only through such less restrictive mea- sures as trip limits, size limits, or days allowed at sea. Because the stat- ute requires ACLs to be implemented by 2011 in all fisheries (by 2010 for fisheries where overfishing is occur- ring), discussion has begun on ways to compute them. Accompanying the discussion of ACLs is the discussion of corresponding annual catch targets (ACTs), levels of catch set as quotas in the fishery. In this study, we propose a method for setting annual catch levels that are treated as targets, but equally well could serve as limits. The meth- od is based on stochastic projection with uncertainty in population dy- namics. It extends usual projection methodology by including uncertainty in the limit reference point and in management implementation, and by making explicit the overfishing risk that managers consider acceptable. This probabilistic approach was de- vised specifically to satisfy the U.S. statute, but we expect it should be useful whenever the management ap- proach is to limit the risk of overfish- ing by controlling catch. From a technical point of view, the requirement to set ACLs is in- teresting in that overfishing is de- fined in terms of a fishery input (i.e., fishing-induced mortality rate), yet the control mechanism is defined in terms of a fishery output (i.e., catch). (Review of inputs and outputs in fishery management can be found in Morison [2004] and Walters and Martell [2004].) Values connecting inputs and outputs mathematically are stock abundance and age struc- ture, which change from year to year. Ideally, then, a method to set catch levels would take into account both uncertainty in the estimates of cur- rent stock abundance and structure and the expectation that abundance and structure will change with time. Current harvest-control rules for fisheries usually depend on a limit reference point, and uncertainty in estimating the limit reference point should also be considered. The limit reference point (typically the fishing rate at maximum sustainable yield (FMSy) or a Proxy for if* is general- ly considered to represent the level at which overfishing occurs (Mace, 2001). Given the uncertainties in popu- lation dynamics, stock assessment, and fishery management, it is argu- ably impossible to fish without some risk of overfishing. Rather than at- tempting to achieve zero probability of overfishing, our approach avoids 226 Fishery Bulletin 106(3) overfishing in a probabilistic sense by keep- ing the expectation of overfishing below a preset level (e.g., 0.1), presumably satisfying the new requirement of the MSRA. The ap- proach is intended for setting annual catch levels while accommodating uncertainties in future stock dynamics, assessment results, and in the implementation of management measures. Materials and methods Probability-based approach to setting catch levels (PASCL) The proposed method acts as a harvest-con- trol rule. It is a probability-based approach to setting catch levels (PASCL), incorporat- ing uncertainties in future stock dynamics, assessment results, and management imple- mentation. Given these uncertainties, PASCL sets annual target levels of catch consistent with the level of risk considered acceptable by managers. The method is based on the ratio- extended approach to setting target reference points (REPAST) of Prager et al. (2003), but is considerably revised 1) to establish reference points in catch, rather than in fishing mortality rate, and 2) to add a stock-projection component, which is needed to set catch for more than one year after a stock assessment. The new method is a general framework that can incor- porate details of almost any stock that is assessed. It is illustrated with gag ( Mycteroperca microlepis), a grouper found off the southeastern United States. Uncertainty in stock dynamics is represented by a stochastic projection model. The projection allows the setting of annual catch levels for more than one year and, if necessary, can account for a lag between the final year of assessment data and the first year of man- agement implementation. The projection model need not carry the assumption of equilibrium dynamics and can include any source of process or estimation uncertainty deemed appropriate, as with projections commonly used in fishery management. Sources often considered are recruitment dynamics and initial numbers of fish at age. Modeling nonequilibrium population dynamics, as here, is critical for developing harvest strategies (Hauser et al., 2006). Stock assessment results generally include estimates of uncertainty. A key stock assessment result used in PASCL is the estimate of Fhm, the limit reference point of fishing mortality rate (F) and its associated uncertainty, described by a probability density function (PDF), which can be either parametric or nonparamet- ric. If a PDF of Flim is unavailable, PASCL can use a point estimate, but ignoring that source of uncertainty can make overfishing more likely (Prager et al., 2003). Another basic assessment result, the estimate of stock abundance at age (with the corresponding estimate of uncertainty), is used to initialize stock replicates in stochastic projection with PASCL. Uncertainty in implementation stems from managers having only partial control of the catch (Rosenberg and Brault, 1993; Caddy and McGarvey, 1996; Prager et al., 2003). A target catch may not be met precisely if catch is monitored with delay, catch is managed indirectly through input controls, regulations are poorly enforced, or fishing behavior is unpredictable. In PASCL, as in REPAST (Prager et al., 2003), the level of risk acceptable to managers (P*) is quantified and explicit. In our study, risk is defined as the prob- ability of overfishing in any year t, i.e., as Pr(F(>Flim). A small value of P* corresponds to risk-averse manage- ment. Always, P*< 0.5 should hold, because at P*=0.5, overfishing is expected in half of all years. When P* is defined as a constant probability, the risk of overfishing in at least one of T years grows with the time horizon as 1-(1-P*)T (Fig. 1). In a simple formulation, the limit fishing mortal- ity rate Flim could be represented by a point estimate. Then, the probability (here equated to risk) of overfish- ing in year t would be a function of Fjim and the PDF of Ft ( Fijm ) = [ Fi(F)dF = l-nFhm), (1) where Flim) = the cumulative distribution of Ft evalu- ated at Flim. A catch level can then be set to position the distri- bution of Ft so that the desired risk is achieved; i.e., Shertzer et al.: Probability-based approach to setting annual catch levels 227 Figure 2 Flowchart of method to compute annual catch level. P* is the acceptable risk of overfishing, Flim is the limit reference point in fishing mortality rate, N is the number of replicate projections, ACT is the target annual catch level, CV is the coefficient of variation of management implementation, Ft is the fishing mortality rate in year t of a single projection replicate, and P=Pr{Ft>Fhm) is the probability of overfishing in year t associated with a trial ACT. Pr (Ft>Fhm)=P*. That catch level becomes the annual catch level in the sense of the MSRA. The full formulation used here is slightly more com- plex (and realistic) in that Flim is described by its PDF, 4>Fi . Then, the probability of overfishing is computed as Pr(Ft > Fhm) = J J Ft(0)de ( F)dF , (2) The goal of PASCL is to set annual catch levels such that Pr {Ft>Flim)=P* in each year of a multiyear se- quence. Extensions from the formulations described by Equations 1 and 2 are twofold: D use of output controls (catches) for management, and 2) a management time frame of more than one year. In what follows, we as- sume that PASCL is used to compute annual catch targets (ACTs). The approach is implemented through a projection model (Fig. 2) with the following steps: where 9 = a dummy variable. Equation 2 is the weighted sum of probabilities com- puted by Equation 1 for all possible values of Flim. Again, the distribution of Ft can be positioned so that Pr(Ft>Flim)=P*. An assumption of Equation 2 is that Fjim and Ft are independent. If correlation is observed or suspected, the probability of overfishing could be computed from the bivariate PDF P „ , rUimFf’ Pr( Ft > Fjjju ) = J J F( dOdF. ( 3 ) o F Although Equation 3 is more general, estimation of Fh F from data may seldom be possible. Fortunately, in many applications, Equation 2 will be a suitable approximation (see Discussion section). 1 Initialize N replicates of the stock, each different in abundance and age structure, to reflect uncertainty in the estimated current state of the stock. 2 Given implementation uncertainty in controlling catch, each ACT will be the central tendency of a probability distribution F| , compute P=Pr(Ft>F]im) from Equation 2. 5 Using a numerical optimization method, adjust p until P=P*. The adjusted p is that year’s ACT. 228 Fishery Bulletin 106(3) 6 Project each replicate one year forward by applying recruitment and natural mortality and taking catch Cn. 7 Repeat steps 2-6 for T years. In general, duration T of the projection will extend until ACTs based on the next assessment can be implemented. The preceding procedure gives an ACT for each year in the period, and the annual probability of overfishing is kept at P*. Setting catch levels of gag To illustrate the method, we applied PASCL to the gag stock off the southeastern United States. The stock was most recently assessed in 2006 from data through 2004 and a statisti- cal catch-age model (Quinn and Deriso, 1999) including the Beverton-Holt spawner-recruit model (Beverton and Holt, 1957). The stock was estimated to be experiencing the effects of over- fishing with a biomass at nearly 90% of that at maximum sustainable yield (SEDAR, 2006). To implement PASCL, we devised a stochastic projection model with structure identical to the age-based assessment model (SEDAR, 2006), in which landings and discards were computed from the Baranov (1918) catch equation. The parameter values chosen were those used or estimated in the assessment. The projection included two sources of uncertainty in stock dynamics. One was stochasticity in recruitment, assumed to be lognormal about the estimated Bever- ton-Holt spawner-recruit model, with parameter values from the assessment. The other was uncertainty in the estimated final numbers at age (Na 2005), which become Figure 3 Probability density (ph ) of the limit reference point (Ejim), defined here as the fishing mortality rate at maximum sus- tainable yield. a, 2005 ). the initial numbers at age in our example (N. In some applications, the variance of Na 2005 would be estimated during the assessment, but SEDAR (2006) provided only point estimates. To include uncertainty, we assumed that multiplicative error in the initial num- bers at age followed a lognormal distribution with mean (in log space) of zero and a standard deviation equal to that of recruitment (<7^): ^ a, 2005 ~ 2005eXP*V) ’ (3) where v~N(ji=0, o=oR). This approach accounts for uncertainty in initial condi- tions, while maintaining strong year classes estimated in the terminal year of the assessment. The first year of the projection was 2005, and new regulations on catch levels were implemented in 2008. For the projection during the premanagement years (2005-07), we applied a fixed level of landings, set to the geometric mean of landings from 2002 through 2004. The duration of the projection was 10 years: three premanagement years followed by seven years of man- aged catch levels (landings plus discard mortalities). Presumably, this duration is generous, spanning a pe- riod until the next assessment. The stochastic projection model was used to generate V=10,000 replicate stocks differing in abundance and age structure. This variation, along with imprecise management implementation, led to V=10,000 values of fishing mortality rate in each year, which were used to characterize the fishery’s annual probability density of Ft . These densities (F) were quantified nonparametri- cally through kernel density estimation with Gaussian kernel and bandwidth equal to the kernel’s standard deviation (Venables and Ripley, 2002). The limit reference point in F was set equal to FMSY (Mace, 2001). For this example, the probability den- sity of Fmsy (F) ) was estimated after the assess- ment through Bayesian analysis of the Beverton-Holt spawner-recruit model, accounting for uncertainty in model parameters. A prior distribution was specified for steepness (h), the parameter controlling how quickly recruitment approaches its unfished level as spawning biomass increases. This prior distribution was based on meta-analysis of steepness values (Myers et al., 1999) from species similar to gag. Species included were those considered to be periodic spawners, as defined by Rose et al. (2001), and limited to marine or anadromous de- mersal fishes, excluding rockfish ( Sebastes spp.) because of their uncharacteristically low steepness values. The estimated prior distribution was lognormal (SEDAR, 2004): h = exp(x) : x ~ N(/u- -0.33, a = 0.28). (4) The resulting posterior distribution of FUSY described F for use in PASCL (Fig. 3). In this example, (f>Fhm Shertzer et at.: Probability-based approach to setting annual catch levels 229 D 2006 2010 2014 B E Figure 4 Example projection with managed risk of overfishing set at P* = 0.1 and uncertainty in management implementation described by its coefficient of variation CV=0.2. (A) Probability of overfishing, (B) annual fishing mortality rate, (C) spawning biomass (100 metric tons It]), (D) landings (100 t), (E) dead discards (100 t), < F ) annual catch level (100 t). In B-F, results are presented as medians (thick lines, circles), along with 10th and 90th percentiles (thin lines), from ,/V=10,000 projection replicates. approaches zero quickly on the right because FMSY is connected tightly with steepness, a bounded parameter with positive probability at its upper bound. To allow uncertainty in management implementation, the annual catch level was assumed to follow a normal distribution with the mean equal to the target annual catch and coefficient of variation (CV) equal to a preset value. In some applications, the CV of implementa- tion might be estimated from data on performance of the fishery; in this application we considered values of 0.1, 0.2, and 0.3, with a focus on the assumption of CV=0.2. The final requirement of PASCL is to specify the al- lowable risk of overfishing. This analysis considered six different levels: P*e[0.05, 0.10, 0.15, 0.20, 0.25, 0.30], Results During the premanagement period (2005-07), overfish- ing was projected to occur in at least half of the projec- tions, and thereafter, at the acceptable P* (Fig. 4A). Because the probability of overfishing in the premanage- ment period was higher than P*, the fishing mortality rate was reduced when annual catch targets took effect (Fig. 4B). This allowed spawning biomass to increase (Fig. 4C). With nearly constant F, the increase in bio- mass provided for an increase in catch, composed mostly of live landings but also dead discards (Fig. 4, D-F). Catch within a year varied by replicate, reflecting uncer- tainty in management implementation, but by design was centered on the annual catch level (Fig. 4F). 230 Fishery Bulletin 106(3) Greater precision in management implementation reduced the variance of Ft for a given catch, which in turn allowed higher fishing mortality rates without an increased probability of overfishing (Fig. 5A). The higher rates then translated into larger annual catch levels (Fig. 5B). In general, higher P* was associated with larger catch (Fig. 6A). Biomass increased over time for all P* examined, but more quickly when risk of overfishing was smaller (Fig. 6B). Consequently, catch increased more quickly for smaller risk, and thus the overall range of catch shrank over time across levels of P*. Discussion work extends common methods by explicitly considering uncertainty in the limit reference point, uncertainty in management implementation, and the level of risk acceptable to managers. A limit reference point used with PASCL can be a single value, such as FMSY or a proxy for jFmsy, but it need not be a single value. For example, the Flim used to manage U.S. west coast groundfish is a func- tion of standing biomass (Punt, 2003). Uncertainty in the limit, whether a single value or function, could be modeled with any appropriate distribution. Similarly, uncertainty in management implementation can be incorporated with flexibility. Choice of harvest-control rules The proposed probabilistic approach to setting annual catch levels, PASCL, is quite flexible. It incorporates many of the projection methods common in stock assessment, which can be based on size-structured, age-structured, or unstructured population models. It can incorporate any sources of uncertainty consid- ered important; for example, environmental influences, demographic stochasticity, and multispecies effects. Our A B 4.4 4.2 - 4.0 - o 3.8 - 03 O 3.6 - 3.4 - 3.2 - 3.0 - 2008 2009 2010 2011 2012 2013 2014 Figure 5 Median annual (A) fishing mortality rate and (B) catch level (100 t) from projections with managed risk of overfishing set at P*=0.1 and uncertainty in management implementa- tion, defined by the coefficient of variation CV, set at CV=0.1 (dashed), CV=0.2 (solid), or CV=0.3 (dotted). PASCL will not be the best choice for setting annual catch levels in every stock. In particular, data-poor stocks will likely require a different approach, such as assemblage management or the use of expert judg- ment. For rebuilding overfished stocks, other projection approaches may be more suitable (Jacobson and Cadrin, 2002; Punt, 2003). During rebuilding, harvest policies are typically based on the probability of stock recovery within a specified time horizon, rather than on the less restrictive constraint of preventing over- fishing. As overfished stocks recover, however, a method such as PASCL could be applied to prevent the stock from another decline and the need for future rebuilding plans. In one school of thought, choice of a harvest- control rule should be based on the likelihood of meeting long-term management objectives. In this regard, the efficacy of PASCL could be compared to that of other control rules by simulating the assessment and management processes in con- junction with stock dynamics (Cooke, 1999; Punt, 2003). Such management strategy evaluations can be useful for shedding light on which control rules work best under various conditions. However, they are complex, and thus difficult to program, verify, explain, and modify as circumstances change. Moreover, most fishery management is in fact based on short-term to medium-term consider- ations, and management strategies are likely to change to meet social, biological, or environmental conditions. When a major objective of management is to avoid overfishing, PASCL should be quite effective, and it may simultaneously meet more complex objectives. An advantage of simple control rules such as PASCL is that they can be applied after each assessment without major redesign. In our example, we computed annual catch levels as targets. With slight modification, the method can be used to compute annual catch limits and targets simultaneously. For example, a catch limit (or acceptable biological catch) might be set to prevent overfishing based on scientific uncertainty (e.g., process and estimation error), and a catch target might then be set lower than the limit to Shertzer et al Probability-based approach to setting annual catch levels 231 account for the implementation of uncertainty. In a projection over multiple years, however, lim- its and targets should remain coupled, because simulated catch feeds back to abundance levels and thus affects catch levels (limits and targets) in the next year. Correlation of fishing mortality rates PASCL, as implemented through Equation 2, implicitly treats the two variables Flim and Ft as independent. In some applications, the two may be correlated. To examine this correlation, we conducted a simulation analysis, in which an age- structured population model was used to generate data that were then used in a catch-age assess- ment model. In the population model, parameter values representing natural mortality, somatic growth, maturity, fishery selectivity, steepness, recruitment variability, landings, and indices of abundance were generated at random by using levels from other simulation studies (Maunder and Deriso, 2003; Williams and Shertzer, 2003). For simplicity, the simulation included one fishery and one index of abundance. The population was initi- ated near its unfished state and was subjected to a 27-year linear increase in fishing mortality rang- ing from 0 to 2 times the rate at FMSY (Fhm). Based on 10,000 simulations, the assessment output pro- vided no evidence of correlation between estimates of Flim and terminal-year Ft (r=-0.004, P=0.694), supporting the assumption of independence. In any given application, however, if correlation between Flim and Ft is considered important, PASCL could be applied by using Equation 3. A B Figure 6 Contours of (A) median annual catch levels (landings plus discard mortalities) and (B) median spawning biomass, both in units of 100 metric tons, and plotted as functions of time (years) and the acceptable risk of overfishing (P*). Uncertainty in management implementation was set at CV=0.2. Implementation uncertainty and bias Uncertainty in implementation is a common, but often ignored, reality of natural resource management (John- son et al., 1997). This source of uncertainty was quan- tified here by a normal distribution with an assumed coefficient of variation. In some cases, the distribution could be estimated from data on performance of the fishery — a scenario we expect to become more common as annual catch targets are applied more widely. Our example showed that more precise management allowed higher fishing rates, and thus larger catches, without increased probability of overfishing. Similarly, larger catches would result from more precision in stock dynamics or assessments. This outcome underscores the economic benefits of timely monitoring, enforcement, and compliance. A notable feature of PASCL is that managers may choose the level of risk that they consider acceptable. This choice could reflect socio-economic considerations, in addition to biological factors such as productivity and vulnerability of the stock. In some cases, higher risk of overfishing may be desired, for example if short term loss of yield outweighs long-term benefits (Shertzer and Prager, 2007). In other cases, managers may be more precautionary. Either way, establishing the level of risk as an explicit choice increases transparency in the management process. A simplifying assumption of our application was that annual catch, although imprecise, was centered about the target catch level. In many fisheries, however, the distribution of annual catch may be asymmetric in either direction of the target. Such asymmetric distri- butions can easily be accommodated in PASCL. When annual catch falls above or below the target, managers may consider adjusting target catch in subsequent years accordingly. Conclusion Over the next several years, as science responds to legislation, a body of practice will be developed to imple- ment annual catch limits and targets. In managing U.S. federal fisheries, new approaches must address the MSRA requirements to end and prevent overfishing. The 232 Fishery Bulletin 106(3) PASCL framework proposed here is intended to satisfy these requirements, and it certainly can be applied more broadly in fisheries where the management of catch levels has the objective of avoiding overfishing. Acknowledgments The authors are grateful for the support of the National Marine Fisheries Service Southeast Fisheries Science Center and for comments from J. Bence, P. Conn, D. Vaughan, and anonymous reviewers. Literature cited Baranov, F. I. 1918. On the question of the biological basis of fisheries. Nauchn. Issled. Ikhtiologicheskii Inst. Izv. 1:81-128. (Translated from Russian by W. E. Ricker.) Beverton, J. H., and S. J. Holt. 1957. On the dynamics of exploited fish populations, 533 p. Chapman and Hall, London, UK. Facsimile reprint, 1993. Caddy, J. F., and R. McGarvey. 1996. Targets or limits for management of fisheries? N. Am. J. Fish. Manag. 16:479-487. Cooke, J. G. 1999. Improvement of fishery-management advice through simulation testing of harvest algorithms. ICES J. Mar. Sci. 56:797-810. Hauser, C. E., E. G. Cooch, and J. D. Lebreton. 2006. Control of structured populations by harvest. Ecol. Model. 196: 462-470. Jacobson, L. D., and S. Cadrin. 2002. Stock-rebuilding time isopleths and constant-F stock-rebuilding plans for overfished stocks. Fish. Bull. 100:519-536. Johnson, F. A., C. T. Moore, W. L. Kendall, J. A. Dubovsky, D. F. Caithamer, J. R. Kelley Jr., and B. K. Williams. 1997. Uncertainty and the management of mallard harvests. J. Wildl. Manag. 61:202-216. Mace, P. M. 2001. A new role for MSY in single-species and eco- system approaches to fisheries stock assessment and management. Fish Fish. 2:2-32. Maunder, M. N., and R. B. Deriso. 2003. Estimation of recruitment in catch-at-age models. Can. J. Fish. Aquat. Sci. 60:1204-1216. Morison, A. K. 2004. Input and output controls in fisheries manage- ment: a plea for more consistency in terminology. Fish. Manage. Ecol. 11: 411—413. MSRA (Magnuson-Stevens Fishery Conservation and Manage- ment Reauthorization Act of 2006). 2006. Pub. L. No. 109-479, 120 Stat. 3575. Myers, R. A., K. G. Bowen, and N. J. Barrowman. 1999. Maximum reproductive rate of fish at low popula- tion sizes. Can. J. Fish. Aquat. Sci. 56:2404-2419. Prager, M. H., C. E. Porch, K. W. Shertzer, and J. F. Caddy. 2003. Targets and limits for management of fisheries: a simple probability-based approach. N. Am. J. Fish. Manag. 23:349—361. Punt, A. E. 2003. Evaluating the efficacy of managing West Coast groundfish resources through simulations. Fish. Bull. 101:860-873. Quinn II, T. J., and R. B. Deriso. 1999. Quantitative fish dynamics, 542 p. Oxford Univ. Press, New York, NY. Rose, K. A., J. H. Cowan, K. O. Winemiller, R. A. Myers, and R. Hilborn. 2001. Compensatory density dependence in fish popu- lations: importance, controversy, understanding, and prognosis. Fish Fish. 2:293-327. Rosenberg, A. A., and S. Brault. 1993. Choosing a management strategy for stock rebuild- ing when control is uncertain. In Risk evaluation and biological reference points for fisheries management (S. J. Smith, J. J. Hunt, and D. Rivard, eds.), p. 243- 249. Can. Spec. Publ. Fish. Aquat. Sci. 120. Shertzer, K. W., and M. H. Prager. 2007. Delay in fishery management: diminished yield, longer rebuilding, and increased probability of stock collapse. ICES J. Mar. Sci. 64:149-159. SEDAR (SouthEast Data Assessment and Review). 2004. SEDAR 4 stock assessment report 1: stock assess- ment of the deepwater snapper-grouper complex in the South Atlantic, 594 p. South Atlantic Fishery Man- agement Council, 4055 Faber Place, Suite 201, North Charleston, SC 29405. 2006. SEDAR 10 stock assessment report 1: South Atlan- tic gag grouper, 485 p. South Atlantic Fishery Man- agement Council, 4055 Faber Place, Suite 201, North Charleston, SC 29405. Venables, W. N., and B. D. Ripley. 2002. Modern applied statistics with S, 495 p. Springer, New York, NY. Walters, C. J., and S. J. D. Martell. 2004. Fisheries ecology and management, 448 p. Princ- eton Univ. Press, Princeton, NJ. 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. 233 Abstract — A nursery site for the Alaska skate (Ba thy raja parmifera) was sampled seasonally from June 2004 to July 2005. At the small nurs- ery site (~2 km2), located in a highly productive area near the shelf-slope interface at the head of Bering Canyon in the eastern Bering Sea, reproduc- tive males and females dominated the catch and neonate and juvenile skates were rare. Seasonal samples showed summertime (June and July) as the peak reproductive time in the nursery although some reproduction occurred throughout the year. Time- series analysis of embryo length fre- quencies revealed that three cohorts were developing simultaneously and the period of embryonic development was estimated at 3.5 years and aver- age embryo growth rate at 0.2 mm/ day. Estimated egg case deposition occurred mainly during summertime and hatching occurred during winter months. Protracted hatching times may be common for oviparous elas- mobranch species and may be directly correlated with ambient temperatures as evident from a meta-data analysis. Evidence indicates that the Alaska skate uses the eastern Bering Sea outer continental shelf region for reproduction and the middle and inner shelf regions as habitat for immature and subadults. Skate nurseries may be vulnerable to disturbances because they are located in highly productive areas and because embryos develop slowly. Manuscript submitted 26 November 2007. Manuscript accepted 21 February 2008. Fish. Bull. 106:233-244 (2008). The views and opinions expressed or implied in this article are those of the author and do not necessarily reflect the position of the National Marine Fisheries Service, NOAA. A nursery site of the Alaska skate {Bathyraja parmifera ) in the eastern Bering Sea Gerald R. Hoff Email address: jerry.hoff@noaa.gov Alaska Fisheries Science Center National Marine Fisheries Service, NOAA 7600 Sand Point Way NE Seattle, Washington 981 1 5 Elasmobranchs are of growing concern worldwide because they are threat- ened by increased fishing and habi- tat disturbances (Musick et al., 2000; Stevens et al., 2000). Characteristic life history traits of these fish include slow growth rates, late maturation, low fecundity, and long life-spans, all of which make them extremely vul- nerable to increased fishing-induced mortality (Dulvy, 1999; Frisk et al., 2002). Species with these life history patterns depend on high juvenile sur- vival and recruitment for population stability. An adequate understanding of reproduction dynamics and habitat requirements are lacking for most spe- cies, yet these may be the most criti- cal biological criteria for successful reproduction. Oviparous species such as skates (Rajidae) use nursery areas for egg deposition, embryo development, and hatching (Hitz, 1964; Hoff, 2007). They produce relatively large colla- gen egg cases (Knight et al., 1996) which contain a large yolk mass and developing embryo. The egg cases are deposited directly onto the seafloor and embryos develop independent of maternal care (Hamlett and Koob, 1999). The embryonic developmental period is unknown for most species of skates, but evidence indicates that it may exceed one year for temperate and deepwater species (Berestovskii, 1994). The Alaska skate (Bathyraja par- mifera) represents >95% of estimated skate biomass on the eastern Bering Sea shelf (20 to 200 m) (Lauth and Acuna, 2007), and estimates of bio- mass and population numbers have indicated nearly a fourfold increase since 1975 (Hoff, 2006). Skates at all life stages are encountered in the shelf environment and the species range is limited to depths of <400 m on the slope (Hoff and Britt, 2003, 2005; Stevenson et al., 2007). The Alaska skate reaches a large size (135 cm) and can be locally abun- dant (Hoff and Britt, 2005). Its dis- tribution pattern and accessibility in relatively shallow waters make the species a likely candidate in target fisheries, and its life histo- ry characteristics make it suscep- tible to population decreases (Matta, 2006; Matta and Gunderson, 2007). A nursery site for the Alaska skate was identified in June 2004 (Fig. 1) where significant numbers of skate egg cases were previously reported by commercial fishing crews and fisheries observers. The nursery ex- ists in an area that has been heavily fished for walleye pollock ( Theragra chalcogramma) and Pacific cod ( Ga - dus macrocephalus) for many years, and the site is frequently disturbed by bottom trawls. Understanding reproductive char- acteristics and essential habitat requirements are necessary to ac- curately predict population stability under changing conditions for elas- mobranch species worldwide. The recent discovery of a nursery site for the Alaska skate allowed a first look at the reproductive details of the species. Specifically, this study esti- mates the timing of egg deposition and duration of embryonic develop- ment through length-frequency mode tracking from a seasonal sampling of embryos. In addition, the habitat of the Alaska skate is examined with respect to nursery site use and life- stage distribution patterns. 234 Fishery Bulletin 106(3) A 168°W 167°W I66°W 165°W 164°W “ I I I I 168°W 167°W 166°W 165°W 164°W Figure 1 (A) Map of the eastern Bering Sea, Alaska, and location of a nursery site (dark circle) for the Alaska skate ( Bathyraja parmifera) at the head of Bering Canyon. (B) Locations at the nursery site for all trawls completed. The area in the box was designated as the index site and targeted for a 14-month sea- sonal study conducted during 2004-05. Materials and methods Sampling at the nursery site The Alaska skate nursery site was sam- pled by methods and gear similar to those used by the annual Alaska Fisheries Sci- ence Center (AFSC) standard eastern Bering Sea bottom trawl survey (see Lauth and Acuna, 2007). The samples were collected with an 83-112 eastern otter trawl with a 25. 3 -m headrope and 34.1-m footrope. The footrope consisted of a single firehose-wrapped chain lacking bobbins or discs, and bottom contact was monitored by using electronic tilt sen- sors designed by personnel at the AFSC. Towing speed varied from 2 to 3 knots and tows were made directly into prevailing currents and wind to help control towing speed. During each tow, starting and ending latitude, longitude, bottom depth, time, vessel speed, net height and width, and bottom temperatures were recorded with NETMIND (vers. 3.0, Northstar Technical Inc., St. John’s, Newfoundland, Canada) acoustic trawl mensuration gear. Area swept was estimated from the aver- age net width and distance fished during each tow. Egg case, fish, and invertebrate densities were estimated from the area swept and from the numbers within each category (egg case, fish, and invertebrate) caught in each trawl haul. An initial exploratory trawl was con- ducted in June 2004 to locate the skate nursery. A subsequent survey to deter- mine the spatial coverage of the nursery was conducted during July-August 2004. The extent of the nursery area was de- termined by using an adaptive trawling approach, where 1) trawls were conducted in each of four directions from 0.5 to 1.5 km apart and 2) a reduced egg case den- sity of <500 eggs/km2 was used as the criterion to indicate the farthest limits of the nursery (Fig. 1). The goal was to map the egg case distribution and estimate the size of the nursery area. The Alaska skate nursery site was sampled a total of eight times over the 14-month period in June, July, Septem- ber, and November of 2004 and January, April, June, and July of 2005. An index area was chosen during the July-August 2004 investigation and was sampled dur- ing each of the subsequent six seasonal sampling periods (September 2004 to July 2005). The index site was defined as an area where the skate eggs were pre- Hoff: A nursery site of Bathyra/a parmifera in the eastern Bering Sea 235 dominately in early stages of development (newly de- posited) and a large percentage were viable and at high densities (>50,000 eggs/km2) to allow tracking of embryonic development. The index site constituted an approximate 1-km2 area where the highest egg case density trawls were located. During each seasonal sam- pling period, a single 5 to 10 min bottom trawl targeted the index site. The data collected from seasonal sam- plings were similar to those previously described for the July-August 2004 trawl investigation; however because of time limitations, trawl data were limited to bottom depth, temperature, distance fished, and start and end latitude and longitude during seasonal sampling. Collection of biological data All skates were weighed and enumerated, or a weighed numerical subsample was used to estimate total num- bers from weighed samples. All egg cases were identified to species and documented as empty (posthatching) or full (prehatching), including eggs that may have been damaged by the trawl. A random sample of full egg cases was fixed in 10% formalin from each sampling period for embryo measurements. Density estimates for skates and egg cases were calculated as the number of eggs encountered per km2 by using area swept by the net and the number of individuals encountered in each trawl. All skates encountered were identified to species and sex, and total lengths (TL, to nearest cm) and weights (to nearest 0.1 g) were recorded. Biological data were collected on randomly selected Alaska skates during all seasonal sampling to determine maturity state, repro- ductive state, and diet composition. During the initial July 2004 sampling, 67 female and 45 male Alaska skates were examined and during all subsequent sam- pling periods from 2 to 12 males and 5 to 17 females were examined. For each skate sampled the species, sex, total fish length, total fish weight, stomach content weight, and general diet composition were recorded. Re- productive state of males and females were determined by following maturity stages detailed in Matta and Gunderson (2007). Embryo length-frequency measurements Formalin-fixed egg cases were neutralized and soaked in tap water for up to four days before measurements were taken. Egg cases were cut open, embryos excised from the yolk, and total lengths (TL) were measured to the nearest 0.5 mm. For analysis, lengths were rounded to the nearest millimeter. Measurements were taken from the anterior tip of the snout or disc to the posterior tip of the tail filament. Growth rates of Alaska skate embryos were estimated by following methods similar to those used for the ju- venile English sole ( Plueronectes vetulus) off the Wash- ington coast (Shi et ah, 1996). Natural mode breaks were used to demarcate cohorts from embryo length frequencies. A mean embryo length was estimated for each cohort at each sampling period and plotted along with the corresponding sampling date. A best-fit linear model was used to determine daily growth rates of each cohort. Embryonic growth was assumed linear throughout development, and the cohort data indicated that the linear model was applicable. Hatching dates were estimated by using a mean hatching size of 224 mm TL (mean of all near hatching embryos, n = 39) and the average growth rate from the linear regres- sion. Hatching-date estimates were defined as the time required to reach 224 mm TL based on the length and date of capture. Egg-deposition date was obtained for each embryo measured by back-calculating the time required to reach a size of 1 mm based on the length and date of capture and the average growth rate. An estimate of time between cohorts was calculated as the difference between mean lengths for each cohort divided by daily growth rate to obtain average time between each depositional event. Analysis of developmental period determined from the literature A review of previously published studies on hatching duration and rearing temperatures was synthesized for comparison with hatching duration and rearing tempera- tures obtained from this study. Species in this analysis were limited to oviparous elasmobranchs, which encom- passed a diverse group of 13 chondrichthyan fishes: a chimera, the spotted ratfish ( Hydrolagus colliei), two catshark species ( Scyliorhinus spp.), and ten species of skates in three genera (Raja, Leucoraja, and Okamejei) (Table 1). Species reviewed were found from subtropi- cal to temperate waters spanning a range of tempera- tures from 4.6°C to 24°C. Developmental periods and rearing temperatures were obtained from the reported literature for each study. When a range was reported for either developmental period or rearing temperature, an arithmetic mean was calculated from those values. Temperature and embryonic development period were plotted and a nonlinear regression algorithm was applied to the data. Habitat use The distribution of life stages of the Alaska skate was investigated by examining bottom trawl survey data from the eastern Bering Sea summertime groundfish survey of the AFSC for years 2000 through 2007 (Lauth and Acuna, 2007). Alaska skate density for each sta- tion was estimated as the summed catch per unit of effort (CPUE, number of skates/km2) obtained at each station for the eight years surveyed. Density estimates were calculated at each trawl station for each life his- tory stage of the Alaska skate: juvenile (<300 mm TL, newly hatched to age +1); immature (301-920 mm TL); and adults (>920 mm TL, average maturity size; Matta and Gunderson, 2007). Distribution maps were produced with ArcMap (vers. 8.3, Environmental System Research Institute (ESRI) Redlands, CA). 236 Fishery Bulletin 106(3) Table 1 Data from reported embryonic developmental periods and developmental temperatures for oviparous elasmobranch species from previously published st udies worldwide. Values are the arithmetic means of the reported values. Common name Scientific name Mean developmental period (days) Mean developmental temperature (°C) Source Thorny skate Raja radiata 912 4.6 Berestovskii, 1994 Little skate Leucoraja erinacea 279 10 Steele et al., 2004 Clearnose skate Raja eglanteria 63 24 Libby and Gilbert, 1960 Clearnose skate Raja eglanteria 82 21 Luer and Gilbert, 1985 Clearnose skate Raja eglanteria 85 20 Luer et al., 2007 Clearnose skate Raja eglanteria 368 9.1 Perkins, 1965 Big skate Raja binoculata 277 11.5 Hitz and Reid, 1968* Spiny rasp skate Okamejei kenojei 137 14.6 Ishihara et al., 2002 Thornback skate Raja clavata 139 14.9 Ellis and Shackley, 1995 Thornback skate Raja clavata 137 15.38 Clark, 1922 Thornback skate Raja clavata 170 14.31 Clark, 1922 Small-eyed ray Raja microcellata 217 14.66 Clark, 1922 Blonde skate Raja brachyura 217 14.13 Clark, 1922 Spotted skate Raja montagui 155 15.93 Clark, 1922 Cuckoo skate Leucoraja naevus 248 13.17 Clark, 1922 Chain catshark Scyliorhinus retifer 256 12.25 Castro et al., 1988 Small spotted catshark Scyliorhinus canicula 165 13.3 Ellis and Shackley, 1995 Small spotted catshark Scyliorhinus canicula 334 10 Thomason et al., 1996 Small spotted catshark Scyliorhinus canicula 205 16 Thomason et al., 1996 Small spotted catshark Scyliorhinus canicula 160 16 Ballard et al., 1993 Spotted ratfish Hydrolagus colliei 300 12.75 Dean, 1906 Alaska skate Bathyraja parmifera 1290 4.4 Hoff (this study) * Indicates an unpublished study for which the author has the original data. Results Sampling of nursery site The nursery site was relatively small in area, covering approximately 2 km2 for the highest egg case densities areas. During the initial July-August 2004 nursery investigation, 21 hauls were conducted and egg case densities ranged between 362 and 148,957 eggs/km2 (mean=19,470 ±36,030 eggs/km2). A single trawl con- taining 148,957 eggs/km'2 possessed >70% viable eggs and was designated as an index site for subsequent seasonal trawl sampling (Fig. 1). The seasonal trawl samples contained between 45,418 and 549,843 eggs/km2 (mean =199,683 ±181,467 eggs/km2) from the index site and 53-84% of the eggs per tow were viable. The Alaska skate and the Bering skate ( Bathyraja interrupta) were both found during most sampling periods. The Alaska skate predominated in abundance (96%) and in egg case composition (99.6%). Although the Bering skate accounted for about 4% of the skates found at the site, their egg cases contributed only 0.4%, indicating that this was mainly a single species nursery site for the Alaska skate. The most abundant fish species encoun- tered throughout the sampling period included walleye pollock, arrowtooth flounder ( Atheresthes stomias), flat- head sole ( Hippoglossoides elassodon), rex sole ( Glypto - cephalus zachirus ), and Pacific cod. The most abundant invertebrate species (from summer 2004 trawls) were Tanner crab ( Chionoecetes bairdi), tentacle-shedding anemone ( Liponema brevicornis), and Oregon triton ( Fusitriton oregonensis). Biological sampling The size composition of Alaska skate, for all samples combined, indicated that males and females of mature sizes and reproductive state used the nursery nearly exclusively of other posthatching stages; immature and newly hatched juvenile skates were rarely found (Fig. 2). Gonad examination revealed developed ovaries and egg cases in the uterus of female Alaska skates, and fully developed claspers and testes in males during all seasons examined, indicating sexual maturity and that skates were in actively reproducing states. Recent stud- ies have provided evidence that the eastern Bering sea populations of Alaska skate reach a mature state around 93 cm TL for both sexes (Matta and Gunderson, 2007); thus nearly all individuals found within the nursery site of the present study were of reproductive size. Seasonal nursery use was evident from trawl samples collected at the index site. The Alaska skate showed Hoff: A nursery site of Bathyra/a parmifera in the eastern Bering Sea 237 Skate length (TL, cm) Figure 2 Length frequency for all Alaska skates (Bathyraja parmifera ) for all trawl samples combined from the nursery site in the southeastern Bering Sea. The vertical dashed line on each graph indicates the skate length at 50% maturity (male, 91.75 cm; female, 93.28 cm) for the eastern Bering Sea Alaska skate population as determined by Matta and Gunderson (2007). trends of increased abundance (skate density) in the nursery area during summer months of June and July in 2004 and 2005 respectively, and few skates were found during the nonsummer months of January, April, September, and November (Fig. 3). Stomach analysis of the Alaska skate revealed wall- eye pollock to be the predominant prey consumed (81% by weight, n=195), followed by other fish species includ- ing flatfish, salmon, and unidentified fishes (14.6%), and invertebrate species (snow crabs [ Chionoecetes spp.] and shrimp that represented the third most important component by weight [4.4%]). Seasonal diet analyses revealed that feeding occurred throughout the year and that skates in advanced reproductive states (gravid) nearly always contained full stomachs. Embryo length-frequency analysis Embryo length-frequency modal shifts showed a mini- mum of three cohorts developing simultaneously during all sampling periods. The mean length of each cohort increased slightly at each subsequent sampling and showed a natural progression of development over time, and individual cohorts appeared and disappeared as time progressed (Fig. 4). The length data showed that a cohort appeared dur- ing the November 2004 sampling and persisted dur- ing the April, June, and July 2005 sampling periods. The approximate developmental period was 180 days from egg deposition (June) until the length samples revealed the presence of the cohort (November). The long period of early development was similar to that of the small spotted catshark ( Scyliorhinus canicula ) (Ballard et al., 1993) and the clearnose skate ( Raja eglanteria) (Luer et al., 2007); for these two species, it takes approximately 15-16% of the early developmen- tal period before an embryo is visible at 8 to 10 mm. This finding was similar to that for the Alaska skate; the smallest embryo visible at 15 mm had taken ap- proximately 14% of the development period to reach this size. Cohorts 1, 2, and 3 showed no difference in linear growth rates throughout their size ranges (test of slopes F=32.11, P=0.129; Fig. 5). The estimated daily growth rates (slopes) obtained for the cohort length data ranged from 0.18 to 0.22 mm/day (mean: 0.20 mm/day [Table 2]). The distribution of expected birthdates and egg deposition dates from the embryo length frequen- cies revealed that although there is continuous hatching and egg case deposition throughout the year, the peak hatching event occurs during fall and winter months (October to February: Fig. 6A) and egg case deposition peaks during spring and summer months (June to Au- gust: Fig. 6B). From average growth rates of 0.20 mm/day, an embry- onic development period of 3.5 years to reach 224 mm 238 Fishery Bulletin 106(3) Table 2 Mean length and standard deviation for each cohort of embryos during each sampling period for the Alaska skate (Bathyraja parmifera) at the eastern Bering Sea nursery site. A linear regression equation and growth rate for each cohort was estimated from mean embryo lengths. Number of embryos included in each mean length (±1 standard deviation) is denoted by n. The time lag between cohorts is the estimated interval between the cohorts — a period determined from mean cohort lengths and growth rates at each sampling period. Cohort 3 had hatched by 18 April 2005 sampling. Cohort 1 Cohort 2 Cohort 3 Sampling period mean length (mm) n mean length (mm) n mean length (mm) n 3 June 2004 41.6 ±8.09 20 91.5 ±15.12 10 187.89 ±7.88 9 27 July 2004 46.96 ±10.43 23 118.13 ±16.76 8 199.53 ±8.86 15 11 September 2004 57.44 ±8.13 73 131.89 ±12.97 46 207.82 ±9.57 11 17 November 2004 65.24 ±10.33 53 141 ±1.41 2 217 ±12.28 8 16 January 2005 84.17 ±6.19 18 156.38 ±13.31 21 233 ±11.31 2 18 April 2005 97.11 ±7.25 28 171.85 ±18.75 13 Hatching 1 June 2005 113.83 ±10.25 24 193.27 ±7.92 27 7 July 2005 116.07 ±11.45 57 200.5 ±0.71 7 Linear equation y=0.1819x+13.40 y=0.2214x+68.17 y=0.1899x+ 158.80 Growth rate of cohort 0.18 mm/day 0.22 mm/day 0.19 mm/day Average growth rate 0.20 ±0.02 mm/day Time lag between cohort 1 and 2 348.79 ±41.27 days Time lag between cohort 2 and 3 411.07 ±44.28 days TL was estimated. The time lag between cohorts 1 and 2 was mean=348 ±41 days and between cohorts 2 and 3 mean=411 ±44 days and an overall mean=371 ±50 days between egg deposition events of cohorts one, two, and three combined (Table 2). These data indicate an annual egg deposition cycle, and each cohort represents the result of a single reproductive event that occurs during the summer months. Hoff: A nursery site of Bathyra/a parmifera in the eastern Bering Sea 239 Figure 4 Length frequency for embryos of the Alaska skate ( Bathyraja parmifera) collected during each sampling date at the nursery index site in the eastern Bering Sea. Embryos were assigned to cohorts 1, 2, or 3 (separated by diagonal lines) following mode frequencies. A fourth cohort was evident that was detected in the November 2004 sampling and persisted through July 2005 (graph left) but was not used for growth-rate estimates. Analysis of published developmental periods Developmental period for the Alaska skate was the lon- gest period yet observed for any oviparous elasmobranch species. Embryonic developmental periods were highly correlated with the rearing temperatures for the other oviparous elasmobranchs included in this review (Fig. 7). Most studies were conducted in >8°C water tempera- tures and developmental periods were one year or less. However, a single study conducted on the thorny skates ( Raja radiata ) (Berestovskii, 1994) in water tempera- tures near those found for the Alaska skate revealed that the developmental periods of these two species to be comparable (thorny skate, 4.6°C, 912 days; Alaska skate, 4.4°C, 1290 days). 240 Fishery Bulletin 106(3) Habitat use Analysis of trawl catch data and depth soundings showed that the nursery site had little benthic struc- ture or habitat diversity and the bottom was generally 200 - -S 120 o Cohort 3 2004 2005 1 1 1 3 June 27 July 1 1 Sept. 17 Nov. 16 Jan. Nursery sampling date 1 1 1 18 April June 7 July Figure 5 Mean embryo total length (±1 standard deviation) at each sampling date for each cohort from the length-frequency data for the Alaska skate ( Bathyraja parmifera). Growth rates were estimated from the slope of the linear relationships for each cohort. _Q 14-1 E LU 12- 10- 4- 80 100 120 140 160 180 200 220 240 260 280 300 320 340 360 Julian date Egg deposition l,ll Jl I, illliH Feb Mar April May June July Month Aug Sept Figure 6 Estimated time of hatching (top) and egg deposition (bottom) frequency for all embryos measured from the Alaska skate ( Bathyraja parmifera) nursery site. Hatching and egg deposition dates were determined from the average growth rate of 0.2 mm/day and an estimated hatching size of 224 mm total length. The x-axis indicates the Julian date (top) and the corresponding month (bottom). flat, and composed of sandy mud. The trawl samples within the nursery did not contain any attached ben- thic invertebrates that would have constituted a unique habitat. Bottom depths varied by only several meters throughout the nursery site (145 to 150 m) and the average bottom depth was 149 m, weighted by egg case density. Bottom temperatures at the nursery index site varied little throughout the year, ranging between a low of 4.1°C in June 2004 and April 2005, to a high of 5.0°C in July of 2004. The mean bottom temperature for the 14-month study period was 4.40°C ±0.327°C. The Alaska skate was widely distributed across the eastern Bering Sea shelf and the immature stages used a different portion of the habitat than that used by newly hatched juveniles and mature adults. Juvenile skates were distributed along the outer continental shelf (100-200 m) and overlapped in distribu- tion with mature adults (Fig. 8). Immature Alaska skates were distributed mainly in the middle and inner shelf regions and were found less on the outer shelf. A model lifetime move- ment pattern for the Alaska skate indicates an ontogenetic shift in habitat use in which there is a cyclical movement across the shelf after hatching to the shallow inner shelf, fol- lowed by a return to the outer shelf as maturity is reached (Fig. 8). Discussion Recent advances in elasmo- branch biology have stressed the importance of identifi- cation and conservation of nursery sites for oviparous elasmobranchs (Ellis et al., 2004) . Understanding habitat requirements for skate repro- duction may be critical for successful management plans for these vulnerable species. The results presented here are the first reported dynamics of a skate nursery with regard to reproductive patterns and habitat use. Because of their inherent low fecundity and slow growth rates, skates may reproduce with distinct seasonal pulses, over protracted periods, or in some cases continuously throughout the year (Temple- man, 1982; Sulikowski et al., 2005) . Results from previous Hoff: A nursery site of Bathyra/a parmifera in the eastern Bering Sea 241 Figure 7 Relationship between rearing temperature and days to hatching determined from 21 published studies and 13 oviparous chondrichthyan species. Open circles rep- resent values from the literature and the shaded circle is estimated embryonic developmental period for the Alaska skate ( Bathyraja parmifera) in the pres- ent study. The hatching time of the Alaska skate was not used to determine the equation parameters. See Table 1 for data sources. studies have shown that the Alaska skate is reproductively active year-round and that peak egg production occurs during summer months (Mat- ta, 2006; Matta and Gunder- son, 2007). A distinct summer- time pulse of egg deposition was evident from the nursery site seasonal skate abundance data, and distinct cohorts of embryos were present in the nursery throughout the year. Embryonic development of the Alaska skate was estimat- ed to take over 3.5 years from egg deposition until hatching, and as a consequence multiple cohorts were developing simul- taneously at the nursery site. Embryonic developmental rates are most likely coupled with environmental temperatures and produce a Q10 effect where there is an exponential change in metabolic processes as tem- perature changes (Schmidt- Nielsen, 1997; Charnov and Gillooly, 2003). The sensitivity of the developmental period to temperature increases is significant; if one uses the regression equation pa- rameters, a mean increase of 0.5°C in environmental temperatures could decrease the developmental period of the Alaska skate by nearly 16% (~6 months) and there would be stronger effects as greater temperature changes occurred. This has dramatic implications on what influence climate change may have on the shelf- slope environment, and skate reproduction and recruit- ment. The dramatic increase or decrease in recruitment success due to environmental changes may become an important model parameter for stock assessments and management plans for elasmobranch species. For the estimate of growth rate for the Alaska skate in this study, linear growth was assumed during the developmental period and an effect of environmental temperature on growth rate was not considered because environmental temperatures varied little during this study. These variables recognizably may influence daily growth estimates and therefore the length of develop- mental period, however averaging across three years of cohorts may provide an accurate estimate of the rela- tively long developmental period for the Alaska skate, as well as for other oviparous species in cold waters. Linear growth during embryonic development for the size range reported here for the Alaska skate was simi- lar to that for the clearnose skate from approximately 18 mm through hatching (Luer et ah, 2007). Site selection criteria for skate nurseries are as of yet unknown, however areas of high biological productivity may be a requirement for nursery sites because of the protracted reproductive activity and energy require- ments of adults. The Alaska skate nursery site is in a region of high slope-shelf water transport and is one of the most productive regions in the eastern Bering Sea (Stabeno et ah, 1999); it has supported walleye pollock and Pacific cod bottom trawl fisheries for more than 25 years. Walleye pollock, the main food source of adult Alaska skates (Lang et ah, 2005), co-exist in the outer-shelf region during summertime (Kotwicki et ah, 2005). Results from the nursery seasonal diet analysis indicated that reproductively active skates feed throughout the year, almost exclusively on walleye pol- lock. A ready supply of food may allow skates to remain near the nursery site and minimize foraging excursions during protracted reproductive cycles. Adequate current flows and stable temperatures such as those encountered in the upper slope area of the eastern Bering Sea may be critical for successful hatching and embryo development. From early stages of development, the embryo is dependent on a constant current of fresh seawater to supply tissues with oxygen, remove metabolic waste (Hamlett and Koob, 1999), and prevent the egg case from being buried in sediment. Although strong currents pose a hazard to egg cases by transporting them out of the nursery site, this does not appear to happen frequently because egg cases are rarely found widely scattered outside the nursery, and within nurseries egg cases often cover a small area and are highly concentrated. The upper slope environ- ment provides a nearly constant bottom temperature through upwelled waters that inundate the outer shelf 242 Fishery Bulletin 106(3) 175°W 170'W I65°W 160°W B n5°w no°w i6s°w i60°w 175°W I70°W 165°W 160°W Figure 8 Distribution in the eastern Bering Sea of the Alaska skate ( Bathyraja parmifera) at three life stages: (A) neonate, (B) immature, and (C) mature. Data are from a summertime (June- July) bottom trawl survey conducted on the eastern Bering Sea shelf from 20-200 m by the Alaska Fisheries Science Center. Dots represent a summation of catch per unit of effort (number skates/km2) at each survey station for the eight survey years 2000-07. Life stages are defined as the following: neonate < 310 mm total length; immature = 310-920 mm total length; and adult > 920 mm total length. (D) Diagram of life-time movement patterns for the Alaska skate in the eastern Bering Sea based on summertime survey data. Young skates are distributed along the outer continental shelf edge where nursery sites occurred. The newly hatched skates move to the middle and inner shelf where they remain until returning to the outer continental shelf at maturity. (Pavlov and Pavlov, 1996; Luchin et al., 1999) and the nursery experiences relatively stable year-round water temperatures from 4.1° to 5. CPC. By comparison, middle shelf bottom waters are extremely cold throughout the year and may reach <0°C (Luchin et al., 1999; Lauth and Acuna, 2007). Annually the upper slope water may provide the most stable and relatively warm environ- ment for embryo development. Hoff: A nursery site of Bathyra/a parmifera in the eastern Bering Sea 243 Habitat requirements of newly hatched juvenile skates may not be the criteria for nursery-site selec- tion because very few newly hatched skates were found at the site. The most likely explanation is that neo- nate skates move out of the nursery area shortly after emergence, possibly to reduce intraspecific competition along the slope edge or to avoid large predators that prey on juvenile skates (Hoff, 2007). In trawl studies with the same designs and methods as those of the present study and conducted across the outer conti- nental shelf region, newly hatched juvenile Alaska skates were found to be common (Kotwicki and Wein- berg, 2005; Lauth and Acuna, 2007). Many individu- als encountered on the outer shelf still possessed tail filaments, providing evidence of recent emergence from the egg case (Hoff, 2007). The reason why juvenile skate move to inner and outer shelf waters remains unclear; however, a pattern in their movements from inner to outer shelf regions in the eastern Bering Sea and in their use of different habitats at different life stages is evident. These patterns indicate that skate nurseries may be occupied by skates at two of most critical life stages (embryos and adults), and conservation efforts tar- geting these stages may have the greatest impact on population protection (Frisk et al., 2002, 2004). Skate nurseries in general may be located in areas of high biological productivity and therefore are susceptible to disturbances caused by increased fishing activities. As oviparous elasmobranch fishing mortality increases, recognition and protection of nursery habitats may be one approach to ensure healthy populations. Likewise, nursery sites may become important locations to be monitored for the health and recruitment potential of skate species. Long-term monitoring of these impor- tant habitats can provide a wealth of information with minimal effort because of the permanence or long-term stability of these nursery sites. Acknowledgments I thank the skippers and crew of the FVs Ocean Explorer, Sea Storm, Nordic Fury, Arcturus, Aldebaran, and Great Pacific. I especially thank D. Stevenson, S. Kotwicki, S. Gaichas, R. Reuter, E. Iwamoto, E. Acuna, E. Jor- genson, C. Gredzens, B. Voss, B. Lauth, R. Nelson, G. Stauffer, and G. Walters for their support. I also thank T. Essington, D. Gunderson, D. Kimura, T. Pietsch, and C. Rooper for their review and helpful suggestions to improve this manuscript. This project was supported by North Pacific Research Board (NPRB grant no. 415), Essential Fish Habitat (EFH), and the Alaska Fisheries Science Center. Literature cited Ballard, W. W., J. Mellinger, and H. Lechenault. 1993. A series of normal stages for development of Scyli- orhinus canicula, the lesser spotted dogfish (Chondrich- thyes: Scyliorhinidae). J. Exp. Zool. 267:318-336. Berestovskii, E. G. 1994. Reproductive biology of the family Rajidae in the seas of the far north. J. Ichthyol. 34(61:26-37. Castro, J. I., P. M. Bubucis, and N. A. Overstrom. 1988. The reproductive biology of the chain dogfish, Scyliorhinus retifer. Copeia 1988:740-746. Charnov, E. L., and J. F. Gillooly. 2003. Thermal time: body size, food quality and the 10°C rule. Evol. Ecol. Res. 5:43-51. Clark, R. S. 1922. Rays and skates (Rajidae). No 1. Egg-capsules and young. J. Mar. Biol. Assoc. U.K. 12(4):577-643. Dean, B. 1906. Chimaeroid fishes and their development. Carn- egie Institution of Washington 32:1-194. Dulvy, N. K., J. D. Metcalfe, J. Glanville, M. G. Pawson, and J. D. Reynolds. 1999. Fishery stability, local extinctions, and shifts in community structure in skates. Conserv. Biol. 14(11:283-293. Ellis, J. R., A. Cruz-Martinez, B. D. Rackham, and S. I. Rogers. 2004. The distribution of chondrichthyan fishes around the British Isles and implications for conservation. J. Northw. Atl. Fish. Sci. 35(5):195-213. Ellis, J. R., and S. E. Shackley. 1995. Observations on egg-laying in the thornback ray. J. Fish Biol. 46:903-904. Frisk, M. G., T. J. Miller, and N. K. Dulvy. 2004. Life histories and vulnerability to exploitation of elasmobranchs: inferences from elasticity, perturbation and phylogenetic analyses. J. Northw. Atl. Fish. Sci. 35(4):27-45. Frisk, M. G., T. J. Miller, and M. J. Fogarty. 2002. The population dynamics of little skate Leucoraja erinacea, winter skate Leucoraja ocellata, and barndoor skate Dipturus laevis: predicting exploitation limits using matrix analysis. ICES J. Mar. Sci. 59:576-586. 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.), p. 398-443. Johns Hopkins Univ. Press, Balti- more, MD. Hitz, C. R. 1964. Observations on egg cases of the big skate ( Raja binoculata Girard) found in Oregon coastal waters. J. Fish. Res. Board Can. 21(4):851-854. Hoff, G. R. 2006. Biodiversity as an index of regime shift in the eastern Bering Sea. Fish. Bull. 104:226-237. 2007. Reproductive biology of the Alaska skate Bathy- raja parmifera, with regard to nursery sites, embryo development and predation. Ph. D. diss., 161 p. Univ. Washington, Seattle, WA. Hoff, G. R., and L. L. Britt. 2003. The 2002 eastern Bering Sea upper continental slope survey of groundfish and invertebrate resources. U.S. Dep. Commer., NOAA Tech. Memo. NMFS-AFSC- 141, 261 p. 2005. Results of the 2004 eastern Bering Sea upper continental slope survey of groundfish and invertebrate resources. U.S. Dep. Commer., NOAA Tech. Memo. NMFS-AFSC-156, 276 p. Ishihara, H., T. Mochizuki, K. Homa, and T. Taniuchi. 2002. Reproductive strategy of the Japanese common 244 Fishery Bulletin 106(3) skate (spiny rasp skate) Okamejei kenojei. In Elas- mobranch biodiversity, conservation and management, p. 236-240. Occasional Paper of the IUCN Species Survival Commission 25. Knight, D. P., D. Feng, and M. Stewart. 1996. Structure and function of the Salachian egg case. Philos. Trans. R. Soc. Lond., Ser. B: Biol. Sci. 71:81-111. Kotwicki, S., and K. L. Weinberg. 2005. Estimating capture probability of a survey bottom trawl for Bering Sea skates (Bathyraja spp.) and other fish. Alaska Fish. Res. Bull. 11(2):135— 145. Kotwicki, S., T. W. Buckley, T. Honkalehto, and G. Walters. 2005. Variation in the distribution of walleye pol- lock (Theragra chalcogramma ) with temperature and implications for seasonal migration. Fish. Bull. 103:574-587. Lang, G. M., P. A. Livingston, and K. A. Dodd. 2005. Groundfish food habits and predation on commer- cially important prey species in the eastern Bering Sea from 1997 through 2001. U.S. Dep. Commer., NOAA Tech. Memo. NMFS-AFSC-158, 230 p. Lauth, R. R., and E. Acuna. 2007. Results of the 2006 eastern Bering Sea continental shelf bottom trawl survey of groundfish and inverte- brates resources. U.S. Dep. Commer., NOAA Tech. Memo. NMFS-AFSC-176, 175 p. Libby, E. L„ and P. W. Gilbert. 1960. Reproduction in the clear-nosed skate, Raja eglanteria. Anat. Rec. 138(3):365. Luchin, V. A., V. A. Menovshchikov, V. M. Lavrentiev, and R. K. Reed. 1999. Thermohaline structure and water masses in the Bering Sea. In Dynamics of the Bering Sea (T. R. Loughlin, and K. Ohtani, eds.), p. 61-91. Univ. Alaska Sea Grant Rep. AK-SG-99-03, Fairbanks, AK. Luer, C. A., and P. W. Gilbert. 1985. Mating behavior, egg deposition, incubation period, and hatching in the clearnose skate. Raja eglanteria. Environ. Biol. Fish. 13:161-171. Luer, C. A., C. J. Walsh, A. B. Bodine, and J. T. Wyffels. 2007. Normal embryonic development in the clearnose skate, Raja eglanteria , with experimental observa- tions on artificial insemination. Environ. Biol. Fish. 80:239-255. Matta, M. E. 2006. Aspects of the life history of the Alaska skate, Bathyraja parmifera, in the eastern Bering Sea. M.S. thesis, 86 p. Univ. Washington, Seattle, WA. Matta, M. E., and D. R. Gunderson. 2007. Age, growth, maturity, and mortality of the Alaska skate, Bathyraja parmifera, in the eastern Bering Sea. Environ. Biol. Fish. 80:309-323. Musick, J. A., G. Burgess, G. Cailliet, M. Camhi, and S. Fordham. 2000. Management of sharks and their relatives (Elasmobranchii). Fisheries 25(3):9-13. 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. 1-96. Univ. of Alaska Sea Grant Rep. AK-SG-96-01, Fairbanks, AK. Perkins, F. E. 1965. Incubation of fall-spawned eggs of the little skate, Raja erinacea (Mitchill). Copeia 1965:114-115. Schmidt-Nielsen, K. 1997. Animal physiology: adaptation and environment, 5th ed., 612 p. Cambridge Univ. Press, Cambridge, LT.K. Shi, Y., D. R. Gunderson, and P. J. Sullivan. 1996. Growth and survival of 0+ English sole, Pleuro- nectes vetulus, in estuaries and adjacentnearshore waters off Washington. Fish. Bull. 95:161-173. Stabeno, P. J., J. D. Schumacher, and K. Ohtani. 1999. The physical oceanography of the Bering Sea. In Dynamics of the Bering Sea (T. R Loughlin, and K. Ohtani, eds.), p.1-28. Univ. Alaska Sea Grant Rep. AK-SG-99-03, Fairbanks, AK. Steele, S. L., P. H. Yancey, and P. A. Wright. 2004. Dogmas and controversies in the handling of nitrog- enous wastes: osmoregulation during early embryonic development in the marine little skate Raja erinacea ; response to changes in external salinity. J. Exp. Biol. 207:2021-2031. Stevens, J. D., R. Bonfil, N. K. Dulvy, and P. A. Walker. 2000. The effects of fishing on sharks, rays, and chimae- ras (chondricthyans), and the implications for marine ecosystems. ICES J. Mar. Sci. 57:476-494. Stevenson, D. E., J. W. Orr, G. R. Hoff, and J. D. McEachran. 2007. Emerging patterns of species richness, diversity, population density, and distribution in the skates (Raji- dae) of Alaska. Fish. Bull. 106:24-39. Sulikowski, J. A., J. Kneebone, S. Elzey, J. Jurek, P. D. Danley, W. H. Howell, and P. C. W. Tsang. 2005. The reproduction cycle of the thorny skate (Ambly- raja radiata) in the western Gulf of Maine. Fish. Bull. 103:536-543. 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. Thomason, J. C., W. Conn, E. L. Comte, and J. Davenport. 1996. Effect of temperature and photoperiod on the growth of the embryonic dogfish, Scyliorhinus canicula. J. Fish Biol. 49:739-742. 245 Determination of population structure and stock composition of chum salmon ( Oncorhynchus keta) in Russia determined with microsatellites Nataly V. Varnavskaya Kamchatka Fishery and Oceanography Research Institute 18 Naberezhnaya Street Petropavlovsk-Kamchatsky 683000, Russia Abstract — Variation at 14 microsat- ellite loci was examined in 34 chum salmon (Oncorhynchus keta) popula- tions from Russia and evaluated for its use in the determination of popu- lation structure and stock composi- tion in simulated mixed-stock fishery samples. The genetic differentiation index (Fst) over all populations and loci was 0.017, and individual locus values ranged from 0.003 to 0.054. Regional population structure was observed, and populations from Pri- morye, Sakhalin Island, and north- east Russia were the most distinct. Microsatellite variation provided evidence of a more fine-scale popu- lation structure than those that had previously been demonstrated with other genetic-based markers. Analy- sis of simulated mixed-stock samples indicated that accurate and precise regional estimates of stock composi- tion were produced when the micro- satellites were used to estimate stock compositions. Microsatellites can be used to determine stock composition in geographically separate Russian coastal chum salmon fisheries and provide a greater resolution of stock composition and population structure than that previously provided with other techniques. Manuscript submitted 18 December 2007. Manuscript accepted 25 February 2008. Fish. Bull. 106:233-256 (2008). The views and opinions expressed or implied in this article are those of the author and do not necessarily reflect the position of the National Marine Fisheries Service, NOAA. Terry D. Beacham (contact author) Email address: Beachamt@pac.dfo-mpo.gc.ca Department of Fisheries and Oceans Pacific Biological Station Nanaimo, British Columbia, Canada V9T 6N7 Khai D. Le Michael H. Wetklo Department of Fisheries and Oceans Pacific Biological Station Nanaimo, British Columbia, Canada V9T 6N7 In Asia, there are two distinct types of chum salmon ( Oncorhynchus keta Walbaum). The early-matur- ing or “summer” chum salmon gen- erally returns to spawn from June through August in streams border- ing Kamchatka, the Sea of Okhotsk, the east coast of Sakhalin Island, and the Amur River. Later-matur- ing or “autumn” chum salmon gener- ally return to spawn from September through November in streams in Japan, the southern Kuril Islands, the west coast of Sahkalin Island, and the Amur River (Sano, 1966). In general, summer chum salmon spawn in areas where egg incubation occurs in subsurface stream flow, whereas autumn chum salmon spawn in areas of groundwater upwelling (Volobuyev et al., 1990). In major river drain- ages, autumn chum salmon generally migrate further up the drainage to spawn than do summer chum salmon, and are larger, younger, and more fecund than the summer-run fish (Sano, 1966). Determination of the origin of salmon in mixed-stock fisheries is im- portant for effective management. For chum salmon in Asia, scale pattern variation has provided a technique for the determination of origin of in- dividuals to large geographic areas (Tanaka et al., 1969; Ishida et al., 1989), and in some cases reportedly to a specific river drainage (Nikolayeva and Semenets, 1983). Trace elements in otoliths have also been reported to be effective for stock identification of Korean populations (Sohn et al., 2005). Stock identification techniques based on scale pattern analysis have generally been replaced by applica- tions based on genetic variation, ow- ing to the increased resolution that is possible by applying genetic variation (see example outlined by Wilmot et al. [1998]). Analyses of genetic variation have been demonstrated to be effec- tive in determining salmonid popula- tion structure, as well as determin- ing origins of salmon in mixed-stock fisheries. For Russian chum salmon, analyses of allozyme variation have indicated differentiation among popu- lations on the east and west coasts of Kamchatka (Winans et al., 1994), and either marignal (Salmenkova et al., 2007) or some level of differentia- tion between populations on Sakhalin Island and populations on the main- land Russian coast (Efremov, 2001). Populations in the far northeastern 246 Fishery Bulletin 106(3) portions of mainland Russia were distinct from popula- tions in western Alaska (Wilmot et al., 1994). Surveys of allozyme variation have generally indicated regional population differentiation among Russian populations. DNA-level markers have substantially increased the number of polymorphic loci that are available to be included in analyses of genetic variation. Initial sur- veys of mitochondrial (mt) DNA variation indicated regional differentiation between Sakhalin Island and mainland populations (Ginatulina, 1992). Later analy- ses of additional mtDNA variation indicated marked differentiation between Japanese and Russian popula- tions (Sato et al., 2004), and some differentiation among Russian populations (Brykov et al., 2003; Polyakova et al., 2006). Limited examinations of minisatellite varia- tion have indicated some level of differentiation between Japanese and Russian populations, but have yielded little evidence of regional structure for Russian popula- tions (Taylor et al., 1994; Beacham, 1996). Analyses of microsatellite variation have been ef- fective for determining salmonid population structure in local areas (Small et al., 1998; Banks et al., 2000; Beacham et al., 2004), as well as broad-scale differences across the Pacific Rim (Beacham et al., 2005, 2006). Microsatellites have also been of considerable value in estimating stock composition in mixed-stock salmon fisheries, on both a population-specific (Beacham et al., 2003) and regional basis (Beacham et al., 2006). Micro- satellite variation in chum salmon provides the means to examine fine-scale population structure (Chen et al., 2005), as well as the means for fine-scale estimation of stock composition in mixed-stock fisheries (Beacham et al., in press). Analyses of microsatellite variation in Russian chum populations would likely be of value by providing increased resolution of population structure compared with that provided by previous techniques, and would likely aid in increasing accuracy and preci- sion of estimates of stock composition in mixed-stock fishery samples. Our objectives were to analyze the variation at 14 microsatellite loci to evaluate population structure of Russian chum salmon populations from the far north eastern coast of Russia to the more southern areas of Primorye and Sakhalin Island, and then to evaluate the use of these loci for the practical purpose of providing accurate and precise estimates of stock composition in mixed-stock fishery samples. Stock composition evalu- ation was accomplished by the analysis of simulated mixed-stock fishery samples. Materials and methods Tissue samples were collected from mature chum salmon at a number of rivers during previous analyses of genetic variation (Winans et al., 1994). Additional tissue samples were sent to the Molecular Genetics Laboratory at the Pacific Biologi- cal Station. The geographic area of the 34 populations sampled ranged from Primorye in the south to northeastern Russia (Fig. 1) and encompassed eight geographic regions (Table 1). DNA was extracted from the tissue samples by a variety of methods, including that with chelex resin outlined by Small et al. (1998), a Qiagen 96-well Dneasy® procedure (Qiagen, Mississauga, Ontario, Canada), or a Promega Wizard SV96 Genomic DNA Puri- fication system (Promega, Madison, WI). Once extracted DNA was available, anal- yses of variation at 14 microsatellite loci were conducted: Ots3 (Banks et al., 1999), Oke3 (Buchholz et al., 2001), Oki2 (Smith et al., 1998), OkilOO (primer sequence 5' to 3' F: GGTGTTTTAATGTTGTTTCCT, R: GTTTCCAGAGTAGTCATCTCTG), Omml070 (Rexroad et al., 2001), Omy 1011 (Spies et al., 2005), One 101, Onel02, Onel04, Onelll, and Onell4 (Olsen et al., 2000), Otsl03 (Nelson and Beacham, 1999), Ssa419 (Cairney et al., 2000), and OtsG68 (Williamson et al., 2002). In general, PCR DNA amplifications were conducted by using DNA Engine Cycler Tet- rad2 (BioRad, Hercules, CA) in 6-pL volumes consisting of 0.15 units of Taq polymerase, 1 pL (25-50 ng) of extracted DNA, lx PCR buf- fer (Qiagen, Mississauga, Ontario, Canada), !40°£ 180° Figure t Map indicating the locations in Russia where chum salmon ( Oncorhyn - chus keta) from 34 populations or sampling sites were collected. Numbers for and locations of populations are indicated in Table 1. Beacham et al: Population structure and stock identification of Oncorhynchus keta 247 Population, sample collection years, number of fish sampled per year, and total number of fish sampled for 34 populations of chum salmon ( Oncorhynchus keta) in eight geographic regions from Russia. Eight regions have been defined, and populations (numbered in brackets) were sampled in each region listed. N = population size. Region and population Years Annual sample size N 1 Primorye Narva [11 1994 17 17 Ryazanovka [2] 1994 49 49 Avakumovka [3] 1994 35 35 2 Amur River Amur River [4] 1994, 2001, 2004 43, 97, 198 338 3 Sakhalin Island Tym [5] 1995 55 55 Naiba [6] 1994, 1995 50, 99 149 Udarnitsa [7] 1994 50 50 Kalininka [8] 1994 49 49 4 Magadan Tugur [9] 2004 98 98 Okhota [10] 2004 94 94 Magadan [11] 1991 79 79 Tauy [ 12] 1990 55 55 Ola [13] 1990, 1992 80, 40 120 5 Northern Sea of Okhotsk Oklan [14] 1993 76 76 Penzhina [15] 1993 43 43 6 West Kamchatka Hairusova [16] 1990, 1993 138, 48 186 Vorovskaya [17] 1991, 1993 79, 170 249 Kol [18] 1991 79 79 Pymta [19] 1992, 1993 40, 59 99 Kikchik [20] 1992, 2005 20, 86 106 Utka [21] 1992 40 40 Bolshaya [22] 2004 96 96 Plotnikova [23] 2001 69 69 7 East Kamchatka Zhypanova [24] 2004 46 46 Kamchatka [25] 1990 76 76 Ivashka [26] 2005 48 48 Nerpichi [27] 1992 39 39 Karaga [28] 2005 42 42 Ossora [29] 1990, 1996, 2005 39,41,48 128 Dranka [30] 2005 44 44 Apuka [31] 2002 47 47 Olutorsky Bay [32] 2002 49 49 8 Northeast Russia Anadyr [33] 1991, 1992 79, 15 94 Kanchalan [34] 1991 79 79 60 juM each nucleotide, 0.40 pM of each primer, and de- ionized water. The thermal cycling profile involved one cycle of 15 minutes at 95°C, followed by 30-40 cycles of 20 seconds at 94°C, 30-60 seconds at 47-65°C, and 30-60 seconds at 68-72°C (depending on the locus). Specific PCR conditions for a particular locus could vary from this general outline. PCR fragments were initially size fractionated in denaturing polyacrylamide gels by using an ABI 377 automated DNA sequencer, and genotypes were scored by Genotyper 2.5 software (Applied Biosystems, Foster City, CA) using an internal lane sizing standard. Later in the study, microsatellites 248 Fishery Bulletin 106(3) were size fractionated in an ABI 3730 capillary DNA sequencer, and genotypes were scored by GeneMapper software 3.0 (Applied Biosystems, Foster City, CA) by using an internal lane-sizing standard. Allele identifi- cation between the two sequencers was standardized by analyzing the same approximately 600 individuals on both platforms and converting the sizing in the gel- based data set to match that obtained from the capil- lary-based set. Data analysis Each population at each locus was tested for depar- ture from Hardy-Weinberg equilibrium (HWE) by using genetic data analysis (GDA). Critical significance levels for simultaneous tests (34 populations, Table 1) were evaluated using Bonferroni adjustment (0.05/34 = 0.0015) (Rice, 1989). All annual samples available for a loca- tion were combined to estimate population allele fre- quencies, as was recommended by Waples (1990). Fst estimates for each locus were calculated with FSTAT (Goudet, 1995), individual locus values were determined by jackknifing over populations, and the overall Fst esti- mate was determined by jackknifing over loci (Goudet, 1995). Inter-regional comparisons of Fst estimates were determined by calculation of all appropriate pairwise point estimates of Fst values, and then determining the mean and standard deviation of these values. The Cavalli-Sforza and Edwards (CSE) (1967) chord distance was used to estimate distances among populations. An unrooted neighbor-joining tree based upon CSE was generated using NJPLOT (Perriere and Gouy, 1996). Bootstrap support (by sampling loci) for the major nodes in the dendrogram was evaluated with the CONSENSE program from PHYLIP (Univ. Washington, Seattle, WA) and based on 500 replicate trees. Computation of the number of alleles observed per locus, as well as allelic diversity standardized to a common sample size, was carried out with FSTAT. Estimation of stock composition Genotypic frequencies were determined for each locus in each population, and the Statistical Package for the Analysis of Mixtures software program (SPAM, vers. 3.7, Debevec et al., 2000) was used to estimate stock composition of simulated mixtures. The Rannala and Mountain (1997) correction to baseline allele frequencies was used in the analysis in order to accommodate the occurrence of fish in the mixed sample that were from a specific population having an allele not observed in the baseline samples from that population. All loci were considered to be in Hardy-Weinberg equilibrium, and expected genotypic frequencies were determined from the observed allele frequencies. Reported stock composi- tions for simulated fishery samples were the bootstrap mean estimate of each mixture of 150 fish analyzed, and mean and variance estimates were derived from 100 bootstrap simulations. Both the baseline popula- tion and the simulated single-population were sampled with replacement in order to simulate random variation involved in the collection of the baseline and fishery samples. Results Variation within and among populations The observed number of alleles observed at a locus ranged from 21 alleles at Oke3 and Oki2 to 138 alleles at Onelll (Table 2). Lower expected heterozygosities were generally observed at loci with fewer alleles. The genotypic frequencies observed at the 14 loci generally conformed to those expected under Hardy-Weinberg equilibrium (HWE) after Bonferroni correction. For the Oke3 and OtsG68 loci, a minor HWE nonconformance of genotypic frequencies was observed, and observed hetero- zygosities were 2-6% less than those expected (Table 2). Genetic diversity, with respect to the number of al- leles observed, was evident among regional groups of chum salmon. Chum salmon populations from Primo- rye, the northern Sea of Okhotsk, and northeast Rus- sia displayed fewer alleles (mean 320 alleles) than did populations in Magadan, west Kamchatka, and east Kamchatka (mean 370 alleles) (Table 3). Chum salmon from the latter regions displayed approximately 16% more alleles than did those from the former regions. The greatest differentiation in allelic diversity was observed at those loci with greater numbers of alleles, particularly at locus Onelll. Population structure Genetic differentiation was evident among chum salmon populations from the different geographic regions sur- veyed. The Fst value over all 34 populations and 14 loci surveyed was 0.017, and individual locus values ranged from 0.003 ( Onel02 ) to 0.054 ( Ots3 ) (Table 2). Chum salmon populations from Primorye and the Amur River were well defined compared with other regional popula- tions (Table 4). Populations from the southwestern por- tion of Russia (Primorye, Amur River, Sakhalin Island) were most distinct from those in more northern and eastern regions (Magadan, Sea of Okhotsk, Kamchatka, northeast Russia). Regional clustering of population samples was ob- served in the analysis of population structure. Strong clustering of population samples from the Primorye region was observed; the three population samples in- cluded in the analysis clustered together in 100% of the trees examined (Fig. 2). Similarly, strong clustering was observed in most of the population samples from Sakhalin Island, as well as the two population samples from the northern coast of the Sea of Okhotsk. The two population samples from northeast Russia clustered to- gether in 100% of the trees examined, together with the Utka River population sample from west Kamchatka. Although there was a general clustering of population samples from east and west Kamchatka, these regional Beacham et al: Population structure and stock identification of Oncorhynchus keta 249 Table 2 Number of alleles per locus, Fst, expected heterozygosity (He), observed heterozygosity (Ho), and percent significant Hardy-Wein- berg equilibrium tests (HWE) for 14 microsatellite loci examined in 34 populations of chum salmon ( Oncorhynchus keta) listed in Table 1. Standard deviation of Fst is shown in parentheses. Locus Number of alleles Fst He Ho HWE Oke3 21 0.050(0.022) 0.80 0.74 8.6 Oki2 21 0.027 (0.005) 0.87 0.87 2.9 Ots3 26 0.054 (0.014) 0.78 0.78 0.0 OkilOO 29 0.015 (0.003) 0.91 0.91 0.0 Ssa419 30 0.010(0.003) 0.87 0.87 0.0 0nel02 33 0.003 (0.001) 0.91 0.91 0.0 0 my 10 11 34 0.010(0.003) 0.93 0.92 0.0 Onel04 41 0.013 (0.003) 0.94 0.94 0.0 OnelOl 46 0.052 (0.009) 0.84 0.84 0.0 One 114 46 0.015 (0.004) 0.93 0.93 5.7 Omml070 50 0.005 (0.001) 0.95 0.95 0.0 Otsl03 51 0.008 (0.002) 0.95 0.94 0.0 OtsG68 53 0.005 (0.001) 0.95 0.93 8.6 Onelll Total 138 0.007 (0.001) 0.017 (0.004) 0.98 0.98 0.0 Table 3 Mean number of alleles observed per locus at 14 microsatellite loci for 34 chum salmon (Oncorhynchus keta) populations from eight regions as outlined in Table 1. Regions were the following: 1) Primorye, 2) Amur River, 3) Sakhalin Island, 4) Magadan, 5) Northern Sea of Okhotsk, 6) West Kamchatka, 7) East Kamchatka, and 8) Northeast Russia. Allele numbers have been stan- dardized to a sample size of 79 fish per region. 1 2 3 4 5 6 7 8 Oke3 9.0 8.5 9.9 8.7 7.9 8.7 9.7 7.0 OkilOO 16.0 18.5 22.3 21.4 17.7 19.6 21.4 19.2 Oki2 14.0 12.1 15.1 14.4 14.7 13.8 16.3 14.5 Omml070 30.1 26.6 29.0 33.3 27.7 32.3 32.9 29.5 OmylOll 22.0 20.0 22.6 23.7 25.2 22.7 21.8 23.6 OnelOl 20.7 28.3 21.9 25.0 16.6 22.8 23.3 23.1 One 102 24.1 19.3 20.1 17.7 16.8 18.5 18.9 14.4 One 104 23.8 27.6 23.5 23.9 23.4 24.6 24.9 22.7 Onelll 51.7 50.8 55.8 71.8 57.9 75.1 71.8 60.7 One 114 23.1 24.6 24.0 28.6 19.2 27.3 26.4 25.5 Ots3 12.0 14.1 14.3 14.0 9.7 14.7 16.1 12.9 OtslOS 30.0 35.3 38.4 34.4 27.3 35.1 34.4 29.1 OtsG68 32.9 34.2 39.3 35.6 28.7 36.3 36.6 32.7 Ssa419 16.5 14.8 16.0 16.7 14.4 18.7 16.2 11.9 Total 325.9 334.7 352.2 369.2 307.2 370.2 370.7 326.8 groupings were not strongly supported in the cluster analysis. Stock identification Genetic differentiation observed among the chum salmon in the regions surveyed was evaluated to determine if it was sufficient for a mixed-stock analysis with the objec- tive of obtaining accurate regional stock compositions. Analysis of simulated single-region samples indicated that estimates of the contribution of chum salmon from that region were usually greater than 89% (Table 5), although there was the expectation that errors in esti- mation for the region in question would be maximized when the single region comprised 100% of the simulated sample. The level of accuracy observed in the estimated 250 Fishery Bulletin 106(3) Table 4 Mean pairwise population Fst values observed over 14 microsatellite loci for 34 chum salmon (Oncorhynchus keta) populations from eight regions. Regions were the following: 1) Primorye, 2) Amur River, 3) Sakhalin Island, 4) Magadan, 5) Northern Sea of Okhotsk, 6) West Kamchatka, 7) East Kamchatka, and 8) Northeast Russia. Boldface font indicates significant difference (P<0.05). A dash indicates that it was not possible to determine within-region pairwise population Fst values for the Amur River because only a single population was sampled. Region 1 2 3 4 5 6 7 8 1 0.012 2 0.032 — 3 0.031 0.035 0.020 4 0.026 0.021 0.031 0.004 5 0.036 0.036 0.046 0.010 0.000 6 0.028 0.028 0.027 0.009 0.014 0.006 7 0.026 0.024 0.029 0.006 0.011 0.006 0.004 8 0.033 0.033 0.041 0.014 0.009 0.016 0.014 0.000 Table 5 Mean estimated contribution (%) of populations of chum salmon ( Oncorhynchus keta) in Russia for simulated mixtures. Esti- mates are for only a single region (correct regional estimate = 100%) and were calculated from the populations listed for each mixture, with each population comprising 50% (100% for Amur River) of each simulated regional mixture. Estimates were deter- mined from 14 microsatellite loci for eight separate regional stocks. The region designation includes percentages allocated to all populations within a region. Simulations were conducted with a 34-population baseline, 150 fish in the mixture sample, and 100 resamplings in the mixture sample and baseline samples. Standard deviations are shown in parentheses. Mixture Region Regional estimate Population Population estimate 1 Primorye 89.5(2.5) Avakumovka 43.9(4.5) Ryazanovka 45.5(4.5) 2 Amur River 98.2(1.1) Amur 98.2 (1.1) 3 Sakhalin Island 94.8(1.8) Naiba 55.1 (4.8) Tym 39.6(4.7) 4 Magadan 90.0(3.0) Ola 44.3(4.5) Okhota 44.0(4.1) 5 Northern Sea of Okhotsk 86.9(3.1) Oklan 47.8 (4.7) Penzhina 39.1 (4.3) 6 West Kamchatka 95.9 (2.2) Vorovskaya 49.2(5.4) Hairusova 43.0 (5.2) 7 East Kamchatka 86.1 (3.2) Ossora 43.9(4.9) Kamchatka 41.2 (4.7) 8 Northeast Russia 89.8(2.9) Anadyr 46.5 (4.9) Kanchalan 43.3(3.7) stock compositions indicated that accurate regional esti- mates of stock composition should be obtained in samples containing individuals from multiple regions. Testing of accuracy of regional estimates of stock composition in multiregion mixtures was conducted by evaluating four simulated fishery samples. Esti- mated stock composition of a simulated mixture con- taining chum salmon from Primorye, Sakhalin Island, the Amur River, and Magadan was within 3% of the actual population composition and 4% of the regional composition (Table 6, mixture 1). Similar results were observed for a simulated mixture comprising chum salmon from Sakhalin Island, Magadan, the northern Sea of Okhotsk, and west Kamchatka, with population estimates within 3% of actual composition and regional estimates within 3% (Table 6, mixture 2). Estimated stock compositions of a simulated mixture of chum salmon from west Kamchatka, east Kamchatka, and northeast Russia were within 6% of actual population composition and within 5% for regional contributions (Table 6, mixture 3). Regional estimated stock composi- tions of a more complex simulated sample of fish from Beacham et al: Population structure and stock identification of Oncorhynchus keta 251 ,0.002, ioor — ! Narva '2 Ryazanovka 3 Avakumovka Primorye 100 € 6 Naiba 7 Udarnitsa 8 Kalininka 47 100 5 Tym “4 Amur | Amur 9 Tugur ” 1 0 Okhota ‘ 1 1 Magadan Magadan 3 Ola — 12 Tauy 14 Oklan 1 5 Penzhina Sakhalin Northern Sea of Okhotsk 100 33 Anadyr r 34 Kanchalan Northeast 21 Utka I West Kamchatka 4: — 32 Oiutorsky Bay 3 1 Apuka — 23 Plotnikova ~ 22 Bolshaya 16 Hairusova 20 Kikchik 19 Pymta 18 Kol 17 Vorovskaya 24 Zhypanova 27 Nerpichi East Kamchatka West Kamchatka “25 Kamchatka — “ 28 Karaga " 30 Dranka 26 Ivashka 29 Ossora East Kamchatka Figure 2 Neighbour-joining dendrogram of the Cavalli-Sforza and Edwards (1967) chord distance for 34 populations of chum salmon (Oncorhynchus keta) analyzed at 14 microsatellite loci. Bootstrap values at major tree nodes indicate the percentage of 500 trees where populations beyond the node clustered together. Numbers for and locations of populations are indicated in Table 1. multiple regions were within 3% of actual regional contributions (Table 6, mixture 4). Accurate regional estimates of stock composition should be obtained when the current baseline is applied to mixed-stock samples of chum salmon taken from Russian coastal waters, provided that the individuals in the mixture originate entirely from Russian populations. Discussion Population structure The range of populations sampled in the study required a concerted sampling effort, and in some locations col- lection of appropriate samples proved to be difficult. The number of fish sampled from a site or population ranged from 17 to 338 individuals. Estimated popula- tion-level allele frequencies will be subject to relatively larger sampling error at smaller population sample sizes, particularly for loci with large numbers of alleles such as Onelll. Small sample size may have contributed to errors on allele-frequency estimates for some popula- tions. Sampling error may obscure genetic relation- ships among related populations, or conversely genetic relationships among some populations may be falsely inferred. However, the available evidence indicates that variation in population sample sizes did not obscure relationships among related populations. For example, we analyzed chum salmon from three sampling sites in Primorye with sample sizes ranging between 17 and 49 fish per site,. Although these sample sizes were small, 252 Fishery Bulletin 106(3) Table 6 Estimated contribution (%) of individual populations and regional estimates of chum chum salmon ( Oncorhynchus keta) in Russia in simulated mixed-population samples. Each mixture of 150 fish was generated 100 times from baseline allele frequencies, and stock compositions of the mixt ures were estimated with parametric resampling of each of the 34 baseline populations to obtain a new distribution of allele frequencies on each iteration. Actual= 100% accurate. Standard deviations are shown in parentheses. Fourteen microsatellites were used to estimate stock compositions of the simulated mixtures. Populations Actual % Estimated % Populations Actual % Estimated % Mixture 1 Mixture 3 Amur 30 30.5 (4.0) Plotnikova 20 14.0(3.3) Avakumovka 10 8.7 (2.0) Kol 10 6.7 (2.5) Ryazanovka 10 9.1 (2.4) Utka 10 5.5 (1.9) Naiba 20 21.0(3.3) Kamchatka 10 8.2 (2.4) Udarnitsa 10 7.6 (2.2) Ossora 10 10.6(2.8) Tauy 10 7.1 (2.5) Olutorsky Bay 10 6.0 (2.6) Ola 10 8.9 (2.5) Anadyr 20 17.9(3.4) Regions Kanchalan 10 12.3 (3.1) Amur 30 30.5(4.0) Regions Primorye 20 17.9 (3.2) West Kamchatka 40 40.8(4.7) Sakhalin Island 30 28.7 (3.8) East Kamchatka 30 25.4(3.9) Magadan 20 16.5(3.6) Northeast Russia 30 30.2(4.1) West Kamchatka 0 4.0 (1.9) Magadan 0 1.4 (1.2) East Kamchatka 0 1.6 (1.2) Northern Sea of Okhotsk 0 1.1 (0.8) Mixture 2 Mixture 4 Naiba 20 19.0(3.2) Avakumovka 10 8.2 (2.3) Magadan 10 7.8 (2.6) Naiba 10 10.1 (2.3) Ola 10 9.7 (2.8) Tym 10 8.1 (2.5) Okhota 10 9.0 (2.5) Ola 15 13.0(2.8) Oklan 10 7.8 (2.5) Okhota 10 8.7 (2.6) Bolshaya 20 16.5(3.3) Hairusova 15 13.7(3.1) Hairusova 10 10.0(3.0) Utka 10 5.5 (2.2) Kikchik 10 7.5 (2.6) Kamchatka 10 8.1 (2.5) Regions Kanchalan 10 11.6(2.6) Sakhalin Island 20 19.2(3.3) Regions Magadan 30 27.1 (4.0) Primorye 10 8.312.3) Northern Sea of Okhotsk 10 8.0 (2.6) Sakhalin Island 20 18.3(3.1) West Kamchatka 40 42.1 (4.7) Magadan 25 22.7 (4.0) East Kamchatka 0 2.9 (1.6) West Kamchatka 25 26.6(3.8) East Kamchatka 10 10.9(2.9) Northeast Russia 10 12.2(2.7) and thus the estimation of allele frequencies would be subject to sampling error, the clustering of these popula- tions was well supported by our bootstrap calculations (100%). Regional clustering of samples or populations is typically observed in chum salmon (Beacham et al., 1987; Winans et ah, 1994), and thus it is unlikely that close genetic relationships among these populations from Primorye were inferred incorrectly. If populations spawn in remote areas, opportunities to collect samples may be limited. In our study, all sam- ples that were available for a specific sampling site or population were combined in order to estimate genetic differentiation among populations. Annual variation in allele frequencies within a population is typically less than the geographic and population differences ob- served; therefore pooling annual samples over time is a reasonable approach to estimate population-level allele frequencies. Relative annual stability of microsatellite allele frequencies is a general feature of microsatel- lite loci in salmonids (Tessier and Bernatchez, 1999; Beacham et ah, 2006). The population structure of chum salmon in Russia has been investigated previously. For example, Winans et al. (1994) after examining 35 allozyme loci, indicated that there were four groups of Russian chum salmon populations, and that one group generally comprised populations from west Kamchatka, a second group com- prised populations from Magadan, the Sea of Okhotsk, and east Kamchatka, the third group was solely from east Kamchatka, and a fourth group comprised the Beacham et al: Population structure and stock identification of Oncorhynchus keta 253 populations from the Utka River in west Kamchatka and the Anadyr River in northeast Russia. In our study, some similarities were observed between Magadan and the Sea of Okhotsk regional population groups, as well as between the groups from east and west Kamchatka. Winans et al. (1994) did not include populations from Primorye or Sakhalin Island in their survey, but Ginat- ulina ( 1992 ) had previously demonstrated clear differen- tiation of mitochondrial genotypes between the popula- tions from these two regions. The results of our study revealed regional separation of populations between these two areas. Our analysis supports the concept of regional groups of populations, and generally supports the concordance in patterns of population differentiation derived from analysis of allozymes and mitochondrial DNA variation. Allendorf and Seeb (2000) reported a concordance between results from allozyme and mic- rosatellite analyses of population structure for sockeye salmon (O. nerka). Genetic differentiation of Russian chum salmon gen- erally follows a regional structure because proximate populations are generally more similar to each other than to more distant populations. However, there were some cases of populations from one region clustering with populations from another region. One example was the Utka River population from west Kamchatka clustering with populations from northeast Russia. An association between the Utka River population and the Anadyr River population was also reported by Winans et al. (1994) in an analysis of allozyme variation. Be- cause Winans et al. (1994) and authors of the present study analyzed the same sample from the Utka River population, concurrence between the allozyme and mi- crosatellite analyses was not unexpected. However, the number of fish sampled for the Utka River population was the fewest for any of the west Kamchatka popula- tions (Table 1), and it may be that an increase in sample size for this population would result in estimated allele frequencies that would be more similar to those of other populations in west Kamchatka. Additionally, the Tugur River population was most closely associated with the population from the Amur River. In recent geologic his- tory, the Tugur River may have been once part of the Amur River drainage, but now flows into Tugur Bay on the Sea of Okhotsk. A common origin between the Amur River and Tugur River populations may account for the current association between the two populations. Distinctive groups of populations surveyed were those from the Primorye, Sahkalin Island, the northern Sea of Okhotsk, and northeast Russia, and a strong regional clustering of these populations was observed in the dendrogram analysis. Most of these population groups were characterized by slightly lower genetic variation compared with other populations surveyed in Russia. For other salmonids, populations from regions with re- duced genetic variation have formed distinctive clusters in dendrogram analysis (Beacham et al., 2006). Russian chum salmon populations displayed on aver- age less genetic differentiation than did populations from western Alaska and adjacent areas. Despite the Russian populations being surveyed from a larger geo- graphic area than were populations in western Alaska (Beacham et al., in press), and thus there was greater likelihood of differentiation due to isolation by distance, comparisons of locus-specific Fst values between the two groups indicated that Russian populations were less differentiated (lower values in 11 of 14 loci, P= 0.057). This result indicates that there may be more straying among Russian populations than among those in west- ern Alaska, possibly as a result of adaptation to harsh enviromental conditions or hatchery development and broodstock transfer in Russia. Alternatively, less dif- ferentiation would also be observed if Russian chum salmon had colonized available habitats more recently than had chum salmon in western Alaska. Stock identification Accurate, economical, and practical methods of stock identification are required to determine the migra- tion pathways of juvenile and maturing salmon, and to manage fisheries that may intercept salmon during their migration to natal spawning grounds. Effective stock identification techniques are based on characters that display stable differentiation among groups to be discriminated, and these characters must be examined easily in a rapid and cost-effective manner. Allozymes provided the characters for initial genetically based population surveys and stock identification of Russian chum salmon (Winans et al., 1994, 1998). Later, single nucleotide polymorphisms (SNPs) were used to estimate the genetic structure of the population (Sato et al., 2001); therefore estimation of stock composition in mixed-stock samples can proceed (Sato et al., 2004). In an analysis of 30 haplotypes from mtDNA, Sato et al. (2004) were able to indicate some level of regional structure in popula- tions in Russia, but the exact nature of the geographic structure was uncertain. In our analysis of microsatel- lite variation, clear differentiation was observed among regional groups of populations, and populations from Primorye were the most distinctive. Accuracy of estimated stock compositions is directly influenced by the baseline used in the estimation pro- cedure, and the level of genetic differentiation among regional groups of populations is a key component. How- ever, sample sizes of populations in the baseline are also an important part because the accuracy of estimation is related to the population sample size (Beacham et al., 2006). Fewer than 60 fish were sampled for many of the populations sampled in our survey, and increasing sam- ple sizes to approximately 150 fish per population would likely lead to all regional estimates of stock composition being in excess of 90% accurate in all simulated single- region mixture samples. Surveys of genetic variation of salmon populations al- low stock identification in mixed-stock fisheries, where the origins of fish contributing to mixed-stock fisheries are determined by comparing the genetic characteristics of fish in the fishery samples to the genetic character- istics of fish from potentially contributing populations. 254 Fishery Bulletin 106(3) Analysis of simulated mixed-stock samples of known origin is an initial practical method to evaluate the potential for applying genetic variation to mixed-stock fishery analysis. Our analysis of simulated mixtures indicated that microsatellite variation provides accurate estimates of regional contributions of chum salmon stocks from Russia, and in some cases provides reli- able estimates of individual populations in simulated mixtures. Microsatellites have previously been reported to provide reliable estimates of stock composition in mixed-stock chum salmon samples of largely North American origin (Beacham et ah, in press), and our results from simulated mixtures indicated that mic- rosatellites should provide reliable estimates of stock composition for chum salmon in coastal waters in Rus- sia. However, if Japanese or Korean chum salmon or po- tentially North American chum salmon are intercepted in coastal or nearshore fisheries in Russia, then clearly a larger baseline than the one examined in the current study would be required to provide reliable estimates of stock composition under these circumstances. The application of microsatellites for the determina- tion of population structure of Russian chum salmon will allow significant regional differentiation among these populations to be employed in estimating region- al contributions to mixed-stock fishery samples from coastal waters. Microsatellites provide similar results for other Pacific salmon species (Beacham et al., 2005, 2006) and are likely to be effective in identifying the origin of Russian chum salmon in mixed-stock fisheries in nearshore and offshore waters. The present analysis of microsatellite variation of Russian chum salmon provides evidence of a more fine- scale population structure than those that have previ- ously been demonstrated with other genetic-based mark- ers such as allozymes (Winans et al., 1994; Efremov, 2001) or mitochondrial based SNPs (Sato et al., 2004). This more fine-scale resolution of population structure was likely due to the larger number of alleles associ- ated with the microsatellite loci than with either the allozyme or SNP loci. Because genetic-based markers generally exhibit annual stability in allele frequencies, they are generally more effective for stock identification applications than are techniques that rely on envi- ronmentally induced variation to discriminate among stocks, such as scale-pattern analysis of trace elements in otoliths. Once the baseline has been established for genetic applications, annual surveys of contributing stocks are not necessary, as is the case with for environ- mentally induced variation. Should greater resolution in stock composition estimates be required than that pro- vided by the 14 microsatellites surveyed in the present study, the addition of markers specifically designed to provide the required resolution will be necessary. These markers could either be additional microsatellites, or perhaps single nucleotide polymorphisms (SNPs) (Smith et al., 2005). It is likely that a combination of microsat- ellites and SNPs can be employed to provide accurate population or regional estimates of stock composition of mixed-stock samples. Acknowledgments A significant effort was undertaken to collect samples from chum salmon populations analyzed in the study. Samples of populations from Primorye and Sakhalin Island, as well as some Magadan samples, were intially provided by V. V. Efremov to the United States National Marine Fisheries Service (NMFS) Auke Bay Laboratory, where R. Wilmot provided access to the Molecular Genet- ics Laboratory (MGL). G. Winans of the NMFS Mont- lake laboratory also provided access to some population samples. C. Wallace and J. Candy of the MGL assisted in the analysis. Funding was provided by Fisheries and Oceans Canada. Literature cited Allendorf, F. W., and L. W. Seeb. 2000. Concordance of genetic divergence among sockeye salmon populations at allozyme, nuclear DNA, and mito- chondrial DNA markers. Evolution 54:640-651. Banks, M. A., M. S. Blouin, B. A. Baldwin, V. K. Rashbrook, H. A. Fitzgerald, S. M. Blankenship, and D. Hedgecock. 1999. Isolation and inheritance of novel microsatellites in chinook salmon (Oncorhynchus tshawytscha). J. Hered. 90:281-288. Banks, M. A., V. K. Rashbrook, M. J. Calavetta, C. A. Dean, and D. Hedgecock. 2000. Analysis of microsatellite DNA resolves genetic structure and diversity of chinook salmon (Oncorhyn- chus tshawytscha) in California’s Central Valley. Can. J. Fish. Aquat. Sci. 57:915-927. Beacham, T. D. 1996. The use of minisatellite DNA variation for stock identification of chum salmon, Oncorhynchus keta. Fish. Bull. 94:611-627. Beacham, T. D., J. R. Candy, K. L. Jonsen, J. Supernault, M. Wetklo, L. Deng, K. M. Miller, R. E. Withler, and N. V. Varnavskaya. 2006. Estimation of stock composition and individual identification of Chinook salmon across the Pacific Rim by use of microsatellite variation. Trans. Am. Fish. Soc. 135:861-888. Beacham, T. D., J. R. Candy, B. McIntosh, C. MacConnachie, A. Tabata, K. Kaukinen, L. Deng, K. M. Miller, R. E. Withler, and N. V. Varnavskaya. 2005. Estimation of stock composition and individual iden- tification of sockeye salmon on a Pacific Rim basis using microsatellite and major histocompatibility complex variation. Trans. Am. Fish. Soc. 134:1124-1146. Beacham, T. D., J. R. Candy, K. J. Supernault, M. Wetklo, B. Deagle, K. Labaree, J. R. Irvine, K. M. Miller, R. J. Nelson, and R. E. Withler. 2003. Evaluation and application of microsatellites for population identification of Fraser River chinook salmon (Oncorhynchus tshawytscha). Fish. Bull. 101:243-259. Beacham, T. D., A. P. Gould, R. E. Withler, C. B. Murray, and L. W. Barner. 1987. Biochemical genetic survey and stock identifica- tion of chum salmon (Oncorhynchus keta) in British Columbia. Can. J. Fish. Aquat. Sci. 44:1702-1713. Beacham et ai: Population structure and stock identification of Oncorhynchus keta 255 Beacham, T. D., M. Lapointe, J. R. Candy, B. McIntosh, C. Mac- Connachie, A.Tabata, K. Kaukinen, L. Deng, K. M. Miller, and R. E. Withler. 2004. Stock identification of Fraser River sockeye sal- mon ( Oncorhynchus nerka) using microsatellites and major histocompatibility complex variation. Trans. Am. Fish. Soc. 133:1106-1126. Beacham, T. D., K. D. Le, M. Wetklo, B. McIntosh, T. Ming, and K. M. Miller. In press. Population structure and stock identification of chum salmon from western Alaska determined with microsatellite and major histocompatibility complex variation. In Sustainability of the Arctic-Yukon-Kus- kokwim salmon fisheries: what do we know about salmon ecology, management, and fisheries? (C. C. Krueger, and C. E. Zimmerman, eds.). Am. Fish. Soc., Symp. Bethesda, MD. Brykov, V. A., N. E. Poliakova, and A. V. Prokhorova. 2003. Phylogenic and geographic analysis of chum salmon Oncorhynchus keta (Walbaum) in Asian populations based on mitochondrial DNA variation. Russ. J. Genet. 39:61-67. Buchholz, W. G., S. J. Miller, and W. J. Spearman. 2001. Isolation and characterization of chum salmon microsatellite loci and use across species. Anim. Genet. 32:160-167. Cairney, M., J. B. Taggart, and B. Hoyheim. 2000. Characterization of microsatellite and minisatel- lite loci in Atlantic salmon ( Salmo salar L.) and cross- species amplification in other salmonids. Mol. Ecol. 9:2175-2178. Cavalli-Sforza, L. L., and A. W. F. Edwards. 1967. Phylogenetic analysis: models and estimation procedures. Am. J. Hum. Genet. 19:233-257. Chen, J.-P, D.-J. Sun, C.-Z. Dong, B. Liang, W.-H. Wu, and S. -Y. Zhang. 2005. Genetic analysis of four wild chum salmon Oncorhyn- chus keta populations in China based on microsatellite markers. Environ. Biol. Fishes 73:181-188. Debevec, E. M., R. B. Gates, M. Masuda, J. Pella, J. M. Reynolds, and L. W. Seeb. 2000. SPAM (version 3.2): Statistics program for analyz- ing mixtures. J. Hered. 91:509-510. Efremov, V. V. 2001. Genetic variation and differentiation of popula- tions of chum salmon Oncorhynchus keta (Walbaum) from the southern Russian Far East. Russ. J. Genet. 37:283-289. Ginatulina, L. K. 1992. Genetic differentiation among chum salmon, Oncorhynchus keta (Walbaum), from Primorye and Sakhalin. J. Fish Biol. 40:33-38. Goudet, J. 1995. FSTAT, a program for IBM = PC compatibles to calculate Weir and Cockerham’s (1984) estimators of F-statistics (version 1.2). J. Hered. 86:485-486. Ishida, Y., S. Ito, and K. Takagi. 1989. Stock identification of chum salmon Oncorhynchus keta from their maturity and scale characters. Bull. Jap. Soc. Sci. Fish. 55:651-656. Nelson, R. J., and T. D. Beacham. 1999. Isolation and cross species amplification of micro- satellite loci useful for study of Pacific salmon. Anim. Genet. 30:228-229. Nikolayeva, Y., and N. I. Semenets. 1983. A contribution to stock differentiation of chum salmon, Oncorhynchus keta ( Salmonidae), by scale structure in the first year of growth. J. Ichthyol. 23: 18-28. Olsen, J. B., S. L. Wilson, E. J. Kretschmer, K. C. Jones, and J. E. Seeb. 2000. Characterization of 14 tetranucleotide microsat- ellite loci derived from Atlantic salmon. Mol. Ecol. 9:2155-2234. Perriere, G., and M. Gouy. 1996. WWW-Query: an on-line retrieval system for bio- logical sequence banks. Biochimie 78:364-369. Polyakova, N. E., A. V. Semina, and V. A. Brykov. 2006. The variability in chum salmon Oncorhynchus keta (Walbaum) mitochondrial DNA and its connec- tion with the paleogeological events in the northwest Pacific. Russ. J. Genet. 42:1164-1171. Rannala, B., and J. L. Mountain. 1997. Detecting immigration by using multilocus geno- types. Proc. Natl. Acad. Sci. USA 94: 9197-9201. Rexroad, C. E., R. L. Coleman, A. M. Martin, W. K. Hershberger, and J. Killefer. 2001. Thirty-five polymorphic microsatellite markers for rainbow trout ( Oncorhynchus mykiss). Anim. Genet. 32:283-319 Rice, W. R. 1989. Analyzing tables of statistical tests. Evolution 43:223-225. Salmenkova, E. A., V. T. Omelchenko, D. V. Politov, K. I. Afa- nasyev, and G. A. Rubtsova. 2007. Genetic diversity of northern Okhotsk Sea coast populations of chum salmon with natural and artificial reproduction. Biol. Morya 33:299-308. Sano, S. 1966. Salmon of the North Pacific Ocean — Part III. A Review of the Life History of North Pacific Salmon. 3. Chum salmon in the Far East. Int. N. Pac. Fish. Comm. Bull. 18:41-57. Sato, S., J. Ando, H. Ando, S. Urawa, A. Urano, and S. Abe. 2001. Genetic variation among Japanese populations of chum salmon inferred from the nucleotide sequences of the mitochondrial DNA control region. Zool. Sci. (Tokyo) 18:99-106. Sato, S., H. Kojima, J. Ando, H. Ando, R. L. Wilmot, L. W. Seeb, V. Efremov, L. LeClair, W. Buchholz, D.-H. Jin, S. Urawa, M. Kaeriyama, A. Urano, and S. Abe. 2004. Genetic population structure of chum salmon in the Pacific Rim inferred from mitochondrial DNA sequence variation. Environ. Biol. Fish. 69:37-50. Small, M. P, T. D. Beacham, R. E. Withler, and R. J. Nelson. 1998. Discriminating coho salmon ( Oncorhynchus kisutch ) populations within the Fraser River, British Colum- bia using microsatellite DNA markers. Mol. Ecol. 7:141-155. Smith, C. T., B. F. Koop, and R. J. Nelson. 1998. Isolation and characterization of coho salmon (Oncorhynchus kisutch) microsatellites and their use in other salmonids. Mol. Ecol. 7:1613-1621. Smith, C. T„ W. D. Ternplm, J. E. Seeb, and L. W. Seeb. 2005. Single nucleotide polymorphisms provide rapid and accurate estimates of the proportions of U.S. and Cana- dian chinook salmon caught in Yukon River fisheries. N. Am. J. Fish. Manag. 25:944-953. Sohn, D., S. Kang, and S. Kim. 2005. Stock identification of chum salmon ( Oncorhynchus keta) using trace elements in otoliths. J. Oceanogr. 61: 305-312. 256 Fishery Bulletin 106(3) Spies, I. B., D. J. Brasier, P. T. L. O’Reilly, T. R. Seamons, and P. Bentzen. 2005. Development and characterization of novel tetra-, tri-, and dinucleotide microsatellite markers in rain- bow trout ( Oncorhynchus mykiss). Mol. Ecol. Notes 5:278-281. Tanaka, S., M. P. Shepard, and H. T. Bilton. 1969. Origin of chum salmon ( Oncorhynchus keta ) in offshore waters of the North Pacific in 1956-1958 as determined from scale studies. Int. N. Pac. Fish. Comm. Bull. 26:57-155. Taylor, E. B., T. D. Beacham, and M. Kaeriyama. 1994. Population structure and identification of North Pacific Ocean chum salmon ( Oncorhynchus keta) revealed by an analysis of minisatellite DNA variation. Can. J. Fish. Aquat. Sci. 51:1430-1442. Tessier, N., and L. Bernatchez. 1999. Stability of population structure and genetic diver- sity across generations assessed by microsatellites among sympatric populations of landlocked Atlantic salmon ( Salmo salar L.). Mol. Ecol. 8:169-179. Volobuyev, V. V., A. Yu. Rogatnykh, and K. V. Kuzishchin. 1990. Intraspecific forms of chum salmon, Oncorhyn- chus keta , along the continental coast of the Sea of Okhotsk. J. Ichthyol. 30:104-114. Waples, R. S. 1990. Temporal changes of allele frequency in Pacific salmon populations: implications for mixed-stock fishery analysis. Can. J. Fish. Aquat. Sci. 47:968-976. Williamson, K. S., J. F. Cordes, and B. P. May. 2002. Characterization of microsatellite loci in chinook salmon (Oncorhynchus tshawytscha) and cross-species amplification in other salmonids. Mol. Ecol. Notes 2:17-19. Wilmot, R. L., R. J. Everett, W. J. Spearman, R. Baccus, N. V. Varnavskaya, and S. V. Putivkin. 1994. Genetic stock structure of Western Alaska chum salmon and a comparison with Russian Far East stocks. Can. J. Fish. Aquat. Sci. 51 (suppl. 1): 84- 94. Wilmot, R. L., C. M. Kondzela, C. M. Guthrie, and M. M. Masuda. 1998. Genetic stock identification of chum salmon har- vested incidentally in the 1994 and 1995 Bering Sea trawl fishery. N. Pac. Anad. Fish Comm. Bull. 1: 285-299. Winans, G. A., P. B. Aebersold, S. Urawa, and N. V. Varnavskaya. 1994. Determining continent of origin of chum salmon (Oncorhynchus keta ) using genetic identification tech- niques: status of allozyme baseline in Asia. Can. J. Fish. Aquat. Sci. 51(suppl. 1):95-113. Winans, G. A., P. B. Aebersold, Y. Ishida, and S. Urawa. 1998. Genetic stock identification of chum salmon in highseas test fisheries in the western North Pacific Ocean and Bering Sea. N. Pac. Anad. Fish Comm. Bull. 1:220-226. 257 Abstract — For many fish stocks, resource management cannot be based on stock assessment because data are insufficient — a situation that requires alternative approaches to manage- ment. One possible approach is to manage data-Iimited stocks as part of an assemblage and to determine the status of the entire unit by a data-rich indicator species. The utility of this approach was evaluated in analyses of 15 years of commercial and 34 years of recreational logbook data from reef fisheries off the southeastern United States coast. Multivariate statisti- cal analyses successfully revealed three primary assemblages. Within assemblages, however, there was little evidence of synchrony in popu- lation dynamics of member species, and thus, no support for the use of indicator species. Nonetheless, assem- blages could prove useful as manage- ment units. Their identification offers opportunities for implementing man- agement to address such ecological considerations as bycatch and species interrelations. Manuscript submitted 6 August 2007. Manuscript accepted 26 February 2008. Fish. Bull. 106:257-269 (2008). The views and opinions expressed or implied in this article are those of the author and do not necessarily reflect the position of the National Marine Fisheries Service, NOAA. Fish assemblages and indicator species: reef fishes off the southeastern United States Kyle W. Shertzer (contact author) Erik H. Williams Email address for K. W. Shertzer: Kyle.Shertzer@noaa.gov National Oceanic and Atmospheric Administration Southeast Fisheries Science Center Center for Coastal Fisheries and Habitat Research 101 Pivers Island Road Beaufort, North Carolina 28516 Most approaches to fishery man- agement rely on results from stock assessment. Data-limited situations, however, may not conform to conven- tional assessment methods, necessitat- ing other approaches to management (Kruse et al., 2005). One possible approach with data-limited stocks is to assign them to assemblages that are managed as units. Ideally, each assemblage would include at least one data-rich species that could be assessed and serve as a status indi- cator of the entire unit. Managing assemblages by means of indicator species is arguably a small but prac- tical step in the direction of ecosys- tem-based management (Hall and Mainprize, 2004). Assemblages may be defined by similarities in such biological charac- teristics as life history, trophic behav- ior, or home range. For the purpose of fishery management, however, an assemblage should consist of spe- cies caught together, if regulations on fishing are to benefit assemblage members. This is particularly true if regulations are focused on an indica- tor species but the intent is also to control the harvest of other species in the assemblage. Indicator species have been used in management of both terrestrial and marine systems (Simberloff, 1998; Zacharias and Roff, 2001). The term “indicator species” has no single definition (Landres et al., 1988); it is used here as suggested by the Na- tional Standard Guidelines of U.S. federal fishery management, which states that where maximum sustain- able yield (MSY) cannot be specified for each stock of a mixed-stock fish- ery, then “MSY may be specified on the basis of one or more species as an indicator for the mixed stock as a whole or for the fishery as a whole.” (Federal Register, 1998) According to this usage, the stock status of the indicator is extrapolated to represent that of other species in the assem- blage, or analogously, other stocks of the same species. Such an approach requires the assumption that popu- lation trends of an indicator species reflect those of others in the assem- blage. The approach of managing multi- species assemblages by means of in- dicator species raises two fundamen- tal questions. First, can assemblages be identified? As mentioned, species of an assemblage would need to be caught together if regulations are to affect the entire unit. Second, if an assemblage can be identified, do its members have similar stock dynam- ics? If not, focusing management on the indicator species may not pro- vide the intended benefits to other stocks. We address both questions, using as a case study the snapper-grouper complex off the southeastern United States. As defined for management, the complex contains 73 species (Ap- pendix), the majority of which cannot be assessed with currently available data. The objectives of this study are 1) to identify assemblages of finfish species within the snapper-grouper complex and 2) to examine synchrony of stock dynamics within assemblag- es. To accomplish the first objective, multivariate statistical techniques 258 Fishery Bulletin 106(3) were applied to data from recreational and commercial fisheries, and to accomplish the second, indices of abun- dance were computed and tested for correlation. Materials and methods Data used in multivariate analyses To identify assemblages of species within fishery land- ings, statistical grouping techniques were applied to two fishery data sets, one recreational (headboat) and one commercial. Both data sets encompassed areas from Cape Hatteras, North Carolina to Key West, Florida. These data were chosen because of their importance for stock assessment of species in the snapper-grouper complex. The recreational sector was represented by logbook data reported by headboat operators and verified by port samplers. Headboats are large, for-hire vessels that typically accommodate 20-60 anglers on half- or full-day trips. Data collection began in 1972 with a focus on coastal waters off North and South Carolina. The area of collection was extended in 1976 to include the coastal waters of Georgia and northern Florida, and again in 1978 to include those of southern Florida. We used 1972-2005 headboat data. Records from each trip contained information on number of anglers, trip duration, date, geographic area, and landings (number fish) of each species. The commercial sector was represented by logbook data reported by commercial anglers with snapper- grouper permits. We used 1992-2006 commercial data; however, 2006 was a partial year (data through Sep- tember). Records contained information similar to those in the headboat data set, but landings were reported in weight (pounds). Excluded were nonsensical records suspected to be misreported or misrecorded. Analyses of commercial data were restricted to trips with han- dline gear (~87% of records) to avoid the possibility of confounding estimated assemblages with effects of gear. Furthermore, these analyses included only trips of one-day duration (~50% of records) to minimize the possibility that catch in a trip was taken from widely separated geographic areas with potentially different assemblages. Species assemblages Following Lee and Sampson (2000), we used more than one statistical technique to identify species assemblages. We applied three techniques: ordination and two types of cluster analysis. For all three techniques, the Sprenson (also called Bray-Curtis) measure of dissimilarity (dis- tance) between species was used (McCune and Grace, 2002). In comparison to other measures, Sprenson dis- tance has been found more robust in ecological studies (Field et ah, 1982; Faith et ah, 1987) and provides more ecologically interpretable results (Beals, 1973). Perhaps for these reasons, it has been considered appropriate in studies of fish assemblages (e.g., Mueter and Norcross, 2000; Gomes et ah, 2001; Williams and Ralston, 2002). To compute dissimilarities, we formatted each data set as a matrix, with rows representing species and columns representing vessel-months. That is, each ele- ment (Cy) of the matrix quantified the amount (in units of number fish for headboat or pounds for commercial) of a species (i) landed by each vessel pooled over one month (vessel-month j). The duration of a month was chosen as a reasonable compromise between maximiz- ing the variety of species landed (longer duration) and minimizing the number of different locations fished (shorter duration). Locations fished per vessel were generally consistent within a month, but could have changed on the time scale of seasons (perhaps follow- ing fish migrations, for example). Species were removed if they appeared in fewer than 1% of all trips because rare species may distort inferred patterns (Koch, 1987; Mueter and Norcross, 2000). This restriction left 25,293 records of vessel-month-species in the headboat data set and 143,426 in the commercial data set. Before computing dissimilarities, data were trans- formed with the root-root transformation to moderate the influence of abundant species: This transformation has been preferred for density and biomass data, particularly when used in connection with the Sprenson measure of distance (Field et al., 1982). After transformation, a matrix of dissimilari- ties between species was computed with the Sprenson measure of distance: \Cij C'hj\ _ (Cij+Chj) (2) where Dlh - the distance between species i and h\ and J = the number of columns (vessel-months). To identify species assemblages, the ordination method of nonmetric multidimensional scaling (NMDS) was ap- plied to the matrix of dissimilarities (Kruskal, 1964). As stated by McCune and Grace (2002), “Nonmetric multi- dimensional scaling is the most generally effective ordi- nation method for ecological community data and should be the method of choice, unless a specific analytical goal demands another method.” NMDS searches for positions of n objects (here, n species) in d dimensions such that dissimilarities in ordination space are close to those of the original space. We extracted the first two dimensions of ordination space (d= 2) for graphical presentation. In addition to ordination, we applied nonhierarchical and agglomerative hierarchical cluster analyses. The nonhierarchical cluster analysis was used to partition species into groups, based on the method of £-medoids, a more robust version of the classical method of Lmeans (Kaufman and Rousseeuw, 1990). The /e-medoids ap- proach attempts to identify k objects from the data set Shertzer and Williams: Reef fishes off the the southern United States 259 that best represent all objects. Clusters are created by assigning each object in the data set to its nearest representative (i.e., medoid). As with any nonhierarchical method, the number of clusters k must be specified a priori. We applied a range of values and selected the k most concordant with the data, as quantified by highest average silhouette width. The silhouette width of each species measures its goodness of clustering. For any given k, silhouette widths averaged over species within clusters indicate relative strength of assemblages; across k, the highest average width computed from all species corresponds to the optimal number of clusters (Rousseeuw, 1987). To examine uncertainty in the optimal number, a bootstrap procedure was applied in which columns (vessel-months) of the original data matrices were resampled with re- placement to produce /i = 1000 bootstrapped matrices of the original dimension, and then n = 1000 average silhouette widths were recomputed for each k. The hierarchical cluster analysis was included to provide a comparison with clusters computed by £-me- doids and to quantify associations among species, as represented by dendrograms. The hierarchical analysis was based on the linkage method of McQuitty (McCune and Grace, 2002). Indices of abundance Indices of abundance were computed to examine synchrony of dynamics among stocks and thus, to investigate the basic assumption that an indicator spe- cies could be used to infer dynamics of other species in the assemblage. This investigation focused on the three strongest assemblages (i.e., strongest coherence among members), as measured by average silhouette widths from the cluster analysis. Because the strongest assemblages were examined, this investigation is a best-case scenario. If strongly associated populations do not exhibit synchronous dynamics, one should not assume that weakly associated populations do otherwise. Ideally, indices of abundance should be computed from fishery independent data; however, for many spe- cies here, such data were unavailable or insufficient. In this study, indices were computed from the headboat data set. Fishing effort from headboats is applied gener- ally toward many species, rather than toward specific targets. Because effort is nondirected, any confounding effects of density-dependent catchability are likely to be minimized, and in this regard, headboat data are similar to fishery-independent data. Indices of abundance were computed from catch and effort data in units of number of fish landed per angler- hour. Data were considered from 1978, the first year of full area coverage, to 2005. For each species, a trip was included only if a species from the relevant assemblage was landed. Thus, many trips were excluded, and some trips were included that had effort but zero catch. This approach represents effective effort more accurately than if all trips were included (a situation that would inflate the assumed effort) or if trips were restricted to those that landed the species in question (a situation that would deflate the assumed effort). To compute indices of abundance, catch and effort da- ta were standardized using a generalized linear model (Hardin and Hilbe, 2001). The explanatory variables for the model were year, month, and geographic area. To ensure adequate sample sizes by geographic area, sam- pling areas were aggregated into four regions: North Carolina, South Carolina, Georgia-northern Florida, and southern Florida (south of Cape Canaveral). The response variable was catch per effort, assumed to be distributed with delta-lognormal error structure (Lo et ah, 1992; Stefansson, 1996; Maunder and Punt, 2004). In this structure, the proportion of positive values is modeled with binomial error, and positive values them- selves are modeled with lognormal error. Indices were not computed for species that were caught in fewer than 20% of trips on the relevant assemblage, to avoid estimation error associated with inflation of zero values (Lampert, 1992). Because this criterion excludes rarely caught species, evidence of synchrony in our results should be viewed as a necessary but not sufficient con- dition for the use of indicator species. Synchrony in dynamics between any two stocks was measured by the Spearman’s rank correlation coeffi- cient, computed both from 1) the indices of abundance and 2) the first-differenced time series of log-abun- dances {zt)\ zt = \ogUt-logUt_1 = \og^J-> (3) U t- 1 where Ut = the index value of a stock at time t. Positive correlation of the indices themselves would indicate similar trends in abundance over time. The use of first differences, as in Equation 3, rather than raw or relative abundance, puts emphasis on annual population growth rates and may reduce spurious correlation (Bjprnstad et al., 1999). Positive correlation of growth rates would indicate that stocks not only have similar patterns of productivity (growth, recruitment, and mortality), but that they also respond similarly to interannual variation in fishing effort or catchability. Significance levels of correlation coefficients were obtained nonparametrically with n = 10,000 randomiza- tions of zt (Prager and Hoenig, 1989; Edgington, 1995; Bjprnstad et al., 1999). A coefficient that ranks suffi- ciently high in relation to the randomizations could be considered significantly positive, and a coefficient that ranks low, significantly negative. Significance was de- termined with a two-tailed test at the a=0.1 level with Bonferroni correction. Results Species assemblages Multidimensional scaling did not reveal strongly isolated groups of species in ordination space (Fig. 1). It did, however, reveal consistency of ordination in the sense 260 Fishery Bulletin 106(3) that many species had similar neighbors across headboat mutton snapper, and yellowtail snapper (see Appendix and commercial data sets. For example, in both data for scientific names). This repeatability of results pro- sets, lane snapper was near blue runner, gray snapper, vides evidence of species assemblages. The £-medoid cluster analyses identified £ = 14 clusters as most compatible with the headboat data and £ = 7 clusters as most com- patible with the commercial data (Fig. 2). These optimal numbers of clusters were not cleanly defined because peaks in average silhouette widths lacked distinction (Fig. 2). In general, assemblages were similar across data sets, at least for species that were pres- ent in both data sets (Table 1). Hierarchical cluster analyses provided as- sociations among species that were consistent with the assemblages of £-medoid analyses. In hierarchical analysis of the headboat data (Fig. 3), three assemblages had the strongest similarities among member species, labeled here as the deepwater assemblage (blueline tilefish, snowy grouper, speckled hind, and yellowedge grouper), southern assemblage (blue runner, gray snapper, lane snapper, mutton snapper, and yellowtail snapper), and northern assemblage (bank sea bass, black sea bass, knobbed porgy, gag, gray triggerfish, greater amberjack, red porgy, red snapper, scamp, tomtate, vermilion snap- per, white grunt, and whitebone porgy). In hierarchical analysis of the commercial data, the same three assemblages were identified with few differences in constituent species (Fig. 4). In both data sets, these assemblages had the strongest coherence among member species, as measured by each cluster’s av- erage silhouette width (Table 1). Thus, the deepwater, southern, and northern assem- blages were examined further for synchrony in indices of abundance. indices of abundance Although data through 2005 were considered, indices of the deepwater assemblage were derived through 1993, because 1994 began regulations that would have invalidated catch per effort as an index of abundance (i.e., one speckled hind per vessel per trip). Deepwater species that met the criterion of at least 20% positive trips were speckled hind, snowy grouper, and blueline tilefish. All southern species met the 20% positive trip criterion, however small sample sizes of these species north of Cape Canaveral, Florida, necessitated combining geographic areas into two regions: southern Florida and all other areas. Northern species that met the 20% criterion were white grunt, gag, tomtate, black sea bass, vermilion snapper, and gray triggerfish. Cotwck YtlSnp BstGnt FrnGnt Schmst o o Prkfsh GrySnp LanSnp BluRun ° °o MtnSnp WhtGnt ... q Dl. n VrmSnp ° oJltP9V BI£GpAroTrf Margat ° 0OceTrf ScyPgy BlkMrg oTomtat RedPgy GryTrf o Gag RedGpr WtbPgy o BndRud o oBankSB RedSnp 5 Scamp oSIkSnp Grvsby“ °SndTlf qjKnbPgy AlnvJck o RckHnd _Hogfsh RedHnd oGrAjck QenTrf o BfnSnp Coney DogSnp NssGpr YlfGpr Sc up AtlSpf SpkHnd o SnwGpr o WrsGpr LgsPgy YlmGpr CbrSnp YdgGpr B o YtlSnp CrvJck GrySnp ° o o LanSnp BluRun MtnSnp FrnGnt o BstGnt WtbPgy KnbPgy GrAjck BlkGpr o RedGpr o Gag o o GryTrf VrmSnp Hogfsh Margat JitPgy SnwGpr o ° BluTIf RedSnp BIckSB o ° o RedPgy o Scamp LsAjck SlkSnp YdgGpr Dimension one Figure 1 Nonmetric multidimensional scaling of species from the (A) head- boat and (B) commercial sectors. Distances between points are approximately proportional to the dissimilarities between species. Abbreviations are explained in the Appendix. Shertzer and Williams: Reef fishes off the the southern United States 261 Figure 2 Average silhouette width (lines with circles) from /z-medoid cluster analysis of species in (A) the headboat sector and (B) the commercial sector. Lower and upper lines (without circles) represent 5th and 95th percentiles, respectively, from n = 1000 bootstrap replicates. Average silhouette width mea- sures goodness of clustering; higher values indicate better concordance with data. In general, indices of abundance were not syn- chronous (Table 2). Within the deepwater as- semblage, snowy grouper was positively but not significantly correlated with blueline tilefish, and neither species was strongly correlated with speckled hind. Within the southern assemblage, correlations between species were mostly negative; however, those between yellowtail, lane, and gray snappers were positive and significant, indicating synchrony in this subset. Within the northern as- semblage, about half of the correlations between species were negative, and only the correlation between vermilion snapper and black sea bass was positive and significant. These results offer little evidence of synchrony in population trends within assemblages. Similarly, first-differenced indices of abundanc- es were out of synchrony (Table 3). Correlations between species were both positive and nega- tive; only one was significantly negative (between snowy grouper and speckled hind), and one was significantly positive (between gray triggerfish and vermilion snapper). These results from first- differenced time series do not support the hypoth- esis of synchrony in annual population growth rates within assemblages. Discussion It is unlikely that sufficient resources will ever be available to monitor, assess, and manage every fish stock individually. Thus, managing assemblages by means of indicator species has intuitive appeal. It begins a shift from single-species management toward ecosystem-based approaches and provides a scientific and managerial shortcut by supplanting the need to monitor and assess every managed stock. In the United States, for example, the Magnuson-Stevens Fishery Conservation and Management Reauthorization Act of 2006 (MSRA, 2006) requires that annual catch limits be established to end and prevent overfishing by 2011 in all fisheries (by 2010 for fisheries where overfish- ing is occurring), yet many stocks cannot be assessed and their status is therefore unknown. Conceivably, setting catch limits by assemblage rather than stock- by-stock could satisfy the statute 1) without substantial new resources devoted to both data collection programs and stock assessment and 2) within the time frame allowed. Despite its possible appeal, the use of indicator spe- cies to extrapolate trends of other species should be viewed with considerable skepticism. From the per- spective of niche theory, fishes that coexist are able to do so, in part, because they have adapted to use different niches in their shared environment (May and MacArthur, 1972; Leibold, 1995). Consequently, species within assemblages differ in reproductive character- istics, foraging behavior, habitat requirements, and population-level responses to such factors as competi- tion, predation, disease, and environmental variation (Landres et al., 1988). Because of these differences, population trends of one species (or stock) do not readily extrapolate to others in the assemblage (e.g., Niemi et ah, 1997; Shaul et al., 2007). For empirical and theo- retical reasons, several authors have concluded that the use of indicator species should be avoided, unless sup- ported by strong evidence from the system in question (Landres et al., 1988; Niemi et al., 1997). From another perspective, even without strong evi- dence of synchrony, indicator species may still be use- ful if applied in a restrictive sense. That is, if fishing effort occurs at the level of assemblages, regulations to reduce effort on one species (the indicator) could transmit to others of unknown status. The cost of this approach would be the forgone yield of any species that could sustain increased rates of exploitation. Ideally, the indicator species should be the weakest link of the assemblage, although defining weakest link could be problematic, along with choosing the correct species (Simberloff, 1998). Furthermore, there may be limited data for the species that is the weakest link of a ma- rine fish assemblage. If achievable, however, such a 262 Fishery Bulletin 106(3) Table 1 Clusters of species in headboat and commercial landings, listed in order of strongest to weakest cluster, as measured by each cluster’s average silhouette width (in parentheses). Clusters were partitioned around £ = 14 (headboat) or k=l (commercial) medoids — values determined by highest average silhouette widths computed from all species. See Appendix for the scientific names of species. Headboat clusters Commercial clusters One (0.22) Six (0.07) One (0.17) Six (0.01) Bank sea bass Cubera snapper Black sea bass Bluestriped grunt Black sea bass Warsaw grouper Gag Crevalle jack Gag Gray triggerfish French grunt Gray triggerfish Seven (0.04) Margate Lane snapper Greater amberjack Blackfin snapper Red grouper Whitebone porgy Knobbed porgy Cottonwick Red porgy Red porgy Sand tilefish Red snapper Seven (0.00) Red snapper Silk snapper Scamp Lesser amberjack Scamp Vermilion snapper Tomtate Eight (0.03) White grunt Vermilion snapper Black margate White grunt Coney Two (0.14) Whitebone porgy Porkfish Black grouper Blue runner Two (0.19) Nine (0.00) Gray snapper Blue runner Bar jack Greater amberjack Bluestriped grunt Black grouper Mutton snapper French grunt Hogfish Yellowtail snapper Gray snapper Margate Jolthead porgy Ocean triggerfish Three (0.07) Lane snapper Red hind Almaco jack Mutton snapper Saucereye porgy Banded rudderfish Red grouper Schoolmaster Blueline tilefish Yellowtail snapper Silk snapper Ten (0.00) Snowy grouper Three (0.18) Atlantic Tilefish Blueline tilefish spadefish Yellowedge grouper Snowy grouper Speckled hind Eleven (0.00) Four (0.04) Yellowedge grouper Crevalle jack Hogfish Jolthead porgy Four (0.11) Twelve (0.00) Rock hind Almaco jack Dog snapper Banded rudderfish Five (0.02) Graysby Thirteen (0.00) Knobbed porgy Queen triggerfish Longspine porgy Ocean triggerfish Rock hind Red hind Fourteen (0.00) Five (0.09) Scup Nassau grouper Yellowfin grouper Yellowmouth grouper restrictive use of indicator species could be considered a precautionary approach to management. In this study of reef fishes off the southeastern United States, we found little evidence of synchrony in popu- lation dynamics, and thus, no support for the use of indicator species. One possible reason for these negative results is that the study area was too broad; however, similar findings have been documented at smaller spa- tial scales (Parker and Dixon, 1998). A second reason is that the indices of abundance did not accurately Shertzer and Williams: Reef fishes off the the southern United States 263 Height 0000000- OJ^cna^ 'vJOocDO I I I I I I I I Nassau grouper Gray snapper Lane snapper Yellowmouth grouper Yellowfin grouper Blackfin snapper Coney — Bar jack Schoolmaster Porkfish Black margate French grunt Sand tilefish , Silk snapper 1 Jolthead porgy . Bluestriped grunt Blue runner Mutton snapper Yellowtail snapper Rock hind Red grouper Black grouper Margate Whitebone porgy Bank sea bass Ocean triggerfish Saucereye porgy Red hind Hogfish Longspine porgy Crevalle jack Atlantic spadefish — Scup Knobbed porgy White grunt Scamp Greater amberjack - Gag 1 Red snapper ' Tomtate 1_ Black sea bass ' Red porgy Vermilion snapper — Gray triggerfish — Banded rudderfish Queen triggerfish Graysby ■ Almaco jack - Speckled hind Warsaw grouper Cubera snapper Yellowedge grouper Snowy grouper Blueline tilefish Figure 3 Dendrogram from hierarchical cluster analysis of species in the headboat sector. Height measures similarity among species within a branch, with a value of 1.0 rep- resenting the lowest similarity. Scientific names are provided in the Appendix. represent actual relative abundances because indices were computed from fishery-dependent (headboat) data (Arreguin-Sanchez, 1996; Harley et ah, 2001). Fishery- independent data would have been preferable; however, for most species in this study, such data were of small sample size, short survey duration, or were nonexistent. We contend that the headboat data set was the best available for computing indices of abundance because of its relatively large sample size, long duration, and wide geographic coverage. In addition, headboat effort is expended generally toward a complex of species rather than specific stocks, and that generality minimizes any confounding effect of density-dependent catchability. A third possible reason for the negative results is that the population dynamics were truly out of synchrony. From a practical perspective, the actual reason, whether it stemmed from inadequate data or real dynamics, is of secondary concern. Foremost, positive evidence of 264 Fishery Bulletin 106(3) Height _L Lesser amberjack Silk snapper Tilefish Yellowedge grouper Snowy grouper Blueline tilefish Banded rudderfish Red grouper White grunt Scamp — Gray triggerfish Vermilion snapper Red porgy Black sea bass Gag Red snapper Almaco jack Greater amberjack Rock hind Hogfish Jolthead porgy Knobbed porgy Red hind Ocean triggerfish Whitebone porgy Bluestriped grunt - Lane snapper - Crevalle jack Blue runner Gray snapper Yellowtail snapper Black grouper Mutton snapper Margate French grunt Figure 4 Dendrogram from hierarchical cluster analysis of species in the commercial sector. Height measures similarity among species within a branch, with a value of 1.0 rep- resenting the lowest similarity. Scientific names are provided in the Appendix. synchrony in this reef fish complex has yet to be estab- lished, and we therefore urge precaution before using indicator species. We did find positive evidence of species assemblages on the basis of landings, but these were not necessar- ily ecological assemblages. Although assemblages in landings may reflect those in nature, the two could differ if some species are preferentially retained from the catch or are more vulnerable to exploitation. Still, assemblages in landings have direct implications from the perspective of managing fisheries, in terms of reducing bycatch and controlling fishing effort across species. Nondimensional scaling analysis revealed that the species assemblages are not strongly coherent. Such loose structure has also been found in assemblages north of our study area (Mahon et ah, 1998). Nonethe- less, agreement between the headboat and commer- Shertzer and Williams: Reef fishes off the the southern United States 265 Table 2 Synchrony in indices of abundance of reef fishes off the southeastern United States from headboat data. Assemblages are labeled as Deepwater, Southern (south of Cape Canaveral, Florida), and Northern (north of Cape Canaveral, Florida). Values are correla- tion coefficients and proportions of coefficients (in parentheses) from «=10,000 randomizations that were smaller than the corre- lations presented, such that values near 1.0 indicate significance of positive coefficients and values near 0.0 indicate significance of negative coefficients. Asterisks note significance at the a=0.1 level after Bonferroni correction (two-tailed test). Abbreviations for species names are explained in the Appendix. Deepwater SnwGpr BluTlf SpkHnd SnwGpr BluTlf SpkHnd 1.00 0.45(0.96) 1.00 0.08(0.62) -0.06 (0.41) 1.00 Southern BluRun YtlSnp LanSnp GrySnp MtnSnp BluRun 1.00 -0.16 (0.21) -0.06 (0.38) -0.11 (0.28) -0.23 (0.11) YtlSnp — 1.00 0.73V1.00) 0.78V1.00) -0.56V0.00) LanSnp — — 1.00 0.82V100) -0.65V0.00) GrySnp — — — 1.00 -0.64V0.00) MtnSnp — — — — 1.00 Northern BlckSB WhtGnt Tomtat GryTrf VrmSnp Gag BlckSB 1.00 -0.67V0.00) 0.24 (0.89) -0.35 (0.04) 0.47V0.99) 0.37 (0.97) WhtGnt — 1.00 -0.41 (0.02) 0.43 (0.99) -0.35 (0.04) -0.35(0.03) Tomtat. — — 1.00 0.03 (0.56) 0.36 (0.97) 0.19 (0.83) GryTrf — — — 1.00 -0.08(0.34) -0.14 (0.23) VrmSnp — — — — 1.00 -0.04(0.42) Gag — — — — — 1.00 cial data sets implies that the assemblages, although loosely structured, are not arbitrary. Moreover, ad- ditional analyses conducted as part of this study re- vealed assemblages that were quite similar to those presented. These analyses included the use of an alter- native transformation [logOc+1)], alternative measure of distance (binary dissimilarities), alternative link- age method with the hierarchical analysis (average linkage), commercial data from multiday trips (two, three, or four-plus days), and data by trip (i.e., trips not aggregated over months). The three clusters with the most coherence were deepwater, southern, and northern assemblages. The ranges of these assemblages likely correlate with physi- cal characteristics (as our chosen labels imply). Several of the assemblage species have been found to be linked through latitude, depth, and hard bottom habitat (Sed- berry and Van Dolah, 1984; Cuellar et al., 1996). Such information should be beneficial for managing assem- blages as units, allowing regulations to be focused on relevant geographic areas. Although the status of many stocks in the snapper- grouper complex is unknown, it is evident from most stock assessments that overfishing is occurring. In ag- gregate, these assessment results indicate overfishing of the ecosystem in general (Murawski, 2000). The av- erage level of overfishing and its variance, along with considerations of life histories and vulnerabilities, may indicate appropriate degrees of reduction in fishing ef- fort across assemblages. The use of multiple species as probes into ecosystem health is likely more robust than the use of a single indicator species. As single-species management loses fashion, its ideal replacement of full ecosystem management remains theoretically appealing, yet impractical given current data and understanding of marine ecosystems. Prog- ress toward ecosystem management will likely occur in increments (Hall and Mainprize, 2004). Where assem- blages exist, managing them as such offers a practicable step for implementing ecosystem considerations, includ- ing bycatch and species interrelations. Although the results of our study do not support the use of indicator species, they provide information on fish communities fundamental to the judicious application of assemblage management. Acknowledgments The authors are grateful for the support of the Southeast Fisheries Science Center (National Marine Fisheries Service), for comments from J. McGovern, M. Prager, C. Taylor, D. Vaughan, and anonymous reviewers, and for cooperation of the many commercial fishermen and 266 Fishery Bulletin 106(3) Table 3 Synchrony in first differences of indices of abundance of reef fishes off the southeastern United States from headboat data. Assemblages are labeled as Deepwater, Southern (south of Cape Canaveral, Florida), and Northern (north of Cape Canaveral, Florida). Values are correlation coefficients and proportions of coefficients (in parentheses) from n=10,000 randomizations that were smaller than the correlations presented, such that values near 1.0 indicate significance of positive coefficients and values near 0.0 indicate significance of negative coefficients. Asterisks note significance at the a=0.1 level with Bonferroni correction (two-tailed test). Abbreviations for species names are explained in the Appendix. Deepwater SnwGpr BluTlf SpkHnd SnwGpr BluTlf SpkHnd 1.00 0.31 (0.87) 1.00 -0.61*(0.01) -0.14(0.31) 1.00 Southern BluRun YtlSnp LanSnp GrySnp MtnSnp BluRun 1.00 -0.25(0.11) -0.10 (0.32) -0.11 (0.31) -0.25 (0.11) YtlSnp — 1.00 0.24 (0.89) -0.08 (0.35) 0.25 (0.90) LanSnp — - 1.00 -0.19 (0.17) 0.37(0.97) GrySnp — — — 1.00 -0.06 (0.38) MtnSnp — — — — 1.00 Northern BlckSB WhtGnt Tomtat GryTrf VrmSnp Gag BlckSB 1.00 0.05 (0.61) 0.16(0.78) 0.18 (0.82) 0.10 (0.70) -0.06(0.38) WhtGnt — 1.00 -0.09 (0.33) 0.20 (0.84) -0.06(0.37) -0.22(0.14) Tomtat — — 1.00 0.08(0.65) 0.23(0.88) 0.02 (0.47) GryTrf — — — 1.00 0.50*(1.00) -0.11 (0.28) VrmSnp — - — — 1.00 0.21 (0.86) Gag — — — — — 1.00 headboat operators who submitted logbook data. The data sets were provided to us by K. Brennan (headboat) and K. McCarthy (commercial). Literature cited Arreguin-Sanchez, F. 1996. Catchability: a key parameter for fish stock assessment. Rev. Fish Biol. Fish. 6:221-242. Beals, E. W. 1973. Mathematical elegance and ecological naivete. J. Ecol. 61:23-35. Bjprnstad, O. N., R. A. Ims, and X. Lambin. 1999. Spatial population dynamics: analyzing patterns and processes of population synchrony. Trends Ecol. Evol. 14:427-432. 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. In Biology, fisher- ies and culture of tropical groupers and snappers (F. Arreguin-Sanchez, J. L. Munro, M. C. Balgos, and D. Pauly, eds.), p. 59-73. Int. Specialized Book Services, Portland, OR. Edgington, E. S. 1995. Randomization tests, 3rd ed., 407 p. Marcel Dekker, Inc., New York, NY. Faith, D. P., P. R. Minchin, and L. Belbin. 1987. Compositional dissimilarity as a robust measure of ecological distance. Plant Ecol. 69:57-68. Federal Register. 1998. Magnuson-Stevens Act provisions, national stan- dard guidelines. Code of Federal regulations, title 50, part 600. U.S. Government Printing Office, Wash- ington, D.C. Field, J. G., K. R. Clarke, and R. M. Warwick. 1982. A practical strategy for analysing multispecies distribution patterns. Mar. Ecol. Prog. Ser. 8:37-52. Gomes, M. C., E. Serrao, and M. F. Borges. 2001. Spatial patterns of groundfish assemblages on the continental shelf of Portugal. ICES J. Mar. Sci. 58:633-647. Hall, S. J., and B. Mainprize. 2004. Towards ecosystem-based fisheries manage- ment. Fish Fish. 5:1-20. Hardin, J., and J. Hilbe. 2001. Generalized linear models and extensions, 245 p. Stata Press, College State, TX. Harley, S. J., R. A. Myers, and A. Dunn. 2001. Is catch-per-unit-effort proportional to abundance? Can. J. Fish. Aquat. Sci. 58:1760-1772. Kaufman, L., and P. J. Rousseeuw. 1990. Finding groups in data: an introduction to clus- ter analysis, 319 p. John Wiley and Sons, Inc., New York, NY. Shertzer and Williams: Reef fishes off the the southern United States 267 Koch, C. F. 1987, Prediction of sample size effects on the mea- sured temporal and geographic distribution patterns of species. Paleobiology 13:100-107. Kruse, G. H., V. F. Gallucci, D. E. Hay, R. I. Perry, R. M. Peterman, T. C. Shirley, P. D. Spencer, B. Wilson, and D. Woodby (eds.). 2005. Fisheries assessment and management in data- limited situations, 958 p. Alaska Sea Grant College Program, Univ. Alaska Fairbanks, Fairbanks, AK. Kruskal, J. B. 1964. Multidimensional scaling by optimizing good- ness of fit to a nonmetric hypothesis. Psychometrika 29:1-27. Lampert, D. 1992. Zero-inflated poisson regression, with an application to defects in manufacturing. Technometrics 34:1-14. Landres, P. B., J. Verner, and J. W. Thomas. 1988. Ecological uses of vertebrate indicator species: A critique. Conserv. Biol. 2:316-328. Lee, Y., and B. Sampson. 2000. Spatial and temporal stability of commercial groundfish assemblages off Oregon and Washington as inferred from Oregon trawl logbooks. Can. J. Fish. Aquat. Sci. 57:2443-2454. Leibold, M. A. 1995. The niche concept revisited: mechanistic models and community context. Ecology 76:1371-1382. Lo, N. C., L. D. Jacobson, and J. L. Squire. 1992. Indices of relative abundance from fish spotter data based on delta-lognormal models. Can. J. Fish. Aquat. Sci. 49:2515-2526. Mahon, R., S. K. Brown, K. C. T. Zwanenburg, D. B. Atkinson, K. R. Buja, L. Claflin, G. D. Howell, M. E. Monaco, R. N. O’Boyle, and M. Sinclair. 1998. Assemblages and biogeography of demersal fishes of the east coast of North America. Can. J. Fish. Aquat. Sci. 55:1704-1738. Maunder, M. N., and A. E. Punt. 2004. Standardizing catch and effort data: a review of recent approaches. Fish. Res. 70:141-159. May, R. M., and R. H. MacArthur. 1972. Niche overlap as a function of environmental variability. Proc. Natl. Acad. Sci. 1109-1113. McCune, B., and J. B. Grace. 2002. Analysis of ecological communities, 300 p. MjM Software Design, Gleneden Beach, OR. MSRA (Magnuson-Stevens Fishery Conservation and Manage- ment Reauthorization Act of 2006). 2006. Pub. L. no. 109-479, 120 Stat. 3575. Mueter, F. J., and B. L. Norcross. 2000. Changes in species composition of the demersal fish community in nearshore waters of Kodiak Island, Alaska. Can. J. Fish. Aquat. Sci. 57:1169-1180. Murawski, S. A. 2000. Definitions of overfishing from an ecosystem perspective. ICES J. Mar. Sci. 57:649-658. Niemi, J. G., J. M. Hanowski, A. R. Lima, T. Nicholls, and N. Weiland. 1997. A critical analysis on the use of indicator species in management. J. Wildl. Manag. 61:1240-1252. Parker, R. O., and R. L. Dixon. 1998. Changes in a North Carolina reef fish commu- nity after 15 years of intense fishing global warming implications. Trans. Am. Fish. Soc. 127:908-920. Prager, M. H., and J. M. Hoenig. 1989. Superposed epoch analysis: A randomization test of environmental effects on recruitment with application to chub mackerel. Trans. Am. Fish. Soc. 118:608- 618. Rousseeuw, P. J. 1987. Silhouettes: a graphical aid to the interpretation and validation of cluster analysis. J. Comp. Appl. Math. 20:53-65. Sedberry, G. R., and R. F. Van Dolah. 1984. Demersal fish assemblages associated with hard bottom habitat in the South Atlantic Bight of the USA. Environ. Biol. Fish. 11:241-258. Shaul, L., L. Weitkamp, K. Simpson, and J. Sawada. 2007. Trends in abundance and size of coho salmon in the Pacific Rim. N. Pac. Anadr. Fish Comm. Bull. 4:93-104. Simberloff, D. 1998. Flagships, umbrellas, and keystones: Is single- species management passe in the landscape era? Biol. Conserv. 83:247-257. Stefansson, G. 1996. Analysis of groundfish survey abundance data: combining the GLM and delta approaches. ICES J. Mar. Sci. 53:557-558. Williams, E. H., and S. Ralston. 2002. Distribution and co-occurrence of rockfishes (family: Sebastidae) over trawlable shelf and slope habi- tats of California and southern Oregon. Fish. Bull. 100:836-855. Zacharias, M. A., and J. C. Roff. 2001. Use of focal species in marine conservation and management: a review and critique. Aquatic Conserv: Mar. Freshw. Ecosyst. 11:59-76. 268 Fishery Bulletin 106(3) Appendix Species in the snapper-grouper complex off the southeastern United States, as managed under the Snapper Grouper Fishery Management Plan of the South Atlantic Fishery Management Council. Common name Scientific name Short name Almaco jack Seriola rivoliana AlmJck Atlantic spadefish Chaetodipterus faber AtlSpf Banded rudderfish Seriola zonata BndRud Bank sea bass Centropristis ocyurus BankSB Bar jack Caranx ruber BarJck Black grouper Mycteroperca bonaci BlkGpr Black margate Anisotremus surinamensis BlkMrg Black sea bass Centropristis striatus BlckSB Black snapper Apsilus dentatus BlkSnp Blackfin snapper Lutjnaus buccanella BfnSnp Blue runner Caranx chysos BluRun Blueline tilefish Caulolatilus mierops BluTlf Bluestriped grunt Haemulon sciurus BstGnt Coney Epinephelus fulvus Coney Cottonwick Haemulon melanurum Cotwck Crevalle jack Caranx hippos CrvJck Cubera snapper Lutjanus cyanopterus CbrSnp Dog snapper Lutjanus jocu DogSnp F rench grunt Haemulon flavolineatum FrnGnt Gag Mycteroperca microlepis Gag Goliath grouper Epinephelus itajara GolGpr Grass porgy Calamus arctifrons GrsPgy Gray snapper Lutjanus griseus GrySnp Gray triggerfish Balistes capriscus GryTrf Graysby Epinephelus cruentatus Grysby Greater amberjack Seriola dummerili GrAjck Hogfish Lanchnolaimus maximus Hogfsh Jolthead porgy Calamus bajonado JltPgy Knobbed porgy Calamus nodosus KnbPgy Lane snapper Lutjanus synagris LanSnp Lesser amberjack Seriola fasciata LsAjck Longspine porgy Stenotomus carprinus LgsPgy Mahogany snapper Lutjanus mahogoni MhgSnp Margate Haemulon album Mar gat Misty grouper Epinephelus mystacinus MstGpr Mutton snapper Lutjanus analis MtnSnp Nassau grouper Epinephelus striatus NssGpr Ocean triggerfish Canthidermis sufflamen OceTrf Porkfish Anisotremus virginicus Prkfsh Puddingwife Halichoeres rad i at us Puddwf Queen snapper Etelis oculatus QenSnp Queen triggerfish Balistes vetula QenTrf Red grouper Epinephelus morio RedGpr Red hind Epinephelus guttatus RedHnd Red porgy Pagrus pagrus RedPgy Red snapper Lutjanus campechanus RedSnp Rock hind Epinephelus adscensionis RckHnd Rock sea bass Centropristis philadelphicus RockSB Sailors choice Haemulon parrai SlsChc Sand tilefish Malacanthus plumieri SndTlf Saucereye porgy Calamus calamus ScyPgy continued Shertzer and Williams: Reef fishes off the the southern United States 269 Appendix (continued) Common name Scientific name Short name Scamp Mycteroperca phenax Scamp Schoolmaster Lutjanus apodus Schmst Scup Stenotomus chrysops Scup Sheepshead Archosargus probatocephalus Shphed Silk snapper Lutjnaus vivanus SlkSnp Smatlmouth grunt Haemulon chrysargyreum SmtGnt Snowy grouper Epinephelus niueatus SnwGpr Spanish grunt Haemulon macrostomum SpnGnt Speckled Hind Epinephelus drummondhayi SpkHnd Tiger grouper Mycteroperca tigris TgrGpr Tilefish Lopholatilus chamaeleonticeps Tilfsh Tomtate Haemulon aurolineatum Tomtat Vermilion snapper Rhomboplites aurorubens VrmSnp Warsaw grouper Epinephelus nigritus WrsGpr White grunt Haemulon plumieri WhtGnt Whitebone porgy Calamus leucosteus WtbPgy Wreckfish Polyprion americanus Wrkfsh Yellow jack Caranx bartholomaei YelJck Yellowedge grouper Epinephelus flavolimbatus YdgGpr Yellowfin grouper Mycteroperca venenosa YlfGpr Yellowmouth grouper Mycteroperca interstitalis YlmGpr Yellowtail snapper Ocyurus clirysurus YtlSnp 270 Abstract — Cape Cod Bay (Massa- chusetts) is the only known winter and early spring feeding area for con- centrations of the endangered North Atlantic right whale ( Eubalaena gla- cialis ) population. During January- May, 1998-2002, 167 aerial surveys were conducted (66,466 km of total survey effort), providing a complete representation of the spatiotemporal distribution of right whales in the bay during winter and spring. A total of 1553 right whales were sighted; some of these sightings were multiple sightings of the same individuals. Right whale distribution and relative abundance patterns were quantified as sightings per unit of effort (SPUE) and partitioned into 103 23-km2 cells and 12 2-week periods. Significant interannual variations in mean SPUE and timing of SPUE maxima were likely due to physically forced changes in available food resources. The area of greatest SPUE expanded and con- tracted during the season but its center remained in the eastern bay. Most cells with SPUE>0 were inside the federal critical habitat (CH) and this finding gave evidence of the need for management measures within CH boundaries to reduce anthropogenic mortality from vessel strikes and entanglement . There was significant within-season SPUE variability: low in December- January, increasing to a maximum in late February-early April, and declining to zero in May; and these results provide support for management measures from 1 Janu- ary-15 May. Manuscript submitted 29 July 2007. Manuscript accepted 21 March 2008. Fish. Bull. 108:270-280 (2008). The views and opinions expressed or implied in this article are those of the author and do not necessarily reflect the position of the National Marine Fisheries Service, NOAA. Spatial and temporal distribution of North Atlantic right whales ( Eubalaena glacialis) in Cape Cod Bay, and implications for management Owen C. Nichols (contact author)1 Robert D. Kenney2 Moira W. Brown1-3 Email address for O.C. Nichols: onichols@umassd.edu 1 Provincetown Center for Coastal Studies 5 Holway Avenue Provincetown, Massachusetts 02657 Present address: School for Marine Science and Technology University of Massachusetts-Dartmouth 200 Mill Road Fairhaven, Massachusetts 02179 2 University of Rhode Island Graduate School of Oceanography Bay Campus Box 41 South Ferry Road Narragansett, Rhode Island 02882 3 New England Aquarium Centra! Wharf Boston, Massachusetts 02110 North Atlantic right whales ( Euba- laena glacialis) have been com- mercially exploited by whalers for centuries and are listed as endan- gered under the U.S. Endangered Species Act (ESA). The remnant popu- lation of approximately 300 individu- als, ranging along continental shelf waters from the southeastern United States to the Scotian Shelf (Winn et al., 1986; Waring et al., 2007), is con- sidered to be a small fraction of its original size (IWC, 2001). Despite a slight increase in the minimum popu- lation estimate from 295 individuals in 1992 to 306 in 2001 (Waring et al., 2007), there has been little sign of recovery since studies began in the 1970s (IWC, 2001; Kraus et al., 2001, 2005). Although estimates of survival probability decreased in the 1990s, a relatively modest reduction in mortal- ity rate could reverse the observed decline and reduce the potential for species extinction (Caswell et al., 1999; Fujiwara and Caswell, 2001; Kraus et al., 2005). The most frequent known anthropogenic causes of right whale mortality are collisions with ships and entanglements in fishing gear. Of 50 documented right whale mortalities from 1986 through 2005, at least 19 (38%) were due to ship col- lisions, and a minimum of six (12%) were attributed directly to entangle- ments (Kraus et al., 2005). Cape Cod Bay (Massachusetts), a semi-enclosed basin approximately 40 km in diameter (Fig. 1), is the only known winter and early spring feeding area for the remaining con- centrations of the right whale pop- ulation (Winn et al., 1986; Waring et al., 2007). Scientific observations of right whales in Cape Cod Bay began in 1955 (Watkins and Schev- ill, 1982; Schevill et al., 1986), and Reeves et al. (1999) reviewed whaling records indicating right whale pres- ence from the early 1600s. During shipboard surveys and commercial whale watching trips, right whales were sighted in Cape Cod Bay and Massachusetts Bay to the north in every month except December from 1978 to 1986. Most of these sight- ings were sparse for most months, and peak abundance occurred from February through April (Hamilton and Mayo, 1990). With the exception of an unusual summer occupancy in 1986, most right whales had departed Nichols et al.: Spatial and temporal distribution of Eubalaena glacialis in Cape Cod Bay 271 70° 40' W 70° 30' W 70° 20' W 70° 10' W 70° 00' W 69° 50' W 42° 10' N 42° 00' N 41° 50' N 41° 40' N Figure 1 Cape Cod Bay, Massachusetts, including the boundaries of the right whale (Eubalaena glacialis ) critical habitat (diagonally hatched area), Seasonal Area Management (SAM) West zone (gray shaded area), limit of state-regulated waters (dotted line), and outside boundaries for shipping lanes (dashed lines). the bays by mid-May (Hamilton and Mayo, 1990). Nearly one third of the population of photographically identi- fied and catalogued individual right whales has been sighted in Cape Cod Bay in recent years (see Hamilton et al., 2007). To reduce right whale mortality from ship strikes and entanglements, state and federal agencies have en- acted a number of management mea- sures. Cape Cod Bay was designated as federal critical habitat under the ESA in 1994 (Federal Register, 1994). The critical habitat includes all of Cape Cod Bay, except the segment of the bay west of 70°30'W, as well as a portion of the waters immediately to the north (Fig. 1), and was defined according to ESA criteria based on the knowledge of right whale distribution and patterns of habitat use available at the time (Hamilton and Mayo, 1990; Federal Register, 1994). In 1997, the state of Massachusetts instituted a Right Whale Conservation Plan that included modifications to fixed-gear fishing practices within state waters and a surveillance-based monitoring program that provided data on right whale presence to mariners and man- agement agencies (Brown et al., 2007). Also enacted in 1997, the federal At- lantic Large Whale Take Reduction Plan (ALWTRP) included measures de- signed to reduce entanglements in U.S. Atlantic waters (Federal Register, 1997). Regulations under both plans have affected fixed-gear fishing activities in Cape Cod Bay, including those of the lobster trap and anchored sink-gillnet fisheries. The most stringent regulations have been implemented within the boundaries of the critical habitat from 1 January through 15 May. Regu- lations governing the critical habitat during this period have included a more comprehensive modification of lobster gear than that required from 16 May to 31 De- cember and the prohibition of all gillnetting operations (Federal Register, 2002b). Additional state and federal regulations have been in effect outside the critical habi- tat, including the seasonal area management (SAM) component of the ALWTRP, which provides additional restrictions on fixed-gear fishing in an area immediately east of the critical habitat (SAM West) from 1 March through 30 April (Fig. 1; Federal Register, 2002a). Al- though the above are the most relevant spatially and temporally defined management measures, there are several additional and often overlapping regulations af- fecting fixed-gear fishing in the study area. Ship-strike reduction strategies currently under consideration in- clude measures to reroute ships and to implement speed restrictions from 1 January through 30 April in Cape Cod Bay, as well as speed restrictions in the waters immediately north and east of the critical habitat, or in the proposed Off Race Point Proposed Management Area (Federal Register, 2004, 2006). As part of the state conservation plan for right whales, a systematic aerial survey program was imple- mented in Cape Cod Bay during winter and spring be- ginning in January of 1998 (Brown et al., 2007). These surveys were designed with an emphasis on detecting right whales and notifying mariners and managers of their presence, and have resulted in a systematic data set that allows right whale occurrence and patterns of habitat use in the bay to be quantified for the first time. The objectives of this study were to use the aerial survey data to describe the spatial and temporal distri- bution of right whales in Cape Cod Bay, and to compare these data with the extent and timing of present and proposed management measures to evaluate their po- tential effectiveness. We examined the following critical management strategies: the specification of location and duration for the use of modified fishing gear; the use of critical habitat boundaries as a management unit: and the specification of critical periods when reduced ship speed would be mandated to reduce the mortality of right whales cause by ship strikes. 272 Fishery Bulletin 106(3) 70° 40' W 70° 30' W 70° 20' W 70°10'W 70° 00' W 69° 50' W 42° 10' N 42° 00' N 41° 50’ N 41° 40’ N 70° 40' W 70° 30’ W 70° 20’ W 70°10'W 70° 00' W 69° 50' W Figure 2 Aerial survey track lines for the Cape Cod Bay right whale (Euba- laena glacialis ) monitoring program (black lines: 15 east-west lines at 1.5 nautical mile (2.7 km) intervals and a sixteenth line paralleling the eastern shore of the Cape approximately 3 nmi from shore). Also shown are the area boundaries for the sightings per unit-of-effort analysis (shaded gray). Materials and methods Aerial surveys Aerial surveys were conducted from a Cessna 337-A Sky- master, a twin-engine, high wing aircraft with retract- able landing gear. The aircraft was flown at a standard altitude of 229 m and a ground speed of approximately 185 km/h, and survey methods were those developed for the Cetacean and Turtle Assessment Program (Scott and Gilbert, 1982) and adapted for right whale surveys (Brown et al., 2007). The aircraft was flown in sea con- ditions up to and including sea state 4 on the Beaufort wind scale, but were aborted in Beaufort sea state 5 or when visibility decreased below two nautical miles (3.7 km) in fog, rain, or snow. Surveys were conducted in winter and spring (Janu- ary through mid-May) of 1998-2002. Additional surveys were conducted in December of some years (for the purpose of this paper, a given year or season includes December of the previous calendar year.) The standard survey design consisted of 16 track lines (Fig. 2). Fif- teen east-west track lines were flown at 2.8-km inter- vals from the mainland to the eastern Cape Cod Bay shoreline. Analysis of the right whale survey data indi- cated that the effective total survey swath of a Cessna 337 was 4.2 km (Kenney et al., 1995); thus the 2.8 km track line spacing of the standard survey plan was designed to provide 100% coverage of the sea surface in the study area to maximize the potential to detect right whales. An additional 65-km track line was flown parallel to the outer coast of Cape Cod from the eastern end of the northernmost track line to a point east of Chatham, Massachusetts, at an approximate distance of 5.6 km from shore (Fig. 2). The entire survey (approximately 568 km of track line distance) was completed as often as conditions permitted (ca. 2 surveys/week), and ad- ditional track lines to the north of the standard survey design were occasionally added to address management concerns outside the bay (e.g., whale distribution in the shipping lanes). Analysis of survey data To quantitatively assess right whale distribution and minimize bias caused by uneven distribution of survey effort, we used the sightings per unit of effort (SPUE) Nichols et al.: Spatial and temporal distribution of Eubalaena glacialis in Cape Cod Bay 273 70° 40' W 70° 30' W 70° 20’ W 70° I O' W 70° 00' W 69° 50' W 42° 10' N 42° 00' N 41° 50' N 41° 40' N Figure 3 North Atlantic right whale ( Eubalaena glacialis) sightings ( + ) recorded during valid aerial survey effort 1998-2002. A sight- ing is defined as one or more whales observed at the same time and location. method (Kenney and Winn, 1986). This method first quantifies survey effort as length of track line sampled, then expresses SPUE as the number of whales sighted per standardized length of track. The boundaries of the study area for this analysis were 42°09'N to 41°39'N and 70°00'W to 70°39'W (Fig. 2). The study area was partitioned into 117 cells mea- suring 3 minutes of latitude (5.6 km) by 3 minutes of longitude (4.1 km). Each cell was 23 km2 in area. Four- teen cells lay completely over land and were eliminated; thus there were 103 cells available to be sampled. Survey data comprised a chronological sequence of latitude and longitude points that described the path flown by the aircraft. Each successive pair of points de- scribed a track segment, and the length of that segment (effort) could be computed from the latitude and longi- tude data (Kenney and Winn, 1986). For each survey, each track segment was partitioned into smaller sec- tions contained within the separate 3-minute cells. In order to standardize effort further, only those segments were completed where visibility was at least 3.7 km, Beaufort sea state was 3 or lower, aircraft altitude was below 325 m, and observers were on watch. Similarly, only right whales sighted under these defined, valid ef- fort conditions were included in our study. We included only sightings identified by observers in the field as definite or probable right whales in the analysis; sight- ings identified as possible right whales were eliminated. A sighting was defined as one or more whales observed at the same time and location; for simple distribution maps (e.g., Fig. 3) all sightings were treated as single points, but the number of whales sighted is factored into the SPUE analysis. The sampling season was de- fined as December through mid-May, and was divided for analysis into 12 two-week periods (1 December-17 May). Total effort and total number of right whales sighted within each cell were summed within periods and across years; then the number of whales sighted was divided by effort to generate the SPUE index, in units of whales sighted per 1000 km of valid effort. Cells with less than 2 km of effort (half the width of a cell) within a 2-week period were considered to have been inadequately sampled and were eliminated from further analyses. For mapping purposes, the SPUE values were ranked into levels. All cells with SPUE>0 were pooled, sorted from highest to lowest SPUE values, and separated by quartiles, representing the top, second highest, third, and bottom quarters of the distribution. The upper quarter was further partitioned by identifying the top 5% of all values. These values were mapped by using different shadings to show the classes. The cells were 274 Fishery Bulletin 106(3) Table t Total acceptable survey effort (i.e., where visibility was at least 3.7 km, Beaufort sea state was 3 or lower, air- craft altitude was below 325 m, and observers were on watch) for right whales (Eubalaena glacialis) in the Cape Cod Bay, Massachusetts, study area, summarized by two- week periods across the entire study, 1998-2002. Period Effort (km) 1-14 December 900 15-28 December 550 29 December-11 January 3203 12-25 January 5476 26 January-8 February 8242 9-22 February 7953 23 February-8 March 7535 9-22 March 4983 23 March-5 April 7965 6-19 April 9006 20 April-3 May 7863 4-17 May 2790 pooled across periods before classification; therefore the maps showing different time periods were comparable. As a test of the definition for critical habitat boundary (i.e., a definition that did not include a potential effort bias), SPUE values were compared between the 3-min- ute cells inside the critical habitat, outside and west of it (western Cape Cod Bay), and outside and east of it (outside of Cape Cod). All computations and statistical analyses were per- formed with Statistical Analysis System procedures (SAS for Windows, vers. 6.13, SAS Inst. Inc., Cary, NC). Calculations of effort and SPUE were performed with custom SAS programs developed at the University of Rhode Island. Whenever possible, we used statistical tests of hypotheses with nonparametric methods (in SAS PROC NPAR1WAY) to avoid the assumptions of normal distributions and homogeneous variances as- sociated with parametric tests. The test most often used was the Kruskal-Wallis nonparametric ANOVA. When the comparison involved only two distributions, the Wilcoxon rank sum test was used (equivalent to the Mann-Whitney U test or a parametric T-test). The only parametric test that we used was a Duncan’s mul- tiple range test (in SAS PROC GLM), which allowed us to perform multiple pairwise T-tests while correcting the critical significance probabilities for the number of comparisons. There is no nonparametric test equivalent available in SAS. Results During the 1998-2002 seasons, 167 aerial surveys were conducted, yielding 66,466 km of acceptable Table 2 Mean (±standard error [SE] ) and maximum number of sightings of North Atlantic right whales (Eubalaena gla- cialis) per unit of effort (no. of whales/1000 km of survey effort) in Cape Cod Bay, Massachusetts, summarized by year. Year Mean SE Maximum n 1998 14.58 2.45 789.4 819 1999 11.48 1.49 417.3 874 2000 18.44 2.49 625.7 728 2001 12.74 1.70 575.7 802 2002 2.11 0.61 210.7 739 All 11.91 0.84 789.4 3962 survey effort. Of the 103 3-minute cells available to be sampled in the study area, 101 (98.1%) were sampled at least once, and 88 (85.4%) were sampled in every year. Because the basic sampling unit represented aggregated data within a cell across a 2-week period, the maximum possible sample size was 6180 (103 cells x 5 years x 12 periods), and the actual number of samples (effort >2 km) was 3962 (64.1% of the maxi- mum possible). Overall mean effort was 658 km of survey track per 3-minute cell (range: 2.5 to 1189.5 km). Spatial distribution of effort was fairly uniform throughout the study area; lowest effort occurred in the peripheral cells, some of which were partially over land. Temporal distribution of effort was relatively consistent during the study period, with the excep- tion of December, when surveys were conducted on an ad hoc basis (Table 1). Effort was lower in May than from January through April because surveys in May were suspended after two or three consecutive surveys when right whales were not sighted. The remaining temporal variability in survey effort was primarily due to weather conditions or sea ice that prohibited or shortened surveys. During the aerial surveys, 1553 right whales were sighted, the majority of which were sighted during February-April. Cumulative sighting totals included multiple sightings of photographically identified indi- vidual whales within and between seasons. The earli- est sighting within a season occurred on 13 December 1998, and the latest sighting occurred on 2 May 1999. The highest number of right whales sighted on a single day was 47 on both 4 March 1998 and 27 March 2000. Sightings of 203 whales occurred during unacceptable survey conditions (mostly due to high sea state), and eight whales were documented for cells with acceptable conditions but during which there was less than 2 km of effort during a particular 2-week period; therefore sightings of 1342 whales were included in the SPUE analysis. Right whales were sighted during acceptable conditions in 69 of the 101 sampled cells (68.3%) over the five years of the study. Sightings were concen- Nichols et ai. : Spatial and temporal distribution of Eubalaena glacialis in Cape Cod Bay 275 70° 40' W 70° 30' W 70° 20' W 70° 10' W 70° 00' W 69° 50’ W - 42° 00' N 42° 10' N 41°50'N 41° 40’ N > 69.04 (top 5%) = 28.95-69.03 (75th— 95th percentile) = 12.31-28.94 (50th— 75th percentile) = 5.37-12.30 (25th— 50th percentile) = 0.01-5.36 (< 25th percentile) = 0 (effort > 0) no effort Figure 4 Overall mean North Atlantic right whale ( Eubalaena glacialis) sightings per unit of effort (SPUE; number of whales/1000 km of survey effort) in Cape Cod Bay, 1998-2002. SPUE values are separated by quartiles, representing the top, second highest, third, and bottom quarters of the distribution. The upper quarter is further partitioned by identifying the top 5% of all values. trated in a band extending across Cape Cod Bay from the southwest to the northeast (Fig. 3). The overall 5-year mean SPUE value for all 101 sampled cells was 11.9 whales per 1000 km of sur- vey effort (i.e., one right whale every 84 km of accept- able survey flown; Table 2). The maximum SPUE in a single cell during any two-week period was 789.4 right whales/1000 km (i.e., one whale every 1.3 km) during 29 December 1997-11 January 1998. An area of high-SPUE cells extended along the eastern side of the bay from the south to the tip of the Cape (Fig. 4). Distinct areas of lower SPUE values radiated out from the highest-value cells, and a gradient of decreasing SPUE values extended to the west. Annual mean SPUE was highest in 2000 and lowest in 2002 (Table 2), and the difference was almost an order of magnitude. The interannual variability was statistically significant (Kruskal-Wallis nonparametric ANOVA, P<0.001). When 2002 data was eliminated, the interannual variability remained statistically sig- nificant (Kruskal-Wallis, P= 0.033). When we tested for differences among the annual means with a parametric Duncan’s multiple range test, using either unweighted or effort-weighted means, 2002 was found to be sig- nificantly different from all other years. Using the un- weighted means, we found that 1998 and 2000 were not different from one another, and neither were 1998, 1999, and 2001. Using the effort-weighted means, we found that 1998, 1999, and 2001 did not differ, but that 2000 was significantly different from all other years. When separated into two-week periods pooled across all years, mean SPUE values varied considerably. SPUE values were low in December and January, in- creased during February, exhibited two peaks in late February-early March and late March-early April, and declined to low values by the end of April and to zero in May (Table 3). The within-season temporal vari- ability was statistically significant (Kruskal-Wallis, .PcO.OOl). The area occupied by right whales expanded and then contracted over the course of the season in 276 Fishery Bulletin 106(3) the same fashion (Fig. 5). The center of right whale oc- currence tended to remain relatively consistent across the season, and there was no statistically significant variability in mean latitude or longitude (for the 2- week intervals, weighted by SPUE) across the periods (Kruskal-Wallis, P-0.210 for latitude, P=0.580 for lon- gitude). Similarly, there was no significant interan- nual variability in mean latitude (P=0.063) or mean longitude (P=0.797). The timing of the SPUE maximum varied markedly between years: late February-early March in 1998, early to mid-April in 1999, late March-early April in 2000, early March to mid-April in 2001, and late Feb- ruary-early March in 2002. This interannual varia- tion was responsible for the apparent bimodal pattern seen over the season when data were pooled across all five years (Table 3). The within-season temporal variation was statistically significant in each of the five years (P<0.001 for each). The duration of right whale occurrence also varied substantially between years: January-April in 1998 (no December 1997 surveys), December-April in 1999, mid- January-mid-April in 2000, and at least late December-mid-April in 2001 (no surveys were undertaken during the first two weeks of December 2000), and only late January-late March (and a second brief period in early April) in 2002. The numbers of consecutive two-week periods when right whales were observed were 9 in 1998, 9 in 1999, 7 in 2000, 9 in 2001, and 5 in 2002. Approximately 97% of the right whales sighted within the study area by aerial survey were within the criti- cal habitat boundaries (Fig. 3). The critical habitat included approximately 75% of the total ocean area within our defined study area. The majority of cells within which right whales were sighted were inside the critical habitat, but 17 cells with SPUE >0 fell partially or entirely outside the critical habitat (Fig. 4). SPUE did vary significantly among the three areas (Kruskal- Wallis, P=0.009). However, the mean values for the cells east of the critical habitat (11.1 unweighted, 10.9 weighted by effort and by the proportion of cell area for the five cells straddling the eastern boundary) fell between the area inside the critical habitat (18.5 un- weighted, 23.1 weighted) and the aa outside to the west (2.3 unweighted, 2.0 weighted). Pairwise nonparametric comparisons (Wilcoxon rank sum tests, weighting was not possible) showed statistically significant differences between the areas east and west of the critical habitat (P=0.023) and between the areas inside and to the west (P- 0.003), but not between the areas inside and to the east (P=0.669). Discussion The spatial pattern of right whale distribution observed during this study was largely similar to that described by Hamilton and Mayo (1990); most sightings were concentrated toward the eastern side of the bay (Fig. 3). The majority of the data collected by Hamilton and Mayo Table 3 Mean (±standard error [SE] ) and maximum number of sightings of North Atlantic right whales ( Eubalaena gla- cialis) per unit of effort (no. of whales/1000 km of survey effort) in Cape Cod Bay, Massachusetts, summarized by two-week periods across the entire study, 1998-2002. Period Mean SE Maximum n 1-14 Dec 4.21 3.53 417.3 121 15-28 Dec 4.47 4.47 339.9 76 29 Dec-11 Jan 7.70 2.87 789.4 328 12-25 Jan 4.53 1.34 288.3 378 26 Jan-8 Feb 10.98 2.67 554.4 417 9-22 Feb 14.08 2.40 510.8 407 23 Feb-8 Mar 23.15 4.60 729.8 333 9-22 Mar 14.79 3.14 625.7 395 23 Mar-5 Apr 23.54 3.47 575.7 410 6-19 Apr 14.70 2.51 552.1 412 20 Apr-3 May 6.63 1.67 293.1 380 4-17 May 0.00 0.00 0.0 305 (1990) was derived from opportunistic sightings recorded by researchers working aboard commercial whale watch- ing vessels departing from Provincetown multiple times per day, bound for Stellwagen Bank to the north of Cape Cod, from mid-spring through early fall. Aerial survey effort in our study was systematic, consistent across years, and spatially uniform throughout the study area and thus provided the first complete representation of the spatial and temporal distribution of right whales in Cape Cod Bay during winter and spring. The SPUE analysis provided a more refined interpretation of raw sighting data than have earlier analyses by reducing bias caused by uneven allocation of effort. During the seasonal expansion of right whale distri- bution in the bay (Fig. 5), right whales may be particu- larly at risk from collisions with ships traveling between the mouth of the Cape Cod Canal at the southwest and ports to the northeast of the bay (Fig. 1; Ward-Geiger et al., 2005). The above risk should be considered when implementing routing measures for shipping traffic. The boundaries of the Cape Cod Bay critical habitat encompassed the areas of highest SPUE values when pooled across all years and periods, and therefore the existing boundaries appear to service as a good man- agement unit. Despite the low number of sightings east of Cape Cod, the area- and effort-weighted statistics indicated that the waters east of the critical habitat may be important habitat for right whales, providing some support for existing fishery management measures in place east of the critical habitat (SAM West; Fig. 1) and the proposed Off Race Point Seasonal Management Area for shipping activity, which includes the portion of SAM West within the boundaries of this analysis (Federal Register, 2006). Nichols et al.: Spatial and temporal distribution of Eubalaena glacialis in Cape Cod Bay 277 70° W 12-25 January ! | > 69.04 (top 5%) MBB = 28.95-69.03 ( 75th— 95th percentile) = 12.3 1-28.94 ( 50th— 75th percentile ) = 5.37-12.30 ( 25th— 50th percentile) = 0.01-5.36 (< 25th percentile) = 0 (effort > 0) no effort Figure 5 North Atlantic right whale ( Eubalaena glacialis ) sightings per unit effort (SPUE; whales/1000 km of survey effort) in Cape Cod Bay, by two-week period (A-L), 1998-2002. SPUE values are separated by quartiles, representing the top, second highest, third, and bottom quarters of the distribution. The upper quarter is further partitioned by identifying the top 5% of all values. The significant interannual variations in mean SPUE and timing of annual SPUE maxima were likely due to physically forced changes in available food resources. An atypical physical environment and zooplankton assemblage were observed in 2002, compared to 2000 and 2001 (DeLorenzo Costa et al., 2006; Jiang et al., 2007), which corresponded to the order-of-magnitude difference in 2002 right whale SPUE compared to that in 2000. Further comparisons of right whale distribu- tional data and environmental variables should be con- ducted to assess the causes of the patterns observed. Despite the interannual SPUE variability, the signifi- cant within-season and interannual spatial stability of right whale occurrence lends support to the use of these data to define the spatial extent of management measures. The results of the aerial surveys and SPUE analysis confirmed earlier findings of Hamilton and Mayo (1990) in that peak occurrence of right whales in the study ar- ea occurred from February through April. The patterns observed in both studies are largely consistent with earlier, more qualitative descriptions of right whale 278 Fishery Bulletin 106(3) 70° W \ - 42° N 23 February - 8 March > 69.04 (top 5%) IHH = 28.95-69.03 ( 75th— 95th percentile) = 12.3 1-28.94 ( 50th— 75th percentile) = 5.37-12.30 (25th-50th percentile) 9 - 22 March = 0.01-5.36 (< 25th percentile) = 0 (effort > 0) no effort Figure 5 (continued) occurrence in Cape Cod Bay (Allen, 1916; Watkins and Schevill, 1982; Schevill et al., 1986). More right whales were sighted in December and January of our study than by Hamilton and Mayo (1990); likely a result of the spatial and temporal expansion of survey effort through the surveillance-based monitoring program. The shipboard survey effort of Hamilton and Mayo (1990) was not consistent throughout all the months that were examined in the present study. In the first five years of their eight-year study, there were no sur- veys in January or February. Right whales were present inside the critical habitat in December, before the comprehensive gear restric- tions were implemented within its boundaries on 1 January, although low December effort makes it dif- ficult to interpret the significance of the sightings. Although gear restrictions were in place through 15 May, no right whales were sighted after 2 May, and SPUE dropped to zero throughout the study area dur- ing the period 4-17 May. Moreover, although there were no sightings of right whales after 2 May during our study, there were a few sporadic sightings in the study area throughout May in other years (Hamilton and Mayo, 1990). Given that the period of peak SPUE occurs from February through April, the duration of existing fixed gear fishing regulations (1 Jan-15 May) appears adequate because it provides a buffer of a few weeks on either side of the peak period of right whale abundance — a buffer that allows for the interannual variation observed. The ship-strike reduction measures currently proposed for the period 1 January-30 April (Federal Register, 2004, 2006) should be extended to Nichols et al.: Spatial and temporal distribution of Eubalaena glacialis in Cape Cod Bay 279 23 March - 5 April 70° W \ - 42° N > 69.04 (top 5%) = 28.95-69.03 ( 75th— 95th percentile) = 12.3 1-28.94 (50th -75th percentile) = 5.37-12.30 (25th 50th percentile) = 0.01-5.36 (< 25th percentile) = 0 (effort > 0) no effort Figure 5 (continued) 15 May to match the duration of fixed-gear fishery restrictions. The efficacy of management measures in Cape Cod Bay has range-wide importance for the right whale because a large portion of the remnant popula- tion frequents the bay and because the possibility of even a slight reduction in mortality may prevent spe- cies extinction. Acknowledgments We thank the aerial observers, the pilots of Ambroult Aviation, and the staff of the New England Aquarium. N. Jaquet (Provincetown Center for Coastal Studies), D. McKiernan and E. Lyman of the Massachusetts Division of Marine Fisheries (MADMF) reviewed ear- lier versions of this manuscript. Surveys were funded by MADMF with support from the National Oceanic and Atmospheric Administration (NOAA) and the Northeast Consortium and conducted under NOAA Fisheries permits 1014 and 633-1483. R. Kenney was supported by the NOAA Cooperative Marine Educa- tion and Research Program. This article is dedicated to those lost in a right whale survey aircraft crash on 26 January 2003. Literature cited Allen, G. M. 1916. The whalebone whales of New England. Mem. Boston Soc. Nat. Hist. 8:107-322. 280 Fishery Bulletin 106(3) Brown, M. W., S. D. Kraus, C. K. Slay, and L. P. Garrison. 2007. Surveying for discovery, science, and manage- ment. In The urban whale: North Atlantic right whales at the crossroads (S. D. Kraus, and R. M. Rolland, eds.), p. 105-137. Harvard Univ. Press, Cambridge, MA. Caswell, H., M. Fujiwara, and S. Brault. 1999. Declining survival probability threatens the North Atlantic right whale. Proc. Natl. Acad. Sci. 96:3308-3313. DeLorenzo Costa, A., E. G. Durbin, C. A. Mayo, and E. G. Lyman. 2006. Environmental factors affecting zooplankton in Cape Cod Bay: implications for right whale dynamics. Mar. Ecol. Prog. Ser. 323:281-298. Federal Register. 1994. Designated critical habitat; northern right whale (final rule). Federal Register 59:28793-28808. Office of the Federal Register, National Archives and Records Administration (NARA), College Park, MD. 1997. Taking of marine mammals incidental to com- mercial fishing operations; Atlantic Large Whale Take Reduction Plan regulations. Federal Register 62:16519-16538. Office of the Federal Register, NARA, College Park, MD. 2002a. Taking of marine mammals incidental to com- mercial fishing operations; Atlantic Large Whale Take Reduction Plan regulations. Federal Register 67:1142-1160. Office of the Federal Register, NARA, College Park, MD. 2002b. Taking of marine mammals incidental to com- mercial fishing operations; Atlantic Large Whale Take Reduction Plan regulations. Federal Register 67:1300-1314. Office of the Federal Register, NARA, College Park, MD. 2004. Advance notice of proposed rulemaking (ANPR) for right whale ship strike reduction. Federal Register 69:30857-30864. Office of the Federal Register, NARA, College Park, MD. 2006. Proposed rule to implement speed restrictions to reduce the threat of ship collisions with North Atlantic right whales. Federal Register 71:32699-36313. Office of the Federal Register, NARA, College Park, MD. Fujiwara, M., and H. Caswell. 2001. Demography of the endangered North Atlantic right whale. Nature 414:537-541. Hamilton, P. K., A. R. Knowlton, and M. K. Marx. 2007. Right whales tell their own stories: the photo-iden- tification catalog. In The urban whale: North Atlantic right whales at the crossroads (S. D. Kraus, and R. M. Rolland, eds.), p. 75-104. Harvard Univ. Press, Cambridge, MA. Hamilton, P. K., and C. A. Mayo. 1990. Population characteristics of right whales ( Euba - laena glacialis) observed in Cape Cod and Massachusetts Bays, 1978-1986. Rep. Int. Whal. Comm. Spec. Issue 12:203-208. IWC (International Whaling Commission). 2001. Report of the workshop on status and trends of western North Atlantic right whales. J. Cetacean Res. Manag. Spec. Issue 2:61-87. Jiang, M., M. W. Brown, J. T. Turner, R. D. Kenney, C. A. Mayo, Z. Zhang, and M. Zhou. 2007. Springtime transport and retention of Calanus finmarchicus in Massachusetts and Cape Cod Bays, USA, and implications for right whale foraging. Mar. Ecol. Prog. Ser. 349:183-197. Kenney, R. D., and H. E. Winn. 1986. Cetacean high-use habitats of the northeast United States continental shelf. Fish. Bull. 84:345-357. Kenney, R. D., H. E. Winn, and M. C. Macaulay. 1995. Cetaceans in the Great South Channel, 1979-1989: right whale ( Eubalaena glacialis). Continental Shelf Res. 15:385-414. Kraus, S. D., M. W. Brown, H. Caswell, C. W. Clark, M. Fujiwara, P. K. Hamilton, R. D. Kenney, A. R. Knowlton, S. Landry, C. A. Mayo, W. A. McLellan, M. J. Moore, D. P. Nowacek, D. A. Pabst, A. J. Read, and R. M. Rolland. 2005. North Atlantic right whales in crisis. Science 309:561-562. Kraus, S. D., P. K. Hamilton, R. D. Kenney, A. R. Knowlton, and C. K. Slay. 2001. Reproductive parameters of the North Atlantic right whale. J. Cetacean Res. Manag. Spec. Issue 2:231-236. Reeves, R. R., J. M. Breiwick, and E. D. Mitchell. 1999. History of whaling and estimated kill of right whales, Balaena glacialis, in the northeastern United States, 1620-1924. Mar. Fish. Rev. 61:1-36. Schevill, W. E., W. A. Watkins, and K. E. Moore. 1986. Status of Eubalaena glacialis off Cape Cod. Rep. Int. Whal. Comm. Spec. Issue 10:79-82. Scott, G. P., and J. R. Gilbert. 1982. Problems and progress in the US BLM-spon- sored CETAP surveys. Rep. Int. Whal. Comm. 32:587-600. Ward-Geiger, L. I., G. K. Silber, R. D. Baumstark, and T. L. Pulfer. 2005. Characterization of ship traffic in right whale crit- ical habitat. Coast. Manag. 33:263-278. Waring, G. T., E. Josephson, C. P. Fairfield, and K. Maze- Foley, eds. 2007. U. S. Atlantic and Gulf of Mexico marine mammal stock assessments — 2006. NOAA Tech. Memo. NMFS NE 201, 378 p. Watkins, W. A., and W. E. Schevill. 1982. Observations of right whales, Eubalaena glacialis, in Cape Cod waters. Fish. Bull. 80:875-880. Winn, H. E., C. A. Price, and P. A. Sorenson. 1986. The distributional biology of the right whale Euba- laena glacialis in the western North Atlantic. Rep. Int. Whal. Comm. Spec. Issue 10:129-138. 281 Abstract — Groundfish fisheries in the southeast Bering Sea in Alaska have been constrained in recent years by management measures to protect the endangered Steller sea lion ( Eumetopias jubatus). There is concern that the present commercial harvest may produce a localized deple- tion of groundfish that would affect the foraging success of Steller sea lions or other predators. A three- year field experiment was conducted to determine whether an intensive trawl fishery in the southeast Bering Sea created a localized depletion in the abundance of Pacific cod (Gadus macrocephalus). This experiment produced strongly negative results; no difference was found in the rate of seasonal change in Pacific cod abundance between stations within a regulatory no-trawl zone and stations in an immediately adjacent trawled area. Corollary studies showed that Pacific cod in the study area were highly mobile and indicated that the geographic scale of Pacific cod move- ment was larger than the spatial scale used as the basis for current no-trawl zones. The idea of localized depletion is strongly dependent on assumed spa- tial and temporal scales and contains an implicit assumption that there is a closed local population. The scale of movement of target organisms is criti- cal in determining regional effects of fishery removals. Manuscript submitted 6 June 2007. Manuscript accepted 28 March 2008. Fish. Bull. 106:281-292 (2008). The views and opinions expressed or implied in this article are those of the author and do not necessarily reflect the position of the National Marine Fisheries Service, NOAA. Effects of commercial fishing on local abundance of Pacific cod ( Gadus macrocephalus ) in the Bering Sea M. Elizabeth Conners (contact author) Peter Munro Email address for M. E. Conners: Liz.conners@noaa.gov National Marine Fisheries Service Alaska Fisheries Science Center 7600 Sand Point Way NE Seattle, Washington 98115-6349 As fishery management in the United States moves toward an ecosystem approach, both the direct and indi- rect effects of commercial harvests on marine food webs are being considered (Schoener, 1993; Murawski, 2000). Localized depletion is the hypothesis that intense fishing pressure may cause small-scale effects on local den- sities of the target fish — effects that are disproportionate to the managed overall harvest mortality rate. The possibility of localized depletion has been considered in developing and implementing regulations to reduce the jeopardy of mortality of Steller sea lions ( Eumetopias jubatus) due to fishing in the Bering Sea. In 1997, the western stock of Steller sea lions was listed as endangered under the U.S. Endangered Species Act. Because the period of decline in sea lion num- bers coincides with the period of rapid expansion of U.S. domestic fisheries in Alaska (Braham et al., 1980; Hiatt, 2005), there has been concern that commercial fisheries may indirectly affect sea lion abundance through competition for available prey (Alver- son, 1992; Fritz et al., 1995). During their consultations regard- ing the U.S. Endangered Species Act, scientists of the National Ma- rine Fisheries Service (NMFS) cited the federally managed fishery for Pa- cific cod (Gadus macrocephalus) as one with a potential to adversely af- fect Steller sea lions. Pacific cod is the third-most valuable commercial fish species in the United States; the value of the 2006 harvest was esti- mated at $197 million. Approximately 40% of Pacific cod catch is taken by trawling in the southeastern portion of the Bering Sea during the winter “A” season, from 20 January through 31 March (Thompson and Dorn, 2005). The trawl fishery has histori- cally been concentrated on an area of the continental shelf north of Unimak Island, where Pacific cod form dense spawning aggregations during the winter season. Although the Pacific cod population in the eastern Bering Sea is managed at sustainable overall harvest levels, intensive trawl fishing in a smaller area has been suspected of causing short-term, small-scale impacts on fish abundance (localized depletion) that are disproportionate to the over- all harvest rate. Particularly worri- some is the possibility that fishing mortality rates in the heavily trawled area may be much higher than those for the total region, and that this disproportionate mortality may cause reduced availability of Pacific cod as prey for Steller sea lions (Fritz et al., 1995). Evidence indicating such a lo- cal effect has been seen in commer- cial catch data in the eastern Bering Sea (Fritz and Brown, 2005). Under the assumption of reduced prey avail- ability from localized fishing, trawl exclusion zones have been placed around Steller sea lion rookeries and haulouts in western Alaska, and sea- sonal and spatial allocation of catch quotas have been used to disperse fishing effort. Our goal was to test for effects of localized depletion of Pacific cod caused by the Pacific cod winter trawl fishery. 282 Fishery Bulletin 106(3) In order to test for the presence of localized depletion, the temporal and spatial scales of the presumed ef- fects must be specified. Localized depletion is a general term that could encompass several types of interactions. Short-term movement rates of the target species in re- lation to the scale of the fishery are a critical factor in determining fishing effects. To illustrate, we offer three conjectures on fishery interaction, where the dynamics of harvest and fish movement result in different effects on fish abundance. For the first conjecture, harvest results in a localized reduction in fish abundance in the immediate vicinity of fishing. This reduction remains geographically stable for some period of time. We refer to this effect as sta- tionary localized depletion. This form of depletion may be envisioned as the action of a dipper to remove cer- tain amount of mud from a bucket; a hole or depression would remain where the mud was removed, eventually filling in but persisting long enough to be observable. Implicit in such stationary localized depletion is the notion that movement by the fish is on a scale smaller than that of the fishery; therefore it does not obscure the geographic imprint of the removal. For the second conjecture, the movement of the fish interacts with a locally intense harvest. In this case, short-term fish movement occurs on a geographic scale greater than that of the fishery and the effect of the removal would quickly dissipate over the area occupied by the fish. This form of depletion would more closely resemble the action of ladle in dipping water out of a bucket, with no persistent depression left behind by the ladle. On the scale of the ladle, the effect is transi- tory and there is no apparent localized depletion, even though there is a measurable removal on the scale of the bucket. For the third conjecture, the short-term fish move- ment includes a net flow in one direction. In this sce- nario, fishing effects would show as a change to the flow of fish or as an area of reduced abundance that is displaced downstream. This effect would be similar to intercepting part of a flowing stream with a dip net. Combinations of random and directed short-term move- ment may also produce effects that are both dispersed and spatially displaced from the location of harvest. In the specific case of Bering Sea Pacific cod, the scale of fishery harvest is known from commercial fish- ing data, but the scale of Pacific cod movement is poorly understood. From an early tagging study (Shimada and Kimura, 1994), it was found that Pacific cod make large-scale movements over the eastern Bering Sea on a seasonal basis. Short-term movement dynamics of Pacific cod, on the other hand, are only beginning to be studied (Nichol and Chilton, 2006). Assumed scales of Steller sea lion foraging are based largely on satellite tagging studies of adult female and juvenile Steller sea lions (Merrick and Loughlin, 1997). Measures enacted to protect Steller sea lions include trawl exclusion zones 18.5 or 37.0 km in diameter around Steller sea lion haulouts and rookeries. Use of trawl exclusion zones to reduce competition between Steller sea lions and trawlers implies an assumption of stationary localized depletion at this scale. Thus, under the current regula- tory framework, the general question of whether fishing causes some form of localized depletion becomes much more specific, namely that of whether or not trawling causes a stationary localized depletion on the geograph- ic scale of the trawl exclusion zones. Our experiment was therefore designed to address the specific mecha- nism that causes stationary localized depletion, and we assumed that this type of depletion would have the greatest potential for negatively affecting Steller sea lion abundance. Materials and methods Study area The study area was located in the southeast Bering Sea near the tip of the Alaska Peninsula (Fig. 1) — an area situated off Unimak Island at Cape Sarichef on the east- ern side of Unimak Pass, at depths of approximately 70 to 110 m. This area is one of the most productive trawl- ing grounds in the Bering Sea, where fisheries focus a great deal of trawling effort for Pacific cod. The trawl exclusion zone around the Steller sea lion haulout at Cape Sarichef intersects this preferred trawling ground and the two areas provided us an opportunity to use a spatially adjacent treatment zone and control zone. The spatial scale of the experiment was determined by the 18. 5 -km boundary of the Cape Sarichef trawl exclusion zone and the extent of preferred depths for trawl fishing. The temporal scale, the length of time an effect must persist to be observable, was taken to be approximately two weeks, which was the time required to conduct each phase of the research fishing. Research pot gear We used the catch of standardized pot gear as an index of local Pacific cod abundance. Pot gear is widely used in commercial fishing for both crab and Pacific cod in the eastern Bering Sea and Gulf of Alaska. Research catches were based on catch from pots, even though the fishery effect being tested for would have been due to trawling, because pots provide large sample sizes and can be deployed at a very high spatial resolution. Pot catches cannot be easily or reliably used to estimate absolute abundance of fish, but provide a consistent index of relative abundance that can be used to make statistical comparisons between different survey areas or times. Commercial pots were used during initial feasibil- ity and pilot studies for this project; a standardized research pot was then developed and used for the ex- periment. The research pots were slightly larger than most commercial pots, and had a smaller net mesh and modified tunnel openings: they were 2.3 m by 2.3 m by 1.2 m, and had 5-cm stretched mesh and two entrance tunnels, each with 68-cm by 23-cm tunnel openings. Conners and Munro: Effects of commercial fishing on local abundance of Gadus macrocephalus in the Bering Sea 283 1 65°30'0"W 165°0'0"W 164°30'0"W Location of the study to determine localized depletion of Pacific cod ( Gadus macrocephalus) in the southeast Bering Sea. The large arc shows the regula- tory notrawl zone around the Steller sea lion ( Eumetopias jubatus ) haulout at Cape Sarichef on Unimak Island. The small triangles indicate station locations for the experiment. Depth contour lines and light shading show the region of 70 to 100 m depth. At the tunnel openings there was a modified Hilti-style stainless-steel trigger, a variation of a fish retention mechanism commonly used in cod-pot fishing. Research pots were fished individually with buoy lines, buoy con- figurations, and other tackle configured similarly to that deployed in commercial pot fishing. Experimental design We used an experimental design similar to the class of designs referred to as “before/after control/impact” (BACI). The capabilities and limitations of B AC I study designs have been discussed extensively in the ecological literature (e.g., Osenberg and Schmitt, 1994; Hewitt et al., 2001). We used many of the components of a BACI design, but the final design differed substantially from the paired designs of Stewart-Oaten et al. (1992). In our study, the measured quantity was not the difference between treatment and control stations replicated over time, but the percentage difference over time at each sta- tion, replicated over space. In this sense, our design was similar to the ANOVA-type designs of Underwood (1994). Our design allowed for expected seasonal and small- scale spatial variability in fish abundance, and provided the necessary replication for hypothesis testing. The experiment was conducted in 2003, 2004, and 2005. In each year, data were collected during two separate cruise legs: in early January (before the trawl- ing season) and immediately after the main trawl har- vest in late March. Sampling was balanced between a treatment area subject to intensive trawl fishing and a control area inside the Cape Sarichef notrawl zone. A total of 80 sampling stations were set along 10 parallel transects intersecting the notrawl zone boundary, and each transect contained four stations inside and four outside the boundary (Fig. 1). Although it was not pos- sible to match treatment and control stations exactly with respect to depth, habitat, bottom currents, etc., this layout provided a similar range of habitats within the two zones. The quantity of interest for each station was the ratio of the mean pot catch (in numbers or weight of Pacific cod) at that station during the March survey (XAfter) to the average catch during the January survey (XBefore). This ratio reflects the percentage change in abundance between the two surveys at a particular station. The seasonal percentage change is referred to as the delta (6) for the station: 5 The ratio is expressed so that <5 will be near zero if the catch is the same at a given station during both the before and after surveys, positive if the catch increases, _ After ' X Before J _ ^ After L Before x l. Before 284 Fishery Bulletin 106(3) and negative if catch decreases over the season. For example, a 8 of -0.50 represents a 50% decrease in catch, and a <3 of +1.00 represents a 100% increase. Sampling considerations and gear deployment There are two important concerns in using catch from fixed-gear such as pots as an index of fish abundance. The relationship between catch and abundance can break down at either very low or very high fish densities (Hubert, 1996). During our pilot studies, soak times and sampling procedures were developed to ensure that the research fishing did not result in either excessive num- bers of empty pots or gear saturation (where the number of fish in the pot became so high that catchability of additional fish was reduced). To guard against gear satu- ration and to provide detail on the timing of fish catch within the soak period, instruments referred to as trig- ger timers were developed. These instruments involved a magnetic reed switch mounted on the triggers of the pot and an electronic event timer that recorded when the triggers were pushed open. A modular trigger assembly allowed instruments to be mounted and dismounted in pots without slowing down the pace of fishing. In order to avoid gear saturation and to increase the number of observations, pot soak times were ini- tially kept short (4-8 h), and fishing was conducted mainly during daylight hours. It was not feasible to fully standardize the soak time or the timing of the launch within the diel and tidal cycles over the 40-50 pots fished each day. In order to compensate for varia- tion from these sources, each day’s sampling included approximately equal numbers of stations inside and outside the notrawl zone. Difficulty in retrieving pots during strong tidal currents led to a change in proce- dure between 2004 and 2005. In 2005 a slightly longer overnight soak was used, with pots being launched at new locations in the afternoon or evening and retrieved the next morning. Results of a pilot study conducted in 2002 showed short-term temporal variation (day-to-day variability in catch rates at a given station) as a larger component of variability than small-scale spatial variation (variabil- ity in catch between stations). In order to smooth over short-term temporal variation, we attempted to fish each station on at least three different days during each survey. Each day’s fishing was balanced with stations in both the treatment and control areas, so that any short-term influences on abundance would affect both treatment and control groups. The goal was to apply fishing methods in such a way that variation in catch due to soak time, diel and tidal cycles, weather, and current patterns would be minimized and distributed evenly between trawled and untrawled areas. Our pilot study results also indicated that pots lo- cated at least 0.11 km apart functioned as independent sampling units (no correlation between catch for pairs of pots at 0.11 km or further distances). Stations for the experiment were spaced 0.11 km apart within each zone (trawled or untrawled), and 1.8 km apart across the notrawl zone boundary. The same layout of study stations was used for all three years. Examination of both pilot study data and pot fishing data collected by fisheries observers indicated that there was not a strong relationship between length of pot soak and catch over a time span of 4-24 hours. For this reason, catch rates were expressed not as CPUE in fish/hour, but simply as total number or weight of Pacific cod caught per standardized pot deployment. The catch measure at each station was the average catch over all of the days that a station was fished during a survey. The use of an average over several days as the measure at each station provided smoothing over day-to-day variation and reduced the likelihood of zero catches in the final data set. Pots were baited with chopped Pacific herring ( Clupea pallasi) contained in meshed bait bags. Bait for each cruise was purchased as a bulk lot so that the same lot of bait was used for the entire cruise. Filled bait bags were weighed and the amount of bait adjusted to within 0.1 kg of the target weight of 5.0 kg. Procedures for se- curing bait bags and triggers, launching, and retrieving pots were consistent in all cruises. Upon retrieval, all catch was sorted, identified, and weighed. A systematic subsample of pots (every sec- ond, third, or fourth pot retrieved) was selected with a random starting point each day; all Pacific cod from selected pots were processed for length frequency, by sex. The sampling interval was adjusted according to average catch rates so that at least 100 fish were mea- sured each day. The condition of the gonad of measured fish was also examined visually and coded according to a 5-stage system, in order to record the approximate frequency of different stages of reproductive maturity of Pacific cod in the catch Data analysis After every study year, average catch rates for each station and cruise were calculated for all valid fishing days at a station. Average catches from the two cruises were used to compute the d for each station. Spatial mapping of both raw catch data and <3s was performed to look for spatial patterns and verify the assumption of independence between study stations. Distance-based correlograms (both anisotropic and isotropic) were plot- ted to check for spatial dependence in catch data. Linear models were also used to look for patterns in untransformed catch data; effects of year, season, treat- ment versus control, station, and fishing day within each cruise were examined. After examination of the distributional characteristics and independence of the 6’s, the nonparametric, rank-based Wilcoxon rank sum test (Ott, 1984) was used to test for a difference in dis- tribution of 8 between stations in the trawled and un- trawled areas. The nonparametric test was selected over the parametric t-test because 8 is a ratio of counts and may have a strongly non-normal statistical distribution. All modeling was conducted in S-Plus (Math Soft Inc., Seattle WA; Venables and Ripley, 2002). Conners and Munro: Effects of commercial fishing on local abundance of Gadus macrocephalus in the Bering Sea 285 Bootstrap methods (Manly, 1991) were used to evaluate the power of the experimental design to correctly identify the presence of a fishing effect. For each study year, daily catch data from all sta- tions and days were pooled for use as a sampling population. Random samples the same size as those used in each year’s experiment were drawn. Half of the simulated sample stations were randomly as- signed to the treatment group, and the mean catch for these stations in the March cruise was decreased by a fixed percentage to simulate a known fishing effect. Percentage changes were then calculated for each station, and we used the Wilcoxon rank sum test to compare stations in treatment and control groups. This algorithm was repeated 1000 times for each level of fishing effect. The P-values computed for each sample were used to compute probabilities of rejecting the null hypothesis for a = 0.05, 0.1, and 0.2 for each of the levels of sampling effort. To check that substantial removals of Pacific cod took place during the study, we evaluated the har- vest in the study area using NMFS catch estimates and haul-specific data collected by the Alaska Fisheries Science Center observer program. Winter-season Pacific cod catch for trawl gear was summarized for each of the two federal reporting areas that intersect the study area. These totals included both catches in the study area and catches in other parts of the reporting area. Observer data did not cover all hauls on all vessels but were recorded with spatial precision to the nearest minute latitude and longitude. Records for all observed hauls within the two reporting areas were extracted, and hauls within a one-degree longitude by 30-minute latitude block around the study area were identified. The proportion of observed hauls and total fishing effort (duration of haul) in the study area as a proportion of observed hauls in each reporting area was calculated. The total catch of Pacific cod was determined from the smaller subset of observed hauls (for which there were catch composition data), and the proportion of Pacific cod catch coming from the study area was also estimat- ed. This fraction was then applied to the total catch for the reporting area to obtain a rough estimate of Pacific cod harvested from the study area by gear type. Tagging studies In conjunction with the localized depletion study, we conducted studies on the feasibility of determining on Pacific cod movements through tagging. Partial results of these studies are reported here to explain the results of the main experiment. Tagging work included develop- ment of methods 1) for capturing and handling Pacific cod with pot gear, 2) for tagging fish and determining data formats, 3) for releasing Pacific cod tagged with archival data-storage tags and standard spaghetti tags, and 4) for studying preliminary tag-induced mortal- ity. The goal of this effort was to collect information on both seasonal migration patterns and small-scale movements of Pacific cod during the spawning season. Figure 2 Example of the frequency distribution of pot catch data for Pacific cod (Gadus macrocephalus) . Catch data are in numbers of fish per pot from all stations during the March 2004 cruise. An understanding of these movements is important for the interpretation of the local abundance study. During a special tagging cruise in February 2003, tagged fish were released at a series of locations, including within the notrawl zone off Cape Sarichef. Tag recapture was conducted entirely by the fishing industry. Date and location of the recaptures were provided by fishermen and fishery observers. In total, 295 archival tags and over 6000 spaghetti tags were released between 2002 and 2004. Approximately 35% of both types of tags were returned over the five years after tag release. Full results of these tagging studies will be presented in a separate publication. Results Field data confirmed that assumptions for the experi- mental design were met. The frequency distribution of the raw catch data (Fig. 2) illustrated the properties of pot catch. Similar results were obtained with either numbers or weight of Pacific cod per pot as the measured quantity. Zero catches (empty pots) were rare, and there was no evidence of gear saturation at high catches (the upper tail of Fig. 2 declines gradually without a sharp cutoff). Although the distribution of pot catch data was slightly skewed, the skewness and heteroscedasticity were much smaller than is typical for many types of fishery data. The level of Pacific cod catch varied strongly between study years and seasons but was fairly consistent within each study cruise. Average catch ranged from 8.8 Pacific cod per pot (27.7 kg) in March 2003 to 31.7 cod per pot (105.5 kg) in March 2004. Coefficients of variation for the raw catch data ranged from 64% for January 2004 to 42% for March 2005. Coefficients of variation for cruise averages at individual stations ranged from 14% to 43%. The average catch rate changed substantially between January and March in each year of the study, 286 Fishery Bulletin 106(3) 165°20'W 165°10'W 165°0'W Spatial pattern in seasonal percent change (6) for the three study years. Refer to Figure 1 for general location of the study area. Symbols show study stations. The arc in the middle of the stations is the Cape Sarichef notrawl zone boundary. Symbol shape and size indicate the size of the percent change in Pacific cod ( Gadus macrocephalus ) catch between “before” (early January) and “after” (immediately after the main trawl harvest in late March) surveys in each year: (A) 2003, (B ) 2004, and (C) 2005. but not always in the same direction. In 2003, the aver- age catch rate decreased 55% from January to March, and nearly all of the individual station changes (6) were less than 1.0. In 2004 and 2005, the average catch rate increased by 73% and 26%, respectively, from January to March. These differences were presumably the result of interannual differences in the timing of seasonal migration and spawning aggregation A check of spatial pattern in the raw data and final 6’s verified the independence of the experimental study stations and the absence of any strong spatial patterns. There was no consistent spatial pattern in raw catch data, either between trawled and untrawled areas or from southwest to northeast within the study area. There was some evidence of serial correlation between adjacent pots within each fishing day; we believe this correlation to be a result of similarities in timing of pot launches within tidal and diel cycles rather than spatial correlation in fish abundance. There was no spatial correlation in the calculated average catch at each station over a cruise. Although there was some evidence of serial correlation between days within a cruise, averaging the catch at each station over the days within a cruise eliminates this correlation. Maps of percentage change values (6) also showed no discern- ible spatial pattern (Fig. 3). Distance-based variograms (both isotropic or anisotropic) were reviewed for the final 6’s and we found no significant spatial correlation as a function of distance between pots. Correlation of pairs of 6’s at all distances of 0.11 km or more were not significant, verifying the assumption of independence between stations. Implementation of the study design was generally suc- cessful, and good replication was obtained. The excep- tion was the January cruise in the first year of the ex- periment (2003), where severe weather and mechanical problems severely curtailed the field effort. Sample size from this cruise (160 pots fished) was small (Table 1). The March “after” cruise in 2003 was, however, success- ful; a total of 475 pots were fished and full replication at all 80 experimental stations was obtained. Both before and after cruises in 2004 and 2005 achieved full cover- age of the 80 stations and good replication. Because of the very low sample size in January 2003, 6’s could be calculated for only 48 of the study stations, and some before values were based on only a single measure- ment. In 2004 and 2005, 6’s at all 80 study stations were calculated from the averages for over three to five replicate fishing days within each cruise. The resulting values ranged from negative numbers, representing a net seasonal decrease at a station, to fairly large posi- tive values, indicating a substantial seasonal increase. In 2004, there was a particularly large outlier (6>4) at one station which had very low catches in January. The shift to overnight soaks in 2005 resulted in not only smaller variance of the raw catch data but in smaller Conners and Munro: Effects of commercial fishing on local abundance of Gadus macrocephalus in the Bering Sea 287 1 65°20'W 1 65° 1 0’VV I65°0’W I ! | | I I 165°I0rW 1 65°0'W Figure 3 (continued) variance of the <5’s. Although the small sample size in January 2003 limited the power of statistical testing in this year, the overall consistency of the results persuad- ed us to include the 2003 data set in final analyses. Analysis of the raw data foreshadowed the overall results of the study. We examined the catch data from individual pots, using standard linear model analysis. The best-fit linear model accounted for only 30% of the variability in the data, reflecting the importance of oth- er sources of variability not included in the model. The treatment effect (trawled vs. untrawled zone) was not significant for the raw data, indicating only slight dif- ferences in baseline abundance between the two zones. The interaction term for treatment and season (which is the term that would reflect a substantial localized depletion in the trawled zone) was strongly not sig- nificant. As pointed out in discussions of simple BACI analysis (Hurlbert, 1984; Stewart-Oaten et al., 1986; Underwood, 1991), the interaction term from the stan- dard ANOVA does not provide the correct error term for a true test of impact; for testing the presence or absence of localized depletion we used the analysis of <5’s. Final results of the study clearly indicated very simi- lar values of seasonal change in Pacific cod abundance (d) in both the trawled and untrawled portions of the study area. We did not see the differences in slope that we would have expected to result from strong localized depletion in the trawled zone. When we used the rank- based Wilcoxon test to look for differences in mean d between the trawled and untrawled regions, P-values 288 Fishery Bulletin 106(3) Table 1 Dates and summary statistics for cruises conducted during the Bering Sea localized depletion experiment for Pacific cod ( Gadus macrocephalus ). Each year’s experiment included a “before” cruise in early January (before the trawling season) and an “after” cruise in late March immediately after the main trawl harvest. Cruise dates Cruise purpose Number of days fishing Number of pots fished Average number of cod/pot Average weight (kg) of cod/pot 30 Mar-25 Apr 2002 Pilot study and tagging 21 703 28.8 103.4 28 Dec 2002-8 Jan 2003 Abundance experiment — before 4 160 22.2 104.4 4-17 Feb 2003 Tagging 11 336 22.3 78.3 12-31 Mar 2003 Abundance experiment — after 14 475 8.8 27.7 2-10 Jan 2004 Abundance experiment — before 9 360 19.3 93.3 15-31Mar 2004 Abundance experiment — after 15 604 31.7 105.5 9-22 Mar 2005 Abundance experiment — before 14 481 21.2 85.3 16-29 Mar 2005 Abundance experiment — after 14 500 25.7 95.4 for the three study years were 0.70, 0.92, and 0.81, respectively. Although the range and mode of the ob- served d’s changed from year to year, in each year the distribution of 5’s over stations within the two zones was very similar (Fig. 4). Power simulations gave us confidence that the strong failure to reject the null hypothesis in 2004 and 2005 reflects a true absence of a treatment effect in the study area. Power was poor in 2003 because of the low sample size and higher variability of the data; only imposed fishing effects of 50% or more would have given a high probability of correctly detecting a difference between treatment and control groups. In 2004 and 2005, how- ever, full replication of the experiment resulted in good power. We were able to detect differences at 25-30% fishing effects in the 2004 experiment and at as low as 20% fishing effects in 2005. We verified that substantial fishing removals occurred within the trawled portion of the study area during our experiment. The reported January-March fish harvest data from the NMFS Alaska Regional Office indicated that the harvest of Pacific cod by bottom trawl gear in the two federal reporting areas around Cape Sarichef was on the order of 25,000 metric tons (t) per year. Based on observed hauls, approximately 45% of the total harvest in these two reporting areas came from the study area (Table 2). Auxiliary biological and tagging data were useful for interpreting our results. Sex ratio and length-frequency data collected during each cruise indicated that the population of fish sampled sometimes changed substan- tially, even within the two-week duration of a study cruise. Sex and maturity class data from systematic subsamples of Pacific cod collected during the March 2004 cruise are shown in Figure 5. Although shifts from developing and prespawning stages to ripe and spent stages may reflect seasonal maturation of indi- vidual fish, differences in the proportion of mature to immature fish can only be explained by movement of Table 2 Estimation of the Pacific cod ( Gadus macrocephalus) har- vest from the study area. Total Pacific cod catch during the winter trawl season in each year is shown for National Marine Fisheries Service reporting areas 509 and 517, which included the study area. Catch data from observed hauls were used to estimate the proportion of cod catch taken from a 1° latitude-longitude block around the study area; this proportion applied to the total harvest was used to show the approximate harvest (in metric tons) from the study area during the experiment. Estimated harvest (metric tons) taken with bottom trawl gear Year Reporting areas Study area Proportion 2003 20,971 10,215 48.7% 2004 25,158 12,295 48.9% 2005 29,870 13,882 46.5% fish into and out of the study area during the two-week period of the cruise. The overall sex ratio for the 15-22 March samples was 0.93 (males/female), whereas the sex ratio for 23-20 March was 0.75. This difference represents a significant change in sex ratio (P=0.0004) over the two-week period. Qualitative tagging studies were conducted concur- rently with the localized depletion experiment. Although these studies are to be reported separately, some of their results help to explain the outcome of the localized depletion experiment. Partial results from the February 2003 tagging studies indicated substantial movement of Pacific cod, both over spans of several months and over shorter time scales (Table 3). Over 70% of the fish released in the study area in February 2003 and recovered within two weeks of release were recaptured Conners and Munro: Effects of commercial fishing on local abundance of Gadus macrocephalus in the Bering Sea 289 more than 18.5 km (10 nmi, the radius of the notrawl zone) from their release loca- tion. The majority of fish recaptures took place east and slightly north of the release site within the Cape Sarichef notrawl zone, and only a few recaptures were documented for the study area immediately outside the notrawl zone. Discussion The results of our experiment were ex- tremely clear. Although the direction and magnitude of the net seasonal change in abundance differed between study years, in each study year direction and magnitude of the percentage change (<5) were similar in the trawled and untrawled areas. Nonpara- metric tests used to compare <3’s from the two zones consistently had P-values over 70%, indicating that there was no evidence that the distribution of the two groups dif- fered— a conclusion that is evident without any statistical testing simply by compar- ing the frequency distributions of the <5’s in the trawled and untrawled zones (Fig. 4). Analysis of the raw catch data with linear models leads to the same conclusion. Checks for more subtle indications of fishing effects, such as spatial patterns in catch within the study region or temporal trends within each two-week cruise, were also negative. These experimental results are inconsistent with the hypothesis of strong stationary localized depletion at the scale of the existing notrawl zones in Alaska. Power simulations indicated that the fail- ure to see a difference between trawled and untrawled areas was not simply due to a lack of resolution in the data. Although the sample size for 2003 was low, results for 2004 and 2005 showed that fishing re- movals that resulted in a 20-30% decline in catch rates in the trawled area would have been detected. There is little infor- mation on the size and duration of prey density decreases that would be necessary to impact Steller sea lion foraging success. It is possible that Pacific cod abundance in the study area was so high that even the substantial fishing removals resulted in a <20% change in local Pacific cod abun- dance. If this is the case, then the question becomes one of whether or not such small changes in prey availability would signifi- cantly affect Steller sea lion foraging. Data from the auxiliary tagging and biological ies clearly indicated that Pacific cod in our study were highly mobile over much shorter time scales A 2003 Trawled Untrawled -100% -25% -50% -125% -200% -500% C 2005 Figure 4 Frequency distribution of seasonal percent change in abundance (d) for the Cape Sarichef localized depletion study. Dark bars are frequen- cies of 6 for untrawled stations inside the notrawl zone, and hatched bars represent trawled stations outside the notrawl zone boundary: (A) 2003, (B) 2004, and (C) 2005. P-values for the nonparametric rank-sum test are shown in each figure. previously assumed. The stationary localized depletion stud- scenario is based on the assumption of a closed local area pool of fish that is reduced by local removals. Both the than tagging data and the observed shifts in maturity and 290 Fishery Bulletin 106(3) A Females — early March Immature 17% Spent 45% Prespawning 11% Spawning 27% C Females — late March Immature 6% Spent 78% Prespawning 3% Spawning 13% B Males — early March Spent 38% Immature 24% Spawning Prespawning 37% D Males — late March Immature 12% Prespawning 20% Spent 68% Spawning 0% Figure 5 Gonad maturity stages for Pacific cod ( Gadus macrocephalus) caught in March 2004. Fish were classified as immature, prespawning, spawning, or spent (immediately postspawning) based on visual inspection of the gonad. (A) females in randomly sampled catch from 15 through 22 March 2004, (B) males from this period, (C) females in randomly sampled catch from the same locations in 23-30 March 2004, and (D) males from this period. sex ratio indicate that the local population of Pacific cod in our study area was not a closed, static pool of fish but a shifting, dynamic mix. In fact, the tagging data indicate that the short-term movement scale for Pacific cod in this region is substantially larger than the 10 nmi notrawl zone. Although fishing removal may have had an immediate localized effect on fish abundance, the effect was obscured by rapid fish movement (less than one week) over a geographic scale greater than that of the fishery removal. Thus, of the three localized depletion scenarios presented earlier (see Introduction), our results strongly disagree with conjecture one (sta- tionary localized depletion). Because the experiment was limited in scale, our results do not eliminate the possibility of conjectures two and three (regional ef- fects over larger spatial scales or displaced effects due to directed movement). Recent hydroacoustic studies of walleye pollock ( Theragra chalcogramma ) abundance in the same region (Barbeaux et al.1) also indicated at- tenuation of fishery-removal effects by rapid fish move- ment. Barbeaux et al. saw a visible pattern in echo sign during fishing, but diurnal fish movements eliminated the pattern within 12-24 hours. The movement of Pacific cod, as qualitatively observed in the tagging data, could appear as a decline in catch rates for catch data measured on the same geographic scale. In a previous study of possible effects of commer- cial fishing on Pacific cod abundance (Fritz and Brown, 2005), commercial catch-per-unit-of-effort (CPUE) data were collected from a large region of the southeast Ber- ing Sea, including our study area. Fritz and Brown interpreted a decreasing trend in fishery CPUE from 1 Barbeaux, S. J., M. Dorn, J. Ianelli, and J. Horne. 2005. Visualizing Alaska pollock (Theragra chalcogramma) ag- gregation dynamics. ICES Council Meeting 2005/U:01. Conners and Munro: Effects of commercial fishing on local abundance of Gadus macrocephalus in the Bering Sea 291 Table 3 Partial results from tagged Pacific cod ( Gadus macrocephalus ) released near Cape Sarichef in February 2003, showing net move- ment of tagged fish. Table columns show distance travelled (km) between release and recapture points of a tagged fish. Table rows show weeks at liberty (the time elapsed between release and recapture dates). Table values show the percentage of recovered tags for which the distance travelled is within the specified range. Distance (km) Number of Weeks at liberty <9.3 9.3-18.5 18.5-37.0 37.0-74.1 74.1-148.2 >148.2 tags recovered <2 12.2% 15.1% 25.9% 18.0% 3.6% 25.2% 139 2-4 7.0% 2.3% 9.3% 34.9% 23.3% 23.3% 43 OO 1 2.9% 5.9% 5.9% 26.5% 32.4% 26.5% 34 8-16 18.8% 12.5% 6.3% 25.0% 6.3% 31.3% 16 13 February through 24 March 2001 as localized deple- tion due to fishing removals. The models used in the Fritz and Brown study (the models of Leslie and Davis, 1939, and DeLury, 1947) are based on the assumptions of a closed population and constant catchability. These models are unable to distinguish between changes in abundance due to fishing removals and those due to fish dispersal or movement across the boundaries of the study area. If the assumption of a closed popula- tion is true, then declining CPUE would indicate a regional-scale effect (conjecture two). Given the high mobility indicated in Table 3, however, we doubt that the model assumptions are met for Pacific cod. Under- standing patterns in Pacific cod abundance must take into account both substantial short-term movement and seasonal processes of migration and aggregation related to spawning. The formal statistical inference presented in the pres- ent study applies only to the study area; extending this inference to other areas is reasonable but can only be considered a qualitative exercise. As with many com- parative environmental studies, this project included only two experimental units, in the sense that the treat- ment (trawling) was applied to one region and the other region (notrawl zone) was used as a control. Hurlbert (1984) pointed out that, in an observational study with only one treatment and one control area, a statistical test constitutes evidence only for a difference between the two observed areas. The observation of an effect (or lack of effect) must be combined with biological knowledge of the system to extend inference from the observed areas to other parts of the system. In the case of Cape Sarichef, the experimental area was selected not as a representative location for the entire Bering Sea, but as the location where trawl fishing was most intensive and most likely to produce measurable local effects. Qualitative inference to other areas will require consideration of similarities in fishing pressure, Pacific cod behavior, and Pacific cod movement. Localized depletion has not been widely discussed in the scientific literature. It has been proposed as a mechanism primarily in coral and sedentary benthic species (Gorfine et al., 2001; Jamieson, 2001; Harriott, 2003; Smith et al., 2004). Our results demonstrate that the mobility of the target organism must be considered in looking for localized spatial effects on groundfish. If localized depletion is to occur, it will result from the interaction of fishing pressure, fish abundance, and fish movement. To evaluate the impacts of fishery removals on other predators, such as Steller sea lions, the relevant scales of fishing, fish movement, and preda- tor feeding must be clearly defined and understood. For Pacific cod, the very small spatial scale associated with the current regulatory notrawl zones appears to be smaller than the relevant scale of fish movement. Potential fishery effects at broader spatial and temporal scales may be more appropriately addressed by continu- ing to manage seasonal and spatial dispersal of the Pacific cod harvest. Acknowledgments Funding for this project was provided by the National Oceanic and Atmospheric Administration, National Marine Fisheries Service (NMFS). S. Neidetcher of the Alaska Fisheries Science Center (AFSC) and O. Ormseth of the University of Alaska were instrumental in organiz- ing and conducting field work for this project. We thank the captains and crews of the charter vessels who partic- ipated in this research and the many NMFS employees and associates who braved the Bering Sea in winter to help us perform the experiment. We thank L. Fritz, W. Stockhausen, and D. Somerton of the AFSC for prelimi- nary reviews of the manuscript. Three anonymous review- ers also contributed improvements to the manuscript. Literature cited Alverson, D. L. 1992. A review of commercial fisheries and the Steller sea lion Eumetopias jubatus: The conflict arena. Rev. Aquat. Sci. 63(3-4)203-256. 292 Fishery Bulletin 106(3) Braham, H. W., R. D. Everitt, and D. J. Rugh. 1980. Northern sea lion population decline in the eastern Aleutian Islands. J. Wildl. Manag. 44:25-3. DeLury, D. B. 1947. On the estimation of biological populations. Bio- metrics 3:145-167. Fritz, L. W., R. C. Ferrero, and R. J. Berg. 1995. The threatened status of Setller sea lions, Eumeto- pias jubatus, under the Endangered Species Act: Effects on Alaska groundfish fisheries management. Mar. Fish. Rev. 57(2):14-27. Fritz, L. W., and E. S. Brown. 2005. Survey- and fishery-derived estimates of Pacific cod ( Gadus macrocephalus) biomass: implications for strate- gies to reduce interactions between groundfish fisheries and Steller sea lions ( Eumetopias jubatus). Fish. Bull. 103(3):501-515. Gorfine, H. K„ B. L. Taylor, and T. I. Walker. 2001. Triggers and targets: What are we aiming for with abalone fisheries models in Australia. J. Shellfish Res. 20(2):803-811. Harriott, V. J. 2003. Can corals be harvested sustainably? AMBIO 32(2):130-133. Hewitt, J. E., S. E. Thrush, and V. J. Cummings. 2001. Assessing environmental impacts: effects of spatial and temporal variability at likely impact scales. Ecol. Appl. 11(5):1502— 1516. Hiatt, T. 2005. Stock assessment and fishery evaluation report for the groundfish fisheries of the Gulf of Alaska and Bering Sea/Aleutian Islands area: Economic status of the groundfish fisheries off Alaska, 2004, 182 p. North Pacific Fishery Management Council, 605 W. 4th Ave., Anchorage, AK 99510. Hubert, W. A. 1996. Passive capture techniques. In Fisheries tech- niques, 2nd ed. (B. R. Murphy, and D. W. Willis, eds.), p. 157-192. Am. Fish. Soc., Bethesda, MD. Hurlbert, S. H. 1984. Pseudoreplication and the design of ecological field experiments. Ecol. Monogr. 54(2):187-211. Jamieson, G. S. 2001. Review of the status of the northern abalone, Haliotis kamtschatkana, in Canada. Can. Field-Nat. 115(4):555-563. Leslie, P. H., and D. H. S. Davis. 1939. An attempt to determine the absolute number of rats on a given area. J. Anim. Ecol. 8:94-113. Manly, B. F. J. 1991. Randomization, bootstrap, and Monte Carlo methods in biology, 91 p. Chapman and Hall, London, U.K. 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. Murawski, S. A. 2000. Definitions of overfishing from an ecosystem perspective. ICES J. Mar. Sci. 57:649-658. Nichol, D. G., and E. A. Chilton. 2006. Recuperation and behavior of Pacific cod after barotrauma. ICES J. Mar. Sci. 63:83-94. Osenberg, C. S., and R. J. Schmitt. 1994. Detecting human impacts in marine habitats. Ecol. Appl. 4(1): 1—2 . Ott, R. L. 1984. An introduction to statistical methods and data analysis, 2nd ed., p 146-150. Duxbury Press, Belmont, CA. Schoener, T. W. 1993. On the relative importance of direct versus indirect effects in ecological communities. In Mutualism and community organization (H. Kawanabe, and J. E. Cohen, eds.), p. 365-415. Oxford Univ. Press, Oxford, U.K. Shimada, A. M., and D. K. Kimura. 1994. Seasonal movements of Pacific cod, Gadus macro- cephalus, in the eastern Bering Sea and adjacent waters based on tag-recapture data. Fish. Bull. 92:800- 816. Smith, P. J., S. M. McVeagh, J. T. Mingoia, and S. C. France. 2004. Mitochondrial DNA sequence variation in deep- sea bamboo coral (Keratoisidinae) species in the southwest and northwest Pacific Ocean. Mar. Biol. 144(2):253-261. Stewart- Oaten, A. S., W. W. Murdoch, and K. R. Parker. 1986. Environmental impact assessment: “pseudoreplica- tion” in time? Ecology 67(4):929-940. Stewart-Oaten, A. S., J. R. Bence, and C. W. Osenberg. 1992. Assessing effects of unreplicated perturbations: no simple solutions. Ecology 73(4):1396-1404. Thompson, G. G., and M. W. Dorn. 2005. Assessment of the Pacific Cod stock in the eastern Bering Sea and Aleutian Islands area. In Stock assess- ment and fishery evaluation report for the groundfish fisheries of the Gulf of Alaska and Bering Sea/Aleutian Islands area 2004, p. 219-330. North Pacific Fishery Management Council, 605 W. 4th Ave., Anchorage, AK 99510. Underwood, A. J. 1991. Beyond BACI — experimental designs for detecting human environmental impacts on temporal variations in natural populations. Aust. J. Mar. Freshw. Res. 42(5):569-587. Underwood, A. J. 1994. On beyond BACI — sampling designs that might reliably detect environmental disturbances. Ecol. Appl. 4(1):3-15. Venables, W. N., and B. D. Ripley. 2002. Modern applied statistics with S, 4th ed., 495 p. Springer-Verlag, New York, NY. 293 Abstract — Defining types of seafloor substrate and relating them to the distribution of fish and invertebrates is an important but difficult goal. An examination of the processing steps of a commercial acoustics analyzing software program, as well as the data values produced by the proprietary first echo measurements, revealed potential benefits and drawbacks for distinguishing acoustically dis- tinct seafloor substrates. The positive aspects were convenient processing steps such as gain adjustment, accu- rate bottom picking, ease of bad data exclusion, and the ability to average across successive pings in order to increase the signal-to-noise ratio. A noteworthy drawback with the pro- cessing was the potential for acciden- tal inclusion of a second echo as if it were part of the first echo. Detailed examination of the echogram mea- surements quantified the amount of collinearity, revealed the lack of standardization (subtraction of mean, division by standard deviation) before principal components analysis (PCA), and showed correlations of individual echogram measurements with depth and seafloor slope. Despite the facil- ity of the software, these previously unknown processing pitfalls and echo- gram measurement characteristics may have created data artifacts that generated user-derived substrate clas- sifications, rather than actual sea- floor substrate types. Manuscript submitted 4 February 2008. Manuscript accepted 28 March 2008. Fish. Bull. 106:293-304 2008). The views and opinions expressed or implied in this article are those of the author and do not necessarily reflect the position of the National Marine Fisheries Service, NOAA. Comparison of echogram measurements against data expectations and assumptions for distinguishing seafloor substrates Mark Zimmermann (contact author) Christopher N. Rooper Email address for M Zimmermann: Mark.Zimmermann@noaa.gov National Marine Fisheries Service Alaska Fisheries Science Center 7600 Sand Point Way NE, Bldg. 4 Seattle, Washington 981 15-6349 Marine natural resource managers must define essential fish habitat (EFH) for federally managed, com- mercially exploited species (Federal Register, 2002) but the best method for fulfilling this mandate across the vast area and significant depths of the U.S. Exclusive Economic Zone remains unknown. A successful acous- tic method for determining EFH would be of great benefit, because single- beam seafloor echosounder reflections are collected simultaneously with fish density estimates during National Marine Fisheries Service (NMFS) stock assessment bottom trawl sur- veys in the Gulf of Alaska (~800 sta- tions among 320,000 km2, <1000 m depth) and the Aleutian Islands (-400 stations among 67,000 km2, <500 m depth). Therefore we conducted an acoustic analysis on data from a small portion from one survey in order to determine if there was a direct cor- relation between substrate classes or echogram measurements with species abundance. We tested a widely used, propri- etary software package (vers. 3.30, QTC IMPACT™), developed by the Quester Tangent Corporation (QTC, Sidney, British Columbia, Canada), to resolve the echosounder reflec- tions into substrate types for com- parison with the survey trawl catch data to determine whether there was a correlation or relationship between seafloor substrate classes and fish- density. This software produces 166 proprietary unitless echogram mea- surements (EMs) on the first seafloor echo for an internal principal compo- nents analysis (PCA), and then uses the first three principal components (PCs), generally accounting for more than 95% of the covariance (Ellingsen et al., 2002; Legendre et al., 2002) in A-means clustering, for dividing the first seafloor echoes into acoustically distinct substrate types. Our initial efforts with Tf-means clustering indicated that a solution of any particular number of classes was not much better than other solu- tions (e.g., four versus five substrate classes), and therefore the 166 EMs were analyzed to determine if they could be used in another analysis for resolving substrate types. Although the general manner in which the 166 EMs, or the data, are acquired, pro- cessed, and divided into substrate classes by QTC software has been well reported in the literature, many specific details are lacking and it was therefore not clear what these 166 EMs represent. To investigate an acoustic method for determining EFH we described the specific details of the processing method that QTC software follows, focusing on potential pitfalls and ad- vantages for the user. We report on new findings based on some simple data explorations on the 166 EMs from echosounder data collected dur- ing a 2003 NMFS research cruise; and our findings are corroborated with four data sets collected indepen- dently from other agencies on other ships. In this analysis we checked the assumption that these 166 EMs have the same scale or range as that normally used in PCA, and the as- 294 Fishery Bulletin 106(3) sumption that these 166 EMs are derived from the first echo only. Because both of these assumptions are typi- cally presumed to be correct for this type of acoustic analysis, these findings may be of use for interpret- ing seafloor substrate classifications for determining EFH. Materials and methods Data collection and conversion Data were collected in the *.raw format from a 38-kHz Simrad single-beam echosounder on the FV Gladiator during the 2003 NMFS bottom trawl survey in the Gulf of Alaska (Table 1). The transducer gain was 24.5 dB, transmit power was 1500 W, beam angle was 9°, pulse length was 4.096 ms, and the sampling interval was 1.024 ms. These Simrad files were calibrated in Echo- View® (vers. 3.30.60.05, SonarData Pty. Ltd, Hobart, Tasmania, Australia), and short (~1.5 km, -1440 pings) seafloor sections corresponding to 15-minute dura- tion bottom tows conducted at 1.54 m/s (3 knots) were exported into binary files by using the Echolmpact export module for import into QTC IMPACT™. This Echolmpact export module was specifically designed by the two companies to convey acoustic data in an appropriate format from EchoView to QTC IMPACT. We also examined EMs recorded directly by QTC VIEW™ (QTC, Sidney, British Columbia, Canada), without any prior EchoView® processing, at preset gains (ping inten- sities or amplitudes) by four external research cruises: the Alaska Department of Fish and Game (ADFG RV Resolution 2003 cruise), the Canadian Department of Fisheries and Oceans (Canadian coast guard ship RV John P. Tully 2002 cruise, and the RV Pallasi 2004 cruise), and the New Zealand National Institute of Water and Atmosphere (RV Rangithi 1999 cruise) (Table 1). Gain settings In automated seafloor echo-processing systems, there may be a mismatch between the seafloor echo strength (gain) and the ability of the processing system to iden- tify the abrupt rise or spike that represents the begin- ning of the seafloor reflection. It is necessary to adjust the gain setting such that the inflection point can be distinguished from the earlier portion of the echo, which is the water column above the seafloor. Therefore sev- eral postprocessing gain settings in QTC IMPACT were applied to subsets of the NMFS 2003 FV Gladiator data sets in order to maximize the number of pings strong enough for automatic bottom detection (or bottom pick- ing) and to minimize the number of pings that would be too strong for the dynamic range of 96 dB of sound that QTC software can process. Otherwise, louder portions of pings would have had to have been automatically decreased to 0 dB, a process known as clipping, and quieter portions of pings would have had to be automati- Zimmermann and Rooper: Comparison of echogram measurements for distinguishing seafloor substrates 295 cally increased to -96 dB. Another convenience of the QTC software was that abnormally weak echoes could be eliminated by specifying a minimal signal strength, and this was set to be equal to 25% of the maximum permissible amount (0 dB). Bottom picks After importing the recorded echoes into the software at an appropriate gain setting, another method for further improving the signal-to-noise ratio would be to assemble a stack of successive echoes, presumably from the same substrate, and average the echo stack into a single echo (Pace and Ceen, 1982). For QTC IMPACT, a minimum stack size of five pings was recommended, which corre- sponded to 2.5 seconds or 3.85 m traveled at 1.54 m/s, for a theoretical yield of 288 stacks per trawl path. The strong, positive benefits of stacking were dependent on correctly aligning events with the successive echoes, and therefore on the software’s interpretation of the seafloor inflection point. Although this data check is not men- tioned in the literature, it is a critical part of the process, because all measurements start at the seafloor inflection point. We examined every bottom pick for appropriate placement, as recommended by QTC guidelines. This process determined that the bottom pick was not inter- polated between sample intervals, and that the natural variability of the depth among a group of pings would be the distance sound travels during half the sampling interval (0.768 m). Generating echogram measurements Once the bottom pick had been located, an automatic determination of the length or extent of any seafloor echo was difficult because rough, steep, soft, and deep areas have longer reflections than smooth, flat, hard, and shallow areas. The QTC IMPACT software uses 256 sound samples of vertical time intervals, or recorded sound intensity within a ping, surrounding the bottom pick. Starting at the bottom pick, five samples (repre- senting the water column) were taken above the seafloor inflection point and 251 samples were taken below the start of the first seafloor reflection (representing the seafloor). If the echograms contained fewer than 251 time intervals below the start of the first seafloor reflec- tion, the last sample was repeated (padded) as many times as needed until the 251 sample requirement was fulfilled. The QTC software then generated 166 EMs for each stack with reference to a specific depth such that depth-related changes in signal protraction were corrected. Optimum substrate classification Organizing the echogram measurements along a contin- uum of measurements or grouping them into a number of acoustically distinct substrate classes is the final step in the process. Ideally this step would identify sub- strate qualities of importance to EFH species, such that researchers could infer essential fish habitat from sub- strate types and use this information for better resource management. The QTC method first uses continua by performing PCA on the 166 EMs and retaining the first three PCs for plotting the location of each stack in three- dimensional space. Then it is up to the user to determine the optimum number of substrate classes on the basis of the A-means clustering of the first three PCs. Examination of the data Because the algorithms for producing the 166 EMs are proprietary, the data values produced by the 166 EMs were exported and viewed in a text editor, which showed that the data values for each stack were displayed as seven decimal-place numbers in four columns, under- neath a stack header. In order to resolve the possible complications of having a single set of 166 EMs for five different stacked pings, a single ping was exported from EchoView® and imported into QTC IMPACT five times, to create one stack of identical pings. The four columns of 166 EMs were reformatted into a single column in a spreadsheet and an examination of the data revealed that the EMs from this single, repeated ping were occur- ring in five groups (von Szalay, 1998). Variability and covariance of echogram measurements Simple data checks, such as checks of averages, vari- ances, minima, and maxima, enabled us to describe each data set and determine the range or scale of each EM. The variance between EMs, or the covariance, was derived to determine the amount of collinearity among the EMs. Correlation of echogram measurements with depth The correlation between each EM versus depth was determined from each of the data sets. This simple analysis, which could provide some useful diagnostics, has not been reported in any of the literature. Angle of incidence The angle of echosounder seafloor reflections has a potentially confounding influence on any depth-cor- relation analysis, because the rate of change of depth and slope vary together. In general, QTC and similar products should be used to analyze normal (90°) incident reflections (see Pace and Ceen, 1982; Orlowski, 1984), and it is expected that severe departures from normality would cause analytical failures. The influence of non- normal (<90°) reflections could not be formally exam- ined in our study because of a lack of knowledge about cross-track slope, vessel pitch and roll, and interactions between seafloor angle and vessel angle. However, more single-beam data were analyzed from the FV Gladiator in 2005 at small study sites in the Aleutian Islands that had been groundtruthed with video and multibeam sonar equipment in 2004, such that the substrate types 296 Fishery Bulletin 106(3) and seafloor slopes were known (Rooper and Zimmer- mann, 2007). Assumption 1: scale of measurements Although as a group the 166 EMs ranged between zero and one (Legendre et al., 2002), this was tested for each of the 166 EMs; and EMs fully extending across this range would indicate that the data had been standard- ized for proper PCA (Manly, 1994; Legendre and Legen- dre, 1998). PCA was also conducted (S-Plus, vers. 6.1, Insightful Corp., Seattle, WA) independently to ensure that the QTC PCA results could be reproduced. Assumption two: first echo Although it is widely reported or implied in the lit- erature that QTC IMPACT software analyzes only the first echo, that conclusion is not strictly correct. QTC IMPACT analyzes the first 251 sound samples beneath the bottom pick, and it is up to the user to ensure that this is a meaningful window. The relationship between the 251 samples and the analysis depth range is directly related to half of the sample interval preset in the echosounder; Analysis depth range (m) = Sample interval (s/sample) x (0.5) x (1500 m/s) x 251 samples, where 1500 m/s = the approximate speed of sound through seawater. This relationship was tested to determine how well the 251 sample size corresponded with the first echo in the NMFS 2003 FV Gladiator data sets and with the external data sets collected by other agencies from other vessels. Results Gain settings Determining the proper gain setting for the NMFS 2003 FV Gladiator data sets was a time-intensive process because a wide range of gains needed to be applied and weak or bad data had not yet been identified. We deter- mined that a gain setting of -18 dB would be appropriate for shallow sites (25-100 m, 68 sites, 97,119 pings) and a gain setting of -17 dB would be appropriate for deep trawl sites (100-200 m, 19 sites, 25,110 pings). Several additional sites with too many weak pings (>50%), an indication of bad data, were identified and eliminated from processing at this stage. Although changing the gain setting within QTC IMPACT by 0.5 dB was equiva- lent to changing the gain setting by 1.0 dB in Echo- View®, use of the gain adjustment within QTC IMPACT was far more convenient for adjusting the echo signal strength to be within the required 96-dB range, and for identifying bad data. Bottom picks In the NMFS 2003 FV Gladiator data sets, there were 72,296 shallow (25-100 m) pings with bottom picks including 3961 that were clipped, and there were 18,021 deep (100-200 m) pings with bottom picks including 1106 of those that were clipped. Thus there was a greater than 70% success rate in bottom picking and approxi- mately 6% clipping among both data sets, indicating that the gain settings were appropriate. Each bottom pick was inspected and found to occur anywhere between the base and the tip of the peak — in the general region of the seafloor location. Thus QTC IMPACT software did an excellent job of bottom picking in the NMFS 2003 FV Gladiator shallow and deep data sets. Generating echogram measurements The NMFS 2003 FV Gladiator shallow data set yielded 14,432 stacks (of five pings) and the deep data set yielded 3598 stacks (of five pings); odd lots of fewer than five pings were not included in stacks, and there- fore the total number of stacks was slightly less than one fifth of the total number of pings with bottom picks. Padding was required at all shallow sites for all of the stacks, and padding was required at 16 of 19 deep sites on a total of 3112 stacks. A reference depth of 50 m was used for the shallow sites and 150 m was used for the deep sites. The EMs for the shallow sites were combined into a single data set for PCA and A-means clustering. The process was repeated for the EMs from the deep sites. Optimum substrate classification The A'-means clustering of the first three PCs indicated that a solution of any specific number of acoustically derived substrate classes would not explain much more of the variance than other solutions. Therefore the data processing was repeated several times to check for errors that may have influenced the results. The main focus was on gaining a better understanding of the numbers that were being created and processed with PCA and A-means clustering, and on exploring factors that may have affected the EMs. Examination of the data Examinations of the spreadsheets of EMs from the NMFS 2003 FV Gladiator shallow and deep data sets, and the four externally collected data sets, showed the same groups as those revealed by the examination of the stack of the single pings repeated five times; EMs 1-23, EMs 24-39, EMs 40-70, EMs 71-101, and EMs 102-166. Across all data sets, the EMs in the first (EMs 1-23) and fifth (EMs 102-166) groups were, in general, highly correlated with their neighbors (e.g., EM 22 versus 23, Fig. 1). In the second group of EMs (EMs 24-39), EM 31 was the sum of EM 32 through EM 39, each of which were fractions of 256 (e.g., 1/256, 2/256). Among the 31 Zimmermann and Rooper: Comparison of echogram measurements for distinguishing seafloor substrates 297 FV Gladiator 2003 shallow FV Gladiator 2003 deep RV Resolution 2003 0.90 Echogram measurement 22 RV Pallasi 2004 f.OO 0.90 0.80 0.70 0.60 0.50 0.40 0.30 0 20 0.10 0.10 0.30 0.50 0.70 0.90 RV John P. Tully 2002 Echogram measurement 22 Figure 1 Significant linear regressions (P<0.0005) between the unitless Quester Tangent Corpora- tion echogram measurements 22 and 23 for the shallow (n = 14,432) and deep (n = 3598) FV Gladiator 2003 data sets, the RV Resolution 2003 cruise 02=3680), the C.C.G.S. RV John P. Tully 2002 cruise (n= 727), the RV Pallasi 2004 cruise (n = 736), and the RV Rangithi 1999 cruise (n = 736). Note the different scales on the x axes. EMs in each of the third (EMs 40-70) and fourth (EMs 71-101) groups, there were 16 original EMs grouped into 15 succeeding sums, with the final EM (EM 40 in the third group and EM 71 in the fourth group) being the sum of the whole group (Fig. 2). In the third group, EM 40, which was always 1.0, was created by the sum of EM 41 and EM 56, which are complements to each other and therefore are entirely dependent. Variability and covariance of echogram measurements There were several unusual observations of variance or covariance among the 166 QTC IMPACT-generated EMs. Three EMs — 16, 31, and 40 — never varied (were always 1.0000000 or 0.9999999) and it is presumed these are the same three that Legendre et al. (2002) stated never varied from one. Additionally, three EMs from the RV Rangithi 1999 cruise data were always zero. Although it is difficult to describe all the depen- dent relationships (the sums and correlations) between the remaining 160 to 163 EMs, it is much simpler to note that among the six data sets, a range of 148 to 155 EMs (Table 1) were fully collinear (causing the variance-covariance matrix determinant to be zero; Neter et al., 1990). Among the eight to 12 EMs within each data set that were not fully collinear, four to 298 Fishery Bulletin 106(3) Initial sums Secondary Tertiary Final sums sums sum 70 + 69 = 68 67 + 66 = 65 68 + 65 = 64 63 + 62 = 61 60 + 59 = 58 61 + 58 = 57 64 + 57 = 56 55 + 54 = 53 52 + 51 =50 53 + 50 = 49 48 + 47 = 46 45 + 44 = 43 46 + 43 = 42 49 + 42 = 41 56 + 41 =40= 1.000000 Figure 2 Schematic diagram depicting a sequence (from left to right) of 15 sums from 16 original unitless echogram measurements — a sequence of sums that leads to a final sum of one. The left column shows how 16 original echogram measurements are summed into eight new echogram mea- surements. The next column shows how these eight sums are summed into four new echogram measurements. The third column shows how these four sums are summed into two new echogram measurements. The fourth column shows how these two sums are summed to produce a final echogram measurement of one. seven EMs had variance inflation factors >10 (a gen- eral threshold indicating high correlations but not full collinearity with the remaining variables; Neter et al., 1990), leaving only three to six relatively independent EMs in each data set. Correlation of echogram measurements with depth There was a significant relationship between depth and some EMs in all data sets (Fig. 3). This relationship translated into significant relationships (LOESS curve fits) between PCI and PC2 versus depth for all six data sets (F-tests, P<0.001), indicating that depth has a direct influence on the QTC substrate classification. Angle of incidence At the Aleutian Islands groundtruth site (FV Gladiator in 2005; Rooper and Zimmermann, 2007), there were significant linear correlations (P<0.05) between slope and EMs for the most common substrate classes of sand- boulder (n = 368 video observations), sand-sand (rc = 351), and bedrock-boulder (n=259), even when the analyses were restricted to low (<5°) slopes (von Szalay, 1998; von Szalay and McConnaughey, 2002). The influence of slope resulted in EMs that were equivalent among different substrates and different slopes. For example, EM 1 on a substrate of bedrock-boulder at 1° slope was equivalent to EM 1 on sand-sand substrate at 4.1° slope, and equivalent to EM 1 on sand-boulder substrate at 4.9° slope (Fig. 4), illustrating how easily substrates can be misclassified at low slopes. Assumption one: scale of measurements The S-Plus version of PCA, conducted after eliminat- ing invariant EMs (16, 31, and 40), confirmed that the QTC IMPACT method of PCA does not use any addi- tional data ranging or standardization. PCA performed in S-Plus with standardization (subtraction of mean, Zimmermann and Rooper: Comparison of echogram measurements for distinguishing seafloor substrates 299 F V Gladiator 2003 shallow to E c 2 I LU 03 Cl) E FV Gladiator 2003 deep 0.090 0.080 0.070 0.060 0.050 0.040 0.030 0,020 0.010 0.000 100 120 140 160 180 200 RV Resolution 2003 RV Pallasi 2004 RV Rangithi 1 999 RV John P. Tully 2002 0.180 0.160 0.140 0.120 0.100 0 080 0.060 0.040 0 020 0.000 60 80 100 120 140 Figure 3 Significant relationships between QTC (Quester Tangent Corporation, Sidney, British Columbia, Canada) IMPACT™ generated echogram measurements and depth for the shallow and deep FV Gladiator 2003 data sets, the RV Resolution 2003 cruise, the C.C.G.S. RV John P. Tully 2002 cruise, the RV Pallasi 2004 cruise, and the RV Rangithi 1999 cruise. Note the different scales on the x axes. division by standard deviation; Manly, 1994; Legen- dre and Legendre, 1998) revealed that more PCs were required to explain the same total amount of variance as the QTC IMPACT PCA method (Fig. 5). Thus the lack of standardization within the QTC IMPACT PCA method can have a strong effect. For example, across all six data sets, some EMs such as EM 15 were always large (>0.938), some such as EM 166 were always small (<0.006), and some such as EMs 1 and 102 generally had larger ranges and were more variable. Within the first group of EMs (1-23), EM 1 was always < EM 2, EM 2 was always < EM 3, etc., up to EM 16, which was always > EM 17, which was always > EM 18, etc., up to EM 23. Thus these variables have constricted ranges which can affect PCA. Additionally, the strong correlations between neighboring variables within the first. (EMs 1-23) and fifth (EMs 102-166) groups indicate that these EMs are either measuring nearly the same echo component, 300 Fishery Bulletin 106(3) or that EMs are based on neighboring EMs. Without standardization of these correlated EMs, the amount of variance explained by the first three PCs is artificially inflated. The inclusion of sums of variables in a PCA, such as the 15 sums of variables in the third (EMs 40-70) and fourth (EMs 71-101) groups, also artificially inflates the amount of variance explained in a PCA. Inclusion of a variable that is a complement of another variable (EM 41 or EM 56) in a PCA does not improve or impair the results and one of these complements could be excluded, along with the three invariant variables. Assumption two: first echo The 251 sampling envelope used with the QTC IMPACT software may be a mismatch for the actual length of the first echo. In the FV Gladiator 2003 shallow and deep data sets (both <200 m), the 251 sampling intervals of Sand-sand Sand-boulder Bedrock-boulder 1.00 0.98 0.96 0.94 0.92 0.90 0.88 0 2 4 6 8 10 12 14 16 Degrees of seafloor slope Figure 4 Significant linear regressions (P<0.05) between low seafloor slopes <5° (•) and QTC (Quester Tangent Corporation, Sidney, British Columbia, Canada) IMPACT™ echogram measurement 1, with observations from greater slopes (O) excluded from the regression analysis, for the three most common substrate types found during the FV Gladi- ator 2005 cruise. Only data values to the left of the 5° mark (dashed vertical line) were included in the regression analysis. 977 Hz or 0.001024 s translated into an excessive and unnecessary 192.8 m analysis depth range below the start of the first sea- floor reflection, a fact not realized during the collection of the echosounder data. However, the recording of most of the *.raw files were truncated before the full 192.8 m distance, before any second echo, and also before the full 251 samples; therefore most of the echoes needed padding (extended repetition of the last sound sample in the echo). The second echo was recorded in the *.raw file and it fell within the 251 sample requirement in 19 of the 68 shallow sites (25-100 m) and four of the 19 deep sites (100-200 m). These same pings were exported from EchoView® to QTC IMPACT with and without the second echo, and our analyses demonstrated that QTC IMPACT treated the second echo as if it were part of the first echo. The acciden- tal inclusion of the seafloor spike from this second echo reduced the values of the first 23 EMs (except EM 16) and had the greatest effect when the seafloor spike of the second echo occurred at the edge of the export window, where it was repeated to fulfill the 251 sample requirement (see Haul 206, Fig. 6). There was less of an effect when more of the second seafloor spike was included, so that the padded value was of lower sound intensity (see Haul 124, Fig. 6). Thus sig- nificant differences in some of the echogram measurements can be created for the exact same substrate type if users are not careful about ensuring that the 251 sample window of QTC IMPACT matches up well with the first echo length. Discussion Although this analysis demonstrated that there are several strong advantages (gain adjustment, bottom picking, bad data exclu- sion, and stacking) in using the partially automated echogram classifying software (QTC IMPACT), there are also several potential pitfalls (dependencies among the 166 EMs, lack of standardization, cor- relation with depth, influence of seafloor slope, and mismatch between 251 sample Zimmermann and Rooper: Comparison of echogram measurements for distinguishing seafloor substrates 301 100 90 - cd 80 c OS Q- 70 CD 60 ■ 50 40 □: 30 20 10 3 4 5 6 7 Number of principal components 10 Figure 5 Percent of variance explained in principal components analysis (PCA) conducted on the echogram measurements produced by QTC (Quester Tangent Corporation, Sidney, British Columbia, Canada) IMPACT™ software when the data were not standard- ized (QTC method ) and when the data were standardized (textbook method ) for the RV Resolution 2003 cruise (□), the C.C.G.S. RV John P. Tully 2002 cruise (O), the RV Pallasi 2004 cruise (O), the RV Rangithi 1999 cruise (+), and the NMFS FV Gladiator 2003 shallow (x) and deep (■&) cruises. intervals versus first echo length), such that it does not function as users would expect for distinguishing substrate types. There are also several processing steps within QTC IMPACT, such as repeating the last sample of short pings (padding), reducing the strength of sections of pings that are too strong, and increasing the strength of sections of pings that are too weak, all of which may affect analyses. The propri- etary nature of the software and the internal processing steps have discouraged user criticism and examination of the QTC IMPACT generated data (Kloser et al., 2001). QTC IMPACT users should export, format, and carefully examine their 166 EMs before substrate classification in order to catch user-generated mistakes, such as accidentally including a second echo, and to identify and remove any constant or collinear EMs. After analysis of the EMs, users may be able to reduce the 166 EMs to fewer than 10 without any loss of information, and compare these against depth, slope, and substrate types, if known, for further data-checking. The first assumption — that the scale or range of the data were appropriate for PCA — was disproved, and users may want to consider whether standard- izing is appropriate for their data. The second assumption — that QTC IMPACT only uses the first echo — is not necessarily true and published QTC IMPACT substrate classes may have been differentiated by the presence or absence of all or part of the second echo. Optimum substrate classification The inability to determine an optimum number of sub- strate classes for the shallow and deep FV Gladiator 2003 data sets is a common problem in seafloor substrate analysis and is not a critique of the particular A-means method within QTC IMPACT software. Our echosounder data could have been too noisy, too coarse, too affected by sea-state or seafloor slope, or our trawl sites could have been too variable or too constant for determining sub- strate classes. Instead, our results, with corroborations from independent data sets, indicated the importance of analyzing the echogram measurements before any PCA and A-means analysis so that depth-related and slope- related errors, second echo or echo envelope errors, and variable range or collinearity errors could be caught. The pros and cons of the QTC IMPACT method of A-means partitioning have already been thoroughly discussed. It was criticized by Legendre et al. (2002) who offered a new A-means method based on Euclidean distance. Preston and Kirlin (2003) responded by defending and elaborating on their A-means clustering method, which is based on Mahalanobis distance, and citing successful QTC IMPACT substrate-typing projects (Anderson, 2001; Morrison et al., 2001; Anderson et al., 2002; Ellingsen et al., 2002). Legendre (2003) offered additional criti- cism of the QTC IMPACT A-means clustering method and added that the QTC IMPACT method of cluster- ing, based on only the first three principal components (PCs), was strongly influenced by depth, since his PCI was strongly related to depth, as opposed to a solution that would include a greater number of PCs. Clearly the A-means method for distinguishing substrate classes is important, and strongly linked to the data values that feed into it. Examination of the data and the variability and covariance of echogram measurements This exploration of the 166 EMs provides the first description of the QTC IMPACT data set that is used for seafloor substrate classification. Before this description, only three EMs were known to be invariant and the rest were highly collinear (Legendre et al., 2002). These EMs were known to carried limited information and were highly redundant (Ellingsen et al., 2002). Researchers collecting data directly into QTC VIEW, such as cor- roborating data from different agencies, did not have the additional processing step of importing the data from EchoView®, and were probably unaware of the 96 dB dynamic range required for QTC IMPACT software. Therefore the effect of clipping portions of pings that were too loud, increasing the sound level of ping portions that were too quiet, or adding or subtracting a constant amount of sound to entire ping data sets (gain adjust- ment), was not widely reported in the literature. Only Anderson et al. (2002) mentioned experimenting with 302 Fishery Bulletin 106(3) FVGIaditor 2003, Haul 124 LU CO +1 FVGIaditor 2003, Haul 206 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 Echogram measurement no. Figure 6 Relationship between the mean of the first 23 echogram measurements (± stan- dard error [SE ] ) for two hauls when part of the second echo is included ( ) or excluded ( ) in the QTC (Quester Tangent Corporation, Sidney, British Columbia, Canada) IMPACT™ analysis. different gain settings and different reference depths. Users have also been unaware of the 251 sample require- ment, or the effects of repeating the last sound sample (padding) in order to fulfill this requirement. Ellingsen et al. (2002) mentioned that truncated acoustic reflec- tions resulted in a loss of some of the 166 EMs. Perhaps results from their field work resulted in modification to the QTC IMPACT software so that the last acoustic sample was repeated (a process known as padding). Correlation of echogram measurements with depth The influence of depth on the EMs, and on the resulting PCs, may be due to improper echosounder calibration or improper depth-correction in QTC IMPACT, rather than to true variation in substrate types. It is not possible for users to determine the origin of the depth influence. Although the reference depth is supposed to compensate for the signal-protraction of pings of different depths within a data set, none of the QTC IMPACT studies in the literature have actually checked to determine if such compensation occurs. It has been reported in the litera- ture that the QTC IMPACT-generated PCI is correlated with depth (Legendre, 2003) and that QTC IMPACT- generated substrate classes are sometimes correlated with depth (Anderson et al., 2002). As with our findings, depth biases were also reported for the El (roughness) and E2 (hardness) measurements made by RoxAnn™ (Sonavision, Aberdeen, Scotland, U.K.) bottom-typing software, as determined by a more thorough study with careful seafloor groundtruthing (Kloser et al., 2001). Angle of incidence Any potential effect due to impact angle of echogram reflection, which is a combination of seafloor slope and vessel motion, is not widely addressed in the literature. Anderson (2001) used QTC VIEW™ to distinguish among substrates on steep slopes, some of which appear to be as steep as 45° (see Anderson, 2001, Figs. 4 and 5), whereas von Szalay (1998) and von Szalay and McConnaughey (2002) reported that Zimmermann and Rooper: Comparison of echogram measurements for distinguishing seafloor substrates 303 seafloor slopes exceeding 5° to 8° caused significant substrate misclassification. In Ellingsen et al. (2002), effects from vessel motion may have been reduced or eliminated by working in calm seas. Our analysis of individual EMs revealed the mechanism whereby substrates may be misclassified in areas with slope and we would suggest that there is greater sensitivity with QTC VIEW than previously noted by von Szalay (1998) and von Szalay and McConnaughey (2002). The QTC IMPACT method of stacking multiple pings could potentially ameliorate the influence of slope-affected pings, but it could also create new substrate classes by combining normal and non-normal reflections. Although it is presumed that QTC IMPACT software could distinguish among substrates at a constant depth in a flat seafloor area with no vessel pitch and roll, this type of situation is not realistic for the NMFS bottom trawl surveys in the Gulf of Alaska and Aleutians Islands. If steep seafloor areas can be dis- tinguished from vessel motion through careful incor- poration of vessel motion measurements, slope may be considered as a substrate modifier or as a significantly different substrate, depending on the species of inter- est for which habitat is being defined. Assumption one: scale of measurements The QTC IMPACT method of PCA (without standard- ization of data) results in a higher amount of variance explained because it is based on a few variables with the highest variance, which are also highly corre- lated. Those EMs without much variance only make a minor contribution to the PCA solution; however, it remains unclear whether forcing EMs to vary through standardization — a process that could possibly include both discriminating (signal) and nondiscriminating measures (noise) of echo energy, timespread, and skew- ness (van Walree et al., 2005) — increases or decreases statistical power for discriminating substrate types. The user is left having to choose between conducting a nonstandardized PCA where nearly all variables are collinear or conducting a standardized PCA that may be based mostly on noise. Including fully collinear (e.g., the sums of) variables and correlated variables in a PCA does not provide additional discriminatory information, but it does change the results. Therefore users may find it beneficial to conduct an additional PCA without these collinear variables and determine how much the substrate groupings change. Perhaps not coincidentally, our findings that only three to six variables within each of the acoustic data sets were somewhat independent (provided discriminatory power) matches well with van Walree et al.’s (2005) descrip- tion of six acoustic algorithms and Kloser et al.’s (2001) description of four algorithms. Our results indicate that the EMs are not standardized before PCA, and because this is not mentioned in the literature, it may be an unexpected problem for users. The lack of standardiza- tion among collinear and correlated variables might partly explain why QTC IMPACT software typically requires only three eigenvectors to explain more than 95% of covariance (Ellingsen et al., 2002; Legendre et al., 2002). Assumption two: first echo Surveys conducted in shallow water with high sam- pling rates and surveys conducted in deep water with low sampling rates (such as that of the NMFS 2003 FV Gladiator) are equally vulnerable to accidentally including second, or later, seafloor reflections in QTC IMPACT analysis. For example, the RV Rangithi 1999 and RV Pallasi 2004 data sets both required 9.4 m below the start of the first seafloor reflection to achieve 251 samples, but the range of these data sets was shal- lower (Table 1). Because both of these data sets were recorded directly into QTC VIEW, the raw data could not be examined to determine if additional echoes were recorded or not. However, both data sets had charac- teristic drops in the values of the first group of EMs between approximately 9 and 4 m depth, indicating a probable increasing inclusion of the second echo with a decrease in depth (see Fig. 3). By 4 m in depth, both data sets should have included most of the second echo. The sharp increase in EM 1 just below 3 m in the RV Pallasi 2004 data set, at the shallowest depth, may indi- cate partial inclusion of the third echo. Thus accidental analysis of more than one echo with QTC IMPACT can cause strong depth-related influences and can create significantly different echogram measurements such that additional substrate classes could be created. To avoid such problems, users need to compare the depth range for their echogram measurement analysis (echo envelope) to the range of depths in their study area. Conclusions The need for a cost-effective approach to classify sea- floor substrates, in order to define EFH across areas such as the NMFS bottom trawl surveys of the Gulf of Alaska and Aleutian Islands, remains strong. Because of the unexpected problems with the QTC IMPACT processing steps and creation of EMs, it seems highly likely that QTC IMPACT users are producing substrate classifications based on problems implementing the software or analyzing the measurements. Although data-gathering or data-processing errors are common across all such analyses, there is little chance to correct such errors when using a black box system. Therefore for future projects more transparent analytical methods will be needed, such as the published algorithms in Kloser et al. (2001) and van Walree et al. (2005), for translating acoustic data into EFH. Acknowledgments We thank D. Urban for supplying the RV Resolution 2003 data, I. Murfitt for the RV Pallasi 2004 data, 304 Fishery Bulletin 106(3) C. Grandin for the C.C.G.S. RV John P. Tully 2002 data, and J. Hewitt for the RV Rangithi 1999 data. The data contributors also provided helpful manu- script reviews, along with D. Somerton, R. McCon- naughey, S. Syrjala, L. Bonacci, and three anonymous reviewers. S. Syrjala, P. Spencer, and others provided helpful statistical advice. The video and multibeam groundtruth data from the Aleutian Islands were col- lected with support from the North Pacific Research Board. Literature cited Anderson, J. T. 2001. Classification of marine habitats using submersible and acoustic seabed techniques. In Spatial processes and management of fish populations (G. H. Kruse, N. Bez, A. Booth, M. W. Dorn, S. Hills, R. N. Lipcius, D. Pelletier, C. Roy, S. J. Smith, and D. Witherells, eds.), p. 377-393. Alaska Sea Grant Rep., Alaska Sea Grant Program, Univ. Alaska Fairbanks. AK-SG-01-02. Anderson, J. T., R. S. Gregory, and W. T. Collins. 2002. Acoustic classification of marine habitats in coastal Newfoundland. ICES J. Mar. Sci. 59:156-167. Ellingsen, K. E., J. S. Gray, and E. Bjprnbom. 2002. Acoustic classification of seabed habitats using the QTC VIEW™ system. ICES J. Mar. Sci. 59: 825-835. Federal Register. 2002. Magnuson-Stevens Act Provisions; Essential Fish Habitat (EFH). 50 CFR Part 600, Federal Regis- ter 67(12):2343-2383. Office of the Federal Register, National Archives and Records Administration (NARA), College Park, MD. Kloser, R. J., N. J. Bax, T. Ryan, A. Williams, and B. A. Barker. 2001. Remote sensing of seabed types in the Austra- lian South East Fishery; development and application of normal acoustic techniques and associated “ground truthing.” Mar. Freshw. Res. 52:475-489. Legendre, P. 2003. Reply to the comment by Preston and Kirlin on “Acoustic seabed classification: improved statistical method”. Can. J. Fish. Aquat. Sci. 60:1301-1305. Legendre, P., K. E. Ellingsen, E. Bjornbom, and P. Casgrain. 2002. Acoustic seabed classification: improved statistical method. Can. J. Fish. Aquat. Sci. 59:1085-1089. Legendre, P., and L. Legendre. 1998. Numerical ecology, 2nd ed., 853 p. Elsevier Sci- ence B.V., Amsterdam. [In English.]. Manly, B. F. J. 1994. Multivariate statistical methods, 2nd ed., 215 p. Chapman and Hall, New York, NY. Morrison, M. A., S. F. Thrush, and R. Budd. 2001. Detection of acoustic class boundaries in soft sedi- ment systems using the seafloor acoustic discrimination system QTC VIEW. J. Sea Res. 46:233-243. Neter, J., W. Wasserman, and M. H. Kutner. 1990. Applied linear statistical models: regression, analy- sis of variance, and experimental designs, 3rd ed., 1181 p. Richard D. Irwin, Homewood, IL. Orlowski, A. 1984. Application of multiple echoes energy measure- ments for evaluation of sea bottom type. Oceanologia 19:61-78. Pace, N. G., and R. V. Ceen. 1982. Seabed classification using the backscattering of normally incident broadband acoustic pulses. Hydrogr. J. 26:9-16. Preston, J. M., and R. L. Kirlin. 2003. Comment on “Acoustic seabed classification: improved statistical method”. Can. J. Fish. Aquat. Sci. 60:1299-1300. Rooper, C. N., and M. Zimmermann. 2007. A bottom-up methodology for integrating under- water video and acoustic mapping for seafloor substrate classification. Cont. Shelf Res. 27:947-957. van Walree, P. A., J. T. Tegowski, C. Laban,, and D. G. Simons. 2005. Acoustic seafloor discrimination with echo shape parameters: A comparison with the ground truth. Cont. Shelf Res. 25:2273-2293. von Szalay, P. G. 1998. The feasibility of using single beam seabed clas- sification systems to identify and quantify slope rockfish habitat in the Gulf of Alaska. M.S. thesis, 158 p. Univ. Washington, Seattle, WA. von Szalay, P. G., and R. A. McConnaughey. 2002. The effect of slope and vessel speed on the perfor- mance of a single beam acoustic seabed classification system. Fish. Res. 56:99-112. 305 Abstract — Over 34,000 age 0-2 ju- venile sablefish ( Anoplopoma fimbria) were tagged and released in southeast Alaska waters during 1985-2005. The data set resulting from this tagging study was unusual because of its time span (20 years) and because age could be reliably inferred from release length (i.e., tagged and released fish were of known age); thus, age-specific movement patterns could be examined. The depth- and area-related recovery patterns supported the concepts that sablefish move to deeper water with age and migrate counterclockwise in the Gulf of Alaska. Availability to the fishery increased rapidly for fish of younger ages, peaked at age 5 to 6, and then gradually declined as sablefish moved deeper with age. Decreased availability with age may occur because of lower fishing effort in deep water and could have sub- stantial implications for sablefish stock assessments because “dome- shaped” availability influences the reliability of abundance estimates. The area-related recovery pattern was not affected by year-class strength; i.e., there was no significant density- dependent relationship. Manuscript submitted 4 December 2007. Manuscript accepted 4 April 2008. Fish. Bull. 106:305-316 (2008). The views and opinions expressed or implied in this article are those of the author and do not necessarily reflect the position of the National Marine Fisheries Service, NOAA. Age-specific movement patterns of sablefish ( Anoplopoma fimbria) in Alaska Nancy E. Maloney (contact author) Michael F. Sigler Email address for N. E. Maloney: Nancy.Maloney@noaa.gov National Oceanic and Atmospheric Administration National Marine Fisheries Service Alaska Fisheries Science Center, Auke Bay Laboratories Ted Stevens Marine Research Institute 17109 Point Lena Loop Road Juneau, Alaska 99801 The general migration pattern of sablefish (Anoplopoma fimbria) in the northeast Pacific Ocean was deduced in the 1980s from several tagging studies (Bracken, 1983; Beamish and McFarlane, 1988; Fujioka et al., 1988;) and enlarged upon in further studies over the following two decades (Heifetz and Fujioka, 1991; Rutecki and Varosi, 1997; Kimura et al., 1998; Maloney, 2004). In southeast Alaska, juvenile sablefish that are spawned offshore appear in inshore waters in late summer or early fall and spend the first year or two of life in shallow coastal bays and inlets before moving into progressively deeper water. At the same time that they are moving into deeper water, many young sable- fish move north and west on a migra- tion path that takes them across the Gulf of Alaska to the Aleutian Islands and Bering Sea. Eventually, most will return to the eastern Gulf of Alaska as adults. The sablefish fishery in the Gulf of Alaska (GOA), eastern Bering Sea, and Aleutian Islands is managed by the National Marine Fisheries Ser- vice (NMFS) in cooperation with the North Pacific Fishery Management Council. Sablefish in these areas are assumed to belong to one population (Kimura et al., 1998), for which a to- tal allowable catch is calculated each year and apportioned among six man- agement areas. The annual quotas for each area are based on the distri- bution of biomass among the areas, which is estimated from longline sur- veys and commercial catches (Heifetz et al., 1997). Because sablefish are known to be migratory, estimates of the rates of migration between ar- eas could affect the apportionment of quotas among management areas (Heifetz et al., 1997). Migration rates between areas have been estimated from tag data by using fish-length classes in the modeling process (Heifetz and Fu- jioka, 1991). Although fish-length data are commonly available, actual age data are generally scarce. Age data are preferable to length data for estimating population age structure (Sigler, 1999), but sablefish are diffi- cult to age, especially for ages great- er than 5 or 6 years (Kimura and Lyons, 1991). Tagging of known-age juveniles before they leave coastal areas offers an opportunity to docu- ment age-specific movements. Age 0-2 (mostly age 1) sablefish have been tagged annually since 1985 in bays and inlets of southeast Alas- ka. The objective of our study was to determine movement patterns of sablefish based on these known-age fish, using a unique 20-year data set of age-specific mark-recapture data. Specifically, we determined 1) how the depth inhabited by sablefish changes with age; 2) how the area inhabited changes with age; 3) how availability to the primary fishery (longline) changes with age; and 4) whether there is a density-dependent effect of year-class strength on the extent of migration of young sable- fish. Results of objectives 1 and 2 largely confirmed the results of pre- vious studies, whereas objectives 3 and 4 were new. 306 Fishery Bulletin 106(3) 1 80°0’0" 160°0'0"W 140°0'0"W 120°0'0"W Figure 1 Recovery locations (o) of sablefish (Anoplopoma fimbria) tagged and released as juveniles in southeast Alaska. St. John Baptist Bay was the most common release site. Vertical and horizontal bars are regulatory area boundaries. Gulf of Alaska is shown as GOA for the Western, Central, and Eastern areas. Materials and methods Juvenile sablefish were captured, primarily with jigging gear (Rutecki and Varosi, 1997) in various bays and inlets of southeast Alaska, for tagging and release from 1985 to 2005. A total of 74 sites were selected during that period, but most tagging after 1987 was undertaken in St. John Baptist Bay near Sitka, Alaska (Fig. 1), because it was easily accessible and juvenile sablefish were consistently found there. Bottom depth is about 30 m and fish were caught on the bottom and throughout the water column. Release and recovery data for these fish are maintained in the NMFS Alaska Sablefish Tag Data- base, which is described in detail by Fujioka et al. (1988). Data criteria Recovered fish had to meet several criteria to be included in this study: recovery year had to be known, recovery depth and location had to be accurate, and the fish had to be at liberty for at least one year. Recovery year was necessary to calculate age at recapture. Accurate recovery depth and location were necessary to reliably assign recovery depth strata and areas. Only recover- ies for which there were reported positions that were precise to within 10 minutes of latitude and longitude were used; this criterion was used to judge the reli- ability of the recovery information. Some tag recoveries had accurate recovery location but no depth informa- tion; these tag recoveries were included in the area analysis, but not the depth analysis. Recovery depths were classified into seven depth strata chosen to reflect general habitat type: 1-100 m (nearshore), 101-200 m (continental shelf), 201-300 m (shelf break), 301-500 m (upper continental slope), 501-700 m (middle slope), 701-1000 m (lower slope), and >1000 m (deep water). Recovery locations were classified into seven areas: Aleutian Islands, Bering Sea, western Gulf of Alaska (western GOA), central Gulf of Alaska (central GOA), eastern Gulf of Alaska outside waters (eastern GOA outside), eastern Gulf of Alaska inside waters (eastern GOA inside), and British Columbia. For some analyses, inside and outside waters were pooled and referred to as eastern GOA. The minimum time at liberty of one year was imposed to exclude short-term movements and to focus on migration. Age at release was determined from fish size and time of year. Depending on the time of year, in most years and tagging areas, no more than two ages of fish, and usually only one, were present at the time of tagging. Ages were readily separable by means of non-overlapping length frequencies and by time of year. Age-0 fish enter into bays from the ocean in the fall of their first year of life, and they average 21-23 cm in length (Rutecki and Varosi, 1997). One-year-old fish in the middle of summer average 31-35 cm and 2-year-old fish average 40-45 cm. The number of years at liberty after release was calculated by subtracting the release year from the recovery year; adding this number to the release age supplied the recovery age. Maloney and Sigler: Age-specific movement patterns of Anoplopoma fimbria 307 Availability to the fishery by age Sablefish move progressively deeper with age, and as they do so, become available to the main commercial fishery (longline), which operates primarily on the con- tinental slope. The fraction of the total population avail- able by age to the commercial fishery was estimated by the following method. The initial number of tagged fish released in year t of age a is N'at. A fraction of the tags, Z = 0.048 (Lenarz and Shaw, 1997), are immediately lost or the fish die from tagging, such that a short time after tagging, some smaller number of tagged fish survive, Nat = (1 - l)N'at. The year following tagging, the number of tagged fish is Na+U+ 1 = exp (-(M + XsaFt + H)), where M = 0.1 (Sigler, 1999; Hanselman et al., 2006) is the instantaneous rate of natural mortal- ity; A = a calibration parameter (Heifetz and Fujioka, 1991) to account for bias in assumed values for the instantaneous rates of annual fishing mortality (Ft); sa = availability (selectivity) to the commercial fishery; and H = 0.03 (Lenarz and Shaw, 1997) is the instan- taneous rate of tag shedding. The Ft values were estimated independently in the Alaska sablefish stock assessment (Hanselman et al., 2006). The fishery captures a number of the tagged fish, Cat, where Cat = XsaFtHM + XsaFt+H) (l-exv(-(M + 7isaFt + H)))Nat The relationship between availability and age was rep- resented by the exponential-logistic function (Thompson, 1994; Sigler, 1999) sa = l/(l-y)((l-y)/y)7 exp ( /fy(a - a)) / (l + exp ( /f( a - a ))) . The exponential-logistic function is flexible, allowing both asymptotic availability when availability increases with age to an asymptote, and dome-shaped availability when availability increases with age to a maximum and then decreases for older fish. The exponential-logistic function automatically scales maximum availability to 1.0 and reduces to asymptotic availability as the param- eter y approaches zero. When y = 0, the parameter a is the age of 50% availability and the slope of the curve equals (4 (5 at a = a. W’hen y > 0, then a and /3 lose bio- logical meaning because a no longer represents the age at 50% availability, and y is a parameter that allows availability to decrease (and form the “dome-shape”) for older ages The fishery switched from open access to individual fishing quotas (IFQ) in 1995. This switch has been shown to affect availability of the fish to the fishery (Sigler and Lunsford, 2001). Thus, we estimated availability parameters, a, and y, as well as the fishing mortality calibration parameter, A, separately for each time period (1984-94, 1995-2005). We assumed that the estimated availability curves represent the commercial longline fishery because most tags (93%) were recovered by longline or other fixed gear types. Not all tagged fish caught in the sablefish fishery are reported (Heifetz and Maloney, 2001). The number of tags reported, R, is related to the number of tagged fish caught, Cat, where Rat - wt Cat and wt is the report- ing rate. Heifetz and Maloney (2001) estimated annual reporting rates for 1980-98 and subsequent reporting rates were estimated of 0.43 for 1999-2001 and 0.52 for 2002-05, which we applied in our analysis. The model parameters (a, (i, y, and A for 1984-94 and 1995-2005) were estimated by maximum likeli- hood. The observed number of tag recoveries in any year-cohort grouping was small (mean of 6, range of 0 to 27); therefore the expected number of tag recoveries, Q , could be approximated by the Poisson distribution (Hilborn, 1990). The negative log-likelihood (-log,L) for all observed recoveries was -log, L(Qat\Rat) = X „ X , ' ( ' Qat - R°* loge ( 1 3* ) + log, ' ( ■ Rat ! )), which was minimized to find the most likely set of parameter estimates. We examined model fit using devi- ance (McCullagh and Nelder, 1983), which for any obser- vation of tag recoveries is devianceia, t) = - 2 { log, L( Qat \Rat ) - log, L( Rat \ Rat J (Heifetz and Fujioka, 1991). We applied the likeli- hood ratio test for nested models (Hilborn and Mangel, 1997) to determine whether model fit was significantly improved by assuming separate parameter sets for the open access and IFQ fisheries. We estimated the 95% confidence intervals of the parameters from their likeli- hood profiles (Hilborn and Mangel, 1997). Density-dependent effect on migration Migration may be affected by abundance if sablefish tend to disperse when abundant. We tested for a den- sity-dependent effect by examining whether recovery patterns by area were influenced by cohort abundance (recruitment strength). Recruitment strength is esti- mated through age-structured population modeling (Hanselman et al., 2006) and is expressed as the number of fish at age 2 (in millions). We tested by linear regres- sion whether more recoveries occurred in western areas for stronger year classes, hypothesizing that more mem- 308 Fishery Bulletin 106(3) Table 1 Total numbers of juvenile sablefish (Anoplopoma fimbria ) released and recovered in southeast Alaska, 1985-2005, by release age (age 0, age 1, and age 2). Also shown are numbers of recoveries of fish with known recovery year, with accurate recovery location and accurate recovery depth, for fish at liberty longer than one year. Year Age 0 Age 1 Age 2 Total releases Total recoveries Accurate recovery location Accurate recovery depth 1985 0 6168 0 6168 853 1986 0 240 936 1176 68 2 1 1987 0 7916 0 7916 314 8 7 1988 1762 2142 1 3905 153 26 20 1989 0 530 1 531 35 47 41 1990 0 0 0 0 0 65 53 1991 789 2580 1 3370 154 56 39 1992 0 1658 0 1658 68 57 47 1993 0 568 26 594 48 66 59 1994 0 1190 8 1198 44 31 27 1995 0 986 0 986 75 50 42 1996 0 1735 0 1735 62 46 42 1997 0 58 0 58 4 59 55 1998 0 1174 0 1174 37 43 38 1999 0 859 5 864 41 61 56 2000 0 559 178 737 41 40 31 2001 0 105 1 106 3 41 34 2002 0 471 2 473 8 37 29 2003 766 0 0 766 0 56 49 2004 0 290 1 291 0 40 34 2005 0 610 0 610 3 29 26 Totals 3317 29,839 1160 34,316 2011 860 730 bers of strong year classes would move westward if cohort density affected migration. Results Over 34,000 juvenile sablefish were tagged and released in southeast Alaska from 1985 to 2005 (Table 1). Most (87%) were tagged and released at age 1. A total of 2011 sablefish tagged as juveniles were recovered, most by the commercial fishery and a few by research vessels. Of these 2011 recoveries, 860 fish had a known recovery year, accurate recovery information, and were at liberty for at least one year, thus qualifying for area-based analyses; 730 fish also had known recovery depth and qualified for depth-based analyses (Table 1). Of the 860 fish recovered, most (85%) were caught by longline, 8% by pots, 6% by bottom trawl, and the remainder (1%) by jig, purse seine, sport fishing gear, or by unknown gear. The largest percentage (45%) of recoveries occurred in the eastern Gulf of Alaska (Table 2). Large percentages of tagged fish also were recovered in the central Gulf of Alaska (30%) and farther westward (18%). About half (51%) of recoveries occurred at depths 501-700 m and nearly all (93%) recoveries occurred at depths from 201 to 1000 m. Sablefish tagged as juveniles in southeast Alaska were recovered as far west as 177°E along the Aleutian Islands, as far north as 60°N in the eastern Bering Sea, and as far south as 48.5°N off Vancouver Island (Fig. 1). Most recoveries were located along the upper continental slope or in cross-shelf gullies such as Spen- cer and Seward Gullies. Having originated in coastal bays, these fish had to cross the continental shelf to reach these areas. Fish recovered in Chatham Strait may have moved there by way of inland waters or may have migrated first to outer coastal waters before mov- ing into the strait. Recovery locations by depth and area Generally, young fish were more common at shallower depths and older fish were more common at greater depths. In depths shallower than 200 m, the most common ages of tagged juvenile sablefish recoveries were 3 and 4 years, and in depths greater than 200 m, the most common ages were 5-8 years (Fig. 2). Median recovery age increased with depth from shallow (2 years) Maloney and Sigler: Age-specific movement patterns of Anoplopoma fimbria 309 Table 2 Recovery area and depth strata (m) for recovered sablefish (Anoplopoma fimbria) tagged as juveniles, number of fish for which depth was unknown, and total number of tagged fish captured in the recovery area. Proportion of recoveries at depth (bottom row) excludes recoveries with unknown depth. Proportion of recoveries by recovery area (rightmost column) includes recoveries with unknown depth. GOA = Gulf of Alaska. Depth strata (m) Recovery area 1-100 101-200 201-300 301-500 501-700 701-1000 >1000 Unknown Total Proportion Bering Sea 4 9 4 1 18 0.02 Aleutian Islands 3 5 39 7 8 62 0.07 Western GOA 4 6 19 28 7 10 74 0.09 Central GOA 25 43 48 90 19 2 28 255 0.30 Eastern GOA 2 12 8 67 177 45 1 75 387 0.45 British Columbia 6 11 26 12 1 8 64 0.07 Total 2 44 63 154 369 94 4 130 860 Proportion 0.00 0.06 0.09 0.21 0.51 0.13 0.01 730 Recovery age (yr) 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 Figure 2 Tag recoveries from sablefish ( Anoplopoma fimbria) tagged as juveniles, by age (years) and depth (m) for all areas pooled. The size of the circles is proportional to the number of recoveries and represents a range from 1 to 57 recoveries. The symbol (x) represents the median age (years). to deep (5-6 years). Only two 2-year-old sablefish that had traveled <6 km from their release location in St. John Bap- tist Bay were recovered in nearshore waters (<100 m water depth); no fish older than 2 years were recovered in nearshore waters. Only one sablefish older than 13 years was found in depths <300 m and most were found deeper than 500 m. The most common areas of recovery were the central GOA and the east- ern GOA, which together accounted for 75% of all recoveries (Table 2). There were more than twice as many recov- eries in outside waters as in inside waters of the eastern GOA. The most common ages of recovered fish in all areas except eastern GOA and British Columbia were 5 to 7 years (Fig. 3). In eastern GOA the most common ages at recovery were 3 to 6 years, and in British Columbia 6 to 9 years. Fish of all ages from 3-10 years old were recovered in all areas except the Ber- ing Sea, where no fish younger than 4 years old were recovered. The Bering Sea had the fewest recoveries (only 18) and the smallest range of ages (4-10 years, except for one recovery of a 19-year-old fish). Most fish 2 years old and most fish older than 12 years were recovered in the eastern GOA (Fig. 3). Most recov- eries in western areas (Bering Sea, Aleutian Islands, western Gulf of Alaska) were 12 years old or less. The large number of recoveries that occurred in the central Gulf of Alaska (30%) and farther west (18%) indicated that nearly half of the population had moved westward from the eastern Gulf of Alaska (Table 2). Movement by age Age-specific movement patterns were discernible even though multiple ages were found within areas and depths. By ages 3 and 4 years, most fish had moved offshore into >100 m water depth (Fig. 2). Some had moved to the inside waters of the eastern GOA or directly south into the waters off British Columbia, but most were found in eastern GOA outside waters or in the central GOA 310 Fishery Bulletin 106(3) Recovery age (yr) Figure 3 Tag recoveries from sablefish ( Anoplopoma fimbria) tagged as juveniles, by age and area. The size of the circles is proportional to the number of recoveries and represents a range from 1 to 43 recoveries. GOA=Gulf of Alaska. (Fig. 3). By ages 5 and 6 years, many fish had reached the western areas. Age 5 was the most common age of recovery for the western GOA and Aleutian Islands, and age 6 for the Bering Sea. Some fish aged 7-9 years remained in the western areas, but most had begun a return to the east. In the central GOA and eastern GOA outside waters, the 7-9-year-olds were mostly found in the 501-700 m depth range and were some of the most numerous fish. Fish aged 6-9 years were the most commonly recovered in the waters off British Columbia and may also have been fish returned from a westward migration (Fig. 4). A few fish may have come from eastern GOA inside waters (Chatham Strait), but Maloney and Heifetz (1997) found that this area has a high proportion of non-migrating fish. The most common depth stratum for fish recovered in British Columbia was the 501-700 m stratum. Availability to the fishery by age The model of availability at age fit the observed pat- tern of tag recoveries well. Deviances were scattered symmetrically around zero for most ages (Fig. 5). Only for ages 2 and 13 were there noticeable biases. The full model that assumed separate selectivity functions for the open access and IFQ fisheries significantly improved model fit, compared to a single selectivity function (Like- lihood ratio test, %2=220.1, df=3, P<0.001). The assump- tion of separate calibration coefficients rather than a single calibration coefficient also significantly improved fit (likelihood ratio test, x2 = 189.7, df=l, P<0.001). Including parameters to allow availability to decrease for older ages (y for 1985-94 and 1995-2005) signifi- cantly improved fit compared to a reduced model with asymptotic availability (likelihood ratio test, %2= 686.7, df =2, PcO.OOl). Juvenile sablefish first became available to the com- mercial fishery at age 2. Availability rapidly increased such that by age 5, nearly all sablefish were avail- able to the commercial fishery (Fig. 6). Both the age at 50% availability and the age at 100% availability values were one year greater (older) in the IFQ fishery than in the open access fishery (4 years versus 3 years and 6 years versus 5 years, respectively). Availability decreased for older ages, such that by 15 years, avail- ability was 50% for the open access fishery and 20% for the IFQ fishery. The degree of dome shape was sensitive to the assumed value of M; for example, by 15 years, availability for the open access fishery was 70% for M = 0.12 compared to 50% for the assumed value of M = 0.10. Density-dependent effect on migration We tested whether year-class strength affected the pro- portion of recoveries in the western areas, hypothesizing that a density-dependent effect would cause more recov- eries in western areas for strong year classes. There was no significant relationship (regression, df=14, P=0.18) because about 20% of recoveries occurred in western areas, regardless of year-class strength (Fig. 7). In the regression, the proportion was transformed by arcsin squar- eroot, as is recommended to normalize data expressed as proportions (Zar, 1984). Maloney and Sigler: Age-specific movement patterns of Anoplopoma fimbria 311 A Eastern GOA outside waters Recovery age (yr) 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 2 o-ioo - 101-200 ■ O O o O o 201-300 - o 0 o o o o 301-500 - °00000 o o O o O O O 501-700 - ° OCXXIjoCDO o o o o o o 701-1000 - O o OO ° o o o ° ° ° ° >1000 - o sz Q. 1000 o Figure 4 Tag recoveries from sablefish ( Anoplopoma fimbria) tagged as juveniles in (A) eastern Gulf of Alaska (GOA) outside waters and (B) central GOA, by age (years), depth (m), and area. The size of the circles is proportional to the number of recoveries and represents a range from (A) 1 to 15 recoveries and (B) 1 to 16 recoveries. Discussion Movement by depth and area with age Sablefish spend 1-2 years nearshore before moving onto the continental shelf where they reside as adults and spawn. In this study, the most common ages in the shal- lower depths were 3-4 years and in the deeper depths 5-8 years, indicating that sablefish are younger on the continental shelf than on the continental slope. This result confirmed other sablefish age data that indicated that fewer sablefish older than 10 years are found on the continental shelf than on the continental slope ( Sigler et al., 1997). Concurrent with the offshore movement with age, many young fish from the eastern areas (British Columbia, eastern GOA) moved in a northerly and west- erly direction through the central GOA to the western areas (western GOA, Bering Sea, and Aleutian Islands). The simultaneous depth and area movements resulted in a general age distribution of younger fish in shallower water in the east, mid-age fish in middle depths in the west, and older fish in deeper waters on the return from western to eastern areas. Age pattern variability was high; there was a sub- stantial overlap of ages within and between depths and areas. For example, fish of ages 2-13 years were recovered in 101-200 m and fish aged 2-20 years were caught in 501-700 m. Likewise, fish of all ages from 3 to 10 years were recovered in all areas except the Ber- ing Sea where the youngest fish recovered was 4 years old. However, the separation of ages by depth, although incomplete, was quite pronounced within some areas, most notably the eastern GOA outside and the central GOA (Fig. 4). In both of these areas the distribution of younger fish in shallower water and older fish in deeper water was evident and, taken together with the general 312 Fishery Bulletin 106(3) Figure 5 Scatterplot representing the fit of observed and predicted availability of sablefish ( Anoplopoma fimbria) to the fishery by age. Deviance is a function of the difference between observed and predicted values; a smaller deviance indicates a better fit of the exponential-logistic model to the observation. Figure 6 Estimated availability (fraction of the total population) of sablefish (Ano- plopoma fimbria) to the fishery by age (years) for the open-access (1985-94, ) and individual fishing quota (IFQ) (1995-2005, — ) fisheries. Param- eter estimates and 95% confidence intervals for 1985-94 are a = 1.50 (1.26, 1.87), p = 3.33 (3.04, 3.69), y = 0.050 (0.023, 0.092), and A = 0.23 (0.22, 0.25) and for 1995-2005 are a = 1.14 (1.05, 1.24), j3 = 4.76 (4.53, 5.01), y = 0.18 (0.15, 0.21), and A = 0.38 (0.36, 0.39). age distribution by area, corroborat- ed the counterclockwise pattern of sablefish migration in the northeast Pacific Ocean. Availability to the fishery by age The pattern of movement from shal- low to deep water with age (Fig. 2) results in increased availability to the fishery as sablefish grow older (Fig. 6). Estimates of availability at age have shown that about half of sablefish are available to the fish- ery by age 3 or 4, depending on the fishery management system, and that most are available by age 5 or 6. The later availability of fish (at ages 5 or 6) under the IFQ manage- ment system (compared to earlier availability of younger fish under the open access system) also was also found to be the case in an analysis of length-frequency data from the two fisheries, and this pattern of avail- ability was likely created because the crowding of fishing vessels during the open access fishery pushed fish- ermen into areas and depths where there were smaller fish (Sigler and Lunsford, 2001). The IFQ fishery, with a longer season and fewer ves- sels, reduced crowding so that fish- ermen were able to avoid shallower depths with smaller, younger fish. Unlike previous analyses, where availability was assumed not to de- crease with age (e.g., Sigler, 1999), our analysis of known-age tag re- coveries showed that sablefish avail- ability decreases with age. An al- ternate explanation is that tag loss increased with time. However, dou- ble-tagging experiments have shown that the rate of tag loss is constant with time (Beamish and McFarlane, 1988; Lenarz and Shaw, 1997). De- creased availability with age may occur because of reduced fishing ef- fort for older age fish. Fishing ef- fort is concentrated at intermediate depths (e.g., half of the recoveries occurred at depths 501-700 m [Table 2] ) but fish exit these depths as they age and move deeper. In addition, older fish were less available for IFQ management compared to open ac- cess management, which also may have been due to reduced crowding of fishing grounds during IFQ man- Maloney and Sigler: Age-specific movement patterns of Anoplopoma fimbria 313 15 20 25 30 35 Number at age 2 (millions) Figure 7 Scatterplot of proportion of sablefish ( Anoplopoma fimbria) tag recoveries in western areas (Bering Sea, Aleutian Islands, Western Gulf of Alaska) versus year-class strength (number at age 2 in millions; Hanselman et al., 2006). agement; for example, recoveries at depths >700 m were fewer during IFQ management (16% of recoveries during 2003-05) than during open access management (25% during 1992-94). This result has poten- tially substantial implications for a stock assessment because “dome- shaped” availability influences the reliability of abundance estimates (Bence et ah, 1993; Sigler, 1999). A logical next step for other research- ers to understand these effects is to complete a migratory catch-age analysis (e.g., Quinn et ah, 1990) that melds sablefish migration (Heif- etz and Fujioka, 1991) and age- structured (Sigler, 1999; Hanselman et al., 2006) analyses. Density-dependent effect on migration Migration is a prominent feature in the life history of many fishes. Well-known examples of migratory fish are Pacific salmon ( Oncorhyn - chus spp.) that return to their natal stream to spawn (Burgner, 1991; Heard, 1991) and Pacific herring ( Clupea harengus ) that consistently follow routes from spawning to feeding grounds (Hourston, 1982; Wheeler and Winters, 1984; Corten, 2002). Presumably the energy expended during migration is compensated for by other benefits. Demonstrated benefits include the ability to take advantage of seasonally available prey (Walters et al., 1986; Livingston, 1993) and avoidance of predation (Carlson, 1980). Benefits of migration for sablefish are not immediately obvious because they are opportunistic feeders and have no need to pursue specific prey, and their rapid growth rate in early life quickly lessens their vulnerability as prey. Nev- ertheless, a substantial proportion of the population migrates each year (Heifetz and Fujioka, 1991; Kimura et ah, 1998). Sablefish are characterized by great variability in year-class strength; and occasional strong year classes dominate the fishery for several years in a row (Sigler, 1999; Hanselman et al., 2006). A higher proportion of a strong year class may migrate and young fish may move farther and faster in order to occupy less crowded areas (density-dependent habitat selection; e.g., Mac- Call, 1990). Beamish and McFarlane (1988) noted dif- fering rates of sablefish movement out of release areas from 1977 to 1985 and theorized that increased den- sity resulting from recruitment of the large 1977 year class may have contributed to an increased rate of movement. However, we found no significant effect of year-class strength on the proportion of recoveries in the western areas, and therefore cohort density does not appear to affect the proportion of a cohort that will migrate. Although we tested a long time series, this time series does not span the full range of observed recruitment variability. Some earlier year classes were substantially stronger (e.g., the exceptional 1977 year class was 44% larger than the strongest year class (1984) that we tested). Migration may be stronger for year classes of such magnitude. Further, movement rates may be affected by total abundance — a possibility that could be tested in a sablefish migratory catch-age analysis such as we suggested earlier. Observed sablefish abundance trends by area during the last 25 years can be explained by their counter- clockwise migration pattern. Overall sablefish abun- dance peaked in the late 1980s and then decreased. The western areas of Bering Sea, Aleutian Islands, and western GOA decreased quickest, as migrating fish matured and turned eastward (Fig. 8). Abundance declined more slowly in the eastern GOA, presumably because fish that migrated westward returned to the eastern GOA. The abundance decline in the central GOA was intermediate, probably because migrating fish pass through in both directions (westward and eastward). This pattern of abundance changes (faster in western areas, slower in central and eastern GOA) supports the conclusion that the eastern GOA and the eastern part of the central GOA are the center of the range for Alaska sablefish (Bracken, 1983; Beamish and McFarlane, 1988; Sigler et al., 2001). Currents and sablefish migration Prevailing currents may play an important role in deter- mining the direction of migration for most young sable- 314 Fishery Bulletin 106(3) A Bering Sea, Aleutian Islands, and Western Gulf of Alaska B Central Gulf of Alaska C Eastern Gulf of Alaska Figure 8 Relative abundance (weight) from sablefish ( Anoplopoma fimbria) longline sur- veys, 1979-2004: Japan-U.S. cooperative longline survey (O) and U.S. (domestic) longline survey (A) (Hanselman et al., 2006). The values for the U.S. survey were adjusted to account for the higher efficiency of the U.S. survey gear. fish. From the time they first venture out of coastal bays onto the continental shelf of northern British Columbia or the eastern GOA, young fish are subject to northward or westward flowing currents. Driven by fresh water runoff, the Alaska Coastal Current (ACC) flows north- westward close to shore toward the head of the Gulf of Alaska (Royer, 1981). From Icy Bay at about 137°W the ACC flows 1500 km to Unimak Pass at the eastern end of the Aleutian Island chain (Stabeno et al., 2004). This inshore current is likely the initial route of most young sablefish leaving nursery areas in southeast Alaska. As the fish move westward, cross-shelf gullies and canyons provide avenues of deeper water leading to the shelf break and the upper continental slope, along which runs the westward-flowing Alaskan Stream. The potential ease of transit from the Alaska Coastal Current on the shelf to the Alaskan Stream on the upper slope may help to explain the considerable overlap in fish ages that we found within and between depths. The direction of migratory movement by young sable- fish may be influenced by prevailing current direction, but the return of adult sablefish along the continental slope to the eastern areas of the GOA is presumably made against the westward-flowing Alaskan Stream and from a lower density area to a higher one. Reed and Schumacher (1987) believed that velocities of the Alaskan Stream are low in water deeper than 300 m, and most of the fish travel within a 500-700 m depth when returning; therefore swimming against the current would not pose a problem for adult fish. The return of most adults to the eastern GOA serves to maintain the center of the population there and likely increases the chance of successful spawning in that area. Maloney and Sigler: Age-specific movement patterns of Anoplopoma fimbria 315 One factor that may make the eastern GOA and British Columbia spawning grounds more favorable is that spawning depths in these areas are closer to the coast than those farther west because of the narrow continental shelf in much of the eastern GOA. Also, the prevailing north-flowing Alaska Current in the eastern GOA may carry pelagic larvae and young fish closer inshore for easier access to coastal nursery areas. In the central and western GOA, spawning depths are farther offshore, increasing predation risk for larvae, and there is no prevailing northerly current to trans- port larvae shoreward. Instead, the Alaskan Stream, up to 100 km wide, flows westward along the shelf break, more or less perpendicular to the route that offshore-spawned larval and juvenile sablefish must travel to reach inshore nursery grounds. Tokranov (2002) believes this current is the source of periodic occurrences of juvenile sablefish off Kamchatka and the Kuril Islands. Winter current direction and sablefish recruitment success are related, and above-average recruitment is more likely in years with northerly drift (59%) than for years with an easterly or southerly drift (25%) (Sigler et al., 2001). All the sablefish in this study originated in the east- ern GOA, but young-of-the-year sablefish have been caught in small numbers on various cruises in the Ber- ing Sea, Aleutian Islands, western and central GOA, as well as the eastern GOA from 1955 to 1999 (Kendall and Matarese, 1987; Sigler et al., 2001). These observa- tions indicate the likelihood of some direct recruitment into each of these areas, in addition to recruitment resulting from migration. Spawners contributing to each area may be migrants returning to the eastern GOA, adult fish that are resident in the area, or adult fish in an adjoining upstream area whose larvae are caught up in the prevailing currents and are carried westward. Although most fish in our study older than 12 years were recovered in eastern GOA outside wa- ters, older fish (13 to 21 years) also were recovered in each of the other areas, indicating that they may have become resident in the new area at some point during migration. Our study corroborated much that is already known or suspected about sablefish migration in Alaska waters. In addition, our data on age by depth and area have refined our knowledge of sablefish movements. Further studies to locate sablefish nursery grounds throughout the GOA and in the Bering Sea and Aleutian Islands and to tag juveniles on these grounds as was done in the eastern GOA for our study would determine whether these movement patterns observed in the present study are similar to movement patterns of sablefish originat- ing in other regions of Alaska. Acknowledgments We thank all those who participated in juvenile sable- fish tagging cruises, as well as the many members of the fishing industry who have returned tags with catch information. J. Heifetz, D. Clausen, D. Hanselman, and three anonymous reviewers reviewed the paper and provided valuable comments and suggestions. Literature cited Beamish, R. J., and G. A. McFarlane. 1988. Resident and dispersal behavior of adult sablefish ( Anoplopoma fimbria) in the slope waters off Canada’s west coast. Can. J. Fish. Aquat. Sci. 45:152-164. Bence, J. R., A. Gordoa, and J. E. Hightower. 1993. Influence of age-selective surveys on the reliability of stock synthesis assessments. Can. J. Fish. Aquat. Sci. 50:827-840. Bracken, B E. 1983. Sablefish migration in the Gulf of Alaska based on tag recoveries. In Proceedings of the second Lowell Wakefield fisheries symposium, Anchorage, AK, p. 185- 190. Alaska Sea Grant Report 83-8, Univ. Alaska, Anchorage, AK. Burgner, R. L. 1991. Life history of sockeye salmon (Oncorliynchus nerka). In Pacific salmon life histories (C. Groot, and L. Margolis, eds.), p. 1-117. Univ. British Columbia Press, Vancouver, B.C., Canada. Carlson, H. R. 1980. Seasonal distribution and environment of Pacific herring near Auke Bay, Lynn Canal, southeastern Alaska. Trans. Am. Fish. Soc. 109:71-78. Corten, A. 2002. The role of “conservatism” in herring migra- tions. Rev. Fish Biol. Fish. 11:339-361. Fujioka, J. T., F. R. Shaw, G. A. McFarlane, T. Sasaki, and B. E. Bracken. 1988. Description and summary of the Canadian, Japa- nese, and U.S. joint data base of sablefish tag releases and recoveries. NOAA Tech. Memo. NMFS F/NWC- 137, 34 p. Hanselman, D. H., C. R. Lunsford, J. T. Fujioka, and C. Rodgveller. 2006. Alaska sablefish assessment for 2007. In Stock assessment and fishery evaluation report for the ground- fish fisheries of the Gulf of Alaska, p. 341-427. North Pacific Fishery Management Council, 605 W 4th Avenue, Suite 306, Anchorage, AK 99510. Heard, W. R. 1991. Life history of pink salmon (Oncorhynchus gor- buscha). In Pacific salmon life histories (C. Groot, and L. Margolis, eds.), p. 119-230. Univ. British Columbia Press, Vancouver, B.C., Canada. Heifetz, J., and J. T. Fujioka. 1991. Movement dynamics of tagged sablefish in the northeastern Pacific. Fish. Res. 11:355-374. Heifetz, J., J. T. Fujioka, and T. J. Quinn, II. 1997. Geographic apportionment of sablefish, Anoplopoma fimbria , harvest in the northeast Pacific Ocean. In Biology and management of sablefish, Anoplopoma fimbria (M. E. Wilkins, and M. W. Saunders, eds.), p. 229-237. NOAA Tech. Rep. NMFS 130, 275 p. Heifetz, J., and N. E. Maloney. 2001. Estimation of tag-reporting rates for sablefish in the northeastern Pacific Ocean. Alaska Fish. Res. Bull. 8:1-11. Hilborn, R. 1990. Determination of fish movement patterns from tag 316 Fishery Bulletin 106(3) recoveries using maximum likelihood estimators. Can. J. Fish. Aquat. Sci. 47:635-643. Hilborn, R., and M. Mangel. 1997. The ecological detective — confronting models with data, 315 p. Princeton Univ. Press, Princeton, NJ. Hourston, A. S. 1982. Homing by Canada’s West Coast herring to man- agement units and divisions as indicated by tag recov- eries. Can. J. Fish. Aquat. Sci. 39:1414—1422. Kendall, A. W. Jr., and A. C. Matarese. 1987. Biology of eggs, larvae, and epipelagic juveniles of sablefish, Anoplopoma fimbria, in relation to their poten- tial use in management. Mar. Fish. Rev. 49(1):1-13. Kimura, D. K., and J. J. Lyons. 1991. Between-reader bias and variability in the age determination process. Fish. Bull. 89:53-60. Kimura, D. K., A. M. Shimada, and F. R. Shaw. 1998. Stock structure and movement of tagged sable- fish (Anoplopoma fimbria) in offshore northeast Pacific waters and the effects of El Nino-Southern Oscillation on migration and growth. Fish. Bull. 96:462-481. Lenarz, W. H., and F. R. Shaw. 1997. Estimates of tag loss from double-tagged sable- fish, Anoplopoma fimbria. Fish. Bull. 95:293-299. Livingston, P. A. 1993. Importance of predation by groundfish, marine mammals and birds on walleye pollock Theragra chalco- gramma and Pacific herring Clupea pallasi in the eastern Bering Sea. Mar. Ecol. Prog. Ser. 102:205-215. MacCall, A. D. 1990. Dynamic geography of marine fish populations, 153 p. Univ. Washington Press, Seattle, WA. Maloney, N. E., and J. Heifetz. 1997. Movements of tagged sablefish released in the eastern Gulf of Alaska. In Biology and management of sablefish, Anoplopoma fimbria (M. E. Wilkins, and M. W. Saunders eds.), p.115-121. NOAA Tech. Rep. NMFS 130, 275 p. Maloney, N. E. 2004. Sablefish, Anoplopoma fimbria, populations on Gulf of Alaska seamounts. Mar. Fish. Rev. 66(3): 1— 12 . McCullaugh, P, and J. A. Nelder. 1983. Generalized linear models, 261 p. Chapman and Hall, New York, NY. Quinn, T. J., II, R. B. Deriso, and P. R. Neal. 1990. Migratory catch age analysis. Can. J. Fish. Aquat. Sci. 47:2315-2327. Reed, R. K., and J. D. Schumacher. 1987. Physical oceanography. In The Gulf of Alaska: physical environment and biological resources (D. W. Hood, and S. T. Zimmerman, eds.), p. 57-75. U.S. Govt. Printing Off., Washington, DC. NTIS No. PB87-103230. Royer, T. C. 1981. Baroclinic transport in the Gulf of Alaska. Part II. Fresh water driven coastal current. J. Mar. Res. 39:251-266. Rutecki, T. L., and E. R.Varosi. 1997. Migrations of juvenile sablefish, Anoplopoma fim- bria, in southeast Alaska. In Biology and management of sablefish, Anoplopoma fimbria (M. E. Wilkins, and M.W. Saunders eds.), p. 123-130. NOAA Tech. Rep. NMFS 130, 275 p. Sigler, M. F. 1999. Estimation of sablefish, Anoplopoma fimbria, abun- dance off Alaska with an age-structured population model. Fish. Bull. 97:591-603. Sigler, M. F., S. A. Lowe, and C. R. Kastelle. 1997. Area and depth differences in the age-length relationship of sablefish, Anoplopoma fimbria, in the Gulf of Alaska. In Biology and management of sable- fish, Anoplopoma fimbria (M. E. Wilkins, and M. W. Saunders eds.), p. 55-63. NOAA Tech. Rep. NMFS 130, 275 p. Sigler, M. F., and C. R. Lunsford. 2001. Effects of individual quotas on catching effi- ciency and spawning potential in the Alaska sablefish fishery. Can. J. Fish. Aquat. Sci. 58:1300-1312. Sigler, M. F., T. L. Rutecki, D. L. Courtney, J. F. Karinen, and M.-S. Yang. 2001. Young of the year sablefish abundance, growth, and diet in the Gulf of Alaska. Alaska Fish. Res. Bull. 8:57-70. Stabeno, P. J., N. A. Bond, A. J. Hermann, N. B. Kachel, C. W. Mordy, and J. E. Overland. 2004. Meteorology and oceanography of the Northern Gulf of Alaska. Cont. Shelf Res. 24:859-897. Thompson, G. G. 1994. Confounding of gear selectivity and the natural mortality rate in cases where the former is a nonmono- tone function of age. Can. J. Fish. Aquat. Sci. 51: 2654-2664. Tokranov, A. M. 2002. On the occurrence of the fry of the sablefish Anoplopoma fimbria (Pallas) (Anoplopomatidae) off Kamchatka. Russ. Acad. Sci. Oceanol. 42:117-119. Walters, C. J., M. Stocker, A. V. Tyler, and S. J. Westrheim. 1986. Interaction between Pacific cod ( Gadus macro- cephalus ) and herring (Clupea harengus pallasi) in the Hecate Strait, British Columbia. Can. J. Fish. Aquat. Sci. 43:830-837. Wheeler, J. P., and G. H. Winters. 1984. Homing of Atlantic herring (Clupea harengus haren- gus) in Newfoundland waters as indicated by tagging data. Can. J. Fish. Aquat. Sci. 41:108-117. Zar, J. H. 1984. Biostatistical analysis, 2nd ed., 718 p. Prentice Hall, Inc. Englewood Cliffs, NJ. 317 Distribution of red deepsea crab (Chaceon quinquedens ) by size and sex in the Gulf of Mexico Morgan J. Kilgour (contact author) Thomas C. Shirley Email address for M. J. Kilgour: Morgan.Kilgour@tamucc.edu Harte Research Institute for Gulf of Mexico Studies Texas A&M University-Corpus Christi 6300 Ocean Drive, Unit 5869 Corpus Christi, Texas 78412-5869 The red deepsea crab ( Chaceon quinquedens (Smith, 1879)) has sup- ported a commercial fishery off the coast of New England since the 1970s (Wigley et al., 1975) and has had annual harvests from 400 metric tons (t) (1996) to 4000 t (2001) (NEFMC, 2002). In 2002, a fishery management plan for the northeast fishery on the Atlantic coast was implemented and total allowable catch was reduced to approximately 2500 t (NEFMC 2002). Although there are manage- ment plans for the golden crab (C. fenneri) and the red deep sea crab for Atlantic coast regions, there is no fishery management plan for red deepsea crabs in the Gulf of Mexico. Successful management for sustain- able harvests should be based on a knowledge of the life history of the species, but C. quinquedens has been a difficult species for which to obtain life history and abundance informa- tion because of its deep distribution. Previous studies used trap surveys to estimate the relative abundance and distribution of C. quinquedens and almost all had similar results (Wigley et ah, 1975; Stone and Bailey, 1980; Lockhart et al., 1990; Lindberg and Lockhart, 1993; Waller et al., 1995; Trigg et al., 1997). Typically, C. quinquedens is found at depths from 100 to 2000 m (Haefner, 1978; Hastie, 1995) in the western Atlantic Ocean and the Gulf of Mexico (Hae- fner and Musick, 1974). The largest populations of C. quinquedens are found on soft sediments along the continental slope (Lindberg and Lock- hart, 1993). Red deepsea crabs typi- cally are found in progressively shal- lower depths as they increase with size (Wigley et al., 1975; Trigg et al., 1997), and this finding indicates that the megalopae, after settling in deeper water, move up the continen- tal slope as they grow. Bathymetric sexual segregation has been reported for many of the Chaceon species, in- cluding C. quinquedens (McElman and Elner, 1982; Lockhart et al., 1990). Females typically are found in shallower waters than those that males inhabit (Wigley et al., 1975; Lockhart et al., 1990), although fe- males have also been found at depths similar to those of males (Wenner et al., 1987). In previous studies, C. fenneri and C. quinquedens were found to have similar distributions. However, most studies in which this comparison was made had relatively larger collections of C. fenneri than C. quinquedens (Lockhart et al., 1990; Lindberg and Lockhart, 1993). The two congeners may be expected to have similar dis- tribution patterns because of their ge- netic and morphological similarities; however, assumptions made about the distribution of C. quinquedens based on the distribution of C. fenneri may lead to inappropriate management strategies. Similarly, populations of crabs in the Gulf of Mexico and Mid- dle Atlantic Bight may use habitat differently and further investigation is warranted (Lockhart et al., 1990). Depth limitations of prior surveys have indicated that juveniles may not have been sampled, as evidenced by the depth-limited NEFSC (North- east Fisheries Science Center) 400 m surveys from 1964 to 1999 where fewer than six crabs smaller than 7- cm carapace width were captured per tow (Steimle et al., 2001). Chaceon quinquedens is closely re- lated to C. fenneri (Weinberg et al., 2003). However, C. fenneri is tan and found at shallower depths than the red C. quinquedens (Lockhart et al., 1990; Lindberg and Lockhart, 1993). Commonly, when one species of Cha- ceon is studied, data are extrapo- lated to other congeners because of a limited sample of those congeners or because data have not yet been collected for the congeners. Most of the life history patterns that have been attributed to C. quinquedens have been attributed because of its close relationship with C. fenneri. Little or no genetic difference be- tween C. fenneri and C. quinquedens populations was found in the east- ern Straits of Florida and the Gulf of Mexico (Weinberg et al., 2003). The hypothesis was that the distribution of C. quinquedens would be similar to the distribution observed for C. fenneri in previous studies. To ex- amine this hypothesis, we compared the size distribution and sex distribu- tion of C. quinquedens near six ship- wrecks at varying depths in the Gulf of Mexico. Because fish communities around these shipwrecks were being examined at the time, we took the opportunity to explore red deepsea crab distributions. Materials and methods From July 29 to August 16, 2004, six World War II era shipwrecks were surveyed with the remotely operated vehicle (ROV) XL-11 deployed from the HOS Dominator in the Gulf of Mexico. The shipwrecks were located Manuscript submitted 2 July 2007. Manuscript accepted 19 February 2008. Fish. Bull 106:317-320 (2008). The views and opinions expressed or implied in this article are those of the author and do not necessarily reflect the position of the National Marine Fisheries Service, NOAA. 318 Fishery Bulletin 106(3) 94°0’0"W 92WW 90°0'0"W 88WW 86”0'0"W The locations of the shipwreck sites surveyed in the Gulf of Mexico with the XL-11 ROV. Red deepsea crabs ( Chaceon quinquedens ) were found at the four deeper sites (Gulf Penn, U-166, Robert E. Lee, and Alcoa Puritan). in the Mississippi Canyon or adjacent areas (Fig. 1) and varied in water depth from 90 to 1970 m. At each wreck, ROV transects were conducted over the wreck and over the debris field. Once transects were completed, traps were set and biological sampling with the ROV began. Transects over the wreck and in parallel lines adjacent to the wreck until about 300 m were mapped before the deployment of the ROV. The ROV and trap sampling time lasted a minimum of eight hours and a maximum of sixteen hours. At each wreck, both a small baitfish trap (95 cm width x 75 cm length x 50 cm height) and a chevron fish trap (150 cm width x 180 cm lengthx60 cm height) baited with herring and squid were set within a few meters of the wreck and approximately 300 m away from the wreck. A baited crab trap (64 cm width x 97 cm length x 51 cm height) was placed near the wreck, and two minnow traps (23 cm diameter x 44 cm length) bound together were placed far from the wreck; both were baited with canned cat food and biolume sticks. Additional collections were made at the direction of biologists using the manipulator arm of the ROV for larger invertebrate specimens and the suction hose of the ROV (approximately 10 cm in diameter) for smaller invertebrate specimens. Each crab specimen collected was measured from the anterolateral spine on one side of the carapace to the anterolateral spine on the opposite side for carapace width (CW) to the nearest 0.1 mm with vernier calipers. Shell condition and missing appendages were noted and sex was recorded. If the crab was female, presence and color of eggs were also recorded. Voucher specimens from each wreck were either preserved in 90% ethanol or frozen. Statistical analyses were performed with SPSS 14.0 ( SPSS, Inc., Chicago, IL) software to determine if crab size or sex was related to the distribution of the crabs. To determine if there was a correlation with size and depth, a linear regression was performed with a 95% confidence interval. To examine the relationship of sex with depth, the Kruskal-Wallis test was used because the data were non-normal. Data were also non-normal for depth distribution, but linear regression was used to determine if a linear correlation existed. Results Chaceon quinquedens was found at the four deepest of the six shipwrecks, and there was no linear distribu- tion of the crabs by sex. A total of 127 specimens was collected at the four sites, with CW ranging from 21.2 mm to 141.1 mm (Table 1). No significant correlation occurred between depth and crab size (R2- 0.051). The smallest crabs were found at the shallowest depth, and NOTE Kilgour and Shirley: Distribution of Chaceon quinquedens by size and sex in the Gulf of 319 Table 1 Carapace width (mm) and numbers of females and males of red deepsea crab Chaceon quinquedens found on four shipwrecks in the Gulf of Mexico. Shipwreck Depth (m) Carapace width (mm) Sex Average Minimum Maximum Female Male Gulf Penn 533 24.1 21.2 31 0 4 Robert E. Lee 1428 95.0 56.5 141.1 27 21 U-166 1428 116.0 61.5 136.5 18 6 Alcoa Puritan 1950 66.5 43.7 120.3 22 26 the largest crabs were found at intermediate depths. The four smallest specimens were collected with the suction arm of the ROV (all at the Gulf Penn shipwreck) and one crab was collected with the manipulator arm of the ROV. All other samples were collected in the fish traps. No crabs were collected in the minnow traps, and crabs were rarely observed on the shipwrecks. Although slightly more than half of the crabs were collected close to the shipwrecks (74 ±7 m), insufficient data were collected to compare wreck effects on crab size or sex distribution. Males and females did not have significantly different distributions. A total of 67 females and 57 males was collected (three of unknown sex collected at the U-166 site) and no depth pattern could be attributed to sex (Table 1; P=0.059). Discussion Previous reports of the distribution of C. quinquedens have been contradictory to our findings, which were influenced by the addition of a different sampling tech- nique, the ROV. Without the ROV, the four smallest crabs would not have been observed or collected at the shallowest depth. Red deepsea crabs are typically found on silt substrate, and although their abundance on the shipwrecks was not analyzed in our study, we noted that crabs were rarely found on the shipwrecks (see also Kilgour, 2007). A strong correlation of crab size with depth has been reported in some studies of C. quinquedens (Wigley et al., 1975; Lockhart et ah, 1990; Lindberg and Lock- hart, 1993; Waller et al., 1995; Trigg et al., 1997), but not in others (Wenner et ah, 1987). The variation in results may have been caused either by a temporal variation in sampling, or a sampling bias from gear. In our study, the smallest crabs were found at the shallowest site and this result was contrary to that of previous reports (Lockhart et al., 1990; Lindberg and Lockhart, 1993; Waller et al., 1995; Trigg et al., 1997). A linear relationship between crab size and depth was not observed in our study, but this result may have been due to the ROV collections which were composed of the smallest crabs; crabs of a similar small size were not collected with traps at the same site. An almost equal ratio of females to males occurred in our study and our sampling did not occur during the reported mating season (Wigley et al., 1975; Haefner, 1978; Biesiot and Perry, 1995). Female C. quinquedens were found only at the three deepest sites, whereas males were found at four of the six sites; this find- ing contradicts that of previous studies where depth segregation by sex was observed (Wigley et al., 1975; Lockhart et al., 1990). Sampling methods to date may have been inadequate because the use of traps has been the primary method for obtaining information for this species. Sampling biases often result in more males being collected, particularly when females are oviger- ous, because females tend to avoid traps while brooding eggs (Howard, 1982; McDonald et al., 2004; Taggart et al., 2004). Additionally, sampling bias exists for larger males and for smaller males and females; smaller crabs are less likely to enter traps when large males are pres- ent (Taggart et al., 2004). Chaceon quinquedens is diffi- cult to study because of its deep distribution; techniques other than trap and trawl sampling may be necessary to accurately understand the movements, spatial distribu- tion, and life history of this species. The bathymetric distribution of many deepwater crab species is suspect because the methods by which these patterns were determined may be biased. We used two different techniques, an ROV and traps; the bathymetric distribution of C. quinquedens was different from that of previous studies where a single sampling technique was used. The best data available are used for fishery management, but many times these data represent sampling techniques that may not provide a complete distribution of the species in question. Our study, in spite of a small sample size, demonstrates the impor- tance of using more than one technique for observing the life history and distribution of a deepwater species to reduce the biases of sampling techniques. Both the ROV and the trap surveys have sampling biases, but in combination, may reveal a more accurate picture of the bathymetric distribution of red deepsea crab. The use of multiple sampling gears to gain knowledge of the life history of deepwater crabs is necessary. 320 Fishery Bulletin 106(3) Acknowledgments We thank A. Baldwin, University of Alaska Fairbanks, for his efforts during the sampling and identifica- tion of red deepsea crabs. This project was funded by Minerals Management Service Contract 1435-01-03- CT-73095 (M03PC00012); National Oceanic and Atmo- spheric Administration, Office of Ocean Exploration; the National Oceanic Partnership Program; and the Harte Research Institute. We thank our collaborators at sea on the Deep Gulf Shipwrecks of World War II Survey and members of the Sonsub crew and HOS Dominator. Literature cited Biesiot, P. M., and H. M. Perry. 1995. Biochemical composition of the deep-sea red crab Chaceon quinquedens (Geryonidae): organic reserves of developing embryos and adults. Mar. Biol. 124:407- 416. Haefner, P. A. 1978. Seasonal aspects of the biology, distribution and relative abundance of the deep-sea red crab Geryon quinquedens Smith, in the vicinity of the Norfolk Canyon, western North Atlantic. Proc. Natl. Shellfish. Assoc. 68:49-61. Haefner, P. A., and J. A. Musick. 1974. Observations on distribution and abundance of red crabs in Norfolk Canyon and adjacent continental slope. Mar. Fish. Rev. 36:31-34. Hastie, L. C. 1995. Deep-water geryonid crabs: a continental slope resource. Oceanogr. Mar. Biol. Annu. Rev. 33:561- 584. Howard, A. E. 1982. The distribution and behaviour of ovigerous edible crabs ( Cancer pagurus ) and consequent sampling bias. ICES J. Mar. Sci. 40:259-261. Kilgour, M. J. 2007. Bathymetric and spatial distribution of deca- pod crustaceans on deep shipwrecks in the Gulf of Mexico. M.S. thesis, 60 p. Texas A&M Univ. — Corpus Christi, TX. Lindberg, W. J., and F. D. Lockhart. 1993. Depth stratified population structure of geryonid crabs in the eastern Gulf of Mexico. J. Crustacean Biol. 13:713-722. Lockhart, F. D., W. J. Lindberg, N. J. Blake, R. B. Erdman, H. M. Perry, and R. S. Waller. 1990. Distributional differences and population simi- larities for two deep-sea crabs (family Geryonidae) in the northeastern Gulf of Mexico. Can. J. Fish. Aquat. Sci. 42:2112-2122. McElman, J. F., and R. W. Elner. 1982. Red crab (Geryon quinquedens) trap survey along the edge of the Scotian Shelf, September 1980. Can. Tech. Rep. Fish. Aquat. Sci. 1084:1-12. McDonald, P. S., G. C. Jensen, and D. A. Armstrong. 2004. Between a rock and a hard place: the ecology of ovigerous green crab, Carcinus maenas (L.), with emphasis on implications for monitoring and control efforts. J. Shellfish Res. 23:657. NEFMC (New England Fishery Management Council). 2002. Fishery management plan for deep-sea red crab ( Chaceon quinquedens) including an environmental impact statement, an initial regulatory flexibility act analysis, and a regulatory impact review: vol. I, 446 p. New England Fishery Management Council, 50 Water St., Mill 2, Newburyport, MA 01950. Steimle, F. W., C. A. Zetlin, and S. Chang. 2001. Essential fish habitat document: red deepsea crab, Chaceon (Geryon) quinquedens, life history and habitat characteristics. NOAA Tech. Memo. NMFS-NE-163, 36 p. Stone, H., and R. F. J. Bailey. 1980. A survey of the red crab resource on the conti- nental slope, N. E. Georges Bank, and western Scotian Shelf. Can. Tech. Rep. Fish. Aquat. Sci. 977:1—9. Taggart, S. J., C. E. O’Clair, T. C. Shirley, and J. Mondragon. 2004. Estimating Dungeness crab ( Cancer magister ) abun- dance: crab pots and dive transects compared. Fish. Bull. 102:488-497. Trigg, C., H. Perry, and W. Brehm. 1997. Size and weight relationships for the golden crab, Chaceon fenneri, and the red crab, Chaceon quinque- dens, from the eastern Gulf of Mexico. Gulf Res. Rep. 4:339-343. Waller, R., H. Perry, C. Trigg, J. McBee, R. Erdman, and N. Blake. 1995. Estimates of harvest potential and distribution of the deep sea red crab, Chaceon quinquedens, in the north- central Gulf of Mexico. Gulf Res. Rep. 9:75-84. Weinberg, J. R., T. G. Dahlgren, N. Trowbridge, and K. M. Halanych. 2003. Genetic differences within and between species of deep-sea crabs ( Chaceon ) from the North Atlantic Ocean. Biol. Bull. 204:318—326. Wenner, E. L., G. F. Ulrich, and J. B. Wise. 1987. Exploration for golden crab, Geryon fenneri, in the South Atlantic Bight: distribution, population structure, and gear assessment. Fish. Bull. 85:547-560. Wigley, L. G., R. B. Theoroux, and H. E. Murray. 1975. Deep-sea red crab, Geryon quinquedens, survey off north-eastern United States. Mar. Fish. Rev. 37:1-21. 321 Catches in ghost-fishing octopus and fish traps in the northeastern Atlantic Ocean (Algarve, Portugal) Karim Erzini (contact author) Luis Bentes Rui Coelho Pedro G. Lino Pedro Monteiro Joaquim Ribeiro Jorge M. S. Goncalves Email address for K. Erzini: kerzini@ualg.pt Centro de Ciencias do Mar (CCMAR), Universidade do Algarve, 8005-139 Faro, Portugal Ghost fishing is the term used to describe the continued capture of fish and other living organisms after a fisherman has lost all control over the gear. Traps may be lost for a variety of reasons including theft, vandalism, abandonment, interactions with other gear, fouling on the bottom (i.e., traps and ropes are caught on rocky sub- strate), bad weather, and human error (Laist, 1995). Annual trap loss can be as high as 20% to 50% of fished traps in some fisheries (Al-Masroori et ah, 2004). Because lost traps can con- tinue to fish for long periods, albeit with decreasing efficiency over time (e.g., Smolowitz, 1978; Breen, 1987, 1990; Guillory, 1993), ghost fishing is a concern in fisheries worldwide. Few studies on the ghost fishing of lost traps have been carried out in European waters, and there has been no information from southern Euro- pean waters. Ghost fishing of parlour pots used to catch lobsters and crabs off the south-west coast of the United Kingdom was studied by Bullimore et al. (2001), and Godpy et al. (2003) carried out an experimental study on much larger, deliberately lost pots for red king crab ( Paralithodes camts- chaticus ) in Norwegian waters. In both cases the effect of ghost fish- ing by parlour pots was deemed to be relatively small compared to the effects of other types of traps used in Canadian and American fisheries (Brown and Macfadyen, 2007). In southern Portugal, pots and traps of various types are among the most widely used gears in the small- scale fisheries. Fishing vessels <9 m (local category) can legally fish up to 500 traps, and coastal category vessels (9-12 m and >12 m in total length) are allowed up to 750 and 1000 traps, respectively. The most widely used traps in the Algarve are 1) metal frame, hard plastic netting, single entry traps for octopus (covo), 2) large, metal frame traps for catch- ing cuttlefish and fish (armadilha), and 3) wire traps (murejona) for catching fish. However, only the covo traps and murejona traps were used in our study. Under the Common Fisheries Pol- icy and the European Community directive on habitats and species, member states are responsible for lo- cal fisheries and are obliged to take measures to minimize or mitigate the negative effects of fishing activ- ity. Concern over the effects of lost gear in European waters has led the European Commission to finance two pan-European projects on ghost fish- ing. The first project focused only on gill nets and trammel nets (Erzini et al., 1997), and the second project in- cluded studies on lost traps in several European areas (Godpy et al., 2003). Here we report the results from one of the studies carried out with two types of traps in the northeast Atlan- tic (south coast of Portugal) (Fig. 1). The catches of deliberately lost traps were monitored and estimates of the number of trap losses and causes of trap losses were obtained through surveys of commercial fishermen. Materials and methods Catches in deliberately lost traps The main gear used to catch octopus is the octopus trap (covo), a small metal framed trap with a single entrance on the top (Fig. 2). To make escape- ment more difficult, the entrance is partially blocked by plastic strips that are easy to push through when enter- ing the trap but not when exiting. A total of 60 octopus traps, each baited with two sardines, were deployed on August 11, 1999, at two sites off Faro where normal fishing activities with octopus traps takes place. The depth at one site was 20 m, and 50 m at the other, and both were situated near rocky reefs. At each location 30 traps were deployed, 15 on soft bottom and 15 on rocky bottom. Because the traps set at 50 m were difficult to retrieve with a grapnel from the hard bottom or were all lost within one month after deployment (on soft bottom), an addi- tional 30 octopus traps were deployed at the shallower depth on soft and hard bottom on 18 May, 2000 and were monitored weekly for 14 weeks. In addition to the 90 octopus traps, 10 fish traps of the murejona type were also deployed on 25 May 2000 at the shallower site (20 m depth) and monitored by scuba divers on a weekly basis for three months. Murejona traps are round, wire traps with a single funnel-shaped Manuscript submitted 9 July 2007. Manuscript accepted 25 March 2008. Fish. Bull. 106:321-327 (2008). The views and opinions expressed or implied in this article are those of the author and do not necessarily reflect the position of the National Marine Fisheries Service, NOAA. 322 Fishery Bulletin 106(3) 9°30'W 9°00'W 8°30'W 8°00W 7°30'W 7°00'W Figure 1 Map of the Algarve region and the Barlavento and Sotavento areas where the catches of deliberately lost traps of two types were quantified in 1999 and 2000, and where information on the numbers of traps lost by com- mercial fishermen and the reasons for trap loss were obtained by means of questionnaire surveys. opening at the top (Fig. 3). Murejonas targeting sea breams (Sparidae) were baited with approximately 0.5 kg of crushed common cockle ( Cerastoderma edule). The octopus traps located in shallow waters were monitored by scuba divers using slates, video, and still photography cameras. Acoustic pingers, an acoustic receiver, and a GPS differential antenna were used to aid divers in locating the experimental traps. Data recorded consisted of the number of the trap, number and identification of the species captured, as well as an estimate of the total length of each individual caught. In order to estimate the total catch (numbers of fish), traps were also inspected for remains of fish that might have died or been eaten while inside the traps. The structural integrity of the traps was evaluated by divers one year after their deployment. The catches were analyzed in terms of target vs. nontarget and prey (both target and non- target) vs. predator species. The target spe- cies were common octopus ( Octopus vulgaris ) for the octopus traps, and Sparidae (axillary seabream ( Pagellus acarne ), common pandora (P. erythrinus), striped seabream ( Lithognathus mormyrus), annular seabream ( Diplodus annu- laris), Senegal seabream ( D . bellottii), common seabream (D. sargus), two-banded seabream (D. vulgaris), black seabream ( Spondylioso - ma cantharus), and blotched picarel ( Spicara maena) for the murejona fish traps. Conger eel ( Conger conger ), Mediterranean moray eel ( Muraena helena), forkbeard ( Phycis phycis), and O. vulgaris were considered predator spe- cies that would feed on trapped small fish and in the case of conger and moray eels, also on octopus. Bail basket Entrance 40 cm Bridle 44 cm Figure 2 View from the top of an iron frame, plastic mesh octopus trap (covo) of typical dimensions (40 x 44 x 25 cm) and mesh size (4-cm plastic square mesh). NOTE Erzini et a!.: Catches in ghost-fishing octopus and fish traps in the northeastern Atlantic Ocean 323 = catch in numbers per trap haul; - soak time in days; and are parameters to be estimated. Bridle 1 00 cm Figure 3 View from the top of a wire fish trap (murejona) showing the funnel-shaped entrance. The diameter is 100 cm, the height 35 cm, and the sides of the triangular wire meshes are approximately 2.5 cm. For each trap type, the Zhou and Shirley (1997) model for the relationship between catch and soak time for baited traps where escapement is possible was fitted by nonlinear least squares regression to the catch-per-trap data with PROC NLIN software (SAS Insti- tute Inc., Cary, NC.) and the equation: C(t) = ab + a(t - b)e~ct, where C(t) t a, b, and c For this model, catch is zero at t = 0, the asymptotic catch after an infinite soak time is the product ab , and maximal catch Cmax is attained at a soak time of fmax = 1/c + 6: Cmnr - ab + ac~1e~(1+cb> . max Quantification of trap loss Questionnaires were used to survey fishing- boat skippers at ten ports of the Algarve, South of Portugal. The questionnaires were divided between the following areas (area — Barlavento, western Algarve and Sotavento, eastern Algarve) and by port and fishing vessel (local or coastal). The question- naires were designed to quantify the number and type of traps used, the number lost per year, reasons for loss, and the degree of success in recovery attempts. Results Catches in deliberately lost traps Lost octopus traps caught six species: O. vulgaris, C. conger, M. Helena, red scorpionfish ( Scorpaena notata), comber ( Serranus cabrilla), and P. phycis. Catch rates were generally low and highly variable (Fig. 4). Most octopus were captured in the first two weeks after trap deployment, and few catches were observed thereafter. For other fishes, namely small red scorpionfish, occa- sional catches were recorded up to three months after deployment. The estimated parameters of the Zhou and Shirley (1997) model were a - 3.8576, b = 0.0318, and c = 2.292. Based on these parameters the maximal catch is attained within 24 hours after deployment (0.47 days), and the asymptotic catch rate is 0.12 individuals per trap. In addition to all six species caught by the octopus traps, fish traps caught damselfish ( Chromis chromis), Mediterranean rainbow wrasse (Coris julis), D. annularis, D. bellottii, D. vulgaris, S. cantharus, Baillon’s wrasse ( Symphodus bailloni), and axillary wrasse (S. mediterra- neus) and a maximum diversity of 10 species was attained 27 days after deployment. The most abundantly caught species was D. vulgaris that accounted for 43% of the fish observed in the traps, followed by D. bellottii (16%). Although most of the species were small, some larger fish, namely C. conger, were also found in the traps. The mean number of individuals per trap peaked approximately two weeks after deployment and was followed by a sharp decrease from week 4 to 5, and then averaged approximately one fish per trap up to the end of the three month monitoring period (Fig. 5). The estimated maximal catch, based on the parameters of the Zhou and Shirley (1997) model (a=1.5397, 6 = 0.5669, and c = 0.1101) occurred 9.7 days after deployment, and the asymptotic catch rate was 0.87 individuals per trap. The same pattern of an initial increase in catches, fol- lowed by a decline, was seen in the catches of the most abundantly caught species ( D . vulgaris) in individual fish traps (Fig. 6). The fish trap predator-to-prey ratio, with predators considered to be C. conger, O. vulgaris, M. Helena, and P. phycis, showed an opposite trend, increasing sharply from week 4 to 5 to a maximum of 2.0 35 days after deployment, then leveling off (Fig. 5). The initial high number of fish observed in the fish traps was largely due to the presence of the target species (Sparidae), whereas the predators, especially the three fish species C. conger, M. Helena, and P. phycis were relatively more abundant 55, 71, and 89 days after deployment. Whereas the iron frame octopus traps retained their structural integrity 12 months after deployment, the wire fish traps were completely destroyed. Quantification of trap loss A total of 84 interviews were conducted, representing 19.4% of the boats registered in the Algarve (southern 324 Fishery Bulletin 106(3) O 06 0.5 -0 0.4 0.3 0.2 01 - 0 0 □ = Octopus 06 A = Other o.5 • = All species combined o.4 H " 0 3 I) a £3 A — B- u 0.2 i 0.1 0 A -a- r a C(t)= ab + a(t-b)e 0 10 20 30 40 50 60 70 80 90 100 Days after deployment (t) A itB □ 10 20 30 40 50 60 70 Days after deployment (t) 80 90 100 Figure 4 Octopus trap catch rates (mean number per trap) over time. Inset figure is the fitted catch model of Zhou and Shirley (1997) with a = 3.8576, b = 0.0318, and c = 2.292. 1 7 14 27 35 43 55 71 89 Days after deployment (t) Figure 5 Murejona catch rates (C) (mean number per trap) over time. Mean ±SE (stan- dard error) number of fish and octopus per trap, and the predator-to-prey ratio. Predators were conger eel ( Conger conger), forkbeard (Phycis phycis), Mediterranean moray eel ( Muraena Helena), and common octopus ( Octopus vulgaris ), whereas prey were all other finfish species. Inset figure is the fitted catch model of Zhou and Shirley (1997) with a = 1.5397, b = 0.5669, and c = 0.1101. region of Portugal) with licenses for fishing with traps. Of these, 13 boats had to be excluded from the survey because traps had not been used during the past year. Thus, questionnaire surveys were completed for 71 fish- ing boats that had been used to fish with traps. The results of the questionnaire survey are summarized in Tables 1 and 2. All skippers surveyed that had fished with octo- pus traps had the particular type of small trap (covo) used to catch octopus. However, some of the boats also possessed other types of traps, generally of a larger size that were used to target other species. Thus, 16 (22.5 %) of the skippers interviewed had also used larger traps, mostly to catch cuttlefish, and two (2.8 %) of the skippers from the western area (Barlaven- to), had used murejona wire fish traps to capture fish, especially sea breams. These results confirmed the relative importance of covo-style traps as a gear. NOTE Erzini et al.: Catches in ghost-fishing octopus and fish traps in the northeastern Atlantic Ocean 325 Table 1 Summary of survey information collected from 84 interviews with skippers: mean depth fished, mean number of traps used by the different fleets (local and coastal) in the two areas (Barlavento and Sotavento), and mean numbers of traps lost per year per fishing vessel. SD = standard deviation. Mean number ( ± SD ) of traps fished Mean number (±SD) of traps lost Fleet region and area Mean (±SD) depth (m) fished Octopus trap Cuttlefish trap Fish trap Octopus trap Cuttlefish trap Fish trap Local (<9 m) Barlavento Sotavento 19.1 ±5.7 21.1 ±5.0 270.3 ±200.5 644.4 ±261.7 149.0 ±145.2 112.5 ±75.0 190 ±7.1 30.9 ±55.5 145.6 ±102.2 78.8 ±147.5 13.5 ±11.1 13.5 ±10.6 Coastal (>9m) Sotavento 25.0 ±5.0 903.8 ±227.7 80.0 318.5 ±207.8 10.0 Table 2 Estimates of numbers of octopus traps lost per year off the coast of Barlavento and Sotavento of southern Portugal based on national statistics and questionnaire surveys. Fleet was separated into a local and coastal category. Number of licenses was the number of trap licenses issued (from national statistics). Number fishing was the number of boats fishing traps (from national statistics), Mean number of traps/boat was the mean number of traps fished per year (determined from questionnaires). Total number of traps in use was the product of the mean number of traps fished and the number of fishing vessels. Proportion of traps lost was estimated from the number of traps lost (from questionnaires) divided by the number of traps fished. Traps lost per year was the product of the proportion lost and the total number used. Fleet Area Number of Licenses Number of boats fishing Mean number of traps/boat Total number of traps used Proportion of traps lost Number of traps lost per year Local (<9 m) Barlavento 190 161 270.3 43,518 0.11 4975 Sotavento 103 87 644.4 56,063 0.23 12,667 Coastal (>9 m) Barlavento 58 49 995 48,755 0.21 10,437 Sotavento 91 77 903.8 69,593 0.35 24,525 Total 442 374 2813.5 217,929 0.24 52,604 Although fish traps are relatively less impor- tant compared to the octopus traps, they are re- stricted to a particular use by the Algarve fishing fleet. The use of large fish traps in the Barlavento area is favored because of the hard bottom where there are larger concentrations of fish. The aver- age numbers of traps used per boat for the three types of traps commonly used in the fishery, by port category (local or coastal) and coastal zone area, are given in Table 1. The octopus traps are by far the most common of all the traps used. We estimated that 52,604 octopus traps were lost in Algarve waters in 2000, with the coastal fleet accounting for more losses than the local fleet, and higher losses in the Sotavento than in the Barlavento area (Table 2). Regarding the big traps used mostly to catch cuttlefish the local fleet lost more such traps than the coastal fleet, and there were more losses in the Sotavento than in the Barlavento area. 0 10 20 30 40 50 60 70 80 90 100 Days after deployment Figure 6 Number of two-banded sea bream ( Diplodus vulgaris ) observed in eight different traps over three months. Each symbol rep- resents the catches of one trap. 326 Fishery Bulletin 106(3) The most important cause for the loss of traps was interaction with other gears (41%), followed by bad weather (39%), and fouling on rough bottom (18%). Skippers also indicated that gear loss could be caused by other factors (2%), especially theft. The main rea- son for trap loss in the local fishery was interference with other gears (42.6%) and fouling on rough bottom (42.4%) in the Sotavento and Barlavento areas. In the case of the coastal fishery, the main reasons for trap loss were bad weather (40.4 %) in the Sotavento area and interference with other gears in the Barlavento area (40.0 %). Discussion In comparison to the octopus traps, fish traps caught a greater variety of species and the average catch per trap (in the period of days to weeks after deployment) was much greater. Groups of individuals of the same species of Sparidae were recorded in the same trap, often on subsequent monitoring dates, indicating that escape- ment rates were low or that individuals that died or escaped were replaced by conspecifics (Bullimore et al., 2001). Abrasions on the head and snout from attempts to escape through the wire mesh also indicated that escapement rates were probably low (Bullimore et al., 2001; Al-Masroori et al., 2004). There was a succes- sion in the capture of species; there were initially high catches of the target sea bream species, followed by the entry of larger predator-type species such as conger eel and fork beard. The predators were probably attracted by the smaller prey species within the trap, and the same individual predators were observed in the traps over weeks and in some cases for more than a month. There have been relatively few studies on fish escape- ment rates from traps, and comparisons have generally not been possible because of differences in trap design and size. Munro (1974) reported that escapement from Antillean fish traps used in the Caribbean averaged 11.6% per day. Scarsbrook et al. (1988) reported a 0% escapement rate for sablefish ( Anoplopoma fimbria). Al- Masroori et al. (2004) assumed a 10% escapement rate from large, single opening wire traps in Oman, and a 95% mortality rate for ghost-fishing traps. Given the design of the fish traps, our own observations of trapped fish, and the typical escapement rates reported in the literature, we believe that ghost fishing mortality rates of fish in the murejona traps are high and are caused by predation in the trap or are the result of injuries and starvation. On the other hand, we assume that octopus escapement rates were 100%. There may have been some trap-related mortality caused by predation because octopus require several minutes to exit a trap through the mesh and are therefore susceptible during that time to the attack of a moray eel or conger eel inside the trap. Catches in octopus traps decline sharply 24 hours after deployment, whereas fish trap catches peak one to two weeks after deployment, and long after the bait has been consumed or has deteriorated. Rapid consumption of bait has been supported by the findings of Castro et al. (2005), who reported that fish discards in this region are completely scavenged within 24 hours, and by the general knowledge that octopus fishermen must rebait their traps frequently. Optimal trap soak times of days or even weeks with asymptotic catch rates have been reported in a num- ber of studies (Munro, 1974; Mahon and Hunte, 2001; Al-Masroori et al., 2004). Typically, as seen with our fish traps, catches tend to decline and stabilize at low rates for long soak times. Munro (1974) reported that for long soak times, catch rates in Antillian fish traps stabilized at the point where daily escapement equaled daily ingress. Based on the relationship between rates of ingress, escapement, catch, and soak time, a variety of models have been used to model trap catches over time (Fogarty and Addison, 1997; Zhou and Shirley, 1997; Al-Masroori et al., 2004). The Zhou and Shirley (1997) is the only model where catches increase to a maximum of days or weeks after deployment and then decline, stabilizing at a low level. This model gave a good fit to the murejona data, where catches peaked two weeks after trap deploy- ment, and then stabilized at a mean of approximately one fish per trap. Octopus trap catches also stabilized at very low catches per trap, but were highest 24 hours after deployment. A simple exponential model (Al-Mas- roori et al., 2004) adequately describes the catches over time but does not model the low residual catches. Thus, we opted to use the Zhou and Shirley (1997) model for the octopus trap data as well. The results of the questionnaire survey showed that interaction with other gears (gear conflict) was the most important cause of trap loss. The large number of traps (often deployed without buoys at the surface to avoid theft) within a limited area where many other fishing vessels are operating simultaneously, coupled with long soak times, may explain these results. From our experience, fishermen who catch a longline of traps in their own gear often will simply cut the lines to dis- entangle the gears. Thus, the traps are often cut loose but fall close to where they had been fishing. The other major cause of trap loss was bad weather, often lead- ing to the loss of entire longlines of traps. This cause is particularly important for the larger coastal vessels, which tend to fish further from their homeports and in deeper waters. Given the fact that fishing with traps in the Algarve takes place in relatively shallow water, underwater sur- veys with divers are an appropriate method for monitor- ing catches in deliberately lost traps and for quantifying gear loss. Despite the problem of the loss of traps due to bad weather and interaction with commercial gear, it is possible to monitor both octopus and fish traps for prolonged periods. The use of divers permits the moni- toring of traps and their catches without disturbance. This method is vital for understanding trap catch dy- namics and the changes in catches after the bait used to attract fish and cephalopods is no longer present in NOTE Erzini et al.: Catches in ghost-fishing octopus and fish traps in the northeastern Atlantic Ocean 327 the traps. However, in order to be able to fully evalu- ate the effects of ghost fishing from the large number of traps that are lost each year in the coastal waters of southern Portugal, it will be necessary to investigate escapement rates, and to estimate mortality rates. Such investigations can be done by tagging trapped fish and monitoring their escapement and survival by divers. Bycatch and ghost fishing mitigation measures for traps generally involve the use of escape mechanisms and the use of degradable materials (e.g., Scarsbrook et al., 1988). In the case of the fish traps, mitigation op- tions are limited because the entire trap is made from wire and the trap door is on the bottom of the trap. Octopus traps have a hatch that can be attached with degradable material and the plastic netting could be replaced with biodegradable netting. However, perhaps the most important measure to reduce mortality would be the implementation of a code of conduct leading to less gear loss from gear interaction and theft. Acknowledgments This work was funded in part by the European Union (Project reference: FAIR-PL98-4338). We would like to thank P. Breen and an anonymous referee for their com- ments that helped improve the manuscript. We would like to thank C. Flor and I. Costa of the fishing vessel Zequinha for their help with the preparation, setting, retrieval, and monitoring of the traps and for help with the diving operations. Literature cited Al-Masroori, H., H. Al-oufi, J. L. Mcllwain, and E. McLean. 2004. Catches of lost fish traps (ghost fishing) from fish- ing grounds near Muscat, Sultanate of Oman. Fish. Res. 69:407-414. Breen, P. A. 1987. Mortality of Dungeness crabs caused by lost traps in the Fraser river estuary, British Columbia. N. Am. J. Fish. Manag. 7:429-435. 1990. A review of ghost fishing by traps and gillnets. In Proceedings of the second international conference on marine debris (R. Shomura, and M. L. Godfrey, eds.), p. 571-599. 2-7 April 1989, Honolulu, Hawaii. U.S. Dep. Commer., NOAA Tech. Memo. NMFS, NOAA-TM- NMFS-SWFSC-154. Brown, J., and G. Macfadyen. 2007. Ghost fishing in European waters: Impacts and management responses. Mar. Policy 31:488-504 Bullimore, B. A., P. B. Newman, M. J. Kaiser, S. E. Gilbert, and K. M. Lock. 2001. A study of catches in a fleet of “ghost-fishing” pots. Fish. Bull. 99:247-253. Castro, M., A. Araujo, and P. Monteiro. 2005. Fate of discards from deep water crustacean trawl fishery off the south coast of Portugal. N. Z. J. Mar. Freshw. Res. 39:437-446. Erzini, K., C. C. Monteiro, J. Ribeiro, M. N. Santos, M. Gaspar, P. Monteiro, and T. C. Borges. 1997. An experimental study of gill net and trammel net “ghost fishing” in the Algarve (southern Portugal). Mar. Ecol. Prog. Ser. 158:257-265. Fogarty, M. J., and J. T. Addison. 1997. Modelling capture processes in individual traps: entry, escapement and soak time. ICES J. Mar. Sci. 54:193-205. Godpy, H., D. M. Furevik, and S. Stiansen. 2003. Unaccounted mortality of red king crab (Para- lithodes camtschaticus) in deliberately lost pots off Northern Norway. Fish. Res. 64:171-177. Guillory, V. 1993. Ghost fishing by blue crab traps. N. Am. J. Fish. Manag. 13:459-466. Laist, D. W. 1995. Marine debris entanglement and ghost fishing: a cryptic and significant type of bycatch? In Solving bycatch: considerations for today and tomorrow: proceed- ings of the Solving Bycatch Workshop, p. 33-39. Univ. Alaska Sea Grant College Program Report 96-03, Univ. of Alaska, Fairbanks, AK. Mahon, R., and W. Hunte. 2001. Trap mesh selectivity and the management of reef fishes. Fish Fish. 2:356-375. Munro, J. L. 1974. The mode of operation of Antillean fish traps and the relationships between ingress, escapement, catch and soak. J. Cons. Int. Explor. Mer 35:337-350. Scarsbrook, J. R., G. A. MacFarlane, and W. Shaw. 1988. Effectiveness of experimental escape mechanisms in sablefish traps. N. Am. J. Fish. Manag. 8:158-161. Smolowitz, R. J. 1978. Trap design and ghost fishing: an overview. U.S. Nat. Mar. Fish. Serv. Mar. Fish. Rev. 40:59-67. Zhou, S., and T. C. Shirley. 1997. A model expressing the relationship between catch and soak time for trap fisheries. N. Am. J. Fish. Manag. 17:482-487. 328 Age- and length-at-maturity of female arrowtooth flounder ( Atheresthes stomias ) in the Gulf of Alaska James W. Stark Resource Assessment and Conservation Engineering Division Alaska Fisheries Science Center National Oceanic and Atmospheric Administration/National Marine Fisheries Service 7600 Sand Point Way NE Seattle, Washington 981 15 Email address: Jim.Stark@noaa.gov Arrowtooth flounder ( Atheresthes sto- mias) has had the highest abundance of any groundfish species in the Gulf of Alaska since the 1970s (Matarese et al., 2003; Turnock et al., 2005; Blood et al., 2007); however, com- mercial catches have been restricted because Pacific halibut (Hippoglossus stenolepis) are caught as bycatch in the fishery. Arrowtooth flounder plays a key role in the ecosystem because it is a dominant organism within the food web, both as an apex predator of fish and invertebrates, as well as an important prey for walleye pol- lock ( Theragra chalcogramma\ Aydin et al., 2002). Walleye pollock is the dominant groundfish in the Bering Sea, a principal groundfish in the Gulf of Alaska, and the primary prey for marine mammals. The distribu- tion of arrowtooth flounder extends from Cape Navarin and the eastern Sea of Okhotsk in Russia, across the Bering Sea, Aleutian Islands, Gulf of Alaska, and south to the coast of central California (Shuntov, 1964; Britt and Martin, 2001; Chetvergov, 2001; Weinberg et al., 2002; Zenger, 2004). Because of the importance of arrowtooth flounder in the marine ecosystem of Alaska, a maturity study of this species was undertaken to determine age-at-maturity, which is essential for age-based stock man- agement models. Before these results, management has had to rely upon a length-at-maturity-based estimate (Zimmermann, 1997) to manage stocks in the Gulf of Alaska (GOA), Bering Sea, and Aleutian Islands. The central GOA was selected as the location for this maturity study because it contains approximately 70% of the total Gulf of Alaska arrow- tooth flounder biomass (1.9xl06 t, age 3 and older) — the highest percentage in the world (Shuntov, 1964; Britt and Martin, 2001; Weinberg et al., 2002; Wilderbuer and Nichol, 2006). Materials and methods All female arrowtooth flounder used in this study were collected with bottom trawls. The central GOA was initially sampled during February 2002 (Fig. 1) from the National Oce- anic and Atmospheric Administration (NOAA) ship Miller Freeman during an Alaska Fisheries Science Center (AFSC) Recruitment Processes Pro- gram cruise (Blood et al., 2007). The area selected for trawling was sampled the prior year by the AFSC and had produced a high abundance of arrow- tooth flounder eggs and larvae. A second collection was made in July 2003 during the AFSC biennial GOA groundfish assessment survey. In both years, whole ovaries and oto- liths were collected. All specimens were selected by using length-strati- fied sampling method so that three to seven females were collected for each cm interval of total body length larger than 18 cm.. The sampling protocol, histological methods, ovary maturity classifications, and aging methods followed those described in Stark (2007). Mature females were those specimens classified with ovary stages ranging from cortical alveoli to postovulatory follicles, which were the same criteria used by Zimmer- mann (1997) for a September 1993 GOA arrowtooth flounder maturity assessment. To investigate the consis- tency of oocyte maturation within the ovary, two additional sections were taken from the anterior and medial regions of both ovaries from 10 speci- mens collected during February 2002 and 10 specimens from July 2003. These sections were compared with the standard sample section taken from the posterior area of one ovary. For all the following procedures, S- Plus software was used (vers. 2000 Professional release 3, MathSoft Inc., Cambridge, MA). Maturity was es- timated as a function of length and age by fitting a logistic function to the maturity data with generalized linear modeling (Venables, 1997). The significance of temporal differences was tested by fitting the model of maturity as a function of total body length (L) and age (A), including the date of sampling, and by recalculat- ing without the date term. Signifi- cance of the date term was deter- mined by using analysis of deviance (Venables, 1997). The variance of age (A50) and total body length (L50) at 50% maturity were estimated for February 2002 and July 2003 by us- ing bootstrapping (Efron and Tibshi- rani, 1993) based on 200 resamplings with replacement of the maturity, age, and length data. February 2002 and July 2003 differences in the A50 and L50were tested with a Z-test (So- kal, 1969). The February 2002 L50 result was also tested against the September 1993-based estimate by Zimmermann (1997) with a Z-test. To assess the temporal progression of ovary maturity between February 2002 and July 2003, ovary maturity classifications were summarized for females that had reached A50 as de- termined by this study. Manuscript submitted 21 March 2008. Manuscript accepted 6 May 2008. Fish. Bull. 106:328-333 (2008). The views and opinions expressed or implied in this article are those of the author and do not necessarily reflect the position of the National Marine Fisheries Service, NOAA. NOTE Stark: Age- and length-at-maturity of female Atheresthes stomias in the Gulf of Alaska 329 I60WW I55°0'0"W Figure t Locations where arrowtooth flounder ( Atheresthes stomias ) were collected by bottom trawl during National Oceanic and Atmospheric Administration Alaska Fisheries Science Center cruises during February 2002 (x) and July 2003 (+) in the Gulf of Alaska. To determine if ovary growth occurred between Feb- ruary and July, the gonadosomatic index (somatic body weight/ovary weight, 7G) of females > A50 was deter- mined by methods described by Stark (2007). The com- parison did not include females with hydrated oocytes or postovulatory follicles. Results In the February 2002 collection, approximately 50% of the females were classified as spawning and 10% were nearing spawning condition (advanced yolk). By July, spawning activity was <5% but continuing, and 20% of the females were at the advanced yolk stage of maturity based on the 2003 collection. Mature female arrowtooth flounder were found in every location where sample collections were made, namely on the outer continental shelf at 101-200 m depths and upper slope at 301-700 m depths, during February 2002. During July 2003, the locations and depths sampled ranged from the intertidal zone at 1- 100 m down to the continental banks, gullies, and outer shelf at 101-200 m. Mature female specimens were as young as 3 years of age and 260 mm L. This study was the first to determine that age was a significant predictor (PcO.001) of maturity for female arrowtooth flounder. The estimated age at 50% matu- rity (A50) was 7.0 years, based on the February 2002 collection (Fig. 2, Table 1). The February 2002 result did not differ significantly (P= 0.38) from the July 2003 result (6.6 years). Of the two results, the February 2002 A50 estimate had a much lower variance (Table 2) and consequently was the most reliable. The estimated length at which 50% of the females were mature was 460 mm for the February 2002 collection (Fig. 3, Table 2) and did not differ ( P= 0.68) for the July 2003 collection (464 mm). However, the variance of the L50 estimate was much lower for the February 2002 collec- tion, and thus February represented the more optimal month for assessing arrowtooth flounder maturity. Oo- cyte development was consistent within the ovaries and between each ovary pair for every compared specimen. No significant growth occurred between February and July, based on the 7G results, which declined from a mean of 3.6 IG in February 2002 (n = 167) to 1.8 IG in July 2003 (n = 115). Discussion This study was the first to establish arrowtooth flounder age-at-maturity. The estimated age at 50% maturity (A50= 7 years) in the GOA was based on the February 330 Fishery Bulletin 106(3) 2002 maturity collection results that were supported by the July 2003 results. Age-based estimates of maturity should be less variable, and less affected by changes in environmental conditions, population abundance, and spawning biomass levels than length-based estimates. Those changes can affect arrowtooth flounder growth rates. Growth was significantly faster for Pacific cod ( Gadus macrocephalus) in the Bering Sea than in the Gulf of Alaska (Stark, 2007). Together with growth, cod L50 differed significantly (P<0.001) between areas; in contrast, cod Ago differed only slightly (P=0.02). There- fore, age-based maturity estimates should generally be considered the more reliable. Although not as reliable, the estimated length-at- maturity of female arrowtooth flounder has remained stable in the GOA, based on the results of this study NOTE Stark: Age- and length-at-maturity of female Atheresthes stomias in the Gulf of Alaska 331 Table 1 Female arrowtooth flounder (Atheresthes stomias) age- at-maturity results based on ovary histological samples ( n ) collected between the spawning and late spawning periods in the Gulf of Alaska by date of collection. The parameters of the logistic equation that were used to fit the data were the following: B (slope of the line) and A (y intercept), variance (the square of the standard devia- tion of B and A), covariance (the product of the standard deviations of B and A and the coefficient of correlation between them), age (years) at which 50% of females were expected to reach sexual maturity (A50), and variance °fA50. Date of collection February 2002 July 2003 n 301 226 B 1.3817 0.6835 A -9.6183 -4.4945 Variance (B) 0.0077 0.0016 Variance (A) 1.7381 1.3528 Covariance (B, A) 0.0070 0.0007 ^50 6.9614 6.5754 Variance (A50) 0.0326 0.1448 Table 2 Female arrowtooth flounder (Atheresthes stomias ) length-at-maturity results based on ovary histology samples (n) collected between the spawning and late spawning periods in the Gulf of Alaska by date of col- lection. The parameters of the logistic equation that were used to fit the data were the following: B (slope of the line) and A (y intercept), variance (the square of the standard deviation of B and A), covariance (the product of the standard deviations of B and A and the coefficient of correlation between them), length (mm) at which 50% of females were expected to reach sexual maturity (L50), and variance of L50. Date of collection February 2002 July 2003 n 303 251 B 0.0455 0.0215 A -20.9220 -9.9786 Variance (B) 1.1171 2.9534 Variance (A) 6.9072 6.3385 Covariance (B, A) 0.0002 -0.0002 ^50 (mm) 459.6917 464.1629 Variance (L50) 27.5705 75.9320 and the 1993 maturity study results by Zimmermann (1997). For this present study, the estimated female L50 was 460 mm, which did not differ significantly (P=0.08) from the September 1993 L50 estimate of 469 mm. Cur- rently, that length-at-maturity estimate is used to man- age the arrowtooth flounder stocks of both the GOA and Bering Sea, because the A50 has not been known. Consequently, with the use of the age-at-maturity esti- mates determined in the present study, the estimates of arrowtooth flounder abundance would be expected to improve significantly within the stock management model. Stock management would also benefit from a deter- mination of the arrowtooth flounder annual spawning period. The arrowtooth flounder spawning period prob- ably begins in December, based on results from the AFSC ichthyoplankton surveys in the Gulf of Alaska, during which developing embryos and larvae were found at the end of January (Blood et al., 2007). The rate of spawning declined from February 2002 (50%) to July 2003 (<5%), according to this study. Spawning may conclude at the end of summer because Zimmermann (1997) found no spawning females during September. Therefore, the overall spawning period of arrowtooth flounder appears to extend for over 8 months or more in the GOA. This is a longer spawning period than has been found for other principal groundfish species in Alaska, which have spawning periods of 6 months or less (Matarese et al., 2003; Stark, 2004, 2007). A protracted spawning period could promote stock re- cruitment by increasing the dispersion of progeny and thereby increasing the probability of placing progeny in a favorable rearing environment. Turnock et al. (2005) estimated that the spawning biomass of female ar- rowtooth flounder has increased annually since 1961 (1.98xl05t) and remains above 1.24 x 106 t in the GOA. Similarly, the spawning biomass was estimated to be at its highest level ever recorded in the Bering Sea and Aleutian Islands, at more than 8.24 x 105 t (Wilderbuer and Nichol, 2006). However, the size of the spawning biomass may be overestimated for the Bering Sea and Aleutian Islands because of the reliance on the Gulf of Alaska L50 estimate for the Bering Sea and Aleutian Islands management model. This overestimate could occur if the rate of female arrowtooth flounder growth was significantly higher for the Bering Sea and Aleutian Islands population than it was for the Gulf of Alaska population. A significantly higher rate of growth could result in a significantly larger L50, which could lower the spawning biomass estimate by excluding smaller females that may have been mature. Therefore, the estimates of the arrowtooth flounder spawning biomass that are determined by stock managers should be more reliable after the current length-based maturity models are replaced with age-based maturity models using the A50 estimate from this study. Conclusions Arrowtooth flounder has consistently been the most abundant groundfish species in the Gulf of Alaska because of its high levels of recruitment and low fish- ing-induced mortality. Age was found to be a significant 332 Fishery Bulletin 106(3) predictor of female maturity in this study and a mean age-at-maturity (A50, 7 years) was established for female arrowtooth flounder that will allow for the development of an improved stock management model. The model should provide more reliable estimates of arrowtooth flounder abundance in the Gulf of Alaska as well as the Bering Sea. This study also documented the total body length (L 460 mm) at which 50% of the females are mature in the Gulf of Alaska — an estimate that did not differ significantly from a 1993-based estimate. Acknowledgments I thank the following Alaska Fisheries Science Center personnel for assisting in the collection of specimens in NOTE Stark: Age- and !ength-at-maturity of female Atheresthes stomias in the Gulf of Alaska 333 2002: D. Blood, A. Matarese, M. Busby, W. Floering, D. Stevenson, and R. Cartwright. Assisting with the 2003 specimen collections were D. Anderl and J. Ferdinand. Age determinations were made by J. Brogan. Research support was provided by D. Somerton, G. Stauffer, and R. Nelson. In-house review was provided by D. Somerton, M. Zimmermann, M. Wilkins, G. Duker, and J. Lee. Literature cited Aydin, K. Y., V. V. Lapko, V. I. Radchenko, and P. A. Livingston. 2002. A comparison of the eastern Bering and western Bering Sea shelf and slope ecosystems through the use of mass-balance food web models. NOAA Tech. Memo. NMFS-AFSC-130, 87 p. Blood, D. M., A. C. Matarese, and M. S. Busby. 2007. Spawning, egg development, and early life history dynamics of arrowtooth flounder (Atheresthes stomias) in the Gulf of Alaska. U.S. Dep. Commer., NOAA Prof. Pap. NMFS 7, 28 p. Britt, L. L., and M. H. Martin. 2001. Data report: 1999 Gulf of Alaska bottom trawl survey. NOAA Tech. Memo. NMFS-AFSC-121, 249 p. Chetvergov, A. A. 2001. Long-jawed flounder Atheresthes stomias (Pleuro- nectidae) from the eastern part of the Sea of Okhotsk. J. Ichthyol. 42:294-299. Efron, B., and R. J. Tibshirani. 1993. An introduction to the bootstrap, 436 p. Chapman and Hall, New York, NY. Matarese, A. C., D. M. Blood, S. J. Picquelle, and J. L. Benson. 2003. Atlas of abundance and distribution patterns of ichthyoplankton from the Northeast Pacific Ocean and Bering Sea ecosystems based on research conducted by the Alaska Fisheries Science Center 1972-1996. U.S. Dep. Commer., NOAA Prof. Pap. NMFS 1, 281 p. Shuntov, V. P. 1964. Distribution of the Greenland halibut and arrow- toothed halibuts in the North Pacific. Soviet Fisheries Investigations in the northeast Pacific, Part 4. Pacific Scientific Research Institute of Marine Fisheries and Oceanography (TINRO). Izvestiya 58, 147-156. [In Russian; 1968 English transl. by Israel Program for Sci. Transl.] Sokal, R. R. 1969. Biometry, 776 p. W. H. Freeman and Company, San Francisco, CA. Stark, J. W. 2004. A comparison of the maturation and growth of female flathead sole in the central Gulf of Alaska and south-eastern Bering Sea. J. Fish Biol. 64:876-889. 2007. Geographic and seasonal variations in maturation and growth of female Pacific cod (Gadus maerocepha- lus) in the Gulf of Alaska and Bering Sea. Fish. Bull. 105:396-407. Turnock, B. J., T. K. Wilderbuer, and E. Brown. 2005. Stock assessment and fishery evaluation: Gulf of Alaska arrowtooth flounder, 38 p. North Pacific Fishery Management Council. 605 West Fourth Avenue, Suite 306, Anchorage, AK 99501. Venables, W. N. 1997. Modern applied statistics with S-Plus, 548 p. Springer-Verlag, New York, NY. Weinberg, K. L., M. E. Wilkins, F. R. Shaw, and M. Zimmermann. 2002. The 2001 Pacific west coast bottom trawl survey of groundfish resources: Estimates of distribution, abun- dance, and length and age composition. NOAA Tech. Memo. NMFS-AFSC-128, 149 p. Wilderbuer, T. K., and D. G. Nichol. 2006. Stock assessment and fishery evaluation: Bering Sea and Aleutian Islands arrowtooth flounder, 50 p. North Pacific Fishery Management Council, 605 West Fourth Avenue, Suite 306, Anchorage, AK 99501. Zenger, H. H., Jr. 2004. Data report: 2002 Aleutian Islands bottom trawl survey. NOAA Tech. Memo. NMFS-AFSC-143, 258 p. Zimmermann, M. 1997. Maturity and fecundity of arrowtooth flounder, Atheresthes stomias , from the Gulf of Alaska. Fish. Bull. 95:598-611. 334 Fishery Bulletin 106(3) Fishery Bulletin Guidelines for authors Content of manuscripts Contributions published in Fishery Bulletin describe original research in marine fishery science, fishery engineering and economics, as well as the areas of marine environmental and ecological sciences (including modeling). Preference will be given to manuscripts that examine processes and underlying patterns. Descriptive reports, surveys, and observational papers may occa- sionally be published but should appeal to an audience outside the locale in which the study was conducted. Although all contributions are subject to peer review, responsibility for the contents of papers rests upon the authors and not on the editor or publisher. Submission of an article implies that the article is original and is not being considered for publication elsewhere. Articles may range from relatively short contributions (10-15 typed, double-spaced pages, tables and figures not included) to extensive contributions (20-30 typed pages). Notes are reports of 5 to 10 pages without an abstract and describe methods or results not supported by a large body of data. Manuscripts must be written in English; authors whose native language is not English are strongly advised to have their manuscripts checked by English-speaking colleagues before submission. Manuscript Preparation Title page should include authors’ full names and mail- ing 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 200 words (one-half typed page), state the main scope of the research, and emphasize the author’s conclusions and relevant findings. Do not review the methods of the study or list the contents of the paper. Because abstracts are circulated by abstract- ing 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 Conclusions), and Acknowl- edgments. Headings within each section must be short, reflect a logical sequence, and follow the rules of multiple subdivision (i.e., there can be no subdivision without at least two items). The entire text should be intelligible to interdisciplinary readers; therefore, all acronyms, abbreviations, and technical terms should be written out in full the first time they are mentioned. Include FAO common names for species in the list of keywords and in the introduction. Regional common names may be used throughout the rest of the text if they are dif- ferent from FAO common names which 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; for fish nomenclature follow the most current issue of the American Fisheries Society’s Common and Scientific Names of Fishes from the United States and Canada. Dates should be written as follows: 11 November 2000. Measurements should be expressed in metric units, e.g., 58 metric tons (t); if other units of measurement 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). Refrain from using the shorthand slash (/), an ambigu- ous symbol, in the general text Literature cited comprises published works and those accepted for publication in peer-reviewed literature (in press). Follow the name and year system for citation format in the “Literature cited” section (that is say, citations should be listed alphabetically by the authors’ last names, and then by year if there is more than one citation with the same authorship). If there is a sequence of citations in the text, list chronologically: (Smith, 1932; Green, 1947; Smith and Jones, 1985). Abbrevia- tions of serials should conform to abbreviations given in the Serial Sources for the BIOSIS Previews Database. Authors are responsible for the accuracy and complete- ness of all citations. Literature citation 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. Cite all software and special equipment or chemical 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). 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. • Do not place outline rules around graphs. • Do not use horizontal lines through graphs to indicate measurement units. • Use a comma in numbers of five digits or more (e.g. 13,000 but 3000). • Maps require a North arrow and degrees latitude- longitude (e.g., 170°E). 335 Tables are often overused in scientific papers; it is seldom necessary or even desirable to present all the data associated with a study. Tables should not be excessive in size and must be cited in numerical order in the text. Headings should be short but ample enough to allow the table to be intelligible on its own. All unusual symbols must be explained in the table legend. Other incidental comments may be footnoted with italic numeral footnote markers. Use asterisks to indicate probability in statisti- cal data. Do not type table legends on a separate page; place them above the table data. Do not submit tables in photo mode. Figures include line illustrations, photographs (or slides), and computer-generated graphs and must be cited in numerical order in the text. Graphics should aid in the comprehension of the text, but they should be limited to presenting patterns rather than raw data. Figures should not exceed one figure for every four pages of text. Figures must be labeled with author’s name and number of the figure. Avoid placing labels vertically (except of y axis). 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 and failure to correspond with editors in a timely manner will delay publication of a manuscript. Copyright law does not apply to Fishery Bulletin, 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 consid- ered correct form (e.g., Source: Fish. Bull 97:105). Submission The Scientific Editorial Office encourages authors to submit their manuscripts as a single PDF (preferred) or Word (zipped) document by e-mail to Fishery. Bulletin@noaa.gov. Please use the subject heading, “Fishery Bulletin manuscript submission”. Do not send encrypted files. For further details on electronic sub- mission, please contact the Scientific Editorial Office directly (see address below). Or you may send your manuscript on a compact disc in one of the above formats along with four printed copies (one original plus three copies [stapled]) to the Scientific Editor, at the address shown below. Dr. Adam Moles Scientific Editor, Fishery Bulletin NOAA/NMFS/AFSC 17109 Point Lena Loop Road Juneau, AK 99801-8626 Once the manuscript has been accepted for publication, you will be asked to submit a final electronic 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, tiff files, or EPS files. Send a copy of figures in the origi- nal software if conversion to any of these formats yields a degraded version. Questions? If you have questions regarding these guidelines, please contact the Managing Editor, Sharyn Matriotti, at Sharyn.Matriotti@noaa.gov Questions regarding manuscripts under review should be addressed to Adam Moles, Scientific Editor, at Adam. Moles@noaa.gov Fishery Bulletin Subscription form Superintendent of Documents Publications Order Form *5178 □ YES, please send me the following publications: Subscriptions to Fishery Bulletin for $36.00 per year ($50.40 foreign) The total cost of my order is $ . Prices include regular domestic postage and handling and are subject to change. (Company or Personal Name) (Please type or print) (Additional address/attention line) (Street address) (City, State, ZIP Code) (Daytime phone including area code) (Purchase Order No.) Charge your order. ITS EASY! Please Choose Method of Payment: ] Check Payable to the Superintendent of Documents 3H GPO Deposit Account -□ 3H 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! Also available online at http://bookstore.gpo.gov/collections/fishery-bulletin SMITHSONIAN INSTITUTION LIBRARIES U.S. Department of Commerce Volume 106 Number 4 October 2008 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 James W. Balsiger, Ph.D. Acting Assistant Administrator for Fisheries Scientific Editor Adam Moles, Ph.D. Associate Editor Elizabeth Siddon Ted Stevens Marine Research Institute Auke Bay Laboratories Alaska Fisheries Science Center 17109 Pt. Lena Loop Road Juneau, Alaska 99801 Managing Editor Sharyn Matriotti National Marine Fisheries Service Scientific Publications Office 7600 Sand Point Way NE, BIN C 15700 Seattle, Washington 98115-0070 The Fishery Bulletin (ISSN 0090-0656) is published quarterly by the Scientific Publications Office, National Marine Fisheries Service, NOAA, 7600 Sand Point Way NE, BIN C15700, Seattle, WA 98115-0070. Periodicals postage is paid at Seattle, WA. POSTMASTER: Send address changes for subscriptions to Fish- ery Bulletin, Superintendent of Docu- ments, Attn.: Chief, Mail List Branch, Mail Stop SSOM, Washington, 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: $36.00 domestic and $50.40 foreign. Cost per single issue: $21.00 domestic and $29.40 foreign. See back for order form. Editorial Committee Jeffrey M. Leis Thomas Shirley David Somerton Mark Terceiro Australian Museum, Sydney, Australia Texas A&M University National Marine Fisheries Service National Marine Fisheries Service Fishery Bulletin web site: www.fishbnll.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 106 Number 4 October 2008 The National Marine Fisheries Service (NMFS) does not approve, recommend, or endorse any proprie- tary product or proprietary material mentioned in this publication. No reference shall be made to NMFS, or to this publication furnished by NMFS, in any advertising or sales promotion which would indicate or imply that NMFS approves, recom- mends, or endorses any proprietary product or proprietary 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. Fishery Bulletin Contents Articles 337-347 Stoner, Allan W., Craig S. Rose, J. Eric Munk, Carwyn F. Hammond, and Michael W. Davis An assessment of discard mortality for two Alaskan crab species. Tanner crab ( Chionoecetes bairdi) and snow crab (C. opilio), based on reflex impairment 348-363 Casazza, Tara L., and Steve W. Ross Fishes associated with pelagic Sargassum and open water lacking Sargassum in the Gulf Stream off North Carolina 364-374 Rodgveller, Cara J., Chris R. Lunsford, and Jeffrey T. Fujioka Evidence of hook competition in longline surveys 375-385 Kastelle, Craig R., Delsa M. Anderl, Daniel K. Kimura, and Chris G. Johnston Age validation of Dover sole (Microstomus pacificus) by means of bomb radiocarbon 386-394 Maschner, Herbert D. G., Matthew W. Betts, Katherine L. Reedy-Maschner, and Andrew W. Trites A 4500-year times series of Pacific cod (Gadus macrocephalus) size and abundance: archaeology, oceanic regime shifts, and sustainable fisheries 395-404 Kane, Emily A., Paula A. Olson, Tim Gerrodette, and Paul C. Fiedler Prevalence of the commensal barnacle Xenobalanus globicipitis on cetacean species in the eastern tropical Pacific Ocean, and a review of global occurrence II Fishery Bulletin 106(4) 405-416 Fowler, Ashley M., Jeffrey M. Leis, and lain M. Suthers Onshore-offshore distribution and abundance of tuna larvae (Pisces: Scombridae: Thunnini) in near-reef waters of the Coral Sea 417-426 Jeffers, Sarah A., William F. Patterson III, and James H. Cowan Jr. Habitat and bycatch effects on population parameters of inshore lizardfish (Synodus foetens ) in the north central Gulf of Mexico 427-437 Brill, Richard, Christopher Magel, Michael Davis, Robert Hannah, and Polly Rankin Effects of rapid decompression and exposure to bright light on visual function in black rockfish ( Sebastes melanops) and Pacific halibut (Hippoglossus stenolepis) 438-456 Asch, Rebecca G., and Jeremy S. Collie Changes in a benthic megafaunal community due to disturbance from bottom fishing and the establishment of a fishery closure 457-470 Vollen, Tone, and Ole T. Albert Pelagic behavior of adult Greenland halibut ( Reinhardtius hippoglossoides ) Notes 471-475 Love, Milton S., Donna M. Schroeder, Linda Snook, Anne York, and Guy Cochrane All their eggs in one basket: a rocky reef nursery for the iongnose skate (Raja rhino Jordan & Gilbert, 1880) in the southern California Bight 476-482 Karlsson, Sten, Mark A. Renshaw, Caird E. Rexroad III, and John R. Gold Microsatellite primers for red drum ( Scioenops ocellatus ) 483 Acknowledgment of reviewers 484 List of titles 486 List of authors 487 Subject index 490 Guidelines for authors 492 Subscription form 337 Abstract — Delayed mortality asso- ciated with discarded crabs and fishes has ordinarily been observed through tag and recovery studies or during prolonged holding in deck tanks, and there is need for a more efficient assessment method. Chi- onoecetes bairdi (Tanner crab) and C. opilio (snow crab) collected with bottom trawls in Bering Sea waters off Alaska were evaluated for reflexes and injuries and held onboard to track mortality. Presence or absence of six reflex actions was determined and combined to calculate a reflex impair- ment index for each species. Logis- tic regression revealed that reflex impairment provided an excellent predictor of delayed mortality in C. opilio (91% correct predictions). For C. bairdi , reflex impairment, along with injury score, resulted in 82.7% correct predictions of mortality, and reflex impairment alone resulted in 79.5% correct predictions. The rela- tionships between reflex impairment score and mortality were indepen- dent of cr. b gender, size, and shell condition, and predicted mortality in crabs with no obvious external damage. These relationships provide substantial improvement over earlier predictors of mortality and will help to increase the scope and replication of fishing and handling experiments. The general approach of using reflex actions to predict mortality should be equally valuable for a wide range of crustacean species. Manuscript submitted 25 March 2008. Manuscript accepted 7 May 2008. Fish. Bull. 106:337-347 (2008). The views and opinions expressed or implied in this article are those of the author and do not necessarily reflect the position of the National Marine Fisheries Service, NOAA. An assessment of discard mortality for two Alaskan crab species. Tanner crab ( Chionoecetes bairdi ) and snow crab (C. opilio), based on reflex impairment Allan W. Stoner (contact author)1 Craig S. Rose 2 J. Eric Munk 3 Carwyn F. Hammond 2 Michael W. Davis 1 Email address for A. W. Stoner: AI.Stoner@noaa.gov 1 Fisheries Behavioral Ecology Program Alaska Fisheries Science Center National Marine Fisheries Service, NOAA 2030 Marine Science Dr Newport, Oregon 97365 2 Conservation Engineering Program Alaska Fisheries Science Center National Marine Fisheries Service, NOAA 7600 Sand Point Way NE, Seattle, Washington 98115 3 Shellfish Assessment Program Alaska Fisheries Science Center, National Marine Fisheries Service, NOAA 301 Research Court Kodiak, Alaska 99615 Numerous species with economic and ecological significance are discarded from fishing operations because of harvest restrictions or low value com- pared with the target species or size groups. Although discarded catch can exceed that retained (Alverson et ah, 1994; Witherell and Pautzke, 1997), discard mortality rates are rarely known. Mortality can occur on deck (immediate and observed mortality) or after the animal is released or escapes capture (delayed and unobserved mor- tality) (Stevens, 1990; Davis, 2002; Suuronen, 2005). Delayed mortal- ity can result directly from physical injury and physiological stress or indi- rectly by an increased susceptibility to disease and predation, and through an inability to feed. Although a signifi- cant effort has been made to reduce unwanted bycatch through improve- ments in gear selectivity and han- dling procedures, unobserved discard mortality continues to be an impor- tant source of uncertainty for fishery management (Harrington et al., 2005; Broadhurst et al., 2006; Coggins et ah, 2007). Directed fisheries for Chionoecetes spp. and Paralithodes camtschaticus (red king crab) in the North Atlan- tic, North Pacific, and Bering Sea are prosecuted with baited pots, the only legal commercial gear for these crabs in the United States and Cana- da. Harvesting is restricted to males within strict size limits; consequently all females and undersized males are discarded. Crabs are also captured and discarded or escape from trawl gear. Given the difficulties of deter- mining mortality for discarded crabs, fixed mortality rates have been as- sumed for specific fishing sectors in most fishery management models on the basis of a handful of experiments (see below). For example, fishery managers in Alaska have assumed fixed values of 20%, 50%, and 20%, respectively, for the discard mortality rates of Chionoecetes bairdi (southern 338 Fishery Bulletin 106(4) Tanner crab), C. opilio (snow crab) and P. camtschaticus in pot fisheries, and a fixed value of 80% for the discard mortality of both red king crab and snow crab in Bering Sea trawl fisheries (Siddeek, 2003). Estimates for discard-related mortality in fishery spe- cies normally require experimental research whereby individuals or populations captured in different fishing operations or under different conditions are monitored for subsequent survival. Mortality can be evaluated through tag recovery studies, but this method requires handling large numbers of individuals, often for low returns. Holding fish or crabs in tanks or cages is a good method for direct evaluation of mortality in dif- ferent fishing operations and handling, but experiments are often limited to relatively small numbers, short fol- low-up periods, and a few specific treatments or fishing variables. A useful predictor for discard mortality may be crab condition. Blood and tissue chemistry (e.g., lactate, glucose, glycogen levels) have been used to evaluate stress in crustaceans (Crear and Forteath, 2001; Har- ris and Andrews, 2005; Ridgway et ah, 2006). Recent experiments with marine fishes, however, show that chemical measures are often poor predictors of mor- tality because they typically reach peak values before any mortality occurs and because chemical measures often respond differently to physiological stress and physical injury (Davis et al., 2001, Davis and Schreck, 2005). Two other types of condition indicators have been explored for predicting mortality rates in crabs: injuries to the exoskeleton and behavioral impairments. In the first category, correlations between externally visible physical injuries and mortality were reported for Chionoecetes spp. (Rosenkranz, 2002) and red king crabs (Zhou and Shirley, 1995). However, internal in- juries and bleeding that result in mortality can occur without apparent external injuries. Given that limita- tion, Stevens (1990) considered both external injuries and an index of spontaneous activity, termed vitality, as possible correlates with delayed mortality in red king crabs and Tanner crabs caught incidentally in Bering Sea trawls. The crabs were assessed and held for 48 hours in shipboard tanks to observe mortality. Logistic models showed that the vitality index was useful in predicting delayed mortality, and injury in- formation did not significantly improve model power. Subsequently, others proposed that impaired right- ing behavior provides a sensitive indicator of freeze- related stress in snow crabs (Warrenchuk and Shirley, 2002), and might be used to predict mortality (van Tamelen, 2005). However, complex behaviors are diffi- cult to quantify at sea because they require space and controlled conditions, and often do not yield graduated results. As an alternative to assessments determined by complex behavioral patterns, Davis and Ottmar (2006) recently discovered that easily acquired obser- vations on a suite of simple reflex actions can provide excellent predictions of mortality in fishes related to both physical (i.e., wounding) and physiological (e.g., thermal stress, air exposure) injury. We hypothesized that an assessment of reflex ac- tions in crabs would directly reflect their condition and provide a good predictor for mortality, independent of external injury. The goals of this study were to identify reflexes for potential use in assessing crab condition, and then test their ability to predict mortality of Chi- onoecetes bairdi and C. opilio captured in Bering Sea trawls. Crabs were tested for reflex actions and injuries, and monitored for mortality in shipboard tanks. These experiments yielded excellent predictions of mortality based on reflex actions. Materials and methods Identification and scoring of individual reflexes Laboratory studies were conducted at the Kodiak Fish- eries Research Laboratory (National Marine Fisheries Service, Alaska Fisheries Science Center) during April 2007, to identify reflex actions of Chionoecetes spp. that could be reliably used in evaluating their likely survival. Thirty-two individuals of C. bairdi were collected by divers from Kodiak nearshore waters between February and April 2007. Males (n - 24) ranged from 79 to 128 mm carapace width (CW). Females (n = 8) ranged from 88 to 98 mm CW. The crabs were transported to the Kodiak Laboratory in buckets and placed in large fiberglass tanks (1.8 m diam., 1 m deep) where they were held with flowing seawater. Temperature ranged 1.5° to 4.6°C during the holding period. The crabs were fed twice each week with a diet of frozen chopped fish and squid. At the time of testing all of the crabs were in new shell condition, having molted in the past few months, and they were active and feeding. The crabs were marked with vinyl spaghetti tags tied loosely around the basi- ischium of the second or third walking leg. During the preliminary experiments seawater temperature was 5.0°C in the laboratory, and air temperature ranged 16° to 18.6°C. From prior experience with various crab species, we expected that C. bairdi would demonstrate stereotypic responses to being lifted and to manipulation of their appendages. The goal was to identify simple reflex ac- tions that could be evaluated rapidly in the tester’s hand (out of water), during shipboard operations, with a high degree of reliability. After one day of manipula- tions we identified six reflex actions that appeared to be reliable (Table 1). These were tested on every individual crab on two consecutive days and were scored as strong, weak, or as no response. Field experiment A field experiment was conducted in June and July 2007 to evaluate the feasibility of using reflex actions as predictors of delayed mortality in Chionoecetes spp. Trawling operations for this study were conducted on the Bering Sea shelf east of St. Paul Island in the Pribi- lof Islands of Alaska (57°12' to 57°25'N, 169°30' to Stoner et al: An assessment of discard mortality for two Alaskan crab species, based on reflex impairment 339 Table 1 Reflexes identified as useful for assessing stress in Chionoecetes spp. The test was the manipulation required to elicit a stereo- typic response. Characteristic strong and weak responses are described. When no motion was detected in response to repeated testing a “No response” score was recorded. Reflex Test Strong response Weak response Leg flare Lift crab by the carapace, dorsum up. All legs spread wide and high, near horizontal orientation. Legs droop below horizontal. Leg retraction While held as above, draw the forward-most walking legs in the anterior direction. Legs respond with a strong retraction in the posterior direction. Leg retraction is diminished. Low resistance to legs pulled forward. Chela closure Observe for motion or hold the chelae in the fingers. Chelae open and close rapidly without manipulation. Manipulation results in immediate strong closure. Chelae close slowly and weakly upon manipulation, or with a delayed response. Low resistance to manual opening of the chelae. Eye retraction Touch the eye stalk with a blunt probe, or lift the eye stalk from its retracted position. Eye stalk retracts strongly in the lateral direction below the carapace hood. Eye stalk retracts weakly or demonstrates low resistance to lifting. Mouth closure If closed, attempt to open (extend) the 3rd maxillipeds with a sharp dissecting probe. If open, draw the maxillipeds downward. 3rd maxillipeds retract quickly and strongly to cover the smaller mouth parts. The maxillipeds droop open or move in an agitated manner, but do not close tightly. Kick With the crab in ventrum-up position, use a sharp dissecting probe to lift the abdominal flap away from the body. Immediate, strong agitation of the legs and chelipeds. Males respond more strongly than females. Testing with the latter often requires greater extension of the entire abdominal flap. Response is diminished or slow. Motion is observed in only the hind-most legs. 169°45'W) in depths ranging from 35 to 75 m. Crabs were collected from 22 locations where the bottom was relatively homogenous muddy sand. Temperature on the bottom and near surface ranged from -0.3° to 1.7°C and from 3.5° to 4.9°C, respectively. An important objective of the trawling operation was to acquire crabs in various states of stress and injury for assessment and monitoring. Also, we wanted to test crabs that had experienced the suite of stressors typical of those produced during encounters with fishing gear used in the Bering Sea bottom-trawl fishery. Therefore, the crabs were collected in various locations around a commercial trawl by means of recapture nets (see Rose, 1999). The main trawl was a two-seam Alfredo bottom trawl (with headrope and footrope lengths of 36 and 54.6 m, respectively) similar to that used from many vessels in the Bering Sea. The center section of the footrope was composed of 46-cm diameter rounded cones separated by approximately 70-cm long sections of 20-cm diameter disks. The forward 14.2 m of the footrope on each wing of the trawl was made of 20-cm disks strung over a 19-mm long link chain. These sec- tions were not directly attached to the netting panel above them, and thus formed so-called “flying wings.” Extending forward from the trawl, 27.4-m bridles, made of bare cable, were attached to the upper wings and the same lengths of cable, covered with 9-cm diameter rub- ber disks, were attached to the lower wings. Ahead of these were 88 m of 4.8-cm diameter combination rope and 27.4 m of bare cable, leading to the trawl doors. The main trawl was towed with an open codend. The recapture nets were small 2-seam trawls with longer headropes than footropes (14.3 m and 12.0 m, respectively). The long headrope maximized escape of fish, and the small diameter (5-cm) footropes were used to enhance crab capture. These nets were fished directly behind the main net sweeps, wings, or footrope. In some cases, two recapture nets were fished simultaneously at different locations. As a control for damage in the recapture nets, the nets were also fished ahead of the main trawl, capturing crabs with no previous damage. Tows were short (15 minutes) so that the stress or dam- age to crabs was created by the main trawl gear, and less by packing within the net or handling. However, evaluation of gear impact was not the primary goal of this cruise. Once a recapture net was hauled on deck, the volume of the catch (normally comprising Gadus macrocephalus 340 Fishery Bulletin 106(4) Table 2 Composition of Chionoecetes bairdi (Tanner crab) and C. opilio (snow crab) populations tested and monitored in the field experi- ment. Values are means and ranges (in parentheses) for the nonrandom experimental population and do not represent the overall results of the fishing operations. Shell condition is a relative index of molt stage, ranging from molting or recently molted crabs with no hardening of the shell (0) to old and heavily encrusted shells (5). Reflex impairment index A is reported, representing the number of reflex actions that were entirely lost; this index can range from 0 to 6. Injury scores can range from 0 to 5, representing crabs with no visible exoskeletal damage to those with badly broken carapaces and limbs. Species and gender n Carapace width (mm) Shell condition Reflex impairment Injury scores Mortality (%) C. bairdi 250 29.2 Female 89 79 (58-97) 2.6 (2-4) 0.94(0-6) 0.85(0-5) 34.8 Male 161 98 (67-140) 2.7 (1-4) 0.73 (0-6) 0.66(0-5) 26.1 C. opilio 399 24.3 Female 74 63 (52-81) 2.7 (2-4) 1.40 (0-6) 0.62 (0-5) 23.0 Male 325 83 (49-133) 2.6 (0-4) 1.42 (0-6) 0.72 (0-5) 24.6 [Pacific cod], Atheresthes stomias [arrowtooth flounder], and other unidentified flatfishes, seastars, and crabs) was estimated and C. bairdi, C. opilio, and P. camts- chaticus were separated by species and gender into bas- kets. Air temperature during the crab handling process ranged 5° to 10°C. Sorting normally took <15 minutes and the sorting baskets were placed in fish totes with flowing seawater if the sorting and subsequent handling exceeded that time limit. The crabs were then tested individually for loss or weakness according to the six selected reflexes (loss=0, weak=l, strong=2) (Table 1), and notes were made on autotomy and obvious injuries such as broken legs, cracked carapace, or torn abdo- men. Later, injuries were scaled from 0 to 5, where 0 = no injuries, l=newly autotomized legs, 2 = broken legs, chelae, or mouthparts, 3=minor carapace or abdo- men damage, 4=major carapace or abdomen damage, and 5=major damage to multiple parts (carapace, leg, and other parts). Crab gender, shell condition, and CW were recorded during the assessment period. Briefly, shell condition in Chionoecetes spp. was scored from 0 to 5, where 0 represented molting or recently molted crabs with no hardening of the shell whatsoever, crabs with shell condition 1 represented a soft flexible shell, condition 2 represented full hardness, and scores 3 to 5 showed increasing stages of discoloration and encrus- tation with shell age. Following assessment the crabs were either discarded overboard or tagged and held for monitoring mortality. Initially, all crabs were marked and held. However, it soon became apparent that a large proportion had relatively low reflex impairment. Thus, as the cruise progressed, emphasis was shifted to gear configurations producing greater damage to crabs, and we chose individuals with observable reflex impair- ment or injury for holding. As a result, the impairment, injury, and mortality rates reported in Table 2 do not represent the fishing conditions, only the crabs included in this analysis. Crabs held for monitoring were marked with uniquely numbered vinyl spaghetti tags tied securely but loosely around the basi-ischium of the third leg, or fourth leg (if the third leg had been autotomized). Tagged crabs were immediately moved to one of 12 large fish totes (98 x 110x85 cm deep; -900 liters) secured on the ship’s trawl deck. Each tote was supplied with a constant flow of seawater (>20 L/minute). Water temperature during 8-11 day holding periods ranged from 2.1° to 5.9°C, and oxygen (monitored morning and evening in every tote) never fell below 100% saturation. Mortality was assessed and dead crabs were removed each afternoon for the first five days of holding, then every other day until the end of the experiment. On the last day of holding all of the remaining crabs were re-assessed for reflex scores. Reflex impairment indices and statistical procedures Scores for reflex actions were combined into impairment indices and used in the analysis. Composites provided robust indices of overall condition for the animal and provided the advantage of reducing the weight of any one reflex (Davis, 2007). Analyses described below were conducted with two different impairment indices. Reflex impairment index A was calculated as the total number of zero scores for individual reflex actions (i.e., lost reflexes), and reflex impairment index B was the total number of scores that were either 0 or 1 (i.e., lost or weak reflexes). Both index scores ranged from 0 to 6. Before analysis of the relationships between reflex impairment, injury, and mortality, we wanted to be certain that the holding period was sufficiently long to provide an accurate estimate for mortality. The time course for mortality was evaluated as a simple cumula- tive curve of deaths for C. bairdi and C. opilio shown as a function of time in days. We used simple linear regression to explore relationships among mortality-re- lated variables, injury scores, and reflex impairment. Logistic regression was used to model mortality, by using potential predictors and mediators including re- flex impairment, injury score, gender, size, and shell Stoner et al.: An assessment of discard mortality for two Alaskan crab species, based on reflex impairment 341 condition. Models were fitted by the method of maxi- mum likelihood for binary data (i.e., dead or alive) with the regression module of Systat 12 (SYSTAT Software, Inc., San Jose, CA) (Peduzzi et al., 1980). A backward stepwise approach was used to determine the most parsimonious model for mortality, with an alpha value of 0.15 to remove a variable from the full model. This model for mortality was described by Loge (p /(I - p)) - a + p'x, Where p - proportion of y - 1; y = 1 if crab was determined to be dead, and 0 if alive; a = intercept; j8' = model coefficients; and x = the model matrix of explanatory variables. The maximum likelihood estimates of mortality (p) were calculated as p = e_ Initially, the data for each species were split randomly into equal halves, one representing a learning set and the other a test set. The most parsimonious logistic model was developed with the learning set and validated with the test set. After cross-validation, a final model was fitted to the entire data set. Finally, the logistic model for each species was used to develop a surface plot showing the probability of mortality based upon fixed values for the key observations of crab condition. Results Laboratory results The six reflex actions identified for testing with C. bairdi (Table 1) were highly reliable and consistent among individuals. Strong responses in leg flare, leg retrac- tion, eye retraction, and mouth close were observed with every individual every time they were tested. Weak chela closure was observed just once, and the kick response to lifting the abdominal flap was weak in eight instances (seven females) and missing entirely in one test (also female). The weak responses generally occurred in the same individuals in duplicate trials. Field evaluation of reflex impairment Crabs collected in the recapture nets demonstrated a wide range of size, shell condition, reflex impairment, and injury level (Table 2), and reflex actions were lost at different frequencies. When just one reflex was absent, kick and leg retraction were the reflexes lost most fre- quently (Table 3). Among the crabs where one to five reflexes were lost, leg retraction, kick, and leg flare were most often lost, and eye retraction and mouth closure were absent least frequently. Table 3 Percentages of reflex actions lost in Chionoecetes bairdi (Tanner crab) and C. opilio (snow crab). When just one reflex was absent in a crab, it was considered the 1st reflex lost. The right column represents the percentage of spe- cific reflexes that were lost among all of the crabs where between one and five reflexes were lost. Reflex .st reflex lost % of total losses C. bairdi (72 = 18) (72 = 147) Kick 44.4 26.5 Leg retraction 44.4 32.0 Leg flare 5.6 19.0 Chelae closure 5.6 14.3 Eye retraction 0 4.1 Mouth closure 0 4.1 C. opilio (72 = 35) (72=210) Kick 51.4 23.3 Leg retraction 31.4 30.0 Leg flare 14.3 24.8 Chelae closure 0 13.8 Eye retraction 2.9 4.8 Mouth closure 0 3.3 Mortality increased with increasing reflex impair- ment, regardless of the specific index used (Fig. 1). However, mortality at a given impairment value was consistently higher with index A than with index B, and the differences were greatest between index values of 1 and 5. Hence, we judged index A, from the count of completely lost reflexes (scores=0), to provide a more sensitive predictor of mortality than the index based on reflexes scored as lost or weak. A lost reflex was more definitive and less ambiguous than a weak reflex; therefore, reflex impairment index A was used for all subsequent analyses. Appropriateness of experimental holding Adequacy of the field experiment depended upon two important experimental requirements. First, hold- ing needed to be sufficiently long to allow detection of delayed mortality in crabs. Second, the holding methods themselves did not induce mortality. Most of the mortality in the two species occurred within the first few days of collection (Fig. 2). For both species, all the crabs with reflex impairment equal to 6 died within the first hour after reflex assessment. Among the balance of the individuals with less impair- ment, 80% of the mortality occurred within 3 days for C. bairdi and 2 days for C. opilio. For both species, >95% died by day 7. Not surprisingly, time to mortality decreased with increasing reflex impairment (Fig. 3). Some crabs with low reflex impairment died in hold- ing, but these crabs tended to have substantial physical injuries. Time to mortality averaged ~3 days for these individuals, and a high variability in time to mortal- 342 Fishery Bulletin 106(4) 0 1 2 3 4 5 6 7 Reflex impairment score Figure 1 Percent mortality of Chionoecetes bairdi (Tanner crab) and C. opilio (snow crab) shown as a function of reflex impairment indices A and B. Index A was calculated as the sum of completely lost reflexes (score = 0) and index B was calculated as the sum of reflexes either lost or weak (scores = 0 or 1). ity was observed for C. opilio. Crabs with high reflex impairment (4 to 6) usually died within 1 to 2 days and there was little variation in this period of mortal- ity. From these results, we concluded that the holding period for the experiment (11 days) was sufficiently long to test the relationship between reflex impairment and mortality. Re-assessment of reflex impairment in surviving crabs at the end of the experiment revealed that condition of the vast majority did not deteriorate during the hold- ing period at sea. Only two of the surviving C. bairdi (n = 176) had increased reflex impairment. One of these had multiple new leg autotomies and the other had multiple weak responses upon capture. Improvement in the reflex impairment index by one or two points occurred in 7.4% of the crabs, whereas the balance showed no changes. Only 8.5% of the survivors had reflex impairments >0 by the end of the study. Among the 295 C. opilio survivors, none had higher or lower reflex impairment, although most of the survivors (98%) 0 2 4 6 8 10 Days after capture Figure 2 Percent cumulative mortality of Chionoecetes bairdi (Tanner crab) and C. opilio (snow crab) over time during holding at sea. Total numbers of C. bairdi and C. opilio dying over the course of the experiment were 73 and 96, respectively. had no impairment initially. These results indicate that the deck tanks provided a suitable means of holding to assess delayed mortality in the subject species. Mortality and its relationship to reflex impairment Logistic regression analysis revealed that mortality in C. bairdi was associated primarily with reflex impair- ment index and injury score (Table 4). Despite a wide range of crab size and shell condition (Table 2), neither these variables nor crab gender was a significant vari- able in the regression model. In the most parsimonious model, containing a constant, reflex impairment and injury score correctly predicted 82.7% of the mortality and survival (Table 4), whereas the full model (with all variables) predicted 82.8% correctly. Cross-validation showed that the model was robust; i.e., the random learning set produced a model that correctly predicted 86.7% of the test set for C. bairdi. When the injury score was removed from the regression model, 79.5% of the observed outcomes were predicted correctly, dem- onstrating the value of the assessed reflexes. A similar analysis for C. opilio revealed reflex impairment to be the only significant predictor of mortality (Table 5). The most parsimonious model, containing a constant and the reflex impairment index, correctly predicted 91.0% of the mortality and survival — a percentage almost identical to that of the full model (82.8%). As with C. bairdi, cross- validation for C. opilio with a learning set resulted in 94.0% correct predictions for the test set. Consequently, models for both species were robust for the field study. Surface plots for probability of mortality with the logistic models for each species (Fig. 4) showed the strong relationships between reflex impairment and mortality. For C. bairdi, mortality increased rapidly Stoner et al.: An assessment of discard mortality for two Alaskan crab species, based on reflex impairment 343 Figure 3 Time to mortality (days) observed for Chionoecetes bairdi (Tanner crab) and C. opilio (snow crab) shown as a function of reflex impairment (index A). Values are mean, ±standard error (in parentheses). Index A was calculated as the sum of completely lost reflexes (score = 0) and index B was calculated as the sum of reflexes either lost or weak (scores = 0 or 1). with increasing impairment, reaching an asymptote at an index equal to 3. Mortality also increased with injury score, but the effect was smaller and more gradual. Probability of mortality in C. opilio was de- termined primarily by reflex impairment (with no ob- vious or significant effect of injury score). Reflex im- pairment and injury scores were correlated in both C. bairdi (P-0.514, Fl 248=89.024, P<0.001) and C. opilio (P-0.725, Fl 397=44i5.504, PcO.OOl). The number of crabs representing each reflex impair- ment index (0 to 6) was not uniform, and we wanted to be certain that these unequal numbers did not bias the resulting relationship between mortality and reflex im- pairment. For example, among the 399 C. opilio tested, reflex impairments equal to 0, 1, and 6 were over-repre- sented. Therefore, we randomly drew up to 20 individu- als for each impairment level, reducing the number of crabs entering the model to 115. The reduced database resulted in a relationship between mortality and reflex impairment that was identical to that incorporating the larger data set. A similar analysis for C. bair'di produced a similar result, showing that the uneven distribution of data did not bias the experimental outcome and that the logistic models shown in Figure 4 were robust. Discussion Behavior of animals reflects a host of internal and exter- nal conditions, and, in the context of fishing-related C. bairdi C. opilio 100 80 60 40 20 0 Figure 4 Surface plots, resulting from logistic regression, show- ing the probabilities of mortality for Chionoecetes bairdi (Tanner crab) and C. opilio (snow crab) with varied levels of reflex impairment (index A) and injury. Index A was calculated as the sum of completely lost reflexes (score = 0) and index B was calculated as the sum of reflexes either lost or weak (scores = 0 or 1). 100 80 jS 60 o £ 40 o .§■ 20 5 £ o O CL 6 5 344 Fishery Bulletin 106(4) Table 4 Results of logistic modeling for mortality in Chionoecetes bairdi (Tanner crab). A backward stepwise approach was used to determine the most parsimonious model for mortality, and an alpha value of 0.15 was required to remove a variable from the full model. Parameter Estimate t-ratio P-value Full model constant -2.346 -1.226 0.220 gender -0.114 -0.383 0.701 shell condition -0.274 -0.805 0.421 size 0.005 0.244 0.807 reflex impairment 1.316 5.364 <0.001 injury score 1.029 4.743 <0.001 Most parsimonious model Parameter constant -2.710 -8.957 <0.001 reflex impairment 1.353 5.731 <0.001 injury score 1.044 4.767 <0.001 Prediction matrix for the most parsimonious model No. dead predicted No. alive predicted Actual total Dead 51.4 21.6 73 Alive 21.6 155.4 177 Total predicted 73 177 250 Correct (%) 70.4 87.8 False (%) 29.6 12.2 Total correct (%) 82.7 Table 5 Results of logistic modeling for mortality in Chionoecetes opilio (snow crab). A backward stepwise approach was used to determine the most parsimonious model for mortality, and an alpha value of 0.15 was required to remove a variable from the full model. Parameter Estimate t-ratio P-value Full model constant -2.716 -1.933 0.053 gender 0.333 0.815 0.415 shell condition -0.089 -0.292 0.770 size -0.016 -1.128 0.259 reflex impairment 1.416 8.240 <0.001 injury score -0.102 -0.384 0.701 Most parsimonious model Parameter constant -3.913 -10.529 <0.001 reflex impairment 1.349 9.622 <0.001 Prediction matrix for the most parsimonious model No. dead predicted No. alive predicted Actual total Dead 78 18 96 Alive 18 285 303 Total predicted 96 303 399 Correct (%) 81.2 94.1 False (%) 18.8 5.9 Total correct (%) 91.0 Stoner et al.: An assessment of discard mortality for two Alaskan crab species, based on reflex impairment 345 injury, provides a neurological integration of both physi- ological stress and physical wounding (Davis, 2002). In fact, a variety of behaviors such as feeding (and growth) (Carls and O’Clair, 1990, 1995; Zhou and Shirley, 1995), molting (O’Brien et al., 1986; Carls and O’Clair, 1990), and righting behavior (Carls and O’Clair, 1990, 1995; Zhou and Shirley 1995; Warrenchuk and Shirley, 2002) are useful for evaluating crab condition. However, all these complex behaviors require some form of holding in seawater tanks for observation. For example, pre- liminary experiments at the Kodiak Fisheries Research Laboratory confirmed that lack of righting behavior in C. bairdi can be a useful predictor of mortality caused by both internal and externally visible injuries, but the evaluation can take >5 minutes, the behavior is highly variable, and yields only a binary response (i.e., yes or no). The vitality index described by Stevens (1990) also has just two possible outcomes — alive or moribund — for live crabs. In contrast, Davis and Ottmar (2006) and Davis (2007) have shown that testing a suite of reflex actions yields a graduated response variable with excel- lent potential for mortality prediction in fishes. They called the resulting relationship a reflex action mortal- ity predictor (RAMP). Our results show that a similar approach is possible for Chionoecetes spp. We identified six reflexes in Chionoecetes spp. that are stereotypic, repeatable, and easy to assess. These reflexes, associated with simple movements of the pri- mary limbs, mouth parts, and eye stalks, represent the most fundamental and involuntary responses of the crabs, and they can be rapidly assessed in hand (i.e., without holding a crab in water). The most sensitive indicators of trawl-related stresses observed in this study were kick and leg retraction. These reflexes were generally lost first, followed by leg flare and chela clo- sure, and reflexes associated with the eyes and mouth were least sensitive. The latter two reflexes probably require the lowest energy expenditures and movements of the mouth parts play an important role in ventilating the gills. Mouth movements were maintained in crabs near death and loss of motion could be used as the final determination of death (Stevens, 1990; this study). The order of reflex loss might shift somewhat under different fishing conditions or specific types of injury; however, a reflex impairment index, calculated by summing reflex actions without weighting individual reflexes (as in this study), has the advantage of representing condition over a wide range of stressor types in fishes (Davis and Ott- mar, 2006; Davis, 2007). The same representation over a wide range of stressors is likely for crabs. The most important finding of this study was that mortality in Chionoecetes spp. was closely correlated with reflex impairment. Although others have evalu- ated injuries to the exoskeleton in an attempt to predict mortality (Stevens, 1990), a thorough assessment of exoskeletal injuries is time consuming and minor and potentially fatal injuries, such as finely cracked cara- paces, are easily missed. On the other hand, internal injuries and bleeding that can occur without external injury can contribute significantly to delayed mortality (Grant, 2003). Rose (1999) reported exoskeletal injury rates for red king crabs captured or struck with dif- ferent types of trawl footropes, but it is now clear that these injury rates may or may not correlate well with mortality. Injury and reflex impairment were correlated in the present study, but our more general assessment of crab condition acquired through testing reflex actions provided a better predictor of mortality and an obvious improvement over tedious and subjective evaluations of shell damage. As observed with fishes (Davis, 2005, 2007), reflex impairment appears to integrate the ef- fects of different kinds of stress and injury that can occur in gear encounters and in routine handling and discard practices. Another key finding of this study was that the reflex impairment index provides a relatively universal indi- cator of condition and likely mortality for Chionoecetes spp. For example, shell condition can affect mortality rates in crabs handled in fishery operations and re- turned to sea (Kruse et al., 1994). However, our logistic analyses showed that reflex impairment indices pro- vided the best predictors of mortality in both C. bairdi and C. opilio, regardless of shell condition and crab gender and size. Similarly, reflex impairment indices have provided excellent predictions of mortality for several different fish species over a wide range of sizes, and a good integration of cumulative stress (Davis and Ottmar, 2006; Davis, 2007). Reflex action mortality predictors have also proven to be robust over a wide range of different stressor types, including tow time in a trawl, air exposure, and temperature shock, at least for fishes. The present study emerged from an interest in evaluating impacts of trawling operations on crab mortality rates and we believe that the response curves provided for C. bairdi and C. opilio should be robust for the kinds of injuries sustained under typical trawl operations in Alaska. The results, however, should be regarded primarily as a first proof of principle for crab species and the best possible RAMP models will depend upon additional fishing experiments conducted over a broader range of stressor types and fishing conditions. For example, other variables, such as freezing temperatures and windchill stress (Carls and O’Clair, 1995; Warrenchuk and Shirley, 2002) encountered in pot fishing warrant experimental investigation and may necessitate modi- fications to the mortality prediction models. Once a RAMP curve is well developed for a species it should be widely applicable in fishing experiments de- signed to improve fishing practices and reduce discard mortality. Instead of the traditional approach, where fishing and handling variables are evaluated directly through multiday holding, either in deck boxes (Stevens, 1990) or in sea cages (Grant, 2003), crabs can be evalu- ated immediately in the field, without holding, and prob- abilities of mortality can be estimated with the RAMP model. For example, we have been interested in mortal- ity of crabs that encounter trawls, but are not captured. In the future, we should be able to use recapture nets in front of and behind different trawl components (see 346 Fishery Bulletin 106(4) Rose, 1999), especially footropes, bridles, and sweeps, to quickly evaluate mortality for Chionoecetes spp. with the use of RAMP estimators. Neither assessments of exo- skeletal injuries nor holding in cages or deck boxes will be required. Similarly, the sublethal and lethal effects of fishing conditions such as bottom type, water and air temperature, net packing, and time in air could be evaluated with a RAMP approach. Effects of soak times for pots and handling procedures could be assessed with the same tool. Also, with an increasing interest in the live market for crabs, the RAMP approach may be useful in perfecting shipping and holding procedures for the best possible live products. The largest direct benefit of RAMP approach will be to increase the scope and replication of experiments on gear and handling. More broadly, discard-related mortality is not constant, and we believe that this new approach will facilitate sub- stantial improvements in fishery models that incorporate mortality for crabs, lobsters, and other crustaceans that are routinely discarded at sea. Acknowledgments This project was funded in part by the National Oceanic and Atmospheric Administration Bycatch Reduction Program and the North Pacific Research Board. D. King and J. Smart prepared the recapture nets, S. McEntire assisted with electronics, S. McDermott loaned the deck boxes for holding crabs, and S. Walters and K.-H. Lee participated in the field study. We are grateful to Cap- tain R. Haddon and crew of the FV Pacific Explorer for assistance with tank set-ups and handling our uncon- ventional fishing gear. T. Hurst provided suggestions on statistical approach and anonymous reviewers helped to improve the manuscript. Literature cited Alverson, D. L., M. H. Freeberg, S. A. Murawski, and J. G. Pope. 1994. A global assessment of fisheries bycatch and discards. FAO (United Nations Food and Agriculture Organization) Fish. Tech. Pap. 339, 235 p. Broadhurst, M. K., P. Suuronen, and A. Hulme. 2006. Estimating collateral mortality from towed fishing gear. Fish Fish. 7:180-218. Carls, M. G., and C. E. O’Clair. 1990. Influence of cold air exposure on ovigerous red king crabs (Paralithod.es camtschatica ) and Tanner crabs ( Chionoecetes bairdi ) and their offspring. In Proceed- ings of the international symposium on king and Tanner crabs, p. 329-343. Univ. Alaska Sea Grant College Program, AK-SG-90-04, Fairbanks, AK. 1995. Responses of Tanner crabs, Chionoecetes bairdi , exposed to cold air. Fish. Bull. 93:44-56. Coggins, L. G. Jr., M. J. Catalano, M. S. Allen, W. E. Pine III, and C. J. Walters. 2007. Effects of cryptic mortality and the hidden costs of using length limits in fishery management. Fish Fish. 8:196-210. Crear, B. J., and G. N. R. Forteath. 2001. Recovery of the western rock lobster, Panulirus cygnus, from emersion and handling stress: the effect of oxygen concentration during re-immersion. J. Shellfish Res. 20:921-929. Davis, M. W. 2002. Key principles for understanding fish bycatch discard mortality. Can. J. Fish. Aquat. Sci. 59:1834- 1843. 2005. Behaviour impairment in captured and released sablefish: ecological consequences and possible substi- tute measures for delayed discard mortality. J. Fish Biol. 66:254-265. 2007. Simulated fishing experiments for predicting delayed mortality rates using reflex impairment in restrained fish. ICES J. Mar. Sci. 64:1535-1542. Davis, M. W., B. L. Olla, and C. B. Schreck. 2001. Stress induced by hooking, net towing, elevated sea water temperature and air in sablefish: lack of con- cordance between mortality and physiological measures of stress. J. Fish Biol. 58:1-15. Davis, M. W., and M. L. Ottmar. 2006. Wounding and reflex impairment may be predic- tors for mortality in discarded or escaped fish. Fish. Res. 82:1-6. Davis, M. W., and C. B. Schreck. 2005. Responses by Pacific halibut to air exposure: lack of correspondence among plasma constituents and mortality. Trans. Am. Fish. Soc. 134:991-998. Grant, S. M. 2003. Mortality of snow crab discarded in Newfoundland and Labrador’s trap fishery: at-sea experiments on the effect of drop height and air exposure duration. Can. Tech. Rep. Fish. Aquat. Sci. 2481: vi + 28 p. Harrington, J. M., R. A. Myers, and A. A. Rosenberg. 2005. Wasted fishery resources: discarded by-catch in the USA. Fish Fish. 6:350-361. Harris, R. R., and M. B. Andrews. 2005. Physiological changes in the Norway lobster Nephrops norvegicus (L.) escaping and discarded from commercial trawls on the west coast of Scotland. 1. Body fluid volumes and haemolymph composition after capture and during recovery. J. Exp. Mar. Biol. Ecol. 320:179-193. Kruse, G. H., D. Hicks, and M. C. Murphy. 1994. Handling increases mortality of softshell Dunge- ness crabs returned to the sea. Alaska Fish. Res. Bull. 1:1-9. O’Brien, J. J., D. L. Mykles, and D. M. Skinner. 1986. Cold-induced apolysis in anecdysial brachy- urans. Biol. Bull. 171:450-460. Peduzzi, P. N., T. R. Holford, and R. J. Hardy. 1980. A stepwise variable selection procedure for nonlin- ear regression models. Biometrics 36:511-516. Ridgway, I. D., A. C. Taylor, R. J. A. Atkinson, E. S. Chang, and D. M. Neill. 2006. Impact of capture method and trawl duration on the health status of the Norway lobster, Nephrops norvegicus. J. Exp. Mar. Biol. Ecol. 339:135-147. Rose, C. S. 1999. Injury rates of red king crab, Paralithodes camts- chaticus, passing under bottom-trawl footropes. Mar. Fish. Rev. 61(2):72-76. Rosenkranz, G. E. 2002. Mortality of Chionoecetes crabs incidentally caught in Alaska’s weathervane scallop fishery. In Crabs in Stoner et al.: An assessment of discard mortality for two Alaskan crab species, based on reflex impairment 347 cold water regions: biology, management, and econom- ics (A. J. Paul, E. G. Dawe, R. Elner, G. S. Jamieson, G. H. Kruse, R. S. Otto, B. Sainte-Marie, T. C. Shirley, and D. Woodby, eds.), p. 717-732. Univ. Alaska Sea Grant College Program, Report AK-SG-02-01, Fairbanks, AK. Siddeek, M. S. M. 2003. Determination of biological reference points for Bristol Bay red king crab. Fish. Res. 65:427-451. Stevens, B. G. 1990. Survival of king and Tanner crabs captured by commercial sole trawls. Fish. Bull. 88:731-744. Suuronen, P. 2005. Mortality of fish escaping trawl gears. FAO (United Nations Food and Agriculture Organization). Fish. Tech. Pap. No. 478, 72 p. van Tamelen, P. 2005. Estimating handling mortality due to air exposure: Development and application of thermal models for the Bering Sea snow crab fishery. Trans. Am. Fish. Soc. 134:411-429. Warrenchuk, J. J., and T. C. Shirley. 2002. Effects of wind chill on the snow crab (Chionoece- tes opilio). In Crabs in cold water regions: biology, management, and economics (A. J. Paul, E. G. Dawe, R. Elner, G. S. Jamieson, G. H. Kruse, R. S. Otto, B. Sainte-Marie, T. C. Shirley, and D. Woodby, eds.), p. 81-96. Univ. Alaska Sea Grant College Program, Report AK-SG-02-01, Fairbanks, AK. Witherell, D., and C. Pautzke. 1997. A brief history of bycatch management measures for eastern Bering Sea groundfish fisheries. Mar. Fish. Rev. 59:15-22. Zhou, S., and T. C. Shirley. 1995. Effects of handling on feeding, activity, and sur- vival of red king crabs, Paralithodes camtschaticus (Tilesius, 1815). J. Shellfish Res. 14:173-177. 348 Abstract — The community structure of fishes associated with pelagic Sar- gassum spp. and open water lacking Sargassum was examined during summer and fall cruises, 1999-2003, in the Gulf Stream off North Caro- lina. Significantly more individual fishes ( n= 18,799), representing at least 80 species, were collected from samples containing Sargassum habi- tat, compared to 60 species (n = 2706 individuals) collected from open- water habitat. The majority (96%) of fishes collected in both habitats were juveniles, and planehead filefish ( Stephanolepis hispidus ) dominated both habitats. Regardless of sam- pling time (day or night), Sargassum habitat yielded significantly higher numbers of individuals and species compared with open-water collections. Overall, fishes collected by neuston net tows from Sargassum habitat were significantly larger in length than fishes collected from open-water habi- tat with neuston nets. A significant positive, linear relationship existed between numbers of fishes and the quantity of Sargassum collected by neuston net. Underwater video record- ings indicated a layered structure of fishes among and below the algae and that smaller fishes were more closely associated with the algae than larger fishes. Observations of schooling behaviors of filefishes (Monacanthi- dae), dolphinfish ( Coryphaena hip- purus), and jacks (Carangidae), and fish-jellyfish associations were also recorded with an underwater video camera. Our data indicate that Sar- gassum provides a substantial nurs- ery habitat for many juvenile fishes off the U.S. southeast coast. Manuscript submitted 12 February 2008. Manuscript accepted 6 June 2008. Fish. Bull. 106:348-363 (2008). The views and opinions expressed or implied in this article are those of the author and do not necessarily reflect the position of the National Marine Fisheries Service, NOAA. Fishes associated with pelagic Sargassum and open water lacking Sargassum in the Gulf Stream off North Carolina Tara L. Casazza (contact author) Steve W. Ross Email address for T.L. Casazza: tarac73@juno.com Center for Marine Science University of North Carolina Wilmington 5600 Marvin Moss Lane Wilmington, North Carolina 28409 In the western North Atlantic Ocean, pelagic brown algae of the genus Sar- gassum form a dynamic, floating habi- tat that supports a diverse assemblage of fishes, invertebrates, sea turtles, pelagic birds, and marine mam- mals. The pelagic species S. natans and S. fluitans provide resources in an otherwise nutrient-poor environ- ment and serve as a nursery area for many juvenile fishes (Wells and Rooker, 2004), some of which are com- mercially or recreationally important, or both (e.g., Coryphaena hippurus [dolphinfish], Caranx spp. [jacks], Seriola spp. [amberjacks]). Sargas- sum habitat appears to be particularly important for early survival of some fishes because the majority of fishes collected from Sargassum habitat in the Gulf of Mexico (Bortone et ah, 1977; Wells and Rooker, 2004) and off the southeastern United States (Dooley, 1972; Moser et ah, 1998) are juveniles. The spatial distribution and quan- tity of Sargassum are highly variable. Sargassum distribution along the U.S. east coast depends on the Florida Current and the Gulf Stream, which entrain pelagic Sargassum from the Sargasso Sea and move it northward. Aggregations of Sargassum range from small, widely dispersed clumps to rafts and large weedlines that con- tinue for many kilometers, and the great variability in the structure of this habitat is due to variations in Gulf Stream flow, storms, tidal cur- rents, and wind-generated waves and currents. Although estimates of Sar- gassum biomass in the western North Atlantic have varied (Howard and Menzies, 1969; Butler and Stoner, 1984), the majority of pelagic Sar- gassum has persisted and reproduced vegetatively in the western North At- lantic Ocean for at least decades and probably for hundreds to thousands of years (Parr, 1939). Sargassum habitat is extremely difficult to sample consistently and quantitatively, and no single meth- od of sampling provides a complete view of the Sargassum community. Even though Moser et al. (1998) rec- ommended using multiple sampling methods, especially visual methods, to survey this ecosystem, most Sargas- sum community studies from the Gulf of Mexico (Bortone et al., 1977; Wells and Rooker, 2004) and the western North Atlantic (Dooley, 1972; Moser et al., 1998) have been based on limited sampling methods. Kingsford (1995) and Dempster and Kingsford (2004) suggested that a weakness in previ- ous studies was a lack of open-water, unvegetated control samples, and to date, in only one study (Moser et ah, 1998) in the western North Atlantic have fishes associated with Sargas- sum habitat been compared to fishes in open-water habitat. Additionally, most samples of Sargassum were col- lected during daytime only (Dooley, 1972; Moser et al., 1998; Wells and Rooker, 2004) or sampling times were not specified (Bortone et ah, 1977; Stoner and Greening, 1984). In some cases, explicit association of samples with Sargassum was unclear because samples were not collected within the algae (e.g., Settle, 1993). Casazza and Ross: Fishes associated with pelagic Sargassum and open water off North Carolina 349 Figure t Surface sampling sites for fishes collected during summer and fall of 1999-2003 off North Carolina. Collections with neuston nets (triangles) and supplemental gears (circles) (i.e. , nightlighting, dip net, hook-and line, longline) from Sargassum (closed symbols) and open-water (open symbols) habitats. The white circles with a black center off Cape Hat- teras represent sites where underwater video recordings were taken. The relationships between the quantity of Sargassum and species richness and abundance or biomass of fishes are high- ly variable. Dooley (1972) and Fedoryako (1980) found no correlation between num- bers of fishes and quantity of Sargassum, but significant positive correlations be- tween fish abundances and quantity of algae have been found in other studies (Moser et al., 1998; Wells and Rooker, 2004). Sampling methods may substan- tially influence these results. Neverthe- less, it is clear that objects floating in the ocean, like Sargassum, attract and concentrate fauna (Kingsford, 1992). Despite several surveys and the wide- spread occurrence of Sargassum habitat, the fishes associated with this habitat have not been extensively documented along the U.S. east coast. Our objective was to describe fish community structure (species composition, day versus night species composition, sizes, and Sargas- sum abundance and fish distribution relationships) within Sargassum and open-water habitats off North Carolina. In addition, behaviors of fishes with- in and below Sargassum habitat were documented to better describe the close associations of fishes with the habitat, the different types of habitat usage, and to provide a three dimensional view of the distribution of fishes within and beneath the Sar- gassum. Our approach was to use consistent methods (supplemented by further sampling) and extensive tem- poral (diel) and spatial sampling across ocean surface habitats with no Sargassum to those with high densities of Sargassum to examine the relative contribution of Sargassum to oceanic fish communities off the south- eastern United States. Materials and methods Sampling Surface waters were sampled during 2-7 August 1999, 20-27 July 2000, 22-28 August 2001, 20-26 September 2001, 6-15 August 2002, and 17-26 August 2003, in the Gulf Stream off Cape Hatteras, Cape Lookout, and Cape Fear, North Carolina (Fig. 1). The primary sampling device, a l.lx2.4-m neuston net (6.4-mm mesh body, 3.2 -mm tailbag), was towed in the upper meter of the water column at slow speeds (<3.7 km/h) for 30 minutes in 1999 and for 15 minutes during 2000-03. Sampling was conducted throughout the 24-h period to compare daylight (0700-2000 h eastern daylight time [EDT]) and nighttime (-0700 h EDT) collections. The neuston net was towed in both open water without Sargassum and in waters with varying amounts of Sargassum. Sargassum is in constant motion in the Gulf Stream, and without aerial surveillance its distribution and den- sity are unpredictable, especially at night. Because we were unable to consistently target a particular habitat, and thus balance sampling effort, the nets were pulled through unknown habitat and the sample was classified later depending on whether Sargassum was present in the sample or not (see Data analysis). In most cases it was also not possible to determine the proximity of one habitat to another. When Sargassum was abundant, the neuston net was towed directly through the clumps, mats, or weedlines, but on some occasions, Sargassum was collected opportunistically. Fishes were sorted from the algae in the neuston tow catches, and the Sargassum was weighed (wet weight) to the nearest 0.1 kg and dis- carded. Because neuston net tows in 1999 were of longer duration and catches were not handled consistently, the catch data from the 1999 neuston net tows were not analyzed statistically (see Data analysis). Additional collection methods supplemented the use of neuston nets, especially when Sargassum was too dense for use of the neuston net. When conditions were favor- able (low wind and waves), stations with nightlighting were sampled by allowing the vessel to drift with the current or maintain its slowest speed into the current. The deck lights of the vessel, plus two 500-W and one 1000-W spotlights, were used to illuminate surface waters around the stern and both sides of the vessel, 350 Fishery Bulletin 106(4) and fishes that swam near the vessel were collected by five or six crew members using small mesh (6.4-mm mesh) dip nets. Each 30-minute segment of time during the drift represented a station. During the nightlight- ing sampling, the presence or absence of Sargassum within the field of view was recorded, and if present, whether the Sargassum was collected in dip nets was recorded. Fishes were also opportunistically collected with dip nets during daylight when dense aggregations of Sargassum were encountered. Limited hook-and-line sampling occurred in both Sargassum and open-water habitats during the day and at night, and each sam- pling period (station) lasted from 15 to 160 minutes. One longline set was made in the Cape Hatteras study area. The line was about 366-m long and contained 104 baited hooks that fished within 1-2 m of the surface. The set was made at night, lasted for 501 min, and drifted for 30 km through open-water habitat. In 1999, underwater video was recorded under a large Sargassum weedline at two stations off Cape Hatteras, North Carolina (Fig. 1). Snorkelers using a handheld color camcorder (SONY model DCR-TRV900, New York, NY) in a waterproof case swam at and just below (<3 m) the surface along the edge of and under the weedline. A total of 62 minutes of video footage was recorded dur- ing the two stations. Analyses of the underwater video footage included identification of species, documentation of behaviors, and placement of fishes within or below the weedline. Specimens were preserved at sea in 10% formalin- seawater solution and later stored in 40% isopropanol. Larval fishes had been collected in previous Sargassum studies, and this fact implied an association with this habitat. However, because distributions of pelagic fish larvae are highly influenced by currents and they gener- ally lack affinity for drift algae (Kingsford and Choat, 1985), their presence in Sargassum collections (Settle, 1993; Wells and Rooker, 2004) is probably coincidental. For this reason and because the neuston net mesh size was inappropriate for sampling larvae, larval fishes (classified according to Richards, 2006) were excluded from this study. Fishes were identified to the lowest pos- sible taxon, counted, measured to the nearest mm for standard length (SL), and weighed (wet weight) to the nearest 0.1 g. Damage to some fishes precluded identi- fication to species and SL measurements. When more than 500 individuals of the same species were collected in a tow, a subsample (approximately 10% of the catch) was measured for SL and wet weight. Data analysis Fish catches from neuston nets were analyzed statisti- cally to assess differences in fish community structure between habitats, and diurnal differences in community structure. Neuston tows without Sargassum were des- ignated as open-water (OW). Because clumps of algae as small as 0.005 kg could influence the distribution and abundance of fishes (Kingsford and Choat, 1985), samples were classified as Sargassum (S) if algae were collected, regardless of the quantity. The number of individuals and number of species collected from Sargas- sum and open-water habitats were log (x+1) transformed before analysis to correct for heterogeneity of variance, to reduce the influence of abundant species, and to enhance the contribution of rare species. If the assump- tions of homogeneity of variance and normality were not satisfied after data transformation, a nonparametric Mann-Whitney test was applied to determine whether there were differences in the number of individuals and species in Sargassum versus open-water habitat. A Kruskal-Wallis test was used to compare day and night fish catches from neuston nets within and across station types (i.e., S versus OW), and a Dunn’s multiple compari- son test was used to determine where significant differ- ences occurred. The relationship of fish abundance and species richness to the quantity of Sargassum collected by neuston nets was evaluated with regression analysis. Length-frequency distributions for dominant species col- lected from Sargassum habitat were compared to the size structures of the same species collected from open-water habitat by using a Kolmogorov-Smirnov test. Habitat type sampled (S versus OW) was also des- ignated for the supplemental methods. If Sargassum was collected by dip net, the station was designated as S; otherwise it was OW. Likewise, if the hook-and- line gear was placed in Sargassum (S), the catch was designated as S; if the gear was placed in unvegetated habitat, the catch was designated as OW. Results Catch composition For all methods and cruises combined, most fishes were collected in samples containing Sargassum habitat. A total of 18,799 fishes, representing 80 species from 28 families, were collected in 162 Sargassum samples, and a total of 2706 fishes, representing 60 species from 23 families, were collected in 80 open-water samples (Fig. 1; Table 1). Both Sargassum and open-water collections were dominated by the families Monacanthidae (75% of S, 45% of OW), Carangidae (13%, 21%), and Exocoetidae (6%, 19%). Individuals of nine species represented 93% of the total Sargassum catch (in decreasing order of abundance): Stephanolepis hispidus (planehead filefish), Caranx crysos (blue runner), Cheilopogon melanurus (Atlantic flyingfish), Balistes capriscus (gray triggerfish), Seriola rivoliana (almaco jack), Parexocoetus brachyp- terus (sailfin flyingfish), Monacanthus ciliatus (fringed filefish), Decapterus punctatus (round scad), and Cory- phaena hippurus (dolphinfish). Individuals of 10 spe- cies represented 92% of the total open-water catch (in decreasing order of abundance): S. hispidus, C. crysos, Clupea harengus (Atlantic herring) (all from a single station), C. melanurus, P. brachypterus, D. punctatus, Prognichthys occidentalis (bluntnose flyingfish), Oxypor- hamphus micropterus (smallwing flyingfish), Istiophorus platypterus (sailfish), and C. hippurus. For all methods Casazza and Ross: Fishes associated with pelagic Sargassum and open water off North Carolina 351 352 Fishery Bulletin 106(4) Casazza and Ross: Fishes associated with pelagic Sargcissum and open water off North Carolina 353 354 Fishery Bulletin 106(4) Casazza and Ross: Fishes associated with pelagic Sargassum and open water off North Carolina 355 combined, 33 species were collected only in association with Sargassum habitat, and 13 species were unique to open-water habitat (Table 1). There was a large discrepancy between Sargassum and open-water catches from the 2000-03 neuston net samples. A total of 14,123 fishes, representing 65 species, were collected in 91 neuston tows in Sargas- sum habitat. Thirteen open-water tows produced no catch, whereas 14 open-water tows yielded 1393 fishes, representing 27 species (Table 2). Dominant families collected by neuston net in both Sargassum and open- water habitats were Monacanthidae (83% of S, 79% of OW), Carangidae (9%, 13%), and Exocoetidae (4%, 6%). Individuals of eight species represented 95% of the total Sargassum catch with neuston nets (in decreasing order of abundance): S. hispidus, C. crysos, C. melanurus, B. capriscus, M. ciliatus, S. rivoliana, P. brachypterus, and C. hippurus. Individuals of four species represented 93% Table 2 Number of fishes collected in neuston net tows from Sargassum and open-water habitat off North Carolina during 2000-03, separated by day and night. Number of samples is in parentheses. Species are listed in phylogenetic order. Sargassum Open water Sargassum Open water Day Night Day Night Day Night Day Night Species (47) (44) (19) (8) Species (47) (44) (19) (8) Cyclothone sp. 1 0 0 0 Caranx bartholomaei 2 1 0 0 Argyropelecus aculeatus 1 0 0 0 Caranx crysos 342 468 31 43 Diaphus dumerilii 1 0 0 0 Caranx ruber 23 33 0 0 Myctophum affine 0 27 0 0 Caranx spp. 2 18 1 1 Myctophum obtusirostre 0 10 0 0 Decapterus punctatus 9 44 78 1 Myctophum punctatum 1 4 0 0 Decapterus spp. 4 7 0 0 Myctophum selenops 0 1 0 0 Elagatis bipinnulata 5 5 0 0 Myctophum sp. 0 1 0 0 Selar crumenophthalmus 2 0 0 0 Histrio histrio 7 2 0 0 Selene setapinnis 0 0 1 0 Mugilidae 0 2 0 0 Seriola dumerili 3 4 0 0 Ablennes hians 6 8 1 2 Seriola fasciata 30 21 4 0 Platybelone argalus 0 2 0 0 Seriola rivoliana 125 35 13 0 Tylosurus acus 1 2 0 0 Seriola spp. 12 9 3 1 Tylosurus spp. 0 2 0 0 Seriola zonata 0 4 1 0 Cheilopogon cyanopterus 0 6 0 0 Lobotes surinamensis 9 1 0 0 Cheilopogon exsiliens 0 2 0 0 Kyphosus incisor 0 2 0 0 Cheilopogon furcatus 0 2 0 0 Kyphosus sectatrix 11 1 0 0 Cheilopogon melanurus 31 365 5 49 Kyphosus sp. 1 0 0 0 Cheilopogon spp. 0 6 0 0 Abudefduf saxatilis 5 5 0 0 Cypselurus comatus 0 1 0 0 Istiophorus platypterus 1 3 1 1 Hirundichthys a f finis 0 13 0 6 Psenes cyanophrys 0 5 0 0 Oxyporhamphus Batistes capriscus 120 109 9 5 micropterus 0 2 0 0 Canthidermis maculata 9 1 0 0 Parexocoetus Canthidermis sufflamen 23 3 0 1 brachypterus 19 140 2 17 Xanthichthys ringens 1 0 0 0 Prognichthys occidentalis 9 26 1 6 Balistidae 0 2 0 0 Euleptorhamphus velox 1 10 0 2 Aluterus heudelotii 48 17 0 0 Hemiramphus balao 3 7 0 0 Aluterus monoceros 1 3 0 1 Hemiramphus brasiliensis 0 21 1 o Aluterus schoepfi 10 7 0 1 Hemiramphus spp. 2 31 0 0 Aluterus scriptus 31 8 0 0 Hyporhamphus Aluterus sp. 0 1 0 0 unifasciatus 0 1 0 0 Cantherhines macrocerus 2 3 0 0 Bryx dunckeri 2 1 1 0 Cantherhines pullus 3 9 2 0 Hippocampus erectus 3 8 0 1 Monacanthus ciliatus 75 97 1 6 Hippocampus reidi 1 1 0 0 Monacanthus tuckeri 9 4 0 0 Hippocampus sp. 1 0 0 0 Monacanthus sp. 1 0 0 0 Syngnathus louisianae 1 0 0 0 Stephanolepis hispidus 7840 3541 681 408 Syngnathus pelagicus 1 0 1 0 Stephanolepis setifer 1 0 0 0 Fistularia sp. 0 0 1 0 Chilomycterus sp. 0 1 0 0 Synagrops bellus 0 1 0 0 Diodon holocanthus 1 5 0 0 Coryphaena equiselis 2 0 0 1 Diodon hystrix 1 1 0 0 Coryphaena hippurus 13 71 1 0 Total 8869 5254 840 553 356 Fishery Bulletin 106(4) of the total open-water catches with neuston nets (in decreasing order of abundance): S. hispidus, D. punc- tatus, C. crysos, and C. melanurus. There were sig- nificantly more individuals (Mann-Whitney test: df=117, P<0.001) and numbers of species (Mann-Whitney test: df=117, P<0.001) in Sargassum habitat compared with open-water habitat. The three most abundant species in neuston net collections containing Sargassum habitat also exhibited the highest frequencies of occurrence: S. hispidus (70% of samples), C. crysos (64%), and C. melanurus (46%), whereas in open-water habitat these species occurred in 41%, 19%, and 22% of samples, respectively. Forty of the total 65 species collected in 2000-03 neuston net tows in Sargassum were unique to this habitat, whereas only two ( Fistularia sp., Selene setapinnis) of the total 27 species collected in open-wa- ter habitat were unique (Table 2). Day versus night catch composition Regardless of sampling time (day or night), Sargassum habitat yielded significantly (Kruskal-Wallis test: df=3, P<0.05) higher numbers of individuals and species than open-water habitat. Daytime neuston net samples from Sargassum habitat (n = 47) yielded 8869 fishes from 48 species, and nighttime neuston net samples from Sargas- sum habitat (n = 44) yielded 5254 fishes from 56 species (Table 2); however, these differences were not statisti- cally significant (Kruskal-Wallis test: df=3, P=0.924). Supplemental methods used in Sargassum habitat (dip nets, hook-and-line, 1999 neus- ton net) yielded different results; slightly more fishes (350 individu- als) were collected at night than during the day. This finding was likely due to the slightly higher effort at night and the attraction of fish by the nightlighting. As above, most fishes from neus- ton net samples in open-water habitat were collected during the day. Ten of the total 19 daytime neuston net tows in open-water habitat yielded 840 fishes, rep- resenting 20 species, and four of the total eight nighttime neuston net samples in open water yielded 553 fishes, representing 18 species (Table 2); however, these differ- ences were not statistically signifi- cant (Kruskal-Wallis test: d f = 3 , P= 0.843). Supplemental methods used in open-water habitat (dip nets, hook-and-line, 1999 neuston net, long line), as above, produced more fishes at night (213 more in- dividuals), probably for the same reasons. Size distributions Sargassum Stephanolepis hispidus 1 000 -i (n=3834) <6 & rlO iS> cP c$> o$> M. cAcsA c- ^ rfO O' N" \° ‘Sop CPCpCp' & '■S° Open water Stephanolepis hispidus (n= 487) 200 20 0 80 - 20 - 60 20 0 ■flFflFfflTT'rrT ' l.a>- Caranx crysos (n= 349) ! 0“i i m n i i i i i s i i t i Cheilopogon melanurus - (n=1 56) ,._ll*flQlT!riTrTTn i m n i i i i i i i i r i im 1 Balistes capriscus - (n=1 9) Seriola rivoliana - (n=15) ■T-rT-ff r~ £ $ $ & & & £ & *0' <£>' rif> ' rp <&' riP

300 m) and shortraker and rougheye rockfish (-250-500 m), indicating that these species could po- tentially be competing with sablefish for longline hooks. Also, experimen- tal evidence indicates that sablefish are adept at finding baited hooks on the longline surveys, even when few Rodgveller et al: Evidence of hook competition in longline surveys 365 Stations sampled (▲) during the annual National Oceanic and Atmospheric Admin- istration Alaska Fisheries Science Center (AFSC) sablefish ( Anoplopoma fimbria) longline surveys, 1979-2003 within the six sablefish management areas (1-Bering Sea, 2-Aleutian Islands, 3-Western Gulf of Alaska, 4-Central Gulf of Alaska, 5-West Yakutat, and 6-East Yakutat and Southeast [Alaska]). The AFSC Gulf of Alaska slope trawl survey stations (not shown) are also located along the continental slope above the 1000-m isobath. remain (Sigler, 2000). Therefore, it is more likely that sablefish are outcompeting grenadier and shortraker and rougheye rockfish for hooks. Negative correlations between catch rates of two spe- cies could be evidence of gear saturation, or could be caused by other factors such as differing habitat pref- erences or direct competition between species. If the cause of negative correlations is competition for hooks, catch rate trends or depth distribution shifts could be confounded. Additionally, it may not be possible to evaluate the effects of factors such as habitat and other environmental variables until competition effects are taken into account. For this study, we examined the cor- relations between sablefish and grenadier and between sablefish and shortraker and rougheye rockfish to iden- tify any negative relationships on the longline surveys. We also compared correlations of catch rates from the longline surveys to correlations of catch rates from the AFSC bottom trawl surveys in the Gulf of Alaska to investigate the cause of the relationships on the longline surveys (Britt and Martin, 2001). Because trawl survey catches are not generally susceptible to gear saturation, catches are a reflection of natural fish densities and therefore a good control to compare to longline catch rates (Gunderson, 1993). Although feeding history, wa- ter temperature, and other variables could affect the strength of competition between sablefish and grenadier and between sablefish and shortraker and rougheye rockfish, we were unable to control for these variables in this analysis. Materials and methods Sampling AFSC longline surveys The continental slope throughout the eastern Bering Sea, the eastern Aleutian Islands, and the Gulf of Alaska have been systematically sampled each summer since 1979 during the annual AFSC long- line surveys. The survey is divided among six sablefish management areas: 1) Bering Sea; 2) Aleutian Islands; 3) Western Gulf of Alaska; 4) Central Gulf of Alaska; 5) West Yakutat; and 6) East Yakutat and Southeast (Alaska). Stations are placed 30-50 km apart and at each station depths from 150 to 1000 m are sampled (Fig. 1). Catches are pooled by management area and an abundance index is computed (Sigler, 2000). The gear on the survey closely resembles gear used by the sablefish fishery in Alaska. The basic unit of gear is a skate; each skate consists of 45 hooks, baited with squid, spaced 2 m apart. At the end of each skate a 3-kg lead ball is 366 Fishery Bulletin 106(4) attached to ensure the gear is fished on the bottom. Each station consists of 160 skates set across depth contours from 150 to 1000 m. At each station there are 7200 hooks set over 16 km. The objective of the setting pattern is to evenly distribute sampling effort over the depths that sablefish inhabit. Gear retrieval begins after the gear has soaked for 3 hours, and fish and baited hooks are counted as they are brought aboard. Catch rates are tab- ulated as the number of fish of each species per skate of gear (no. fish/45 hooks). Bottom depth is recorded every five skates as the gear is hauled. Interpolated depths are assigned to skates that do not have a recorded depth. Shortraker and rougheye rockfish catches were pooled for this analysis because they are not distinguished from one another during gear retrieval. These species are very similar in appearance and difficult to distin- guish, share the same habitat on the upper continental slope, and often are found in fishery hauls (Clausen and Fujioka, 2005). In this article we will refer to both species as rockfish. AFSC trawl surveys The AFSC groundfish trawl sur- veys sample the continental shelf and upper continental slope of the Gulf of Alaska at depths to 500 m, and in some years as deep as 1000 m, during the summer (Britt and Martin, 2001; Fig. 1). The surveys follow a stratified random sampling pattern with 49 strata that are categorized by depth, geological area (e.g., gully, slope), and management area (i.e., latitude and longi- tude). Trawls are hauled at a vessel speed of 3 knots for 15 minutes, or in some years, for 30 minutes, and tows are conducted along a constant bottom depth. An average depth is assigned to each trawl haul. All fish caught on the surveys are enumerated and a catch rate is calculated by dividing the number of fish caught by the area swept by the net (no. fish/km2). Analysis Longline correlations Average catch rates from the 1979-2003 annual longline surveys were calculated by 50-m depth increments to determine the preferred depth range for grenadier and rockfish in each sablefish man- agement area. Analyses were done separately by man- agement area because preferred depth ranges and catch rates have differed in each area for each fish species. The preferred depth ranges of grenadier and rockfish in each area were defined as the range where the average catch rate, in all years, was at least 20% of the highest catch rate. This method resulted in disregarding depths where the catch rate was less that 20% of the average peak catch rate. Our intention was to consider only depths where grenadier or rockfish were prevalent. An average catch rate was then calculated for the preferred depth range for grenadier and rockfish in each management area in each year at each longline survey station. Catch rates were also calculated for sablefish in the grenadier and rockfish preferred depth ranges so that sablefish catch rates could be compared to grenadier and rockfish catch rates. Preferred depth ranges were calculated from longline survey catch rates instead of trawl survey catch rates because there is little fishing effort below 500 m on the trawl surveys. In addition to calculating the average catch rate for the whole preferred depth range, we also calculated the average catch rate by 50-m increments within the preferred depth range in order to make more fine-scale species comparisons. Before tests for statistical significance, catch rates were transformed by using natural logarithm, square root, or fourth root transformations to help meet as- sumptions of normality. Pearson’s correlation coefficients and tests of significance were calculated. Catch rates calculated as catch per skate did not meet the assump- tions of normality due to a prevalence of zeros. There- fore, average catch rates of the preferred depth range by station were used to calculate correlation coefficients. Longline versus trawl correlations Because longline and trawl data are collected at the same time, areas, depths, and from the same habitats, trawl correlations are a good comparison for longline correlations. In the trawl data, a positive or zero correlation indicates that species do not have differing habitat preferences and that the species do not directly compete with each other. A negative correlation on longline gear and a positive or zero correlation in trawl gear would indicate that the negative correlations in longline gear could be caused by hook competition and not by these other factors. Trawl catch rates were computed for each haul for each species in each year the trawl surveys occurred (1984, 1987, 1990, 1993, 1996, 1999, 2001, 2003, and 2005). Because depth ranges varied throughout the Gulf of Alaska, the entire depth distribution of grena- dier and rockfish throughout the Gulf of Alaska was included in this analysis. The depth distribution was rounded to the nearest 50 m because longline survey catches were already summarized by 50 m increments. For grenadier the range was 300-700 m and for rock- fish the depth range was 250-550 m. Catches deeper than 700 m were not included because of the limited trawl survey effort below this depth. Trawl catch rates were also transformed to help meet the assumptions of normality. In one case, namely for the rockfish trawl data, the data could not be transformed to fit a normal distribution, but the transformation was retained be- cause it aided in the visual analysis of the data. In this instance, nonparametric correlation tests, Spearman’s p and Kendall’s r, were used to calculate correlation coefficients and to test for significance. Longline catch rates were summarized to be comparable with trawl catch rates; average catch rates were computed for the same depth ranges and years at each station. Results Longline catch rates by area Preferred depths of grenadier and rockfish differed by management area (Table 1). Peak catch rates also dif- Rodgveller et al: Evidence of hook competition in longline surveys 367 Table 1 Preferred depth ranges (m) and average peak number of fish caught per 45 hooks (one skate of gear) for giant grenadier ( Alba - trossia pectoralis) (grenadier) and rockfish (shortraker [Seiostes borealis] and rougheye [Sefoasies aleutianus] rockfish) caught during the National Oceanic and Atmospheric Administration Alaska Fisheries Science Center annual longline surveys in the six management areas, 1979-2003. Preferred depth range is defined as the range in which the catch rate was at least 20% of the average peak catch rate (peak catch). Area Grenadier Rockfish Preferred depth (m) Peak catch Preferred depth (m) Peak catch Bering Sea 350-1000 5.3 250-550 1.0 Aleutian Islands 400-1000 9.8 250-500 4.1 Western Gulf 400-1000 14.4 250-450 2.8 Central Gulf 350-1000 8.8 300-500 1.8 West Yakutat 350-1000 5.5 250-500 4.2 East Yakutat and Southeast 550-1000 3.5 300-600 4.8 Table 2 Average number of giant grenadier ( Albatrossia pectoralis) (grenadier) and rockfish (shortraker [Sebastes borealis ] and rough- eye [Sebastes aleutianus ] rockfish) caught per 45 hooks (one skate of gear) in their preferred depth range in each management area during the National Oceanic and Atmospheric Administration Alaska Fisheries Science Center annual longline surveys, 1979-2003. Preferred depth range is defined as the range in which the catch rate was at least 20% of the average peak catch rate. The average sablefish (Anoplopoma fimbria) catches (SB catch) and the average number of baited hooks retrieved in the preferred depth ranges are also shown. Total catch is the sum of the average number of grenadier and sablefish caught per skate or the sum of the average number of sablefish and rockfish caught per skate. Catches and baited hooks are an average from 1995 through 2003 because baited hooks were not counted before 1995. Grenadier Rockfish Area Grenadier catch SB catch Baited hooks Total catch Rockfish catch SB catch Baited hooks Total catch Bering Sea 6.6 3.9 14.9 10.5 3.5 8.6 18.9 12.1 Aleutian Islands 15.2 5.0 7.9 20.2 4.3 3.0 9.0 7.3 Western Gulf 18.1 6.4 7.2 24.5 3.0 9.2 4.0 12.2 Central Gulf 9.1 10.5 9.1 19.6 1.4 8.4 10.2 9.8 West Yakutat 3.8 10.6 15.3 14.4 4.3 5.5 19.8 9.8 East Yakutat and Southeast 3.1 11.3 17.9 14.4 3.3 9.5 16.4 12.8 fered substantially by species in each area. For example, in the East Yakutat and Southeast area the preferred rockfish depth range was 300-600 m and the peak catch rate was 4.8 fish/45 hooks, but in the Western Gulf area it was 250-450 m and the peak catch rate was 2.8 fish/45 hooks. The highest grenadier catch rates occurred in the Western Gulf area and were also relatively high in the Central Gulf and the Aleutian Islands areas. The highest rockfish catch rates were in the East Yakutat and Southeast area, West Yakutat, and the Aleutian Islands areas (Table 1). The number of baited hooks retrieved was similar at grenadier and rockfish preferred depths in each area; however, in the Aleutian Islands, Western Gulf, and Central Gulf areas, the total number of sablefish and rockfish caught at rockfish preferred depths was less than the number of sablefish and grenadier caught at grenadier preferred depths (Table 2). This finding indi- cates that there were many fish species other than sa- blefish and rockfish caught at rockfish preferred depths than at grenadier preferred depths. Longline correlations In 11 out of 12 cases, there were significant negative cor- relations between sablefish and grenadier and between sablefish and rockfish catch rates. The correlations between sablefish and grenadier catch rates were nega- tive in all six management areas. Correlations between sablefish and rockfish catch rates were negative in five of the six management areas (Table 3). Sample sizes varied between grenadier and rockfish in some areas because, 368 Fishery Bulletin 106(4) Table 3 Correlations between longline catch rates of giant grenadier ( Albatrossia pectoralis ) (grenadier), sablefish ( Anoplopoma fimbria), and rockfish (shortraker [Sebastes borealis ] and rougheye rockfish [Sebastes aleutianus] combined) caught during the National Oceanic and Atmospheric Administration Alaska Fisheries Science Center annual longline surveys in the six management areas, 1979-2003. The Pearson’s correlation coefficient (r) and the P-value associated with the significance of the correlation (P) is shown as well as the sample size (n). Grenadier- Sablefish Rockfish-Sablefish Area r P n r P n Bering Sea -0.23 0.0006 243 -0.31 <0.0001 243 Aleutian Islands -0.37 <0.0001 264 0.03 0.6626 264 Western Gulf -0.38 <0.0001 243 -0.27 <0.0001 243 Central Gulf -0.36 <0.0001 375 -0.22 <0.0001 368 West Yakutat -0.38 <0.0001 200 -0.28 <0.0001 200 East Yakutat and Southeast -0.46 <0.0001 269 -0.43 <0.0001 400 Table 4 Strongest correlations between giant grenadier ( Albatrossia pectoralis) (grenadier) and sablefish (Anoplopoma fimbria) catch rates and between rockfish (shortraker [Sebastes borealis ] and rougheye [Sebastes aleutianus ] combined) and sablefish catch rates by 50-m depth increments. Data were collected during the National Oceanic and Atmospheric Administration Alaska Fish- eries Science Center annual longline surveys, 1979-2003. The Pearson’s correlation coefficient (r), the P-value associated with the significance of the correlation (P), and the sample size in) are reported, n is often larger for rockfish because in some years the deeper areas, where grenadier reside, were not sampled at each station. Catch rates from the Bering Sea for giant grenadier and sablefish could not be transformed to fit a normal distribution; therefore, no data are presented. Area Grenadier-Sablefish Rockfish-Sablefish Depth (m) r P n Depth (m) r P n Bering Sea 350-400 -0.35 <0.0001 199 Aleutian Islands 750-800 -0.44 <0.0001 218 400-450 -0.09 0.1751 218 Western Gulf 450-500 -0.51 <0.0001 239 250-300 -0.31 <0.0001 244 Central Gulf 600-650 -0.51 <0.0001 356 300-350 -0.40 <0.0001 366 West Yakutat 700-750 -0.50 <0.0001 187 350-400 -0.35 <0.0001 199 East Yakutat and Southeast 600-650 -0.44 <0.0001 341 400-450 -0.53 <0.0001 393 at some stations in some years, gear was not set in the preferred depth range for grenadier or rockfish. For example, in the East Yakutat and Southeast area there were 269 station/year combinations for grenadier and 400 for rockfish because gear was not set deep enough for the preferred grenadier depth range at some stations in some years. At certain 50-m depth intervals, within the preferred depth range, correlations were more strongly negative than catch rate correlations for the entire pre- ferred depth range (Table 4). When sablefish catch rates were high, grenadier and rockfish catch rates were low, and vice versa on the longline surveys. To illustrate an example of these negative correlations, we plotted the transformed catch rates in the East Yakutat and Southeast area (Fig. 2). The 90% density ellipses demonstrated that the trend in the data was negative in both the sablefish and grenadier and sablefish and rockfish scatter plots. Raw, untransformed catch rates also illustrated that the trend between the catch rates was negative. As an example, grenadier and rockfish catch rates were plot- ted against sablefish catch rates in the East Yakutat and Southeast area (Fig. 3). In both figures, sablefish catch rates were high when grenadier or rockfish were low and vice versa. When the catch rates of all the stations were av- eraged within a management area, the negative cor- relation was evident in the time series. For example, the time series of the average catch rates in the East Yakutat and Southeast area showed that grenadier and rockfish catch rates were higher than average when sablefish catch rates were lower than average and vice versa (Fig. 4). This trend was evident in all areas where there was a negative correlation. Rodgveller et al: Evidence of hook competition in longlme surveys 369 Figure 2 Scatterplots of grenadier (giant grenadier [Albatrossia pectoralis ]) versus sablefish (Anoplo- poma fimbria) catch rates and rockfish (shortraker \Sebastes borealis] and rougheye [ Sebastes aleutianus ] rockfish) versus sablefish catch rates from the National Oceanic and Atmospheric Administration Alaska Fisheries Science Center annual longline surveys, 1979-2003. Only data from the East Yakutat and Southeast management area are shown as an example. Each data point represents the average number of fish caught per 45 hooks (one skate of gear) at preferred depths at each station per year. Ninety percent density ellipses are included to demonstrate the trend of the relationship. Catch per unit of effort is abbreviated as CPUE. Data transformations are also abbreviated: square root (SqRt) and natural logarithm (Ln). Table 5 Correlation coefficients (coefficient) between giant grenadier ( Albatrossia pectoralis ) (grenadier) and sablefish ( Anoplopoma fim- bria) catch rates and between rockfish (shortraker [ Sebastes borealis] and rougheye [ Sebastes aleutianus] combined) and sable- fish catch rates in the Gulf of Alaska. Data were collected during the National Oceanic and Atmospheric Administration Alaska Fisheries Science Center Alaska longline and trawl surveys (1984, 1987, 1990, 1993, 1996, 1999, 2001, 2003, and 2005). The Pearson’s correlation coefficient of the relationship between sablefish and either grenadier or rockfish is listed. The coefficients for nonparametric tests (Spearman’s p and Kendall’s r) are listed for the rockfish and sablefish comparison in the trawl data because these data were not normally distributed. The P-value associated with the significance of the correlation tests are shown as well as the sample size in). The sampling gear is specified as either longline (LL) or trawl (Trawl). Grenadier-Sablefish Rockfish-Sablefish Gear Test Coefficient P n Coefficient P n LL Pearson’s -0.45 <0.0001 327 -0.24 <0.0001 355 Trawl Pearson’s 0.51 <0.0001 249 Trawl Spearman’s p -0.08 0.1070 419 Trawl Kendall’s t -0.05 0.1102 419 Longline versus trawl correlations When longline data were summarized in the same way as the trawl data, the correlations between sablefish and grenadier and sablefish and rockfish catch rates were significantly negative (Table 5), just as they were in the analyses that were stratified by sablefish management area. Scatter plots of the joint catch rates of sablefish and grenadier (Fig. 5A) and sable- fish and rockfish (Fig. 5B) showed that when catch rates were low for grenadier and rockfish, they were high for sablefish, and vice versa. The 90% density ellipses illustrated the negative trend in the joint catch rates. Correlation coefficients between sablefish and grena- dier catch rates from the Gulf of Alaska trawl surveys were significantly positive and the correlations between sablefish and rockfish catch rates were not significantly different from zero. Because sablefish and rockfish catch rates within the rockfish preferred depth range could not be transformed to fit a normal distribution, nonparametric methods and tests of significances were 370 Fishery Bulletin 106(4) used. Both the Spearman’s p and Kendall’s r had the same result of no significance (Table 5). Correlations between sablefish and grenadier or rock- fish catch rates in trawl gear were either positive or not different from zero. A scatter plot of sablefish and gren- adier catch rates showed that there were very few hauls when no sablefish or no grenadier were caught (Fig. 5C). It also demonstrated that when sablefish catch rates were relatively high, grenadier catch rates were high as well. The positive trend in the data is illustrated by the 90% density ellipse. However, the scatter plot of sablefish and rockfish catch rates (Fig. 5D) showed that there were three distinct groups of data: one where there were very few sablefish and many rockfish, one where there were few rockfish and many sablefish, and one where there were high catch rates of both species. The 90% density ellipse showed that the overall trend was neutral. The data for the sable- fish and rockfish comparison in trawl gear did not conform to a normal distribution after the data were transformed, although the transformation was retained for ease in viewing the data. Lengths of sablefish, grenadier, shortraker, and rougheye rockfish were similar in trawl and longline gear. Because large fish were caught by both types of gear, we can assume that both gears are catching adult cohorts of fish. Neither gear appears to be selecting only small juveniles (Fig. 6). Discussion Negative correlations on the longline surveys indi- cated that there was likely competition for hooks between sablefish and grenadier and also possibly between sablefish and rockfish. The catch rate cor- relations on longline gear between these species were negative in all six of the management areas, with the exception of the correlations for sablefish and rockfish in the Aleutian Islands. The compari- son of longline and trawl catch rates in the Gulf of Alaska demonstrated that, in the same areas, depths, and habitats, the relationship between sablefish and grenadier and sablefish and rockfish catch rates on longline gear were negative, but positive for sablefish and grenadier, and neutral for sablefish and rockfish caught in trawl gear. The lack of negative correlations for trawl gear and the presence of negative correlations for longline gear indicated that longline gear was not catching these species in proportion to their abundance. The trawl comparison is important because trawl gear is thought to catch fish in proportion to their abundance, even if the gear may be size selective (Gunderson, 1993). Therefore, hook competition is likely the cause of the negative correlations between sablefish and grenadier catch rates and is also likely one of the factors causing negative correlations between sablefish and rockfish catch rates. If the cause of the negative correlations on longline gear was due to differing habitat prefer- ences or direct competition, the trawl correlations would also have been negative. In the trawl data, there were very few hauls when zero sablefish or grenadier were caught, demonstrating that sable- 40 n A 35 - 30 - 3 25 O ,# ♦ ,*♦ **% ♦♦ ♦ 1 1 , * |t •• ♦ **5*.\r . 10 15 20 25 Sablefish CPUE 30 35 40 Figure 3 Untransformed catch rates, in number of fish caught per 45 hooks (one skate of gear), for (A) grenadier (giant grenadier [Albatrossia pectoralis ]) versus sablefish (. Anoplopoma fimbria) and (B) rockfish (shortraker [Sebastes borealis ] and rough- eye [ Sebastes aleutianus ] rockfish) versus sablefish caught during the National Oceanic and Atmospheric Administration Alaska Fisheries Science Center annual longline surveys from 1979-2003. Catch per unit of effort is abbreviated as CPUE. Because multiple data points coincide, a random number was added to each point so that they could be spread slightly to illustrate data point frequency. Only catches from the East Yakutat and Southeast management area from 2003 are shown as an example. Only preferred depths were included for grena- dier (550-1000 m) and rockfish (300-600 m). Rodgveller et al: Evidence of hook competition in longline surveys 371 -70-* 1978 1982 1986 1990 1994 1998 2002 Figure 4 Percent deviation from the average catch rate of the time series for grenadier (giant grenadier [ Albatrossia pectoralis]), rockfish (shortraker [Sebastes borealis ] and rougheye [Sebastes aleutianus] rockfish), and sablefish (Anoplopoma fimbria) caught during the National Oceanic and Atmospheric Admin- istration Alaska Fisheries Science Center annual sablefish longline surveys from 1979 to 2003. (A) includes only the grenadier preferred depths and (B) includes only the preferred rockfish depths. Data from the East Yakutat and Southeast management area are presented as an example. The lines that intersect the y-axis at 0 indicate the average catch rate; anything below this line is lower than average and anything above is higher than average. fish and grenadier are found in the same habitats consistently. Additionally, the correlation between catch rates of these species in the trawl gear was positive, also indicating they both use the same habitats. Because catch rates from multiple fixed stations, where the same habitats are sampled each year in several geographic areas for 25 years, consistently show negative correlations, it is un- likely that the negative relationship is due to dif- fering habitat preferences. For example, the time series of catch rates at all stations in the East Yakutat and Southeast area shows that grenadier and rockfish densities are above average when sablefish densities are below average and vice versa (Fig. 4). This result cannot be explained by differing habitat preferences because the same habitats were sampled each year. Sablefish and rockfish catch rate correlations in the trawl gear showed that there may be three distinct habitat types at rockfish preferred depths, and that negative correlations on the longline sur- veys may be partially influenced by habitat pref- erences of sablefish and rockfish at these depths. Hov/ever, although there appeared to be some dif- ferent habitat preferences, relatively large num- bers of both species were caught in most hauls. Also, nonparametric correlations of trawl catch rates were not significantly different from zero, indicating that any differences in habitat prefer- ences were not great enough to cause negative cor- relations for the longline gear. Additionally, just as with sablefish and grenadier catch rate trends for longline gear, the rockfish catch rates were above average in years when sablefish were be- low average, and vice versa, even when the same habitats were sampled each year. It is also unlikely that sablefish are directly competing with grenadier and rockfish, caus- ing negative correlations. If this were true there would be negative correlations for the trawl gear as well as the longline gear. Also, if sablefish were pushing grenadier or rockfish out of their preferred depth range, grenadier and rockfish catch rates would increase at other depths when sable- fish catch rates increased; however, this has not been observed. Or, if their population numbers were actu- ally being depressed because of competition, it would likely take longer time periods for adult populations of long-lived fish species like grenadier and rockfish to rebound. Although selective differences between longline and trawl gear for species and size have been reported for the Atlantic Ocean (e.g., Hovgard and Riget, 1992; Huse et ah, 2000), both gear types caught adult cohorts in these studies. Although gear selectivity of longline and trawl gear in Alaska has not been compared in field studies, all four of the species analyzed in this article were caught in longline and trawl gear in significant numbers. Very few small fish are caught, indicating that both gears catch adult fish (Fig. 6). Moreover, this finding indicates that the same cohorts are being se- lected by both gear types. Therefore, correlation coef- ficients between these species in these gear types can be compared fairly. Some environmental and fish specific variables, such as feeding history (e.g., Lpkkeborg et al., 1995; Stoner and Strum, 2004) may affect the strength of the effect of competition for hooks; however, we could not address these variables. It is possible, however, that at warmer water temperatures fish would have greater metabolic demands and increased hunger, causing more intense competition if prey availability was similar in each scenario (i.e., at colder and warmer temperatures). It is likely that the ability of fish to locate bait and their hunger affects catch rates. To assess the effects of these variables on competition, more laboratory studies of fish behavior would be needed. 372 Fishery Bulletin 106(4) Figure 5 Scatterplots of catch rates from the National Oceanic and Atmospheric Administration Alaska Fisheries Science Center Gulf of Alaska longline and trawl surveys (1984, 1987, 1990, 1993, 1996, 1999, 2001, 2003, and 2005). (A) Plot of grenadier (giant grenadier [Albatrossia pectoralis ]) and sablefish ( Anoplopoma fimbria) catch rates on longline gear; (B) rockfish (shortraker [ Sebastes borealis] and rougheye [Sebastes aleutianus] rockfish) and sablefish catch rates on longline gear; (C) catch rates of grenadier and sablefish from the trawl surveys; (D) rockfish and sablefish catch rates from the trawl surveys. Ninety percent density ellipses are included to demonstrate the trend. Catch per unit of effort is abbreviated as CPUE. Data transformations are also abbreviated: square root (SQRT), fourth root (4th rt), and natural logarithm (LN). Effects of competition on grenadier and rockfish catch rates will vary by area because of spatial abundance dif- ferences of these species and other groundfish. Negative correlations between sablefish and grenadier catch rates on longline gear varied with the density of sablefish, and the strongest negative relationships were found in the West Yakutat area and East Yakutat and Southeast area where sablefish catch rates and abundances were highest (Hanselman et al., 2006). Negative correlations between sablefish and rockfish catch rates were weaker than the correlations between sablefish and grenadier catch rates. Correlations between sablefish and rockfish catch rates were likely weaker because there were a variety of species caught in relatively large numbers in rockfish preferred depths, such as Pacific halibut (Hippoglossus stenolepsis), arrowtooth flounder (Ather- esthes stomias), and Pacific cod ( Gadus macrocephalus) . In areas where there were many other species caught at preferred rockfish depths (Aleutian Islands, West- ern Gulf, and Central Gulf areas), the relationship between sablefish and rockfish catch rates was weaker than in other areas. Hence, sablefish and rockfish may compete with multiple species for hooks at rockfish pre- ferred depths and dampen the direct competition with each other. The negative correlations at specific depths where grenadier or rockfish catch rates were highest were even greater than the negative correlations of the entire preferred depth ranges. This finding also shows that hook competition will vary depending on the depth, area, and possibly the abundance of the species of inter- est as well as other species. As our data and data from other studies indicate, aggressive predators may out-compete other species for hooks on longline gear (e.g., Skud, 1978; Zenger and Si- gler, 1992). For example, the proportion of Pacific hali- but caught increased as hook spacing increased because when hooks were widely spaced and less abundant, Pacific halibut out-competed other species for baited hooks (Skud 1978). This experiment also indicates that halibut are caught in proportion to their abundance, while other less aggressive species may not be. Simi- larly, Zenger and Sigler (1992) and Sigler and Zenger (1994) reported that catches of shortspine thornyhead ( Sebastolobus alascanus) and grenadier were low in Rodgveiier et a!: Evidence of hook competition in longlme surveys 373 areas where sablefish catches were high, and vice versa, during the AFSC sablefish longline surveys. They concluded that because sablefish are more mobile and aggressive, they are likely out-com- peting other species for baited hooks. If competi- tion between less aggressive groundfish and more energetic predators like sablefish is occurring on the AFSC longline surveys, as our data indicate, catch rate trends may not be proportional to ac- tual trends in fish densities. Directed sablefish experiments have shown that the decrease in probability of catching a sable- fish as the number of baited hooks decreases is not linear. However, in existing models where longline catch rates are adjusted for competition, this decrease is assumed to be linear (Murphy, 1960; Rothschild, 1967). In these models, when the number of recovered baited hooks is high, the magnitude of competition is low and model adjust- ments to catch rates are minimal. Conversely, as fewer baited hooks remain, competition increases, resulting in a greater need for model adjustments. Sigler (2000) documented the time until capture of sablefish on a hook-by-hook basis on longline gear and observed a nonlinear relationship between hooking probability and the number of baited hooks. The probability of hooking a fish remained constant until approximately half of the baited hooks were left; it then decreased steeply toward zero. This indicates that hook competition does not affect sablefish catch rates until fewer than 50% of the baited hooks remain, and that the decrease in hooking probability with a decrease in baited hooks is not linear. In another ex- periment, Sigler (2000) examined 12- to 42-m hook spacings that represented a condition of only 17% and 5% baited hooks remaining, respectively, and found that sablefish catch per hook decreased only 8%. The hook- spacing experiment showed that there is little decrease in catch probability, indicating that competition affects sablefish catch rates very little and also indicates that the decrease in catch probability does not decrease lin- early to zero as the number of baited hooks decreases; it may decrease very slowly until very few baited hooks remain and then drop off quickly. If the effect of baited hooks on catch probability is not linear, as Sigler (2000) showed, results from the Murphy (1960) and Rothschild (1967) competition models used to adjust catch rates would be biased. Assuming a linear relationship, when it is in fact nonlinear, will underestimate relative abun- dance when fish densities are high, and will overesti- mate relative abundance when fish densities are low. Competition for hooks likely occurs during the AFSC longline surveys. Both Skud (1978) and Sigler (2000) found that groundfish catch rates can be affected by hook competition; therefore it is likely that the catch rates of more docile groundfish, such as grenadier and rock fish, would also be affected. Currently gear satu- ration effects are not taken into account when relative population sizes of groundfish are calculated. Hook timing studies, such as Sigler’s (2000), have not been 90 80 70 I 60 £ 50 o> c £ 40 ■ 15 ,0 30 f— 20 10 0 ♦ Longline □ T rawl Sablefish Grenadier Shortraker Rougheye Figure 6 Average total length (cm) and associated 95% confidence intervals for sablefish (Anoplopoma fimbria), giant grena- dier ( Albatrossia pectoralis ), shortraker rockfish ( Sebastes borealis), and rougheye rockfish ( Sebastes aleutianus) caught during the National Oceanic and Atmospheric Administration Alaska Fisheries Science Center longline and trawl surveys in 2003. directed at other species caught during the longline surveys. To accurately assess competition effects on grenadier and rockfish, directed experiments on hook spacing or with hook timers would be needed to develop alternate models of the relationship between catch prob- ability and number of baited hooks. Acknowledgments We thank M. Sigler, J. Heifetz, P. Rigby, and D. Han- selman from the National Oceanic and Atmospheric Administration, Alaska Fisheries Science Center for their helpful reviews with earlier versions of this manu- script. We also thank the anonymous reviewers for their insightful comments, which greatly improved this manuscript. Literature cited Britt, L. L., and M. H. Martin. 2001. Data report: 1999 Gulf of Alaska bottom trawl survey. U.S. Dep. Commer., NOAA Tech. Memo. NMFS- AFSC-121, 249 p. Clark, W. G„ and S. R. Hare. 2006. Assessment and management of Pacific halibut: data, methods, and policy. Int. Pac. Halibut Comm. Sci. Rep. 83, 104 p. Clausen, D. M. 2006a. Grenadiers in the Gulf of Alaska, Bering Sea, and the Aleutian Islands, appendix F. In Stock assessment and fishery evaluation report for the groundfish fisher- ies of the Gulf of Alaska, p. 563-600. North Pacific Fishery Management Council, 605 W 4th Avenue, Suite 306, Anchorage, AK 99501. 374 Fishery Bulletin 106(4) 2006b. Gulf of Alaska shortraker rockfish and other slope rockfish. In Stock assessment and fishery evaluation report for the groundfish resources of the Gulf of Alaska, p. 685-725. North Pacific Fishery Management Council, 605 W 4th Avenue, Suite 306, Anchorage, AK 99501. Clausen, D. M., and J. T. Fujioka. 2005. Variability in trawl survey catches of Pacific ocean perch, shortraker rockfish, and rougheye rockfish in the Gulf of Alaska. In Biology, assessment, and manage- ment of North Pacific rockfishes (J. Heifetz, J Dicosimo, A. J. Gharrett, M. S. Love, V. M. O’Connell, and R. D. Stanley, eds.), p. 411-428. Alaska Sea Grant, Univ. Alaska Fairbanks, Fairbanks, AK. Cook, M. 2007. Population dynamics, structure and per-recruit analyses of yellowedge grouper, Epinephelus flavolim- batus, from the northern Gulf of Mexico. Ph. D. diss., 172 p. Univ. Southern Mississippi, Hattiesburg, MS. Gunderson, D. R. 1993. Surveys of fisheries resources, 248 p. John Wiley and Sons, Inc. New York, NY. Hanselman, D. H., C. R. Lunsford, J. T. Fujioka, and C. J. Rodgveller. 2006. Alaska sablefish assessment for 2007. In Stock assessment and fisheryevaluation report for the ground- fish fisheries of the Gulf of Alaska, p. 195-312. North Pacific Fishery Management Council, 605 W 4th Avenue, Suite 306, Anchorage, AK 99501. Hovgard, H., and F. F. Riget. 1992. Comparison of longline and trawl selectivity in cod surveys off west Greenland. Fish. Res. 13:323-333. Huse, I., S. Lokkeborg, and A. V. Soldal. 2000. Relative selectivity and effects of fishing strategy in trawl, longline and gillnet fisheries. ICES J. Mar. Sci. 57:1271-1282. Kohler, N. E., J. G. Casey, and P. A. Turner. 1998. NMFS cooperative shark tagging program, 1962- 93: an atlas of shark tag and recapture data. Mar. Fish. Rev. 60:1-87 Lokkeborg, S., B. L. Olla, W. H. Pearson, and M. W. Davis. 1995. Behavioral responses of sablefish, Anoplopoma fimbria , to bait odor. J. Fish Biol. 46:142-155. Menezes, G. M., M. F. Sigler, H. M. Silva, and M. R. Pinho. 2006. Structure and zonation of demersal fish assem- blages off the Azores Archipelago (mid-Atlantic). Mar. Ecol. Prog. Ser. 324:241-260. Murua, H., and E. Cardenas. 2005. Depth-distribution of deepwater species in Flemish Pass. J. Northwest Atl. Fish. Sci. 37:1-12. Murphy, G. I. 1960. Estimating abundance from longline catches. J. Fish. Res. Board Can. 17:33—40. Musick, J. A., S. Branstetter, and J. A. Colvocoresses. 1993. Trends in shark abundance from 1974 to 1991 for the Chesapeake Bight region of the U.S. Mid-Atlantic coast. U.S. Dep. Commer., NOAA Tech. Rep. NMFS 115, 18 p. Rothschild, B. J. 1967. Competition for gear in a multiple-species fishery. J. Cons. Perm. Int. Explor. Mer 31:102-110. Shotwell, S. K., D. H. Hanselman, and D. M. Clausen. 2006. Gulf of Alaska rougheye rockfish. In Stock assess- ment and fishery evaluation report for the groundfish fisheries of the Gulf of Alaska, p. 675-734. North Pacific Fishery Management Council, 605 W 4th Avenue, Suite 306, Anchorage, AK 99501. Sigler, M. F., and H. H. Zenger. 1994. Relative abundance of Gulf of Alaska sablefish and other groundfish based on the domestic longline survey, 1989. U.S. Dep. Commer., NOAA Tech. Memo. NMFS-AFSC-40, 79 p. Sigler, M. F. 2000. Abundance estimation and capture of sablefish (Anoplopoma fimbria ) by longline gear. Can. J. Fish. Aquat. Sci. 57:1270-1283. Simpfendorfer, C. A., R. E. Hueter, U. Bergman, and S. M. H. Connett. 2002. Results of a fishery-independent survey for pelagic sharks in the western North Atlantic, 1977-1994. Fish. Res. 55:175-192. Skud, B. E. 1978. Factors affecting longline catch and effort: III. Bait loss and competition. Int. Pac. Halibut Comm. Sci. Rep. No. 64:25-42. Sogard, S. M., and B. L. Olla. 1998a. Contrasting behavioral responses to cold temp- eratures by two marine fish species during their pe- lagic juvenile interval. Environ. Biol. Fishes 53:405- 412. 1998b. Behavior of juvenile sablefish, Anoplopoma fim- bria (Pallas), in a thermal gradient: balancing food and temperature requirements. J. Exp. Mar. Biol. Ecol. 222:43-58. Stoner, A. W., and E. A. Sturm. 2004. Temperature and hunger mediate sablefish ( Anoplo- poma fimbria ) feeding motivation: implications for stock assessment. Can. J. Fish. Aquat. Sci. 61:238-246. Woll, A. K., J. Boje, R. Holst, and A. C. Gundersen. 2001. Catch rates and hook and bait selectivity in long- line fishery for Greenland halibut ( Reinhardtius hip- poglossoides, Walbaum) at East Greenland. Fish. Res. 51:237-246. Zenger, H. H., and M. F. Sigler. 1992. Relative abundance of Gulf of Alaska sablefish and other groundfish based on National Marine Fisheries Service longline surveys, 1988-90. U.S. Dep. Commer., NOAA Tech. Memo. NMFS-AFSC-216, 103 p. 375 Abstract — -We used bomb radiocarbon (14C) in this age validation study of Dover sole ( Microstomus pacificus). The otoliths of Dover sole, a com- mercially important fish in the North Pacific, are difficult to age and ages derived from the current break-and- burn method were not previously validated. The otoliths used in this study were chosen on the basis of esti- mated birth year and for the ease of interpreting growth zone patterns. Otolith cores, material representing years 0 through 3, were isolated and analyzed for 14C. Additionally, a small number of otoliths with difficult-to- interpret growth patterns were ana- lyzed for 14C to help determine age interpretation. The measured Dover sole 14C values in easier-to-inter- pret otoliths were compared with a 14C reference chronology for Pacific halibut (. Hippoglossus stenolepis) in the North Pacific. We used an objec- tive statistical analysis where sums of squared residuals between otolith 14C values of Dover sole and the refer- ence chronology were examined. Our statistical analysis also included a procedure where the Dover sole 14C values were standardized to the ref- erence chronology. These procedures allowed an evaluation of aging error. The 14C results indicated that the Dover sole age estimates from the easier-to-interpret otoliths with the break-and-burn method are accurate. This study validated Dover sole ages from 8 to 47 years. Manuscript submitted 24 March 2008. Manuscript accepted 13 June 2008. Fish. Bull. 108:375-385 (2008). The views and opinions expressed or implied in this article are those of the author and do not necessarily reflect the position of the National Marine Fisheries Service, NOAA. Age validation of Dover sole ( Microstomus pacificus ) by means of bomb radiocarbon Craig R. Kastelie (contact author) Defsa M. Ancierl Daniel K. Kimura Chris G. Johnston Email address for C. R. Kastelie: Craig.Kastelle@noaa.gov National Oceanic and Atmospheric Administration National Marine Fisheries Service Alaska Fisheries Science Center 7600 Sand Point Way NE. Seattle, Washington 98115-6349 The otoliths of Dover sole ( Micros- tomus pacificus), a commercially important fish in the North Pacific, are difficult to interpret. Ages derived with the current otolith break-and- burn method have not been previously validated. The age data are important for population modeling and setting the total allowable catch (Stockhau- sen et ah, 2005). The necessity of age validation studies is widely recog- nized (Beamish and McFarlane, 1983; Campana, 2001) and age validation has become the focus in an expand- ing number of studies at the Alaska Fisheries Science Center (AFSC) (e.g., Kastelie and Kimura, 2006; Kimura et al., 2006; Hutchinson et ah, 2007; Kastelie et al., 2008). Two methods of Dover sole otolith preparation are used by various agen- cies. At the AFSC; the Groundfish Program of the Pacific Biological Sta- tion, Nanaimo, B.C.; and the North- west Fisheries Science Center, Dover sole are aged by the break-and-burn method (Chilton and Beamish, 1982) and are estimated to have a maxi- mum age of 54 years. An alternative aging method employed at the South- west Fisheries Science Center uses transverse thin sections of otoliths (Hunter et ah, 1990). The maximum age estimated with thin sections is 58 years (Hunter et ah, 1990). The two methods appear to produce simi- lar results but any similarity has not been tested quantitatively, and nei- ther method has been validated. Workshops on the interlaboratory calibration of methods and on otolith interpretation for determining the age of Dover sole have been held periodi- cally among agencies responsible for the management of this species. How- ever, age validation has not been a focus of these workshops. Because the otoliths of this fish species are small and the species has a relatively long life expectancy, the precision of mul- tiple readings on a sample of otoliths can be poor. The precision measured by percentage coefficient of variation (CV) between two age readers is re- ported to be 9.64% which is higher than that for most species aged at the AFSC (Kimura and Anderl, 2005). For many North Pacific Pleuronecti- formes the CV is under 4% (Kimura and Anderl, 2005). The poor precision (high CV) is an indication of the dif- ficulty in reading otoliths from Dover sole, and indicates that validating the accuracy of the ages is necessary. The goal of this study was to use bomb-produced radiocarbon (14C) to validate the ages of Gulf of Alaska (GOA) Dover sole determined by the otolith break-and-burn method. Otoliths were selected on the basis of estimated birth year and two de- scriptive categories: 1) otoliths with uniform growth zones that were easy to enumerate, and 2) otoliths with growth zones that were difficult to enumerate. The first category was used to validate general aging crite- ria and the second was used to help 376 Fishery Bulletin 106(4) determine otolith interpretation. Otolith material corre- sponding to a time near birth was isolated by extracting the core up to the first 3 years and measured for its 14C content. We analyzed the AUC from the first category of otolith cores using statistical methods first reported by Kastelle et al. (2008). Bomb radiocarbon age validation has been used on an increasing number of species and is considered one of the best methods to confirm the accuracy of fish ages (Campana, 2001). Recent uses in the North Pacific in- clude that on the white shark ( Carcharodon carcharias ) (Kerr et al., 2006), quillback rockfish ( Sebastes malin- ger) (Kerr et al., 2005), canary rockfish (S. pinniger) (Piner et al., 2005; Andrews et al., 2007), bocaccio rock- fish (S. paucispinis) (Andrews et al., 2005; Piner et al., 2006), Pacific halibut ( Hippoglossus stenolepis) (Piner and Wischniowski, 2004), and Pacific ocean perch (S. alutus) (Kastelle et al., 2008). Radiocarbon fish age validation relies on a time refer- ence provided by production of 14C from atomic bomb testing. The above-ground testing of atomic bombs that introduced 14C into the atmosphere and marine environment began in the 1950s and continued into the 1960s (Kalish, 1993; Nydal, 1993). This caused a rapid increase in marine 14C lasting through about 1970 — an increase that is recorded in calcified marine organisms and otoliths and provides a necessary time reference. To validate ages from a “validation species” (in this case Dover sole), a 14C “reference chronology” is used, where the exact time frame of the 14C increase is con- sidered known. Two reference chronologies have been developed for the North Pacific Ocean: one from Pacific halibut (Piner and Wischniowski, 2004) and one from yelloweye rockfish (S. ruberrimus) (Kerr et al., 2004). The posited birth years for the validation species are calculated from ages estimated by otolith growth zone counts and date of collection. Specimens representing the validation species are chosen such that the range of posited birth years spans the period of rapid marine 14C increase. Otolith core material deposited in the first one or two years of life from the validation spe- cies is analyzed and each core provides one 14C data point. To evaluate the ages, the 14C from the cores of the validation species is plotted with respect to the posited birth years and compared to the known 14C values in the reference chronology. If there is a timing difference between the 14C increase in the validation species and the reference chronology, then the esti- mated ages of the validation species are often assumed to be in error. Alternatively, if a timing difference is not present, the ages from the validation species are considered accurate. In a recent bomb radiocarbon age validation study of Pacific ocean perch, a series of new procedures was used to compare the 14C measurements in the validation samples to the reference chronology (Kastelle et al., 2008). We used the same methods here — purposely biasing the ages to be validated by ±0, 1, 2, and 4 years; standardizing the validation sample 14C values to the reference chronology; and evaluating the residuals between the validation samples and the reference chronology to see if inaccuracies in the age estimates were present. There are two important assumptions when validat- ing fish ages with the bomb radiocarbon method (Cam- pana and Jones, 1998; Piner and Wischniowski, 2004; Piner et al., 2005; Kastelle et al., 2008). Assumption 1 is that the validation species must be biologically and environmentally similar to the species in the reference chronology during the first years of life. If both species are receiving their 14C from the same sources, then the magnitude and timing of the 14C increase should be similar (Andrews et al., 2007). A reference chronology based on the same species as that being investigated is best, and occasionally available (Campana, 1997; Cam- pana et al., 2002; Piner and Wischniowski, 2004). As- sumption 2 requires that the otolith core used for each 14C analysis be uncontaminated and that it constitute a closed system. Therefore, an accurate extraction of the core without contamination from other carbon sources or different years is necessary. Dover sole otoliths pre- sented a unique challenge in this regard because of their small size. For further information regarding radiocarbon age validation studies, one can consult the earlier mentioned studies from the North Pacific along with Kalish (1993, 1995) and Campana (1997). Materials and methods Otolith selection and coring procedures The Gulf of Alaska (GOA) Dover sole otoliths used in this study were collected either during AFSC survey cruises or by AFSC fishery observers aboard commercial vessels. The survey cruises took place in 1984 and 2005; the otoliths were removed from the fish at sea, stored in a glycerin and thymol mixture, and archived for future age determination. The specimens collected from commercial harvests were caught in 1998 and treated similarly, except they were first stored dry for up to 3 months before storage in a glycerin and thymol mixture. The glycerin and thymol mixture is not expected to be a contaminant in otolith 14C measurements (Campana et al., 2003). After the archival period, the otoliths were aged at the AFSC for stock assessments. The initial ages were determined by the break-and-burn method (Chilton and Beamish, 1982) with the blind-side otolith. Assumed annual growth zones were counted by enumerating the translucent zones. The otolith growth over the course of one year is assumed to consist of an opaque zone and a translucent zone. After growth zones were read, the otoliths were archived again for varying durations up to 14 years. Samples where the initial age estimate placed the birth year near the era of marine 14C increase were re-examined by age readers experienced in the interpre- tation Dover sole otolith growth zones, and considered for possible 14C measurement. In the re-examination process otoliths were re-aged to assign a “final age” and were placed into two subjective categories based on the ease of interpretation of the growth zones: Kastelle et at: Age validation of Microstomus pacificus by means of bomb radiocarbon 377 Figure 1 Category-1 otolith (specimen number 82) from Dover sole ( Microstomus pacificus) treated by the break-and- burn method. The series of dots indicate the translucent zones that were counted to estimate a final age of 32 years. The “T” arrow points to the transition region from fast early growth to slower later growth. The sulcus region of the otolith is indicated as a reference point. 1 Category-1 otoliths had clear growth zones that were easy to interpret, in that most of the translucent zones appeared with minimal or no splitting in the proximal growth axes in at least one region (dorsal or ventral) on the break-and-burn cross section (Fig. 1). Splitting is defined as the branching of a single translucent zone into two or more translucent zones. Typically the translucent zones were spaced evenly, but with decreasing intervals, as the fish became older. Some samples in this category may have pre- sented interpretative options, and different reading axes in the cross section could be chosen, but typi- cally only small differences in age (of 1 or 2 years) would result. 2 Category-2 otoliths had growth zones that were dif- ficult to interpret and that made these fish difficult to age. Many translucent zones had obvious splits and uneven spacing (Fig. 2). Widely different ages could be generated depending upon the interpretation chosen and which reading axis was used. A sample selection process for 14C measurement occurred after re-examination of the ages, and this selection process relied on two factors. First, in category 1 our intention was that the initial age and final age should agree within 3 years. For otoliths in category 2, which are more common for Dover sole than are category- 1 otoliths, we did not use age agreement as a selection factor. Second, the estimated birth years, which were based on the final ages, had to be evenly distributed from about 1951 to 1977, to bracket the era of marine 14C increase. For several specimens in category 1, the first factor was relaxed to evenly populate the years of marine 14C increase. This process provided 43 speci- mens for Z\14C analysis: 38 otoliths in category 1 and 5 in category 2 (Table 1). The range in catch years from 1984 to 2005 generated a large range in Dover sole ages for potential validation. In the remainder of this article, unless specified differently, the ages referred to are these final ages. For the selected Dover sole specimens, the otolith core was extracted from the remaining whole eyed-side otolith for 14C analysis. The core represented material deposited only in the first three years of life. Previous studies have often used a 1-year core (e.g., Campana, 1997), but that was not possible with Dover sole because their otoliths are very small, therefore, the minimum mass required for 14C analysis mandated a 3-year core. For the otolith coring procedure we used a Buehler® EcoMet® (Buehler Ltd., Lake Bluff, IL) grinder with 320 grit wet or dry sandpaper to first remove otolith material on the proximal surface (the main growth axis in older otoliths). On some otoliths a small amount of material was also removed on the distal surface. Next, the grinder was used to remove material on the perim- eter, in the anterior-posterior and dorsal-ventral axes, beyond the third year. After the exterior layers were removed with this procedure, the location of the third year’s growth zone in each otolith became easier to see, and its location served as a primary guide in the coring 378 Fishery Bulletin 106(4) Figure 2 Category-2 otolith (specimen 225) from Dover sole ( Microstomus pacificus) treated by the break-and-burn method. The “T” arrow points to the transition region from fast early growth to slower later growth. In the enlarged section, the two series of dots represent different options of interpreting the translucent zones; the final age of 20 years and the maximum age of 32 years are shown. The “S” points to a translucent zone that splits to form two translucent zones. The sulcus region of the otolith is indicated as a reference point. process. Each core was tailored to the size and shape of the apparent third year. A secondary guide for the coring process was the weight and size of otoliths from 3-year-old fish in the 2005 survey collection, and an additional survey collection (not used for 14C measure- ments) from 2003. These otoliths from 3-year-old fish had an average weight of 9.5 mg (±0.4 mg standard er- ror) and an average size of 2.41 x 3.84 x 0.68 mm (?i=21). The core would occasionally break into several pieces as material was removed, but all salvageable pieces were retrieved and used. The core size was recorded for all intact cores and compared to otoliths from young fish. Finally, the cores were cleaned in an ultrasonic cleaner, dried, weighed, and then stored in acid-washed vials before 14C analysis. 14C analysis The 14C and 13C of Dover sole otolith cores were mea- sured at the National Ocean Sciences Accelerator Mass Spectrometry Facility at the Woods Hole Oceanographic Institution, Woods Hole, MA. Samples were treated at the Woods Hole Oceanographic Institution with a routine acid hydrolysis procedure to produce a graphite target, and analyzed with accelerator mass spectrometry. We report results as 414C, which is defined in Stuiver and Polach (1977) as the relative difference between an inter- national standard (base year 1950) and sample activity. The zi14C is normalized to 1950 and corrected for isotopic fractionation with the d13C measurement and normalized to a d13CVPDB value of -25 %c. To evaluate the Dover sole ages, we compared the oto- lith 414C results from the otolith cores with the Pacific halibut reference chronology using several procedures. Initially, the 414C from the Dover sole otolith cores in both categories was plotted along with a loess (locally weighted least squares) smoothed curve of the 414C in Pacific halibut. All loess-smoothed curves in this study were fitted by using Splus (Insightful Corp., Seattle, WA) (Chambers and Hastie, 1992) with a span of 2/3 and a degree of 2. To further analyze the 414C results in comparison with the reference chronology, we used three procedures first introduced by Kastelle et al. (2008). The first procedure was to purposely bias the category-1 ages by 0, ±1, ±2, and ±4 years, generating seven sets of ages. For each of these seven sets of biased ages, pos- ited birth years were calculated and a sum of squared residuals ( SSR ) between the A14 C in Dover sole otoliths and the loess smoothed data in the Pacific halibut ref- erence 414C chronology was calculated. The smallest of the seven SSR s indicated which purposely biased set of ages represented the best fit, and thereby indicated if an overall aging error in the Dover sole ages existed. We call this procedure the “unstandardized” analysis. In the second procedure used to investigate the ac- curacy of Dover sole ages in category 1 we performed a “standardization” of the measured 414C values (Kas- telle et ah, 2008). This is a linear transformation of the 414C values in the Dover sole otolith cores which removes any difference in scale between the valida- tion sample 414C measurements and the reference chronology. It does not change the timing of the val- idation sample 414C values or their relative magni- tude. For this standardization, let {v ^1 be the series of j- 1, ..., n validation observations of 414C, where y\j] refers to the year core j was formed. We defined the standardized series for {v^} as {v^^j=(vy[jj + p)/a|, where p and o can be estimated by a least squares fit (i.e., by minimizing the SSR) to the loess-smoothed curve of the reference chronology data set {Z ^j}: Kastelle et al. : Age validation of Microstomas pacificus by means of bomb radiocarbon 379 TabSe 1 Age estimates (in years) and radiocarbon measurements from Dover sole (Microstomus pacificus) otoliths collected from the Gulf of Alaska. Category 1 represents easy-to-age specimens and category 2 represents difficult-to-age specimens. Estimated birth year was calculated from final age and known capture date. The final ages are re-evaluated ages undertaken in the current study after initial aging for stock assessments. Postmeasurement min. -max. ages were generated after the radiocarbon values were known. Results are reported as 414C (which is defined in Stuiver and Polach [1977] as the relative difference between an international standard [base year 1950] and sample activity), 414C 95% confidence intervals (Cl) were derived from accelerator mass spectrometry error, and the <513C measurements were used to correct for natural effects of isotopic fractionation. Specimen no. Category Estimated birth year Age estimates (years) Postmeasurement Final min. -max. h14C (%o) 414C 95% Cl <513C (%c) 3 1 1972 12 80.7 7.2 -1.34 11 1 1965 33 43.8 6.8 -1.86 17 1 1968 30 80.9 8.4 -1.99 18 1 1963 35 27.2 7.8 -1.6 20 1 1974 24 57.6 7.0 -1.5 21 1 1957 41 -86.3 6.2 -0.97 24 1 1958 47 -68.0 6.0 -1.65 34' 1 1953 45 -108.1 5.6 -0.97 36 1 1974 10 80.3 7.2 -1.77 48 1 1959 39 -64.9 6.6 -1.12 49 1 1969 29 81.5 7.4 -1.06 54 1 1957 27 -76.2 6.6 -0.98 66 1 1960 24 -65.2 8.0 -1.11 73 1 1964 20 49.2 10.2 -2.26 82 1 1966 32 68.0 7.2 -2.03 84 1 1977 21 36.2 8.0 -1.88 85 1 1966 32 86.4 8.2 -1.26 89 1 1959 39 -84.7 7.0 -1.14 102 1 1967 31 73.7 8.2 -1.12 130 1 1964 34 38.3 7.8 -0.76 138 1 1969 15 72.7 6.6 -1.4 143 1 1953 31 -108.2 5.6 -0.7 157i 1 1970 28 57.5 6.6 -0.95 159 1 1976 8 72.6 10.0 -1.88 160 1 1962 36 21.4 7.6 -2.07 210 1 1951 33 -109.5 5.4 -0.57 273 1 1963 21 9.5 8.6 -2.06 302 1 1975 9 64.9 9.8 -1.53 33 1 1959 39 -52.7 6.2 -1.92 34' 1 1963 35 41.5 9.0 -1.6 35 1 1970 28 54.3 6.4 -1.17 52 1 1969 29 83.6 8.4 -1.66 71 1 1955 43 -86.7 5.6 -0.71 157' 1 1974 10 62.2 6.8 -1.06 201 1 1955 29 -104.3 6.4 -1.32 257 1 1966 18 38.0 6.8 -1.34 271 1 1972 12 67.3 8.2 -1.41 310 1 1967 17 52.3 7.2 -2.78 95 2 1957 48 37-50 -120.1 5.8 -1.59 146 2 1958 26 16-26 75.7 10.2 -0.84 188 2 1953 52 37-59 -105.1 6.8 -0.92 225 2 1964 20 17-32 56.9 7.2 -1.33 317 2 1961 23 19-23 54.7 10.2 -1.41 1 Specimens with repeated identification numbers do not represent repeat measurements but are different specimens with the same identification numbers from different collection years. 380 Fishery Bulletin 106(4) ;-Z< SSR - > (vyu]~^u]) :X[((V^'] + ^,/C7)-/-V[7]]2- An iterative process to estimate p and o can be found in Kastelle et al. (2008). Again, to evalu- ate for aging error, we purposely biased the ages by 0, ±1, ±2, and ±4. For each of these seven sets of ages, we estimated p and o, standardized the validation sample AUC observations with {vy[j]l, and calculated the SSR. As before, the smallest SSR indicated the best fit to the reference chronol- ogy and indicated if any overall aging bias (error) existed. It should be noted that if p and a are not estimated, but instead are set to p = 0 and o= 1, this process becomes the unstandardized proce- dure described earlier. The third and final method we employed was estimation of confidence intervals around the loess-smoothed reference chronology (Kastelle et al., 2008). For the estimated confidence intervals, simultaneous Bonferonni statistical inference was used (Miller, 1966) that calculates simultaneous (a=0.01) confidence intervals whose width is dependent on the number of distinct years at which comparisons are made between the reference chronology and valida- tion samples, and on the variability in the reference chronology. As an aid in our comparison between the standardized Dover sole otolith 414C values and the Pacific halibut reference chronology, not only were the SSRs evaluated, the standardized 4UC values were also viewed graphically when plotted against the posited birth years. This comparison generated seven plots, one for each purposeful bias, and included the loess- smoothed Pacific halibut reference chronology with con- fidence intervals. The category-2 otoliths were re-examined again after the 414C results were known. The goal of this re-aging process was to use the z\14C results as a guide for re- fining current otolith aging criteria. To facilitate this process, the 414C in the category-2 otoliths was plotted with a loess-smoothed curve of category-1 results. This process allowed the age reader to learn how difficult- to-interpret otoliths are best aged. In this last re-aging process a minimum and maximum age were estimated for the category-2 otoliths. The 3-year core in Dover sole otoliths needed to be taken into consideration when analyzing the results. We assumed a mid-point of approximately 1.5 years for core deposition because the core represents material from the first 3 years of life. This means that we as- sumed linear otolith growth during the first three years of life. Linear otolith growth may not be accurate, but was assumed for simplicity and because any error from this assumption is trivial. The Pacific halibut reference chronology is based on material from only the first year of life; therefore, we assumed a mid-point of 0.5 years 1950 1960 1970 1980 Birth year Figure 3 414C %c in otolith cores plotted against birth year of Dover sole (Microstomus pacificus), by category 1 (•) and category 2 (A), with a 95% confidence interval shown for each point. Two category-! specimens are overlapping at -108 h14C and birth year 1953. The line ( ) is a loess-smoothed curve of the Pacific halibut ( Hippoglossus stenolepis) reference chronology (Piner and Wischniowski, 2004). for core deposition. These assumptions mandated an approximate allowance of a (1.5-0. 5 =) 1.0-year shift in the comparison of the Dover sole results with the Pacific halibut reference chronology. Results Otolith selection and coring procedures The selection process generated specimens for which final ages agreed with initial ages and was followed by successful coring. In all but three of the specimens in category 1, the agreement between the initial age and final age was within 3 years. In this category, the maximum final age was 47 years, and the minimum age was 8 years (Table 1). In category-2 specimens, four out of five had discrepancies of over 3 years between the initial age and final age. The percentage CV (Kimura and Anderl, 2005) between initial ages and final ages in both categories was 4.21%. The average core weight across all categories was 5.4 mg (±0.2 mg standard error) and the average size was 1.93x2.97x0.49 mm, which was smaller than the guide provided by the 3- year-olds described earlier. ,4C analysis The 414C in Dover sole otolith cores from category 1 followed the expected general pattern of initial low 14C before atmospheric testing followed by an increase syn- chronous with testing. It displayed a rise in about 1955 (from below -100 %c) and peaked in 1966 at over 85%o (Table 1, Fig. 3). This trend in Dover sole radiocarbon Kasteile et a!.: Age validation of Microstomas pacificus by means of bomb radiocarbon 381 Table 2 Sum of squared residuals ( SSR ) between Au C in category-1 Gulf of Alaska Dover sole ( Microstomus pacificus) otoliths and loess- smoothed curve of the Pacific halibut ( Hippoglossus stenolepis) reference chronology (Finer and Wischniowski, 2004). Results are tabled by unstandardized (No) or standardized (Yes) A14C values. The u and u are coefficients in the linear standardiza- tion defined for the series as {v | -j-i=(vyy^+ p)/ o) , where p and o were estimated by a least squares fit to the loess-smoothed curve of the reference chronology data set, or set to 0 and 1 respectively when unstandardized. When standardized, SSR was minimized with respect to p and a. Age bias was applied to each final age estimate such that -4 represents younger ages and +4 represents older ages. V ■(.; Standardized to reference Parameter -4 -2 Age bias (years) -1 0 +1 +2 +4 No Yes p o SSR p a SSR 0 1 40,429 28.47 1.10 15,544 0 1 13,179 9.59 0.94 7969 0 1 11,651 1.63 0.91 9533 0 1 20,140 -6.17 0.90 16,834 0 1 38,281 -13.97 0.92 29,191 0 1 66,696 -22.80 0.95 45,706 0 1 153,202 -45.09 1.11 84,612 Figure 4 Series of seven plots for A14C %o in otolith cores of Dover sole ( Microstomus pacificus) category-1 (#) where the A14C is standardized and each plot, (A) through (G), corresponds to a purposeful age bias of -4 to +4 years, respec- tively, with a loess-smoothed curve ( ) of the Pacific halibut ( Hippoglossus stenolepis) reference chronology (O) (Finer and Wischniowski, 2004) and 99% simultaneous confidence intervals around the mean smooth. values parallels those seen in the Pacific halibut reference chronology but the values appear to be shifted earlier in time by 1 or 2 years. Also, from 1969 on, the Dover sole values are mostly lower than those in the reference chronology (Fig. 3). In the analysis of the category- 1 ages, a purposeful bias of -1 or -2 years provided the lowest SSRs. In the unstandardized procedure, where the ages were purposely bi- ased -1 years, the SSR was small- est at 11,651 (Table 2). In the stan- dardized procedure when the ages were purposely biased -2 years, the SSR was smallest at 7969, where p-9.59 and a=0.94 (Table 2). For the standardized analysis, a se- ries of plots is presented, one plot for each of the seven sets of pur- posely biased ages (Fig. 4). Both the overall time shift and the low bias after 1969 were removed by the standardization procedure when the purposeful age bias was -2 years (Fig. 4B). Also, at this purposeful age bias and standardization, the Dover sole validation specimens were all within the 99% confidence intervals around the loess-smoothed reference chronology. The general difficulty in aging category-2 specimens is evident in the Ai4C results that are not synchronous with those from the category-1 specimens. The range of possible ages for category-2 specimens (Table 1) is shown by horizontal bars in Figure 5. Two of the three younger category-2 speci- mens were likely over-aged because only the right end of the horizontal bar is close to the values for the other specimens or the values of the loess-smoothed curve of category-1 specimens. Therefore, for these three speci- mens the choice of a younger age is more accurate. In 382 Fishery Bulletin 106(4) the two older category-2 specimens, the results are less definitive because the A14C values fall in the stable pre-bomb era. Figure 2 is an example of how a category-2 otolith may be aged given the different possible interpretive options. Discussion Age validation This study is the first published age validation for GOA Dover sole. Ages in the range of 8 to 47 years were validated. Accurate ages were indicated by both the unstandardized and standardized Ai4C results for category 1 when a purposeful age bias of -1 or -2 years produced the lowest SSR s. The shift of approximately 1 year (due to core size, as explained earlier) was expected, and when con- sidered with the purposeful age bias and lowest SSi?s, indicated that ages estimated by the break- and-burn method were accurate. All category- 1 specimens were within the 99% simultaneous confidence intervals on the reference chronology (Fig. 4B) and this result also provides strong evi- dence for accurate ages. The difference between the purposeful age bias of -1 and -2 was not resolved. The resolution of this bomb 14C age validation study was limited by the ap- proximate nature of the expected 1-year shift, variabil- ity in 14C due to geographic location, and variability in 14C measurement, but the general accuracy of the ages was validated. In using bomb Z\14C for age validation, assumption 1 is the largest contributor to any concerns about resolution due to variability in A14C because of depth or geographic area (Nydal, 1993; Kalish, 1995; Andrews et al., 2007). Our method of separating the specimens into two categories addressed specific objectives. Use of only the clearest specimens to validate the age estimates for category-1 specimens is common practice (Piner and Wischniowski, 2004; Kerr et al., 2005; Kastelle et al., 2008). In a long-lived species such as Dover sole, age determination is difficult (Chilton and Beamish, 1982; Kimura and Anderl, 2005) and subjective interpreta- tions of growth patterns (Fig. 2) must be made. When determining the age of a species on a routine basis, a set of species-specific interpretive rules or “aging cri- teria” are applied to all specimens, i.e., to specimens that are easy to interpret and to those that are dif- ficult to interpret (Kastelle et al., 2008). By choosing validation samples like category 1 where there is little ambiguity, the basic methods and aging criteria can be validated and then applied to all samples, includ- ing ones like category 2. If the specimens were chosen randomly, without consideration for the difficulty in estimating age or interpreting otoliths, the spread of validation sample points around the reference chronol- ogy would increase. This spread would provide less in- formative results regarding the basic aging criteria, as 150 100 50 9 0 -50 -100 -150 - 1950 1960 1970 1980 Birth year Figure 5 A14C %c in otolith cores of Dover sole ( Microstomus pacificus) plotted against birth year, by category 1 (•) and category 2 (A). Horizontal lines on category-2 specimens represent- ing the minimum and maximum postmeasurement age, and a loess smoothed curve ( ) has been fitted to category-1 data points. shown by the spread in the category-2 specimens. The types of specimens represented by category 2 are more common in Dover sole; hence this sample design was selected to provide the most informative results. The exact percentage represented by the two categories was not determined in this study. A further reason for us- ing different categories was to provide a tool to develop aging criteria for the hard-to-age specimens. Following the validation, the less common category-1 specimens were used to provide the loess fit in Figure 5 to which the category-2 specimens were compared. The A14C results displayed a gap in specimens at a birth year of 1961. Similar gaps have been seen in other studies (Kerr et al., 2004; Piner and Wischniowski, 2004; Piner et al., 2005; Kastelle et al., 2008) and are likely due to two reasons. First, in the early- to mid- 1960s, marine 14C was likely increasing so quickly that even if otoliths were accurately aged, a gap would likely be present because of the limited time range when these mid-range values of A14C existed. Second, in our sample selection process, otoliths that could be categorized unambiguously as category 1 and aged such that they represented the 1961 birth year were not present. The utility of A14C standardization is apparent in this study. Without the standardization, most of the valida- tion sample h14C points before about 1968 are above the reference chronology, or to its left, and below the refer- ence chronology after 1969. With the A14C standard- ization and the purposeful bias of -2 years applied to the category-1 ages, it is clear that the Dover sole 414C points and Pacific halibut reference chronology are in synchrony and that ages are accurate. As previously ex- plained, the shift to the left is due to the core size. The consistently low Dover sole h14C points after 1969 are Kastelle et at: Age validation of Microstomus pacificus by means of bomb radiocarbon 383 likely due to different environmental regimes or biologi- cal differences of the two species (Kalish, 1993, 1995; Nydal, 1993; Andrews et al., 2007). The standardization procedure is ideal for correcting this type of bias, where a difference in range of A14C exists. Previously, this procedure was used by Kastelle et al. (2008) to re-ana- lyze validation data for black drum (Pogonias cromis), originally presented by Campana and Jones (1998). For the re-analysis, the black drum A14C values were stan- dardized to a Northern Hemisphere atmospheric Au C reference chronology in a comparison where they were dramatically different in scale but similar in timing (Kastelle et al., 2008). If little difference in range ex- ists when the standardization is applied, the estimated values of p and a will be close to 0 and 1, respectively, provided the overall fit is good. This situation would indicate that the standardization had little effect and that the correct evaluation of any aging error will still be made by considering the SSRs. This was the case for Pacific ocean perch analyzed previously with this method (Kastelle et al., 2008). Therefore, we feel this standardization method can be applied generally. We chose the Pacific halibut reference chronology for several reasons. First, this reference chronology is based mostly on juvenile fish (Piner and Wischniowski, 2004). It also represents a wide geographic area in the GOA, similar to that for Dover sole. Finally, although the early life history of Dover sole in the GOA is not well understood, the pelagic larvae are found in the upper 30 m of the water column, and immature fish are known to concentrate in nearshore areas and shallow waters over the continental shelf (Abookire et al., 2001; Abookire and Bailey, 2007). Juvenile Pacific halibut are typically found in shallow nearshore areas (Norcross et al., 1995; Abookire et al., 2001); therefore comparisons with Dover sole for this age validation were reasonable. The otolith cores from the validation samples were smaller than the measured guide otoliths from 3-year- olds. Some of this difference may be explained by the presence of newly deposited opaque material beyond the third translucent zone in the measured 3-year-olds. In the cored otoliths, this same material was often ground away to expose the third translucent zone, thereby pro- ducing a size difference. Also, a few of the cores may have been incorrectly centered during the grinding pro- cess, and therefore may have incorporated a little mate- rial from beyond the third year. Conversely, too much material could have been removed, down to the second year’s growth zone. The latter is more likely the case as evidenced by the small core weights. On the proxi- mal side, the coring process may have inadvertently removed some material belonging to the third year in an effort to remove all material from later years in the region of the sulcus groove. We considered the average age of the material represented by the cores to be ap- proximately 1.5 years. However, if the cores were too small and some material inside the third translucent zone was removed, than the average age of the mate- rial may have been closer to 1 year, indicating less of a required shift. The difference was only 0.5 years; hence we considered any error from this consideration to be negligible. Less of a required shift was also indicated if the otoliths were accumulating more mass during their first year than in subsequent years (see Materials and methods where linear otolith growth is assumed). As mentioned previously, this age validation method can not resolve either type of error when potentially very small. However, it is probably not coincidental that the purposeful age bias of -1 year for the unstandardized A14C results was the best fit. Dover sole otoliths often display a transition in growth rate typically seen as a pattern of decreased spacing between presumed annual growth zones. This occurs when the fish is estimated to be about 6- to 8-years-old, with growth zones deposited prior to the transition rep- resenting younger and faster growth and post-transition zones representing slower growth (Figs. 1 and 2). An association between the transition timing and maturity has not been documented in Dover sole, but Abookire and Macewicz (2003) reported that 50% maturity occurs at 6.7 years which roughly coincides with the observed transition. They used specimens aged by the same ex- perienced age readers as in this study; hence some level of circularity exists. However, our studies’ results lead us to believe that the timing of the transition pattern is likely associated with the onset of maturation. In other species such as orange roughy ( Hoplostethus at- lanticus) a decrease in the annual growth zone width is documented to correspond to the onset of maturity (Francis and Horn, 1997). Considerations concerning category-2 results Category-2 A14C results confirm that Dover sole are often a difficult-to-age species. This difficulty was exemplified by the range in possible ages of even the youngest cat- egory-2 specimens and was especially evident in the two older category-2 specimens. The CV of 4.21% for these hand-picked specimens, where the majority (38 out of 43) of otoliths were deemed to be clear, although better than the typical CV of 9.64% for Dover sole, was high in comparison to that for many other flatfish species aged at the AFSC (Kimura and Anderl, 2005). The category- 2 otoliths are typical of many Dover sole samples aged at the AFSC where subjective decisions are made by necessity in the age reading process. Results from the five difficult-to-age specimens in category 2 did not indicate consistent over-aging or under-aging. Correct decisions in how to interpret the growth zone patterns were made for specimens 225 and 188, and a reasonably good choice was made in specimen 95 especially when its high age was consid- ered. It is clear that the choice of a mid-range or older age was the most accurate for specimens 95 and 188 when compared to the loess-smoothed curve of cat- egory-1 specimens. However, specimens 146 and 317 were demonstrated to be over-aged in the comparison to category-1 specimens. The underlying difficulty in aging and interpreting Dover sole otoliths lies in the framework of splits that can make up a single trans- 384 Fishery Bulletin 106(4) lucent zone (Fig. 2). In specimens from category 1, the translucent zones were usually compact and well defined; hence any apparent splits could easily be in- terpreted and decisions could be made as to how they should be enumerated. In category-2 specimens, the potential annual zones were not well defined, often due to splits that created interpretative options. Splits in a potential annual zone that occur before the transition to slow growth are especially problematic. In hindsight, it was reasonable to conclude that the two specimens (numbers 146 and 317) that diverged from the loess- smoothed category-1 data were over-aged probably be- cause of broad splits. The specimen with the largest discrepancy, number 146, was one where the pattern in the broken-and-burnt cross section could be aged at 16 years (more in line with the loess-smoothed category-1 data), or on a second reading axis could be interpreted as 26 years. Similarly, in Figure 2 an age of 32 years was estimated along an axis closer to the sulcus, but a more correct age estimate of 20 years was chosen from an adjacent reading axis. Some of this discrep- ancy could be the result of splitting translucent zones. In reality, the five specimens in category 2 are not enough to draw firm conclusions on how these difficult otoliths should be aged. A further study of category-2 type specimens may help to refine the aging criteria for Dover sole otoliths. The region in the otolith cross section before the tran- sition to slow growth can be an area of splitting or dif- fuse translucent zones. This is a situation that can lead to over-aging; therefore the reader must exercise care to count only prominent translucent zones. Splitting and diffuse translucent zones are a frequent problem in age reading many species (e.g., Francis and Horn, 1997; Gregg et ah, 2006; Hutchinson et al., 2007). The fish in category 1 were correctly aged by counting only the prominent translucent zones preceding the transi- tion to slow growth. Conclusions The ages estimated for the GOA Dover sole were vali- dated as accurate based on the easy-to-age otoliths in category 1. When the age bias of -1 or -2 years was applied, the A14C in the validation samples had the same timing as the z\14C in the Pacific halibut reference chro- nology and was consistent with the expected 1-year core- size shift. An age-structured stock assessment model is used for management of the GOA Dover sole com- mercial fisheries and hence the age data validated here are important for population modeling and setting the total allowable catch (Stockhausen et ah, 2005). In the future, analysis of additional difficult-to-age category-2 otoliths may help to further answer questions regard- ing aging criteria for these specimens. In reality, the lower-than-average between-reader precision for growth zone counts will likely persist, but now we have a high degree of confidence in the accuracy of ages estimated from specimens with clear growth zones. Acknowledgments We thank the staff at the Age and Growth Program of the Alaska Fisheries Science Center for support during this study. We also wish to thank T. Wilderbuer and W. Stockhausen of the Alaska Fisheries Science Center for helpful reviews and comments on early versions of this manuscript. K. Mckinney provided photographic sup- port for which we are grateful. We thank A. Andrews of Moss Landing Marine Laboratories and two anonymous reviewers for insightful comments on the manuscript. S. Handwork and K. Elder of the National Ocean Sciences Accelerator Mass Spectrometry Facility at the Woods Hole Oceanographic Institution provided support and technical advice regarding the A14C measurements for which we are grateful. Literature cited Abookire, A. A., and K. M. Bailey. 2007. The distribution of life cycle stages of two deep- water pleuronectids, Dover sole ( Microstomus pacificus) and rex sole ( Glyptocephalus zachirus), at the northern extent of their range in the Gulf of Alaska. J. Sea Res. 57:198-208. Abookire, A. A., and B. J. Macewicz. 2003. Latitudinal variation in reproductive biology and growth of female Dover sole ( Microstomus pacificus) in the North Pacific, with emphasis on the Gulf of Alaska stock. J. Sea Res. 50:187-197. Abookire, A. A., J. F. Piatt, and B. L. Norcross. 2001. Juvenile groundfish habitat in Kachemak Bay, Alaska, during late summer. Alaska Fish. Res. Bull. 8(1 ):45 — 56. Andrews, A. H., E. J. Burton, L. A. Kerr, G. M. Cailliet, K. H. Coale, C. C. Lundstrom, and T. A. Brown. 2005. Bomb radiocarbon and lead-radium disequilibria in otoliths of bocaccio rockfish ( Sebastes paucispinis ): a determination of age and longevity for a difficult-to-age fish. Mar. Freshw. Res. 56:517-528. Andrews, A. H., L. A. Kerr, G. M. Cailliet, T. A. Brown, C. C. Lundstrom, and R. D. Stanley. 2007. Age validation of canary rockfish (Sebastes pin- niger) using two independent otolith techniques: lead- radium and bomb radiocarbon dating. Mar. Freshw. Res. 58:531-541. Beamish, R. J., and G. A. McFarlane. 1983. The forgotten requirement for age validation in fish- eries biology. Trans. Am. Fish. Soc. 112(6):735 — 743. Campana, S. E. 1997. Use of radiocarbon from nuclear fallout as a dated marker in the otoliths of haddock Melanograrnmus aeglefinus. 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(2):185-192. Campana, S. E., L. J. Natanson, and S. Myklevoll. 2002. Bomb dating and age determination of large pelagic sharks. Can. J. Fish. Aquat. Sci. 59:450-455. Kastelle et al.: Age validation of Microstomus paaficus by means of bomb radiocarbon 385 Campana, S. E., R. D. Stanley, and S. Wischniowski. 2003. Suitability of glycerin-preserved otoliths for age validation using bomb radiocarbon. J. Fish Biol. 63:848-854. Chambers, J. M., and T. J. Hastie. 1992. Statistical Models in S, 608 p. Wadsworth and Brooks/Cole Advanced Books and Software, Pacific Grove, CA. 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. Francis, R. I. C. C., and P. L. Horn. 1997. Transition zone in otoliths of orange roughy (Hoplo- stethus atlanticus) and its relationship to the onset of maturity. Mar. Biol. 129:681-687. Gregg, J. L., D. M. Anderl, and D. K. Kimura. 2006. Improving the precision of otolith-based age esti- mates for Greenland halibut ( Reinhardtius hippoglos- soides) with preparation methods adapted for fragile sagittae. Fish. Bull. 104:643-648. Hunter, J. R., J. L. Butler, C. Kimbrell, and E. A. Lynn. 1990. Bathymetric patterns in size, age, sexual maturity, water content, and caloric density of Dover sole, Micros- tomus pacificus. Calif. Coop. Oceanic Fish. Invest. Rep. 31:132-144. Hutchinson, C. E., C. R. Kastelle, D. K. Kimura, and D. R. Gunderson. 2007. Using radiometric ages to develop conventional ageing methods for shortraker rockfish ( Sebastes borealis). In Biology, assessment, and management of North Pacific rockfishes (J. Heifetz, J. DiCosimo, A. J. Gharrett, M. S. Love, V. M. O’Connell, and R. D. Stanley, eds.), p. 237-249. Alaska Sea Grant College Program, AK-SG-07-01, Univ. Alaska Fairbanks, Fairbanks, AK. Kalish, J. M. 1993. Pre- and post-bomb radiocarbon in fish otoliths. Earth Planet. Sci. Lett. 114:549 — 554. 1995. Radiocarbon and fish biology. In Recent develop- ment in fish otolith research (D. H. Secor, J. M. Dean, and S. E. Campana, eds), p. 637-653. Belle W. Baruch Library in Marine Science Number 19. Univ. South Carolina Press, Columbia, S.C. Kastelle, C. R., and D. K. Kimura. 2006. Age validation of walleye pollock ( Theragra chalcogramma) from the Gulf of Alaska using the dis- equilibrium of Pb-210 and Ra-226. ICES J. Mar. Sci. 63:1520-1529. Kastelle, C. R., D. K. Kimura, and B. J. Goetz. 2008. Bomb radiocarbon age validation of Pacific ocean perch (Sebastes alutus ) using new statistical methods. Can. J. Fish. Aquat. Sci. 65:1101-1112. Kerr, L. A., A. H. Andrews, G. M. Cailliet, T. A. Brown, and K. H. Coale. 2006. Investigations of A14C, <513C, and 615N in verte- brae of white shark ( Carcharodon careharias ) from the eastern North Pacific Ocean. Environ. Biol. Fishes 77:337-353. 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 ( Sebastes ruberrimus ): a reference time series for the coastal waters of southeast Alaska. Can. J. Fish. Aquat. Sci. 61:443-451. Kerr, L. A., A. H. Andrews, K. Munk, K. H. Coale, B. R. Frantz, G. M. Cailliet, and T. A. Brown. 2005. Age validation of quillback rockfish (Sebastes malinger) using bomb radiocarbon. Fish. Bull. 103:97- 107. Kimura, D. K., and D. M. Anderl. 2005. Quality control of age data at the Alaska Fisheries Science Center. Mar. Freshw. Res. 56:783-789. Kimura, D. K., C. R. Kastelle, B. J. Goetz, C. M. Gburski, and A. V. Buslov. 2006. Corroborating the ages of walleye pollock (Theragra chalcogramma). Mar. Freshw. Res. 57:323-332. Miller, R. G. 1966. Simultaneous statistical inference, 272 p. Mc- Graw-Hill, New York, NY. Norcross, B. L., B. A. Holladay, and F. J. Muter. 1995. Nursery area characteristics of Pleuronectids in coastal Alaska, USA. Neth. J. Sea Res. 34:161-175. Nydal, R. 1993. Application of bomb 14C as a tracer in the global carbon cycle. Trends Geophys. Res. 2:355-364. Piner, K. R., O. S. Hamel, J. L. Menkel, J. R. Wallace, and C. E. Hutchinson. 2005. Age validation of canary rockfish ( Sebastes pin- niger) from off the Oregon coast (USA) using the bomb radiocarbon method. Can. J. Fish. Aquat. Sci. 62:1060-1066. Piner, K. R., J. R. Wallace, O. S. Hamel, and R. Mikus. 2006. Evaluation of ageing accuracy of bocaccio (Sebastes paucispinis) rockfish using bomb radiocarbon. Fish. Res. 77:200-206. Piner, K. R., and S. G. Wischniowski. 2004. Pacific halibut chronology of bomb radiocarbon in otolliths from 1944 to 1981 and a validation of ageing methods. J. Fish Biol. 64:1060-1071. Stockhausen, W. T., B. J. Turnock, Z. T. A’mar, M. E. Wilkins, and M. H. Martin. 2005. Gulf of Alaska Dover sole. In Appendix B, Stock Assessment and Fishery Evaluation Report for the Groundfish Resources of the Gulf of Alaska, p. 351- 397. North Pacific Fishery Management Council, P.O. Box 103136, Anchorage, AK 99510. Stuiver, M., and H. A. Polach. 1977. Discussion: reporting of 14C data. Radiocarbon 19(3):355-363. 386 Abstract — A 4500-year archaeologi- cal record of Pacific cod ( Gadus mac- rocephalus) bones from Sanak Island, Alaska, was used to assess the sus- tainability of the modern fishery and the effects of this fishery on the size of fish caught. Allometric reconstruc- tions of Pacific cod length for eight prehistoric time periods indicated that the current size of the nearshore, commercially fished Pacific cod stocks is statistically unchanged from that of fish caught during 4500 years of subsistence harvesting. This finding indicates that the current Pacific cod fishery that uses selective harvest- ing technologies is a sustainable commercial fishery. Variation in rela- tive Pacific cod abundances provides further insights into the response of this species to punctuated changes in ocean climate (regime shifts) and indicates that Pacific cod stocks can recover from major environmental perturbations. Such palaeofisheries data can extend the short time-series of fisheries data (<50 yr) that form the basis for fisheries management in the Gulf of Alaska and place current trends within the context of centen- nial- or millennial-scale patterns. Manuscript submitted 21 June 2007. Manuscript accepted 16 June 2008. Fish. Bull. 106:386-394 (2008). The views and opinions expressed or implied in this article are those of the author and do not necessarily reflect the position of the National Marine Fisheries Service, NOAA. A 4500-year time series of Pacific cod (Gadus macrocephalus ) size and abundance: archaeology, oceanic regime shifts, and sustainable fisheries Herbert D. G. Maschner (contact author)1 Matthew W. Betts2 Katherine L. Reedy-Maschner1 Andrew W. Trites3 ' Department of Anthropology Idaho State University 921 S. 8th Avenue, Stop 8005 Pocatello, Idaho 83209-8005 2 Archaeology and History Division Canadian Museum of Civilization 100 Laurier Street, Box 3100, Station B Gatineau, Quebec, J8X 4H2 Canada 3 Fisheries Centre, Room 247 Aquatic Ecosystems Research Laboratory, 2202 Main Mall University of British Columbia Vancouver, B C. V6T 1Z4 Canada Email address for H. D. G. Maschner: maschner@isu.edu Fishing is a size-selective process that tends to remove larger fish and reduce the life span and mean size of individuals, especially in heavily exploited systems (Shin et al., 2005; Trites et al., 2006). Reductions in size-at-age and age-at-maturation have been reported for a number of heavily exploited species of fish (Trip- pel, 1995; Rochet, 1998; Bianchi et al., 2000) and may cause evolutionary change (Ernande et al., 2004; Hutch- ings, 2005; Law, 2007). Knowledge of body size is therefore an impor- tant metric for fisheries management because it can affect market price and reflects the spawning potential of the fish population (Shin et al., 2005). Most fisheries management and conservation research has been re- stricted to interpreting trends and data sets that span just a few de- cades (Pauly et al., 1998; Worm et al., 2006), and tend not to incorporate the longer-term perspective that can be obtained by including palaeoeco- logical data (Willis and Birks, 2006). Combining palaeoecological data with biological data can provide a signifi- cantly longer time series for measur- ing population and ecosystem health for conservation biology and resource management, and for understanding how current decadal trends fit within the context of centennial- or millen- nial-scale cycles (Jackson et al., 2001; Braje et al., 2006). Humans have been sampling species and ecosystems for thousands of years and have left rich data sets of both natural ecosystem dynamics and human interactions with animal populations that are preserved in archaeofaunal remains (animal bones found on archaeologi- cal sites). Sanak Island is the center of a small, low-lying archipelago on the edge of the continental shelf in the western Gulf of Alaska (Fig. 1). This island group was a hub of the North Pacific Pacific cod ( Gadus macroceph- alus) fishery from the 1870s to the 1930s (Reedy-Maschner, 2004). Lo- cal Aleuts reported that Pacific cod disappeared in commercial quantities in this region between 1942 and 1975 but returned and have supported the modern Pacific cod fishery since the 1975 oceanic regime shift. Maschner et al.: A 4500-year time series of Gadus macrocephalus size and abundance 387 165°0'0"W 160°0'0”W The North Pacific and the western Gulf of Alaska region showing Alaska and the location of Sanak Island where bones were collected at archeological sites to determine whether a change in fish size was evident over the 4500 archeological record. The Sanak Pacific cod fishery now includes mostly fishermen of Aleut-Scandinavian descent who use jigs, pots, and longlines in nearshore waters, but the fishery has deep, prehistoric roots (Tews, 2005). Archaeological data indicate that Aleuts have occupied Sanak Island for over 6000-7000 years and Pacific cod bones domi- nate the matrices of all preserved midden deposits from over 120 prehistoric village sites found on the island. These remains provide a novel laboratory for construct- ing long-term time series of Pacific cod populations. The goal of this study was to compare archeological data with modern fisheries data to assess whether in- dustrialized fishing has changed the size of Pacific cod in the North Pacific Ocean and to investigate whether or not Pacific cod size and abundance may be influenced by climate change. Pacific cod are among the most heavily exploited and consumed species of fish in the Northern Hemisphere and have declined significantly in most parts of the world in the past 30-40 years. In eastern Canada, fishing resulted in Atlantic cod ( Gadus morhua) becoming smaller and reaching sexual maturi- ty at younger ages as fisheries reduced the populations by more than 90% (Fanning et ah, 2003; Hutchings, 2005; Lilly et al., 2005). Evolutionary reductions in body size caused by fishing have also been reported for Baltic cod ( Gadus morhua callarias) (Andersen et al., 2007). In the Gulf of Alaska, Pacific cod are generally thought to be fished at sustainable levels, but there is a limited time series to assess whether fisheries have changed the size of fish caught over the past 50 years (Thompson et al., 2006). It is, however, possible to use paleofisheries data to assess whether the lengths and, by correlation, average fecundity of fish caught today are comparable to those caught thousands of years ago before the advent of industrialized fisheries, and to use lengths and frequencies of Pacific cod harvested in prehistory to assess the role of climate in the structure of Pacific cod populations. Materials and methods All of the samples used to estimate the sizes of fish over the prehistoric time sequence were derived from shell-midden deposits that were excavated by trowel and sieved through 6-mm mesh, an appropriate sieve size given specimen sizes in the region (e.g., see Discussion in Cannon, 1999), and our results were substantiated by direct comparisons with bulk samples processed with 3-mm mesh). In our analysis, we controlled for possible 388 Fishery Bulletin 106(4) spatial differences in procurement of samples by includ- ing only sites located at the eastern end of the island that were associated with rocky intertidal coastlines. We further controlled for possible seasonal differences in procurement by including only midden deposits that were stratigraphically associated with winter villages (as denoted by the presence of large semisubterranean house ruins). Sample sizes for Pacific cod in each of these faunal assemblages ranged between 48 and 3219 specimens (individual bones). Actual sample sizes for measured Pacific cod skeletal elements ranged from a low of 10 to a high of 507 elements. By archaeological standards, the number of fish bones recovered from these middens was large, and thus the data set was relatively robust. Nevertheless, we recognize the possibility of potential sample-size effects, and present the entire data set as a case study. The majority of Pacific cod bones recovered contained significant biometric information because they were not fragmented and came from well-preserved deposits. Because length figures prominently in modern fisheries management research (Shin et al., 2005; for the Gulf of Alaska see Thompson et al., 2006), allometric rela- tionships developed by Orchard (2001, 2003) were used to establish the live length of an individual from the skeletal elements of Pacific cod. As an exercise to pres- ent the types of data that can be reconstructed from ancient Pacific cod bones, the fecundities of Pacific cod were estimated from their length distributions, by means of regressions developed by Karp (1982). More accurate reconstructions could be based on average fecundity relationships specifically recorded for Pacific cod in the Gulf of Alaska, although these data were not available at the time of publication of this article. Thus, the assemblages of skeletal remains recovered from middens in our study contained information about Pacific cod populations that could be compared to information on modern population structures for a large-scale view of changes in size and abundance over time. Measurements were taken from the ascending process of premaxillas and the centra of trunk vertebrae from Pacific cod derived from eight archaeological sites span- ning the period ca. 2550 BC to 1540 AD (radiocarbon years were calibrated to calendar years by using the methods of Stuiver and Reimer, 1993). Although these data were time averaged, the samples were from single stratigraphic units dated by multiple radiocarbon mea- surements and likely represent a single depositional episode that took place over one or a few seasons. From these data, the estimated length of the individuals that contributed each element was computed from Orchard’s (2001, 2003) regression equations and the combined measures were used to produce length-frequency dis- tributions for each time slice. The modern length distribution of Pacific cod was obtained from the longline survey done by the Alaska Fisheries Science Center in the Gulf of Alaska in 2005 (n = 3308; Thompson et al., 2006). Longline surveys are a fishing method that may be most comparable with prehistoric jigging practices. However, it should be noted that modern longline gear may subtly select for smaller size fish, as discussed by Halliday (2002). The mean longline survey data (Thompson et al., 2006) are presented as frequencies in specific size bins, and mean length was calculated by using procedures in Gedamke and Hoenig (2006). We also measured the relative abundance of Pa- cific cod in the prehistoric middens using a measure known as an abundance index (Al). This measure can be used to track shifts in the relative abundance of taxa in relation to other taxa in a faunal assemblage and represents a normed ratio of a highly ranked (in terms of human foraging efficiency) taxa A to a lower ranked taxa B, measured as AI=A/\A+B ] (Bayham, 1979; Ugan and Bright, 2001). Values close to 0 indi- cate a complete absence of taxa A, and values close to 1 indicate a dominating presence. We used an abun- dance index because they are generally robust and are resistant to taphonomic (conditions affecting preser- vation) and collection biases, as long as these biases are systemic to all the assemblages being compared (Ugan and Bright, 2001) — a situation applicable to the Sanak Island data. Although simple to calculate, the measures can be powerful. For example, shifts in the index can reflect changes in human foraging efficiency because they incorporate body-size-based caloric relationships. In the model, it is assumed that larger bodied prey are often more highly ranked than smaller bodied prey (because they contain more calo- ries per unit of effort); therefore when a highly ranked or large-bodied prey is compared to a lower-ranked, or smaller-bodied prey, the values can be used to de- termine the occurrence of resource depressions, or declines in foraging efficiency. Abundance indices are a well-established and peer-reviewed method of measuring changes in the frequencies of taxa through time from archaeological data (Broughton, 1994, 1997; Butler, 2000; Ugan and Bright, 2001; Nagaoka, 2002; Betts and Friesen, 2006). Problems can arise in the interpretation of Al mea- sures when inappropriate B taxa, or lower-ranked taxa, are chosen for comparison with higher ranked A taxa. Betts and Friesen (2006) demonstrated that these is- sues can be overcome by 1) carefully selecting B taxa that occur in moderate, but relatively stable frequen- cies in the assemblages, and 2) comparing multiple low-ranked B taxa and assessing overall trends. Here we compared Pacific cod to smaller-bodied species of fish — Cottidae, Hexagrammidae, Pleuronectidae, and Oncorhynchus. The former three taxa were found in the same resource patch (i.e., in the nearshore jig fishery), and likely entered the procurement system as bycatch. The last taxon, salmon, was likely captured at stream mouths and head waters with weirs and nets, and was included in our analyses for comparative purposes. Mean lengths of Pacific cod recovered from different eras were compared by using t-tests and analysis of variance, and linear regressions were used to deter- mine the significance of trends over time. Maschner et al.: A 4500-year time series of Gadus macrocephalus size and abundance 389 Table T Comparisons of mean fork lengths of Pacific cod ( Gadus macrocephalus) caught by Aleuts from 1540 AD to 2550 BC compared to the mean fork lengths of fish captured in 2005 by longline surveys in the Gulf of Alaska. Comparisons were made by using L tests, and values in bold indicate no significant difference at the P<0.01 level. Sample size n is the number of measured bone elements from the prehistoric eras. The site numbers reflect individual winter village sites on Sanak Island from which the archaeological samples were derived. Mean fork length (cm) Site number Era Prehistoric Modern Lvalue df P n 111 2550 BC 67.5 67.5 0.003 26 0.997 27 054 1750 BC 70.7 67.5 5.932 573 <0.001 467 036 840 BC 72.1 67.5 5.471 340 <0.001 313 061 595 BC 64.1 67.5 -1.181 9 0.267 10 058 80 BC 69.7 67.5 3.801 601 0.012 507 056 Upper 520 AD 65.5 67.5 -2.011 175 0.046 166 056 Lower 1030 AD 66.6 67.5 -0.981 257 0.327 241 110 1540 AD 62.8 67.5 -6.223 515 <0.001 464 Figure 2 Box plots showing the fork-length (mm) distributions of Pacific cod (Gadus macrocephalus) caught in the Gulf of Alaska since 2550 BC. The prehistoric distributions were reconstructed from skeletal elements (premaxilla and centra from trunk verte- brae) by using allometric relationships established by Orchard (2003). The modern length distributions were compiled from data collected from longline surveys conducted in the Gulf of Alaska by the Alaska Fisheries Science Center between 1978 and 2006 (Thompson et al., 2006). Results and discussion Measuring changes in Pacific cod size and abundance The mean size of Pacific cod has varied consid- erably in the Gulf of Alaska over the last 4500 years (Fig. 2; range 62.8-72.1 cm; F{ 8 5503)=11.97, P<0.001). However, modern mean lengths did not differ significantly from the mean lengths of fish caught around 2550 BC, 595 BC, and 1030 AD (Table 1 ; Ltests P>0.05), or when compared with lengths of those taken around 80 BC and 520 AD (Table 1; Ltests P>0.01). Fecundities of Pacific cod (estimated from body length distributions) varied dramatically, ranging between 2.1 and 3.9 million eggs per individual (Fig. 3). Average fecundity varied consistently with changes in mean length as expected, but the apparent trend shown in Figure 3 towards decreasing fecundity over 4500 years was not significant (r=-0.42, P=0.35). The Al measures revealed changes in the rela- tive abundance of Pacific cod over the past 4500 years (Fig. 4). All of the AIs were highly variable between 2550 and 2580 BC, but were synchro- nized after 520 AD. They indicated high relative abundance of Pacific cod in middens spanning 2550-1750 BC and showed an opposite trend in the abundance of salmon, which was consistent with the inverse relationship previously noted for this region and time period (e.g., Tews, 2005; Misarti, 2007). From 840 BC to 80 BC another period of variability occurred when salmon Al was opposite that of the groundfish taxa. During this time, the AIs for Pleuronectidae and Hexagrammidae indicated a slight decline in relative abundance of Pa- cific cod, whereas the Cottidae index indicated a greater decline in Pacific cod abundance. Taken in tandem, however, all of the groundfish taxa recovered from the 390 Fishery Bulletin 106(4) middens indicate that there were slight or moderate declines in Pacific cod during this early period. On the basis of the remains in the middens, there was a sharp decline in the relative abundance of Pacific cod (compared to the other three taxa) beginning around 520 AD. This declining trend intensified drastically in the 1030 AD assemblage, and was followed by a sharp increase in Pacific cod abundance at 1540 AD. The declining AIs strongly indicate that a form of resource depression occurred from 520 through 1030 AD (for a similar interpretation of declining fish AIs, see Nagaoka, 2001, 2002). Resource depressions can be related to a number of factors, such as exploitation pressure, behavioral changes, and microhabitat (range) shifts (e.g., Charnov et al., 1976). The apparent decrease in Pacific cod abundance that we noted is intriguing given the lack of any technological, procurement, or other cultural changes (e.g., increased territoriality) that could have influenced the encounter or success rate of the Aleut Pacific cod fishery. Instead we suspect the decline in Pacific cod reflects an environmentally-driven natural change in abundance and note that the increase in mean length during this period corroborates this interpretation (see Shin et al., 2005). Interpreting the effects of climate on Pacific cod size and abundance Punctuated shifts in ocean climate (regime shifts) are believed to explain many of the changes observed in abundances of some species in the North Pacific Ocean during the past century (Hare and Mantua, 2000; Benson and Trites, 2002; Trites et al., 2007) and these shifts may explain the variations noted in Figure 4. Pacific cod populations may be susceptible to fluctuations in oceanic regimes to the extent that they periodically disappear in significant numbers from the ecosystem, only to reappear in greater numbers at a later date. This condition has deep historical roots; the ancient Aleut name for Pacific cod translates literally into “the fish that stops” because this species periodically disappears (Black, 1981), a situation that occurred, according to traditional Aleut knowledge, at least once in the mid-19th century, and again in approximately 1942. Several small variations in mean Pacific cod length were evident over the temporal sequence. Correlating the shifts in body size with climatic shifts does not appear to explain the fluctuating sizes of Pacific cod between 2550 BC and 80 BC, which occurred during the generally cool and wet conditions of the Neoglacial period (the first major postglacial cooling period from approximately 2500 BC to AD 1). The average lengths of Pacific cod (Fig. 2) increased as the warming and drying of the Medieval climatic anomaly began (a period of hemispheric climatic fluctuations ca. AD 1000-1300 across the Northern Hemisphere) and decreased slightly during the cool and wet conditions of the Little Ice Age (ca. AD 1400-1850). Marine productivity during the Medieval climatic anomaly and the Little Ice Age, as recorded by Finney et al. (2002) and Misarti (2007), appears to be inversely related to Pacific cod lengths but positively related with Pacific cod numbers (Fig. 4). Such inverse relationships between mean fish length, abundance, and productivity have been noted by Shin et al. (2005) (i.e., increased productivity and abundance is usually associ- ated with decreases in mean length, presum- ably because of the increase of juvenile fish in the population). The impact of these types of processes on fish lengths over the long time scales has never been assessed, and so this must remain a working hypothesis. Regard- less, it is also notable that the mean length of Pacific cod during the Medieval climatic anomaly was not statistically different from that of the modern era codfish (Table 1) — two of the warmest periods in the last 4500 years. The effects of oceanic regime shifts that have dominated the dynamics of marine eco- systems in the North Pacific for the past cen- tury (Benson and Trites, 2002; Polovina, 2005) are difficult to monitor in the prehistoric pe- riod because of an important complicating fac- tor— an overwhelming lack of data. Although climatic and oceanic warming AD 1000 to 1300 have been documented around the North Pacific (Calkin et al., 2001; Hu et al., 2001), and its effects are beginning to be understood (Jones et al., 1999), the full effects of climate change on Pacific cod stocks over the entire period are difficult to measure because not a Figure 3 Mean estimated fecundity of Pacific cod ( Gadus macrocephalus) over time based on the relationships between body length and fecundity established by Karp (1982). This graph indicates that estimated mean fecundity of Pacific cod has varied widely over the past 4500 years, despite concomitantly small changes in mean Pacific cod length (compare with Fig. 2). Although the graph indicates an apparent trend toward decreasing fecundity, the relationship is not significant; r=-0.42, P=0.35. Maschner et al.: A 4500-year time series of Gadus macrocephalus size and abundance 391 Figure 4 Abundance indices (e.g., AI=Pacific cod / [ comparison taxon + Pacific cod]), calculated from the number of identified specimens (bone fragments) for each taxon recovered from nine archaeologi- cal deposits on Sanak Island. The index ranges between 0 and 1, with 0 indicating the complete absence from the middens and 1 indicating a complete dominance of Pacific cod ( Gadus macro- cephalus). Sample sizes from the deposits ranged between 426 and 8339 identified bone specimens. All indices indicate a decline in Pacific cod abundance beginning at 520 AD, except for the Gadi- da e-Oncorhynchus spp. index, which is included to show the differ- ence between riverine ( Oncorhynchus spp.) and nearshore marine fish (the other families of fish represented in the graph) during this period. This decline in abundance continues throughout the period until 1030 AD, after which we see a recovery in Gadidae abundance at ca. 1540 AD. The correspondence in the multiple measures negates the mathematical conundrum of closed arrays and indicates real changes in the relative abundance of Pacific cod in Sanak middens. single preserved Pacific cod bone dating from AD 1100 to 1300 in the Sanak archaeological data has been recovered despite the presence of village sites. The period between AD 1100 and 1300 were the only centuries during the last 5000 years of extensive global warming that occurred before the modern era. Pacific cod returned in the middens only during the hemispheric cooling associated with the Little Ice Age. Interestingly, regardless of periodic regime shifts in ocean climate, Pacific cod populations appeared to have returned to the western Gulf of Alaska with approximately the same length structure as that before they disappeared. The perturbations in the body lengths of Pacific cod did not correlate with changes in their frequencies of occurrence in the middens (Fig. 4). For example, the most extreme shift in mean Pacific cod length occurred between 2550 BC and 80 BC, yet Pacific cod abun- dances varied minimally during this time. It is noteworthy that this entire period occurred during the Neoglacial, a period of generally cooler and wetter oceanic conditions. This re- lationship may indicate that perturbations in abundance, at least during cold periods, were not related to systemic fluctuations in the length or fecundity structure of the Pacific cod populations in the Gulf of Alaska. At the end of the temporal sequence ca. 1030 AD and 1540 AD, an inverse relationship is noted between mean length and abundance of Pacific cod. The mean lengths increased during the Medieval climatic anomaly and decreased during the Little Ice Age, whereas overall relative abundances decreased during the Medieval climatic anomaly and increased during the Little Ice Age. This finding would indicate that Pacific cod were more abundant but smaller during cool periods and less abundant but larger during warm- er periods. Although the inverse relationship between length and abundance is intriguing, it is uncertain how they are related. The shift in mean Pacific cod length between these periods was small (less than 3 cm), yet abundances fluctuated widely. Climatic regime shifts appear to have had minimum effects on the length structure of Pacific cod populations in the Gulf of Alaska, but have had dramatic effects on their numbers and possibly geographic range. The apparent drop in Pacific cod numbers during periods of warming may reflect a shift in Pacific cod distribution (e.g., Charnov et al., 1976), and may explain why the Aleut word for Pacific cod is the “fish that stops.” General observations regarding the prehistoric Pacific cod data Additional insights about Pacific cod can be made from the analyzed bones, which show, for example, that Pacific cod have long-term, millennial-scale population dynam- ics that appear to have a complex relationship with cli- matic shifts. Pacific cod seem to be vulnerable to major climatic regime shifts, particularly warming conditions, and appear to rapidly recover from major perturbations to their environment. This conclusion would signify that oceanic regime shifts have no long-term (centennial- to millennial-scale) ramifications on the overall structure of Pacific cod populations. Furthermore, Pacific cod stocks appear to have largely maintained their popula- tion structure over some 4500 years despite the pres- sures of commercial harvesting in the modern era. The trends uncovered for Pacific cod in the North Pa- cific are significantly different from those in the North Atlantic. Swain et al. (2007) concluded that length dis- tributions can be a key indicator of genetic changes in Pacific cod as a result of overfishing, and that these changes can persist for decades despite low harvest pres- sure. In particular, they found that rapid evolutionary changes can result from size-selective overfishing, and these changes can far outpace the levels expected for natural mortality as a result of disease or regime shifts. 392 Fishery Bulletin 106(4) The archaeological record in the western North Atlan- tic shows that the average length of Pacific cod caught today by commercial fisheries is at least 40 cm smaller than that of prehistoric times, and indicates that the Pacific cod populations have been impacted by modern fisheries (Kenchington and Kenchington, 1993; Jackson et ah, 2001). Similar shifts have been discovered in the central North Atlantic where Amorosi et al. (1994) demonstrated significant differences in the length dis- tributions of Pacific cod between the Medieval period and the modern era. In contrast, we found that the minimum difference in the average length of Pacific cod taken by longline fisheries was only 3-4 cm larger than the smallest prehistoric mean size fish caught by the ancient Aleuts using jigs. The implications of these regional differences are intriguing given the dif- ferent historical trajectories in commercialization of the groundfish fisheries in these two regions, namely a different emphasis on harvesting technologies (i.e., a much higher proportion of the North Pacific catch is taken with cod pots, compared to the North Atlantic, where cod pots have seen minimal use). The North Pacific cod pot fishery, in particular, reduces the catch of juvenile fish. This study points to ways in which ancient archaeo- logical deposits can be used to significantly extend the time-depth of fish population studies and can provide important insights into long-term sustainability. Such palaeofisheries research is important for placing modern decadal population trends within a longer-term perspec- tive. In the case of Pacific cod, the paleoecological re- cords indicate that today’s Pacific cod are comparable in size to fish that inhabited the Gulf of Alaska thousands of years ago. This finding indicates that modern fisher- ies have not altered the average length distributions of Pacific cod in the Gulf of Alaska when measured over long time scales. The fact that modern Pacific cod length distributions are within the range of precommercial variability in- dicates that the current commercial Pacific cod fishery has sustained the length structure (and therefore aver- age fecundity structure) of the prehistoric (noncommer- cial) Gulf of Alaska Pacific cod population. Compared to the North Atlantic, the changes we observed in fish length between the prehistoric and modern eras in the Gulf of Alaska are more consistent with natural fluctua- tions than with harvesting pressure. The fact that the present-day size-structure of Gulf of Alaska Pacific cod stocks is consistent with a sustainable fishery (Thomp- son et al., 2006) and that the current size-structure is comparable to that of pre-industrially fished populations indicates that current Pacific cod fisheries management policies and harvesting techniques in the western Gulf of Alaska are working. That is, management practices and harvests are maintaining a natural Pacific cod population structure based on mean size as a compara- tive metric of sustainability. However, we caution that global warming may become a complicating factor for modern management practices given that Pacific cod populations may be vulnerable to pressures caused by increasing oceanic temperatures, as appear to have oc- curred circa AD 1000-1300. Acknowledgments This research was funded by the National Science Foundation under awards NSF OPP 0326584 and NSF BE/CNH 0508101. L. Ford, Interim Vice President of Research at Idaho State University, provided gener- ous postdoctoral support for this project. We are also grateful for the support provided to A.W. Trites by the North Pacific Marine Science Foundation through the North Pacific Universities Marine Mammal Research Consortium. We thank G. Thompson and four anony- mous reviewers for their valuable comments, and grate- fully acknowledge the student laboratory assistants who sorted nearly 100,000 fish bones. The fieldwork on Sanak Island was done with the permission of the Sanak Corporation and in cooperation with the Pauloff Harbor Tribe. Literature cited Amorosi, T., T. McGovern, and S. Perdikaris. 1994. Bioarchaeology and cod fisheries: a new source of evidence. ICES Mar. Sci. Symp. 198:31-48. Andersen, K., K. Farnsworth, U. Thygesen, and J. Beyer. 2007. The evolutionary pressure from fishing on size at maturation of Baltic cod. Ecol. Model. 204:246-252. Bayham, F. 1979. Factors influencing the archaic pattern of animal utilization. Kiva 44:219-235. Benson, A. J., and A. W. Trites. 2002. Ecological effects of regime shifts in the Bering Sea and eastern North Pacific Ocean. Fish Fish. 3:95-113. Betts, M., and T. M. Friesen. 2006. Declining foraging returns from an inexhaustible resource? Abundance indices and beluga whaling in the western Canadian Arctic. J. Anthropol. Archaeol. 25:59-81. Bianchi, G., H. Gislason, K. Graham, L. Hill, X. Jin, K. Koran- teng, S. Manickchang-Heileman, I. Paya, K. Sainsbury, F. Sanchez, and K. Zwanenburg. 2000. Impact of fishing on size composition and diver- sity of demersal fish communities. ICES J. Mar. Sci. 57:558-571. Black, L. T. 1981. Volcanism as a factor in human ecology: the Aleu- tian case. Ethnohist. 28:313-340. Braje, T. J., J. M. Erlandson, D. J. Kennett, and T. C. Rick. 2006. Archaeology and marine conservation. Archaeol. Rec. 6:14-19. Broughton, J. 1994. Declines in mammalian foraging efficiency during the Late Holocene, San Francisco Bay, California. J. Anthropol. Archaeol. 13:371-401. 1997. Widening diet breadth, declining foraging efficiency, and prehistoric harvest pressure: ichthyofaunal evidence from the Emeryville Shellmound, California. Antiquity 71:845-862. Maschner et al.: A 4500-year time series of Gadus macrocephalus size and abundance 393 Butler, V. 2000. Resource depression on the northwest coast of North America. Antiquity 74:649-661. Calkin, P. E., G. Wiles, and D. J. Barclay 2001. Holocene coastal glaciation of Alaska. Quart. Sci. Rev. 20:449-461. Cannon, M. 1999. A mathematical model of the effects of screen size on zooarchaeological relative abundance measures. J. Arch. Sci. 26:205-214. Charnov, E., G. Orians, and K. Hyatt. 1976. Ecological implications of resources depression. Am. Nat. 110:247-259. Ernande, B., U. Dieckmann, and M. Heino. 2004. Adaptive changes in harvested populations: plastic- ity and evolution of age and size at maturation. Proc. Roy. Soc. B 271:415-423. Fanning, L. P., R. K. Mohn, and W. J. MacEachern. 2003. Assessment of 4VsW cod to 2002, 45 p. Research Document 2003/027. Fisheries and Oceans Canada, Communications Branch, 200 Kent Street, Station 13E228, Ottawa, Ontario, Canada. Finney, B. P, I. Gregory-Eaves, M. S. V. Douglas, and J. P. Smol. 2002. Fisheries productivity in the northeastern Pacific Ocean over the past 2,200 years. Nature 416:729- 733. Gedamke, T., and J. M. Hoenig 2006. Estimating mortality from mean length data in nonequilibrium situations, with application to the assessment of goosefish. Trans. Am. Fish. Soc. 135:476-487 Halliday, R. G. 2002. A comparison of size selection of Atlantic cod ( Gadus morhua) and haddock ( Melanogrammus aegle- fitius) by bottom longlines and otter trawls. Fish. Res. 57(1): 63— 73 . Hare, S. R., and N. J. Mantua. 2000. Empirical evidence for North Pacific regime shifts in 1977 and 1989. Prog. Oceanogr. 47:103-146. Hu, F. S., E. Ito, T. A. Brown, B. B. Curry, and D. R. Engstrom 2001. Pronounced climatic variations in Alaska during the last two millennia. Proc. Natl. Acad. Sci. 98(19): 10552-10556. Hutchings, J. A. 2005. Life history consequences of overexploitation to population recovery in Northwest Atlantic cod ( Gadus morhua). Can. J. Fish. Aquat. Sci. 62:824-832. Jackson, J. B. C., M. X. Kirby, W. H. Berger, K. A. Bjorndal, L. W. Botsford, B. J. Bourque, R. H. Bradbury, R. Cooke, J. Erland- son, J. A. Estes, T. P. Hughes, S. Kidwell, C. B. Lange, H. S. Lenihan, J. M. Pandolfi, C. H. Peterson, R. S. Steneck, M. J. Tegner, and R. R. Warner. 2001. Historical overfishing and the recent collapse of coastal ecosystems. Science 293:629-638. Jones, T. L., G. M. Brown, L. M. Raab, J. L. McVickar, W. G. Spaulding, D. J. Kennett, A. York, and P. L. Walker. 1999. Environmental imperatives reconsidered: demo- graphic crises in western North America during the Medieval climatic anomaly. Curr. Anthrop. 40:137- 170. Karp, W. 1982. Biology and management of Pacific cod ( Gadus macrocephalus) in Port Townsend, Washington. Ph.D. diss., 120 p. Univ. Washington, Seattle, WA. Kenchington, T. J., and E. L. Kenchington. 1993. An eighteenth century commercial length-fre- quency sample of Atlantic cod, Gadus morhua , based on archaeological data. Fish. Res. 18:335-347. Law, R. 2007. Fisheries-induced evolution: present status and future directions. Mar. Ecol. Prog. Ser. 335:271-277. Lilly, G. R., J. Brattey, N. G. Cadigan, B. P. Healey, and E. F. Murphy. 2005. An assessment of the cod ( Gadus morhua) stock in NAFO Divisions 2J3KL in March 2005, 217 p. Research Document 2005/018. Fisheries and Oceans Canada, Communications Branch, 200 Kent Street, Station 13E228, Ottawa, Ontario, Canada. Misarti, N. 2007. Six thousand years of change in the Northeast Pacific: an interdisciplinary view of maritime eco- systems. Ph.D. diss., 214 p. Univ. Alaska Fairbanks, Fairbanks, AK. Nagaoka, L. 2001. Using diversity indices to measure changes in prey choice at the Shang River Mouth site, southern New Zealand. Int. J. Osteoarchaeol. 11:101-111. 2002. Explaining subsistence change in southern New Zealand using foraging theory models. World Archaeol 34:84-102. Orchard, T. 2001. The role of selected fish species in Aleut palaeo- diet. M. S. thesis, 228 p. Dep. Anthropology, LTniv. Victoria, Victoria, British Columbia. 2003. An application of the linear regression technique for determining length and weight of six fish taxa: the role of selected fish species in Aleut paleodiet, 123 p. Archaeopress, Oxford, UK. Pauly, D., V. Christensen, J. Dalsgaard, R. Froese, and F. Torres Jr. 1998. Fishing down marine food webs. Science 279:860- 863. Polovina, J. J. 2005. Climate variation, regime shifts, and implications for sustainable fisheries. Bull. Mar. Sci. 76(2):233- 244. Reedy-Maschner, K. L. 2004. Aleut identity and indigenous commercial fisher- ies. Ph.D. diss., 256 p. Univ. Cambridge, Cambridge, UK. Rochet, M. J. 1998. Short-term effects of fishing on life history traits of fishes. ICES J. Mar. Sci. 55:371-391. Shin, Y.-J., M.-J. Rochet, S. Jennings, J. G. Field, and H. Gis- lason. 2005. Using size-based indicators to evaluate the eco- system effects of fishing. ICES J. Mar. Sci. 62:384- 396. Stuiver, M., and P. J. Reimer 1993. Extended 14C data base and revised CALIB 3.0 14C age calibration program. Radiocarbon 35(1):215- 230. Swain, D. P., A. F. Sinclair, and J. M. Hanson 2007. Evolutionary response to size-selective mortal- ity in an exploited fish population. Proc. Roy. Soc. B 274:1015-1022. Tews, A. M. 2005. Prehistoric eastern Aleut subsistence: 5,000 years of faunal variations. M. S. thesis, 243 p. Idaho State Univ., Pocatello, ID. 394 Fishery Bulletin 106(4) Thompson, G. G., M. W. Dorn, and D. G. Nichol 2006. Assessment of the Pacific cod stock in the Gulf of Alaska. In Stock assessment and fishery evalua- tion report for the groundfish resources of the Gulf of Alaska, p. 147-220. North Pacific Fisheries Manage- ment Council, 605 West 4th, Suite 306, Anchorage, AK 99501-2252. Trippel, E. A. 1995. Age at maturity as a stress indicator in fisher- ies. BioScience 45:759-771. Trites, A. W., V. Christensen, and D. Pauly. 2006. Effects of fisheries on ecosystems: just another top predator? In Top predators in marine ecosystems: Their role in monitoring and management (I. L. Boyd, K. Camphuysen, and S. Wanless, eds.), p. 11-27. Cam- bridge Univ. Press, Cambridge. Trites, A. W., A. J. Miller, H. D. G. Maschner, M. A. Alexander, S. J. Bograd, J. A. Calder, A. Capotondi, K. O. Coyle, E. D. Lorenzo, B. P. Finney, E. J. Gregr, C. E. Grosch, S. R. Hare, G. L. Hunt, J. Jahncke, N. B. Kachel, H.-J. Kim, C. Ladd, N. J. Mantua, C. Marzban, W. Maslowski, R. Mendelssohn, D. J. Neilson, S. R. Okkonen, J. E. Overland, K. L. Reedy-Maschner, T. C. Royer, F. B. Schwing, J. X. L. Wang, and A. J. Winship. 2007. Bottom-up forcing and the decline of Steller sea lions ( Eumetopias jubcitus) in Alaska: assessing the ocean climate hypothesis. Fish. Oceanogr. 16:46-67. Ugan, A., and J. Bright. 2001. Measuring foraging efficiency with archaeological faunas: the relationship between relative abundance indices and foraging returns. J. Arch. Sci. 28:1309- 1321. Willis, K. J„ and H. J. Birks. 2006. What is natural? The need for a long-term perspec- tive in biodiversity conservation. Science 314:1261- 1265. Worm, B., E. B. Barbier, N. Beaumont, J. Duffy, C. Folke, B. Halpern, J. Jackson, H. Lotze, F. Micheli, S. Palumbi, E. Sala, K. Selkoe, J. Stachowicz, and R. Watson. 2006. Impacts of biodiversity loss on ocean ecosystem services. Science 314:787-790. 395 Abstract — Distribution and preva- lence of the phoretic barnacle Xenobal- anus on cetacean species are reported for 22 cetaceans in the eastern tropical Pacific Ocean (21 million km2). Four cetacean species are newly reported hosts for Xenobalanus: Bryde’s whale ( Balaenoptera edeni), long-beaked common dolphin ( Delphinus capen- sis), humpback whale ( Megaptera novaeangliae), and spinner dolphin ( Stenella longirostris). Sightings of Xenobalanus in pelagic waters are reported for the first time, and con- centrations were located within three productive zones: near the Baja Cali- fornia peninsula, the Costa Rica Dome and waters extending west along the 10°N Thermocline Ridge, and near Peru and the Galapagos Archipelago. Greatest prevalence was observed on blue whales (Balaenoptera musculus ) indicating that slow swim speeds are not necessary for effective barnacle settlement. Overall, prevalence and prevalence per sighting were gener- ally lower than previously reported. The number of barnacles present on an individual whale was great- est for killer whales, indicating that Xenobalanus larvae may be patchily distributed. The broad geographic distribution and large number of cetacean hosts, indicate an extremely cosmopolitan distribution. A better understanding of the biology of Xeno- balanus is needed before this species can be used as a biological tag. Manuscript submitted 28 January 2008. Manuscript accepted 17 June 2008. Fish. Bull 106:395-404 (2008). The views and opinions expressed or implied in this article are those of the author and do not necessarily reflect the position of the National Marine Fisheries Service, NOAA. Prevalence of the commensal barnacle Xenobalanus globicipitis on cetacean species in the eastern tropical Pacific Ocean, and a review of global occurrence Emily A. Kane (contact author)1' 2 Paula A. Olson 2 Tim Gerrodette2 Paul C. Fiedler2 Email address for E A. Kane: ekane@tamu.edu 1 Southampton College 239 Montauk Highway Southampton, New York 11968 2 National Marine Fisheries Service, NOAA Southwest Fisheries Science Center 8604 La Jolla Shores Dr. La Jolla, California 92037 Present address for contact author (E. A. Kane): Texas A&M University at Galveston 5007 Avenue U Galveston, Texas 77551 Barnacles of the superfamily Coronu- loidea live as obligate commensals on sea turtles, cetaceans, sirenians, sea snakes, and crustaceans (Newman and Ross, 1976). The monotypic Xeno- balanus globicipitis Steenstrup, 1851 (herein referred to by genus) is spe- cialized for living as a commensal on whales and dolphins (Darwin, 1854). The typical six-plate balanomorph shell is small and is imbedded into the skin of the cetacean host. The membrane supporting the operculum is greatly elongated, so that externally Xenobalanus resembles a pedunculate barnacle. This species is most com- monly observed on the trailing edges of the dorsal fin, pectoral flippers, and tail fluke of the host, although it has been reported in areas such as the ros- trum and the area between the teeth (Samaras, 1989). Xenobalanus does not receive nutrition from its cetacean host and therefore is not considered a parasite. Instead, as a suspension- feeding cirriped, it uses the water flow around swimming cetaceans and ben- efits from being transported by its host (phoresis). This species is highly spe- cialized to live on cetaceans (Seilacher, 2005) and it has been suggested that its hermaphroditic reproduction may be synchronized with that of its host (Dollfus, 1968; Fertl, 2002). A five- to six-month reoccurrence cycle has been reported for Xenobalanus (Van Waerebeek et al., 1993; Orams and Schuetze, 1998), which may indicate that its life span may be of similar length or that occurrence is correlated with seasonal environmental condi- tions. Xenobalanus has been reported on 30 cetacean species worldwide and has a prevalence ranging from 0.5% to 55% of individuals in each sighting. However, intensity is highly variable, and there are some reports of greater than 100 barnacles on a single host (Aznar et al., 2005). We examined the presence of Xeno- balanus on cetaceans in the eastern tropical Pacific Ocean (ETP). Based on photographs taken during research cruises from 1977 through 2003, mean prevalence, mean intensity, and geographic distribution are described for Xenobalanus on 22 host cetacean species. In addition, peer-reviewed lit- erature on this subject is examined, updating a previous summary of the cetacean hosts of this barnacle (Raja- guru and Shantha, 1992). 396 Fishery Bulletin 106(4) Xenobalanus presence or absence for 445 cetacean sightings in the eastern tropical Pacific Ocean (ETP) in 2003 as determined from analysis of identification photographs. Dots (•) indicate cetacean sightings with no Xenobalanus observed; circles (O) indicate sightings with one or more barnacles observed; the solid line indicates the border of the ETP study area. Presence or absence is overlaid on a background of graded shading representing the volume of chlorophyll-a (mg/m3) averaged from September to November 2003. Materials and methods Data collection Cetaceans were photographed during a 2-ship, 4-month research cruise in 2003, covering 26,000 km of tran- sects surveyed for marine mammals (boundaries shown in Figs. 1 and 2). Camera equipment included Canon EOS 10D and D60 digital cameras (Canon USA, Lake Success, NY) with 75-300 mm image-stabilized zoom and 400-mm fixed lenses. Date, latitude and longitude, cetacean species (as identified by trained cruise person- nel), and unique sighting number per cetacean group were recorded with each photograph. In the laboratory, additional data were recorded upon examination of pho- tographs, including the number of usable photographs in the sighting (as described below), number of individual cetaceans identified in the sighting, number of individu- als infested with Xenobalanus, and number of Xenobala- nus present. If barnacles were clumped in such a way as to compromise the accuracy of the count, the maxi- mum number of discernible barnacles was recorded. The resolution of the digital photographs was such that in most cases, individual barnacles were easily identified. In external appearance, Xenobalanus may be confused with the parasitic copepod Pennella balaenoptera (Ev- ans, 1994) or the stalked barnacle Conchoderma virga- tum (Ruppert et al., 2004). However, the much larger P. balaenoptera usually occurs along the flanks, whereas Xenobalanus is generally found along the trailing edges of the dorsal fin, pectoral flippers, and the fluke, as has been described for stranded and live cetaceans. Similar to P. balaenoptera, C. virgatum requires a less specific position for attachment, requiring any hard substrate (such as another barnacle, a tooth, or exposed bone), and C. virgatum is considerably lighter in coloration than Xenobalanus. Digital photographic quality was sufficient for accurate identification of the commensal Xenobala- nus-, no specimens were obtained for direct examination. During the cruise, typically many photographs were taken for each sighting. For our study, usable photo- graphs had 1) to be in focus, 2) to be of sufficient resolu- tion to identify a barnacle if present, and 3) to include at least one cetacean dorsal fin. For large schools, only one photograph per school was used in order to prevent recounting individuals. For small schools, only photo- graphs of animals identifiable as individuals, either from field notes or from unique markings or pigmen- Kane et al. : Prevalence of Xenobalanus globiclpitis on cetacean species in the eastern tropical Pacific Ocean 397 180°W 60°N 30°N 0° 30°S“ 60°S 180°W 150°W 150°W 120°W 120°W 90"W 90 °W 60°W 30°W 60°S _____ _ . XX 60 °W 30°W 30° E 60°E 90° E 120°E 1 50°E 180°E 30° E 60° E 90°E 120°E 150°E 180°E 60°N 30°N 30°S eastern tropical Pacific Ocean (ETP) Figure 2 Documented sightings of Xenobalanus worldwide on various cetacean hosts compiled from lit- erature review, and an outline of the current study area (eastern tropical Pacific Ocean). Ovals indicate the geographic region where barnacles have been reported and their size does not indicate intensity of infestation. Refer to Table 3 for the corresponding citations for each region. tation, were used. Photographs used represent a sub- sample of individuals and did not include every animal observed. Killer whales ( Orcinus orca) were studied in additional detail, by using photographs taken on previous cruises dating back to 1977, and including photographs from the ETP in the California and Mexico killer whale catalog (Black et al., 1997). Data analysis Two measures of prevalence of Xenobalanus were calcu- lated for each cetacean species with at least three sight- ings with usable photographs. Prevalence was calculated for each sighting by dividing the number of individual whales or dolphins with barnacles by the total number of individual cetaceans identified. Mean prevalence and its standard error for each species were calculated from these values. Prevalence per sighting was calculated for each species by dividing the number of sightings with barnacles present by the total number of sightings. Mean barnacle intensity was calculated as the total number of barnacles observed on a host species divided by the number of infested hosts of the same species (Bush et al., 1997). To relate barnacle presence to primary productivity, the presence or absence of Xenobalanus at each cetacean sighting in 2003 was plotted on a map of average surface chlorophyll concentration in the ETP that was based on data provided by the SeaWiFS Project, NASA/Goddard Space Flight Center, and GeoEye. Several statistical tests were performed to determine significant differences among the data. A nonparamet- ric Mann Whitney-G test was used to determine if the rates of prevalence differed between Mysticetes and Odontocetes and a nonparametric Kruskal-Wallace test was used to determine if the prevalence rates differed significantly among species. Chi-square goodness-of-fit test was used to determine if the number of barnacles per killer whale followed a Poisson distribution. After normalization, linear regression was used to determine if the number of barnacles observed was predicted by the total number of animals observed in the sighting. Results Within the ETP over 10,000 photographs of 22 cetacean species and 2510 individuals revealed that 132 individu- als of 14 species were host to Xenobalanus (Table 1). Out of 497 photographed sightings, 445 were determined to be usable in the analysis, and of these sightings, 47 displayed Xenobalanus: 38 odontocete sightings and 9 mysticete sightings. Xenobalanus was not observed on seven cetacean species and on one genus for which the species could not be identified: pygmy killer whale 398 Fishery Bulletin 106(4) Table 1 Summary of photographic data collected for cetaceans in the eastern tropical Pacific Ocean from 1977 through 2003, including the number of usable photographs (“photographs”), number of sightings (“sightings”), number of identifiable individual ceta- ceans (“individuals”), and number of identifiable individuals infested with Xenobalanus (“infested individuals”). * Denotes newly reported hosts of Xenobalanus that were determined from this study. Infested Species Common name Years Photographs Sightings Individuals individuals Balaenoptera edeni* Bryde’s whale 2003 415 30 43 3 Balaenoptera musculus blue whale 2003 513 17 24 9 Balaenoptera physalus fin whale 2003 92 4 6 1 Delphinus capensis* long-beaked common dolphin 2003 228 13 69 2 Delphinus delphis short-beaked common dolphin 2003 1146 48 287 3 Feresa attenuata pygmy killer whale 2003 25 1 11 0 Globicephala macrorhynchus short-finned pilot whale 2003 1551 34 297 9 Grampus griseus Risso’s dolphin 2003 155 12 58 1 Lagenodelphis hosei Fraser’s dolphin 2003 34 1 6 0 Lagenorhynchus obliquidens Pacific white-sided dolphin 2003 5 1 2 0 Lagenorhynchus obscurus dusky dolphin 2003 245 11 68 4 Megaptera novaeangliae* humpback whale 2003 504 12 34 1 Mesoplodon spp. unidentified Mesoplodon 2003 40 2 3 0 Orcinus orca killer whale 1977-2003 1160 49 354 69 Peponocephala electro melon-headed whale 2003 17 1 9 0 Physeter macrocephalus sperm whale 2003 215 9 19 0 Pseudorca crassidens false killer whale 2003 49 2 11 0 Stenella attenuata pantropical spotted dolphin 2003 1281 76 326 4 Stenella coeruleoalba striped dolphin 2003 845 51 319 18 Stenella longirostris* spinner dolphin 2003 770 39 271 3 Steno bredanensis rough-toothed dolphin 2003 118 12 41 0 Tursiops truncatus bottlenose dolphin 2003 600 61 252 5 Total 10,008 486 2510 132 ( Feresa attenuata ); Fraser’s dolphin ( Lagenodelphis hosei)\ Pacific white-sided dolphin ( Lagenorhynchus obliquidens ); melon-headed whale (Peponocephala elec- tro)', sperm whale ( Physeter macrocephalus ); false killer whale ( Pseudorca crassidens), rough-toothed dolphin ( Steno bredanensis ); and three unidentified beaked whale individuals (Mesoplodon spp.) Of these, the barnacle has been previously reported throughout its worldwide range on pygmy killer whale, false killer whale, two species of beaked whales (Mesoplodon spp., Rajaguru and Shantha, 1992), Pacific white-sided dolphin (Dailey and Walker, 1978), and rough-toothed dolphin (Addink and Smeenk, 2001). Four cetacean species seen in the ETP had not previ- ously been reported as hosts of Xenobalanus'. Bryde’s whale ( Balaenoptera edeni ), long-beaked common dol- phin ( Delphinus capensis), humpback whale (Megap- tera novaeangliae), and three forms of spinner dolphin ( Stenella longirostris ): eastern (S. longirostris orienta- lis) and the forms known commonly as whitebelly and southwestern spinner dolphins (Table 2). For Bryde’s whales and humpback whales, the dorsal fin was the only visible appendage, as opposed to the long-beaked common dolphins and spinner dolphins for which pecto- ral flippers and tail flukes were also visible. The hump- back whale individual that displayed a single specimen of Xenobalanus appeared to have a damaged dorsal fin. Prevalence and intensity of the barnacle Blue whales (Balaenoptera musculus ) had the highest mean prevalence of the barnacle, followed by fin whales (B. physalus) and killer whales (Fig. 3A). There was a significant difference in mean prevalence among species (^!o_5o.6, PcO.Ol) and Mysticetes had a higher mean prevalence of the barnacle than Odontocetes (5.1% vs. 0.8%). Standard error was greatest for blue (13.8) and fin whales (16.5); all other species had a standard error less than 1.5. Blue, fin, and killer whales also had the highest prevalence per sighting, and 38% of killer whale sightings had barnacles (Fig. 3B). Prevalence per sight- ing was similar for Mysticetes and Odontocetes (12.7% vs. 14.0%) — a nonsignificant difference (P=0.19). Of the three species most often infested, killer whales repre- Kane et al : Prevalence of Xenobalanus globicipitis on cetacean species In the eastern tropical Pacific Ocean 399 Table 2 Dates and geographic locations (latitude and longitude) for sightings of newly documented cetacean hosts of Xenobalanus in the eastern tropical Pacific Ocean in 2003. Data are the following: total number of individuals of the species photographed (“individ- uals”), total number of these individuals observed with barnacles (“infested individuals”), and barnacle intensity and anatomical location on the host (“intensity and location on the host”). Host Individuals Infested individuals Intensity and location on the host Date Geographic location Balaenoptera edeni Bryde’s whale 64 3 4 on dorsal fin 03 November 2003 09.012°S 079.302°W Delphinus capensis long-beaked common dolphin 69 1 1 on right pectoral flipper 12 August 2003 25.620°N 109.456°W Megaptera novaeangliae humpback whale 34 1 1 on dorsal fin 05 November 2003 06.414°S 081.176°W Stenella longirostris orientalis eastern spinner dolphin 99 1 1 on left pectoral flipper 15 August 2003 21.448°N 108.084°W Stenella longirostris hybrid whitebelly spinner dolphin 91 1 1 on dorsal fin 20 August 2003 08.857°N 145.098°W Stenella longirostris southwestern southwestern spinner dolphin 32 1 2 or more on right pectoral flipper 14 October 2003 05.084°S 097.974°W sented the majority of individuals used in the analysis (14%), whereas blue whales (1%) and fin whales (0.2%) were rarely encountered. The number of barnacles was, therefore, independent of number of individuals observed (R2= 0.00, F=0.10, P=0.08). Xenobalanus was found in coastal as well as offshore waters of the ETP (Fig. 1). All 22 species were repre- sented in offshore sightings and 28% of individuals encountered were seen in waters greater than 600 km from land, and at a maximum distance of 4287 km from land. Of these offshore occurrences, 39 Xenobala- nus were observed on 18 individuals comprising seven species. Xenobalanus was primarily observed in three areas: 1) waters around the Baja California peninsula, 2) the Costa Rica Dome and waters extending west along the 10°N Thermocline Ridge, and 3) waters off Peru and the Galapagos Archipelago. All three areas are known as areas of increased primary productivity within the ETP (Fig. 1, Fiedler et al., 1991; Pennington et al., 2006). For killer whales, which were examined in more de- tail, of the 68 whales infested with 130 barnacles, the mean intensity of infestation was 1.9 barnacles per whale. The greatest numbers of killer whales were pho- tographed in 1998-2003, and these whales also had the greatest intensity of barnacles. This observed increase in intensity was most likely the result of improved pho- tographic techniques. The observed numbers of killer whales with 0, 1, 2, 3, and >3 barnacles were 286, 38, 15, 7, and 8, respectively. This is significantly different from the expected 245, 90, 17, 2, and 0 infested whales, respectively, as predicted by a Poisson distribution with mean 130/354 = 0.367 (%|>300, PcO.00001). The vari- ance (0.941) was much larger than the mean (0.367). Literature review A chart of the worldwide distribution of Xenobalanus was generated from a review of the literature documenting regional occurrences of this genus (Fig. 2). Except for the ETP, Xenobalanus has been reported only within approximately 600 km from land, including the Faroe Islands and the Azores (sites 4 and 8). Figure 2 also demonstrates that Xenobalanus is highly cosmopolitan and has been reported in all oceans, namely in tropical, temperate, and polar waters. The literature review updates a previous review conducted by Rajaguru and Shantha (1992). Eighteen peer-reviewed accounts have been published since that review, including that of the present study (Table 3). Additionally, ten records of Xenobalanus had not been included in Rajaguru and Shantha’s (1992) review. An additional 14 cetacean species are now included: minke whale {Balaenoptera acutorostrata), Bryde’s whale, long- beaked common dolphin, Pacific white-sided dolphin, dusky dolphin (Lagenorhynchus obscurus ), right whale dolphin (Lissodelphis borealis), humpback whale, va- quita ( Phocoena sinus), Burmeister’s porpoise ( Pho - coena spinipinnis), franciscana (Pontoporia blainvillei), clymene dolphin ( Stenella clymene), spinner dolphin, 400 Fishery Bulletin 106(4) 1 ^ % mean prevalence B % prevalence per sighting 5 10 15 20 25 30 35 40 5 10 15 20 25 30 35 40 * • 'Blue Whale (24) Killer Whale (68) • Blue Whale (17) . Killer Whale (354) • Fin Whale (4) — 1 Bryde's Whale (64) • Pilot Whale (34) -•-i Long-beaked Common Dolphin (69) • Dusky Dolphin (11) » Striped Dolphin (319) • Striped Dolphin (51) " Dusky Dolphin (68) • Risso’s Dolphin (12) - Pilot Whale (297) • Humpback Whale (12) • Pantropical Spotted Dolphin (326) • Spinner Dolphin (38) w Humpback Whale (34) • Long-beaked Common Dolphin (13) « Short-beaked Common Dolphin (287) • Short-beaked Common Dolphin (48) ■ Bottlenose Dolphin (284) • Bottlenose Dolphin (70) » Risso’s Dolphin (58) • Pantropical Spotted Dolphin (76) < Spinner Dolphin (271) • Bryde's Whale (38) Figure 3 Prevalence of Xenobalanus on cetaceans in the eastern tropical Pacific Ocean in 2003 as deter- mined from cetacean identification photographs; (A) percent mean prevalence of Xenobalanus ±2 standard errors, and (B) percent prevalence of Xenobalanus per sighting (number of sightings with barnacles/total number of sightings of a species; . The number in parentheses (n) represents the total number of individuals and sightings, respectively, observed for each cetacean species on which Xenobalanus was observed. rough-toothed dolphin, and Indo-Pacific bottlenose dol- phin (Tursiops aduncus). Discussion Prevalence and intensity of the barnacle We describe four new host species for the cetacean-spe- cific phoretic barnacle Xenobalanus and document that the barnacle is present on cetaceans far offshore as well as in coastal areas. The fact that Xenobalanus has now been reported on 34 species of cetaceans in both coastal and offshore waters, from the Arctic to Antarctic, either 1) indicates that the barnacle is extremely cosmopolitan (Newman and Ross, 1976; Spivey, 1981), or 2) may sug- gest that more than one species of the genus Xenobala- nus is involved. Mean prevalence results from this study were lower than those from previous published accounts: 0.2% vs. 4-19% for short-beaked common dolphins ( Delphinus delphis) (Pilleri, 1970; Dailey and Walker, 1978), 0.2% vs. 43-56% for bottlenose dolphins ( Tursiops truncatus) (Di Beneditto and Ramos, 2004; Toth-Brown and Hohn, 2007), and 0.7% vs. 33-43% for striped dolphins ( Stenel - la coeruleoalba) (Pilleri, 1970; Aznar et ah, 2005). Two striped dolphins have been reported with an intensity of more than 100 Xenobalanus (Aznar et al., 2005), but the greatest intensity observed in our study was seven, on killer whales. Although some differences in prevalence are due to previous reports of maximum, rather than mean, rates, the prevalence of Xenobalanus infestation reported in this study is underestimated because not all barnacles present on the animals were visible in our photographs. Only one side of the animal was photographed, and often part of the body was in the water. On the other hand, prevalence reported in many previous studies may have been overestimated when rates were based on mortality events and strandings. Because stranded ani- mals are not usually healthy, the reported rates could represent an abnormal presence of the barnacle, as was observed in Aznar et al. (2005). Differences may also be related to habitat. Moreover, previous reports have Kane et al.: Prevalence of Xenobalanus globicipitis on cetacean species in the eastern tropical Pacific Ocean 401 Table 3 Geographic regions and corresponding citations for each region where Xenobalanus has been documented on cetaceans world- wide. “Circle” refers to the regions encircled in Figure 2. References cited in Rajaguru and Shantha (1992) have been omitted individually, but are included under the citation for Rajaguru and Shantha (1992). Additional data about new hosts determined in the present study are available in Table 2. Circle Geographic region Citation 1 Pacific Northwest United States and Canada Rajaguru and Shantha (1992) 2 Greenland Rajaguru and Shantha (1992) 3 Northern Scandanavian peninsula Rajaguru and Shantha (1992) 4 Feroe Islands Rajaguru and Shantha (1992) 5 Scotland and Shetland Islands Rajaguru and Shantha (1992) 6 Belgium Rajaguru and Shantha (1992) 7 Western Mediterranean and Iberian Peninsula Raga and Carbonell (1985), Rajaguru and Shantha (1992), Aguilar and Raga (1993), Resendes et al. (2002), Aznar et al. (2005) 8 Azores Rajaguru and Shantha (1992) 9 East coast United States and the Bahamas Rajaguru and Shantha (1992), Toth-Brown and Hohn (2007) 10 Gulf of Mexico Spivey (1981), Jefferson et al. (1995) 11 Southern California and Baja Peninsula Dailey and Walker (1978), Brownell et al. (1987), Samaras (1989), This study 12 Pelagic ETP This study 13 Costa Rica Dome, Galapagos, and Peru Van Waerebeek et al. (1990, 1993), Reyes and Van Waerebeek (1995), Palacios et al. (2004), This study 14 Southeast coast of Brazil and Uruguay Brownell (1975), Young (1991), Rajaguru and Shantha (1992), Di Beneditto and Ramos (2000, 2001, 2004) 15 Northwestern coast of Africa Van Bree (1971), Rajaguru and Shantha (1992), Addink and Smeenk (2001) 16 South Africa and Namibia Rajaguru and Shantha (1992) 17 Southern India Rajaguru and Shantha (1992), Karuppiah et al. (2004) 18 Philippines, South China Sea, Hong Kong Parsons et al. (2001) 19 Japan Uchida and Jun (2000), Sakai et al. (2006) 20 East coast of Australia Rajaguru and Shantha (1992), Orams and Schuetze (1998) 21 Mawson and Davis seas, Antarctica Bushev (1990) 22 Riiser-Larsen and Lazarev Seas, Antarctica Bushev (1990) 23 Shetland Islands, northwest Weddell Sea, Antarctica Bushev (1990) 24 Bellinghausen Sea, Antarctica Bushev (1990) primarily been composed of data from coastal areas, whereas our study was focused on pelagic waters. Factors affecting the presence of Xenobalanus Other behavioral and environmental factors may also affect barnacle presence on cetaceans within the ETP. Swimming speed of the host has been shown to corre- late negatively with intensity of the whale lice Isocya- mus delphini (Balbuena and Raga, 1989) and has been hypothesized as an inversely proportional factor in Xeno- balanus settlement (Orams and Schuetze, 1998; Aznar et al., 2005). In our study, blue whales had the greatest mean intensity of Xenobalanus and have been shown to sustain cruising speeds up to 33 km/hr (Yochem and Leatherwood, 1985), indicating that swimming speed may not be a primary factor in host species selection for Xenobalanus. Abrasive breaching and slapping behavior of the host may scour barnacles and inhibit settlement; however, some barnacles appear resistant (Felix et al., 2006; Sakai et al., 2006). In the ETP, deep-diving sperm whales ( Physeter macrocephalus) and beaked whales ( Mesoplodon spp.) were not hosts, indicating that dive depth of the host may limit the settlement of the barnacle on these species. Orams and Schuetze (1998) and Toth- Brown and Hohn (2007) have suggested an environmen- 402 Fishery Bulletin 106(4) tal correlation in the distribution of Xenobalanus that is similar to that observed between primary production and barnacle presence in the ETP. Plankton abundance in oligotrophic areas of the ETP may be below a critical threshold for the filter-feeding barnacles and may thus indirectly limit the presence of Xenobalanus. The intensity of barnacles on killer whales in the ETP was not randomly distributed. There were more whales with no barnacles and with three or more barnacles than would be expected if barnacles settled randomly on killer whales, indicating that if Xenobalanus larvae settle, it is most often in groups of three or more. This aggregated or contagious distribution could occur as a result of: 1) a chemical cue emitted from the host that induces settlement (Nogata and Matsumura, 2005), 2) a chemical cue emitted from conspecifics that induces settlement (Knight- Jones, 1953), which was suggested for Xenobalanus by Aznar et al. (2005), 3) patchily dis- tributed barnacle larvae, or 4) an inability of the host to slough newly settled larvae (Ridgway et al., 1997). The low variance in prevalence and the nonuniform distribution of Xenobalanus sightings within the ETP indicate that most species are equally selected and that barnacle recruitment may be the result of patchily distributed larvae. Xenobalanus has been reported on a wide variety of cetacean hosts, and this apparent lack of specialization could provide insight into evolutionary age of Xenobala- nus. Various species of cyamid whale lice are highly specialized for a particular species of right whale ( Eu - balaena spp., Kaliszewska et al., 2005). Xenobalanus is more of a generalist than whale lice, given its apparent ability to settle on various cetacean hosts, which may indicate that its evolution and relationship with ceta- ceans may be more recent than that of other cetacean commensals, and that its specialization to host species has not yet occurred. Coronulid whale barnacles did not appear in the fossil record until approximately 23 million years ago (Newman and Ross, 1976; Seilacher, 2005), after the appearance of Mysticetes and Odon- tocetes in the fossil record approximately 35 million years ago. However, it is unknown at what point the genus Xenobalanus arose, and presently no data exist on the evolutionary age of cyamids for comparison. In the ETP, Xenobalanus, appearing on almost every ce- tacean species encountered, did not exhibit the degree of host specialization observed in whale lice. With a lack of data on evolutionary age, these findings support only the hypothesis that Xenobalanus is a generalist cetacean barnacle. Biological tags The relationship between commensals and their hosts, which can indicate host movement and host distribu- tional patterns, is often used to make inferences into the biology and ecology of the host. Comparison of internal parasite fauna has helped distinguish stocks, determine stock associations, track large-scale movements, and identify new recruits to populations in many species of fish, elasmobranchs, invertebrates, and marine mam- mals (Williams et ah, 1992). Among marine mammals, intestinal parasites, whale lice, and barnacles have proven useful for tracking migrations of gray whales (Eschrichtius robustus ; Killingley, 1980) and identifying stocks and the social structure of pilot whales (Globi- cephala melas; Balbuena and Raga, 1993), and have been useful for tracking general movement patterns of wide ranging, elusive cetacean populations without the use of expensive tagging equipment. Our results indicate that Xenobalanus, however, would not be useful as a biological tag. Within the ETP, Xe- nobalanus is widely distributed and a single, definitive source or home range was not determined for this spe- cies. However, this is the first study where distribution of Xenobalanus has been systematically examined on a large scale and it is possible that the few offshore obser- vations within the ETP are not representative of global distribution. Although Xenobalanus could not be used as a biological tag to track the movements of cetaceans within the ETP in this study, the potential use of Xe- nobalanus as a biological tag should not be abandoned completely. Increased knowledge of the biology of the barnacle, such as host-selection criteria, environmental tolerance limits, and early life history strategies could provide a finer resolution of the phoretic relationship with cetacean species that would enable the use of Xe- nobalanus as a biological tag in future studies. This and other research on Xenobalanus will form a useful part of the study of cetacean biology and ecology. Acknowledgments Funding for travel and living expenses for E. A. Kane was provided by the Evan Frankel Foundation. We thank J. Barlow, M. Kretzmann, and W. Perrin for reviewing the manuscript before submission, as well as D. Fertl and two anonymous reviewers for their helpful com- ments. We also thank N. Black and the many scientists and researchers who donated photographs, references, and information for the project. Literature cited Addink, M. J., and C. Smeenk. 2001. Opportunistic feeding behavior of rough-toothed dolphins Steno bredanensis off Mauritania. Zool. Verh. (Leiden) 334:37-48. Aguilar, A., and J. A. Raga. 1993. The striped dolphin epizootic in the Mediterranean Sea. Ambio 22:524-528. Aznar, F. J., D. Perdiguero, A. Perez del Olmo, A. Repulles, C. Agusti, and J. A. Raga. 2005. Changes in epizoic crustacean infestations during cetacean die-offs: the mass mortality of Mediterranean striped dolphins Stenella coeruleoalba revisited. Dis. Aquat. Org. 67:239-247. Balhuena, J., and J. A. Raga. 1989. Ecology and host relationships of the whale-louse Kane et al.: Prevalence of Xenobalanus globicipitis on cetacean species in the eastern tropical Pacific Ocean 403 Isocyamus delphini (Amphipoda: Cyamidae) parisitiz- ing long-finned pilot whales ( Globicephala melas ) off the Faroe Islands (Northeast Atlantic). Can. J. Zool. 69:141-145. 1993. Intestinal helminths as indicators of segregation and social structure of pods of long-finned pilot whales ( Globicephala melas ) off the Faroe Islands. Can. J. Zool. 72:443-448. Black, N., A. Schulman-Janiger, R. Ternullo, and M. Guerrero- Ruiz. 1997. Killer whales of California and western Mexico: a catalog of photo-identified individuals. NOAA Tech. Rep. NMFS-SWFSC 247, 174 p. Brownell, R. 1975. Progress report on the biology of the Franciscana, Pontoporia blainvillei, in Uruguayan waters. J. Fish. Res. Board Can. 32:1073-1078. Brownell, R. L., L. T. Findley, O. Vidal, A. Robles, and S. Manzanilla. 1987. External morphology and pigmentation of the vaquita, Phocoena sinus (Cetacea, Mammalia). Mar. Mamm. Sci. 3:22-30. Bush, A. O., K. D. Lafferty, J. M. Lotz, and A. W. Shostak. 1997. Parasitology meets ecology on its own terms: Mar- golis et al. revisited. J. Parasitol. 83:575-583. Bushev, S. 1990. A study of the population structure of the southern minke whale ( Balaenoptera acutorostrata Lacepede) based on morphological and ecological variability. Rep. Int. Whal. Comm. 40:317-324. Dailey, M. D., and W. A. Walker. 1978. Parasitism as a factor (?) in single strandings of southern California cetaceans. J. Parasitol. 64:593-596. Darwin, C. 1854. A monograph on the subclass Cirripedia, II the Balanidae, Verrucidae, etc., 684 p. Royal Soc. Pubis., London, England. Di Beneditto, A. P. M., and R. M. A. Ramos. 2000. Records of the barnacle Xenobalanus globi- cipitis ( Steen str up, 1851) on small cetaceans of Brazil. Biotemas 13:159-165. 2001. Biology and conservation of the franciscana ( Pon- toporia blainvillei) in the north of Rio de Janeiro State, Brazil. J. Cetacean Res. Manag. 3:185-192. 2004. Biology of the marine tucuxi dolphin ( Sotalia flu- viatilis) in south-eastern Brazil. J. Mar. Biol. Assoc. UK 84:1245-1250. Dollfus, R. 1968. Xenobalanus globicipitis Steenstrup (Cirripedia, Thoracica): collected on Tursiops truncatus (Montagu) near the northern coast of Morocco. Bull Inst. Peches Marit. Maroc. 16:55-59. I In French.] Evans, W. E. 1994. Common dolphin, white-bellied porpoise Delphi- nus delphis Linnaeus, 1758. In Handbook of marine mammals, vol. 5: The second book of dolphins and the porpoises (S. H. Ridgway, and R. Harrison, eds.), p. 216-217. Academic Press, San Diego, CA. Felix, F., B. Bearson, and J. Falconi. 2006. Epizoic barnacles removed from the skin of a humpback whale after a period of intense surface activity. Mar. Mamm. Sci. 22:979-984. Fertl, D. 2002. Barnacles. In Encyclopedia of Marine Mammals (W. F. Perrin, B. Wiirsig, and J. G. M. Thewissen, eds.), p. 75-78. Academic Press, San Diego, CA. Fiedler, P. C., V. Philbrick, and F. P. Chavez. 1991. Oceanic upwelling and productivity in the eastern tropical Pacific. Limnol. Oceanogr. 36:1834-1850. Jefferson, T. A., D. Odell, and K. Prunier. 1995. Notes on the biology of the clymene dolphin ( Stenella clymene) in the northern Gulf of Mexico. Mar. Mamm. Sci. 11:564-573. Kaliszewska, Z. A., J. Seger, V. J. Rowntree, S. G. Barco, R. Benegas, P. B. Best, M. W. Brown, R. L. Brownell, A. Carribero, R. Harcourt, A. R. Knowlton, K. Marshall- Tilas, N. J. Patenaude, M. Rivarola, C. M. Schaeff, M. Sironi, W. A. Smith, and T. K. Yamada. 2005. Population histories of right whales (Cetacea: Eubalaena) inferred from mitochondrial sequence diver- sities and divergences of their whale lice (Amphipoda: Cyamus). Mol. Ecol. 14:3439-3456. Karuppiah, S., A. Subramanian, and P. Obbard. 2004. The barnacle Xenobalanus globicipitis (Cirripedia, Coronulidae) attached to the bottlenose dolphin Tursiops truncatus (Mammalia, Cetacea) on the southeastern coast of India. Crustaceana 7:879-882. Killingley, J. S. 1980. Migrations of California gray whales tracked by oxygen-18 variations in their epizoic barnacles. Science 207:759-760. Knight- Jones, E. W. 1953. Laboratory experiments on gregariousness during settling in Balanus balanoides and other barnacles. J. Exp. Biol. 30:584-598. Newman, W. A., and A. Ross. 1976. Revision of the Balanomorph barnacles; including a catalog of species. S. Diego Soc. Nat. Hist. Mem. 9:1-108. Nogata, Y., and K. Matsumura. 2005. Larval development and settlement of a whale barnacle. Biol. Lett. 2:92-93. Orams, M. B., and C. Schuetze. 1998. Seasonal and age/size-related occurrence of a bar- nacle (Xenobalanus globicipitis) on bottlenose dolphins (Tursiops truncatus). Mar. Mamm. Sci. 14:186-189. Palacios, D. M., S. K. Salazar, and D. Day. 2004. Cetacean remains and strandings in the Galapa- gos Islands, 1923-2003. Lat. Am. J. Aquat. Mamm. 3:127-150. Parsons, E. C. M., R. M. Overstreet, and T. A. Jefferson. 2001. Parasites from Indo-Pacific hump-backed dolphins (Sousa chinensis) and finless porpoises (Neophocaena phocaenoides) stranded in Hong Kong. Vet. Rec. 148:776-780. Pennington, J. T., K. L. Mahoney, V. S. Kuwahara, D. D. Kolber, R. Calienes, and F. P. Chavez. 2006. Primary production in the eastern tropical Pacific: a review. Prog. Oceanogr. 69:285-317. Pilleri, G. 1970. Xenobalanus globicipitis Steenstrup on Delphinus delphus, Stenella styx, Tursiops truncatus in the western Mediterranean. In Investigations on Cetacea, vol. 2 (G. Pilleri, ed.), p. 248-249. Waldau-Berne, Berne, Switzerland. Raga, J. A., and E. Carbonell. 1985. New dates about parasites on Stenella coeruleoalba (Meyen, 1833) (Cetacea: Delphinidae) in the western Mediterranean Sea. In Investigations on Cetacea, vol. 17 (G. Pilleri, ed.), p. 207-213. Waldau-Berne, Berne, Switzerland. 404 Fishery Bulletin 106(4) Rajaguru, A., and G. Shantha. 1992. Association between the sessile barnacle Xeno- balanus globicipitis (Coronulidae) and the bottlenose dolphin Tursiops truncatus (Delphinidae) from the Bay of Bengal, India, with a summary of previous records from cetaceans. Fish. Bull. 90:197-202. Resendes, A., C. Juan-Salles, S. Almeria, N. Majo, M. Domingo, and J. Dubey. 2002. Hepatic sarcocystosis in a striped dolphin ( Stenella coeruleoalba ) from the Spanish Mediterranean coast. J. Parasitol. 88:206—209. Reyes, J. C., and K. Van Waerebeek. 1995. Aspects of the biology of the Burmeister’s por- poise from Peru. Rep. Int. Whaling Comm. Spec. Issue 16:349-364. Ridgway, S. H., E. Lindner, K. A. Mahoney, and W. A. Newman. 1997. Gray whale barnacles Cryptolepas rhachianecti infest white whales, Delphinapterus leucas, housed in San Diego Bay. Bull. Mar. Sci. 61:377-385. Ruppert, E., R. Fox, and R. Barnes. 2004. Invertebrate zoology: a functional evolutionary approach, 963 p. Brooks/Cole, Belmont, CA. Sakai, M., T. Hishii, S. Takeda, and S. Kohshima. 2006. Flipper rubbing behaviors in wild bottlenose dol- phins (Tursiops aduncus). Mar. Mamm. Sci. 22:966- 978. Samaras, W. 1989. New host record for the barnacle Cryptolepas rha- chianecti Dali, 1872 (Balanomorpha, Coronulidae). Mar. Mamm. Sci. 5:84-87. Seilacher, A. 2005. Whale barnacles: exaptational access to a forbid- den paradise. Paleobiology 31:27-35. Spivey, H. R. 1981. Origins, distribution, and zoogeographic affinities of the Cirripedia (Crustacea) of the Gulf of Mexico. J. Biogeogr. 8:153-176. Toth-Brown, J., and A. A. Hohn. 2007. Occurrence of the barnacle, Xenobalanus globi- cipitis, on coastal bottlenose dolphins (Tursiops trunca- tus) in New Jersey. Crustaceana 80:1271-1279. Uchida, A., and A. Jun. 2000. The ectoparasites and endoparasites in the minke whale, Balaenoptera acutorostrata, from the western north Pacific Ocean. J. Jap. Vet. Med. Assoc. 53:85- 88. Van Bree, P. 1971. The rabbit-eared barnacle, Conchoderma auritum, on the teeth of the dolphin Stenella frontalis. Rep. J. Mamm. 36:316-317. [In German.] Van Waerebeek, K., J. C. Reyes, and J. Alfaro. 1993. Helminth parasites and phoronts of dusky dol- phins Lagenorhynchus obscurus (Gray, 1828) from Peru. Aquat. Mamm. 19:159-169. Van Waerebeek, K., J. C. Reyes, A. J. Read, and J. S. McKinnon. 1990. Preliminary observations of bottlenose dolphins from the Pacific coast of South America. In The bottle- nose dolphin (S. Leatherwood, and R. Reeves, eds.), p. 143-154. Academic Press, San Diego, CA. Williams, H. H., K. MacKenzie, and A. M. McCarthy. 1992. Parasites as biological indicators of the population biology, migrations, diet, and phylogenetics of fish. Rev. Fish Biol. Fish. 2:144-176. Yochem, P. K., and S. Leatherwood. 1985. Blue Whale Balaenoptera musculus (Linnaeus, 1758). In Handbook of marine mammals, vol. 3: The sirenians and baleen whales (S. H. Ridgway, and R. Harrison, eds.), p. 193-240. Academic Press, San Diego, CA. Young, P. S. 1991. The superfamily Coronuloidea Leach (Cirripe- dia, Balanomorpha) from the Brazilian coast, with redescription of Stomatolepas species. Crustaceana 61:190-212. 405 Onshore-offshore distribution and abundance of tuna larvae (Pisces: Scombridae: Thunnini) in near-reef waters of the Coral Sea Email address for A M. Fowler: ashley.m.fowler@student.uts.edu.au 1 Fisheries and Marine Environmental Research Laboratory School of Biological, Earth, and Environmental Sciences The University of New South Wales Kensington, New South Wales, 2052, Australia Present address for A. M. Fowler: Department of Environmental Sciences University of Technology, P. O Box 123 Broadway Sydney, New South Wales 2007 Australia 2 Ichthyology, Australian Museum 6 College Street Sydney, New South Wales, 2010, Australia Abstract — The on-offshore distri- butions of tuna larvae in near-reef waters of the Coral Sea, near Lizard Island (14°30'S, 145°27'E), Australia, were investigated during four cruises from November 1984 to February 1985 to test the hypothesis that larvae of these oceanic fishes are found in highest abundance near coral reefs. Oblique bongo net tows were made in five on-offshore blocks in the Coral Sea, ranging from 0-18.5 km offshore of the outer reefs of the Great Bar- rier Reef, as well as inside the Great Barrier Reef Lagoon. The smallest individuals (<3.2 mm SL) of the genus Thunnus could not be identified to spe- cies, and are referred to as Thunnus spp. We found species-specific distri- butional patterns. Thunnus spp. and T. alalunga (albacore) larvae were most abundant (up to 68 larvae/100 m2) in near-reef (0-5.5 km offshore) waters, whereas Katsuwonus pelamis (skipjack tuna) larvae increased in abundance in the offshore direction (up to 228 larvae/100 m2, 11.1-18.5 km offshore). Larvae of T. albacares (yellowfin tuna) and Euthynnus affinis (kawakawa) were relatively rare throughout the study region, and the patterns of their distributions were inconclusive. Few larvae of any tuna species were found in the lagoon. Size-frequency distribu- tions revealed a greater proportion of small larvae inshore compared to off- shore for K. pelamis and T. albacares. The absence of significant differences in size-frequency distributions for other species and during the other cruises was most likely due to the low numbers of larvae. Larval dis- tributions probably resulted from a combination of patterns of spawning and vertical distribution, combined with wind-driven onshore advection and downwelling on the seaward side of the outer reefs. Manuscript submitted 8 January 2008. Manuscript accepted 23 June 2008. Fish. Bull. 106:405-416 (2008). The views and opinions expressed or implied in this article are those of the author and do not necessarily reflect the position of the National Marine Fisheries Service, NOAA. Ashley M. Fowler (contact author)’- 2 Jeffrey M. Leis2 lain M. Suthers’ Large-scale (100s km) distributions of tuna larvae (family Scombridae), par- ticularly of the commercially impor- tant genera Thunnus and Katsuwonus , have been extensively investigated because of the need to identify spawn- ing locations and the possibility of estimating spawning stock biomass from surveys of larvae (Strasburg, 1960; Richards, 1976; Scott et al., 1993). Despite this effort, and the apparent abundance and fecundity of T. albacares (yellowfin tuna) and K. pelamis (skipjack tuna) in the western and central Pacific Ocean, relatively low mean concentrations (2-4 larvae/ 100 m3, Leis et al., 1991) and abun- dances of tuna larvae (2-8 larvae/10 m2, Leis et al., 1991) have been found during most sampling programs. Low numbers of tuna larvae in any one site or region may simply reflect a broadly distributed, but sparse, spawning pattern of adults, as indicated by the wide geographical ranges of T. alba- cares and K. pelamis larvae. Other possible explanations, however, are that previous sampling scales of 100s km between samples were too coarse to account for spatial variability in abundances (Davis et al., 1990a), and that tuna larvae are more abundant in undersampled near-reef areas (Leis et al., 1991). Nearly all sampling of tuna lar- vae has been conducted in the open ocean because of the oceanic distri- butions of adults; however relatively high concentrations of tuna larvae have been found in previously under- sampled tropical near-reef (<5 km offshore) locations in three studies (Miller, 1979; Leis et al., 1991; Boe- hlert and Mundy, 1994). Concentra- tions of larval T. albacares were up to two orders of magnitude greater within 2 km of leeward (west) Oahu Island, Hawaii, than published con- centrations in surrounding oceanic waters (Miller, 1979), and Thunnus spp. and K. pelamis larvae were up to 100 times more concentrated within 200 m of coral reefs in French Polyne- sia than in oceanic waters of the cen- tral Pacific Ocean (Leis et al., 1991). Because adjacent oceanic locations were not simultaneously sampled in these near-reef studies, a direct com- parison of these two habitats was not possible. Furthermore, comparisons between the near-reef values of lar- val tuna concentration determined by Miller (1979) and Leis et al. (1991), and those determined for oceanic hab- itats in the central Pacific Ocean in other studies, may also be confounded because of differences in sampling methods employed between the two 406 Fishery Bulletin 106(4) habitats. Near-reef concentrations were determined by using short (<15 min.) horizontal tows at or near the surface (Miller, 1979; Leis et ah, 1991), whereas oceanic sampling in the region has generally involved tows of longer duration (>30 min.), with nets towed either horizontally at the surface or obliquely from depths reaching 200 m (Matsumoto, 1958; Nakamura and Mat- sumoto, 1966). Because distributions of tuna larvae are vertically stratified, and Thunnus spp. larvae are rarely found at depths greater than 50 m (Davis et ah, 1990b; Boehlert and Mundy, 1994), oblique tows to depths in excess of 100 m most likely underestimate concentra- tions in shallower strata. Because of this situation, it is appropriate to use the measure of abundance (i.e., the number of larvae below a standard surface area of water) rather than concentration (numbers of larvae per standard volume) to compare numbers of larvae among sites where tows were taken to different depths. Abundances of tuna larvae were determined at both near-reef (2 km offshore) and oceanic sites (~30 km off- shore) around the Hawaiian island of Oahu by Boehlert and Mundy (1994) by oblique sampling to 200 m depth. The abundance of Thunnus spp. larvae decreased with increasing distance offshore, but only on the leeward side of Oahu Island. Interestingly, the abundance of K. pelamis larvae actually increased with distance from the reef on both sides of the island (Boehlert and Mundy, 1994), indicating the possibility of important taxa-spe- cific differences in near-reef distributions of tuna larvae. The possibility that tuna larvae are generally con- centrated near islands and reefs, and not in the open ocean, has important implications for the protection of these vulnerable life stages. Large numbers of tuna lar- vae near shore have been found in only two regions in the central Pacific Ocean, and only near oceanic islands; therefore further investigation of the generality of this phenomenon is required. Considering that patches of highly concentrated (10,945 larvae/500 m3) T. maccoyii (southern bluefin tuna) larvae in the North East Indian Ocean were only 5-15 km in diameter (Davis et al., 1990a), further investigation of fine-scale patterns of larval tuna distribution is also warranted. The aim of the present study was to investigate the near-reef abundance and on-offshore distributions of tuna larvae in the Coral Sea, near the Great Barrier Reef, Australia, over a fine (1-10 km) scale. Although the sampling design was originally intended for inves- tigation of distributions of reef fish larvae (Leis, 1986; Leis and Reader, 1991), the sampling scale and grada- tion of habitat from near-reef to oceanic in the offshore direction also made it appropriate for investigating the near-reef distributions of tuna larvae. Materials and methods Study area and experimental design Larval fish samples were collected in an area of the Coral Sea between Lizard Island (14°30'S, 145°27'E) and 19 km seaward of the outer ribbon reefs of the Great Barrier Reef, Australia (Fig. 1). Lizard Island is situated approximately halfway across the Great Barrier Reef Lagoon (hereafter “lagoon”) where water depths range from 25 to 40 m. The outer reefs lie on the continental shelf break, beyond which depth increases rapidly reach- ing 2000 m within 12 km. There is an abrupt change from shallow, protected waters of the lagoon to oceanic conditions in the offshore direction. Winds were usu- ally from the E to SE during the study period (common for this region); therefore the near-reef waters that we sampled were on the windward side of the outer reefs. Four cruises were conducted to investigate the hori- zontal distribution of fish larvae: 1) 2-5 November 1984 (early November cruise), 2) 17 and 20-22 November 1984 (late November cruise), 3) 30 January-2 Febru- ary 1985 (early February cruise), and 4) 9-13 February 1985 (late February cruise). On each cruise, six samples were taken in each of five on-offshore blocks defined by distance (nautical miles, nmi) from the outer reef crest: A) 0-0.25 nmi (0-0.46 km), B) 0.25-1.0 nmi (0.46-1.85 km), C) 1. 0-3.0 nmi (1.85-5.56 km), D) 3. 0-6.0 nmi (5.56-11.1 km), and E) 6.0-10.0 nmi (11.1-18.5 km, Fig. 1). On each cruise, six samples were also taken in the lagoon between Lizard Island and the outer reefs (Fig. 1). Samples were taken over four consecutive days on each cruise. Offshore transects were planned to be conducted over three days, and two samples to be taken in each block at randomly chosen distances from the reef on each day. All samples in the lagoon were taken on the same day. Transects were centered on a different reef each day and were started from opposite ends on alternate days. Distance from the reef was determined by radar reflection off the waves breaking on the outer reef crest. Because of this method of measurement, ac- tual distance from the reef varied by approximately 100 m depending on the tide and sea state. Because of bad weather on the second cruise, two days elapsed between the sampling of lagoon and offshore waters, and only two samples could be taken in block A. Offshore tran- sects were conducted over four days during the fourth cruise because of mechanical problems. Sampling procedure Quantitative, double-oblique plankton tows were made from a 14-m catamaran with a bongo net (cylinder-cone mesh design) with 0.85 m mouth diameter and 0.5-mm mesh. The net was towed at approximately 1 m/s and was fitted with both a calibrated mechanical flowme- ter and a calibrated mechanical depth and distance recorder. Tows usually filtered 1000-2000 m3 of water with a mean volume (and standard deviation) of 1554 (585), 1644 (629), 1637 (306), and 1348 (278) m3 for each cruise, respectively. All tows were completed during daylight, between one hour after sunrise and one hour before sunset. Tows were taken to a target depth of 200 m on the first cruise and to 120 m thereafter, except in the lagoon and block A where they were taken as close to the bottom as considered safe. The net hit bottom Fowler et al.: Distribution and abundance of tuna larvae in near-reef waters of the Coral Sea 407 Map of the study area in the Great Barrier Reef Lagoon and Coral Sea, near Lizard Island, Australia. Lagoon samples were taken between Lizard Island and the outer barrier reefs, within the boundaries indicated by dashed lines. Offshore samples were taken in five blocks, A-E, in the Coral Sea. The outer reefs are 1) Day, 2) Carter, 3) Yonge, 4) No Name, and 5) Number 10 Ribbon. Map adapted from Leis et al. (1987). occasionally in block A because of great variation in water depth in this area. Samples were fixed in formalin (5-10% in seawater) in the field. Laboratory procedure Larvae from both the port and starboard sides of the bongo net were sorted, except for those in the lagoon samples from the first and second cruises where the catch from only one randomly chosen side was sorted because of high plankton volumes. Samples from block D were not sorted because of funding cuts to the research program. Larvae were removed with the aid of a dis- secting microscope and transferred to 70% ethanol for storage. Tuna larvae (family Scombridae) were identi- fied to species, when possible, by using the descrip- tions of Fritzsche (1978) and Nishikawa and Rimmer (1987). Larvae of T. alhacares and T. alalunga (albacore) <3.2 mm standard length (SL) could not be separated and were identified as Thunnus spp. larvae. For larger larvae, Richards et al. (1990) advocate also using an osteological character, rather than relying solely on pigment, when identifying Thunnus larvae to species. However, their study was primarily concerned with T. atlanticus (blackfin tuna), a species that is not found in our study area. Further, the osteological character in T. alalunga and T. alhacares seems to vary about as much as does the pigment character, and the former cannot be used in specimens <6 mm SL (Richards et al., 1990). Therefore, we relied on the pigment character to separate our T. alalunga and T. alhacares larvae >3.2 mm SL. We note, however, that one-third of the Thunnus larvae >6 mm SL that we cleared and stained had pigment inconsistent with the osteological characters listed in Table 1 of Richards et al. (1990). Therefore, according to the criteria of Richards et al. (1990, Table 1), at least 67% of our Thunnus larvae are correctly identified to species, and between 0 and 33% may be misidentified to species. Given the variability in osteological features noted by Richards et al. (1990, their Tables 2 and 3), we cannot be more precise about this “uncertain” 33%. Larvae of the genus Auxis cannot currently be identified to species (Nishikawa and Rimmer, 1987), and adults of both A. thazard (frigate tuna) and A. rochei (bullet tuna) inhabit the region of the study area (Collette and Nauen, 1983). Auxis larvae were therefore all identified as Auxis spp. Larvae of Auxis spp. and Euthynnus affinis (kawakawa) <2.3 mm SL could not be separated and were identified as Auxis-Euthynnus larvae. Notochord length and SL were measured to the nearest 0.1 mm for preflexion and postflexion larvae, respectively, by using a calibrated ocular micrometer. No correction was made for shrinkage of the larvae. Statistical analyses Abundances (no. of larvae/100 m2) were calculated by 1) calculating larval concentration (larvae/m3), 2) multiplying concentration by the depth (m) sampled, and 3) multiplying the result by 100 to obtain appro- 408 Fishery Bulletin 106(4) Table 1 Species composition of tuna larvae (family Scombridae) caught during four cruises in the Coral Sea, near the Great Barrier Reef, between November 1984 and February 1985. Values are numbers of larvae caught. Taxa are ordered to allow comparison of larval numbers among groups comprising the same genera. Taxon Early November cruise Late November cruise Early February cruise Late February cruise Species total Proportion of total catch (%) Katsuwonus pelamis 55 89 109 94 347 34 Thunnus spp. 3 24 116 29 172 17 Thunnus albacares 11 19 49 16 95 9 Thunnus alalunga 69 62 26 14 171 17 Auxis-Euthynnus 13 2 72 9 96 10 Euthynnus affinis 49 4 54 1 108 11 Auxis spp. 5 0 10 7 22 2 Total 205 200 436 170 1011 priately scaled values. Abundance values incorporate a depth component, and are therefore appropriate for comparing oblique samples taken down to different depths. Statistical analyses of on-offshore patterns of distribution of tuna larvae were performed on natu- ral log-transformed abundances, following inspection of the data for normality and heterogeneity of vari- ance. A count of 1 was added to all data-points before transformation, to allow transformation of zero values. Preflexion and postflexion larvae were combined for on-offshore analyses. For each taxon, analysis was done only for cruises with an average larval abundance >1 larvae/100 m2 to avoid problems associated with numerous zeros. Lagoon samples were not included in calculation of the cruise average because few larvae of any tuna taxon were caught in the lagoon. Only K. pelamis larvae were sufficiently abundant for analysis on all four cruises, and their abundance among on- offshore blocks (including the lagoon) and cruises was compared by using a two-factor analysis of variance (ANOVA). For other taxa, abundance among blocks was analyzed with a one-factor ANOVA for each cruise with larval abundance >1 larvae/100m2. When ANOVA tests yielded significant (P<0.05) results, pairwise dif- ferences between blocks were analyzed using Tukey’s test. Auxis spp. larvae were not sufficiently abundant on any cruise to allow for statistical analysis. Size-frequency data from blocks A and B (inshore zone, 0-1.85 km from the reef) were pooled and com- pared with size-frequency data pooled from blocks C and E (offshore zone, >1.85 km from the reef) using the Kolmogorov-Smirnov (K-S) test. Data were pooled to increase n , because few taxa had sufficient numbers of larvae in each block to provide adequate statistical power. K-S tests were done on data from individual cruises. Larvae from the lagoon were excluded from size-frequency analysis because of low abundance. The significance level used for all statistical tests was 0.05. Not enough E. affinis larvae were caught in both zones on any cruise to allow a statistical comparison of size distributions. Results Species composition and abundance Over 1000 tuna larvae were caught, comprising at least five species and four genera (Table 1). Larvae of K. pelamis were the most abundant, making up over one third of all tuna larvae caught. Numerous small (<3.2 mm SL) Thunnus spp. larvae were caught on the early February cruise, coinciding with a peak in abundance of less common T. albacares larvae. In contrast, T. alalunga larvae were most abundant on the November cruises, and only 26 individuals were caught on the early February cruise. It is, therefore, likely that most of the Thunnus spp. larvae were T. albacares. Larvae of other Thunnus species (e.g., T. obesus [bigeye tuna] or T. tonggol llongtail tuna]) were not caught and would have been distinguishable from T. albacares and T. alalunga, even at small (<3.2 mm SL) sizes (Fritzsche, 1978). Euthynnus affinis larvae were most common on the early November and early February cruises, and Auxis-Euthynnus larvae were most abundant on the early February cruise. Only 22 Auxis spp. larvae were caught during the study, therefore Auxis-Euthynnus larvae were most likely primarily E. affinis larvae. On-offshore distribution Tuna larvae generally had near-reef distributions, as greatest abundances usually occurred within 5.6 km of the outer reefs of the Great Barrier Reef in the Coral Sea. Patterns of on-offshore distribution differed among taxa, however. Species of Thunnus had the most consistent near- reef distribution among the taxa of tuna larvae, with Fowler et al.: Distribution and abundance of tuna larvae in near-reef waters of the Coral Sea 409 highest abundances found within 5.6 km of the outer reefs on most cruises with sufficient numbers of lar- vae for statistical analysis. Thunnus spp. larvae were more abundant in block C than in either the lagoon or block E on the late November cruise (Tukey’s test, P<0.01, Fig. 2) (blocks A and B being intermediate), and more abundant in blocks A and B than in the three other blocks, which did not differ, on the early Febru- ary cruise (Tukey’s test, P<0.03, Fig. 2). Thunnus spp. larvae showed a similar inshore distribution on the late February cruise; however the only significant difference was a greater abundance of larvae in block B than in the lagoon (Tukey’s Test, P<0.03, Fig. 2). Thunnus alalunga larvae were more abundant in blocks B and C than in the lagoon on the early November cruise (Tukey’s test, P<0.02, Fig. 3) (the the other blocks be- ing intermediate) and more abundant in block A than in either the lagoon or block E on the early February cruise (Tukey’s test, P<0.04, Fig. 3). A similar pat- tern of abundance of T. alalunga larvae occurred on the late November cruise; however differences among blocks were not quite significant (Tukey’s test, P=0.051, Fig. 3). Thunnus albacares larvae were most abundant in block A on the early February cruise; however there were no significant differences among blocks (ANOVA, P=0.06, Fig. 4). Distributions of E. affinis larvae and Auxis-Euthynnus larvae were similar to that of T. alalunga, and appeared to be most abundant within 5.6 km of the outer reefs. Differences in larval abundance among blocks were not, however, significant for the early November (ANOVA, P=0.14, Fig. 5) and early February (ANOVA P=0.11, Fig. 5) cruises analyzed for E. affinis, or the early Feb- ruary (ANOVA, P=0.06, Fig. 5) cruise analyzed for Auxis-Euthynnus . In contrast to other tuna taxa, the abundance of K. pelamis larvae increased in the offshore direction. De- spite a significant interaction among blocks and cruises (ANOVA, P= 0.004), K. pelamis larvae were more abun- dant in block E than in blocks A, B, and the lagoon on the early November cruise (Tukey’s test, P<0.02, Fig. 6), more abundant in blocks C and E than in the lagoon on the early February cruise (Tukey’s test, P<0.002, Fig. 6), and more abundant in block E than the lagoon on the late February cruise (Tukey’s test, P<0.05, Fig. 6). The statistical interaction among blocks and cruises was most likely caused by a relatively great abundance (mean 12.1 larvae/100 m2) of K. pelamis larvae in the lagoon on the late November cruise, compared with abundance in the lagoon on the early November, early February, and late February cruises (means 0.6, 0.0, and 3.5 larvae/100 m2, respectively, Fig. 6). Apart from K. pelamis and T. alalunga larvae on the late November cruise, few larvae of any tuna taxon were found in the lagoon. Size-frequency distribution Tuna larvae ranged in size from 1.7 to 15 mm SL; however most taxa had a strong size mode at 2-4 mm 100 10 _ 1 + 100 E o o 0) 5 10 03 0) O c 05 TD § 1 < 100 10 1 0 4 8 12 16 20 Distance from outer reefs (km) Figure 2 Mean abundance (larvae/100 m2+l [± standard error]) of Thunnus spp. larvae with distance from the outer reefs of the Great Barrier Reef during the late November, early February, and late February cruises in the Coral Sea. Data points indicate the midpoint of sampling blocks (A-E) by distance, except for the Great Barrier Reef Lagoon (L on the x axis), the display of which is categorical and does not reflect the true distance from the other blocks. They axis is log10 scale. The x axis shows the width of each sampling block (km). The hatched area on the x axis indicates the position of the outer reef area of the Great Barrier Reef. Within a cruise, if data points share a lowercase letter, they were not significantly different from each other according to Tukey’s post hoc test. Abun- dance data was not obtained for block D. SL. For all species, larvae were of similar size on each cruise, which were at least seven days apart. Differences in size-frequency distributions of lar- vae between inshore and offshore zones of the Coral Sea were taxa-specific, however for those tuna taxa where significant differences were detected, there was a greater proportion of small larvae in the inshore zone. A greater proportion of small (2-3.5 mm SL) K. pela- mis larvae was found in the inshore zone, compared with the offshore zone, on both the early February 410 Fishery Bulletin 106(4) 0 4 8 12 16 20 Distance from outer reefs (km) Figure 3 Mean abundance (larvae/100 m2+l [± standard error]) of Thunnus alalunga (albacore) larvae with distance from the outer reefs of the Great Barrier Reef during the early November, late November, and early February cruises in the Coral Sea. Data points indicate the midpoint of sampling blocks (A-E) by distance, except for the Great Barrier Reef Lagoon (L on the x axis) the display of which is categorical and does not reflect the true distance from the other blocks. The y axis is log10 scale. The x axis shows the width of each sampling block (km). The hatched area on the x axis indicates the position of the outer reef area of the Great Barrier Reef. Within a cruise, if data points share a lowercase letter, they were not signifi- cantly different from each other according to Tukey’s post hoc test. Abundance data were not obtained for block D. (K-S test, P<0.01, Fig. 7A) and late February (K-S test, P<0.02, Fig. 7B) cruises. A similar pattern was found for the late November cruise, with the result approach- ing significance (K-S test, 0.050.2, Fig. 9A) or late November (K-S test, P> 0.2, Fig. 9B) cruises. There was also no significant difference in the size of Auxis-Euthynnus larvae between the inshore and offshore zones on the early February cruise (K-S Test, P>0.2, Fig. 9C). Discussion Mean near-reef (<4 km offshore) abundances of tuna larvae in the Coral Sea (range 20-120 larvae/100 m2) were similar to those found around the Hawaiian island of Oahu (-3-80 larvae/100 m2, Boehlert and Mundy, 1994) and similar to estimates of near-reef abundance from French Polynesia (45-75 larvae/100 m2, Leis et ah, 1991). Although these values were not much greater Fowler et al.: Distribution and abundance of tuna larvae in near-reef waters of the Coral Sea 411 0 4 8 12 16 20 Distance from outer reefs (km) Figure 5 Mean abundance (larvae/100 m2 + l T± standard error]) of larvae with distance from the outer reefs of the Great Barrier Reef for Euthynnus affinis (kawakawa, full line) larvae during the early November and early February cruises, and for Auxis-Euthynnus (dashed line) larvae during the early February cruise, in the Coral Sea. Data points indicate the midpoint of sampling blocks (A-E) by distance, except for the Great Barrier Reef Lagoon the display of which (L on the x axis) is categorical and does not reflect the true distance from the other blocks. The y axis is log10 scale. The x axis shows the width of each sampling block (km). The hatched area on the x axis indicates the position of the outer reef area of the Great Barrier Reef. No significant differences were found among blocks on either the early November (ANOVA, P=0.14) or early February (ANOVA, P=0.11) cruises for E. affinis larvae, or on the early February cruise (ANOVA, P=0.06) for Auxis-Euthynnus larvae. Auxis-Euthynnus larvae were not sufficiently abundant on the first cruise for statistical analysis. Abundance data were not obtained for block D. than those determined for oceanic sites elsewhere (24-80 larvae/100 m2, Strasburg, 1960; Nakamura and Matsu- moto, 1966), the larvae of most tuna species were more abundant within 5.6 km of the Great Barrier Reef than further offshore in the Coral Sea. This pattern was consistent among cruises for particular taxa, indicating that near-reef larval distributions persist over seasonal time scales. Larvae of the genus Thunnus may generally be more abundant in near-reef waters than farther offshore, because in all studies of tuna larvae in near-reef wa- ters consistently high concentrations or abundances of Figure 6 Mean abundance (larvae/100 m2+l [± standard error]) of Katsuwonus pelamis (skipjack tuna) larvae with distance from the outer reefs of the Great Barrier Reef during four cruises in the Coral Sea. Data points indicate the midpoint of sampling blocks (A-E) by distance, except for the Great Barrier Reef Lagoon the display of which (L on the x axis) is categorical and does not reflect the true distance from the other blocks. The y axis is log10 scale. The x axis shows the width of each sampling block (km). The hatched area on the x axis indicates the position of the outer reef area of the Great Barrier Reef. Within a cruise, if data points share a lowercase letter, they were not significantly different from each other according to Tukey’s post hoc test. Abundance data were not obtained for block D. Thunnus larvae have been found there. Greatest abun- dances of Thunnus larvae were often found within 2 km of the outer Great Barrier Reef in the present study, 412 Fishery Bulletin 106(4) Figure 7 Size-frequency distributions for Katsuwonus pelamis (skipjack tuna) larvae from inshore and offshore zones during the (A) early February, (B) late February, and (C) late November cruises in the Coral Sea. Standard length (mm) of larvae was measured to the nearest 0.1 mm. P-values refer to the significance of Kolmogorov-Smirnov (K-S) tests conducted between the inshore and offshore zones within a cruise. and, as with our findings, small (<3.0 mm SL) Thun- nus spp. larvae were -10 times more abundant 1.8 km offshore, than 9.3 km offshore, of the leeward side of Oahu Island (Boehlert and Mundy, 1994). In an earlier study at Oahu Island high concentrations (up to 220 larvae/500 m3) of T. albacares larvae were found within 2 km of shore (Miller, 1979), and similar concentrations (up to 224 larvae/500 m3) of Thunnus spp. larvae (ei- ther T. albacares or T. alalunga ) were found in samples taken within 200 m of reefs in French Polynesia (Leis et al., 1991). The present study is the first to investigate near-reef distributions of tuna larvae outside of central Pacific Ocean island environments and thus extends the near-reef distributional pattern to include continental slope environments of the western Pacific Ocean, but further research on on-offshore patterns of abundance in other regions is required to confirm the generality of this phenomenon. Considerable differences in on-offshore larval distri- butions may exist among genera of tuna, perhaps even among species, and these differences may not neces- sarily reflect similarities in adult distributions. We have confirmed the opposing on-offshore distributions of Thunnus spp. and K. pelamis larvae previously discov- ered in Hawaii, where Thunnus spp. larvae were more abundant near the reef on the leeward side of Oahu Island, while K. pelamis larvae increased in abundance in the offshore direction (Boehlert and Mundy, 1994). Interestingly, the on-offshore distributions of Thunnus spp. larvae in the Coral Sea were more similar to the larval distributions of E. affinis (and possibly Auxis spp.) than to the distributions of K. pelamis larvae. Fowler et al.: Distribution and abundance of tuna larvae in near-reef waters of the Coral Sea 413 0.25 0.20 0.15 •c 0.10 o 8. 0.05 o f; 0.25 O § 0.20 cr 03 £ 0.15 0.10 0.05 0.00 3 4 5 6 7 8 Size (mm) Figure 8 Size-frequency distributions for Thunnus albacares (yellowfin tuna) larvae from inshore and offshore zones during the early February cruise in the Coral Sea. Standard length (mm) of larvae was measured to the nearest 0.1 mm. The P-value refers to the significance of a Kolmogorov-Smirnov (K-S) test conducted between the inshore and offshore zones. Inshore zone n= 33 P<0.02 m Offshore zone n=1 6 -I ‘l I This finding is remarkable, considering T. albacares and K. pelamis are considered to be truly oceanic species with similar adult distributions, whereas adult E. af- jin is and Auxis spp. have coastal distributions (Collette and Nauen, 1983). Because of potential distributional differences, further research on the near-reef larval distributions of other tuna species is required; however this may be difficult considering the relative rarity of the larvae of some species (e.g., T. tonggol). The greater abundances of small Thunnus spp. and Auxis -Euthynnus larvae within 5.6 km of the outer Great Barrier Reef indicates that these species may have spawned more intensely or more frequently (or both) in this area, than farther offshore, during the study period. Larvae of Thunnus spp. were all <3.2 mm SL because of the limits of our ability to identify small larvae, and therefore their near-reef distribution observed on at least three cruises was most likely the result of near-reef spawning activity of T. albacares (which likely comprised most of the Thunnus spp. lar- vae). In support of this conclusion, there was a greater proportion of small T. albacares larvae within 1.85 km of the outer Great Barrier Reef than farther offshore during the early February cruise. Auxis spp. or E. af- finis (or both) may have also spawned near the reef in early February, as indicated by the greater abundance of small (<2.3 mm SL) Auxis-Euthynnus larvae within 5.6 km of the outer GBR, which approached significance (ANOVA, P=0.06, Fig. 5). Their narrow size range (1.9- 2.2 mm SL) did not, however, allow for a comparison of sizes between inshore and offshore zones. Although it is likely that initial spawning distributions of the larvae of these two taxa would have been modified to some degree by subsequent physical or biological processes, or both (see below), their small size (and likely young age) would have minimized the time between spawn- ing and capture and therefore would have reduced the potential effect of subsequent modification on their ob- served distributions. The greater abundance and size of K. pelamis larvae offshore indicates that observed distributions of this species most likely arose from considerable modifica- tion of initial spawning distributions. Like T. albacares, K. pelamis likely spawned more intensely or more frequently, or both, within 1.85 km of the outer Great Barrier Reef during the study period because there was a greater proportion of small larvae within the inshore zone than in the offshore zone, on two, possibly three, cruises. Larval abundance of this species increased with increasing distance from the outer Great Barrier Reef, however, indicating that larvae may have accumulated in the offshore area. A similar pattern of increasing abundance offshore, combined with smaller larvae near the reef, was found for K. pelamis on the leeward side of Oahu Island, Hawaii (Boehlert and Mundy, 1994); however no mechanism was suggested to account for these patterns. Differential growth or mortality, or both, for K. pelamis larvae may have occurred between near-reef and offshore areas; however we believe that offshore transport by means of physical mechanisms (see below) provides the best explanation of observed distributions, at least in the Coral Sea. The scarcity of larvae of any tuna taxon in the Great Barrier Reef Lagoon, even when offshore abundances were quite high, indicates that little, if any, spawning occurred there. The moderate abundances of K. pelamis (13.1 larvae/100 m2) and T. alalunga (10.3 larvae/100 m2) larvae in the lagoon on the late November cruise were most likely caused by advection of larvae through the inter-reef passages from spawning locations on the seaward side of the outer reefs, either by onshore winds, or by tidal movement (Leis et al., 1987). We cannot exclude the possibility that some individuals of these species spawned inside the lagoon during the study period, but such spawning would have represented only a small proportion of total spawning effort. The high abundances of Thunnus spp. and T. alalunga larvae found near the reef in the present study likely resulted, at least in part, from onshore advection due to wind-driven currents interacting with the surface- orientated distribution of the larvae. The relatively consistent light to moderate onshore (E-SE) winds dur- ing the study period in the Coral Sea most likely re- sulted in shoreward advection of surface water layers. Onshore advection of surface water and subsequent downwelling on the windward side of the outer reefs, combined with a shallow vertical distribution of larvae, was suggested as a possible mechanism resulting in 414 Fishery Bulletin 106(4) Size-frequency distributions for larvae from inshore and offshore zones of the Coral Sea of Thunnus alalunga (albacore) during the (A) early November, and (B) late November cruises, and for Auxis-Euthynnus during the (C) early February cruise. Standard length (mm) of larvae was measured to the nearest 0.1 mm. P-values refer to the significance of Kolmogorov-Smirnov (K-S) tests conducted between the inshore and offshore zones within a cruise. near-reef (within 2 km offshore) distributions of Mak- aira indica (black marlin), M. mazara (blue marlin), and Istiophorus platypterus (Indo-Pacific sailfish) larvae in the Coral Sea (“the anstau hypothesis,” Leis et ah, 1987). The istiophorid larvae examined by Leis et al. (1987) for horizontal distributions were taken from the same samples used in this study. Like billfish, larvae of Thunnus spp. also have relatively shallow distributions; greatest abundances were found in the upper 20 m of the water column around Oahu Island (Boehlert and Mundy, 1994) and higher concentrations were found at 5 m depth than at 10 m depth in French Polynesia (Leis et al., 1991). Larvae of T. atlanticus (blackfin tuna) were caught in greatest numbers in the upper 20 m of the water column in the northern Caribbean Sea, and few larvae were caught below 40 m depth (Hare et ah, 2001). We cannot confirm this hypothesis, however, because we did not take direct measurements of either currents or the vertical distributions of tuna larvae during the present study period. Downwelling on the seaward side of the outer reefs in the Coral Sea could account for the simultaneous occur- rence of opposing on-offshore distributions of K. pelamis and Thunnus larvae because of known differences in the vertical distributions of larvae between these two genera. Larvae of K. pelamis have deeper distributions Fowler et al.: Distribution and abundance of tuna larvae in near-reef waters of the Coral Sea 415 than Thunnus spp. larvae (Boehlert and Mundy, 1994; Hare et al., 2001), and migrate into deeper water during the day, at which time larvae of Thunnus spp. move into surface layers (Richards and Simmons, 1971; Davis et ah, 1990b). It is therefore possible that while Thunnus spp. and T. alalunga larvae in the present study were advected shoreward by wind-driven surface currents, and accumulated there by a tendency to remain near the surface, larvae of K. pelamis were advected offshore by deeper return flow originating from downwelling near the outer reefs. At the least, K. pelamis larvae would not accumulate near the reef front because they would not be expected to counter the putative down- welling at those locations. The larger size and greater abundance of K. pelamis larvae offshore indicate that the larvae of this species likely accumulated there, providing support for the hypothesis of offshore physical transport at depth for this species. And although T. albacares and T. alalunga larvae were not larger inshore, as would be expected if larvae were transported onshore and accumulated near the reef, larvae of these species >3.5 mm SL were common near the reef, which was not the case for K. pelamis. It is possible that the size distributions of T. albacares and T. alalunga larvae in the Coral Sea may have been affected by greater mortality of larger size classes in the inshore zone than in the more offshore waters. It has been hypothesized that predation rates of larval fish are higher in near-reef waters than in the open ocean (Johannes, 1978), and direct observations of late-stage reef fish larvae have shown that larvae near reefs feed less and are preyed upon more often than larvae farther offshore (Leis and Carson-Ewart, 1998). The patterns of on-offshore distribution of tuna lar- vae documented here support the hypothesis that at least some tuna species have high larval abundances near reefs in the Tropical Pacific Ocean. We conclude that fine-scale (1-10 km) on-offshore distributions of tuna larvae found in the Coral Sea were most likely the result of relatively near-reef spawning patterns of adults (<10 km offshore) subsequently modified by wind-driven onshore currents and presumed down- welling in front of the outer reefs of the Great Barrier Reef. To account for different horizontal distributions of larvae among taxa, we suggest that putative opposing flow directions between the surface layers and deeper water may have interacted with the taxa-specific verti- cal distributions of larvae. An investigation of physical and biological factors, vertical distributions of larvae, and the abundance and distribution of spawning adults near reefs is required to further our understanding of the primary causes of on-offshore distributions of tuna larvae. Regardless of how distributions occurred, near-reef areas may generally be more important than offshore areas for the production of T. albacares and T. alalunga larvae, and possibly other large pelagic species. It is now evident from four studies that larvae of T. alba- cares and T. alalunga are abundant in near-reef (<5 km offshore) waters, and in the two studies where larval tuna abundances near a reef were compared with larval tuna abundance in offshore areas, higher abundances of Thunnus spp. larvae were found near the reef (the pres- ent study; Boehlert and Mundy, 1994). These studies also indicate that K. pelamis may, at least, spawn close to shore, although their larvae are not most abundant there. Larvae of other large pelagics, such as billfishes, may also be generally more abundant near reefs, as indicated by the near-reef abundance of larvae of three species in our study area (Leis et al., 1987). Near-reef areas have not received much attention in studies of distribution and abundance of larvae of large pelag- ic predators like tunas and billfishes. If the patterns found thus far are a general occurrence in tropical regions, larval abundance surveys that do not include these areas may underestimate true abundances. It must be kept in mind, however, that near-reef areas are much smaller than oceanic areas. Therefore, in spite of higher abundances (per unit of area) of larvae near reefs, the offshore areas may provide the bulk of the recruits to adult populations, because of the vast areas involved. As yet, there are no data on the survival rates of larvae near reefs compared to the survival rates of larvae offshore, or on their relative contributions to spawning populations. Acknowledgments We would like to thank M. McGrouther for providing access to the samples held in the Australian Museum collection. We also thank T. Trnski and A. Hay for help with sorting and identification of the tuna larvae. We are grateful for A. Poore’s advice on appropriate statisti- cal analyses. Literature cited Boehlert, G. W., and B. C. Mundy. 1994. Vertical and onshore-offshore distributional pat- terns of tuna larvae in relation to physical habitat features. Mar. Ecol. Prog. Ser. 107:1—13. Collette, B. B., and C. E. Nauen. 1983. Scombrids of the world: an annotated and illus- trated catalogue of tunas, mackerels, bonitos and related species known to date, 137 p. FAO Fish. Synop. 125. Food Agr. Organ. U.N., Rome. Davis, T. L. O., G. P. Jenkins, and J. W. Young. 1990a. Patterns of horizontal distribution of the larvae of southern bluefin (Thunnus rnaccoyii ) and other tuna in the Indian Ocean. J. Plankton Res. 12:1295-1314. 1990b. Diel patterns of vertical distribution in larvae of southern bluefin Thunnus rnaccoyii, and other tuna in the East Indian Ocean. Mar. Ecol. Prog. Ser. 59:63-74. Fritzsche, R. A. 1978. Development of fishes of the Mid-Atlantic Bight, an atlas of egg, larval, and juvenile stages, vol 5. Chae- todontidae through Ophidiidae, 340 p. Fish. Wildl. Serv. -Off. Biol. Serv. no. 78/12. U.S. Government Printing Office, Washington, DC. 416 Fishery Bulletin 106(4) Hare, J. A., D. E. Hoss, A. B. Powell, M. Konieczna, D. S. Peters, S. R. Cummings, and R. Robbins. 2001. Larval distribution and abundance of the family Scombridae and Scombrolabracidae in the vicinity of Puerto Rico and the Virgin Islands. Bull. Sea Fish. Inst. 153:13-29. Johannes, R. E. 1978. Reproductive strategies of coastal marine fishes in the tropics. Environ. Biol. Fish. 3:65-84. Leis, J. M. 1986. Vertical and horizontal distribution of reef fish larvae near coral reefs at Lizard Island, Great Barrier Reef. Mar. Biol. 90:505-516. Leis, J. M., and B. M. Carson-Ewart. 1998. Complex behaviour by coral-reef fish larvae in open- water and near-reef pelagic environments. Environ. Biol. Fish. 53:259-266. 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. Leis, J. M., and S. E. Reader. 1991. Distributional ecology of milkfish, Chanos chanos , larvae in the Great Barrier Reef and Coral Sea near Lizard Island, Australia. Environ. Biol. Fish. 30:395- 405. Leis, J. M., T. Trnski, M. Harmelin-Vivien, J. P. Renon, V. Dufour, M. K. El Moudni, and R. Galzin. 1991. High concentrations of tuna larvae (Pisces: Scomb- ridae) in near-reef waters of French Polynesia (Society and Tuamoto Islands). Bull. Mar. Sci. 48:150-158. Matsumoto, W. M. 1958. Description and distribution of larvae of four species of tuna in central Pacific waters. Fish. Bull. 58:31-72. Miller, J. M. 1979. Nearshore abundance of tuna (Pisces: Scombri- dae) larvae in the Hawaiian Island. Bull. Mar. Sci. 29:19-26. Nakamura, E. L., and W. M. Matsumoto. 1966. Distribution of larval tunas in Marquesan waters. Fish. Bull. 66:1-7. Nishikawa, Y., and D. W. Rimmer. 1987. Identification of larval tunas, billfishes and other scombroid fishes (suborder Scombroidei): an illustrated guide, no. 186, 20 p. CSIRO Marine Laboratories, Hobart, Tasmania. Richards, W. J. 1976. Spawning of bluefin tuna (Thunnus thynnus) in the Atlantic Ocean and adjacent seas. Int. Comm. Conserv. Atl. Tunas 5:267-278. Richards, W. J., T. Potthoff, and J. M. Kim. 1990. Problems identifying tuna larva species (Pisces: Scombridae: Thunnus) from the Gulf of Mexico. Fish. Bull. 88:607-609. Richards, W. J., and D. C. Simmons. 1971. Distribution of tuna larvae (Pisces, Scombridae) in the northwestern Gulf of Guinea and off Sierra Leone. Fish. Bull. 69:555-568. Scott, G. P., S. C. Turner, C. B. Grimes, W. J. Richards, and E. B. Brothers. 1993. Indices of larval bluefin tuna, Thunnus thynnus , abundance in the Gulf of Mexico; modeling variability in growth, mortality, and gear selectivity. Bull. Mar. Sci. 53:912-929. Strasburg, D. W. 1960. Estimates of larval tuna abundance in the Central Pacific. Fish. Bull. 60:231-248. 417 Abstract — We examined the effect of habitat and shrimp trawl bycatch on the density, size, growth, and mor- tality of inshore lizardfish (Synodus foetens), a nonexploited species that is among the most widespread and abundant benthic fishes in the north central Gulf of Mexico. Results of quarterly trawl sampling conducted from spring 2004 through spring 2005 revealed that inshore lizardfish are most abundant on sand habitat, but larger fish are more common on shell rubble habitat. There was no significant difference in fish density between habitats exposed to shrimp trawling on the open shelf versus those habitats within a permitted artificial reef zone that served as a de facto no-trawl area; this find- ing indicates that either inshore liz- ardfish experienced minimal effects from trawling or, more likely, that fish moved between trawled and non- trawled habitats. Exploitation ratio (bycatch mortality/total morality) estimates derived from catch curve analysis ranged from 0.43 inside the artificial reef zone to 0.55 out- side the reef zone, thus indicating that inshore lizardfish are subject to significant fishing mortality in the north central Gulf of Mexico despite the lack of a directed fishery for the species. We infer from this result that effects of shrimp trawl bycatch may be significant at the population level for nonexploited species and that a broader ecosystem-scale examination of bycatch effects is warranted. Manuscript submitted 7 November 2007. Manuscript accepted 23 June 2008. Fish. Bull. 106:417-426(2008), The views and opinions expressed or implied in this article are those of the author and do not necessarily reflect the position of the National Marine Fisheries Service, NOAA. Habitat and bycatch effects on population parameters of inshore lizardfish ( Synodus foetens ) in the north central Gulf of Mexico Sarah A. Jeffers1 William F. Patterson, III (contact author)1 James H. Cowan Jr.2 Email address (for W. F Patterson): wpatterson@uwf.edu 1 Department of Biology University of West Florida 11000 University Parkway Pensacola, Florida 32514 2 Department of Oceanography and Coastal Studies Louisiana State University Baton Rouge, Louisiana 70803 Habitat degradation and bycatch of adult or younger life-stage indviduals of nontargeted species are among the greatest ecosystem effects of fishing (Hall et al., 2000; Pauly et al., 2002, 2003; Thrush and Dayton, 2002). In fact, bycatch of nontargeted species or life stages may pose an even greater threat to marine ecosystem health than direct the harvest of targeted species (Crowder and Murawski, 1998; Baum et al., 2003; Harrington et al., 2005). Therefore, biologists, conservation groups, and fisheries agencies have called for a transition away from single species manage- ment to a more holistic ecosystem- based fisheries management (EBFM) approach (Zabel et al., 2003; Francis et ah, 2007; Marasco et ah, 2007). A single-species approach traditionally has been applied to estimate maxi- mum sustainable yield (MSY) and to subsequently set total allowable catch (TAC) for most fisheries, but that approach does not account for or mitigate against direct and indirect ecosystem effects of fishing. The shortcomings of single species management are evident when con- sidering management of the penaeid shrimp trawl fishery in the northern Gulf of Mexico (GOM). That fishery has operated near its estimated MSY since the 1950s, averaging approxi- mately 130 xlO6 lbs/yr in total land- ings of brown (Farfantepenaeus az- tecus), white ( Litopenaeus setiferus), and pink (Farfantepenaeus duorarum) shrimps, but greater than 1 x 109 lbs/ yr of bycatch have also been associ- ated with those landings (Diamond, 2004; Ortiz et ah, 2000; Strelcheck and Hood, 2007). Although that level of bycatch is clearly substantial, its effects on fishes at the population level have been studied for only a few commercially important fishes (Dia- mond et ah, 2000; Porch, 2007). Per- haps a more comprehensive approach would be to analyze the effects of by- catch on the community (e.g., Wells, 2007), as well as to examine species- specific effects for nonexploited but ecologically important fishes. Inshore lizardfish ( Synodus foetens) is an ideal model species for examin- ing the effects of shrimp trawl bycatch on the population demographics and dynamics of an ecologically but not commercially important species. In- shore lizardfish are ubiquitous on the northern GOM shelf and are among the ten most frequently encountered benthic fishes in trawl samples there (Wells, 2007). The ecological impor- tance of inshore lizardfish has been assumed to be significant given their abundance on the shelf and carnivo- rous habits (Garcia-Abad et al., 1999); however, relatively little is known about their population ecology in the 418 Fishery Bulletin 106(4) 89.5W 89. 0W 88.5W 88 OW 87.5W Figure 1 Map of areas sampled for inshore lizardfish ( Synodus foetens) in the northern Gulf of Mexico from spring 2004 through spring 2005. Large polygons on the shelf indicate artificial reef zones. Habitats inside the artificial reef zones served as de facto no trawl areas. Habitats A and F are predominantly sand; habitats B and E are low relief (1-2 m) shell rubble; habitats C and G are high relief (2-3 m) shell habitat; and habitats D and H are reef habitats. Filled circles indicate sites where inshore lizardfish were sampled opportunistically for otolith marginal condition and age analyses. The 200-m isobath is shown to indicate the shelf edge. northern GOM. As part of a broader study examining the effects of habitat type and shrimp trawl bycatch on demersal and benthic fishes in the north central GOM, the objectives of the work presented here were to ex- amine inshore lizardfish life history and ecology and to estimate potential impacts of shrimp trawl bycatch on them. Specifically, we estimated the effect of habitat type and shrimp trawl bycatch on inshore lizardfish density, size, age, and mortality in various habitats on the north central GOM continental shelf. Materials and methods Inshore lizardfish were sampled within four habitat types in the north central GOM: sand, low-relief shell rubble (low shell), high-relief shell rubble (high shell), and reef. Habitats were mapped with sidescan sonar and groundtruthed with boxcores and benthic grab samplers during previous studies (Schroeder et al., 1988; Dufrene, 2005; Strelcheck et al., 2005). Dufrene (2005) charac- terized shell rubble habitats as ridges standing 1-m (low-relief) or 2-3 m (high-relief) above the surround- ing seabed and extending up to 200 m across. Boxcore sediment samples revealed that ridges are composed of >50% calcium carbonate (CaC03), and shell fragments from both marine and estuarine taxa, the latter of which come from lower sea levels that date from the Holocene Epoch. Reef sites were characterized by Schroeder et al. (1988) as reef-like outcrops of rock rubble and shell hash supporting a diverse epifaunal assemblage. Each of the four study habitat types was found on both the open shelf and within an artificial reef zone (AR zone) off Alabama (Fig. 1). Shrimp trawling is prevalent on the open shelf, but the AR zone functions as a de-facto no trawl area (NRC, 2002, Fig. B.9 in that report). Therefore, by sampling inside and outside the AR zone the effect of shrimp trawl bycatch on inshore lizardfish population parameters could be tested. Inshore lizardfish were sampled quarterly from May 2004 to April 2005 in the habitats described above us- ing a bottom trawl rigged and fished according to the Southeastern Monitoring and Assessment Program’s (SEAMAP) trawling protocol, except that trawl sam- ples were not taken during winter and spring 2005 in the high shell habitat outside the AR zone because of weather constraints. The sampling gear was a 12.8-m, four-seam semiballoon otter trawl rigged with 2.4-m x 1-m doors, a 54.9-m bridle, a tickler chain set 1.1 m shorter than the footrope, and a codend with 4-cm Jeffers et al.: Habitat and bycatch effects on population parameters of Synodus foetens in the Gulf of Mexico 419 mesh. The sole deviation from SEAMAP protocol was the inclusion of a 0.7-cm mesh bag within the codend to retain small fish. Three replicate tows were made in each habitat type inside and outside the AR zone during quarterly sampling. Before trawl sampling, a hydrographic cast was made at each station with a Sea- bird 19plus CTD (Sea-Bird Electronics Inc., Bellevue, WA) to measure temperature, salinity, and dissolved oxygen. Then, trawling was conducted at 4.6 km/h for approximately 10 minutes. Up to ten fish were randomly sampled from each trawl sample and immediately fro- zen in ziploc bags and brought to the laboratory for processing. Individuals were thawed, weighed to the 0.1 g, measured to the nearest mm total length (TL), and their sex was determined by macroscopic examination of gonads. Both sagittal otoliths were then extracted for age analysis as detailed below. Additional fish were opportunistically sampled with trawls approximately monthly from the north central GOM (Fig. 1). In the laboratory, sampled individuals were measured, weighed, and their sex determined as above. Their sagittae then were extracted and stored. The purpose of opportunistic collections was to provide monthly samples to identify the timing of otolith opaque zone formation by means of marginal condition analysis, but those samples were also included in age and growth analyses. Inshore lizardfish density was estimated as the num- ber of individuals caught divided by the area swept by the trawl. Trawl width under tow was estimated as 0.8x12.8 m, which was based on National Marine Fisheries Service (NMFS) unpublished data of tow- ing conditions for their standard trawl survey gear. Distance towed was estimated with GPS coordinates recorded while the trawl was on the seafloor. Analysis of variance (ANOVA; 5x4x2 factorial design) allowed us to test sampling quarter, habitat, and exposure to trawling effects, as well as their interactions, on in- shore lizardfish density and size. All statistical analy- ses were computed with SAS (SAS Institute Inc., Cary, NC). Density values were transformed by taking the natural log of (density +1) and size was log-transformed to meet parametric assumptions of normality and het- eroscadascity. Analysis of variance test statistics (F- tests) were computed for both density and fish size with type-III sums of squares despite the two missing cells (i.e., no high shell data outside the AR zone for either winter or spring 2005) in the factorial design noted above. However, additional ANOVAs were also computed after removing all high shell habitat data to confirm results from the unbalanced models computed with the full data set. One sagitta from each fish was used to determine age. Otoliths were set in epoxy and a 0.5-mm transverse sec- tion including the core was cut with a Buehler (Buehler Ltd., Lake Bluff, IL) Isomet saw fitted with a diamond blade. Each section was mounted on a glass slide with thermoplastic cement, sanded with 3200 grit wet-or-dry sand paper, and polished with an alumina powder slur- ry on a felt polishing cloth. Marginal condition analysis was performed to verify that opaque zones formed an- nually in adults (Beckman et al., 1990). Opaque zones were then counted to determine age. Otolith thin sec- tions were read with a compound microscope with both reflected and transmitted light. Two different readers counted opaque zones in each sample; readers counted opaque zones independently without knowledge of fish size or the other reader’s age estimate to ensure that no bias occurred during process. If counts for a given oto- lith differed, the otolith was read a second time by both readers. Average percent error (APE) was computed to assess precision between readers (Campana, 2001). Length-at-age was plotted and then least-squares fits of the nonlinear von Bertalanffy growth curve were com- puted to estimate inshore lizardfish growth for males, females, and both sexes jointly: Lt = Lx( l-e-K«-*o», (1) where Lt = estimated length at age t\ L^= asymptotic length; K = growth coefficient; / = age in years; and t0 = hypothetical age at zero length. A likelihood ratio test was computed to test whether sex-specific growth functions were significantly different from one another (Kimura, 1980). Three independent estimates of natural morality (M) were computed based on maximum observed longev- ity according to the methods of Royce (1972), Hoenig (1983), and Pauly (1980): M = 4.6 l(tmax) (Royce, 1972), (2) In (M) = 1.44 -0.982 In (tmax) (Hoenig, 1983), (3) ln(M) = -0.0152 -0.2791n(LJ + 0.65431n(JO + 0.4631n(T) (Pauly, 1980), (4) where M = natural mortality/yr; tmax ~ the oldest aged fish in sample; and T - mean temperature °C. Total instantaneous annual mortality (Z) was estimated with catch curve analysis (Ricker, 1975). Bycatch mortal- ity (Fb) was computed by subtraction: Fb - Z - M (5) where Fh = instantaneous annual fishing (bycatch) mortality; Z = instantaneous annual total mortality; and M = instantaneous annual natural mortality. The difference in total mortality between habitats exposed to trawling versus nontrawled habitats was tested with analysis of covariance (ANCOVA) test for equal slopes. Exploitation ratio ( E ) was estimated as the ratio of Fb/Z (Ricker, 1975). 420 Fishery Bulletin 106(4) Table 1 Mean environmental parameters measured in habitats sampled for inshore lizardfish, Synodus foetens, in the northern Gulf of Mexico during 2004 and 2005 sampling. Habitat C (high relief shell rubble outside the artificial reef zone) was not sampled in winter or spring 2005. Descriptions and locations of habitat types are provided in Figure 1. Quarter Sampling dates Habitat Depth (m) Temperature (°C) Salinity (psu) Dissolved oxygen (mg/L) Spring 2004 16 May 04 A 17.3 22.2 34.2 7.0 16 May 04 B 30.3 19.9 35.2 6.5 19 May 04 C 39.6 19.5 35.7 6.3 18 May 04 D 19.0 24.4 34.6 6.9 17 May 04 E 29.0 20.6 35.2 7.3 18 May 04 F 27.3 20.7 35.1 7.1 18 May 04 G 24.0 21.1 35.0 7.5 17 May 04 H 28.0 20.6 35.2 7.3 Summer 2004 3 Aug 04 A 18.0 23.5 36.0 3.3 3 Aug 04 B 30.6 22.9 36.4 4.2 4 Aug 04 C 37.2 20.7 36.4 5.0 5 Aug 04 D 19.8 24.0 36.0 4.6 4 Aug 04 E 28.8 22.6 36.1 3.4 4 Aug 04 F 28.0 22.9 36.1 3.6 5-Aug 04 G 23.0 22.3 36.2 1.8 4 Aug 04 H 30.7 22.5 36.1 2.9 Fall 2004 29 Oct 04 A 17.7 26.0 35.5 3.9 27 Oct 04 B 30.7 23.1 36.3 4.9 26 Oct 04 C 37.4 28.1 36.2 5.8 28 Oct 04 D 18.3 26.0 35.3 5.4 27 Oct 04 E 28.8 27.0 36.3 5.7 28 Oct 04 F 26.5 26.7 36.3 5.2 28 Oct 04 G 27.5 25.8 35.8 4.7 27 Oct 04 H 29.0 26.5 36.3 5.2 Winter 2005 27 Jan 05 A 18.0 20.1 35.6 5.9 25 Jan 05 B 30.8 20.1 35.6 5.9 26 Jan 05 D 19.0 18.1 34.5 6.7 25 Jan 05 E 29.0 20.6 35.7 5.9 26 Jan 05 F 26.5 20.8 35.7 5.9 26 Jan 05 G 23.5 19.3 35.2 6.4 25 Jan 05 H 31.5 19.8 35.5 6.3 Spring 2005 27 Apr 05 A 18.0 20.2 36.3 3.7 27 Apr 05 B 31.0 20.2 36.3 3.7 28 Apr 05 D 18.6 20.3 35.5 3.9 28 Apr 05 E 29.5 20.5 36.0 4.0 27 Apr 05 F 27.0 20.2 36.3 3.7 28 Apr 05 G 24.0 20.4 35.9 3.6 28 Apr 05 H 31.0 20.6 36.0 4.1 Results All habitat types were sampled each quarter, with the exception noted above that high shell habitat outside the artificial reef zone was not sampled during winter and spring 2005. Hydrographic parameters generally were consistent among sampled habitats within sampling quarters. Water depth of trawl samples ranged from a mean of 18 m for sand habitat outside the AR zone to 38 m for high shell habitat outside the AR zone (Table 1). Temperature ranged from 19° to 26°C among quarters (mean = 22.1°C), and salinity ranged between 35 and 36 psu. Dissolved oxygen fluctuated by season and was lowest during summer months. A total of 1239 inshore lizardfish were caught in trawl samples in study habitats; 749 of those fish were randomly sampled and brought to the laboratory for further analysis. An additional 221 fish were sampled Jeffers et al.: Habitat and bycatch effects on population parameters of Synodus foetens in the Gulf of Mexico 421 opportunistically in various locations to enhance sample sizes and the temporal coverage for oto- lith marginal condition analysis and aging (Fig 1). Within study habitats, inshore lizardfish ap- peared in 91% of all trawl samples. Fish density was significantly different among sampling quar- ters and habitats (ANOVA, P<0.001 for both), but not between trawled and untrawled habitats (ANOVA, P=0.2754; Fig. 2). However, interpreta- tion of the main effects on density is complicated because of significant first-order interactions be- tween sampling quarter and trawl effects (ANO- VA, P<0.001) and habitat and trawling effects (ANOVA, P<0.001); there was no difference in sig- nificance for main effects or interactions between tests computed with or without the high shell habitat data. Inshore lizardfish were more abun- dant in sand habitat (mean density ±standard error [SE]=22.5 [±4.0] fish/ha) than in the other three habitats (mean density <13 fish/ha), but the level of difference was driven by two sand habitat samples taken outside the AR zone in summer 2004 that yielded the highest estimated densities (105 and 65 fish/ha, respectively). Among study habitats not exposed to trawling, inshore lizard- fish densities were similarly high for sand and high shell habitats (Fig. 2A). Inshore lizardfish caught during quarterly sam- pling ranged in TL from 49-404 mm (Fig. 3A), whereas opportunistically sampled fish ranged in TL from 76 to 472 mm (Fig. 3B). Fish length was significantly different among sampling quarters (ANOVA, P<0.001), habitats (ANOVA, P<0.001), and levels of the trawling effect (ANOVA, P= 0.006; Fig. 3). However, interactions between sampling quarter and habitat (ANOVA, P<0.001) and habitat and the trawl effect (ANOVA, P<0.001) complicated inter- pretation of the main effects; there was no difference in significance for the main effects, or their interactions on fish size between tests computed with and without the high shell habitat data. Overall, mean (±SE) fish size was smaller in habitats exposed to trawling (230.7 [±3.2] mm TL) than in habitats sampled within the AR zone (242.9 [±2.2] mm TL). However, high shell habitat outside the AR zone sampled in spring 2004 had the largest mean (±SE) size (320.0 [±44.4] mm TL) among all factor level combinations, although only three fish were captured in that habitat in spring 2004. Fish size was smallest (mean TL ±SE=204.7 [±4.6| mm) during summer 2004 among all habitats. That trend was most pronounced in sand habitat both inside and outside the AR zone where high densities (Fig. 2) of mostly small fish (Fig. 4) were encountered. Of the 970 otoliths prepared for age estimation (749 from quarterly sampling and 221 opportunistically sam- pled), age could be determined for 967. Marginal condi- tion analysis demonstrated that otolith opaque zones generally began forming in November and continued to do so until February (Fig. 5). All samples were at least one year of age. The oldest fish sampled was a 424-mm- o Sand A Low-relief shell □ High-relief shell V Reef 60 - 40 - 20 - € 0 - •J £ ii $ n l f a o o° W >. Spring 04 Summer 04 Fall 04 Winter 04 Spring 05 (/) S ioo- Q B 80 ■ 60 - « > 40 ■ 5 20 - 0 ■ *ov 2 o 2 2 £ n A * Spring 04 Summer 04 Fall 04 Winter 04 Spring 05 Quarter Figure 2 Mean (±standard error) density (fish/ha) of inshore lizardfish ( Synodus foetens) sampled from spring 2004 through spring 2005 in study habitats in the northern Gulf of Mexico for (A) areas inside the artificial reef zone that were not subjected to trawling and (B) areas outside the artificial reef zone that were subjected to trawling. TL 9-year-old female that was sampled in June 2006 onboard during the SEAMAP trawl survey. Of the fish sampled within study areas, the oldest was 8 years and nearly half of all fish were 3-year-olds. Reader agree- ment was judged to be good with an APE of 5.94%. Von Bertalanffy growth functions were not signif- icantly different between sexes (likelihood ratio %2- test, P- 0.998); thus size-at-age data were modeled jointly between sexes. The resultant growth equation was Lt = 290.8(1 - e-0-486 (f-o.204)> (nonlinear regression, P<0.001; r2 (coefficient of determination? If so, lower- case)^.22) (Fig. 6). Estimates of M based on an ob- served tmax of 9 yr were 0.51/yr, 0.49/yr, and 0.53/yr from the methods of Royce (1972), Hoenig (1983), and Pauly (1980), respectively. Before conducting catch curve analysis, the age distribution of the aged fish (n- 749) sampled in our study habitats was expanded to the samples collected that were not aged (>z = 490). This was accomplished by computing habitat- and sampling quarter-specific age distributions and by assigning age by means of a random number table for fish collected that were not aged. Then, catch curve analysis was computed for ages 3 yr and older because those were the fully recruited ages (Fig. 7). Total mortality of inshore lizardfish sampled within the AR zone was Z = 0.93/y 422 Fishery Bulletin 106(4) (linear regression, PcO.OOl, r2=0.993); estimated total mortality outside the AR zone was Z = 1.10/yr (linear regression, PcO.OOl, r2=0.971). Therefore, fish outside the AR zone were exposed to higher Z than those inside, but the slopes were not significantly different between the two catch curves (ANCOVA test for equal slopes, P=0.232). By subtraction, Fb estimates ranged from 0.40 to 0.44/y inside the AR zone and 0.57 to 0.61/yr outside the AR zone. Resultant E estimates ranged from 0.43 to 0.47 for fish sampled inside the AR zone and 0.52 to 0.55 for fish outside the AR zone. Discussion Inshore lizardfish were nearly ubiquitous throughout study areas. Fish abundance was greater in summer, especially for smaller individuals, possibly indicating that recruitment to study habitats from estuarine and inshore habitats occurred in summer. Small inshore lizardfish begin recruiting from estuaries to Campeche Bay in the southern GOM in June and continued to do so until October (Garcia-Abad et al., 1999). A similar pattern was observed in the present study where both size and inshore lizardfish age were lowest in summer. Overall, catch-at-age data indicated that inshore liz- ardfish did not fully recruit to the study sites until age three, but the mean age of fish sampled in summer was Figure 3 Frequency histograms of total length (mm) distributions of inshore lizardfish ( Synodus foetens) collected within (A) study habitats and (B) for fish sampled opportunistically in the north central Gulf of Mexico during spring 2004 through spring 2005. Sample sizes ( n ) are shown for each historgram. slightly less than 3 yr. Fish size, as well as age, was greatest in high shell habitat, but density was lowest there. Cruz-Escalona et al. (2005) reported that adult inshore lizardfish in the southern GOM preferred sand habitat, and our data indicated a similar trend in the northern GOM. However, smaller fish may avoid more complex habitats, such as shell rubble ridges, because of the presence of predators or increased competition for food. Fish were larger, on average, inside the AR zone, but fish density was not significantly different inside and outside the AR zone, which may have resulted from the movement of inshore lizardfish among trawled and untrawled habitats. No direct observations of inshore lizardfish movement were made in this study, but in- dividuals are known to move 10s of km as they recruit to adult habitats on the shelf from inshore estuarine nursery habitats (Cruz-Escalona et al., 2005). Other lizardfishes also move significant (10s to 100s of km) distances (Sweatman, 1984; Golani, 1993). Further- more, experimental closure of areas to bottom trawling in Australia’s northern prawn fishery did not yield sig- nificant differences in lizardfish density between areas open to trawling and those closed to trawling because fish moved into trawled areas after trawling occurred, thus restoring high densities there (Stobutzki et al., 2003). Therefore, the lack of differences observed in inshore lizardfish density in areas exposed to trawl- ing and those not exposed to trawling during the present study may have resulted from fish moving between trawled and untrawled areas. Marginal conditional analysis confirmed that the timing of opaque zone formation in inshore liz- ardfish otoliths is similar to a range of demersal (e.g., black drum [Pogonias cromis] and red snap- per [ Lutjanus campechanus ]) and benthic (e.g., southern flounder [Paralichthyes lethostigma ]) fish- es in the northern GOM (Beckman et al., 1990; Patterson et al., 2001; Fischer and Thompson, 2004) and that annual opaque zones are laid down from November to February. Based on counts of annuli, the maximum longevity observed for in- shore lizardfish (9 years) was similar to the maxi- mum longevity reported for other synodontid spe- cies. For example, Yoneda et al. (2002) reported that a Saurida species from the East China Sea lived to 11 years and Thresher et al. (1986) aged a Saurida species from the northwestern Australian shelf to be 7 years. The low regression coefficient (r2 = 0.22) of the von Bertalanffy growth function computed with size-at-age data reflects the substantial vari- ability in inshore lizardfish size-at-age. Hood and Johnson (1999) reported vermilion snapper ( Rhomboplites aurorubens) displayed similar vari- ability in size-at-age in the GOM. They indicat- ed vermilion snapper were difficult to age, but their high reader agreement indicated that the observed variability in size-at-age was not an artifact of inaccurate aging. Similarly, we infer Jeffers et al. : Habitat and bycatch effects on population parameters of Synodus foetens in the Gulf of Mexico 423 the variability in size-at-age reported for inshore lizardfish is representative of the variability in the population given the high APE computed be- tween reader age estimates. Results from catch curve analysis and calcula- tions of Fb appear to indicate that shrimp trawl bycatch mortality was substantial for inshore lizardfish. However, that interpretation depends on the assumption that the sampling gear has a logistic selectivity-at-age function for inshore lizardfish, and that there was no other source of fishing mortality beyond that of shrimp trawl bycatch. If the selectivity-at-age function was domed shaped, for example, then some of the decline in numbers of larger, older fish in our sample would have been due to older fish not being fully selected by the gear; however, our interpretation was that the observed decline, be- yond that due to M, was caused by considerable bycatch mortality. Experiments could be designed to estimate the selectivity function of our sam- pling gear for inshore lizardfish, but no data currently exist to evaluate selectivity. However, given the small size of even the largest inshore lizardfish sampled in relation to the size of the sampling trawl, it seems reasonably safe to as- sume that a logistic selectivity function existed. Furthermore, it seems unlikely that other sources of fishing mortality, beyond shrimp trawl bycatch, are substantial for inshore lizardfish in the north central GOM. Inshore lizardfish are not targeted by recreational fishermen in the region and we have no knowledge of them being captured in commercial or recreational hook-and-line fisher- ies as bycatch. The difference in Z, hence Fb, was not statisti- cally significant between fish sampled inside and those sampled outside the AR zone. Nonetheless, a difference of Z = 0.17/yr clearly is biologically meaningful. The lack of a larger difference may indicate that either the assumption that no trawl- ing occurs inside the reef zone was false, or that inshore lizardfish moved between habitats inside and outside the AR zone. Results of electronic tracking of shrimp trawler GPS coordinates in- dicate some shrimping effort may occur inside the AR zone (NRC, 2002, Fig. B.9 in that report). However, shrimping effort was shown to be orders of magnitude greater outside than inside the AR zone. Therefore, the lack of difference in Z, hence F b, between habitats inside and outside the AR zone most likely was caused by movement of in- shore lizardfish into and out of the AR zone. Exploitation ratios computed for inshore lizard- fish appear to indicate that the species is heavily fished in the northern GOM despite the lack of a directed fishery for it. A general rule of thumb is that an E approaching 0.5 indicates that a fished population is fully exploited, whereas a ratio greater than 0.5 indicates heavy fishing pressure 400 350 300 250 200 150 100 o Sand A Low-relief shell □ High-relief shell V Reef > ft Spring 04 Summer 04 Fall 04 Winter 04 Spring 05 I 3.0 - a. (/) ° 2.5 - “ 2.0 - 300 mg/kg) of sodium pentobarbital ( Beuthanasia-D, Scher- ing-Plough Animal Health Corp., Union, NJ) injected intramuscularly. After drug injections, individual fish were moved into a light-tight enclosure and placed on a perforated rub- ber sling stretched across an acrylic box. The majority of the body was submerged; only a small portion of the head and the eye receiving the light stimulus remained above the water. The box was supplied with running seawater (12°C) and a small submersible pump continu- ously circulated water over the gills of the fish. Fish were allowed to acclimate to the dark for at least one hour before any measurements were taken. Silver-silver chloride electrodes, constructed from teflon-coated silver wire, were used for recording the ERGs. The active electrode was lightly placed on the corneal surface and the reference electrode either in one of the nares or on the skin over the head. (Electrodes were positioned under dim red light [peak wavelength 660 nm] produced by light-emitting diodes [LEDs].) The recording system was grounded to the seawater through a stainless steel plate. ERG signals were am- plified by using a 10,000x gain with 1 Hz high pass and 1 kHz low pass filter settings (amplifier model DAM 50, World Precision Instruments, Sarasota, FL). The resul- tant signal was further filtered with a Humbug® active electronic filter to remove 60 Hz noise (Quest Scientific, North Vancouver, BC, Canada), and digitized at 1 kHz sampling frequency with a multifunction data-acquisi- tion card (model 6024E, National Instruments, Austin, TX). Data recording and stimulus presentations were controlled by a custom-designed software developed by Eric Warrant (University of Lund, Lund, Sweden) us- ing the LabVIEW graphical programming system for measurement and automation (National Instruments, Austin, TX). In order to account for any influence of circadian rhythms on visual responses (Mangel, 2001), experiments were conducted during the hours the fish holding tanks were lit (herein referred to as “day”), and then repeated on the same individual during the hours the fish holding tanks were in darkness (herein referred to as “night”). As a result, the night experi- 430 Fishery Bulletin 106(4) ments on Pacific halibut were conducted approximately 10-12 hours after exposure to bright light. Three separate procedures were conducted to test for treatment effects on visual function: responses to increasing light intensities (V-log I response curves), flicker fusion frequency, and spectral sensitivity. For the first two procedures, light stimuli were produced by a circular (3.8 cm diameter) light source (model SL2420, Advanced Illumination, Rochester, VT) com- prising 20 white LEDs. The LEDs were mounted behind a thin diffuser and collimating lens to produce an even (±10% edge to edge) illumination field. Light output was controlled by an intensity controller (model CS410, Advanced Illumination, Rochester, VT), which in turn was controlled the analog output of the data acquisition card. A series of neutral density filters (Kodak Optical Products, Rochester, NY) were used to extend the range of light levels as needed. For the determination of spectral sensitivity, the output of a xenon fiberoptic light source (model Y1603, CVI Laser Spectral Products, Albuquerque, NM) was controlled with a monochrometer (model CM110, CVI Laser Spectral Products, Albuquerque, NM), two filter wheel assemblies (model AB301, CVI Laser Spectral Products, Albuquerque, NM) containing quartz neu- tral density filters, and an electronic shutter (model LS6, Uniblitz, Vincent Associates, Rochester, NY). The monochrometer produced single wavelength light (with an 8 nm 50% bandwidth) between 300 and 700 nm. The filter wheel assemblies allowed the attenua- tion of light from 0 to 5 log units in 0.2 log-unit steps. The monochrometer and filter wheel assemblies were controlled by using serial (RS232) interfaces, and the shutter was controlled by the digital output of the data acquisition card. Light from the xenon light source was conducted through the monochrometer — >- filter wheel — > shutter assembly, and from there directed at the eye through 3-mm light guides. The LED light source and exit point of the light guide were placed approximately 5 cm from the corneal surface. The outputs of the LED and xenon light sources (the lat- ter measured at exit of the light guide after passage through the monochrometer — > filter wheel —>■ shutter assembly) were calibrated with a research radiometer (model IL 1700, International Light, Inc., Newbury- port, MA). To construct V-log I response curves, light intensities were increased in 0.2 log-unit steps from levels that produced no measurable responses, to those that pro- duced maximal responses. Stimuli consisted of a train of five 200-ms duration light flashes 200 ms apart. The trains of stimuli at each light intensity were presented every five seconds and repeated five times. The ERG responses to the final flash of each train were recorded and averaged. The data were subsequently normalized by expressing the average response to an intensity step as a fraction of the maximum observed average response. Normalized ERG responses versus log light intensities (candela/meter2) were plotted to construct V-log I response curves. Mean V-log I curves for each species were constructed by averaging the normalized curves of all individuals within a treatment group re- corded during the day or during the night. Flicker fusion frequencies were determined by using five-second sinusoidal light stimulus trains, followed by five seconds of darkness. The maximum light intensity of the sinusoidal stimulus was that required to produce a response that was 50% of the maximal response. This value was determined by eye during experiments from the individual V-log I response curves. Stimulus trains were repeated five times at each frequency and the re- sponses averaged. Stimulus frequencies were increased from 1 Hz (0 log units) to 63.1 Hz (1.8 log units) in 0.2 log-unit steps. The flicker fusion frequency was deter- mined by comparing the power spectrum of the aver- aged response at each stimulus frequency (signal) to the power spectrum of a neighboring frequency (noise) and was defined as the frequency at which the power of the signal fell below the power of the noise. Spectral response curves were determined using monochromatic light flashes produced by a xenon light source and monochrometer —> filter wheel — > shutter assembly. Approximately isoquantal light stimuli from 300 to 650 nm, regulated by the monochrometer and neutral density filters, were presented to subjects in 10-nm wavelength steps. Five stimuli of 40 ms duration were presented at each wavelength, and five seconds were allowed between each light flash. The responses to the five flashes were averaged and mean amplitudes recorded. Spectral sensitivity curves were subsequently cal- culated from the spectral response curves as follows. First, spectral response data were corrected for dif- ferences in lamp output at specific wavelengths, as well as for differences in neutral density filter values from their nominal values, through the application of correction factors. The correction factors were based on the calibration curves developed with the research radiometer described above. Second, responses to ex- actly isoquantal intensities were predicted by adjusting the response amplitude at each wavelength by using spectral V-log I response curves generally following the methods described in Coates et al. (2006). The only exception was that a fourth-order polynomial was fitted to the individual spectral V-log I response data. Spec- tral V-log I response curves were recorded immediately after measurement of the spectral response curves by using a series of increasingly intense flashes at the wavelength which generated the largest response. Five flashes (200 ms duration, five seconds apart) were delivered at each intensity step and the response am- plitudes averaged. Intensities increased in 0.2 log-unit steps and ranged from those producing no response, to those producing a maximal response. Individual spectral sensitivity curves were normal- ized by using the maximal response at each wavelength. Mean spectral sensitivity curves were constructed by averaging the spectral sensitivity curves of all individu- als within a treatment group recorded during the day or during the night Brill et al.: Effects of rapid decompression and exposure to bright light on visual function in Sebastes melanops and Hippoglossus stenolepis 431 Table 1 Mean (and 95% confidence bands) intensities (log candela/meter2) of broad spectrum white light required to produce electro- retinogram (ERG) responses 50% of the maximal observed responses in black rockfish ( Sebastes melanops) and Pacific halibut (Hippoglossus stenolepis). The 50% response points were calculated to explore day-night differences, species-specific light sen- sitivities, and treatment effects. The mean 50% response point in control Pacific halibut was more than an order of magnitude lower than that of black rockfish, indicating the significantly greater light sensitivity of the former. There was no effect of rapid decompression on the 50% response point of black rockfish, whereas 15 minutes of exposure to simulated sunlight increased the 50% response points of Pacific halibut, clearly indicating a diminished light sensitivity in the retina. Day Night Black rockfish, control Black rockfish, rapid decompression Pacific halibut, control Pacific halibut, light exposed 2.1 (2. 1-2.2), n=l 2.3 (2. 2-2. 3), n = 6 0.09 (0.01-0.19), n = 6 0.56 (0.47-0.63), n = 5 2.0 (1. 8-2.1), n=l 1.9 (1. 8-2.0), n = Q -0.07 (-0.16 -0.05), n = 5 0.78(0.61-0.91), n=5 Statistical procedures To explore species differences and treatment effects on retinal light sensitivity, the Weibull four-parameter sig- moid function contained within SigmaPlot for Windows version 10 (Systat Software, San Jose, CA) were fitted to the mean V-log I response data. This function was chosen because ERG responses generally exhibit a sig- moid response to increasing light intensities (Kobayashi, 1962). Light intensities required to produce responses that were 50% of that needed to produce a maximal response (and their 95% confidence bands) were taken from the predicted values produced by the curve-fitting program. Differences in the mean flicker fusion frequency data collected during the day and night were tested with paired-7 tests (SigmaStat 3.1, Systat Software, San Jose, CA) because they represented “before-and-after” trials on the same animal. In those instances where no differences were found, day and night data were combined. Differences in the flicker fusion frequency between control animals and those subjected to rapid decompression (black rockfish) or simulated sunlight (Pacific halibut), were likewise treated with 7-tests. In instances where no treatment effects were present, data were combined. The Mann-Whitney rank sum test was used to test for species differences because the combined data sets were not normally distributed. In all instances P<0.05 was taken to indicate significant differences. To explore the specific effects of bright light expo- sure on retinal function in Pacific halibut, the vita- min Al rhodopsin absorbance templates developed by Stavenga et al. (1993) were fitted to the normal- ized spectral sensitivity data assuming the pres- ence of two visual photopigments. Unknown model parameters (photopigment absorption maxima and their weighting proportions) were estimated by using maximum likelihood within the software package R (vers. 2.7.0, R Foundation for Statistical Computing, Vienna, Austria). Results Effects of rapid decompression Five of eight black rockfish exposed to simulated cap- ture showed severe exophthalmia immediately after rapid decompression, and one fish in this condition also displayed corneal emphysemas. All eight fish showed other external signs of barotraumas, such as gas bubbles under the branchiostegal membrane, and seven of the eight showed esophageal eversion. None showed any gas bubbles within the vitreous humor. Exophthalmia and other signs of barotrauma disappeared in all fish immediately after recompression. There were no obvi- ous external anatomical abnormalities or evidence of disease in any of the fish at the time they were used in an experiment. Responses to increasing light intensities The amplitude of ERG responses of both black rock- fish and Pacific halibut increased with increasing light intensities (Fig. 2) and produced the expected sigmoid V-log I response curves (Kobayashi, 1962). Based on the overlap of mean 50% response points and the 95% confidence bands (Table 1), there were no significant day-night differences in the light sensitivity of either species within treatments with the possible exception of black rockfish exposed to rapid decompression. The mean 50% response points in the control Pacific halibut were more than an order of magnitude lower than those of the black rockfish (Table 1), indicating a significantly greater light sensitivity in the former. There was no effect of rapid decompression on the 50% response point for black rockfish. In contrast, the responsiveness of Pacific halibut retinas was signifi- cantly diminished by exposure to bright light. During both day and night recordings, the V-log I response curves for Pacific halibut exposed to simulated sun- light were clearly right-shifted compared to controls (Fig. 2). The mean 50% response points for treated 432 Fishery Bulletin 106(4) Log I Figure 2 Electroretinogram (ERG) responses to increasing light intensities (I, in log can- dela/m2) in black rockfish (Sebastes melanops ) and Pacific halibut (Hippoglossus stenolepis). To construct response curves, ERG responses to 200-ms duration light flashes were recorded as light intensity from a white LED light source was increased in 0.2 log-unit steps from levels that produced no measurable responses, to those that produced maximal responses. Responses from individual trials were normalized by expressing them as a fraction of the maximal response. Data recorded at light levels beyond those that produced a maximal response have been omitted. Results obtained during the day and during the night were analyzed separately. Data points are means ±standard error; open circles show results obtained during the day and filled circles results obtained during the night. The Weibull four-parameter function contained within SigmaPlot for Windows vers. 10 (Systat Software, San Jose, CA) was fitted to the mean V-log I response data. A sigmoidal curve was chosen because the ERG responses generally approached zero at low light intensities, became saturated (i.e., approached a maximum) at high light levels, and varied in between. Dotted lines show the predicted values based on data collected during the day and solid lines show predicted values based on data collected during the night. For illustrative purposes, vertical dotted and solid lines have been added showing the light intensities required to produce responses 50% of the maximal response during the day and night, respectively. fish were approximately 0.5 to 0.8 log units higher than the mean values for control fish (Table 1). In other words, the light levels required to reach the 50% response points in Pacific halibut exposed to simulated sunlight were three to seven times higher than the light levels required to reach the 50% response points in control fish. Flicker fusion frequency There were no day-night differences in the flicker fusion frequencies for either black rockfish or Pacific halibut. There was also no detectable influence of rapid decom- pression on flicker fusion frequency for black rockfish, nor an apparent effect of exposure to bright light on Brill et al. : Effects of rapid decompression and exposure to bright light on visual function in Sebastes melanops and Hippoglossus stenolepis 433 c/i CD Wavelenath (nm) Figure 3 Spectral sensitivity curves for black rockfish ( Sebastes melanops ) and Pacific halibut (Hippoglossus stenolepis) derived from electroretinogram (ERG) responses to 40-ms flashes of incremental wavelengths (10-nm steps). Responses from individual trials were normalized by expressing the response to a given wavelength as a fraction of the maximal response recorded during that trial. Data points are means ±standard error. The spectral response curves of black rockfish were unaffected by rapid decompression, whereas the ERG responses of Pacific halibut to green wavelengths (=520-580 nm) were diminished by 15 minutes of exposure to simulated sunlight. Color bars (truncated at 400 and 650 nm) have been added to illustrate the approximate spectral colors corre- sponding to the various wavelengths. the flicker fusion frequency of Pacific halibut. Within- species data were therefore combined to test for cross- species differences. The median flicker fusion frequency for Pacific halibut (30 Hz) was significantly lower than that of black rockfish (49 Hz) (PcO.001, Mann-Whitney rank sum Test). Spectral curves The spectral sensitivity curves of black rockfish and Pacific halibut were very similar (Fig. 3). Both species showed strong sensitivity to blue-green wavelengths (480-590 nm), and a range of responses from 380 nm (violet) to 610 nm (orange). There was no appreciable sensitivity to the shorter (UV-A, 350-380 nm) nor longer (red, >620 nm) wavelengths in either species. There was no indication that the spectral sensitiv- ity of black rockfish was affected by rapid decom- pression (Fig. 3), whereas the spectral sensitivity of Pacific halibut was clearly influenced by exposure to simulated sunlight (Fig. 3). Vitamin Al rhodopsin absorbance templates developed by Stavenga et al. (1993) were, therefore, used to further assess these changes. In all cases, the templates provided rea- sonable fits to the observed ERG data (Fig. 4). The parameters showing the largest effects of bright light exposure were clearly the ratios of the long wave- length to short wavelength photopigment weighting factors that are indicative of the relative value of the two photopigments to the composite curve. In control Pacific halibut, this ratio ranged from 0.75 during the day to 0.62 during the night, indicating only a slight predominance of the shorter wavelength photopig- ment (Fig. 4). In contrast, in Pacific halibut exposed to simulated sunlight these ratios were 0.48 and 0.12 during the day and night, respectively. These results indicate a much reduced functional importance of the longer wavelength photopigment, especially during the night (i.e., approximately 12 hours after exposure to simulated sunlight). 434 Fishery Bulletin 106(4) Figure 4 Vitamin A1 rhodopsin absorbance templates (Stavenga et al., 1993) fitted to Pacific hali- but (Hippoglossus stenolepis ) electroretinogram (ERG) data by the maximum likelihood method. The circles show the normalized spectral sensitivity data and the solid lines show the predicated combined photopigment absorption curves. The predicted individual photopigment absorption curves for the short and long wavelength pigments are shown the by the dotted and dashed lines, respectively. The predicted absorption maxima are in reasonable agreement with family Pleuronectidae (Evans et al., 1993; Jokela-Maatta et al,. 2007). Note the dramatically diminished importance of the longer wavelength pig- ment in halibut exposed to 15 minutes of simulated sunlight, especially at night. The spectral colors corresponding to the various wavelengths are shown in Figure 3. Discussion Our observations with black rockfish do not support our original contention that exophthalmia and other internal events associated with rapid decompression compromise retinal function. We found no differences in light sensitivity (V-log I response curves), flicker fusion frequency, or spectral sensitivity between control and experimental fish. Our procedures would not, however, detect damage to the optic nerve or other parts of the central nervous system associated with vision. Experi- ments involving visually evoked potentials (Bullock et al., 1991), predator- and prey-sighting distance, the optomotor response, or other behavioral procedures (Douglas and Hawryshyn, 1990; Vogel and Beauchamp, 1999; Herbert and Wells, 2002) are needed to confirm the lack of detrimental effects of barotrauma on visual function in black rockfish. Rockfish simply discarded on the surface after being subjected to rapid decompression during capture frequently float. They are than subjected to high rates of avian predation. Returning rockfish to depth (repressurization) reverses the external signs of barotrauma (Parker et al., 2006). Our observations confirm that releasing rockfish at depth is an effective procedure for minimizing postrelease mortality, except for fish with severe hook injuries (St. John and Syers, 2005). In contrast to the apparent lack of effects of baro- trauma on vision in black rockfish, there are multiple lines of evidence that 15 minutes of exposure to simu- lated sunlight dramatically affects the visual func- tion of Pacific halibut. First, exposure to bright light causes a large reduction in sensitivity (Table 1) indi- cated by the approximate 3 to 7x increase in amount of light required to achieve an ERG response to broad spectrum white light that is 50% of the maximal re- sponse (Table 1). We strongly suspect that low-light vision was also affected. Measuring low-light sensitiv- ity (i.e., minimal detectable ERG response) requires a specific set of procedures different from the ones we employed (Reilly and Thompson, 2007) and we did not have sufficient support for this project to allow us to Brill et al.: Effects of rapid decompression and exposure to bright light on visual function in Sebastes melanops and Hippoglossus stenolepis 435 conduct the requisite experiments. Pacific halibut are less able to detect baits in near total darkness than at brighter light levels (Stoner, 2003). By extension, we conclude that individuals whose visual function has been compromised by exposure to bright light will be less able to feed and avoid predators than normal animals. Behavioral tests quantifying the effects of bright light exposure on the ability of Pacific halibut to locate and capture prey or detect predators are clearly warranted. To the best of our knowledge, there are no published descriptions of retinal anatomy in Pacific halibut. The retinas of Atlantic halibut (Hippoglossus hippoglossus ) contain both rods and cones (Kvenseth et al., 1996), and we strongly suspect that the retinas of Pacific halibut do also. Likewise, we know of no microspectro- photometry studies detailing the absorbance maxima of photopigments in juvenile or adult Pacific halibut retinas. Microspectrophotometry studies of other mem- bers of the family Pleuronectidae ( Platichthys flesus [flounder] and Pseudopleuronectes americanus [win- ter flounder]) show that the peak absorbance of the photopigment in rod cells is =510 nm, that in single cones is =450 nm, and that in double cones is =530 or 550 nm (Evans et al., 1993; Jokela-Maatta et al., 2007). From these lines of evidence, and the results obtained by fitting rhodopsin absorbance templates to the ERG data by the maximum likelihood method (Fig. 4), we concluded that Pacific halibut retinas contain two photopigments with absorbance maxima similar to other members of the family Pleuronectidae. Our conclusion is also in general agreement with the maxi- mum spectral sensitivities of a range of coastal and continental shelf species (Levine and MacNichol, 1979; Bowmaker, 1990). The change in spectral sensitivity curves of Pacific halibut (Fig. 3) caused by exposure to bright light and the results of photopigment template fitting (Fig. 4) imply that it is the functional properties of a single class of cells (those containing the longer wavelength =520 to 540 nm photopigment) which are predominate- ly disrupted by exposure to bright light. The retinas of larval Atlantic halibut and adults of the related species — winter flounder — are dominated by the so- called “green cones” (Evans et ah, 1993; Helvik et al., 2001). If this is also the case in Pacific halibut, our results indicate disruption of the photoreceptor cells which normally provide maximal quantal absorption in the green-light dominated coastal waters (Levine and MacNichol, 1979; Lythgoe, 1975, 1980; Crescitelli, 1991). This detriment to vision is likely to have severe consequences for postrelease predator avoidance and foraging success. Moreover, the results from phot- opigment template fitting indicate that the specific functional deficit persists for at least 10 to 12 hours after the exposure to bright light and worsens with time (Fig. 4). If this diminished functionality is due to apoptosis induced by photic injury (Wu et ah, 2006), it is likely that it will continue to worsen progressively and be permanent. Hook-and-line caught Pacific halibut can be discarded quickly and are not subject to significant time periods out of the water (Kaimmer and Trumble, 1998). Be- cause they would not be exposed to direct sunlight for prolonged periods, minimal or no reduction in visual function would be expected. In contrast, trawl-caught Pacific halibut can be exposed to bright light during prolonged sorting operations (Davis and Olla, 2001). Increased mortality has been demonstrated in Pacific halibut that remain on deck for 20-40 min (Trumble et al., 1995). Although the exact causes are unknown, our results clearly imply that these fish could have had significant visual impairment when discarded. De- velopment of procedures to improve survival of Pacific halibut may therefore require not only reducing sort- ing time (and therefore reducing exposure to air and extreme temperatures), but also sheltering the fish from bright light. The low flicker fusion frequency and high light sen- sitivity of Pacific halibut are characteristics of a visual system adapted to function at low light levels (War- rant, 1999). We hypothesize that these features make Pacific halibut, and other demersal fishes with similarly structured visual systems, more susceptible to damage by exposure to direct sunlight than species normally inhabiting brightly lit environments. Should this be the case, there would be significant implications for fishery management, although obviously many significant ques- tions remain. For example, we do not know how long in- dividuals have to be exposed to direct sunlight to incur deficits in visual acuity, nor do we know the threshold of light intensity that causes retinal damage. Research is also warranted on exactly which of the various cell types within the retina are being damaged, the effect(s) of light-induced visual deficits on predator-avoidance and prey-finding behaviors, the resultant changes in rates of mortality, and possible effects at the popula- tion level. Acknowledgments This project was funded by Alaska Fisheries Science Center, National Marine Fisheries Service, National Oceanic and Atmospheric Administration. All proce- dures described herein were approved by the Oregon State University Institutional Animal Care and Use Committee (ACUP 3577), and comply with all appli- cable U.S. laws and regulations. The views expressed herein are those of the authors and do not necessarily reflect the views of NOAA or any of its sub-agencies. The authors gratefully acknowledge the hospitality of the staff of the Hatfield Marine Science Center (Newport, Oregon), and A. Horodysky for his efforts to initiate this project, his comments on earlier drafts of this manuscript, and his help with the photopigment template fitting procedures. R. Latour developed the R-code required for photopigment template fitting. This is contribution 2938 from the Virginia Institute of Marine Science. 436 Fishery Bulletin 106(4) Literature cited Berenbrink, M., P. Koldkjaer, O. Kepp, and A. R. Cossins. 2005. Evolution of oxygen secretion in fishes and the emergence of a complex physiological system. Science 307:1752-1757. Bowmaker, J. K. 1990. Visual pigments of fishes. In The visual system of fish (R. Douglas, and M. Djamgoz, eds.), p. 82- 107. Chapman and Hall, London. Brown, K. T. 1968. The electroretinogram: its components and their origin. Vision Res. 8: 633—677. Bullock, T. H., M. H. Hofmann, J. G. New, and F. K. Nahm. 1991. Dynamic properties of visual evoked potentials in the tectum of cartilaginous and bony fishes, with neu- roethological implications. J. Exp. Zool. 5:142-155. Coates, M. M., A. Garm, J. C. Theobald, S. H. Thompson, and D.-E. Nilsson. 2006. The spectral sensitivity of the lens eyes of a box jellyfish, Tripedalia cystophora (Conant). J. Exp. Biol. 209:3758-3765. Crescitelli, F. 1991. The scotopic photoreceptors and their visual pig- ments of fishes: Functions and adaptations. Vision Res. 31:339-348. Davis, M. W. 2002. Key principles for understanding fish bycatch discard mortality. Can. J. Fish. Aquat. Sci. 59:1834- 1843. 2007. Simulated fishing experiments for predicting delayed mortality rates using reflex impairment in restrained fish. ICES J. Mar. Sci. 64:1535-1542. Davis, M. W., and B. L. Olla. 2001. Stress and delayed mortality induced in Pacific halibut by exposure to hooking, net towing, elevated sea- water temperature and air: implications for management of bycatch. N. Am. J. Fish. Manag. 21:725-732. Davis, M. W., and M. L. Ottmar. 2006. Wounding and reflex impairment may be predic- tors for mortality in discarded or escaped fish. Fish. Res. 82:1-6. Davis, M. W., and S. J. Parker. 2004. Fish size and exposure to air: potential effects on behavioral impairment and mortality rates in discarded sablefish. N. Am. J. Fish. Manag. 24:518-524. Davis, M. W., and C. B. Schreck. 2005. Responses by Pacific halibut to air exposure: lack of correspondence among plasma constituents and mortality. Trans. Am. Fish. Soc. 134:991-998. Douglas, R. H., and C. W. Hawryshyn. 1990. Behavioural studies of fish vision: an analysis of visual capabilities. In The visual system of fish (R. H. Douglas, and M. B. A. Djamgoz, eds.) p. 373-418. Chap- man and Hall, London. Evans, B. L, F. I. Harosi, and R. D. Fernald. 1993. Photoreceptor spectral absorbance in larval and adult winter flounder, ( Pseudopleuronectes americanus). Visual Neurosci. 10:1065—1071. Fonner, D. B., J .R. Hoffert, and P. O. Fromm. 1973. The importance of the counter current oxygen multiplier mechanism in maintaining retinal function in the teleosts. Comp. Biochem. Physiol. A 47:559-567. Hannah, R. W., S. J. Parker, and K. M. Matteson. 2008. Escaping the surface: the effect of depth of cap- ture on submergence success of surface-released Pacific rockfish. N. Am. J. Fish. Manag. 28:694—700. Hannah, R. W., and K. M. Matteson. 2007. Behavior of nine species of Pacific rockfish after hook-and-line capture, recompression, and release. Trans. Am. Fish. Soc. 136:24-33. Helvik, J. V., 0. Drivenes, T. Harboe, and H.-C. Seo. 2001. Topography of different photoreceptor cell types in the larval retina of Atlantic halibut ( Hippoglossus hippoglossus). J. Exp. Biol. 204:2553-2559. Herbert, N. A., and R. M. G. Wells. 2002. The effect of strenuous exercise and /3-adrenergic blockade on the visual performance of juvenile rain- bow trout, Oncorhynchus mykiss. J. Comp. Phyisol. B 172:725-731. Jokela-Maatta, M., T. Smura, A. Aaltonen, P. Ala-Lurila, and K. Donner. 2007. Visual pigments of Baltic Sea fishes of marine and limnic origin. Visual Neurosci. 24:389-398. Kaimmer, S. M., and R. J. Trumble. 1998. Injury, condition, and mortality of Pacific halibut by catch following careful release by Pacific cod and sablefish longline fisheries. Fish. Res. 38:131-144. Kobayashi, H. 1962. A comparative study on electroretinogram in fish, with special reference to ecological aspects. J. Shi- monoseki Univ. Fish. 11:408-538. Kvenseth, A. M., K. Pittman, and J. V. Helvik. 1996. Eye development in Atlantic halibut ( Hippo- glossus hippoglossus ): differentiation and develop- ment of the retina from early yolk sac stages through metamorphosis. Can. J. Fish. Aquat. Sci. 53:2524- 2532. Lalli, C. M., and T. R. Parsons. 1997. Biological Oceanography: an introduction, 2nd ed., 314 p. Butterworth and Heineman, Oxford, UK. Levine, J. S., and E. F. MacNichol. 1979. Visual pigments in teleost fishes: effects of habitat, microhabitat, and behavior on visual system evolution. Sens. Process. 3:95-131. Loew, E. R. 1976. Light, and photoreceptor degeneration in Norway lobster, Nephrops norvegicus. Proc. R. Soc. Lond. Ser. B. 193:31-44. Lythgoe, J. N. 1975. Problems of seeing colours under water. In Vision in fishes: new approaches in research (M. A. Ali, ed.), p. 619-634. Plenum Press, New York, NY. 1980. Vision in fishes: ecological adaptations. In Envi- ronmental physiology of fishes (M. A. Ali, ed.), p. 31- 445. Plenum Press, New York, NY. Mangel, S. C. 2001. Circadian clock regulation of neuronal light responses in the vertebrate retina. Prog. Brain Res. 131:505-518. Mesa, M. G., T. P. Poe, D. M. Gadomski, and J. H. Petersen. 1994. Are all prey created equal? A review and syn- thesis of differential predation on prey in substandard condition. J. Fish Biol. 45:81-96. Meyer-Rochow, V. B. 1994. Light-induced damage to photoreceptors of spiny lobsters and other crustaceans. Crustaceana 67:95- 109. Midtlyng, P. J., M. Ahrend, E. Bjerkas, R. Waagbp, and T. Wall. 1999. Current research on cataracts in fish. Bull. Eur. Assoc. Fish Pathol. 19:299-301. Brill et al.: Effects of rapid decompression and exposure to bright light on visual function in Sebastes melanops and Hippog/ossus steno/epis 437 Pacific Fishery Management Council. 2004. Status of the Pacific coast groundfish fishery through 2004, stock assessment and fishery evaluation: stock assessments and rebuilding plans, 506 p. Pacific Fishery Management Council, 7700 NE Ambassador Place, Suite 101, Portland, OR 97220-1384. Parker, S. J., H. I. McElderry, P. S. Rankin, and R. W. Hannah. 2006. Buoyancy regulation and barotrauma in two species of nearshore rockfish. Trans. Am. Fish. Soc. 135:1213-1223. Parker, S. J, J. M. Olsom, P. S. Rankin, and J. S. Malvitch. 2008. Patterns in vertical movements of black rockfish Sebastes melanops. Aquat. Biol. 2:57-65. Parker, S. J, and P. S. Rankin. 2003. Tag location and retention in black rockfish: fea- sibility of using PIT tags in a wild marine species. N. Am. J. Fish. Manag. 23:993-996. Parker, S. J., P. S. Rankin, R. W. Hannah, and C. B. Schreck. 2003. Discard mortality of trawl-caught lingcod in rela- tion to tow duration and time on deck. N. Am. J. Fish. Manag. 23:530-542. Parrish, F. A., and R. B. Moffitt. 1993. Subsurface fish handling to limit decompres- sion effects on deep-water species. Mar. Fish. Rev. 54:29-32. Reilly, C. R. L., and S. H. Thompson. 2007. Temperature effects on low-light vision in juvenile rockfish (Genus Sebastes ) and consequences for habitat utilization. J. Comp. Physiol. A 193:943-953. Rogers, B. L., C. G. Lowe, E. Fernandez-Juricic, and L. R. Frank. 2008. Utilizing magnetic resonance imaging (MRI) to assess the effects of angling-induced barotrauma on rockfish (Sebastes). Can. J. Fish. Aquat. Sci. 65:1245- 1249. Rummer, J. L. 2007. Factors affecting catch and release (CAR) mortal- ity in fish: Insight into CAR mortality in red snapper and the influence of catastrophic decompression. Am. Fish. Soc. Symp. 60:113-132. Rummer, J. L., and W. A. Bennett. 2005. Physiological effects of swim bladder overexpansion and catastrophic decompression on red snapper. Trans. Am. Fish. Soc. 134:1457-1470. St John, J., and C. J. Syers. 2005. Mortality of the demersal West Australia dhufish, Glaucosoma hebraicum (Richardson 1845) following catch and release: The influence of capture depth, venting and hook type. Fish. Res. 76:106-110. Stavenga, D. G., R. P. Smits, and B. J. Hoenders. 1993. Simple exponential functions describing the absor- bance bands of visual pigment spectra. Vision Res. 33:1011-1017. Stoner, A. W. 2003. Hunger and light level alter response to bait by Pacific halibut: laboratory analysis of detection, location and attack. J. Fish Biol. 62:1176-1193. Trumble, R. J., G. H. Williams, and S. E. Hughes. 1995. Methods to improve survival of Pacific halibut bycatch discarded from a factory trawler. In Proceed- ings of the international symposium on North Pacific flatfish, p. 591-610. Alaska Sea Grant College Program, Univ. Alaska Fairbanks, Fairbanks, AK. Vogel, J. L., and D A. Beauchamp. 1999. Effects of light, prey size, and turbidity on reaction distances of lake trout (Salvelinus mamaycush) to sal- monid prey. Can. J. Fish. Aquat. Sci. 56:1293-1297. Warrant, E. J. 1999. Seeing better at night: life style, eye design and the optimum strategy of spatial and temporal summation. Vision Res. 39:1611-1630. Wu, J., S. Seregard, and P. V. Algvere. 2006. Photochemical damage of the retina. Surv. Oph- thalmol. 51:461-481. 438 Abstract — Trawling and dredging on Georges Bank (northwest Atlantic Ocean) have altered the cover of colo- nial epifauna, as surveyed through in situ photography. A total of 454 pho- tographs were analyzed from areas with gravel substrate between 1994 and 2000 at depths of 40-50 m and 80-90 m. The cover of hydroids, bushy bryozoans, sponges, and tubeworms was generally higher at sites undis- turbed by fishing than at sites classi- fied as disturbed. The magnitude and significance of this effect depended on depth and year. Encrusting bryo- zoans were the only type of colonial epifauna positively affected by bottom fishing. Species richness of noncolo- nial epifauna declined with increased bottom fishing, but Simpson’s index of diversity typically peaked at inter- mediate levels of habitat disturbance. Species that were more abundant at undisturbed sites possessed charac- teristics that made them vulnerable to bottom fishing. These characteristics include emergent growth forms, soft body parts, low motility, use of com- plex microhabitats, long life spans, slow growth, and larval dispersal over short distances. After the pro- hibition of bottom fishing at one site, both colonial and noncolonial species increased in abundance. Populations of most taxa took two years or more to increase after the fishing closure. This finding indicates that bottom fishing needs to be reduced to infre- quent intervals to sustain the ben- thic species composition of Georges Bank at a high level of biodiversity and abundance. Manuscript submitted 7 August 2007. Manuscript accepted 1 July 2008. Fish. Bull. 106:438-456 (2008). The views and opinions expressed or implied in this article are those of the author and do not necessarily reflect the position of the National Marine Fisheries Service, NOAA. Changes in a benthic megafaunal community due to disturbance from bottom fishing and the establishment of a fishery closure Rebecca G. Asch (contact author)1 Jeremy S. Collie1 1 Graduate School of Oceanography University of Rhode Island Narragansett, Rhode Island 02882 'Present address for R. G. Asch: Scripps Institution of Oceanography University of California San Diego 9500 Gilman Drive, Mailcode 0208 La Jolla, California 92093-0208 Commercial fishing with mobile gear (e.g., otter trawls, beam trawls, scal- lop dredges) has become such a wide- spread practice that it is estimated that an area equivalent to approxi- mately 75% of the world’s continental shelves is trawled annually (Kaiser et al., 2002). Because bottom fishing can kill and injure benthic invertebrates that are either caught as bycatch or crushed underneath fishing gear, heavily fished areas often exhibit decreased abundance, biomass, and diversity of epifauna (Collie et al., 1997). Trawling and dredging can modify the composition of marine sed- iments through the dispersal of piles of boulders (Collie et al., 1997; Freese et al., 1999) and the resuspension of fine sediments. The latter process may subsequently release contaminants and excess nutrients, expose anoxic sediment layers, smother filter feed- ers, and alter biogeochemical fluxes (Kaiser et al., 2002). Bottom fishing also affects benthic community struc- ture by augmenting the abundance of scavengers (Ramsay et al., 1998) and depleting organisms that serve as ecosystem engineers or keystone spe- cies (Auster et al., 1996). Similarly, ecosystem function and population dynamics can be indirectly altered because bottom fishing may reduce levels of epifaunal productivity (Herm- sen et al., 2003), lower the diversity of prey available to demersal fish (Jiang and Carbines, 2002), and decrease the structural complexity of the benthic environment (Auster et al., 1996). A meta-analysis of the effects of bottom fishing revealed that less than one-half of the studies address- ing this topic examined the long-term recovery of the benthic community (Collie et al., 2000b). In studies where recovery rates were investigated, most tracked recovery for a period of less than two years, which may not be long enough to evaluate the final outcome of ecological succession in disturbed habitat patches. For ex- ample, at the completion of a study that involved experimental trawling in a Scottish sea loch, physical signs of trawl damage had disappeared af- ter 18 months, but differences in the benthic community structure at treat- ment and reference sites persisted (Tuck et al., 1998). Similarly, at the end of a six-month study of the effects of experimental trawling, the poly- chaete Terebellides atlantis and the nemertean Cerebratulus lacteus had not yet returned to pretrawling levels of abundance (Sparks-McConkey and Watling, 2001). Knowledge of epifau- nal recovery from natural and fish- ing-induced disturbance is especially lacking on the continental shelf, an area subjected to heavy bottom fish- ing. Compared to intertidal zones and shallow subtidal areas where benthic responses to natural disturbance have long been studied, relatively little re- search has been undertaken to track the colonization of sessile epifauna to hard substrata at depths greater than 30 m after natural or anthro- pogenic disturbances. In the Gulf of Maine, those studies on this subject (Sebens et al., 1988; Witman, 1998) had relatively short durations (less than two years), which may not have Asch and Collie: Changes in a benthic megafaunal community due to disturbance from bottom fishing 439 provided adequate time for the completion of ecological succession. We evaluated fluctuations in the abundance of colonial and noncolonial epifauna in benthic photographs taken between the years 1994 and 2000 in areas of Georges Bank (northwest Atlantic) that have been classified as either disturbed or undisturbed by bottom fishing. Colonial epifauna are ecologically important because many taxa generate three-dimensional microhabitats that augment the structural complexity of the benthic environment. These microhabitats may directly benefit invertebrates and demersal fishes by aggregating food sources and providing refuge from visual predators (Henkel and Pawlik, 2005). Comparisons were made between two depth strata in order to determine how depth may modulate the response of epifauna to bot- tom-fishing disturbance. Because noncolonial species of megafauna were examined in both photographs and previously analyzed benthic samples collected at the same study sites with a 1-m wide “naturalist” dredge (a dredge design described in Eleftheriou and Holme, 1984), trends in noncolonial epifauna distribution, abun- dance, and species composition are analyzed to facilitate comparisons between sampling techniques. The effect of bottom fishing on the diversity of noncolonial species is also examined. One of our study sites was located in an area where bottom fishing has been prohibited since December 1994. Because this site had previously been fished with mobile gear, much of its resident epifauna had been destroyed at the beginning of our study. This situation allowed us to examine patterns of ecological recovery. Our research represents one of the first published stud- ies conducted in this geographic region on the long-term processes (i.e., greater than two years) affecting ecologi- cal recovery from disturbance among colonial epifauna on the continental shelf. Materials and methods Description of study sites In December 1994, bottom fishing was prohibited in three large sections of Georges Bank and the southern New England continental shelf to decrease levels of fish- ing mortality on Atlantic cod ( Gadus morhua ), haddock ( Melanogrammus aeglefinus), and yellowtail flounder ( Limanda ferruginea). Closed (fishing) area II (CA-II), which is the focus of the current research, is located 150 km offshore on the Northeast Peak of Georges Bank (Fig. 1A). The northern section of CA-II where the seafloor is principally covered by gravel substrate was designated as a habitat area of particular concern (HAPC) in 1998. The gravel substrate in this area forms an important habitat for Atlantic sea scallops {Placopecten magellani- cus), provides a nursery ground for juvenile G. morhua and M. aeglefinus, and serves as a spawning ground where Atlantic herring ( Clupea harengus ) deposit their demersal eggs (Lough et al., 1989). Sections of the gravel substrate contain dense patches of sponges, hydroids, bushy and encrusting bryozoans, colonial ascidians, and the lacy tubeworm ( Filograna implexa) (Collie et al., 2000a; Valentine et ah, 2007). During the year before the establishment of CA-II, scientists from the University of Rhode Island (URI) and the United States Geological Survey (USGS) trav- eled to this area on two occasions to conduct sidescan sonar surveys, collect samples of benthic megafauna, and record the fauna observed on the seafloor along transect lines by means of video and still cameras. Ini- tial assessments of the benthic megafauna and colonial epifauna at these study sites were published in Collie et al. (1997, 2000a). Between 1994 and 2000, the URI and USGS research team returned to Georges Bank on a nearly annual basis to track the recovery of the megafaunal community in CA-II and to evaluate inter- annual variations in community structure at other sites (Collie et al., 2005). Because some sites on Georges Bank contain scat- tered boulders that can damage fishing gear, certain ar- eas are generally avoided by fishermen and can, there- fore, be classified as undisturbed by bottom fishing. To identify which regions were disturbed by bottom fishing, a 10-km2 area at six different sites was surveyed with a 100-Khz high-resolution sidescan sonar in April 1994. Sonograms from these surveys were inspected for trawl and dredge marks. Areas containing such marks were classified as disturbed, and areas with no visible signs of recent trawling or dredging activity were considered undisturbed. Sediment maps were consulted to ensure that only sites dominated by gravel substrate were sam- pled (Valentine et al., 1993). Disturbed and undisturbed sites were examined at shallow depths (40-50 m) and at deeper depths (80-90 m) in U.S. and Canadian waters (Collie et al., 1997). The original classification of study sites as disturbed or undisturbed was validated with data on scallop dredging and bottom trawling effort provided by the National Marine Fisheries Service (NMFS) and the Canadian Department of Fisheries and Oceans (DFO) (Table 1). For sites located within the U.S. Economic Exclusive Zone, data from the NMFS satellite vessel monitoring program were used to compute the num- ber of hours that scallop fishing boats spent dredging each square nautical mile of Georges Bank from 1998 through 2000 (Collie et al., 2005). Before 1998, the clas- sification of sites as disturbed or not disturbed on the U.S. portion of Georges Bank was verified with NMFS data on scallop dredging effort which had a 10-minute latitude resolution. At shallow sites, both NMFS data sets consistently confirmed the accuracy of the original disturbance classifications that had been based on sid- escan sonar surveys. The disturbance classifications at Canadian sites were authenticated with data from DFO logbook reports from the commercial scallop fishery that had a 1-minute latitude resolution. Because the number of days fished at deep sites varied both inside individual study sites and between years (Collie et al., 2005), we decided that the classifications of sites located at the 440 Fishery Bulletin 106(4) 68 °W 67°W i ■ i ' i i ■ i i 67°W 67°W I i i i I i i i I I- >--t -i- -l- i i i. J Key: Wi Sandy Transects © Undisturbed Transects • Disturbed Transects Figure 1 (A) Map of the Northeast Peak of Georges Bank displaying closed area II and the study sites surveyed between 1994 and 2000. (B) and (C) Maps of the starting points of each photographic transect conducted at shallow (B) and deep (C) sites. Transects were clas- sified as sandy if more than 50% of the photographs analyzed contained >50% cover of sand. Numbers on each map identify study sites. The classification of disturbance at some sites (i.e., 17 and 18) changed throughout the duration of this study. The lines in the background of each map show the bathymetry of Georges Bank; each line represents a 20-m change in depth. deeper depth strata should be recategorized. According to our revised classification scheme, any photographed transect at a deep site that was fished with a scallop dredge for one day or less per year was classified as undisturbed. Based on disturbance classification, sediment compo- sition, and depth, five 5 kmx 10 km sites were selected for examination during the first year of this study (Fig. 1). Three of these study sites (i.e., sites 10, 13, and 20) were located in deeper water on the Canadian sec- tion of Georges Bank. At the 80-90 m depth stratum, site 20 was classified as undisturbed, site 13 was clas- sified as disturbed for all years except 1998 when it contained two undisturbed transects, and site 10 was sampled only in 1994 when it contained one disturbed and one undisturbed transect (Table 1). On the shal- lower U.S. portion of the bank, two study sites were originally surveyed, one corresponding to an area with little disturbance from bottom fishing (i.e., site 18) and the other representing a heavily fished zone (i.e., site 17). After the incorporation of site 17 into CA-II, much of the fishing effort formerly concentrated at this site was displaced into areas that were unaffected by the closure, including site 18. Consequently, the initial clas- Asch and Collie: Changes in a benthic megafaunal community due to disturbance from bottom fishing 441 Table 1 Classifications of disturbance and mean scallop dredging effort at study sites based on data from the National Marine Fisheries Service (NMFS) and the Canadian Department of Fisheries and Oceans (DFO). Dashes indicate years when no photographic samples were collected. D and U indicate shallow sites that were classified as “disturbed” or “undisturbed,” respectively, before the development of the NMFS satellite vessel monitoring program in 1998. Classifications made at shallow sites before 1998 were based on trawl and dredge marks seen in side can sonar images. No photographic data were gathered in 1995. Dredging effort from U.S. and Canadian sites cannot be easily compared because data were reported in different units and the type of fishing gear used by each country may have varied. Site Classification of disturbance 1994 1996 1997 1998 1999 2000 Shallow sites: Mean dredging effort in hours fished per year within two miles (3.2 km) of site 16 Disturbed — D D — — — 17 Disturbed before establishment of a fishery closure (i.e., 1994) and undisturbed afterwards (i.e., 1996-2000) D U U 0 0 0 17W Disturbed — — D 172.6 58.1 — 18 Undisturbed before establishment of a closed area (i.e., 1994) and disturbed afterwards (i.e., 1996-2000) U D D 97.4 1.0 1.2 Deep sites: Mean dredging effort in days fished per year within a square minute latitude of a transect 10 Mildly to moderately disturbed 2.5 — — — — — 13 Disturbed, except for two transects surveyed in 1998 3.5 5.0 4.0 2.0 — — 20 Undisturbed 0 0.2 0.1 0 0 — sification of bottom-fishing-induced disturbance at sites 17 and 18 was reversed in 1995. In order to evaluate the effectiveness of the fishery closure, other disturbed sites were sampled surrounding CA-II (i.e., sites 16 and 17W) in later years. Fieldwork and laboratory procedures Photographs of the seafloor of northeastern Georges Bank were collected during seven research cruises, which were conducted on a nearly annual basis from November 1994 through November 2000. During all cruises, except for one occurring in July 1995, the SEABed Observation and Sampling System (SEABOSS; Woods Hole Science Center, U.S. Geological Survey, Woods Hole, MA) was deployed to take 35-mm pho- tographs of the seafloor. The SEABOSS consists of a tethered van Veen grab sampler that is equipped with two Hi-8 video cameras, one still camera, accompany- ing lights, and a pressure-depth sensor (Blackwood and Parolski, 2001). The SEABOSS also contains two parallel lasers placed 20 cm apart, which are used to gauge the size of objects on the seabed. The frame of the SEABOSS instrument was attached to a winch and allowed to drift under the research vessel as it traveled at a speed of 1-2 knots (0. 5-1.0 m/s). At each study site surveyed during a research cruise, three transects were generally conducted with the SEABOSS. However, depending on weather conditions, research priorities, and available ship time, anywhere between one and ten transects were filmed. Because of changes in tidal speed, transect length varied between 183 and 4316 m (mean distance of 921 m, standard deviation [SD] = 527 m). Along transects, photographs were taken at 30-60 s intervals. In addition to the SEABOSS, during the 1996 and 1999 research cruises, videos and photographs of the seafloor were also taken with a MaxRover MK1 remotely operated vehicle (ROV; Deep Sea Systems International, Inc., Falmouth, MA) provided by the National Undersea Research Center at the University of Connecticut. This ROV system included a set of two 35-mm still cameras and two Hi-8 video cameras. As with the SEABOSS, two parallel lasers placed 20 cm apart were used to estimate the size of objects on the seafloor. During the 30-minute ROV transects (approximately 1000 m in length), photographs of the seabed were snapped at 30-second intervals. Each pho- tograph taken with the SEABOSS covered an area of approximately 0.25-0.27 m2 of the seafloor, whereas approximately 0.31 m2 of the sea bottom was examined in each ROV photo. Because of the impracticality of analyzing the hun- dreds of photographs taken during each cruise, we ran- domly selected a subsample of 12-30 photographs per site (average = 16) to be examined in detail each year. As a result of the scarcity of quality photos (i.e., not blurry or sandy), a reduced number of photographs were chosen from site 13 in 1996 and site 20 in 1999. A total of 454 benthic photographs from 94 transects surveyed from 1994 and 2000 were analyzed. These photographs covered an area of 120.8 m2 of the seafloor. Following the method described in Collie et al. (2000a), we over- 442 Fishery Bulletin 106(4) laid photographs with a transparency containing a grid, in which each grid cell represented a 5 cm x 5 cm area of the seafloor. From 1996 through 2000, the percent cover of hydroids, bushy and encrusting bryozoans, sponges, and F. implexa was recorded in each grid cell. Data were summed across grid cells to calculate the total percent cover of each type of colonial organism per photograph. In 1994, the percent frequency of several types of colo- nial epifauna was measured, instead of percent cover. Results of the 1994 surveys were presented in Collie et al. (2000a). In addition, noncolonial, megafaunal species were enumerated in each grid cell and identified to the lowest possible taxonomic level. Sediment type and the number of pieces of cultch (i.e., broken bivalve shells) were also recorded during analyses of the photographs. Like colonial epifauna, some fish species use the three- dimensional structure generated by cultch to obtain shelter from predators (Auster et al., 1995). Because hard bottom (e.g., cobble and gravel) and soft bottom (e.g., sand, silt, and mud) environments support fundamentally different benthic communities, we decided to remove from our data set those transects where sand constituted a large percentage of the sub- stratum. Transects where the majority of photographs contained greater than 50% sand cover were usually clustered around a distinct area or were located at the far edge of our study sites (Fig. IB). To guarantee that only gravel habitat was examined, 34 photographs from sandy transects were removed from the data set. Once photographs that were sandy, blurry, or overexposed had been selected and removed, 386 photographs re- mained, covering an area of 100.1 m2. Throughout this study, benthic photographs were analyzed by five observers. To evaluate the extent to which between year and within-year fluctuations in epifauna abundance might reflect observer bias, calibra- tion tests were performed during which two observers examined the same photograph(s). T-tests, in which photographs analyzed by different observers were treat- ed as matched pairs, were used to determine whether observer bias affected estimates of epifaunal abun- dance. Results indicated that, with the exception of the coiled worm ( Spirorbis spp.), patterns of observer bias did not match the direction and magnitude of between- year and within-year variations in epifaunal cover or megafaunal abundance (Asch, 2006). Data on Spirorbis spp. are not presented here because of concerns about observer bias. Statistical analyses Colonial epifauna A series of two-way analysis of vari- ance (ANOVA) tests were conducted to investigate which taxa of colonial epifauna exhibited significant differences in percent cover between disturbance cat- egories and years. Because large variations in organ- ismal abundance with depth may overshadow subtler fluctuations related to bottom fishing, each depth stra- tum was considered separately in the ensuing analyses. Response variables included six measures of the cover of colonial epifauna taxa and cultch. A previous exami- nation of spatial patterns indicated that autocorrelation existed between photos from the same transect (Asch, 2006). Therefore, percent cover from photos were aver- aged across transects, allowing us to use transects as our primary sampling unit when conducting parametric tests. All response variables were arcsine square-root transformed in order to ensure that percent cover data would conform to the normal probability density func- tion. Bottom-fishing disturbance classifications and year were used as factors in the ANOVAs. At shallow sites, the interaction term {disturbance x year) was used to evaluate whether differences between sites inside and outside of CA-II increased over time because of the continued recovery of organisms in CA-II. Since no data from a deep, disturbed site were collected in 1999, this particular year was removed from the data set when we considered the deeper depth stratum, so that the experimental design would be balanced. Because of the nonorthogonal design of these ANOVAs, type-III sums of squares were used. When interannual differences were indicated to be an important factor affecting a particular type of colonial epifauna, Tukey-Kramer multiple comparisons tests were applied to determine which years differed significantly (S-Plus 6, MathSoft, Inc., Seattle, WA). Noncolonial organisms Although many noncolonial organisms were identified to the species level, other organisms could only be identified to genus or family because of the limited resolution of photos or the need for microscopic examination of distinguishing features. In cases where some organisms belonging to a par- ticular taxonomic group could be identified to species, but others could not, all members of the taxonomic group in question were lumped together in order to guarantee that classifications were mutually exclu- sive. Ninety-eight percent of the noncolonial organisms identified in photographs belonged to five extremely abundant taxa: Spirorbis spp., the tubeworm Protula tubularia, the jingle shells Anomia spp., the barnacles Balanus spp., and an unidentified species of burrowing anemones (order Ceriantharia). For each depth and disturbance category, the R statistical package (Free Software Foundation, Boston, MA) was used to fit the negative binomial distribution to the data on the abun- dance of these five species. Next, a U test was applied to evaluate the goodness-of-fit of the negative binomial distribution (Krebs, 1999). Because in all cases but one (i.e., Anomia spp. at deep, undisturbed sites) the nega- tive binomial distribution provided an adequate fit, we determined that these five highly abundant species have an aggregated spatial distribution, whereby they were absent from many photographs but obtained very high densities in a few areas. For example, Anomia spp. and Balanus spp. were not identified in two-thirds of the photographs sampled but were occasionally observed at concentrations as high as 1328 per m2 and 760 per m2, respectively. Because random sampling of such dense aggregations could result in the detection of artificial Asch and Collie: Changes in a benthic megafaunal community due to disturbance from bottom fishing 443 differences among depths, disturbance categories, and years, data on the abundance of these five species are not presented in this manuscript. However, information on trends in the abundance of these five taxa can be referenced in Asch (2006). All other noncolonial taxa were collectively exam- ined as a multivariate data set. Differences in the species composition of these noncolonial organisms among depth strata, years, bottom-fishing disturbance categories, and sites were explored by constructing nonmetric multidimensional scaling (MDS) plots in PRIMER 6 (Plymouth Marine Lab, Plymouth, UK). MDS plots were constructed with a Bray-Curtis simi- larity matrix containing information on the mean number of individuals of each taxon per photograph, aggregated across each transect (Clarke and Warwick, 2001). To ensure that patterns in the MDS plot were not dictated solely by trends in the abundance of a few ubiquitous species, data were square-root transformed before this analysis and all other multivariate statis- tical routines (i.e., analysis of similarity [ANOSIM] and similarity of percentages [SIMPER]). ANOSIM tests were used to evaluate whether differences in noncolonial species composition between disturbance categories and years were statistically significant (Clarke and Warwick, 2001). In cases where signifi- cant differences were detected, the SIMPER routine in PRIMER 6 was used to determine the mean percent dissimilarity that each species contributed to differ- ences between disturbance categories (Clarke and Warwick, 2001). In addition to examining the effect of bottom-fish- ing-induced disturbance on species composition, its influence on two measures of noncolonial diversity was investigated. Jackknife species richness and Simpson’s index of diversity (1 — A') were calculated according to the methods described in Krebs (1999). These two diversity indices were selected, because 1) they are either unaffected by differences in sample size (e.g., Simpson’s index) or incorporate estimates of the num- ber of rare species that are unlikely to be sampled (e.g., jackknife species richness) (Clarke and Warwick, 2001), and 2) together these indices measure several aspects of biodiversity (e.g., richness, heterogeneity, and evenness). The diversity indices were computed for five levels of disturbance at both shallow and deep sites. Data were binned into five disturbance levels to achieve sufficient sample sizes because the jack- knife species richness index requires that the size of samples is large enough to include at least half of the species in an area (Krebs, 1999). Only data from shallow sites collected during years when the NMFS satellite vessel monitoring program was in operation (i.e., 1998-2000) were analyzed. Linear and polynomial regressions were performed to evaluate the relationship between bottom-fishing disturbance and these two measures of noncolonial diversity. The most parsimonious regression model was selected on the basis of the results of the analysis of deviance test included in the R statistical package. Results Colonial epifauna At shallow sites, the cover of five out of six taxa of colonial epifauna differed significantly between undis- turbed and disturbed sites located, respectively, inside and outside of CA-II (Fig. 2). Hydroids were the only taxon of colonial epifauna whose percent cover was not significantly affected by disturbance at shallow sites. During most years, structurally complex taxa of colonial epifauna, such as sponges and bushy bryozoans, were more abundant at shallow, undisturbed sites, whereas encrusting taxa, such as encrusting bryozoa, cultch, and F. implexa, exhibited a higher percent cover at shallow, disturbed sites (Fig. 2). With the exception ofF. implexa whose cover at each shallow site remained fairly stable throughout the time series, all other colonial epifauna taxa demonstrated significant between-year variations in abundance at shallow sites. Several of these taxa showed marked changes in abundance beginning in 1997 and 1998, indicating that it took at least two years for these species to respond to the establishment of CA-II or to the increased bottom-fishing effort at sites that remained open to mobile fishing gear. For example, sponges were absent from all photographs taken at the shallow site in CA-II during 1994 and 1996 but were seen in 97% of the photographs taken in this area in 1997 (Figs. 2 and 3A). Similarly, bushy bryozoans experienced a temporary spike in their percent cover at the shallow, undisturbed site in 1997. The cover of encrusting bryozoans was fairly similar at all shallow sites at the initiation of this study, but then began to increase in disturbed areas beginning in 1998. By the year 2000, 62% of the substratum at shallow, disturbed sites was covered by encrusting bryozoa (Fig. 3B). Tem- poral variations in the abundance of hydroids and cultch are unlikely to reflect changes in bottom-fishing effort because similar between-year changes were seen at all shallow sites regardless of closure status. The only two taxa that clearly exhibited significant differences between disturbed and undisturbed sites at the deep depth stratum were F. implexa and hydroids, both of which consistently had an elevated percent cover in undisturbed areas (Fig. 4). Bushy bryozoans also had a higher cover in 1997 and 1998 in deep areas that experienced little to no bottom fishing. However, this pattern was reversed in 1996 when deep, disturbed sites exhibited slightly greater cover of bushy bryozoans than deep, undisturbed sites. Cultch and encrusting bryozoans at deep sites were frequently more abundant in disturbed areas, but this trend was inconsistent be- tween years, causing ANOVA results to be only margin- ally significant (i.e., P=0.08 and P=0.07 for cultch and encrusting bryozoans, respectively). Sponge abundance was depressed at all deep sites from 1996 through 1998, and its mean percent cover never exceeded 0.5% during these years. Temporal variations in colonial epifauna cover appeared to be of lesser importance at deep sites than at shallow sites because encrusting bryozoans were 444 Fishery Bulletin 106(4) Shallow sites Filograna implexa Encrusting bryozoans 0.8 -| 0.6 - 0.4 - 0.2 o May-96 May-97 Fdisi = 5.22*; F , Undisturbe^ May-98 May-99 May-00 V' : 0.14; F. dist x yr = 0.19 Hydroids Bushy bryozoans Sponges Cultch frequency Figure 2 Temporal trends in colonial epifaunal percent cover and cultch percent frequency at shallow sites (40-50 m). Trend lines from sites that were either disturbed or undisturbed by bottom fishing are shown. Shallow undisturbed data came from site 17, whereas shallow disturbed data were derived from sites 16, 17W, and 18. The scale of the y axis varies between graphs of different epifaunal taxa. Markers (i.e., black squares and black diamonds) are back-trans- formed means, and error bars are 95% confidence intervals. Different letters above data points represent annual means that differ significantly from each other in multiple comparison tests. F statistics from two-way analyses of variance (ANOVAs) on the effects of disturbance (dist), year (yr), and disturbance x year (distxyr) are shown below each graph (df=l,4,4). Significance levels from ANOVA tests are indicated as follows: *P<0.05; **P <0.01; ***P<0.001. the only taxon to have a significant year term in the two-way ANOVAs performed on deep sites (Fig. 4). The greater between-year variability at shallow sites can be interpreted as a sign of the effectiveness of CA- II. Because there can be subtle, pre-existing differences in habitat quality between areas inside and outside a marine reserve, a conclusive demonstration of reserve effectiveness typically requires that biological distinc- tions between fished and unfished areas become am- plified over time (Murawski et al., 2004). As a result, variables truly affected by the establishment of CA-II should have significant interaction terms in their corre- sponding two-way ANOVA, as well as exhibit significant differences between disturbance categories and among years. Sponges and encrusting bryozoans were the two colonial epifauna taxa that exhibited significant in- teraction terms in the ANOVAs performed for shallow sites. No interactions between disturbance and year were detected at the deep sites. This finding reflects the fact that bottom-fishing effort was fairly consistent over time at deep sites (Table 1; Collie et ah, 2005), whereas fishing effort at shallow sites showed a more distinct temporal trajectory related to the establish- ment of CA-II. Asch and Collie: Changes in a benthic megafaunal community due to disturbance from bottom fishing 445 Figure 3 Photographs of the seafloor taken in 1999 at sites 17 (A) and 18 (B). (A) Sponges (sp) cover 10.4% of the seafloor in this photograph, and bushy bryozoans (bb) cover an additional 2.5%. Species of noncolonial epifauna seen in this photograph include the sea scallop iPlacopecten magellanicus [pm]), the Northern seastar (Asterias vulgaris [av]), the green sea urchin ( Strongylocentrotus droebachiensis [sdj ), and a bur- rowing anemone (ba). (B) Encrusting bryozoans cover 74.2% of the seafloor in this photo from a shallow, disturbed site. Areas where the gravel substrate is a light grey color contain encrusting bryozoans. Two pieces of cultch (cu) are visible, as well as A. vulgaris (av) and a northern cyclocardia ( Cyclocardia borealis [ c b ] ) . Photographs by Dann Blackwood, Woods Hole Science Center, U.S. Geological Survey, Woods Hole, MA. Noncolonial organisms In the 454 benthic photographs that were analyzed, a total of 117,315 noncolonial organisms were identified, which included 73 distinct taxa. After removing data on five abundant species with extremely aggregated spatial distributions, we found that this data set contained 1561 individuals. Clear differences between deep and 446 Fishery Bulletin 106(4) Deep sites Filograna implexa Encrusting bryozoans Hydroids Bushy bryozoans Sponges Cultch frequency Figure 4 Temporal trends in colonial epifaunal percent cover and cultch percent frequency at deep sites (80 — 90 m). Trend lines from sites that are either disturbed or undisturbed by bottom fishing are shown. Data representing deep, disturbed sites originated from sites 10 and 13, and data representing deep, undisturbed sites were primarily from site 20, but also included a few unfished transects from sites 10 and 13. The scale of the y axis varies between graphs of different epifaunal taxa. Markers (i.e. , black squares and black diamonds) are back-transformed means, and error bars are 95% confidence intervals. Different letters above data points represent annual means that differed significantly from each other in multiple comparison tests. F statistics from two-way analyses of variance (ANOVAs) on the effects of disturbance ( dist ), year (yr), and disturbance x year ( distxr ) are shown below each graph ( df= 1,2,2). Significance levels from ANOVA tests are indicated as follows: ?* P<0.10; *P<0.05; **P<0.01. There are no lines connecting data from 1998 and 1999 at deep, undisturbed sites because the 1999 data were not included in two-way ANOVAs. shallow transects in terms of noncolonial megafaunal abundance were apparent when we examined MDS plots (Fig. 5A). Compared to the evident effect of depth on noncolonial species composition, the influence of bottom- fishing disturbance in MDS plots was subtle. Transects recorded at disturbed site 18 could be distinguished from transects at all other shallow sites. Relatively few differences in noncolonial species composition could be seen among sites 16, 17, and 17W (Fig. 5B). At the deep depth stratum, disturbed sites 10 and 13 generally exhibited a community composition that was distinct from undisturbed site 20, as indicated by the position- ing of disturbed transects to the left of the MDS plot (Fig. 5C). The two transects from site 13 that were more interspersed among transects from site 20 in the MDS were filmed in the northwestern extent of this study site, which was subjected to less bottom-fishing pressure. The greater dispersion of deep, undisturbed transects across the MDS revealed a wider variation in species composi- tion at site 20 than at other deep sites. Asch and Collie: Changes in a benthic megafaunal community due to disturbance from bottom fishing 447 T ransect depth A Shallow ▼ Deep Shallow transects Site 16 A Site 17 ♦ Site 17W • Site 18 Deep transects A Site 10 disturbed A Site 10 undisturbed ■ Site 13 disturbed □ Site 13 undisturbed O Site 20 undisturbed Figure 5 Nonmetric multidimensional scaling (MDS) plot describing varia- tions in noncolonial species composition among photographic transects. MDS is an ordination technique that produces two- dimensional “maps” of samples reflecting differences in their bio- logical communities. The distance between points is proportional to the dissimilarity between different photographic transects in terms of their noncolonial epifaunal community structure. The axes on MDS diagrams are arbitrary and can be flipped and rotated. Stress refers to the ordination’s goodness-of-fit when multidimensioal data are projected onto a two-dimensional plane. (A) All transects, (B) Shallow transects, (C) Deep transects. The majority of noncolonial species were most abun- dant at undisturbed sites, but hard-shelled mollusks and scavengers tended to dominate in heavily fished areas. Significant variations in community structure were detected at both depths between different dis- turbance levels and years by means of ANOSIM tests (Tables 2A and 3A). A total of six taxa contributed to 50% of the cumulative dissimilarity between disturbed and undisturbed sites at the shallow depth stratum (Table 2B). Of these six taxa, P. magellanicus, the green sea urchin ( Strongylocentrotus droebachiensis), and the hermit crabs Pagurus spp. exhibited higher levels of abundance at shallow sites inside CA-II. Scavenging seastars ( Asterias spp.) and bivalves (astartes Astarte spp. and the northern cyclocardia Cyclocardia borealis) were more abundant along shallow, disturbed transects recorded outside the closed area. The elevated numbers of the latter two species at site 18, as well as siphons from an unidentified, infaunal bivalve, explained why transects from site 18 were separated from other shal- low sites in the MDS (Fig. 5B). Of the four species that defined 88% of the similarity among undisturbed 448 Fishery Bulletin 106(4) Table 2 (A) Effects of disturbance from mobile fishing gear and year on the composition of noncolonial organisms at shallow sites. In this two-way crossed analysis of similarity (ANOSIM) test, the sample global R is the test statistic. Data were per- muted 9999 times to generate an empirical frequency distribution for global R. Significance levels: *P<0.05; ***P<0.001. (B) Similarity of percentages (SIMPER) results identifying species that are characteristic of disturbed and undisturbed sites at the shallow depth stratum. Cumulative percentage refers to the contribution that species make to either the cumulative similar- ity within each group or the cumulative dissimilarity between two groups. Results are presented in terms of similarity among disturbed transects, similarity among undisturbed transects, and dissimilarity between disturbance groups. A ANOSIM results 95th percentile Factor of permuted global R’s Sample global R Disturbance 0.14 0.20* Year 0.12 0.30*** B SIMPER results Similarity among disturbed transects (i.e., outside CA-II) Mean density Cumulative Scientific name per photograph percentage Pagurus spp. 0.40 27.55 Asterias spp. 0.32 48.41 Placopecten magellanicus 0.59 63.17 Siphons of unidentified bivalves 0.51 74.24 Cyclocardia borealis 0.24 82.70 Similarity amongst undisturbed transects ( i.e., inside CA-II) Mean density Cumulative Scientific name per photograph percentage Placopecten magellanicus 0.72 43.58 Pagurus spp. 0.42 67.24 Strongylocentrotus droebachiensis 0.30 78.10 Asterias spp. 0.27 88.39 Dissimilarity between disturbance groups Mean density Mean density Cumulative Scientific name at disturbed sites at undisturbed sites percentage Placopecten magellanicus 0.59 0.72 12.68 Siphons of unidentified bivalves 0.51 0.09 21.74 Pagurus spp. 0.40 0.42 30.46 Asterias spp. 0.32 0.27 39.07 Strongylocentrotus droebachiensis 0.04 0.30 46.32 Astarte spp. 0.47 0.06 52.49 transects inside CA-II (i.e., P. magellanicus, Pagurus spp., S. droebachiensis, and Asterias spp., Table 2B), all increased in abundance from 1994 through 2000 (Figs. 3A and 6). Of particular note was the nearly tenfold increase in the mean density of P. magellanicus during this time period. At deep sites, seven of the ten taxo- nomic groups that contributed the most to dissimilar- ity among disturbance categories were more abundant along undisturbed transects (Table 3B). As was the case at shallow sites, hard-shelled mollusks ( Astarte spp. and the waved whelk [. Buccinum undatum ]) and scavengers, such as Pagurus spp., characterized deep, disturbed sites. At both depths, species richness of noncolonial epi- fauna peaked along undisturbed transects (Fig. 7). The negative relationship between species richness and hot- Asch and Collie: Changes in a benthic megafaunal community due to disturbance from bottom fishing 449 Table 3 (A) Effects of disturbance from mobile fishing gear and year on the composition of noncolonial organisms at deep sites. In this two-way crossed analysis of similarity (ANOSIM) test, the sample global R is the test statistic. Data were permuted 9999 times to generate an empirical frequency distribution for global R. Significance levels: *P<0.05; ***P<0.001. (B) Similarity of percent- ages (SIMPER) results identifying species that are characteristic of disturbed and undisturbed sites at the deep depth stratum. Cumulative percentage refers to the contribution that species make to the cumulative similarity within each group or the cumu- lative dissimilarity between these groups. Results are presented in terms of similarity among disturbed transects, similarity among undisturbed transects, and dissimilarity between disturbance groups. A ANOSIM results 95th percentile Factor of permuted global R’s Sample global R Disturbance 0.25 0.28* Year 0.11 0 4i*** B SIMPER results Similarity amongst disturbed transects Mean density Cumulative Scientific name per photograph percentage Astarte spp. 0.55 21.75 Pagurus spp. 0.55 42.10 Caridea unident. 1.03 59.65 Gastropoda unident, (shelled) 0.48 70.37 Zoantharia unident. 0.23 79.04 Placopecten magellanicus 0.15 86.93 Similarity amongst undisturbed transects Mean density Cumulative Scientific name per photograph percentage Zoantharia unident. 1.74 35.80 Astarte spp. 0.64 47.03 Pagurus spp. 0.44 55.14 Caridea unident. 3.52 62.44 Modiolus modiolus 0.53 69.59 Gastropoda unident, (shelled) 0.39 76.16 Hyas coarctatus 0.33 81.47 Dissimilarity between disturbance groups Mean density Mean density Cumulative Scientific name at disturbed sites at undisturbed sites percentage Zoantharia unident. 0.23 1.74 11.32 Caridea unident. 1.03 3.52 20.25 Astarte spp. 0.55 0.64 26.33 Modiolus modiolus 0 0.53 32.24 Thelepus cincinnatus 1.14 0.25 38.12 Pagurus spp. 0.55 0.44 43.90 Gastropoda unident, (shelled) 0.48 0.39 49.52 Hyas coarctatus 0.05 0.33 53.97 Ophiuroidea unident. 0.05 0.75 58.34 Polychaeta unident, (tubicolous) 0.29 0.33 62.68 450 Fishery Bulletin 106(4) > -a c o Sampling date Figure 6 Trends in the abundance of four noncolonial species characteristic of site 17 in closed area II (CA-II). Each black square represents a single photographic transect, and solid lines connect the means from each sampling date. The shaded area indicates the time period before bottom fishing was restricted in CA-II. tom-fishing effort was statistically significant at shal- low sites. Average species richness was greater at deep sites than at shallow sites (Sd = 22 and Sshallow = 13), but these two depths did not differ in terms of mean values of Simpson’s index of diversity (1— A'd =0.77 and 1— A's/iaWoUJ=0.76). The highest values of Simpson’s index at shallow sites were recorded in areas with 50-100 hours of scallop dredging per year. At greater levels of bottom-fishing disturbance, a decline in Simpson’s index was observed. At deep sites, Simpson’s index varied relatively little with disturbance, although its peak value occurred in an area with an intermediate amount of disturbance (i.e., three days of dredging per year). Trends in Simpson’s index were not significant when modeled with linear and polynomial regression. Discussion Colonial epifauna Because sessile, colonial organisms cannot avoid the path of mobile fishing gear nor can they quickly immi- grate into recently disturbed areas, they may be more adversely affected by bottom fishing than motile spe- cies. All six taxa of colonial epifauna examined in this study were affected by bottom fishing, although the magnitude and direction of this effect often depended on depth and varied between years. Because the effect of disturbance on noncolonial species composition was fairly subtle, our study indicates a greater sensitivity of colonial epifauna to disturbance by mobile fishing gear. Similarly, studies conducted in the Gulf of Alaska and Irish Sea show that most motile organisms are less severely affected by chronic and experimental trawling than are anthozoans, sponges, bryozoans, hydroids, tubi- colous polychaetes, and barnacles (Freese et al., 1999; Bradshaw et ah, 2002). Nevertheless, sessile colonial epifauna exhibit variable morphological forms and pos- sess diverse life history characteristics that can result in different responses to bottom fishing. Colonial organisms whose physical structure can be described as branching, sheet-forming, or mound-forming generally grow slowly but are better competitors for space on the seafloor than encrusting and stoloniferous epifauna (Hughes, 1989). Because of their slow growth patterns, organisms with branching, sheet-forming, and mound-forming structures are typically adapted to stable environments where they are less likely to be detached from their substrate by physical forces associated with natural disturbances. Anthropogenic disturbance causes formerly stable envi- ronments to become unstable, thus leaving colonial epi- fauna with a branching structure maladapted to their new environment. On Georges Bank, sponges appear to be the colonial epifaunal taxa most negatively affected by bottom fish- Asch and Collie: Changes in a benthic megafaunal community due to disturbance from bottom fishing 451 Shallow sites Figure 7 Species richness (number of species) and Simpson’s index of diversity fl-A') for noncolo- nial epifauna at shallow and deep sites. Black squares indicate mean species richness or diversity for a given level of disturbance; standard error bars are shown. Different units of measurement are plotted on the x axis for shallow and deep sites, reflecting the fact that fishing effort was estimated by different methods in U.S. and Canadian waters. Results of linear regression for species richness versus disturbance at shallow sites (A): F=17.69; df=l,3; P= 0.02. All other linear and polynomial regressions were not significant. ing at shallow depths. Most sponges at our study sites had branched or mounding-forming structures, which contributed to their sensitivity to mobile fishing gear. Other characteristics that make sponges vulnerable to mobile fishing gear are aperiodic recruitment and a perennial life cycle that allows colonies to persist over several years (Hughes, 1989). In the southern portion of CA-II that is dominated by sandy substrate, sponges are one of the microhabitats whose abundance differs sig- nificantly between fished and unfished areas (Lindholm et al., 2004) as was also recorded in our study. Several studies from other regions have documented that spong- es are among the most sensitive phyla to bottom fishing (Van Dolah et al., 1987; Freese et al., 1999). Arborescent epifauna react inconsistently to bottom fishing, reflecting the fact that these organisms possess life history traits that both facilitate and hinder their recovery from physical disturbances. Past research in- dicates that the effect of bottom fishing on arborescent epifauna varies between regions and species (Auster et al., 1996; Bradshaw et al., 2002). Like sponges, hy- droids and bushy bryozoans have an emergent growth form that makes them vulnerable to being knocked over by trawls and dredges. Yet these types of arbores- cent epifauna do not generally grow as tall as sponges, thus providing them with a degree of resistance to disturbance. When mobile fishing gear do succeed in removing arborescent epifauna from their substrate, recolonization may occur quickly because these organ- isms are known for rapid growth and typically have short life spans, ranging in length from ten days to one year (Boero, 1983; Hughes, 1989). Other species of hydroids and erect bryozoans are perennial but exhibit seasonal regression. Despite the swift turnover of hy- droid and bushy bryozoan colonies, the spatial extent of recovery may be restricted because many species have limited larval dispersal (i.e., <50 m) or tend to settle close to parent colonies (Hughes, 1989; Bradshaw et al., 2002). The dispersal potential of other species, such as hydroids in the family Sertularidae, is even more con- strained, because they do not have a pelagic medusa stage (Boero, 1983). Our study indicates that depth may be an additional factor affecting how hydroids react to disturbance caused by bottom fishing, because deeper 452 Fishery Bulletin 106(4) sites display a more pronounced, negative response to increased fishing effort. Depth has also been shown to influence the abundance and biomass of other ben- thic taxa (i.e., amphipods and the brittle star Ophiura robusta) in the southern portion of CA-II (Link et al., 2005). Species in the class Polychaeta play diverse ecologi- cal roles (e.g., carnivores, deposit feeders, suspension feeders), use different habitats (e.g., infauna, epifauna), and display different mobility patterns (e.g., errant and tubicolous lifestyles), all of which influences how they respond to bottom fishing. Many polychaetes exhibit a high intrinsic rate of growth allowing them to quickly colonize barren substrate and recover rapidly from disturbances. This high rate of growth explains why the free-living polychaetes in the North Sea remain unaffected by bottom fishing at trawling frequencies as high as six times per year (Jennings et al., 2002). In contrast, tubicolous polychaetes are often heavily affected by bottom fishing because of both their pro- duction of calcareous and sediment-encrusted tubes that can be crushed by fishing gear and their need for stable substrate on which to build tubes. This appears to be the case at the deep depth stratum on Georges Bank, where F. implexa was less abundant at disturbed sites. Similarly, the inception of bottom fishing was implicated in the declining abundance of four spe- cies of tubicolous serpulid polychaetes in the Irish Sea (Bradshaw et al., 2002) and the sabellid polychaete Myxicola infundibulum on Fippennies Ledge in the Gulf of Maine (Langton and Robinson, 1990). A few species do prove that there are exceptions to the rule that most tubicolous polychaetes are negatively affected by mobile fishing gear. Such examples include the tube- heads formed by the serpulid polychaete Pomatoceros spp. that were not significantly affected by biannual beam trawling in the eastern Irish Sea (Kaiser et al., 1999) and the Spirorbis spp. tubes whose abundance was elevated at fished sites on Georges Bank (Collie et al., 2000a; Asch, 2006). The small size of Spirorbis spp. tubes makes them extremely difficult to remove from their substrate and may provide this species with a competitive advantage over other species more sensi- tive to bottom fishing (Collie et al., 2000a). Similarly, the resistance of the Pomatoceros spp. to bottom fishing may be related to the fact their tubes are small enough to pass through the 80-mm mesh used by Kaiser et al. (1999) during experimental trawling. Because of their low, encrusting growth form, en- crusting bryozoans have proven to be resistant to physi- cal disturbances from natural sources (Sebens et al., 1988) and appear to be less sensitive to anthropogenic disturbances, as well. The cover of encrusting bryozo- ans is generally greatest in disturbed areas of Georges Bank. This effect is significant at shallow sites and marginally significant at deep sites. Encrusting bryo- zoans whose substrate may be overturned by a trawl or dredge are capable of recovering quickly because of their fast growth rates and rapid ability to repair structural damage (Bradshaw et al., 2002). This set of life history characteristics may provide encrusting bryozoans with a competitive advantage in highly dis- turbed environments. The abundance of cultch in disturbed areas of north- east Georges Bank can be explained by the fact that most cultch at our study sites consists of P. magellani- cus shells that were either discarded by scallop fishing crews or killed by mobile fishing gear, but not landed. In areas where cultch accumulation is linked to high levels of predation on bivalves or to current patterns concentrating shell fragments, bottom fishing may have a negative effect on this resource. Such is the case in studies of the effect of mobile fishing gear on patches of cultch on Stellwagen Bank and in a more southern area of Georges Bank (Auster et al., 1996; Lindholm et al., 2004). Noncolonial organisms As part of a related research project, naturalist dredge samples of noncolonial megafauna have been collected since 1994 at the same study sites where benthic pho- tographs in the present study were taken (Collie et al., 2005), allowing for comparisons to be made between these two sampling techniques. The results of the cur- rent study are similar to those obtained by Collie et al. (2005) in that they both indicated that P. magel- lanicus, S. droebachiensis, and Pagurus spp. increased in abundance inside CA-II. The increased density of P. magellanicus in the closed area is likely related to both direct and indirect effects of bottom fishing. As a commercially targeted species, P. magellanicus is removed from areas where scallop dredging occurs. The adverse effect of dredging and otter trawling on sponge cover may also serve to reduce P. magellanicus abundance, because some scallop species maintain a mutualistic relationship with sponges that helps them escape predation. In a laboratory experiment where this mutualistic relationship was examined, scallops with sponges encrusted on their shells had to exert 20-30 times less effort to overcome the adherence of the tubefeet of a seastar predator than did scallops whose shells were cleaned of sponges (Bloom, 1975). A third factor influencing the distribution of P. magellanicus may be that the seastars (Asterias spp.), which are key scallop predators, obtained slightly higher abundance outside the closed area. Asterias spp. may be particu- larly abundant at disturbed sites because members of this genus are also scavengers that have been reported to feed upon organisms damaged by trawls (Ramsay et al., 1998). The sea urchin S. droebachiensis may have also ben- efited from the elevated cover of colonial epifauna in CA-II because this species is known to eat sponges, hydroids, bryozoans, tunicates, and amphipod and poly- chaete tube complexes at locations where macroalgae and kelp (the preferred diet of sea urchins) are absent (Briscoe and Sebens, 1988). The elevated number of S. droebachiensis in photos taken in CA-II corroborates the findings of Hermsen et al. (2003), who identified Asch and Collie: Changes in a benthic megafaunal community due to disturbance from bottom fishing 453 this species as one of the taxa most responsible for increased megafaunal production in CA-II. Along shal- low, disturbed transects outside CA-II, bivalves, such as Astarte spp. and C. borealis, were more abundant. The hard shells of these two species likely confer some resistance to disturbance by bottom fishing. Similarly, the unidentified infaunal bivalve species (most likely the razor clam Ensis directus) whose siphons were fre- quently seen along transects open to fishing may be less vulnerable to this form of disturbance because it resides deep in the substrate below depths affected by otter trawls. Many of the species most commonly observed at deep, undisturbed sites belong to the classes Anthozoa, Malacostraca, and Ophiuroidea, which are the classes that a meta-analysis has identified as the taxonomic groups most adversely affected by mobile fishing gear (Collie et al., 2000b). At deep sites, we and Collie et al. (2005) found that bottom fishing results in reduced abundance of multiple species of shrimp, brittle stars (e.g., Ophiopholis aculeata ), tubicolous polychaetes (e.g., Potamilla neglecta), and the toad crab Hyas co- arctatus. Another species that helps define the dis- similarity between disturbed and undisturbed areas at deep sites is the horse mussel Modiolus modio- lus, which is a long-lived, thin-shelled bivalve that is known to be sensitive to bottom fishing (Bradshaw et al., 2002; Collie et al., 2005). M. modiolus may play an important role in the ecology of Georges Bank because anecdotal information from fishermen indicates that the fishes G. morhua and M. aeglefinus congregate around their beds (Leach, 1998). As was the case at shallow sites, hard-shelled mollusks (i.e., Astarte spp. and B. undatum) and scavengers, such as Pagurus spp., are among the species characteristic of deep, disturbed sites. Both Pagurus spp. and B. undatum readily consume organisms injured by bottom fishing (Ramsay et al., 1998). The exact form of the relationship between bottom- fishing disturbance and the diversity of noncolonial spe- cies is highly dependent upon the particular diversity index under consideration. At both depth strata, in- creased bottom fishing results in a concurrent decrease in noncolonial species richness. However, Simpson’s index tends to exhibit its highest value at intermediate levels of bottom fishing. This latter pattern is consistent with the intermediate disturbance hypothesis, which proposes that a moderate amount of disturbance can augment diversity by creating more heterogeneous habi- tats and reducing the likelihood that a single climax species will dominate an area (Connell, 1978). Because bottom fishing is not boosting the absolute number of species present, as indicated by species richness trends, then the high value of Simpson’s index at intermediate disturbance levels must be due to a commensurate in- crease in species evenness resulting from the reduced dominance of a few abundant species. A similar pat- tern was identified in a study of the effect of scallop dredging on the benthic community of the Irish Sea (Bradshaw et al., 2002). Recovery period and recovery rates after disturbance Time periods associated with the recovery of benthic organisms after disturbance caused by bottom fishing are quite variable. Because of adaptations to high levels of natural disturbance, communities living in unconsoli- dated sand have been predicted to recover from bottom fishing in as little as 100 days (Collie et al., 2000b). However, recovery in structurally complex habitats and among particularly vulnerable species (i.e., species that are long lived, poorly adapted to withstand frequent natural disturbances, or highly susceptible to capture or removal by mobile fishing gear) may require longer time periods. For example, it is estimated to take at least 15 years for several species of sponges off the northwest shelf of Australia to grow to a height of 25 cm (Sains- bury et al., 1997). Similarly, the fig sponge ( Suberites ficus) became more abundant in areas protected from bottom fishing within a period of 4.5 years (Lindholm et al., 2004). Deep sea corals, such as the samples of Desmophyllum cristagalli caught off Western Ireland and calculated to be 4000-5000 years old (Hall-Spencer et al., 2002), are the type of marine fauna requiring the longest recovery time. Some estimated recovery rates from bottom fishing reported in the scientific literature may be overly opti- mistic because of biases in the sampling design of many studies. Frequently, these studies involve trawling a small area (<50 m width) located inside a largely un- disturbed site. In the ecological literature, small-scale disturbances surrounded by large, unaffected areas are often referred to as type-1 disturbances. Because of the small spatial scale of a type-1 disturbance and the low temporal frequency of such disturbances, recoloni- zation may occur through immigration from adjacent undisturbed areas or vegetative growth (Auster and Langton, 1999; Kaiser et al., 2002). Recovery through localized immigration and vegetative growth requires less time than would be necessary if recovery were to occur through larval settlement and in situ reproduc- tion of remaining organisms within the disturbed area. In the case of type-2 disturbances where the ecological community is perturbed across large areas interspersed amongst small unaffected patches, recovery usually proceeds by the slower process of larval settlement, a process where most larvae originate from either distant areas or from the small unaffected patches. This latter scenario more realistically describes the recovery pro- cess after disturbance from large-scale bottom-fishing operations on Georges Bank. Our study is noteworthy because little research has been published on the long-term processes governing recovery from disturbance among colonial epifauna on the continental shelf of the northwest Atlantic. After the establishment of CA-II, several colonial and non- colonial taxa underwent successive increases and de- clines in abundance at site 17, thus, providing potential evidence of ecological succession. During 1997, bushy bryozoans briefly peaked in abundance at the shallow undisturbed site. Bushy bryozoans were able to react 454 Fishery Bulletin 106(4) quickly after establishment of a closed-area because the life span of individual colonies typically does not exceed one year (Hughes, 1989). The subsequent decline in the percent cover of this taxon may be linked to either increased interspecific competition or greater predation pressure at site 17. Organisms with life spans of five years or greater (i.e., sponges, P. magellanicus, Pagu- rus spp., Asterias spp., and 15 hours; and Hijk = the number of hooks. 460 Fishery Bulletin 106(4) Table 1 Overview of recaptured (Rec.) archival tags from Greenland halibut (Reinhardtius hippoglossoides) . Fish total length (TL, cm) measured during tagging. F=female; M=male. Rec. no. Interval between recording (min) Release date No. of days recording Fish TL Fish sex 1 60 30.11.02 687 67 2 60 08.12.02 334 64 F 3 60 08.12.02 24 68 — 4 60 08.12.02 188 61 — 5 60 08.12.02 12 54 M 6 60 12.12.02 18 67 F 7 60 13.12.02 16 67 F 8 60 13.12.02 70 74 F 9 10 12.08.03 151 57 — 10 10 14.08.03 149 57 — 11 15 14.08.05 306 76 F 12 15 14.08.05 120 63 F 13 15 15.08.05 54 66 F 14 15 15.08.05 67 77 F 15 15 15.08.05 296 74 F 16 15 15.08.05 10 57 — 17 15 15.08.05 8 73 F 18 15 15.08.05 51 55 F 19 15 16.08.05 54 63 F 20 15 16.08.05 61 75 F 21 15 16.08.05 177 77 — 22 15 16.05.06 261 67 — 23 15 16.05.06 261 47 F 24 15 16.05.06 99 68 F 25 15 17.05.06 196 65 — Total effort within ij was the sum of Eljk for k=(\ to AL), where Ni; is the number of vertical longlines in ij. Demersal longlines Length-frequency distribution data of Greenland halibut on the bottom (separated by sex) were collected by using demersal longlines. In August 2005 the demersal long- lines were 3.5 km long and were deployed in an east-west direction perpendicular to the continental slope. Fishing depth ranged from 450 to 950 m (mean of 655 [±155] m). Soak time ranged from 3 to 22 hours (mean of 10.7 [±5.2] hours). The hooked section on demersal longlines was identical to that of vertical longlines. Archival tags On four occasions (November 2002, August 2003, August 2005, and May 2006) a total of 503 Greenland halibut were tagged with archival tags (DST milli or DST Pitch and Roll, Star-Oddi, Reykjavik, Iceland). All individuals were tagged from demersal longlines within the central study area. The longlines were carefully hauled and only individual fish in apparently good shape were tagged and then immediately released. Recaptures were made with bottom trawls, mainly along the continental slope (Fig. 1). The tags were programmed to record time and ambient depth (pressure) every 10, 15, or 60 minutes, and the depth trajectories were recovered after recap- ture. Up to October 2007, a total of 38 (7.6%) of the released tags were recaptured. Of the recaptured tags, 13 had been destroyed and there was no possibility of data extraction. Of the 25 tags containing data, three tags had stopped recording after being exposed to depths greater than 1000 m, and six tags had stopped record- ing because the memory was full. For the remaining 16 tags, data were recovered for the whole period from release to recapture. Table 1 shows recording sequence, number of data sampled, as well as sex and length of the tagged individual for all 25 tags used in the analysis. These tagging experiments will form the basis for more detailed future publications; here we restrict ourselves to consider the vertical activity and compare it with the results from the vertical longline experiments. The tags, by design, conveyed no direct information on bottom depth or the distance from the bottom. How- ever, it was assumed that readings of constant ambient depth can only occur when the fish are on the bottom, Vollen and Albert: Pelagic behavior of adult Remhardtius hippoglossoides 461 whereas data showing frequent changes in vertical position may indicate pelagic swimming behavior. The vertical activity was analyzed in two ways: standard- ized vertical distance and vertical swimming activity. Standardized vertical distance (VD) between succeeding observations of ambient depths (D) were standardized to equal time intervals: VD( = 3600 x abs (D-Di ^/s, where s = the time step (in seconds) between observation z'— 1 and i. For each individual fish and day, the extent of vertical swimming activity (VA) was estimated as VA=I(Dmax r Dmin i)/12, Where Dmax • and Dmin ■ represents the maximum and minimum depths recorded within a two-hour time inter- val z, and the summation is across the twelve two-hour time intervals within each day. Thus VA is not only dependent on the vertical range traveled during a day, but also on the frequency of vertical excursions throughout the day. Whereas VD measures the speed during a depth change, VA is a measure of activity through the day. Additional surveys Trawl acoustic surveys were conducted in the Barents Sea and Norwegian Sea in August 2003-2005, as part of a regular monitoring program of pelagic fish stocks in these areas. In the present study, data from these investigations were extracted for the extended survey area. The surveys were conducted with a Simrad EK 60 echosounder, frequency 38 kHz (Kongsberg Maritime AS, Horten, Norway). The acoustic recordings were scru- tinized each day by using the Bergen Echo Integrator, and allocation of SA (nautical area scattering coefficient) values on species was made by trained personnel using standard procedures (Korneliussen, 2004). Bathymetric distributions of selected species of fish in bottom trawls were used for comparisons with the pelagic records. Data were gathered from the annual Norwegian Greenland halibut surveys in August 2003- 2005, within the extended study area. In these surveys an Alfredo-5 commercial bottom trawl with rockhopper ground gear (Engas and Godp, 1989) and a measured mean vertical opening of 3.8-4 m was used at towing speeds of 3. 5-4.0 knots. Mesh size in the codend was 60 mm. Stomach-contents data were collected from the same surveys, as well as on similar surveys in November 2003 and March 2004. A standard biological sampling (length, weight, sex, maturity, and stomach fullness) was performed for two individuals in every 5 -cm length group. Individuals sampled were chosen randomly, ex- cept in August 2005 when only individuals with stom- achs containing food were sampled. These stomachs were frozen and analyzed in the laboratory. Methods were the same as those used for stomachs sampled from vertical longline catches. Results Pelagic catches of Greenland halibut Greenland halibut were caught on vertical longlines in most parts of the water column, but despite high effort, the species was rarely recorded in the upper few hundred meters of the water column. Throughout the experimental period a total of 283 individuals were caught, and the species was recorded during all sur- veys but one. Catch depth ranged from 208 to 934 m (Fig. 4). Seventy percent of all catches were from the main survey in August 2005, when 93% of the indi- viduals were caught between 400 m and 800 m depth. There seemed to be an upper catch limit which was relatively stable across bottom depths. In August 2005 this limit was 300 m at 600-700 m bottom depth, 400 m at 800-1000 m bottom depth, and 500 m at 1200 m bottom depth. The westward distribution boundary could not be identified because of inadequate coverage of large depths. Greenland halibut was caught at shallower depths in November than in either March or August. For March and August surveys, mean catch depth above 400-700 m bottom depth ranged from 484 m to 554 m. For the two November surveys, the corresponding numbers were 378 m and 363 m, respectively, differing significantly from all individual March and August surveys (t-tests, P<0.001). The two November surveys were also the only surveys where fish were captured in the region above 300 m catch depth (Fig. 4). There were no clear or significant changes in length composition with catch depth, but rather with bottom depth and with distance from the bottom (data from Au- gust 2005, Fig. 5). This was partly due to males, which are smaller than females, being distributed higher up in the water column (Gtests, PcO.001) and partly to a significant decrease of female length (linear regression, P<0.05). Below 100 m from the bottom, approximately 50% of the catches were males, whereas above 300 m, 75% were males. The length range of individuals caught on vertical longlines was broad (41-85 cm), but the length and sex composition seemed to have more in common with bot- tom trawl catches than bottom longline catches (Fig.6). Bottom longlines caught few fish smaller than 50 cm, which were primarily males. Bottom trawls, on the other hand, caught fish as small as 30 cm length, indi- cating that both sexes were available at the bottom in approximately equal shares. During individual surveys in March and August, the proportion of females in the pelagic longline catches varied from 8% to 48% (mean of 35%). In November no females were caught — a finding that differed signifi- cantly from all March and August surveys (%2, df=l, 462 Fishery Bulletin 106(4) 0 Greenland halibut O Fishing effort 200 300 400 500 600 700 O O O O o O o o o O o o O MAR 04 ° O O O - 1 o • .0 o • O o ’o Cl O o O 0 o o NOV 03 4/ 500 600 700 500 600 0 100 200 300 400 500 600 700 O O O O • o o o o o o o O o o o o O o o o O o O \o • o* AUG 04 • P sO 100 200 300 400 500 600 700 700 o O o o o o o O O o o* o- o o ° o o o o o a o \0 6 © NOV 04 0 100 200 300 400 500 600 700 O 1 o O o o ° O o O o ° O o O o ° o o O o • *o o JD o O *> MAR 05 q 0 100 200 300 o O' o o'O o o O O o o o O o O lP-O OO0 400 ^ Q.; 400 500 600 700 Bottom depth (m) 500 600 700 800 900 1000 1100 1200 o'O o'O o o o o o o o o O-Q o Q o23 ° ■' Q*. AUG 05 500 600 700 800 900 1000 Bottom depth (m) 1100 O O o o o o o Q O o o o 1200 1300 Figure 4 Pelagic catches (noted as individuals) of Greenland halibut (ReinharcLtius hippoglossoides) on vertical longlines. Individual catches are shown by bottom depth (m) and catch depth (m). Fishing effort is shown for each 100-m catch depth interval and 50-m bottom depth interval. The area of the fishing effort symbol is proportional to fishing effort (hooks x hours/100). The straight diagonal line represents the bottom. P<0.01). In November all Greenland halibut caught were classified as late maturing or ripe, whereas in catches from March and August all but three individu- als were immature or early maturing. Individual depth trajectories Of the 25 recaptured fish with intact archival tags, the time recorded varied from a few weeks up to two years, yielding a total of 296,000 depth recordings. All tags were both released and recaptured at or close to the continental slope (Fig. 1). For all recordings combined, the 25th, 50th, and 75th percentiles of depth were 454, 522, and 617 m, respectively. Six percent of all record- ings were from depths above 300 m, from 11 recaptures. For individual tags this percentage ranged from zero to 20%, with a mean of 2.8%. One individual was regularly recorded above 100 m depth, and seven individuals were recorded below 1000 m. Only one of the 16 individuals, for which sex was determined, was male. Most of the recaptured tags revealed alternating pe- riods of distinctly different vertical activity by Green- land halibut. Such periods were apparent on both small (hours-days) and large (days-months) scales. Diurnal variation was most apparent at depths above 300 m (Fig. 7). Here the occurrence of large stan- dardized vertical distance (VD) between successive records was three times higher at night (21h-03h) than during the midday (09h-15h), and records of vertical inactivity (VD< 0.5 m per hour) were twice as numerous during daytime (Fig. 7). Above 300 m, the total number of observations was 19% higher at night Vollen and Albert: Pelagic behavior of adult Reinhardtius hippog/ossoides 463 8 47 70 35 22 5 1 Catch depth (m) 80 70 60 50 40 400 600 800 1000 1200 1400 Bottom depth (m) 0 200 400 600 800 Distance from bottom (m) Figure 5 Line plot of sex composition and box plot of length of Greenland halibut ( Reinhardtius hippoglos- soides) caught on vertical longlines, by catch depth, bottom depth, and catch depth in distance from the bottom (all in 100-m intervals). Data are from August 2005. The number of individuals is given above plots. The plots show median, inter quartile range, upper and lower fence, and outliers as defined in SYSTAT 11 (SYSTAT Software Inc., San Jose, CA). Note that % males and % females add up to 100%. time than during the day, indicating that many of the Greenland halibut in the upper 300 m during night may migrate to deeper waters during the day. The higher number of observations early in the day in the 300-500 m depth zone, and later in the day at depths below 500 m support this (Fig. 7). The verti- cal activity increased and the diurnal signal weakened with increasing depth. The num- ber of recaptures was not considered high enough to warrant analyses of individual and seasonal differences in the diurnal activity pattern. The larger-scale time-variability is illus- trated in Figure 8, which shows the depth trajectory of five tagged fish, together repre- senting the typical pattern seen in all tags. In most cases periods of high vertical activity began with a sudden descent to much deeper depths than those occupied at any time dur- ing the periods of low vertical activity. This pattern is clearly apparent from examples Al, A2, and Cl in Figure 8, where periods of low vertical activity at depth less than 500 m sud- denly changed to periods of high-frequency changes in depth between 500 and 800 m. Example A3 shows a trajectory from a tag that stopped recording after a sudden descent to depths greater than 1000 m, which is the depth limit of the tag. This cessation in re- cording happened to three of the 25 recovered time series. The degree of high vertical swimming activ- ity (VA) varied throughout the year (Fig. 9). The proportion of days with high vertical ac- tivity was greatest during August-October and least in January and February. 15 10 5 15 10 C 10 m □ 0.510 m), medium (0.5-10 m), and low (<0.5 m) VD, respectively. Prey composition The main prey categories found in Greenland halibut stomachs were crustaceans, cephalopods, and fish. These species all occurred in 35-47% of stomachs with con- tents (all trawl surveys combined, Table 2). Crustaceans were mainly shrimps (primarily Pasiphaea spp. and Pandalus borealis) and gammarid amphipods, whereas Gonatus fabricii was the predominant cephalopod. Her- ring and blue whiting were the most common fish prey. One stomach contained a fish tag that one month earlier had been attached to a 19-cm salmon ( Salmo salar), but whether this tag represents a direct or indirect foraging link between salmon and Greenland halibut, remains unknown (Rikardsen et ah, 2008). Of the three main categories of prey, Crustacea were a relatively stable contributor, whereas frequency of occurrence of cephalopods and fish varied between sur- veys (x2, df=4, P<0.01). This temporal variability was mainly due to three species or species groups. Gonatus fabricii was recorded more frequently in August than in March and November, whereas herring was most frequent in August 2003 and March 2004, and blue whiting in August 2003 and 2004. Figure 10 shows how the diet changed with the size of the predator. In terms of weight, the contribution of cephalopods, shrimps, and most other crustaceans decreased with increasing length of Greenland hali- but. On the other hand, the contribution of herring and most other fish increased with increasing preda- tor length. Eelpouts and flatfish were found only in Greenland halibut specimens larger than 65 cm, and redfish were eaten only by individuals larger than 75 cm. Vollen and Albert: Pelagic behavior of adult Remhardtius hippoglossoides 465 Figure 8 Examples of individual depth trajectories determined from archival tags on recaptured Greenland halibut ( Reinhardtius hippoglossoides). Examples A1-A3 from release in November 2002, example B1 from release in August 2003, and example Cl from release in August 2005. For A3 the recordings stopped before recapture, probably because the fish descended beyond the depth limit that the tag was designed to withstand. Of all Greenland halibut caught on vertical long- lines in August 2005, only 27 individuals, or 14%, had recently been feeding, as evidenced by the presence of prey in the stomachs. This percentage ranged from 14% to 40% for fish caught in bottom trawls (Table 2). Stomachs from specimens caught on vertical long- lines contained food from all major prey groups that were also found in stomachs of trawl-caught fish. Of these specimens, 33% contained fresh prey, i.e. prey without signs of digestion. For trawl-caught fish dur- ing the same period this percentage was significantly lower; only 7% contained fresh prey (%2, df=l, P<0.01). In terms of percentage by number, 17% of the prey items from vertical longlines and 5% from bottom trawl catches were classified as fresh (x2, df=l, P <0.01). The fresh prey of Greenland halibut caught by vertical longlines were G. fabricii and the hyperiid amphipod Parathemisto abyssorum. The Greenland halibut were all captured in waters above an 800-900 m bottom depth, and 70% of them were caught more than 300 m above the bottom. The fresh contents from trawl catches consisted of hyperiid amphipods, G. fabricii, and Pasiphaea sp., as well as one case of offal. These individuals were caught at 600—1200 m depth. Pelagic distribution of potential prey The acoustic profiles from the extended study area in August 2003—05 were all quite similar, with an acous- tic layer in the upper 50 m and another around 300 m depth (Fig. 11). Data from 2005 showed that this pattern continued out to the deeper part of the slope. That year the echo sounder was regrettably set to record only down to 500 m depths, but data from 2003 and 2004 strongly indicated that very little backscatter should be expected below 500-600 m. 466 Fishery Bulletin 106(4) Table 2 Percent frequency of occurrence of prey categories determined from stomach contents of Greenland halibut ( Reinhardtius hip- poglossoides) caught on five individual bottom trawl surveys in August 2003-05, November 2003, and March 2004, and on the vertical longline survey in August 2005, and for all bottom trawls combined. Aug 2003 bottom trawl Nov 2003 bottom trawl Mar 2004 bottom trawl Aug 2004 bottom trawl Aug 2005 bottom trawl Aug 2005 vertical longline Combined bottom trawl Crustaceans Amphipods 4.5 5.4 6.7 7.1 14.7 22.2 7.8 Euphausiids 0.0 5.4 0.7 2.4 0.0 0.0 1.1 Shrimps 28.8 35.1 15.3 21.4 36.0 48.1 24.6 Undetermined 3.0 8.1 24.7 7.1 13.3 3.7 14.9 Cephalopods Gonatus fabricii 37.9 10.8 20.0 35.7 58.7 40.7 34.6 Rossia spp. 0.0 0.0 1.3 2.4 1.3 3.7 1.1 Undetermined 1.5 0.0 4.7 2.4 4.0 0.0 3.5 Fish Salmon 1 0.0 0.0 0.0 0.0 1.3 0.0 0.3 Herring 18.2 8.1 36.0 2.4 5.3 0.0 20.0 Blue whiting 9.1 5.4 2.7 14.3 2.7 0.0 5.4 Sebastes spp. 3.0 0.0 0.7 0.0 0.0 0.0 0.8 Other fish 0.0 5.4 1.3 4.8 2.7 0.0 2.2 Undetermined 21.2 32.4 31.3 14.3 12.0 3.7 23.8 Offal 3.0 0.0 0.0 4.8 4.0 0.0 1.9 Other 3.0 2.7 1.3 11.9 1.3 0.0 1.8 Number of stomachs examined 327 98 378 298 — — 198 Stomachs with contents 66 37 150 42 75 370 27 Empty stomachs 80% 62% 60% 86% — — 86% 1 A Carlin dangler tag from a tagged salmon was found in one Greenland halibut stomach. 10 - _ 0 - J FMAMJ J ASOND Month Figure 9 Seasonal variation in mean vertical activity (VA) for Greenland halibut ( Reinhardtius hippoglossoides) tagged with archival tags. For each month, the number of days of individual archival tag recordings where VA was greater than 5 (bold), 10 (regular), and 15 m (dotted line) respectively, is given as a percentage of the total number of days recorded by all tags combined. During these acoustic surveys, most of the total backscatter was allocated to herring in the upper layer and blue whiting in the lower layer. As usual with these types of surveys, the allocation was based on visual recogni- tion of typical patterns, guided by data on target strength of individual echoes and catch composition from sporadic pelagic trawl hauls (mostly outside our extended survey area). Figure 11 shows how catches of redfish and gadoids on vertical longlines compare with the results from the acoustic surveys. The gadoids Atlantic cod ( Gadus morhua), saithe ( Polla - chius virens), and blue whiting were generally caught within the depth range of the lower acoustic layer, i.e. from 200 to 500 m. The low-target-strength species, the deepwater redfish (Sebastes mentella), was caught in the lower part of this layer and below, mostly from 400 to 600 m depth. The bathymetric distribu- tion of Greenland halibut catches from pelagic longlines overlapped to a large extent with the distribution of the prey species redfish and blue whiting, much less with other gadoids, and not with herring. Vollen and Albert: Pelagic behavior of adult Reinhardtius hippoglossoides 467 3 15 36 51 7159 37 39 23 15 8 Offal - %N ° ° ° (- %w ° U- c %FO ° O O (_ Sebastes spp. oC oC OC Blue whiting ° o o o o o C cO°00 O o o o o OO O C Herring ooocXXQX occosM ooaXXOX Salmon 0 - o Other fish o o o o OC o oo<3c o oooOC Undetermined fish - OooocCXXO= QooOoO gHDo - OoCCGCBXX Cephalopoda 5]DooOooooc $ OOWlDoo o . - 38DfflDccooc Caridea <3docXX)oOC G0OO o o O o o o cGOCCOoooc Amphipoda oq0xD° - o o O o o . o - OOOOOoO Euphausiacea oO ° • o - O o o Other Crustacea 0OOOOOO O Vo o O O O o j£xXX)000 o Undetermined - O o O O O 0 i ■■■' i 1 r 1 i 1 i 1 -T - • o • • - - i — — 1 — i — 1 — r-1 — i — >— r - O o o o o o 1 — 1—l — 1 — 1 — 1 1 ' 1 1 — T 30 40 50 60 70 80 30 40 50 60 70 80 30 40 50 60 70 80 Length group (cm) Length group (cm) Length group (cm) Figure 10 Prey composition by 5-cm length groups of Greenland halibut ( Reinhardtius hippoglossoides) from bottom trawl, all periods combined. Bubble size is proportional to percentage composition by numbers (%N), weight (%W), and frequency of occurrence (%FO). Number of stomachs with contents is given above left panel. The abundance of Greenland halibut in bottom trawl was highest between 500 and 800 m, and the species was virtually absent below 1000 m. Figure 11 also shows that the backscatter from the 10-m high bot- tom-zone almost disappeared below 500 m depth, when Greenland halibut, which does not have a swim-bladder and thus a very low target-strength, dominated almost completely over all other fish species. Above 500 m, blue whiting and several demersal species were numerous, whereas Greenland halibut catches were near zero. Discussion Greenland halibut caught in pelagic waters ranged in length from 41 to 84 cm, which is within the same range as that found for those caught on the bottom. This is in contrast to a study by Jprgensen (1997), who found that although one- and two-year-old Greenland halibut from West Greenland nursery areas were found in high abundance in pelagic waters, larger fish (>52 cm) were not found in the pelagic zone of the area inhabited by adult fish. However, Jprgensen commented that the pelagic occurrence of larger specimens may have been underestimated if these fish were able to avoid the pelagic trawl. Support for the ability of large individu- als to avoid bottom trawls has subsequently been noted (Albert et ah, 2003). The influence of gear selectivity processes was ap- parent in length-frequency distributions of catches from demersal longlines and bottom trawls — distribu- tions that closely resembled those found by Huse et al. (1999). The results from trawl catches indicated that small individuals were present on the bottom even though these fish were not caught by demersal longlines. The reason for this could be that the com- mercial longlines used in the experiment targeted large individuals by factors such as hook- and bait- size, which are important in the size-selectivity of longlines (Bjordal and Lpkkeborg, 1996). Also, dif- ferences in swimming speed and in the competitive ability between small and large fish may impart a selection bias, bringing a relatively high proportion of large individuals into contact with the gear (Bjordal and Lqkkeborg, 1996). Therefore, because small fish were common in pelagic longline catches, this may imply that these fish were more abundant than catch rates indicated. The depth recordings from archival tags supported the findings from vertical longline results. The re- cordings showed periods of distinctly different verti- cal activity, both on a small scale (diurnal) and large scale (typically several weeks). Although the depth-re- cordings bore no information about the distance from the sea floor, it seems reasonable to assume that the high vertical activity periods were associated, at least partly, with pelagic distribution. The diurnal signal was most apparent at shallower depths, but could also be interpreted as an increase in migration to lower depths during daytime. Depth recordings from the 468 Fishery Bulletin 106(4) Figure 11 Total acoustic backscatter (bubbles) and catch (noted as individuals) on vertical longlines. Upper panel shows total backscatter together with longline catches of redfish ( Sebastes mentella ) (■) and gadoides (X); lower panel shows total backscatter and longline catches of blue whiting (Micromesistius poutassou) (X) and Greenland halibut ( Reinhardtius hippoglossoides) (■)). Data from August 2005. Diagonal line represents the bottom, and bubbles beneath this line are acoustic recordings from the bottom channel, i.e., from the water column from the bottom up to 10 m off the bottom. high vertical activity periods were comparable to depth distribution of vertical longline catches. Individual and seasonal differences in diurnal activity patterns could not be investigated further due to the low number of recaptures, and due to the unbalanced sex distribution of recaptured fish. However, diurnal vari- ability in bottom trawl catches associated with small individuals migrating into the water column at night time has already been reported off the coast of Labrador during summer (Bowering and Parsons, 1986). Such behavior is probably related to foraging, and should be expected to vary according to prey availability. This may explain why diurnal variability was not previously observed in bottom trawl catches of Greenland halibut from West Greenland waters (Jorgensen, 1997). The predominant prey species or prey groups in stom- achs of Greenland halibut captured with bottom trawl were herring, blue whiting, shrimps, amphipods, and the cephalopod G. fabricii. This finding is in agreement with that of previous studies from the Barents Sea-Nor- wegian Sea continental slope (Michalsen and Nedreaas, 1998; Bjelland et al., 2000; Hovde et al., 2002). The acoustic surveys demonstrated that the vertical distri- bution of blue whiting overlapped with the distribution of Greenland halibut as determined from vertical long- lines. For herring, which were the most important fish prey, no overlapping distributions were evident. Herring were not recorded below 200 m depth during the trawl acoustic surveys — a depth that is shallower than any recording of Greenland halibut from vertical longlines, but archival tag recordings showed that Greenland halibut may be found higher up in the water column, up to above 100 m depth. A broad overlap could be seen between the vertical distributions of redfish and Greenland halibut. Despite this overlap, redfish occurred only in stomachs of fish larger than 75 cm. This finding may be due to the size selectivity of prey as seen in West Greenland waters, where Greenland halibut up to 64 cm total length fed only on redfish smaller than 15 cm, even though larger specimens were present (Pedersen and Riget, 1993). In our area, redfish smaller than 15 cm have been vir- tually absent because recruitment has been very low since 1990. Voilen and Albert: Pelagic behavior of adult Remhardtius hippoglossoides 469 The cephalopod G. fabricii and the shrimps Pasiphaea spp. and P. borealis, all important prey species, may also have been encountered in the pelagic zone. Gonatus fabricii juveniles are distributed in the surface layers, living gradually deeper as they become older. As adults, G. fabricii are largely benthic at depths of 200—3000 m, performing upwards migrations at night (Kristensen, 1983; Bjqrke, 2001). The mesopelagic shrimp Pasiphaea spp. is widely distributed in the Norwegian Sea and have the highest biomasses around 200-600 m depth (Dalpadado et al., 1998), whereas the shrimp P. bo- realis is known to perform diurnal vertical migrations, ascending in the water column in the evening and re- turning to the sea bottom in the morning (see Garcia, 2007). In our study, Greenland halibut larger than 65 cm were the only ones feeding on demersal species such as eelpouts and other flatfish, but they also fed on epipe- lagic species such as herring. This finding differs from that of previous studies where the diet of large Green- land halibut was dominated by large groundfish, even when smaller size individuals fed heavily on smaller pelagic fish such as capelin (Yang and Livingston, 1988; Bowering and Lilly, 1992; Solmundsson, 2007). Because their pelagic distribution is supposed to be linked to foraging activity, the presence of suitable pe- lagic prey should be expected to influence the pelagic behavior of Greenland halibut. Important prey species such as cephalopods, capelin, and herring have strong seasonal migration patterns, and the abundance of these within any given geographic area is prone to seasonal and annual variations. One example would be herring, which was found to be prominent as prey on the conti- nental slope in October 1997, whereas in January 1998 it was not identified as prey at all (Hovde, 2002). Heavy feeding on capelin, another migratory small pelagic species, has also been observed for Greenland halibut (Smidt, 1969; Bowering and Lilly, 1992). The vertical activity level of individual Greenland halibut showed a clear seasonal pattern, being highest in August-October. This is the period preceding the main spawning season, when mature Greenland halibut gather in the slope area (Albert et ah, 2001). It is also the period when annual surveys are made for popula- tion monitoring of the Northeast Arctic stock. Concluding remarks This is the first published account of adult Greenland halibut being caught in pelagic waters in considerable quantities in any part of its distribution area. In the present study, adult individuals were found to be widely distributed in the water column. Pelagic activity was related to fish size and dial and seasonal cycles and was influenced by variability in the distribution of prey species. Variations in availability to the survey trawl could be biasing the length- and sex-distribution of these catches as well as obscuring trends in population abundance and structure as derived from stock assess- ment. The methods used in our study are not suitable for quantifying pelagic behavior, but both diet composition and vertical activity from archival tags indicate that use of the pelagic zone may be significant. In order to improve abundance estimates, and thereby the basis for management decisions, it is important to develop methods to estimate the availability of adult Greenland halibut to bottom sampling trawl. Acknowledgments Four anonymous reviewers and the journal editor are thanked for valuable comments and suggestions on an earlier version of the manuscript. Literature cited Albert, O. T., A. Harbitz, and A. S. H0ines. 2003. Greenland halibut observed by video in front of survey trawl: behaviour, escapement, and spatial pattern. J. Sea Res. 50:117-127. Albert, O. T., E. M. Nilssen, A. Stene, A. C. Gundersen, and K. H. Nedreaas. 2001. Maturity classes and spawning behaviour of Green- land halibut ( Reinhardtius hippoglossoides ). Fish. Res. 51:217-228. Bjelland, O., O. A. Bergstad, J. E. Skjaeraasen, and K. Meland. 2000. Trophic ecology of deep-water fishes associated with the continental slope of the eastern Norwegian Sea. Sarsia 85(2 ):101— 117. Bjordal, A., and S. Lokkeborg. 1996. Longlining, 156 p. Fishing News Books, Oxford. Bjprke, H. 2001. Predators of the squid Gonatus fabricii (Lichten- stein) in the Norwegian Sea. Fish. Res. 52 ( 1-2 ):1 13-120. Bowering, W. R., and G. R. Lilly. 1992. Greenland halibut ( Reinhardtius hippoglossoides) off southern Labrador and northeastern Newfoundland (Northwest Atlantic) feed primarily on capelin ( Mallotus villosus). Neth. J. Sea Res. 29:211-222. Bowering, W. R., and K. H. Nedreaas. 2000. A comparison of Greenland halibut (Reinhard- tius hippoglossoides (Walbaum)) fisheries and distribu- tion in the Northwest and Northeast Atlantic. Sarsia 85:61-76. Bowering, W. R., and D. G. Parsons. 1986. Diel variability in trawl catches of Greenland halibut from the channels off coastal Labrador and implications for resource assessment. N. Am. J. Fish. Manag. 6(2):149-155. Christensen, O., and W. H. Lear. 1977. Bycatches in salmon drift-nets at West Greenland in 1972. Meddelelser om Grpnland. 205(5):1— 38. Dalpadado, P., B. Eilertsen, W. Melle, and H. R. Skjoldal. 1998. Summer distribution patterns and biomass esti- mates of macrozooplankton and micronekton in the Nordic Seas. Sarsia 83(2):103-116. Dawe, E. G., W. R. Bowering, and J. B. Joy. 1998. Predominance of squid (Gonatus spp.) in the diet of Greenland halibut (Reinhardtius hippoglossoides ) on the deep slope of the northeast Newfoundland continental shelf. Fish. Res. 36:267-273. 470 Fishery Bulletin 106(4) de Groot, S. J. 1970. Some notes on an ambivalent behaviour of Green- land halibut Reinhardtius hippoglossoides (Walb.). Pisces: Pleuronectiformes. J. Fish Biol. 2:275-279. Engas A., and O. R. Godo. 1989. Escape of fish under the fishing line of a Nor- wegian sampling trawl and its influence on survey results. ICES J. Mar. Sci. 45(31:269-276. Garcia, E. G. 2007. The northern shrimp ( Pandalus borealis) offshore fishery in the Northeast Atlantic. Adv. Mar. Biol. 52:147-266. Hovde, S. C., O. T. Albert, and E. M. Nilssen. 2002. Spatial, seasonal and ontogenetic variation in diet of Northeast Arctic Greenland halibut ( Reinhardtius hippoglossoides). ICES J. Mar. Sci. 59:421-437. Huse, I., A. C. Gundersen, and K. H. Nedreaas. 1999. Relative selectivity of Greenland halibut (Rein- hardtius hippoglossoides, Walbaum) by trawls, longlines and gillnets. Fish. Res. 44:75-93. Jorgensen, O. A. 1997. Pelagic occurrence of Greenland halibut, Rein- hardtius hippoglossoides (Walbaum), in West Greenland waters. J. Northwest Atl. Fish. Sci. 21:39-50. Korneliussen, R. J. 2004. The Bergen echo integrator post-processing system, with focus on recent improvements. Fish. Res. 21(1- 3):159-169. Kristensen, T. K. 1983. Gonatus fabricii. In Cephalopod life cycles, vol. 1. Species accounts. (Boyle, P. R., ed.), p. 159-173. Aca- demic Press, London. Merrett, N. R., and R. L. Haedrich. 1997. Deep-sea demersal fish and fisheries, 282 p. Chap- man and Hall, London. Michalsen, K., and K. H. Nedreaas. 1998. Food and feeding of Greenland halibut (Reinhard- tius hippoglossoides, Walbaum) in the Barents Sea and East Greenland waters. Sarsia 83:401-407. Pedersen, S. A., and F. Riget. 1993. Feeding-habits of redfish (Sebastes spp.) and Green- land halibut ( Reinhardtius hipploglossoides) in West Greenland waters. ICES J. Mar. Sci. 50(4):445-459. Rikardsen, A. H., L. P. Hansen, A. J. Jensen, T. Volien, and B. Finstad. 2008. Do Norwegian Atlantic salmon feed in the northern Barents Sea? Tag recoveries from 70 to 78°N. J. Fish Biol. 72:1792-1798. Smidt, E. L. B. 1969. The Greenland halibut, Reinhardtius hippoglos- soides (Walb ), biology and exploitation in Greenland waters. Medd. Danm. Fisk. Havunders., N.S. 6(4):79-148. Solmundsson, J. 2007. Trophic ecology of Greenland halibut (Reinhardtius hippoglossoides) on the Icelandic continental shelf and slope. Mar. Biol. Res. 3:231-242. Yang, M. S., and P. A. Livingston. 1988. Food-habits and daily ration of Greenland halibut, Reinhardtius hippoglossoides, in the Eastern Bering Sea. Fish. Bull. 84:675-690. 471 All their eggs in one basket: a rocky reef nursery for the longnose skate (Raja rhina Jordan & Gilbert, 1880) in the southern California Bight Milton S. Love (contact author)1 Donna M. Schroeder2 Linda Snook1 Anne York3 Guy Cochrane4 Email address for M. S. Love: love@lifesci.ucsb.edu 1 Marine Science Institute University of California Santa Barbara, California 93106 2 Minerals Management Service 770 Paseo Camarillo Camarillo, California 93010 3 6018 Sycamore Avenue Seattle, Washington 98107 4 United States Geological Survey 400 Natural Bridges Dr. Santa Cruz, California 95060 Skates (family Rajidae) are ovipa- rous and lay tough, thick-walled eggs. At least some skate species lay their eggs in spatially restricted nursery grounds where embryos develop and hatch (Hitz, 1964; Hoff, 2007). After hatching, neonates may quickly leave the nursery grounds (Hoff, 2007). Egg densities in these small areas may be quite high. As an example, in the eastern Bering Sea, a site <2 km2 har- bored eggs of Alaska skate (Bathyraja parmifera ) exceeding 500,000/km2. All skate nursery grounds have been identified over soft sea floors (Lucifora and Garcia, 2004; Hoff, 2007). In 2005, while conducting fish surveys using a manned submers- ible over natural reefs in the Santa Barbara Channel, southern Califor- nia, we found an area of high skate egg density, located on the edge of Hueneme Submarine Canyon. Until that date, we had rarely observed skate eggs in our southern California surveys. For example, in 362 other submersible dives, in waters between 18 and 365 m deep and encompass- ing 395 km of transects over a wide range of habitats, we had observed only 44 skate eggs. In 2006, we re- turned to the Hueneme Submarine Canyon site and examined this nurs- ery ground more closely. Material and methods We conducted the study on 24 October 2006 on a feature located on the west side of Hueneme Submarine Canyon (Fig. 1). The study area is a rocky outcrop located at approximately 34°02.3'N, 119°18.1'W. Rocks exposed at the site are likely a submerged extension of the Miocene volcanic rocks that make up Anacapa Island (Vedder et al., 1986). These rocks form gently north-dipping strata of volca- nic flows, breccias, conglomerates, or tuffs. Faults and joints observed on the Island, and from the submersible at the site, increase rock resistance against gravitational failure. Bathy- metric high spots, such as that of the skate nursery grounds, are north- dipping volcanic-layer outcrops that have not collapsed because of the local strength of the rock substrate in com- bination with the buttressing struc- ture formed by faulting and jointing. Several diminutive taxa, including squarespot ( Sebastes hopkinsi), sword- spine ( Sebastes ensifer), and pygmy ( Sebastes wilsoni) rockfishes, domi- nate the fish fauna. Among structure- forming invertebrates, the volcanic outcrop also harbors high densities of barrel, flat, foliose, and vase sponges, gorgonian corals, the large anemone Metridium sp., basketstars, and the deep-water antipatharian coral Anti- pathes dendrochristos. This survey was conducted aboard the small (4.8 m in length) research submersible Delta (Delta Oceano- graphies, Ventura, CA). During the dive, we tried to maintain a constant distance within 1 meter of the sea- floor and a constant speed between 0.5 and 1.0 knot. The survey was made during daytime hours and we documented the egg density of long- nose skates in that area with an ex- ternally mounted hi-8 video camera positioned above the middle viewing- porthole on the starboard side of the submersible. The scientific observer conducted a belt-transect survey through this same starboard viewing port, verbally recording onto the vid- eotape all skate eggs observed within 2 m of the submersible. Navigation fixes (latitude and lon- gitude coordinates) were received from a Thales GeoPacific Winfrog ORE Trackpoint 2 USBL (Fugro- Pelagos, San Diego, CA) system at two-second intervals, and a Winfrog DAT file was generated for the re- search dive. Distance and duration between fixes were calculated to obtain a point-to-point submersible speed; errant navigation fixes were removed when speed exceeded 2 m/ sec. The navigation fixes were then smoothed by using a nine-point mov- ing average, and transect length was Manuscript submitted 14 February 2008. Manuscript accepted 13 May 2008. Fish. Bull. 106:471-475 (2008). The views and opinions expressed or implied in this article are those of the author and do not necessarily reflect the position of the National Marine Fisheries Service, NOAA. 472 Fishery Bulletin 106(4) (Lower map) The general location (black box) of the longnose skate ( Raja rhina) nursery ground on the edge of Hueneme Submarine Canyon, south- ern California Bight. (Upper map) The area within the black box in the lower map has been enlarged and the white box represents the nursery ground within that area; the other white lines in the upper map denote transects followed by the Delta submersible. Transects outside the box did not contain skate eggs. estimated from the total distance between the smoothed points. Transect length was divided by transect dura- tion to obtain an average transect speed. The length of individual habitat patches was estimated from average speed of the submersible during each transect. During the survey, the observer verbally noted each egg or egg aggregation, estimated the number of eggs in the aggregation, and whether the eggs were living or dead. We assumed that yellow or green, shiny and full eggs still contained embryos and that those that were dark, deflated, and heavily biofouled were empty. It should be noted that in other studies dark egg cases have been found to still hold living embryos and thus our estimates of viability may be low. At that time, the observer also described whether the egg(s) had been laid on rock, cobble, sand, or on such structure-forming animals as sponges and gorgonians. Lastly, in order to identify the species using this nursery ground, five eggs were collected. In the laboratory, the abiotic habitats of the nurs- ery ground were also characterized from the survey videotape. We defined these habitats using the differ- ent categories of substrata and standard geological definitions defined by Yoklavich et al. (2000). In order NOTE Love et at: A rocky reef nursery for Raja rhina in the southern California Bight 473 Table 1 Expected and actual (observed) longnose skate (Raja rhina) egg numbers by habitat type. Observations were made from the Delta submersible, eastern Santa Barbara Channel, October 2006. Habitat types were rock ridge (R), boulder (B), cobble (C), and sand (S). A two-character code was assigned each time a distinct change in substra- tum type was noted. The first character of the code rep- resents the substratum that accounted for at least 50% of the survey patch, and the second character represents the substratum accounting for at least 20% of the patch. Expected numbers were calculated by assuming that the density of skate eggs was not different across habi- tat types. Expected numbers were rounded to the near- est integer. Habitats are ordered from most to least area surveyed (m2). Habitat Area surveyed (m2) Expected egg numbers Actual egg numbers RR 2219.0 1513 1721 RC 208.6 142 13 RB 76.1 52 2 RS 37.6 26 4 CR 10.4 7 0 of decreasing particle size, these substrata were the following: rock ridge (R), boulder (B), cobble (C), and sand (S). A two-character code was assigned each time a distinct change in substratum type was noted. The first character of the code represented the substratum that accounted for at least 50% of the survey patch, and the second character represented the substratum accounting for at least 20% of the patch. Thus, RB rep- resented a patch composed of at least 50% rock ridge and at least 20% boulders. Results In the nursery area, we surveyed 2551.8 m2 of sea floor in waters between 125 and 151 m depth (average depth 139 m) and with bottom temperatures ranging from 9.1° to 10.1°C (Fig. 1). The study site was primarily composed of high-relief rocky ridge (RR), along with lesser amounts of rock-cobble (RC), rock-boulder (RB), rock-sand (RS), and cobble-rock (CR) (Table 1). We observed 1740 eggs, of which 238 were character- ized as intact. As far as we could ascertain, all of the eggs observed were those of the longnose skate (Raja rhina Jordan & Gilbert, 1880) (Fig. 2). We observed a range of egg states, from those that appeared to be new- ly deposited to ones that had almost disintegrated. The eggs did not appear to have been randomly laid over the substrata because significantly more eggs were observed over the highest relief areas (habitat RR) than over the other habitats (chi-square=13.1, df=4, P=0.01) (Table 1). Eggs also appeared to be clumped in their distributions. Although many of the eggs were deposited singly, we Table 2 Number of observations (from the Delta submersible) of each size of longnose skate ( Raja rhina) egg cluster (no. of eggs in cluster) in the nursery ground on the edge of Hue- neme Submarine Canyon, October 2006, categorized by habitat type. Habitat types were rock ridge (R), boulder (B), cobble (C), and sand (S). A two-character code was assigned each time a distinct change in substratum type was noted. The first character of the code represents the substratum that accounted for at least 50% of the patch, and the second character represents the substratum accounting for at least 20% of the patch. Number of eggs in cluster Number of observations RB RC RR RS 1 0 13 145 4 2 1 1 38 0 3 0 0 23 0 4 0 0 9 0 5 0 0 6 0 6 0 0 4 0 8 0 0 1 0 10 0 0 6 0 12 0 0 1 0 20 0 0 3 0 24 0 0 1 0 25 0 0 1 0 30 0 0 4 0 40 0 0 2 0 >50 0 0 8 0 Total 1 14 1722 4 observed several aggregations with 300 or more eggs (Table 2). Most of the eggs were laid on rocks and rela- tively few on such invertebrates as sponges, deep-water corals, and sea anemones (Table 3). However, the eggs that were laid on invertebrates were significantly more likely to be intact than were those on rocks (intact on invertebrates = 95%, 95% confidence interval, 83-99%; intact on rock=12%, 95% confidence intervals 10-13%; chi-square=225, df=l, PcO.0001). We estimated the total number both of all eggs and only those that were intact. We assumed that our sur- vey transect path demarcated the outer perimeter of the nursery grounds because we noted eggs all along the transect path. Using densities of both all eggs (0.68183/ m2) and only intact ones (0.093 eggs/m2) and the area bounded by our transects (27,878 m2), we estimated that there were 19,008 eggs in the nursery grounds, of which 2593 were intact. Discussion According to our many years of visual observations, skate nursery grounds are uncommon in southern 474 Fishery Bulletin 106(4) Figure 2 (A) An intact longnose skate ( Raja rhina) egg, (B) an aggregation of eggs (see arrows), many of which are covered in fine sediment. Table 3 Numbers of intact and empty longnose skate (Raja rhina) eggs on abiotic (rocks and fishing line) and biotic sub- strata, observed from the Delta submersible, eastern Santa Barbara Channel, October 2006. Substrata Number of eggs intact Number of eggs empty Abiotic Rocks 197 1500 Fishing line 1 0 Total abiotic 198 1500 Biotic Sponges 35 1 Gorgonians 3 1 Corals 1 0 Sea anemones 1 0 Total biotic 40 2 California waters (at least at depths <360 m), and like that observed by Hoff (2007) the nursery site that we observed was relatively small in area because visual transects conducted within 1 km over similar habitats yielded no eggs (Fig. 1). However, whereas previously described skate nursery grounds lay on soft sea floors, the study site was a rocky outcrop sitting on the edge of a submarine canyon. This reef appears to be a relatively high-energy area (little sediment was found on the substrata) with high densities of structure-forming invertebrates. Because so few visual surveys have been conducted on the Pacific Coast, we do not know if the nursery ground that we observed was atypical or if there are other such grounds in Pacific Coast waters. Longnose skates do not randomly lay their eggs over the sea floor. Among the various habitat types in the nursery area, rock ridges contained statistically more eggs and these eggs were often found in clumps. It would be expected that eggs laid on the highest relief would be most exposed to currents and arguably less NOTE Love et al. : A rocky reef nursery for Ra/a rhina in the southern California Bight 475 available to at least some predators than eggs lying on low substrata. Skate reproduction, with its emphasis on highly spa- tially restricted nursery grounds, appears to be an example of “predator swamping” (Van Montfrans et ah, 1995). In addition, the observation that intact skate eggs are more likely to be found on sponges than on bare rocks may reflect an antipredator strategy by adult skates. Boring snails, including those in the fam- ilies Muricidae, Naticidae, and likely Ranellidae, are major predators on skate eggs and predation rates in nursery grounds can reach 40% or more (Lucifora and Garcia, 2004; Hoff, 2007). Snails of all of these families are found in southern California waters (McLean and Gosliner, 1993). Although they could not identify them in their trawl-based study, Lucifora and Garcia (2004) speculate that microhabitat differences in egg place- ment could lead to variable protection from predators. It is possible that sponges, with their spicule-rich skel- etons and potent chemical defenses, may deter snails or other egg predators. The Bering Sea skate nursery ground that Hoff (2007) investigated was notable for having 1) high currents and productivity and 2) the presence of only mature fish. The nursery site of our study was com- posed of high-relief rocks situated on the edge of a submarine canyon. Because of the relative absence of fine particulates on the substrata and the presence of unusually high densities of sponges, gorgonians, and other structure-forming invertebrates, we surmised that this nursery ground is bathed by high currents and is likely quite productive. In addition, we observed no juvenile or adult skates, providing additional evi- dence for Hoff’s (2007) hypothesis that newly hatched skates quickly leave their nursery grounds. In the context of this nursery ground much about longnose skate reproduction remains unknown. For instance, we do not know when the eggs are laid, their incubation period, nor the fate of the juveniles. We do not know how many skates use this habitat and how they migrate. Lastly we note that the extent to which female skates seek out sponges is unknown. However, if skate eggs deposited on sponges are at a competitive advantage, damage to these invertebrate communities (i.e., through destructive fishing practices) and the subsequent increase in egg predation, would have a detrimental effect on skate reproductive success. Acknowledgments D. Ebert identified the skate eggs and J. Hoff provided a great deal of useful background information on skate nursery grounds and provided us with a copy of his Ph.D. thesis. Also very useful were the discussions with D. Cadien on sponge and snail ecology. We thank the pilots of the research submersible Delta, J. Lilly and C. Ijames, and the crew of the research vessel Velero for support in the field. Literature cited Hitz, C. R. 1964. Observations on egg cases of the big skate (Raja binoculata Girard) found in Oregon coastal waters. J. Fish. Res. Board Can. 21:851-854. Hoff, G. R. 2007. Reproductive biology of the Alaska skate Bathy- raja parmifera, with regard to nursery sites, embryo development and predation. Ph.D. diss., 160 p. Univ. Washington, Seattle, WA. Lucifora, L. O., and V. B. Garcia. 2004. Gastropod predation on egg cases of skates (Chondrichthyes, Rajidae) in the southwestern Atlan- tic: quantification and life history implications. Mar. Biol. 145:917-922. McLean, J. H., and T. M. Gosliner. 1993. Mollusca, vol. 9, part 2. In Taxonomic atlas of the benthic fauna of the Santa Maria Basin and the western Santa Barbara Channel (J. A. Blake, and P. V. Scott, eds.), 162 p. Santa Barbara Museum of Natural History, Santa Barbara, CA. Van Montfrans, J., C. E. Epifanio, D. M. Knott, R. N. Lipcius, D. J. Mense, K. S. Metcalf, E. J. Olmi III, R. J. Orth, M. H. Posey, E. L. Wenner, and T. West. 1995. Settlement of blue crab postlarvae in western North Atlantic estuaries. Bull. Mar. Sci. 57:834-854. Vedder, J. G., J. K. Crouch, and J. Junger. 1986. Geologic map of the mid-southern California con- tinental margin. In California continental margin geo- logic map series, 3A (H. G. Greene, and M. P. Kennedy, eds.), 4 p. Calif. Dep. Conserv., Sacramento, CA Yoklavich, M. M., H. G. Greene, G. M. Cailliet, D. E. Sullivan, R. N. Lea, and M. S. Love. 2000. Habitat associations of deep-water rockfishes in a submarine canyon: an example of a natural refuge. Fish. Bull. 98:625-641. 476 Microsatellite primers for red drum (Sciaenops ocellatus ) Email address for J.R. Gold: goldfish@tamu.edu 1 Center for Biosystematics and Biodiversity Texas A&M University, TAMU-2258 College Station, Texas 77843-2258 2 USDA/ARS National Center for Cold and Cool Water Aquaculture 11861 Leetown Road Kearneysville, West Virginia 25430 Sten Karlsson1 Mark A. Renshaw1 Caird E. Rexroad III2 John R. Gold (contact author)1 In this note, we document polymerase- chain-reaction (PCR) primer pairs for 101 nuclear-encoded microsatel- lites designed and developed from a genomic library for red drum ( Sciae - nops ocellatus ). Details of the genomic library construction, the sequencing of positive clones, primer design, and PCR protocols may be found in Karls- son et al. (2008). The 101 microsatel- lites (Genbank Accession Numbers EU015882-EU015982) were amplified successfully and used to genotype 24 red drum obtained from Galveston Bay, Texas (Table 1). A total of 69 of the microsatellites had an uninter- rupted (perfect) dinucleotide motif, and 30 had an imperfect dinucleo- tide motif; one microsatellite had an imperfect tetranucleotide motif, and one had an imperfect and compound motif (Table 1 ). Sizes of the cloned alleles ranged from 84 to 252 base pairs. A ‘blast’ search of the Genbank database indicated that all of the primers and the cloned alleles were unique (i.e., not duplicated). Summary genotypic data, based on 22-24 assayed red drum, also are given in Table 1 and include number (and size range) of alleles detected, observed and expected heterozygos- ity, and probability values from tests for conformity to Hardy-Weinberg equilibrium expectations. One mic- rosatellite (Soc734) was monomor- phic; the number of alleles detected at the remaining 100 (polymorphic) microsatellites ranged from 2 to 26. Estimates of observed and expected heterozygosity and tests for confor- mity to Hardy-Weinberg and geno- typic equilibrium expectations were performed with Genepop (Raymond and Rousset, 1995). Observed het- erozygosity (polymorphic microsat- ellites) ranged from 0.042 (Soc706) to 1.000 (11 microsatellites, Table 1) and averaged (±standard deviation [SD] ) 0.775 ±0.211; expected hetero- zygosity ranged from 0.042 (Soc706) to 0.971 (Soc636) and averaged 0.806 ±0.201. After Bonferroni correction (Rice, 1989), genotypes at 99 of the polymorphic microsatellites did not differ significantly from Hardy-Wein- berg equilibrium expectations. At one locus, Soc 706, there were only two al- leles, one of which was observed only in a heterozygote; this microsatellite was not tested for Hardy-Weinberg equilibrium. Analysis with MICRO- CHECKER (Van Oosterhout et al., 2004) indicated the possible occur- rence of null alleles at nine of the microsatellites, and single base-pair shifts (i.e., alleles differing by only a single base pair) were observed at five of the microsatellites (Table 1). Tests of genotypic disequilibrium were nonsignificant after Bonferroni correction. Given that red drum pos- sess 24 haploid chromosomes (Gold et al., 1988), several of the microsatel- lites undoubtedly are linked; deter- mination of linkage will await formal mapping studies. Along with PCR primers for red drum microsatellites developed pre- viously by O’Malley et al. (2003), Saillant et al. (2004), and Karlsson et al. (2008), the primers developed here will be useful in a variety of applications (Liu and Cordes, 2004), including analysis of stock structure, monitoring and assessment of red drum stock enhancement, parentage analysis as employed in aquaculture, and the generation of a genetic map for red drum. A table of the 269 PCR primers developed for red drum may be found at under the file name “PCR prim- ers for red drum (Sciaenops ocellatus) microsatellites.” Acknowledgments We thank C. Abbey for technical assistance with the Q-bot (Genetix), E. Saillant for assistance in the lab- oratory and helpful comments on a draft of the paper, and R. Vega for encouragement and support. Work was supported by the Coastal Conser- vation Association and Central Power and Light (CCA/CPL) Marine Devel- opment Center of the Texas Parks and Wildlife Department, the Coastal Conservation Association — Texas, and the Texas Agricultural Experiment Station (Project H-6703). This note is number 64 in the series “Genetic studies in marine fishes” and contri- bution no. 155 from the Center for Biosystematics and Biodiversity at Texas A&M University. Manuscript submitted 10 June 2008. Manuscript accepted 12 June 2008. Fish. Bull. 106:476-482(2008). The views and opinions expressed or implied in this article are those of the author and do not necessarily reflect the position of the National Marine Fisheries Service, NOAA. NOTE Karlsson et al.: Microsatellite primers for Sciaenops oce/latus 477 478 Fishery Bulletin 106(4) NOTE Karlsson et al.: Microsatellite primers for Sciaenops ocellatus 479 480 Fishery Bulletin 106(4) NOTE Karlsson et al.: Microsatellite primers for Sciaenops ocellatus 481 482 Fishery Bulletin 106(4) Literature cited Gold, J. R., K. M. Kedzie, D. A. Bohlmeyer, J. D. Jenkin, W. J. Karel, N. Iida, and S. M. Carr. 1988. Studies on the basic structure of the red drum ( Sciaenops ocellatus) genome. Contrib. Mar. Sci. 30 (suppl.):57-64. Karlsson, S., M. A. Renshaw, C. E. Rexroad CE III, and J. R. Gold. 2008. PCR primers for 100 microsatellites in red drum ( Sciaenops ocellatus). Mol. Ecol. Res. 8:393-398. Liu, Z. J., and J. F. Cordes. 2004. DNA marker technologies and their appli- cation in aquaculture genetics. Aquaculture 238:1-37. O’Malley, K. G., C. A. Abbey, K. Ross, and J. R. Gold. 2003. Microsatellite DNA markers for kinship analysis and genetic mapping in red drum, Sci- aenops ocellatus (Sciaenidae, Teleostei). Mol. Ecol. Notes 3:155-158. Raymond, M, and F. Rousset. 1995. Genepop (version 2.1): Population genetics software for exact tests and ecumenicism. J. Hered. 86:248-249 Rice W. R. 1989. Analyzing tables of statistical tests. Evo- lution 43:223-225. Saillant, E., K. Cizdziel, K. G. O’Malley, T. F. Turner, C. L. Pruett, and J. R. Gold. 2004. Microsatellite markers for red drum, Sci- aenops ocellatus. Gulf Mexico Sci. 2004:101- 107. Van Oosterhout, C., W. F. Hutchinson, and P. Shipley. 2004. Micro-Checker: software for identifying and correcting genotyping errors in microsatel- lite data. Mol. Ecol. Notes 4:535-538 483 Acknowledgment of reviewers The editorial staff of Fishery Bulletin would like to acknowledge the scientists who reviewed articles published in 2007-2008. Their contributions have helped ensure the publication of quality science. Mr. Stuart Poss Dr. Andre E. Punt Dr. Juan Antonio Raga Dr. Stephen Ralston Dr. Aaron N. Rice Dr. William J. Richards Dr. Larry G. Allen Dr. Paul Hamer Dr. Bernhard M. Riegl Dr. Laura Rogers-Bennett Mr. Arnold Ammann Dr. Patrick J. Harris Dr. Christopher N. Rooper Dr. Allen H. Andrews Dr. Lee C. Hastie Ms. Jodie L. Rummer Dr. Peter J. Auster Dr. Scott A. Heppell Dr. Gerald R. Hoff Mr. Nick Sagalkin Dr. Raymond T. Bauer Mr. Peter B. Hood Dr. Peter C. Sakaris Dr. Terry D. Beacham Mr. John Hyde Mr. Thor Saunders Dr. James R. Bence Dr. Frederick S. Scharf Dr. Jesper Boje Dr. John Jacobs Dr. Thomas C. Shirley Dr. Stephen Bortone Dr. Simon Jennings Dr. Michael F. Sigler Dr. Paul A. Breen Dr. Christian Jorgensen Dr. Greg Silber Dr. Virginia Butler Dr. Ole Jorgensen Dr. David G. Smith Dr. Terje Jorgensen Dr. Susan M. Sogard Dr. Martin Castonguay Dr. Paul Spencer Dr. Robert W. Chapman Dr. Yoshiaki Kai Dr. Jill St. John Dr. William Cheung Ms. Lisa A. Kerr Ms. Michelle D. Staudinger Dr. Seinen Chow Dr. Daniel K. Kimura Dr. Allan W. Stoner Dr. William G. Clark Dr. Michael J. Kingsford Dr. Christoph Stransky Dr. Martin A. Connaughton Mr. Trey Knott Dr. James A. Sulikowski Ms. Melissa Cook Dr. Iain M. Suthers Mr. Daniel Wr. Cooper Dr. Han-Lin Lai Dr. Theodore Switzer Dr. Tara Cox Dr. Christopher M. Legault Dr. Jeffrey M. Leis Dr. Grant G. Thompson Dr. Edward E. DeMartini Dr. Joe Liddle Dr. Fausto Tinti Dr. David Demer Mr. Frank Lockhart Dr. Galen Tromble Dr. Sandra L. Diamond Dr. Milton S. Love Dr. Vlad Troynikov Dr. Robert Diaz Dr. Theresa Tsou Dr. Miriam J. Doyle Dr. David A. Mann Dr. Richard S. McBride Dr. James H. Uphoff Dr. Nicholas G. Elliott Dr. Susanne McDermott Mr. Dan Urban Dr. Karim Erzini Mr. Wayne McFee Dr. Richard D. Methot Mr. James M. Vasslides Dr. Dagmar Fertl Ms. Joyce Miller Dr. Miguel Vazquez Archdale Ms. Kerstin Fritsches Dr. Thomas J. Miller Mr. Ben Victor Dr. Todd Gedamke Dr. Tony Miskiewicz Dr. Beatriz Morales-Nin Dr. John R. Waldman Dr. Anthony J. Gharrett Dr. John A. Musick Dr. Harvey J. Walsh Dr. R. Grant Gilmore Ms. Leslie Ward-Geiger Dr. John R. Gold Dr. Shawn Narun Dr. Gordon T. Waring Dr. Kenneth J. Goldman Dr. William A. Newman Dr. Eric Warrant Ms. Kelly Goodwin Dr. Edward J. Noga Mr. Jon Warrenchuk Dr. Rachel T. Graham Dr. Douglas Nowacek Mr. William Watson Dr. Churchilll B. Grimes Dr. Dara H. Wilber Dr. Victoria M. O’Connell Dr. Dick Wilmot Dr. Alexei M. Orlov Dr. Gary Winans Dr. Vincent Guillory Dr. Hazel A. Oxenford Mr. Anthony D. Wood Dr. Donald R. Gunderson Dr. Carrie Worton Dr. Jim Hain Mr. Doug Pengilly Dr. Harriet M. Perry Dr. John Zardus Mr. Owen Hamel Dr. Dan J. Pondella 484 Fishery Bulletin 106(4) Fishery Bulletin Index Volume 106(1-4), 2008 List of titles 106(1) 1 An ecological analysis of rockfish (Sebastes spp.) assemblages in the North Pacific Ocean along broad- scale environmental gradients, by Christopher N. Rooper 12 Stock assessment of protogynous fish: evaluating measures of spawning biomass used to estimate biological reference points, by Elizabeth N. Brooks, Kyle W. Shertzer, Todd Gedamke, and Douglas S. Vaughan 24 Emerging patterns of species richness, diversity, population density, and distribution in the skates (Rajidae) of Alaska, by Duane E. Stevenson, James W. Orr, Gerald R. Hoff, and John D. McEachran 40 Efficiency and catch dynamics of collapsible square and conical crab pots used in the red king crab ( Para - lithodes camtschaticus ) fishery, by Stian Stiansen, Anders Ferno, Dag Furevik, Terje Jprgensen, and Svein Lqkkeborg 47 The trophic dynamics of summer flounder ( Parali - chthys dentatus) in Chesapeake Bay, by Robert J. Latour, James Gartland, Christopher F. Bonzek, and RaeMarie A. Johnson 58 Using an inverse-logistic model to describe growth increments of blacklip abalone ( Haliotis rubra ) in Tasmania, by Malcolm Haddon, Craig Mundy, and David Tarbath 72 Effects of commercial fishing regulations on stranding rates of bottlenose dolphin ( Tursiops truncatus), by Barbie L. Byrd, Aleta A. Hohn, Fentress H. Munden, Gretchen N. Lovewell, and Rachel E. Lo Piccolo 82 Age, growth, and reproduction of dolphinfish (Cory- phaena hippurus) caught off the coast of North Caro- lina, by Kara L. Schwenke and Jeffrey A. Buckel 93 Importance of shoreface sand ridges as habitat for fishes off the northeast coast of the United States, by James M. Vasslides and Kenneth W. Able 106(2) 111 Species of the rougheye rockfish complex: resurrec- tion of Sebastes melanostictus (Matsubara, 1934) and a redescription of Sebastes aleutianus (Jordan and Evermann, 1898) (Teleostei: Scorpaeniformes), by James W. Orr and Sharon Hawkins 135 The influence of elemental chemistry on the widths of otolith increments in the neon damselfish (Poma- centrus coelestis), by Michael J. Kingsford, Heather M. Patterson, and Matthew J. Flood 143 Seasonal, diel, and lunar spawning periodicities and associated sound production of white seabass (. Atractoscion nobilis), by Scott A. Aalbers 152 Spatial and temporal variability in the relative con- tribution of king mackerel ( Scomberomorus cavalla ) stocks to winter mixed fisheries off South Florida, by Todd R. Clardy, William F. Patterson III, Douglas A. DeVries, and Christopher Palmer 161 Ichthyoplankton community in the Columbia River plume off Oregon: effects of fluctuating oceano- graphic conditions, by Maria M. Parnel, Robert L. Emmett, and Richard D. Brodeur 174 Interactions between adult migratory striped bass ( Morone saxatilis) and their prey during winter off the Virginia and North Carolina Atlantic coast from 1994 through 2007, by Anthony S. Overton, Charles S. Manooch III, Joseph W. Smith, and Kenneth Brennan 183 Identification of larval sea basses (Centropristis spp.) using ribosomal DNA-specific molecular assays, by Mark W. Vandersea, R. Wayne Litaker, Katrin E. Marancik, Jonathan A. Hare, Harvey J. Walsh, Siya Lem, Melissa A. West, David M. Wyanski, Elisabeth H. Laban, and Patricia A. Tester 194 The vulnerability of reproductively active squaretail coralgrouper (Plectropomus areolatus) to fishing, by Kevin L. Rhodes and Mark H. Tupper 204 Mitochondrial DNA markers to identify com- mercial spiny lobster species (Panulirus spp.) from the Pacific coast of Mexico: an application on phyllosoma larvae, by Francisco J. Garcia-Rodri- guez, German Ponce-Diaz, Isabel Munoz-Garcia, Rogelio Gonzalez-Armas, and Ricardo Perez- Enriquez 213 Advantages of using crest nets to sample presettle- ment larvae of reef fishes in the Caribbean Sea, by Cormac J. Nolan and Bret S. Danilowicz 106(3) 225 A probability-based approach to setting annual catch levels, by Kyle W. Shertzer, Michael H. Prager, and Erik H. Williams 233 A nursery site of the Alaska skate ( Bathyraja parmifera) in the eastern Bering Sea, by Gerald R. Hoff List of titles 485 245 Determination of population structure and stock composition of chum salmon (Oncorhynchus keta) in Russia, determined with microsatellites, by Terry D. Beacham, Nataly V. Varnavskaya, Khai D. Le, and Michael H. Wetklo 257 Fish assemblages and indicator species: reef fishes off the southeastern United States, by Kyle W. Shertzer and Erik H. Williams 270 Spatial and temporal distribution of North Atlantic right whales ( Eubalaena glacialis) in Cape Cod Bay, and implications for management, by Owen C. Nichols, Robert D. Kenney, and Moira W. Brown 281 Effects of commercial fishing on local abundance of Pacific cod (Gadus macrocephalus) in the Bering Sea, by M. Elizabeth Conners and Peter Munro 293 Comparison of echogram measurements against data expectations and assumptions for distinguish- ing seafloor substrates, by Mark Zimmermann and Christopher Rooper 305 Age-specific movement patterns of sablefish ( Anoplo - poma fimbria ) in Alaska, by Nancy E. Maloney and Michael F. Sigler 317 Distribution of red deepsea crab (Chaceon quinque- dens) by size and sex in the Gulf of Mexico, by Morgan J. Kilgour and Thomas C. Shirley 321 Catches in ghost-fishing octopus and fish traps in the northeastern Atlantic Ocean (Algarve, Portugal), by Karim Erzini, Luis Bentes, Rui Coelho, Pedro G. Lino, Pedro Monteiro, Joaquim Ribeiro, and Jorge M. S. Congalves 328 Age- and length-at-maturity of female arrowtooth flounder ( Atheresthes stomias) in the Gulf of Alaska, by James W. Stark 106(4) 337 An assessment of discard mortality for two Alaskan crab species, Tanner crab (Chionoecetes bairdi ) and snow crab (C. opilio), based on reflex impairment, by Allan W. Stoner, Craig S. Rose, J. Eric Munk, Carwyn F. Hammond, and Michael W. Davis 348 Fishes associated with pelagic Sargassum and open water lacking Sargassum in the Gulf Stream off North Carolina, by Tara L. Casazza and Steve W. Ross 364 Evidence of hook competition in longline surveys in Alaska, by Cara J. Rodgveller, Chris R. Lunsford, and Jeffrey T. Fujioka 375 Age validation of Dover sole ( Microstomus pacificus ) by means of bomb radiocarbon, by Craig R. Kastelle, Delsa M. Anderl, Daniel K. Kimura, and Chris G. Johnston 386 A 4500-year times series of Pacific cod (Gadus macrocephalus) size and abundance: archaeology, oceanic regime shifts, and sustainable fisheries, by Herbert D. G. Maschner, Matthew W. Betts, Kather- ine L. Reedy-Maschner, and Andrew W. Trites 395 Prevalence of the commensal barnacle Xeno- balanus globicipitis on cetacean species in the east- ern tropical Pacific Ocean, and a review of global occurrence, by Emily A. Kane, Paula A. Olson, and Paul C. Fiedler 405 Onshore-offshore distribution and abundance of tuna larvae (Pisces: Scombridae: Thunnini) in near- reef waters of the Coral Sea, by Ashley M. Fowler, Jeffrey M. Leis, and Iain M. Suthers 417 Habitat and bycatch effects on population para- meters of inshore lizardfish (Sy nodus foetens) in the north central Gulf of Mexico, by Sarah A. Jeffers, William F. Patterson III, and James H. Cowan Jr. 427 Effects of rapid decompression and exposure to bright light on visual function in black rockfish ( Sebastes melanops) and Pacific halibut (Hippo- glossus stenolepis). by Richard Brill, Christopher Magel, Michael Davis, Robert Hannah, and Polly Rankin 438 Changes in a benthic megafaunal community due to disturbance from bottom fishing and the establish- ment of a fishery closure, by Rebecca G. Asch, and Jeremy S. Collie 457 Pelagic behavior of adult Greenland halibut ( Rein - hardtius hippoglossoides ), by Tone Vollen and Ole T. Albert 471 All their eggs in one basket: a rocky reef nursery for the longnose skate (Raja rhina Jordan & Gilbert, 1880) in the southern California Bight, by Milton S. Love, Donna M. Schroeder, Linda Snook, Anne York, and Guy Cochrane 476 Microsatellite primers for red drum (Sciaenops ocel- latus), by Sten Karlsson, Mark A. Renshaw, Caird E. Rexroad III, and John R. Gold 486 Fishery Bulletin 106(4) Fishery Bulletin Index Volume 106(1-4), 2008 List of authors Aalbers, Scott A. 143 Able, Kenneth W. 93 Albert, Ole T. 457 Anderl, Delsa M. 375 Asch, Rebecca G. 435 Beacham, Terry D. 245 Bentes, Luis 321 Betts, Matthew W. 386 Bonzek, Christopher F. 47 Brennan, Kenneth 174 Brill, Richard 427 Brodeur, Richard D. 161 Brooks, Elizabeth N. 12 Brown, Moira W. 270 Buckel, Jeffrey A. 82 Byrd, Barbie L. 72 Casazza, Tara L. 348 Clardy, Todd R. 152 Cochrane, Guy 471 Coelho, Rui 321 Collie, Jeremy S. 435 Congalves, Jorge M. S. 321 Conners, M. Elizabeth 281 Cowan Jr., James H. 417 Danilowicz, Bret S. 213 Davis, Michael 327 DeVries, Douglas A. 152 Emmett, Robert L. 161 Erzini, Karim 321 Ferno, Anders 40 Fiedler, Paul C. 395 Flood, Matthew J. 135 Fowler, Ashley M. 405 Fujioka, Jeffrey T. 364 Furevik, Dag 40 Garcia-Rodriguez, Francisco J. 204 Gartland, James 47 Gedamke, Todd 12 Gold, John R. 476 Gonzalez-Armas, Rogelio 204 Haddon, Malcolm 58 Hammond, Carwyn F. 337 Hannah, Robert 427 Hare, Jonathan A. 183 Hawkins, Sharon 111 Hoff, Gerald R. 24,233 Hohn, Alefa A. 72 Jeffers, Sarah A. 417 Johnson, RaeMarie A. 47 Johnston, Chris G. 375 Jorgensen, Terje 40 Kane, Emily A. 395 Karlsson, Sten 476 Kastelle, Craig R. 375 Kenney, Robert D. 270 Kilgour, Morgan J. 317 Kimura, Daniel K. 375 Kingsford, Michael J. 135 Laban, Elisabeth H. 183 Latour, Robert J. 47 Le, Khai D. 245 Leis, Jeffrey M. 405 Lem, Siya 183 Lino, Pedro G. 321 Litaker, R. Wayne 183 Lo Piccolo, Rachel E. 72 Lpkkeborg, Svein 40 Love, Milton S. 471 Lovewell, Gretchen N. 72 Lunsford, Chris R. 364 Magel, Christopher 427 Maloney, Nancy E. 305 Manooch, Charles S., Ill 174 Marancik, Katrin E. 183 Maschner, Herbert D. G. 386 McEachran, John D. 24 Monteiro, Pedro 321 Munden, Fentress H. 72 Mundy, Craig 58 Munk, J. Eric 337 Munoz-Garcia, Isabel 204 Munro, Peter 281 Nichols, Owen C. 270 Nolan, Cormac J. 213 Olson, Paula A. 395 Orr, James W. 24, 111 Overton, Anthony S. 174 Palmer, Christopher 152 Parnel, Maria M. 161 Patterson, Heather M. 135 Patterson, William F., Ill 152, 417 Perez-Enriquez, Ricardo 204 Ponce-Dlaz, German 204 Prager, Michael H. 225 Rankin, Polly 427 Reedy-Maschner, Katherine L. 386 Renshaw, Mark A. 476 Rexroad III, Caird E. 476 Rhodes, Kevin L. 194 Ribeiro, Joaquim 321 Rodgveller, Cara J. 364 Rooper, Christopher N. 1, 293 Rose, Craig S. 337 Ross, Steve W. 348 Schroeder, Donna M. 471 Schwenke, Kara L. 82 Shertzer, Kyle W. 12, 225, 257 Shirley, Thomas C. 317 Sigler, Michael F. 305 Smith, Joseph W. 174 Snook, Linda, 471 Stark, James W. 328 Stevenson, Duane E. 24 Stiansen, Stian 40 Stoner, Allan W. 337 Suthers, Iain M. 405 Tarbath, David 58 Tester, Patricia A. 183 Trites, Andrew W. 386 Tupper, Mark H. 194 Vandersea, Mark W. 183 Varnavskaya, Nataly V. 245 Vasslides, James M. 93 Vaughan, Douglas S. 12 Vollen, Tone 457 Walsh, Harvey J. 183 West, Melissa A. 183 Wetklo, Michael H. 245 Williams, Erik H. 225, 257 Wyanski, David M. 183 York, Anne 471 Zimmermann, Mark 293 487 Fishery Bulletin Index Volume 106(1-4), 2008 List of subjects Abalone, blacklip 58 Abundance cetacean barnacle 395 relative 364 Acoustics 93 telemetry 194 Age and growth (dolphinfish) 2 -at-maturity 328 known (sablefish) 305 -specific movement patterns 305 validation 375 Aggregation 395 Alaska 1, 364 southeastern 305, 364 Albacore 405 Albatrossia pectoralis - see grenadier, giant Aleutian Islands Rajidae 24 Allometric reconstruction 396 Anchoa mitchilli — see anchovy, bay Anchovy bay 174 northern 161 Anoplopoma fimbria - see sablefish Archaeology 386 Atheresthes stomias - see flounder, arrowtooth Atlantic Ocean northeastern 321 Atractoscion nobiiis - see sea bass, white Australia 405 Tasmania 58 BACI experimental design 281 Barents Sea 40 Barotrauma 427 Bass, striped 174 Bathyraja parmifera - see skate, Alaska Behavior pelagic 457 Bering Sea 281, 337 Rajidae 24 Bioacoustics 143 Biological reference points (BRPs) 12 Biological tag 395 Bomb radiocarbon 375 Bottom fishing 438 Brevoortia tyrannus — see menhaden, Atlantic Bycatch crab 337 fisheries 72 inshore lizardfish 417 Canonical correspondence analysis 93 Caribbean Sea 213 Catch -and-release 427 limits, annual 225 rates 364 Catchment areas 194 Centropristis ocyurus - see sea bass, bank philadelphica - see sea bass, rock striata - see sea bass, black Cetacean barnacle 395 Chaceon quinquedens - see crab, red deepsea Chesapeake Bay 47 Chionoecetes spp. 337 bairdi - see crab, Tanner opilio - see crab, snow Cluster analysis 257 Cod, Pacific 281,386 Colonial epifauna 435 Columbia River 161 Commensal 395 Community benthic 438 biological 161 Coral Sea 405 Coralgrouper, squaretail 194 Coryphaena hippurus - see dolphinfish Crab 337 red deepsea 317 red king 40 snow 337 Tanner 337 Critical habitat 270 Daily growth increments (dolphinfish) 82 Damselfish, neon 135 Diet diversity (summer flounder) 47 striped bass 174 Distribution cetacean barnacle 395 pelagic (Greenland halibut) 457 prey 457 sex (red deepsea crab) 317 size (red deepsea crab) 317 spatial (North Atlantic right whale) 270 temporal (North Atlantic right whale) 270 tuna larvae 405 Diversity 213 Dogfish, spiny 72 Dolphin, bottlenose 72 Dolphinfish 82 Drum, red 476 Eastern Bering Sea 233 Echosounder 293 Egg case (Rajidae) 233,471 Elasmobranch 471 Electroretinogram (ERG) 427 Engraulis mordax - see anchovy, northern Environmental factors 93 gradients 1 variables 1 Eubalaena glacialis - see right whale, North Atlantic Eumetopias jubatus — see sea lion, Stellar Filefish, planehead 348 Fish assemblages 1, 93, 257 coral reef 213 distributions 1 juvenile 348 protogynous 12 reef 135 sound production 143 Fish spawning aggregation (FSA) 194 Fishery closure 438 management 72, 225, 257 regulation 270 Flounder arrowtooth 328 summer 47 Fourier analysis 152 Gadus TJiacrocephalus - see cod, Pacific Gag 225 Gear saturation 364 Genetics, identification (lobster) 204 Georges Bank 438 488 Fishery Bulletin 106(4) Ghost fishing 321 Great Barrier Reef 135 Grenadier, giant 364 Growth inshore lizardfish 417 seasonal (blacklip abalone) 58 Gulf of Alaska 328,386 Rajidae 24 Gulf of Mexico 152,317,417 Gulf Stream 348 Habitat disturbance 438 fish associations 93 inshore lizardfish 417 nursery 348 open-water 348 Rajidae 233 Sargassum 348 Halibut Greenland 457 Pacific 427 Haliotis rubra - see abalone, blacklip Harvest strategy 225 Hatching date (dolphinfish) 82 Hippoglossus stenolepis - see halibut, Pacific Hook competition 364 Host, cetacean barnacle 395 Ichthyoplankton 161 Identification, lobster 204 Indicator species 257 Internal transcribed spacer 183 Katsuwonus pelamis - see tuna, skipjack Larval fish 213 tuna 405 Length-at-maturity 328 Lizardfish, inshore 417 Lobster, spiny 204 Localized depletion 281 Longline, vertical 457 Lunar periodicity 213 Mackerel, king 152 Marine protected areas (MPAs) 194 Maximum sustainable yield ( MSY ) 1 2 Menhaden, Atlantic 174 Microsatellite primers 476 chum salmon 245 Microstomus pacificus — see sole, Dover Migration (sablefish) 305 Mitochondrial DNA, spiny lobster 204 Modeling, inverse-logistic 58 Monitoring (North Atlantic right whale) 270 Morone saxatilis — see bass, striped Mortality discard (crab) 337 inshore lizardfish 417 Multidimensional scaling (MDS) 257 Mycteroperca microlepis - see gag Nets channel 213 crest 213 New Jersey 93 North Carolina 82 Nursery grounds (Rajidae) 471 site (Rajidae) 233 Oceanographic conditions 161 Oncorhynchus keta - see salmon, chum Otolith 135,375 elemental chemistry 135 increment widths 135 inshore lizardfish 417 king mackerel 152 water chemistry 135 Overfishing 225 Oviparous, Rajidae 233 Pacific Ocean Eastern tropical 395 North 1, 111 Palaeofisheries 386 Panulirus spp. 204 Paralichthys dentatus - see flounder, summer Paralithodes camtschaticus — see crab, red king Phoresis 395 Photography, in situ 438 Phyilosoma larvae 204 Plectropom us areolatus - see coralgrouper, squaretail Polymerase chain reaction ( PCR) 183, 476 Pomacentrus coelestis - see damselfish, neon Population, structure 245 Portugal 321 Pot catch dynamics (crab) 40 design (crab) 40 efficiency (crab) 40 sex selection (crab) 40 size selection (crab) 40 soak time (crab) 40 Predation, striped bass 174 Principal components analysis (PCA) 293 Raja rhina - see skate, longnose Rajidae 24,233,471 Reflex impairment (crab) 337 Reinhardtius hippoglossoides - see halibut, Greenland Reproduction dolphinfish 82 migratory corridors 194 Restriction fragment length polymorphism (RFLP) 183 spiny lobster 204 Right whale, North Atlantic 270 Rockfish 1, 111 black 427 blackspotted 111 rougheye 111, 364 shortraker 364 Russia 245 Sablefish 364 juvenile 305 Salmon, chum 245 Sciaenidae 143 Sciaenops ocellatus - see drum, red Scomberomorus cavalla — see mackerel, king Scombridae 495 Sea bass bank 183 black 183 rock 183 white 143 Sea lion, Stellar 281 Seafloor substrates 293 Sebastes spp. 1, 111 aleutianus - see rockfish, rougheye borealis - see rockfish, shortraker melanops - see rockfish, black melanostictus - see rockfish, blackspotted Serranidae 194 Ship strikes (North Atlantic right whale) 270 Shoreface sand ridge 93 Skates 24 Alaska 24, 233 longnose 471 Snapper-grouper complex 257 Sole, Dover 375 Southeastern United States 257 Southern California Bight 471 List of subjects 489 Spawning biomass 12 flounder, arrowtooth 328 periodicity 143 Squalus acanthias — see dogfish, spiny Stephanolepis hispidus - see filefish, planehead Stock assessment 12 composition 245 data-limited 257 identification 245 management 328 mixing 152 Stomach content analysis 174 Stranding marine mammal 72 rates (bottlenose dolphin) 72 Survey aerial 270 longline 364 Synodus foetens — see lizardfish, inshore Systematics (rockfish) 111 Tag archival 457 sablefish 305 Taxonomy, rockfish 111 Temperature 161 Thunnus spp. 405 alalunga - see albacore albacares - see tuna, yellowfin Traps fish 321 octopus 321 Trophic dynamics 174 (summer flounder) 47 ecology (summer flounder) 47 Tuna skipjack 405 yellowfin 405 Tursiops truncates - see dolphin, bottlenose Uncertainty 225 Vision (fish) 427 Xenobalanus globicipitis - see cetacean barnacle Zoogeography (Rajidae) 24 490 Fishery Bulletin 106(4) Fishery Bulletin Guidelines for authors Content of manuscripts Contributions published in Fishery Bulletin describe original research in marine fishery science, fishery engineering and economics, as well as the areas of marine environmental and ecological sciences (including modeling). Preference will be given to manuscripts that examine processes and underlying patterns. Descriptive reports, surveys, and observational papers may occa- sionally be published but should appeal to an audience outside the locale in which the study was conducted. Although all contributions are subject to peer review, responsibility for the contents of papers rests upon the authors and not on the editor or publisher. Submission of an article implies that the article is original and is not being considered for publication elsewhere. Articles may range from relatively short contributions (10-15 typed, double-spaced pages, tables and figures not included) to extensive contributions (20-30 typed pages). Notes are reports of 5 to 10 pages without an abstract and describe methods or results not supported by a large body of data. Manuscripts must be written in English; authors whose native language is not English are strongly advised to have their manuscripts checked by English-speaking colleagues before submission. Manuscript Preparation Title page should include authors’ full names and mail- ing 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 200 words (one-half typed page), state the main scope of the research, and emphasize the author’s conclusions and relevant findings. Do not review the methods of the study or list the contents of the paper. Because abstracts are circulated by abstract- ing 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 Conclusions), and Acknowl- edgments. Headings within each section must be short, reflect a logical sequence, and follow the rules of multiple subdivision (i.e., there can be no subdivision without at least two items). The entire text should be intelligible to interdisciplinary readers; therefore, all acronyms, abbreviations, and technical terms should be written out in full the first time they are mentioned. Include FAO common names for species in the list of keywords and in the introduction. Regional common names may be used throughout the rest of the text if they are dif- ferent from FAO common names which 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; for fish nomenclature follow the most current issue of the American Fisheries Society’s Common and Scientific Names of Fishes from the United States and Canada. Dates should be written as follows: 11 November 2000. Measurements should be expressed in metric units, e.g., 58 metric tons (t); if other units of measurement 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). Refrain from using the shorthand slash (/), an ambigu- ous symbol, in the general text Literature cited comprises published works and those accepted for publication in peer-reviewed literature (in press). Follow the name and year system for citation format in the “Literature cited” section (that is say, citations should be listed alphabetically by the authors’ last names, and then by year if there is more than one citation with the same authorship). If there is a sequence of citations in the text, list chronologically: (Smith, 1932; Green, 1947; Smith and Jones, 1985). Abbrevia- tions of serials should conform to abbreviations given in the Serial Sources for the BIOSIS Previews Database. Authors are responsible for the accuracy and complete- ness of all citations. Literature citation 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. Cite all software and special equipment or chemical 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). 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. • Do not place outline rules around graphs. • Do not use horizontal lines through graphs to indicate measurement units. • Use a comma in numbers of five digits or more (e.g. 13,000 but 3000). • Maps require a North arrow and degrees latitude- longitude (e.g., 170°E). 491 Tables are often overused in scientific papers; it is seldom necessary or even desirable to present all the data associated with a study. Tables should not be excessive in size and must be cited in numerical order in the text. Headings should be short but ample enough to allow the table to be intelligible on its own. All unusual symbols must be explained in the table legend. Other incidental comments may be footnoted with italic numeral footnote markers. Use asterisks to indicate probability in statisti- cal data. Do not type table legends on a separate page; place them above the table data. Do not submit tables in photo mode. Figures include line illustrations, photographs (or slides), and computer-generated graphs and must be cited in numerical order in the text. Graphics should aid in the comprehension of the text, but they should be limited to presenting patterns rather than raw data. Figures should not exceed one figure for every four pages of text. Figures must be labeled with author’s name and number of the figure. Avoid placing labels vertically (except of y axis). 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 and failure to correspond with editors in a timely manner will delay publication of a manuscript. Copyright law does not apply to Fishery Bulletin, 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 consid- ered correct form (e.g., Source: Fish. Bull 97:105). Submission The Scientific Editorial Office encourages authors to submit their manuscripts as a single PDF (preferred) or Word (zipped) document by e-mail to Fishery. Bulletin@noaa.gov. Please use the subject heading, “Fishery Bulletin manuscript submission”. Do not send encrypted files. For further details on electronic sub- mission, please contact the Scientific Editorial Office directly (see address below). Or you may send your manuscript on a compact disc in one of the above formats along with four printed copies (one original plus three copies [stapled]) to the Scientific Editor, at the address shown below. Dr. Adam Moles Scientific Editor, Fishery Bulletin NOAA/NMFS/AFSC 17109 Point Lena Loop Road Juneau, AK 99801-8626 Once the manuscript has been accepted for publication, you will be asked to submit a final electronic 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, tiff files, or EPS files. Send a copy of figures in the origi- nal software if conversion to any of these formats yields a degraded version. Questions? If you have questions regarding these guidelines, please contact the Managing Editor, Sharyn Matriotti, at Sharyn.Matriotti@noaa.gov Questions regarding manuscripts under review should be addressed to Adam Moles, Scientific Editor, at Adam. Moles@noaa.gov Superintendent of Documents Publications Order Form *5178 □ YES, please send me the following publications: Subscriptions to Fishery Bulletin for $36.00 per year ($50.40 foreign) The total cost of my order is $ . Prices include regular domestic postage and handling and are subject to change. (Company or Personal Name) (Please type or print) (Additional address/attention line) (Street address) (City, State, ZIP Code) (Daytime phone including area code) (Purchase Order No.) Charge your order. IT’S EASY! Please Choose Method of Payment: | | Check Payable to the Superintendent of Documents | | GPO Deposit Account | 1 1 1 | | | ] — Q ^ | VISA or MasterCard Account your orders | | - | ■ | j- | i"- • -| | | | | | | | | j— j— | (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 22 August, 2008. The Bulletin is published quarterly (four issues annually) with an annual subscription price of $36.00 (sold by the Superintendent of Documents, U.S. Government Printing Office, Washington, DC 20402). The complete mailing address of the office of publication is NMFS Scientific Publications Office, NOAA, 7600 Sand Point Way NE, Seattle, WA 98115. The complete mailing address of the headquarters of the publishing agency is National Marine Fisheries Service, NOAA, Department of Commerce, 1315 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, Seattle, WA 98115. The owner is the U.S. Department of Commerce, 14th St. N.W., Washington, DC 20230; there are no bond- holders, mortgages, or other security holders. 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 1038 and the actual number of copies of the single issue pub- lished nearest to the filing dates was 1034. Paid circulation (B) is handled by the U.S. Government Printing Office, Washington, DC 20402, and the total number printed for sales (mail subscriptions and individual sales) was 225 for the average number of copies each issue during the preceding 12 months and 225 the actual number of copies of the single issue published nearest to the filing date (C). Free distribution (D) by mail; samples, compli- mentary and other free copies (average number of copies each issue during the preceding 12 months) was 773 and the actual number of copies of the single issue published nearest to the filing date was 769. Free distribution outside the mail (E) by carriers or other means was 0 for both average number of copies and actual number of copies. Total free distribution (F) was 0 for both average number of copies and actual number of copies of the single issue published nearest the filing date. The total distribution (G: sum of D and B) (average number of copies each issue during the preceding 12 months) was 1016 and the actual number of copies of the single issue published nearest to the filing date was 994. There were 40 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: 1038 and 1034 copies, respectively. I certify that the statements made by me above are correct and complete: (Signed) Willis Hobart, Publisher. SMITHSONIAN INSTITUTION LIBRARIES 3 9088 01444 9623