r%Sf2^r^^s^7ri^iyS^2^ U.S. Department of Commerce Volume 105 Number 1 January 2007 Fishery Bulletin U.S. Department of Commerce Carlos M. Gutierrez Secretary National Oceanic and Atmospheric Administration Vice Admiral Conrad C. Lautenbacher Jr., USN (ret) Under Secretary for Oceans and Atmosphere National Marine Fisheries Service William T. Hogarth Assistant Administrator for Fistieries Scientific Editor Adam Moles, Ph.D. Associate Editor Elizabeth Calvert National Marine Fisheries Service, NOAA 11305 Glacier Highway Juneau, Alaska 99801-8626 ^°""\ ^ATES 0» ' \ / Managing Editor Sharyn Matriotti National Marine Fishenes Sen/ice Scientific Publications Office 7600 Sand Point Way NE, BIN C15700 Seattle, Washington 98115-0070 The Fishery Bulletin (ISSN 0090-0656) is published quarterly by the Scientific Publications Office, National Marine Fisheries Service, NOAA, 7600 Sand Point Way NE, BIN C 15700, 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: $28.00 domestic and $35.00 foreign. See back for order form. Editorial Committee Harlyn O. Halvorson, Ph.D. Ronald W. Hardy, Ph.D. Richard D. Methot, Ph.D. Theodore W. Pietsch, Ph.D. Joseph E. Powers, Ph.D. Harald Rosenthal, Ph.D. Fredric M. Serchuk, Ph.D. George Walters, Ph.D. University of Massachusetts, Boston University of Idaho, Hagerman National Marine Fisheries Service University of Washington, Seattle National Marine Fisheries Service Universitat Kiel, Germany National Marine Fisheries Service National Marine Fisheries Service Fishery Bulletin web site: www.fishbull.noaa.gov The Fishery Bulletin carries original research reports and technical notes on investigations in fishery science, engineering, and economics. It began as the Bulletin of the United States Fish Commission in 1881; it became the Bulletin of the Bureau of Fisheries in 1904 and the Fisheiy 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 105 Number 1 January 2007 Fishery Bulletin Contents Articles 1-18 ^N Massachusero Wexler, Jeanne B., Seinen Chow, Toshie Wakabayashi, Kenji Nohara, and Daniel Margulies Temporal variation in growth of yellowfin tuna (.Thunnus albacares) larvae in the Panama Bight, 1990-97 19—29 McDermott, Susanne F., Katherine P. Maslenikov, and Donald R. Gunderson Annual fecundity, batch fecundity, and oocyte atresia of Atka mackerel (Pleurogrammus monopterygius) in Alaskan waters The conclusions and opinions ex- pressed in Fishery Bulletin are solely those of the authors and do not represent the official position of the National Marine Fisher-ies Ser- vice iNOAAl or any other agency or institution. The National Marine Fisheries Service (NMFSl 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. 30-38 Hobbs, James A., William A Bennett, Jessica E Burton, and Bradd Baskerville-Bridges Modification of the biological intercept model to account for ontogenetic effects in laboratory-reared delta smelt {Hypomesus transpacificus) 39—48 Laidig, Thomas E., James R. Chess, and Daniel F. Howard Relationship between abundance of juvenile rockfishes (Sebastes spp.) and environmental variables documented off northern California and potential mechanisms for the covariation 49—61 Harley, Shelton J„ and Jenny M, Suter The potential use of time-area closures to reduce catches of bigeye tuna (Thunnus obesus) in the purse-seme fishery of the eastern Pacific Ocean 62-73 Secor, David H., and Philip M. Piccoli Oceanic migration rates of Upper Chesapeake Bay striped bass (Morone saxatiiis), determined by otolith microchemical analysis Fishery Bulletin 105(1) 74—87 Matkin, Craig O., Lance G. Barrett-Lennard, Harald Yurk, David Ellifrit, and Andrew W. Trites Ecotypic variation and predatory behavior among killer whales (Orcinus orca) off the eastern Aleutian Islands, Alaska 88—101 Stockhausen, William T., and Michael Fogarty Removing observational noise from fisheries-independent time series data using ARIMA models 102—115 Pitcher, Kenneth W., Peter F. Olesiuk, Robin F. Brown, Mark S. Lowry,. Steven J. Jeffries, John L. Sease, Wayne L. Ferryman, Charles E. Stinchcomb, Lloyd F. Lowry Abundance and distribution of the eastern North Pacific Steller sea lion (Eumetopias jubatus) population 116—120 Gerritsen, Hans D., and David McGrath Precision estimates and suggested sample sizes for length-frequency data 121—130 Farley, Edward V. Jr, James M. Murphy, Mile D. Adkison, Lisa B. Eisner, John H. Helle, Jamal H. Moss, and Jennifer Nielsen Early marine growth in relation to marine-stage survival rates for Alaska sockeye salmon (.Oncorhynchus nerka) 131—139 Ganias, Konstantinos, Cristina Nunes, and Yorgos Stratoudakis Degeneration of postovulatory follicles in the Ibersian sardine Sordino pilchardus: structural changes and factors affecting resorption Notes 140—146 Arrizabalaga, Haritz, Victoria Lopez-Rodas, Eduardo Costas, and Alberto Gonzalez-Garces Use of genetic data to assess the uncertainty in stock assessments due to the assumed stock structure: the case of albacore (Thunnus olalunga) from the Atlantic Ocean 147—152 Cherel, Yves, Richard Sabatie, Michel Potier, Francis Marsac, and Frederic Menard New information from fish diets on the importance of glassy flying squid (Hyoloteuthis pelagica) (Teuthoidea: Ommastrephidae) in the epipelagic cephalopod community of the tropical Atlantic Ocean 153—157 Conti, Stephane G., Benjamin D. Maurer, Mark A. Drawbridge, and David A. Demer Measurements of total scattering spectra from bocaccio (Sebostes paucispinis) 158-159 Guidelines for authors Abstract — Tuna larvae (at flexion, postflexion, and transformation stages) were collected by dip net and light traps at night in the northwest- ern Panama Bight during the season of reduced upwelling (June-September) of 1990, 1991, 1992, and 1997. The larvae were identified as yellowfin tuna {Thunnus alhacares) by mtDNA analysis. Ichthyoplankton data from bongo and Tucker trawl tows were used to examine the potential prey abundance in relation to the mean size-at-age and growth rates of the yellowfin tuna larvae and their oto- liths. The most rapid growth rates occurred during June 1990 when plankton volumes were at their highest levels. The lowest plankton volumes coincided with the lowest growth rates and mean sizes-at-age during the August-September 1991 period. High densities of larval fish were prevalent in the ichthyoplankton tows during the 1991 period; therefore intra- and interspecific competition for limited food resources may have been the cause of slower growth (den- sity-dependent growth) in yellowfin tuna larvae The highest mean sea- surface temperature and the lowest mean wind stress occurred during an El Nino-Southern Oscillation (ENSO) event during the 1997 period. There appeared to be no clear association between these environmental fac- tors and larval growth rates, but the higher temperatures may have caused an increase in the short-term growth of otoliths in relation to larval fish size. Temporal variation in growth of yellowfin tuna iThunnus aibacares) larvae in the Panama Bight, 1990-97 Jeanne B. Wexler (contact author)' Seinen Chow^ Toshie WakabayashP Kenji Nohara^ Daniel Margulies' Email address for J. B, Wexler: iwexler@iattc.org ' Inter-Amencan Tropical Tuna Commission 8604 La Jolla Stiores Drive La Jolla, California 92037-1508 2 National Research! Institute of Fisheries Science Nagai 6-31-1 Yokosuka Kanagawa 283-0316 Japan ^ National Research Institute of Far Seas Fisheries 5-7-1 Shimizu-Orido Shizuoka 424-8633 Japan Manuscript submitted 22 October 2004 to the Scientific Editor's Office. Manuscript accepted 14 March 2006 by the Scientific Editor. Fish. Bull. 105:1-18 (2007). Yellowfin tuna iThunnus aibacares) larvae inhabit the mixed layer of all tropical and subtropical oceans of the world (Ueyanagi, 1969; Nishikawa et al., 1985). When recruited to the com- mercial fishery, yellowfin tuna are one of the most important tuna species worldwide (Collette and Nauen, 1983; FAO, 2004). Near-daily spawning of yellowfin tuna, and the subsequent dispersal of fertilized eggs, appears to be largely dependent on the occur- rence of surface water temperatures equal to or greater than 24°C (Schae- fer, 1998). In the eastern Pacific Ocean (EPO), yellowfin tuna spawn continuously between 0° and 20°N (Schaefer, 20011. Despite widespread spawning of yellowfin tuna through- out the EPO, the larvae are patchy in distribution (Ahlstrom, 1971), and relatively large numbers have been collected only near islands (Graves et al., 1988; this study) and near shore (Gonzalez Armas, 2002). The larvae of Thunnus are difficult to identify by meristic, morphological, or pigmentation characteristics (Mat- sumoto et al., 1972; Potthoff, 1974; Richards et al., 1990; Lang et al., 1994). In the EPO, the late-larval and early-juvenile stages of yellowfin and bigeye (T. obesus) tuna co-exist and cannot be differentiated by these con- ventional methods. However, allozyme (Graves, et al., 1988) and recent mo- lecular (Takeyama et al., 2001; Chow et al., 2003) analyses have made it feasible to identify larvae of these two species that inhabit the EPO. The growth dynamics of yellowfin tuna during early life stages may have a profound effect on cohort strength (Houde, 1987), but growth rates have not been described for the larvae in the Pacific Ocean. Larval and juvenile stage durations and corresponding growth rates (Houde, 1989), starvation rates (Margulies, 1993), and larval transport and pre- dation (Grimes, 2001) may be strongly influenced by biological and physical processes that would affect prerecruit survival in yellowfin tuna. Standing stocks of phytoplankton and zoo- plankton in the EPO, where yellow- fin tuna larvae are found are season- ally variable (Blackburn et al., 1970; Owen and Zeitschel, 1970; Lauth and Olson, 1996; Gonzalez Armas, 2002) and influenced by interannual events such as El Nino-Southern Oscilla- tion (ENSO) conditions (Dessier and Donguy, 1987; Fiedler, 1992; Chavez et al., 1999; Strutton and Chavez, 2000). In the northwestern Panama Bight of the EPO, nearshore ichthyo- plankton surveys (from 1989 to 1993) Fishery Bulletin 105(1) Table 1 Collections of yellowfi n tuna larvae by n ight-lighting (NL) and light traps (LT) near Frailes del Sur in the northwestern Panama Bight, 1990-1997. Number used for ( ) Number used and identified Standard Number of Number of for age and as T. albacares by Age range length Sampling period sampling dates larvae collected growth analyses PCR-RFLP analysis (days) range (mm) 21-26 June 1990 3 97 (NL) 25 (5)1 8-18 6.2-19.6 5-25 July 1991 5 13(NL)9(LT) 13 (13)10 11-15 9.1-12.7 4-7 September 1991 2 126 (NL) 43 (34)26 12-20 7.1-12.4 24 June-3 July 1992 3 47 (NL) 22 (34)19 10-14 7.6-12.0 7 August 1997 1 98 (NL) 69 (71169 11-18 8.7-14.5 (lATTC; IATTC2; Lauth and Olson, 1996; OwenS) and experiments with captured scombrid larvae (from 1986 to 1997) at the Achotines Laboratory of the Inter-Ameri- can Tropical Tuna Commission (lATTC) (Olson and Scholey, 1990; Margulies, 1993; Scholey, 1993; Wexler, 1993) have provided an opportunity to explore factors controlling prerecruit growth and survival of scom- brids. These small- and fine-scale studies may provide some understanding of the recruitment variability of yellowfin tuna in the Panama Bight, considering that yellowfin tuna exhibit limited, small-scale movements within the EPO (Schaefer, 1991; Wild, 1994) and that processes important to recruitment probably occur at small scales (Fortier and Leggett, 1985). The Panama Bight is characterized by distinct sea- sonal and interannual variations in atmospheric and oceanic conditions (Wooster, 1959; Smayda, 1963, 1966; Forsbergh, 1963, 1969). The climatological and physical oceanographic properties that occur within the Pana- ma Bight are determined by the north-south seasonal movement of the northeast trade winds of the Atlantic Ocean, the equatorial calm belt (i.e., the doldrums), the southeast trade winds of the Pacific Ocean, and the convergence of these trade wind systems within the doldrums (i.e., the intertropical convergence zone, ITCZ) (Smayda, 1966). From January through April, the ITCZ is displaced to the south and strong north- erly trade winds create a dry season and produce local upwelling. From about May through December, the ITCZ is displaced to the north and the Panama Bight is dominated by southeast trade winds and a rainy season characterized by reduced upwelling, higher sea- surface temperatures (SSTs), lower ocean salinities, and a deeper thermocline and mixed layer (Lauth and Olson, 1996). The growth and subsequent survival of yellowfin tuna larvae that occur during the reduced upwelling season may be regulated more by the spatial patchiness of prey organisms coincident with lower plankton volumes (Owen, 1989). ENSO events could further affect the seasonal availability of nutrients and food organisms during this period (Barber and Chavez, 1986; Dessier and Donguy, 1987; Fiedler, 1992; Chavez et al., 1999). A mild ENSO event occurred during our sampling periods in 1991-92 (Barber et al., 1996) and a strong event occurred in late 1997 (Chavez et al., 1999; Strutton and Chavez, 2000; Glynn et al., 2001). The objectives of this study were 1) to identify the species of Thunnus sampled in the northwestern Pan- ama Bight by molecular analysis, 2) to determine ages and compare the size-at-age data of yellowfin tuna lar- vae collected during the periods of reduced upwelling of 1990, 1991, 1992, and 1997, and 3) to explore relation- ships between the temporal variation in growth rates and measured levels of plankton and physical processes in the Panama Bight. Materials and methods ' lATTC (Inter-American Tropical Tuna Commission). 1992. Annual report of the Inter-American Tropical Tuna Commission 1990, 261 p. lATTC, 8604 La Jolla Shores Drive, La Jolla, CA 92037. 2 lATTC (Inter-American Tropical Tuna Commission). 1992. Annual report of the Inter-American Tropical Tuna Commission 1991, 271 p. lATTC, 8604 La Jolla Shores Drive, La Jolla, CA 92037. ^ Owen, R. W. 1997. Oceanographic atlas of habitats of larval tunas in the Pacific Ocean off the Azuero Peninsula, Panama, 32 p. Inter-American Tropical Tuna Commission Data Report 9. lATTC, 8604 La Jolla Shores Drive, La Jolla, CA 92037. Larval fish collections Fish larvae were collected in the northwestern Panama Bight (Fig. 1) during the seasons of reduced upwelling in June 1990, July and September 1991, June and July 1992, and August 1997 (Table 1). Most of the larvae were collected with a dipnet just below the ocean surface after they were attracted with an underwater light at night (night-lighting, NL) (Olson and Scholey, 1990) near Frailes del Sur in the vicinity of the 100- and 200-meter isobaths. Larvae were also collected in this area in July 1991 by a light trap (LT) (design described in Thorrold, Wexler et al.: Temporal variation in larval growth of Thunnus olbocares in the Panama Bight 30' yN - 1993) deployed near the surface. All larvae were fixed in 95% ethyl alcohol shortly after capture, except for some that were caught alive and used in laboratory experiments. Fish used in laboratory experiments were not used for the age and growth analyses. SSTs were recorded with a bucket ther- mometer, and the salinity of a sample of water taken just below the surface was measured with a handheld salinometer. Visual observations of environmental con- ditions (e.g., wind, currents, and weather) were recorded at the time of sampling. Laboratory procedures and analyses Larvae of the genus Thunnus were sorted from other scombrid larvae by the morpho- logical features and meristics described in Nishikawa and Rimmer (1987) and Ambrose (1996). The standard length (SL) of each larva was measured in distilled water before the sagittal otoliths were removed for aging and before the remaining tissue of each individual was placed in 95% ethyl alco- hol for species identification. The sagittae were removed, cleaned of tissue with chlo- rine bleach, rinsed in distilled water, dried, and embedded distal side up with Eukitt (O. Kindler, Freiberg, Germany) mount- ing medium on a glass slide. The diameter along the longest axis of each sagitta was measured with an ocular micrometer and light microscope. The sagittae were pol- ished at the surface until the increments were clearly visible with transmitted light at a magnification of 480 or 720x. Daily increments (previously validated in Wexler et al., 2001) of the left and right sagittae were counted "blindly" (i.e., repeated counts were made without prior knowledge of the previous counts) by the first author until the same number of increments were counted at least three times in one of the sagittae. The number of increments in the sagitta that was more clearly read (which usually resulted in a higher count) was used as a direct estimate of age for that fish. The temporal variation in growth was examined by comparing the size-at-age data of the larvae and their otoliths among collection periods through analysis of covariance (ANCOVA) and a multiple range comparison test (Tukey HSD) (XLSTAT vers. 7.5.2, Addinsoft USA, New York, NY) (a=0.05). DNA analysis and species identification The flanking region between ATPase 6 and cytochrome oxidase subunit I (COI) genes of mtDNA was ampli- fied by using the polymerase chain reaction (PCR), 20' 80° W V;^"" i" Caribbean Sea \ iCoslaV ( \ V^^Panama ^-^"^ V'^ Azuero^ \,^p1 \xotomb« Peninsula Pacific Ocean *~' j I'N 20' 80°W Figure 1 Locations where yellowfin tuna iThunnus albacares) larvae were col- lected with an underwater light at night (cross hatched; from 1990-92 and 1997) and where ichthyoplankton sampling occurred (during 1990-92) near the Achotines Laboratory, on the Azuero Peninsula of the northwestern Panama Bight. Ichthyoplankton sampling stations along the Punta Mala and Morro Puercos transects are the following: Mala abyss (MAB), Mala slope (MSL), Mala shelf break (MSB), Mala shelf iMSH), Puercos abyss (PAB), Puercos slope (PSL), Puercos shelf break (PSB), Puercos shelf (PSHJ. and restriction fragment length polymorphism (RFLP) patterns were used to identify the species of Thunnus larvae according to protocols of Takeyama et al. (2001) and Chow et al. (2003). Albacore (T. alalunga), yellowfin, and bigeye tunas in the Pacific Ocean can be identified by the diagnostic restriction profile of Mse I digestion (Chow and Inoue, 1993), and this enzyme assay was used to identify the species of larvae collected in 1990-92. Chow et al. (2000) found, however, that many specimens of bigeye tuna in the Atlantic Ocean shared the same restriction profile with yellowfin tuna; this also occurred in the Pacific Ocean, but at a much lower frequency (1 out of 144 individuals examined). Takeyama et al. Fishery Bulletin 105(1) Table 2 Maximum distances traveled for each collection group of yellowfin tuna larvae recruited to the sampling area based on back- | calculated spawning dates, the period 3f time over which the larvae were exposed to environmental conditions during their life history, and an average, maximum current speed and d irection for the Panama Bight region (Fiedler, 2002). The average lati- ] tudinal and longitudinal degrees traveled were used to estimate an area occupied by ' arvae of all collection groups. The mean sea surface temperature (SST) is based on monthly averages within 1- by 1.5-degree ares for each collection group period within | the estimated area. Monthly mean Time period Number Maximum Maximum First Last SST(SE) exposed to of meters nautical Collection spawn sample and ranges ambient SST days traveled miles group date date (°C) (days) feeding @ .25 m/sec traveled Degrees I 6/6/1990 6/26/1990 27.84(0.090 26.4-28.7 20 17 432,000 233 3.89 II 6/19/1991 7/25/1991 27.90(0.063) 26.6-28.7 36 33 777,600 420 7.00 III 8/14/1991 9/7/1991 27.60(0.059) 26.5-28.4 24 21 518,400 280 4.66 IV 6/9/1992 7/3/1992 28.00(0.066) 26.4-29.0 24 21 518,400 280 4.66 V 7/19/1997 8/7/1997 29.10(0.046) 19 16 410,400 222 3.69 28.3-30.0 mean degrees 4.78 (2001) found another restriction enzyme {Tsp 5091) that was diagnostic for bigeye tuna regardless of where the specimens came from. Therefore, in addition to using Mse I digestion, Tsp 5091 was also used for all individu- als collected in 1997. Back-calculated dates Spawning dates were back-calculated for each larva by subtracting the number of otolith increments counted from the date the larva was collected. An additional day was also subtracted because the first increment in yellowfin tuna is present at hatching approximately 20 hours after fertilization (senior author, personal commun.) and the second increment does not form until the third day after fertilization (approximately two days after hatching); increments are formed daily thereafter (Wexler et al., 2001). During the reduced upwelling season when SSTs are warmer, first feed- ing of the larvae occurs at first light, approximately three days after hatching (Margulies et al., in press) when, on average, three increments are present in the sagittae. Therefore, three days were subtracted from the estimated spawning date to estimate the time period that the larvae of each collection group were feeding until they were collected (Table 2). "Collec- tion-group period" is defined as the time period from the time of first spawning to the time when larvae were sampled. Estimated area occupied by larval cohorts The yellowfin tuna larvae that were collected near the Frailes Islands may be recruited locally from offshore areas; this conjecture is based on measurements of the mean monthly fields of velocity and direction of the North Equatorial Countercurrent (NECC) (up to 0,25 m/s) (Fiedler, 2002), southerly surface winds (up to 5 m/s) (Fiedler, 2002), and the location and proportion of reproductively active female yellowfin tuna (Schaefer, 1998) that are found during June-September in the Panama Bight area (Fig. 2). The earliest back-calcu- lated spawning date for a larva within each collection group was used to estimate the maximum amount of time the larvae within that group were exposed to environmental and feeding conditions (Table 2). This amount of time and the maximum current speed and direction during this season were used to calculate maximum average distances traveled and the potential area occupied by each collection group until sampled at the Frailes Islands (Table 2, Fig, 2). Ichthyoplankton and oceanographic surveys During 1990-92 ichthyoplankton and oceanographic sampling were conducted from a 25-ft Boston whaler along the Morro Puercos (P) and Punta Mala (M) tran- sects (Fig. 1) (lATTCi; IATTC';Lauth and Olson, 1996). Data collected from these surveys were used to describe the temporal variation of conditions within the plank- tonic community that may correspond to that of larval yellowfin tuna growth rates. In 1990, oblique bongo tows were made from the surface to 50 m along both transects with 335-f(m mesh nets (Lauth and Olson, 1996). As a measure of relative abundance, standard- ized plankton volumes under 10 m- of sea surface were calculated by following procedures of Smith and Rich- ardson (1977), and the estimates for each side of the Wexler et al Temporal variation in larval growth of Thunnus albocares In the Panama Bight 15°l bongo were averaged. Beginning in 1991. a 0.6- m- Tucker trawl equipped with a 335-mm mesh net, flow meter, and temperature-depth logger was used to sample ichthyoplankton at discrete depths at only the Punta Mala shelf break (MSB). These surveys were designed to study the vertical distribution and in situ growth and starvation rates of tuna larvae and the abundance of their zooplankton prey (lATTCi; IATTC-). Two replicate tows of 4 to 5 minutes were made at each of three or four depth strata: 0-5 (stratum 1), 5-20 (stra- tum 2), 20-40 (stratum 3), and 40-60 m (stratum 4). Plankton volumes were standardized (Smith and Richardson, 19771 at each depth stratum and were added together for each sampling day to compare the mean plankton volumes collected by the Tucker trawl with those collected by the bongo tows of the previous year. Mean plankton volumes were compared between collection group periods by using a one-way analysis of variance (ANOVA), the Student-Newman-Keuls multiple range com- parison test (SNK test), and a ^test for unequal variance (o=0.05) when appropriate (Zar, 1984). In 1991, all four depth strata were sampled, but in 1992 only the first three strata were sampled. Additionally, a 73-f(m mesh net with a mouth area of 0.014 m- was nested inside the Tucker trawl in 1992 to collect microzooplankton simultaneously with all other plankters. The displaced volume of the microzooplankton was included in the total standardized plankton volume for each sampling day. Water temperatures, surface wind speeds (m/s), and salinity values (psu) were measured (described in Lauth and Olson, 1996) during each sampling day. Plankton displacement volumes for all years were also standardized as plankton volume per volume of water filtered (mL/m^) to compare mean values between years and with literature values. The mean of each standardized volume for the 1990 oblique tows (0-50 m) and for discrete depths between and 40 m of the 1991 and 1992 data were compared between collection group pe- riods by using ANOVA, the SNK test, and a ^test for unequal variance (a=0.05) (Zar, 1984). Sea-surface temperatures and wind stress climatology The oceanographic surveys provided physical data within a limited portion of the area where Thunnus larvae potentially occurred since hatching. Therefore, area- and time-specific (monthly averages within 1- by 1.5-degree areas) SSTs to 5 m depth and wind stress cli- matology data (all data sets based on a hindcast ocean analysis system model described by Ji et al. [1995]) for the estimated area of each collection group period (Table 2, Fig. 2) were accessed from the internet (IRI''). Wind velocities in m/s were calculated from wind stress values based on a constant drag coefficient of 1.3x10"'^ (Sverdrup et al., 1942; Large and Pond, 1981; Ji et al.. Suilace winds August ^ PCif ( l(! i:lU::^p/, /, A ,, y -, . tiiiil. 100"W 95 =W 85°W Surface currents 80°W 95 -W 85'W 80 W Figure 2 Monthly fields of surface wind velocity (for August) and sur- face current velocity (for September) representative of the seasonal extremes during the reduced upwelling period in the Panama Bight (after Figure 3 of Fiedler, 2002). Shading indicates surface wind divergence (intertropical convergence zone) during August, NEC = North Equatorial Current, SEC = South Equatorial Current, NECC = North Equatorial Counter Current, and CRCC = Costa Rica Coastal Current. The area between the vertical and horizontal lines and the land mass represents the estimated maximum average area (in degrees) (see Table 2) potentially occupied by each larval yellowfin tuna cohort during its life history. The spawning distribution of yellowfin tuna within the area is presented as the proportions (P) of reproductively active females in relation to the total numbers of mature females captured within 1-degree areas during the second and third quarters between 1987 and 1989 (from Schaefer. 1998). ^ International Research Institute for Climate Prediction (IRI). 2006. Website: http://ingrid.ldeo.columbia.edu/ SOURCES/.NOAA/.NCEP/.EMC/.CMB/.Pacific/.monthly/ (accessed on 14 October 2005). Fishery Bulletin 105(1) 8 12 Group 1 June 1990 □ Group II July 1991 6 2-19 6 mm SL 10 9.1-12.7 mm SL 6 8 Group III September 1991 7,1-12.4 mm SL 4 6 2^ 1 1 4 1 p 1 0) > III 1 1 1 1 n 1 o 3 10 12 14 16 18 20 8 10 12 14 16 18 20 Number 03 O 22 Group IV June- July 1992 ,„ 7.6-12,0 mm SL 14 Group V August 1997 8.7-14 5 mm SL 6 10 4 1 6 2 1 . 1 1 . 1 II 2 III II 3 10 12 14 16 18 20 B 10 12 14 16 18 20 Age (d Figure 3 Age distributions and standar d length ranges of yellow fin tuna [Thunnus albacares) larvae for each collection group period during 1990-92 and 1997. Larv ae were collected near the Frailes Islands in the Panama Bight. 1995). Physical data were compared between collection group periods using ANOVA and the SNK test. Results Collections and identification The occurrence of fairly large numbers (approximately 100 or more) of larval and early-stage juvenile Thun- nus was sporadic, but not uncommon, in night-light collections during certain months of each year of the reduced upwelling season (Table 1). Based on back- calculated spawning dates and the average surface current speed, the collection site within an 8-degree latitude by 8-degree longitude area (between 2-10°N and 77-85°W) was estimated as the average maximum area potentially occupied by yellowfin tuna larvae of each collection group during their early life history (Table 2, Fig. 2). At the time of collection, sizes of larvae ranged from 6.2 to 19.6 mm SL (Table 1), and all were either in the flexion, postflexion, or transformation stages of development (stages described in Ambrose, 1996). Other scombrid species (\.e.,Auxis sp., Euthynnus lineatus, and Scomberomorous sierra) were also found when Thunnus larvae were collected, but were not usually predominant in the collections. Successful PCR amplification occurred in 80% of the larvae analyzed, and subsequent RFLP analysis in- dicated that the Thunnus larvae collected near the Frailes Islands were T. albacares (Table 1). Size-at-age and growth The ages of yellowfin tuna larvae collected ranged from 8 to 20 days (Table 1, Fig. 3). The age range was mostly limited between 11 and 14 days for the larvae collected in July 1991 (collection group II) and in June- July 1992 (collection group IV); therefore growth models were not fitted to the data (Figs. 3 and 4). However, a comparison of the size-at-age between all five groups within this lim- ited age range indicated that both SLs and otoliths were significantly smaller for larvae collected in September 1991 (collection group III) (ANCOVA and Tukey multiple comparison test, P<0.0001), and that SLs were similar between the larvae of 1990 (collection group I) and July Wexler et al Temporal variation in larval growth of Thunnus albacares in the Panama Bight w June 1990 (n=25) A September 1991 (n=43) V'=2.98e0i06X ?= 3,66 eO 061 'f r2 = 0.86 r2 = 0.59 A July 1 991 (n = 1 3) o June-July 1 992 (n = 22) D August 1997 (n=69) /= 4,44 e 063X r2 = 0.68 Age (d) Figure 4 Exponential relationships between standard length and age in days estimated from otolith increment counts of larval yellowfin tuna tThunnus albacares) collected during June 1990, September 1991, and August 1997. Data are also presented for yellowfin tuna larvae collected during July 1991 and June-July 1992, but growth models were not fitted to the limited range of data. 1991 (collection group II) and between those of 1992 and 1997 (collection groups IV and V). The length-at-age data were used to examine and compare growth relationships of all yellowfin tuna lar- vae collected in June 1990, September 1991, and August 1997 (collection groups I, III, and V, respectively). An exponential model provided the best fit to each of the three groups of data (Fig. 4). The variances were homo- geneous after log transformation of the length data for each group, and the slopes were compared. The slope and the average growth rate obtained through differ- entiation of the exponential equation of the 1990 data (1.28 mm/d, SE = 0.134) were significantly greater than those of the September 1991 (0.60 mm/d, SE =0.033) and 1997 (0.71 mm/d, SE = 0.038) data (ANCOVA, P<0.0001, Tukey multiple comparison test). The elevations (i.e., adjusted means or intercepts) of the 1991 and 1997 data were significantly different (P<0.0001) and indi- cated that the mean length-at-age was significantly smaller for larvae of the September 1991 collection group (Fig. 4). Similar results were obtained for otolith growth rates (based on the exponential relationships between otolith diameter and age) for the three years (Fig. 5). The slopes were compared after log transformation of the otolith data for each group. The growth rate of the 1990 data was significantly faster than that of the 1991 and 1997 data (ANCOVA, P<0.0001, Tukey multiple com- parison test), and the elevations were different between the 1991 and 1997 data, indicating that otoliths were significantly smaller for larvae of the September 1991 collection group (Fig. 5). A comparison of the linear relationships between oto- lith diameter and SL (ANCOVA, P<0.0001, Tukey mul- tiple comparison test; Fig. 6) revealed that the otoliths of the 1997 group were larger and grew significantly faster in relation to fish size than those of the 1990 and 1991 groups. Otoliths of the fastest (1990 group) and slowest (1991 group) growth periods were grow- ing at the same rate in relation to fish length, but the slower-growing group had significantly larger otoliths in relation to fish size than those of the faster-growing group (P<0.0001; Fig. 6). Standing stocks of ichthyoplankton Ichthyoplankton and physical parameters were mea- sured at four stations along the P and M transects on two sampling days in June 1990 (16 tows total, each to 50 m), at the MSB station on the M transect on six sampling days in June and July 1991 (12 tows each depth strata 1-4), at the MSB station on three sampling days in August 1991 (six tows each strata 1-4 and four tows each strata 1-3), and at the MSB station on two sam- pling days in June and July 1992 (six tows each strata 1-3) (Table 3, Fig. 1). The sampling days lay within the Fishery Bulletin 105(1) period of time when each collection group of larvae could have been feeding in the estimated area of occurrence (Table 2, Fig. 2). Ichthyoplankton tows were not made in the Panama Bight during 1997. Mean standardized plankton volumes were variable and significantly different (ANOVA, P<0.001) among collection-group periods (Fig. 7). The mean plankton volume (±SE) in 1990 (157.3 ±13.53 mL) when the fast- est larval growth rate occurred was greater and differ- ent from all other sampling or collection periods (SNK test), and ranged from 106.5 to 310.4 mL under 10 m- of sea surface (Table 3, Fig. 7). The mean plankton volumes during June-July 1991 (82.3 ±3.46 mL) and during August-September 1991 (62.8 ±5.86 mL), when the slowest growth rate occurred, were similar and less than those for all other periods (SNK test); volumes ranged from 43.7 to 102.4 mL under 10 m- of sea sur- face (Table 3, Fig. 7). Mean plankton volumes, expressed as the amount fil- tered per volume of water sampled within the first three depth strata, were also significantly different (ANOVA, P<0.001) among collection group periods. Means were similar between the 1990 and 1992 groups and be- tween the two 1991 groups (SNK test). Volumes ranged from 0.199 to 0.559 mL/m^ during 1990 and 1992 and from 0.075 to 0.235 mL/m'^ during the two 1991 periods (Table 3). Plankton volumes included relatively large numbers of fish larvae (predominantly preflexion stages) during the least (August-September 1991) and most (June 1990) o O 700 - r ♦June 1990 (n=25) /= 50.4 e ° 138X r2 = 0,87 A September 1991 (n=43) Y= 54 6 eoossx r2 = 70 D August 1997 (n=69) V'= 89,1 e0089X r2 = 0,60 600- A July 1991 (n=13) June-July 1992 (n = 22) / A 500 400 300- D 4 i / D A 200- 100- ^^^ A n 1 1 1 1 — 1 1 , 1 • 1 1— . \ 1 rapid growth periods, and mean values were not signifi- cantly different (^test for unequal variances, P>0.20). Numbers of larvae under 10 m^ of sea surface ranged from 686.8 to 4786.1 and from 934.5 to 2685.6 in 1990 and 1991, respectively. Few scombrid larvae occurred in the ichthyoplankton samples for each of the two years, but were greatest during the 1991 period. The num- ber of scombrid larvae under 10 m'^ ranged from to 2.7 and from to 12.7 in 1990 and 1991, respectively. Thunnus larvae (preflexion stage) were collected only in August 1991, and the numbers ranged from 0.5 to 5.5 larvae under 10 m- of sea surface. Environmental effects Mean SSTs were significantly different among all col- lection group periods (ANOVA, P<0.0001). SSTs were similar between the local sampling area (Fig. 1) and the estimated region of each collection group (Fig. 2) in that they were significantly lower for the August-September 1991 period (group III) and higher for the July-August 1997 period (group V) (ANOVA, P<0.0001, SNK test; Table 2, Fig. 7). The mean wind stress was significantly lower for the 1997 collection-group period (group V) when com- pared with all other group periods (ANOVA, P<0.0001, SNK test; Fig. 7). The monthly means within each 1- by 1.5-degree area were similar and ranged from 0.084 to 0.517 dynes/cm- for group-collection peri- ods I-IV, and for the 1997 period (group V), they ranged from 0.027 to 0.462 dynes/cm-. Wind velocities cal- culated from the wind stress values were low to moderate, ranging from 1.59 to 3.94 m/s (modes of 2.45 and 2.60 m/s) for groups I-IV and from 0.90 to 3.72 m/s (modes of 1.2 and 1.7 m/s) for group V. Higher salinity values, rang- ing from 33 to 34 psu, occurred during July-August 1997 (group V) during an ENSO event, and during the 1990-92 collection-group periods (I-IV) values ranged from 29 to 32 psu, when both the fastest and slowest growth rates of yellow- fin tuna larvae occurred. 6 7 9 11 13 15 17 19 21 Age (d) Figure 5 Exponential relationships between sagittal otolith diameter and estimated age in days oflarval yellowfin tuna {Thunnus albacares) collected during June 1990, September 1991, and August 1997. Data are also presented for yellowfin tuna larvae collected during July 1991 and June-July 1992, but growth models were not fitted to the limited range of data. Discussion This study describes the first in situ growth rates for yel- lowfin tuna larvae occurring in the Pacific Ocean. Previ- ous efforts to age and describe growth of yellowfin tuna during the early stages of develop- Wexler et al : Temporal variation in larval growth of Thunnus albocares in the Panama Bight E a T3 600^ r ♦June 1990 (n=25) V'=32.27X- 75.37 r2 = 0.94 A September 1991 (n=43) y'=31.54X-44.03 r2 = 0.65 □ August 1997 (n=69) / = 41. 10X- 129.67 r2 = 0.89 > 500 ^ ^^ 400 300- ^e^^^^^^*^^ 200 100 /^^^ A ^ 1 1 A* —\ i 1 i h i 1 y 1 Standard length (mm) Figure 6 Linear relationships between otolith diameter and standard length for larval yellowfin tuna (Thunnus albacares) collected during June 1990, September 1991. and August 1997. Plankton volume I...... I SST for local area li^^Siy^ SST for estimated area Wind stress E i- E a ~ to OJ »- p » June 1990 June -July 1991 August -Sept, 1991 June • July 1992 July -August 1997 Collection group period Figure 7 Standardized mean plankton volumes and monthly mean wind stress and sea surface temperature data for the local sampling area (Fig. 1) and the estimated area iFig. 2) potentially occupied by each larval yellowfin tuna (Thunnus albacares) cohort during its life history. The standard errors of the means are indicated for plankton and wind stress, and the ranges for the mean water temperatures are also indicated. Plankton data were collected from ichthyoplankton tows in the vicinity of the Frailes Islands (see Fig. 1) and area- and time-specific physical data were extracted from the internet iIRPl. 10 Fishery Bulletin 105(1) Table 3 Standardized plankton displacement volumes Morro Puercos (P) transects during 1990-1992 collected from ichthyoplankton surveys conducted along the Punta Mala (M) and in the Panama Bight. Station definitions are described in Figure 1. Sampling date Transect Station Tow depth (m) mL plankton volume/lOm^ of sea surface mL/m^ Sum of Sum of depth strata depth strata 1-4 1-3 19 June 1990 M MAB 0-50 121.3 0.219 MAB 0-50 242.3 0.425 MSL 0-50 123.9 0.226 MSL 0-50 118.5 0.208 MSB 0-50 143.0 0.260 MSB 0-50 156.0 0.280 MSH 0-50 167.6 0.328 MSH 0-50 310.4 0.559 20 June 1990 P PAB 0-50 140.9 0.252 PAB 0-50 129.1 0.264 PSL 0-50 205.8 0.393 PSL 0-50 163.6 0.301 PSB 0-50 113.2 0.199 PSB 0-50 125.6 0.240 20 June 1990 P PSH 0-40 106.5 0.245 PSH 0-40 149.5 0.333 17 June 1991 M MSB 0-5.7 6.6 0.117 5.3-25.8 21.5 0.105 21.2-27.0 39.6 0.147 43.1-48.1 2.4 0.049 70.1 67.7 MSB 0-5.4 8.4 0.155 5.2-21.9 20.9 0.125 21.2-48.2 38.2 0.142 42.5-67.2 10.2 0.041 77.7 67.5 12 July 1991 M MSB 0-5.7 13.0 0.229 5.4-21.9 31.5 0.190 22.9-44.5 23.9 0.111 43.8-67.2 10.5 0.045 96.9 68.4 MSB 0-6.1 14.4 0.235 4.9-23.7 32.8 0.175 22.9-48.2 32.0 0.127 43.8-67.2 10.5 0.045 89.6 79.1 16 July 1991 M MSB 0-5.7 9.9 0.176 4.9-21.9 19.9 0.117 21.2-44.5 24.7 0.106 40.2-67.2 12.4 0.046 67.0 54.5 MSB 0-5.7 4.6 0.081 5.3-23.7 19.7 0.107 21.2-44.5 25.9 0.111 40.2-67.2 12.5 0.046 62.8 50.3 19 July 1991 M MSB 0-5.4 6.2 0.114 5.2-21.9 24.4 0.146 20.2-44.5 36.4 0.150 40.2-64.0 15.5 0.065 82.5 67.0 MSB 0-5.4 5.2 0.096 5.3-20.9 31.6 0.203 21.2-46.0 31.3 0.126 43.1-70.1 13.2 0.049 81.3 68.0 conliinied Wexler et al Temporal variation in larval growth of Thunnus albacares in the Panama Bigtil Table 3 (continued) Sampling date Transect Station Tow depth ( m I mL plankton volumelOm- of sea surface mL/m-' Sum of depth strata 1-4 Sum of depth strata 1-3 23 July 1991 M MSB 0-5.4 5.5-20.9 21.2-46.5 79 21.0 40.2 0.147 0.137 0.159 43.1-61.8 11.4 0.061 80.6 69.1 MSB 0-5.7 5.5-22.9 21.2-46.5 97 18.5 40.8 0.171 0.116 0.162 41.8-67.2 15.0 0.059 83.9 68.9 25 July 1991 M MSB 0-5.7 5.3-25.8 22.1-46.5 9.2 30.9 44.9 0.163 0.150 0.184 43.1-70.1 17.4 0.065 102.4 84.9 MSB 0-5.4 5.1-20.9 21.2-38.9 9.4 23.3 34.5 0.174 0.147 0.195 43.1-62.3 25.9 0.135 93,2 67.2 15 August 1991 .\1 MSB 0-5.7 5.3-19.2 5.7 15.4 0.101 0.111 21.2-44.5 32.0 0.137 53.1 15. August 1991 M MSB 0-6.7 5.3-21.9 21.2-44.5 8.6 24.3 29.5 0.129 0.146 0.126 36.0-56.5 15.6 0.076 78.0 62.4 MSB 0-5.7 5.3-21.9 21.2-40.2 93 18.8 22.5 0.165 0.113 0.119 43.8-67.2 15.3 0.066 66.0 50.6 MSB 0-5.7 5.7-21.9 8.6 20.2 0.151 0.125 22.9-43.8 27.7 0.133 56.5 MSB 0-6.1 5.3-21.9 8.4 0.138 0.134 21.2-44.5 36.0 0.154 66.6 MSB 0-5.7 6.1-19.2 12.3 18.5 0.217 0.142 21.2-48.2 53.6 0.199 84.3 21 .August 1991 M MSB 0-5.7 5.3-21.9 21.2-44.5 6.5 12.8 19.2 0.115 0.077 0.082 45.0-672 79 0.035 46.3 38.4 MSB 0-6.4 5.5-21.9 21.2-44.5 5.4 12.3 18.5 0.084 0.075 0.079 43.8-67.2 7.5 0.032 43.7 36.2 23 .A.ugust 1991 M MSB 0-5.7 5.3-21.9 21.2-44.5 8.1 19.7 32.6 0.143 0.119 0.140 continued 12 Fishery Bulletin 105(1) Table 3 (continued) Sampling date Transect Station Tow depth (m) mL plankton volume/lOm^ of sea surface mL/m^ Sum of depth strata 1-4 Sum of depth strata 1-3 23 August 1991 (continued) 45.9-67.2 9.5 0.044 69.9 60.4 MSB 0-5.7 5.3-23.7 21.2-44.5 7.3 23.4 31.2 0.128 0.127 0.134 43.1-67.2 10.9 0.046 72.9 61.9 22 June 1992 M MSB 0-5.0 5.4-20.2 14.6 34.6 0.294 0.234 20.2-40.0 60.0 0.303 109.2 MSB 0-5.0 5.0-20.2 11.5 39.3 0.233 0.259 20.2-40.0 52.4 0.264 103.3 MSB 0-5.0 5.0-20.2 16.9 56.3 0.342 0.370 20.2-40.0 70.9 0.358 144.1 MSB 0-5.0 5.0-20.2 23.4 54.5 0.473 0.359 20.2-40.0 54.2 0.274 132.1 2 July 1992 M MSB 0-5.0 4.5-20.2 19.7 48.9 0.397 0.312 20.2-40.0 49.8 0.252 118.4 MSB 0-5.0 5.4-18.7 10.7 37.5 0.216 0.280 20.2-40.0 65.9 0.333 114.0 ment may have been precluded by the patchiness in their distribution and the difficulties in species iden- tification. The only other growth study done on yel- lowfin tuna larvae was conducted in the Gulf of Mexico (Lang et al., 1994), but the species identifications were based on morphology and meristics, and the larvae were younger than those in our study. Results from our mtDNA analysis enabled us to examine species- specific growth rates of older larval stages of yellowfin tuna and associated factors affecting their growth and distribution. Distribution Yellowfin tuna larvae have consistently appeared in the night-light collections near the Frailes Islands during the reduced upwelling season, but not during the season when strong upwelling occurs and other spe- cies of scombrid larvae and plankton levels are more abundant (Smayda, 1966; Forsbergh, 1969; Lauth and Olson, 1996). The absence of yellowfin tuna larvae from our sampling area during the upwelling season may be associated with a cessation of spawning by yellowfin tuna during this period (Schaefer 1998, 2001; Margulies et al., in press) and with the temperature threshold of their larvae. Lower mean water temperatures typically occur during the upwelling season (Lauth and Olson, 1996) and have ranged from 17.3° to 25.8°C within the upper 50 m (Owen'^). In the laboratory, survival of first-feeding yellowfin tuna larvae is poor at ambient water tempera- tures of <21°C and at dissolved oxygen levels <2.2 mg/L (<33.0 % of oxygen saturation) (Margulies et al."^). These temperature and dissolved oxygen requirements probably determine and limit the distribution of yellowfin tuna larvae within the mixed layer and determine whether or not they can survive during the upwelling season when water temperatures are lower. The distribution of yellowfin tuna larvae during the upwelling season may also be strongly infiuenced by the occurrence of strong westerly directed currents and northerly winds resulting in larval transport away from the coastal areas of the Panama Bight during this season. The area of larval distribution since hatching may ac- tually be smaller or larger than what we have estimat- ed, depending on the amount of passive transport and ' Margulies, D., V. P. Scholey, J. B. Wexler, R. J. Olson, J. M. Suter, and S. Hunt. In press. A review of lATTC research on the early life history and reproductive biology of scom- brids conducted at the Achotines Laboratory from 1985 to 2005. Inter-American Tropical Tuna Commission, Special Report 16. lATTC, 8604 La Jolla Shores Drive, La Jolla, CA 92037. Wexler et al,: Temporal variation in larval growth of Thunnus albacares in the Panama Bight 13 the swimming behavior of the larvae within the mixed layer. Passive transport would probably occur only dur- ing the egg, yolk-sac, and first-feeding stages (the first 8-10 days after fertilization) because yellowfin tuna larvae are competent swimmers and can hold their posi- tion against strong currents in the laboratory beginning at around 8-10 mm SL (D. Margulies, personal com- mun.). Although the maximum average area of larval yellowfin tuna distribution from the time of hatching is probably our best estimate, the physical and biological processes that occur in such a large area may not be representative of processes occurring on much smaller scales that may be more specific to conditions affecting larval transport, growth, and survival (Owen, 1989). Prey abundance Our ichthyoplankton data were collected within a 0.5- degree area that included the Frailes Islands where larvae were sampled and may provide an index of prey abundance (at least for the first one or two weeks of feeding until piscivory occurs). Although our data were spatially limited, the measured plankton volumes pro- vide the only available estimates of zooplankton levels for the periods of interest. The use of different gear types (i.e., the bongo and Tucker trawl) for ichthyoplankton collections during 1990-92 may have affected the amount of microzoo- plankton sampled during the different years. Microzoo- plankton abundance has not been compared between these two types of sampling nets. However, Shima and Bailey (1994) reported that the bongo and 1-m Tucker nets caught similar numbers and size distribution of larval walleye pollock (Theragra chalcogramma). The higher plankton volumes collected by the bongo in 1990 may be an underestimate of plankton abundance com- pared to the volumes of water being sampled by the Tucker trawl with a larger mouth opening (McGowan and Fraundorf, 1966). Given that plankton volumes were probably under-represented in the bongo tows, the difference in magnitude between the amounts of plank- ton sampled by each net type may actually be greater. Another potential bias in comparing plankton vol- umes among the different years was that more areas (stations) were sampled with the bongo in 1990 than with the Tucker trawl in other years (MSB station on- ly). However, the mean plankton volume would have been similar in 1990 if only the MSB station had been used in the analysis (Table 3j. Growth Daily growth rates estimated from the exponential models for each of the three years (1990, 1991, and 1997) ranged from 0.46 to 2.06 mm/d and were generally greater than those reported for other congeners (Jen- kins and Davis, 1990; Lang et al, 1994). However, the larvae represented in those studies were predominantly younger and in earlier stages of development (and thus would exhibit slower absolute growth) than the flexion and postflexion larvae and transitional juvenile stages of yellowfin tuna collected in our sampling area. The slower growth rates observed in southern bluefin tuna (Thun- nus maccoyii) larvae were also associated with density- dependent and oligotrophic conditions in the East Indian Ocean (Rochford, 1962; Jenkins and Davis, 1990; Young and Davis, 1990). Our growth rates, however, were comparable to similar developmental stages of other scombrids that inhabit relatively similar, productive nearshore waters, such as king and Spanish mackerels {Scomberomorus cavalla and S. maculates, respectively; DeVries et al., 1990), black skipjack (Euthynnus lineatus; Wexler, 1993), and little tunny {Euthynnus alletteratus; Allman and Grimes, 1998). Distinct differences in the average size-at-age and growth rates were very apparent between our 1990 and September 1991 collections of yellowfin tuna larvae. Size-dependent processes (i.e., predation and starva- tion; Pepin, 1988; Grimes and Isely, 1996) or density- dependent growth and survival (Jenkins et al., 1991) may affect the size-frequency distributions of surviving larvae. A simulation model (Pepin, 1988) demonstrated that with increased food abundance, the mean and vari- ance in larval growth rates increases, but, as predator abundance increases, the variance in growth rates de- creases for any given mean. Instantaneous growth rates for yellowfin tuna larvae of a similar age in 1990 were 2 to 3 times higher than those in 1991, and plankton volumes were 2 to 7 times higher than those in 1991. Increases in food availability, such as that during 1990, may also attract predators and result in greater rates of mortality of the slowest-growing individuals, so that they are not represented in the sampled population. Although the larvae in 1991 were growing more slowly than those in 1990, they probably do not represent the slowest-growing larvae of their cohort. Typically, postflexion larval and early-stage juvenile scombrids collected during the reduced upwelling season in our sampling area have exhibited more variable growth (Wexler, 1993), but have been predominantly healthy (Margulies, 1993). Therefore, slower or faster growing survivors at this stage may be independent of their nutritional condition, and larvae collected by the sam- pling method we used represent the survivors and most competent individuals of their cohort. Growth may have been slower in 1991 because of higher larval densities, limited food availability, and available prey composition. A strong inverse relation- ship exists between growth rates and stocking densities of yellowfin tuna larvae and early-stage juveniles (up to 18 days after hatching) fed a constant food supply in the laboratory (lATTC, lATTC'; Margulies et al.*^). The re- '' lATTC (Inter-American Tropical Tuna Commission). 2000. Annual report of the Inter-American Tropical Tuna Commission 1998, 357 p. lATTC, 8604 La Jolla Shores Drive, La Jolla, CA 92037. " lATTC (Inter-American Tropical Tuna Commission). 2002. Annual report of the Inter-American Tropical Tuna Commission 2001, 148 p. lATTC, 8604 La Jolla Shores Drive, La Jolla, CA 92037. 14 Fishery Bulletin 105(1) lationship may be even more pronounced under limited food conditions (Jenkins et al., 1991). Although we have only indirect evidence for density-dependent growth of the 1991 cohorts, the occurrence of more yellowfin tuna larvae sampled at the surface with the night light and the large numbers of other fish larvae collected in the ichthyoplankton tows coincident with the lower plank- ton volumes may indicate growth-limited conditions during this period. The slower growth of the late-stage larvae in 1991, when plankton abundance was much lower, may also be indicative of the types or species of preferred prey (zooplankters and fish larvae) that were available in the area that the larvae occupied. In the laboratory, yellowfin tuna larvae have predominantly selected all stages of cyclopoids over other types of co- pepods when offered a mixed assemblage of zooplankton prey (Margulies et al., 2001) and have become piscivo- rous beginning at approximately 6-7 mm in SL (Margu- lies et al.''). During this transitional stage in their diet, growth becomes much more rapid and variable (Kaji et al., 1999; Margulies et al.^), and the availability of specific types of fish larvae may influence their ability to switch to piscivory in the ocean. Yellowfin tuna lar- vae readily consume other, smaller conspecifics in the laboratory, but it is not known if there is a preference or growth advantage for consuming certain species of fish larvae during the transition to a more piscivorous diet. Although the available prey composition could affect the growth of late-stage yellowfin tuna larvae, intra- and interspecific competition for limited food resources dur- ing the 1991 period may have been the principal cause of slower growth. The temporal variation in size-at-age within the same season and year (1991) may be related to the physical and biological characteristics of the area occupied by each group of larvae since hatching. Larval distribu- tion could be determined by the location and timing of yellowfin tuna spawning and the small- and large- scale dynamics of physical oceanographic processes. Average sizes of the larvae and their otoliths of the 1991 August-September group were distinctly and sig- nificantly smaller than those of all other groups. The back-calculated first-feeding dates of this group coin- cided with the lowest plankton volumes measured in our local sampling area and with the only collection of first-feeding yellowfin tuna larvae from our ichthyo- plankton tows. In contrast, the mean sizes of larvae and otoliths of the 1991 July group were similar to those of the fastest-growing group of 1990, despite low plankton volumes similar to those of the September 1991 period. We believe that the September 1991 group may have been spawned nearer to our local sampling area (Fig. 2) and that they were more exposed to feeding conditions in the vicinity of the Frailes Islands than were the faster-growing larvae of the July 1991 group. Physical effects on growth The probability of feeding success in marine larvae, and subsequent growth rates and survival, may increase with moderate levels of wind-induced microscale tur- bulence in their feeding environment (Rothschild and Osborn, 1988; Cury and Roy, 1989; Ware and Thomson, 1991; MacKenzie et al., 1994; lATTC*, lATTC^). Pre- liminary estimates of wind speeds that produce optimal turbulent velocities and maximum survival of first-feed- ing yellowfin tuna larvae in the laboratory are moderate to high (D. Margulies, personal commun.) compared to wind speeds measured in the Panama Bight during this study. This estimate of optimal wind speeds is based on the assumption that maximum abundances of yellowfin tuna larvae in the EPO occur at depths of to 20 m. However, the data on wind stress and velocities in the estimated area of larval distribution may not represent the frequency of optimal wind speeds associated with areas where first feeding of each cohort occurred. Addi- tionally, wind event durations and frequencies, for which data were not available, may also play a significant role in the optimal survival of marine larvae (Wroblewski et al., 1989). Although moderate to high wind-induced turbulence may enhance early larval survival, growth rates may actually be slower during a portion of the larval phase as a result of higher larval densities and increased competition for limited resources. Temperature-limited or -enhanced growth was not clear from our analyses of the yellowfin tuna larvae collected. Although a parabolic relationship between growth rates and SSTs was evident, it is unlikely that the two slowest-growing groups represented by the up- per (29.1°C; group V) and lower (27.6°C; group III) mean temperatures in our study (Table 2) are approach- ing thermal tolerance limits for yellowfin tuna larvae. In the laboratory, successful hatching of yellowfin tuna larvae still occurs at upper temperatures of 32-34°C and first-feeding larvae are able to survive and feed be- tween temperatures of 21° and 32°C (Margulies et al.^^). Lower temperatures during the August-September 1991 (group III) period (Table 2) may have resulted in slower growth than that of the other collection periods, but we have only found significantly slower growth rates and differences in the mean sizes at first feeding when mean temperatures were less than 27°C in the laboratory (senior author, personal commun.). In con- trast to the most rapid growth rate in 1990, the larvae in 1997 were growing more slowly when a strong ENSO event and the highest SSTs occurred. The optimum temperature range for growth of yellowfin tuna larvae in the Gulf of Mexico was 29-29.5°C (Lang et al., 1994), which was a similar temperature range for larvae of the 1997 period in our study. We do not, however, have information on relative food abundances during this * lATTC (Inter-American Tropical Tuna Commission). 2001. Annual report of the Inter-American Tropical Tuna Commission 1999, 183 p. lATTC, 8604 La Jolla Shores Drive, La Jolla, CA 92037. ^ lATTC (Inter-American Tropical Tuna Commission). 2002. Annual report of the Inter-American Tropical Tuna Commission 2000, 171 p. lATTC, 8604 La Jolla Shores Drive, La Jolla, CA 92037. Wexler et al.: Temporal variation m larval growth of Thunnus albacores in the Panama Bight 15 time. Interactive effects of food availability and tem- perature may have a profound effect on growth, and we would expect that energetic demands become greater at the higher water temperatures, in which case the potential for faster growth would be attained through increased food consumption (Houde, 1989), providing food resources are not limited. Thus, it may be more reasonable to assume that food availability has a more significant impact on growth than SSTs during the reduced upwelling season when temperatures are con- sistently greater than 27°C. Although water temperature has been shown to regu- late the formation and short-term growth of otoliths in some marine fish species (Barber and Jenkins, 2001), the causal factor affecting otolith growth could also be associated with food availability (Govoni et al., 1985; Johnson et al., 2002) and composition (Woodbury, 1999). Otolith growth appears to be a conservative measure of somatic growth (i.e., otoliths continue to grow even with decreases in somatic growth; Campana and Neilson, 1985); therefore a significant change in otolith growth could signal a dramatic change in the larval feeding environment. We observed significant interannual differences between relationships of otolith size and fish size. Otoliths were disproportionately larger in slower growing groups of larvae, but addi- tionally, during the period of 1997 when SSTs were abnormally high, otoliths were growing at a greater rate in relation to fish size (Fig. 6). Elevated water temperatures have been shown to increase short-term otolith growth (Hoff and Fuiman, 1993; Barber and Jenkins, 2001). Althoughtemperature may have af- fected otolith growth in 1997, lower food levels may have been the causal factor for a significantly slower otolith growth rate and the smaller mean otolith di- ameter of the 1991 group. Recruitment implications The probability of survival from early stages of devel- opment to recruitment in marine fishes is thought to be influenced by prerecruit starvation and predation mortality associated with slower growing and nutrition- ally weakened individuals (Gushing, 1975; Houde, 1987; Margulies, 2001) and predator-prey interactions and densities (Cowan and Houde, 1992). However, Peterman et al. (1988) have demonstrated that the survival rate of prerecruits older than 19 days of age is more vari- able than earlier life stages, and it is this stage that determines recruitment strength in northern anchovy (EngrauUs mordax). Yellowfin tuna in the Pacific Ocean exhibit a pattern of reproduction that has strong poten- tial for regulation of recruitment during prejuvenile stages, when initial numbers in a cohort are quite large and vital rates (e.g., growth, mortality) are high (Houde, 1987; Margulies, 2001). However, the poten- tial for recruitment fluctuations is also high for the relatively long juvenile stage of yellowfin tuna as well (Houde, 1987; Margulies, 2001). In the EPO, yellowfin tuna are recruited to the fishery at a fork length of about 30 cm and at an age of approximately 6 months (Wild, 1994; Maunder and Harley^°). Recruitment estimates calculated at quarterly intervals for yellowfin tuna in the EPO are variable and may be influenced by envi- ronmental fluctuations (Maunder and Harley^"). It is not clear what effect slower or faster growing cohorts may have on recruitment, but the growth- and stage-specific mortality rates (Houde, 1987; Pepin, 1991; Comyns et al., 2003) of a cohort may determine whether it survives to recruitment or when it enters the fishery. We were unable to estimate mortality rates for the larvae col- lected in each of the three years because during some years they were collected on single sampling dates (Essig and Cole, 1986). Nonetheless, the recruitment estimate (ca. 1.44x10^ individuals) following the period in 1991, when the smallest size-at-age and the slowest- growing larvae were present was approximately half the amount estimated (ca. 3.11x0' individuals) following the period in 1990 (Maunder and Harley^"; Maunder"), when larvae were larger and growing more rapidly. The recruitment estimate following the 1997 period (ca. 3.06x10' individuals) was slightly less than that fol- lowing the 1990 period. Larvae of the 1997 group were growing at a rate similar to that of the 1991 group, but the mean size-at-age was significantly greater, which may indicate a size advantage favorable for prerecruit survival (Miller et al., 1988). The growth rates and conditions estimated for yellowfin tuna larvae within the small scale area of our study may apply to local- ized recruitment estimates within the Panama Bight and not those of the entire EPO, given that restricted movements of yellowfin tuna occur in the EPO (Schaefer, 1991; Wild, 1994). Although these inferences may be applicable only to recruitment within the Panama Bight, they may still indicate that growth rates and the mean size-at-age during the larval and early-juvenile stages are a contributing factor to recruitment variability of yellowfin tuna in the EPO. Our study provides the first examination of factors affecting larval growth and possibly prerecruit sur- vival of yellowfin tuna in the Pacific Ocean. Future research is necessary to better understand small-scale variability in growth and mortality rates of yellowfin tuna larvae within their feeding environment. To that end, we are conducting further studies at the Achotines Laboratory in Panama to examine vital rates of this species and the interactions of these vital rates with biological and physical processes to complement our field measurements and to gain more insight into pre- recruit survival of yellowfin tuna. 1" Maunder, M. N., and S. J. Harley. 2004. Status of yellowfin tuna in the eastern Pacific Ocean in 2002 and outlook for 2003, p. 5-119. Inter-American Tropical Tuna Commission, stock assessment report 4. lATTC, 8604 La Jolla Shores Drive, La Jolla, CA 92037. 1' Maunder, M. 2004. Personal commun. Inter-American Tropical Tuna Commission, 8604 La Jolla Shores Drive, La Jolla, CA 92037. 16 Fishery Bulletin 105(1) Acknowledgments We are grateful to the entire staff of the lATTC's Achotines Laboratory who made this study possible. We especially thank R. Lauth, D. Soliz, M. Samaniego, A. Cano, J. Stadler, and M. Vergara, the principal crew members of the Achotines III during frequently hazardous field sampling. We also thank I. Diaz, D. Dominguez, K. Herrera, and A. Garcia for processing and sorting ichthyoplankton samples. T. Foreman, R. Owen, E. Espinoza, D. Ballesteros, Y. Ballesteros, and P. Vergara also assisted in the field sampling. Logistic and maintenance support was provided by J. Budria, V. Scholey, and E. Espinoza. T. Foreman, R. Olson, R. Lauth, and R. Owen were principally involved in the development of the ichthyoplankton and oceanographic sampling program, and the raw data were processed by R. Lauth, J. Stadler, R. Owen, and R. Olson. We also thank M. Samaniego, E. Espinoza, D. Mancilla, S. Thorrold, R. Jope, C. Vergara, D. Soliz, P. Vergara, B. Ballesteros, Y. Ballesteros, D. Ballesteros, and A. Cano, who assisted with or conducted night-light and light-trap operations. Helpful discussions were appreciated and provided by D. Ambrose, R. Deriso, P. Fiedler, S. Harley, R. Olson, R. Owen, P. Smith, and P. Tomlinson. M. Maunder provided the quarterly recruitment estimates. C. Patnode helped with final preparations of the figures. W. Bayliff, S. Harley, R. Owen, K. Schaefer, and two anonymous reviewers pro- vided helpful review comments. Finally, we thank the former and present directors, J. Joseph and R. Allen, and the chief scientist of the Tuna-Billfish Program, R. Deriso, of the lATTC for their continued support of this research. Literature cited Ahlstrom, E. H. 1971. Kinds and abundance offish larvae in the eastern tropical Pacific, based on collections made on EAST- ROPAC I. Fish. Bull. 69:3-77. Allman, R. J., and C. B. Grimes. 1998. Growth and mortality of little tunny (Euthynnus alletteratus) larvae off the Mississippi River plume and Panama City, Florida. Bull. Mar. Sci. 62:189-197. Ambrose, D. A. 1996. Scombridae: Mackerels and tunas. In Guide to the early stages of fishes in the California Current region (H. G. Moser, ed.), p. 1270-1285. Calif Coop. Ocean. Fish. Inves., CALCOFI Atlas 33. Barber, M. C, and G. P. Jenkins. 2001. Differential effects of food and temperature lead to decoupling of short-term otolith and somatic growth rates in juvenile King George whiting. J. Fish Biol. 58:1320-1330. Barber, R. T., and F. P. Chavez. 1986. Ocean variability in relation to living resources during the 1982-83 El Nino. Nature 319:279-285. Barber, R. T., M. P. Sanderson, S. T. Lindley, ¥. Chai, J. Newton, C. C. Trees, D. G. Foley, and F. P. Chavez. 1996. Primary productivity and its regulation in the equatorial Pacific during and following the 1991-1992 El Nino. Deep-Sea Res. II 43:933-969. Blackburn, M., R. M. Laurs, R. W. Owen, and B. Zietzschel. 1970. Seasonal and areal changes in standing stocks of phytoplankton, zooplankton and micronekton in the eastern tropical Pacific. Mar. Biol. 7:14-31. Campana, S. E., and J. D. Neilson. 1985. Microstructure of fish otoliths. Can. J. Fish. Aquat. Sci. 42:101-1032. Chavez, F. P., P. G. Strutton, G. E. Friederich, R. A. Feely, G. C. Feldman, D. G. Foley, and M. J. McPhaden. 1999. Biological and chemical response of the equato- rial Pacific Ocean to the 1997-98 El Nino. Science 286:2126-2131. Chow, S. and S. Inoue. 1993. Intra- and interspecific restriction fragment length polymorphism in mitochondrial genes of Thun- nus tuna species. Bull. Nat. Res. Inst. Far Seas Fish. 30:207-224. Chow, S., H. Okamoto, N. Miyabe, K. Hiramatsu, and N. Barut. 2000. Genetic divergence between Atlantic and Indo- Pacific stocks of bigeye tuna {Thiinruis obesus) and admi.x- ture around South Africa. Mol. Ecol. 9:221-227. Chow, S., K. Nohara, T. Tanabe, T. Itoh, S. Tsuji, Y. Nishikawa, S. Ueyanagi, and K. Uchikawa. 2003. Genetic and morphological identification of larval and small juvenile tunas (Pisces: Scombridae) caught by a mid-water trawl in the western Pacific. Bull. Fish. Res. Agency 8:1-14. Collette, B. B., and C. E. Nauen. 1983. FAO species catalogue. Vol. 2, Scombrids of the world. An annotated and illustrated catalogue of tunas, mackerels, bonitos, and related species known to date. FAO Fish. Synop. 125, 137 p. FAO, Rome. Comyns, B. H., R. F. Shaw, and J. Lyczkowski-Schultz. 2003. Small-scale spatial and temporal variability in growth and mortality offish larvae in the subtropical northcentral Gulf of Mexico: implications for assessing recruitment success. Fish. Bull. 101:10-21. Cowan, J. H., Jr., and E. D. Houde. 1992. Size-dependent predation on marine fish larvae by ctenophores, scyphomedusae and planktivorous fish. Fish. Oceanogr.l:113-126. Cury, P., and C. Roy. 1989. Optimal environmental window and pelagic fish recruitment success in upwelling areas. Can. J. Fish. Aquat. Sci. 46:670-680. Gushing, D. H. 1975. Marine ecology and fisheries, 278 p. Cambridge Univ. Press, Cambridge, England. Dessier, A., and J. R. Donguy. 1987. Response to El Nino signals of the epiplanktonic copepod populations in the eastern tropical Pacific. J. Geophys. Res. 92:14,393-14,403. De Vries, D. A., C. B. Grimes, K. L. Lang, and D. B. White. 1990. Age and growth of king and Spanish mackerel larvae and juveniles from the Gulf of Mexico and U.S. South Atlantic Bight. Envir. Biol. Fish. 29:135-143. Essig, R. J., and C. F Cole. 1986. Methods of estimating larval fish mortality from daily increments in otoliths. Trans. Amer. Fish. Soc. 115:34-40. FAO (Food and Agriculture Organization). 2004. Fishery statistics: capture production, 2002. FAO Yearbook, Fish. Stat. 94/1, 654 p. Wexler et al.: Temporal variation in larval growth of Thunnus albacares in the Panama Bight 17 Fiedler, P. C. 1992. Seasonal climatologies and variability of east- ern tropical Pacific surface waters. NOAA Tech. Rep. NMFS 109, 65 p. 2002. The annual cycle and biological effects of the Costa Rica Dome. Deep-Sea Res. I, 49:321-338. Forsbergh, E. D. 1963. Some relationships of metorological, hydrographic, and biological variables in the Gulf of Panama. Inter- Am. Trop. Tuna Comm. Bull. 7:1-109. 1969. On the climatology, oceanography and fisheries of the Panama Bight. Inter-Am. Trop. Tuna Comm. Bull. 14:45-385. Fortier, L. and W. C. Leggett. 1985. A drift study of larval fish survival. Mar. Ecol. Prog. Ser. 25:245-257. Glynn, P. W., J. L. Mate. A. C. Baker, and M. O. Calderon. 2001. Coral bleaching and mortality in Panama and Ecua- dor during the 1997-1998 El Nino-Southern Oscillation event: Spatial/temporal patterns and comparisons with the 1982-1983 event. Bull. Mar. Sci. 69:79-109. Gonzalez Armas, R. 2002. Aggregation of fish larvae in El Bajo Espiritu Santo in the Gulf of California, their changes in the patterns of distribution and the oceanographic processes that influence them. (In Spanish]. Ph D. diss, 216 p. Centro de Investigaciones del Noroeste S. C. La Paz, Baja California South. Govoni, J. J., A. J. Chester, D. E. Hoss, and P. B. Ortner. 1985. An observation of episodic feeding and growth of larval Leiostomus xanthurus in the northern Gulf of Mexico. J. Plankton Res. 7:137-146. Graves, J. E., M. A. Simovich, and K. M. Schaefer. 1988. Electrophoretic identification of early juve- nile yellowfin tuna, Thunnus albacares. Fish. Bull. 86:835-838. Grimes, C. B. 2001. Fishery production and the Mississippi River discharge. Fisheries 26(8):17-26. Grimes, C. B.. and J. J. Isely. 1996. Influence of size-selective mortality on growth of gulf menhaden and king mackerel larvae. Trans. Am. Fish. Soc. 125:741-752. Hoff, G. R., and L. A. Fuiman. 1993. Morphometry and composition of red drum oto- liths: changes associated with temperature, somatic growth rate, and age. Comp. Biochem. Physiol. 106A:209-219. Houde, E. D. 1987. Fish early life dynamics and recruitment vari- ability. Am. Fish. Soc. Symp. 1989. Comparative growth, mortality, and energetics of marine fish larvae: temperature and implied latitudinal effects. Fish. Bull. 87:471-495. Jenkins, G. P., and T L. O. Davis. 1990. Age, growth rate, and growth trajectory deter- mined from otolith microstructure of southern bluefin tuna Thunnus maccoyii larvae. Mar. Ecol. Prog. Ser. 63:93-104. Jenkins. G. P., J. W. Young, and T. L. O. Davis. 1991. Density dependence of larval growth of a marine fish, the southern bluefin tuna, Thunnus maccoyii. Can. J. Fish. Aquat. Sci. 48:1358-1363. Ji, M., A. Leetmaa, and J. Derber. 1995. An ocean analysis system for seasonal to interan- nual climate studies. Mon. Wea. Rev. 123:460-481. Johnson, M. W., J. R. Rooker, D. M. Gatlin III, and G. J. Holt. 2002. Effects of variable ration levels on direct and indi- rect measures of growth in juvenile red drum iSciaenops ocellatus). J. Exp. Mar. Biol. Ecol. 274:141-157. Kaji, T., M. Tanaka, M. Oka, H. Takeuchi, S. Ohsumi, K. Teruya, and J. Hirokawa. 1999. Growth and morphological development of labora- tory-reared yellowfin tuna Thunnus albacares larvae and early juveniles, with special emphasis on the digestive system. Fish. Sci. 65:700-707. Lang, K. L., C. B. Grimes, and R. F. Shaw. 1994. Variations in the age and growth of yellowfin tuna larvae, Thunnus albacares, collected about the Missis- sippi River plume. Environ. Biol. Fish. 39:259-270. Large, W. G., and S. Pond. 1981. Open ocean momentum flux measurements in moderate to strong winds. J. Phys. Oceanogr. 11:324- 336. Lauth, R. R., and R. J. Olson. 1996. Distribution and abundance of larval scombridae in relation to the physical environment in the north- western Panama Bight. Inter-Am. Trop. Tuna Comm. Bull. 21:125-167. MacKenzie, B. R., T. J. Miller, S. Cyr, and W. C. Leggett. 1994. Evidence for a dome-shaped relationship between turbulence and larval fish ingestion rates. Limnol. Oceanogr. 39:1790-1799. Margulies. D. 1993. Assessment of the nutritional condition of larval and early juvenile tuna and Spanish mackerel (Pisces: Scomb- ridae) in the Panama Bight. Mar. Biol. 115:317-330. 2001. Preface. Early life history studies of yellowfin tuna, Thunnus albacares. Inter-Am. Trop. Tuna Comm. Bull. 22:5-8. Margulies, D., J. M. Suter, S. L. Hunt,, R. J. Olson, V. P. Scholey, J. B. Wexler, and A. Nakamura. In press. Spawning and early development of captive yellowfin tuna, Thunnus albacares. Fish. Bull. Margulies, D., J. B. Wexler, K. T. Bentler, J. M. Suter, S. Masuma, N. Tezuka, K. Teruya, M. Oka, M. Kanematsu, and H. Nikaido. 2001. Food selection of yellowfin tuna, Thunnus albacares, larvae reared in the laboratory. Inter-Am. Trop. Tuna Comm. Bull. 22:9-51. Matsumoto, W. M., E. H. Ahlstrom, S. Jones, W. L. Klawe, W. J. Richards, and S. Ueyanagi. 1972. On the clarification of larval tuna identifica- tion particularly in the genus Thunnus. Fish. Bull. 70:1-12. McGowan, J. A., and V. J. Fraundorf. 1966. The relationship between size of net used and estimates of zooplankton diversity. Limnol. Oceanogr. 11:456-469. Miller, T. J., L. B. Crowder, J. A. Rice, and E. A. Marschall. 1988. Larval size and recruitment mechanisms in fishes: toward a conceptual framework. Can. J. Fish. Aquat. Sci. 45:1657-1670. Nishikawa, Y., M. Honma, S. L'eyanagi, and S. Kikawa. 1985. Average distribution of larvae of oceanic species of scombroid fishes, 1956-1981. Far Seas Fish. Res. Lab., S Ser. 12, 99 p. Nishikawa, Y., and D. W. Rimmer. 1987. Identification of larval tunas, billfishes and other scombroid fishes (Suborder Scombroidei): an illustrated guide, 20 p. CSIRO Mar. Lab. Rep. 186, Melbourne, Australia. 18 Fishery Bulletin 105(1) Olson, R. J., and V. P. Scholey, 1990. Captive tunas in a tropical marine research labora- tory: growth of late-larval and early-juvenile black skip- jack Euthynnus lineatus. Fish. Bull. 88:821-828. Owen, R. W. 1989. Microscale and finescale patchiness of small plank- ton in coastal and pelagic environments. J. Mar. Res. 47:197-240. Owen, R. W., and B. Zeitzschel. 1970. Phytoplankton production: seasonal change in the oceanic eastern tropical Pacific. Mar. Biol. 7:32-36. Pepin, P. 1988. Predation and starvation of larval fish: a numerical experiment of size-and growth-dependent survival. Biol. Ocean. 6:23-44. 1991. Effect of temperature and size on development, mortality, and survival rates of the pelagic early life history stages of marine fish. Can. J. Fish. Aquat. Sci. 48:503-518. Peterman, R. M., M. J. Bradford, N. C. H. Lo, and R. D. Methot. 1988. Contribution of early life stages to interannual variability in recruitment of northern anchovy (Engrautis mordax). Can. J. Fish. Aquat. Sci. 45:8-16. Potthoff, T. 1974. Osteological development and variation in young tunas, genus Thunnus (Pisces, Scombridae), from the Atlantic Ocean. Fish. Bull. 72:563-588. Richards, W. J., T. Potthoff, and J. Kim. 1990. Problems identifying tuna larvae species (Pisces: Scombridae: Thunnus) from the Gulf of Mexico. Fish. Bull. 88:607-609. Rochford, D. J. 1962. Hydrology of the Indian Ocean. Part II: the surface waters of the southeast Indian Ocean and the Arafura Sea in the spring and summer. Aust. J. Mar. Fresh- water Res. 13:226-251. Rothschild, B. J., and T. R. Osborn. 1988. Small scale turbulence and plankton contact rates. J. Plankton Res. 10:465-474. Schaefer, K. M. 1991. Geographic variation in morphometric charac- ters and gill-raker counts of yellowfin tuna Thun- nus alhacares from the Pacific Ocean. Fish. Bull. 89:289-297. 1998. Reproductive biology of yellowfin tuna {Thunnus albacares) in the eastern Pacific Ocean. Inter-Am. Trop. Tuna Comm. Bull. 21:201-272. 2001. Reproductive biology of tunas. 7n Fish physiology. Vol. 19, Tuna: physiology, ecology, and evolution (B. A. Block, and E. D. Stevens, eds.), p. 225-270. Academic Press, San Diego, CA. Scholey, V. P. 1993. Effects of temperature and food concentration on growth and survival of late-larval and early-juvenile black skipjack, Euthynnus lineatus. M.Sci. thesis, 56 p. Univ. Washington, Seattle, WA. Shima, M., and K. M. Bailey. 1994. Comparative analysis of ichthyoplankton sampling gear for early life stages of walleye pollock (Theragra chalcogramma). Fish. Oceanogr. 3:50-59. Smayda, T. J. 1963. A quantitative analysis of the phytoplankton of the Gulf of Panama I. Results of the regional phytoplankton surveys during July and November, 1957 and March, 1958. Inter-Am. Trop. Tuna Comm. Bull. 7:191-253. 1966. A quantitative analysis of the phytoplankton of the Gulf of Panama III. General ecological condtions, and the phytoplankton dynamics at 8°45'N, 79''23'W from November 1954 to May 1957. Inter-Am. Trop. Tuna Comm. Bull. 11:353-612. Smith, P. E., and S. L. Richardson. 1977. Standard techniques for pelagic fish egg and larva surveys. FAO Fish. Tech. Paper, 175, 100 p. FAO, Rome. Strutton, P. G., and F. P. Chavez. 2000. Primary productivity in the equatorial Pacific during the 1997-98 El Nino. J. Geophys. Res. 105:26,089-26,101. Sverdrup, H. U., M. W. Johnson, and R. H. Fleming. 1942. Dynamics of ocean currents. Chapter 13. /n The oceans their physics, chemistry, and general biology, p. 431-515. Prentice-Hall, Inc., New York, NY. Takeyama, H., S. Chow, H. Tsuzuki, and T. Matsunaga. 2001. Mitochondrial DNA sequence variation within and between tuna Thunnus species and its application to species identification. J. Fish Biol. 58:1646-1657. Thorrold, S. R. 1993. Post-larval and juvenile scombrids captured in light traps: preliminary results from the central Great Barrier Reef lagoon. Bull. Mar. Sci. 52:631-641. Ueyanagi, S. 1969. Observations on the distribution of tuna larvae in the Indo-Pacific Ocean with emphasis on the delin- eation of the spawning areas of albacore, Thunnus alalunga. Bull. Far Seas Fish. Res. Lab. 2:177-256. 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. Wexler, J. B. 1993. Validation of daily growth increments and estima- tion of growth rates of larval and early-juvenile black skipjack, Euthynnus lineatus, using otoliths. Inter-Am. Trop. Tuna Comm. Bull. 20:399-440. Wexler, J. B., D. Margulies, S. Masuma, N. Tezuka, K. Teruya, M. Oka, M. Kanematsu, and H. Nikaido. 2001. Age validation and growth of yellowfin tuna, Thun- nus albacares, larvae reared in the laboratory. Inter- Am. Trop. Tuna Comm. Bull. 22:52-91. Wild, A. 1994. A review of the biology and fisheries for yellow- fin tuna, Thunnus albacares, in the eastern Pacific Ocean. FAO Fish. Tech. Paper 336:52-107. FAO, Rome. Woodbury, D. 1999. Reduction of growth in otoliths of widow and yel- lowtail rockfish (Sebastes entomelas and S. flavidus) during the 1983 El Nino. Fish. Bull. 97:680-689. Wooster, W. S. 1959. Oceanographic observations in the Panama Bight, "Askoy" Expedition, 1941. Bull. Am. Mus. Nat. Hist. 118:119-151. Wroblewski, J. S., J. G. Richman, and G. L. Mellor. 1989. Optimal wind conditions for the survival of larval northern anchovy, Engraulis mordax — a modeling investigation. Fish. Bull. 87:387-398. Young, J. W., and T. L. O. Davis. 1990. Feeding ecology of larvae of southern bluefin, alba- core and skipjack tunas (Pisces: Scombridae) in the east- ern Indian Ocean. Mar. Ecol. Prog. Ser. 61:17-29. Zar, J. H. 1984. Biostatistical analysis, 2"12 mm SL) with 12 mm SL as the biological intercept. The mean back-cal- culated size-at-age trajectory was compared to measured standard lengths at time of sampling. We evaluated the fit of each model to the standard length-at-sampling data by comparing the r~ values, as well as the minimum and maximum percent deviation of the mean back-calculated size-at-age value from the mean length-at-sampling. Hobbs et al,: Modification of the biological intercept model to account for ontogenetic effects in Hypomesus transpacificus 33 Table 2 Summary of the three back-calculation models examined in this study: the time-varying growth (TVG) model (Siroiset al. 1998), modified Fry (MF) model (Vigliola et al. 2000). and the biological intercept (BI) model (Campana 1990). L=standard length; if=otolith radius; Lop = standard length-at-biological intercept; L, = standard length-at-age i\ L ,=standard length-at-capture; i?(,p=otolith radius-at-biological intercept; fi, = otolith radius-at-age i\ iJj,p,=otolith radius-at-capture; W=mean otolith increment width during each life stage; W,=otolith increment width at i; G^.=growth effect; and a = allometric shape parameter. Back-calculation models Equation Reference Time-varying growth (TVG) Modified Fry (MF) Biological intercept (BI) A = -Lop J(W^, + G,( W, - W))(L^^, - L,^iR,^, - R„^)-^ Siroiset al. (1998) L^= a + exp ilniLg^ -a) + IniL^^i -a) -IniLg^ - a)) Un(R^)-tn(Rgp)){ln{R^p,)-ln{Rgp)r^) Vigliolaet al. (2000 L, = L.^, + ^n, - fl.p,) tL^p,-Lg^HR,p, - Rg^}-^ Campana (1990) Results Validation of daily otolith increment formation The relationship between the number of increments and days after hatching of delta smelt larvae are shown in (Table 1; Fig. 1). The slope of the regres- sion of increment count on known-age was not sig- nificantly different from one and thus indicated that increment formation occurred daily. However, the intercept was significantly different from zero (P<0.001), indicating that the first increment was not laid at hatching, rather that ring formation began 6 dah. This observation was confirmed by examination of larvae sampled at one and five dah (Table 1). Mean somatic and otolith growth All somatic otolith-growth relationships were best described by life-stage-specific linear regression models, where larval (0-20 dah, 5-12 mm SL) and juvenile (>20 dah, >12 mm SL) life stages were considered separately. Calculated Akaike informa- tion criterion (Sokal and Rohlf, 1973) for the linear models were lower than polynomial models ranging from the 2"'' to 9'^ orders. Somatic growth showed varia- tions in growth over time: fast growth occurred from hatching to 40 dah, followed by a period of slowed growth from 40 dah up to 80 dah. After 80 dah, fish experienced a period of rapid somatic growth associated with the juvenile stage (Fig. 2, A and C). Otolith growth showed a different trend. Otolith growth was slow from hatching to 40 dah, which then increased exponentially from 40 dah to 100 dah, indicating that the relationship between otolith growth and fish growth changes abruptly around 40 dah with the completion of caudal flexing (Fig. 2B). Finally, the relationship between otolith size and fish size was best described by a stage-specific linear regres- sion (Fig. 2D), which accounted for the lack of constant linear proportionality of otolith growth to fish growth. It is important to note that some patterns in the residuals were apparent in the early larval stages. However, we do 100- ^ ' 90- y'oX ,' Ji 80- 1 70- 1 60- o B 50- o 1 40- E i 30- .f// 20- /'[X^ 10- ,'''jr ""1 1 1 ' 1 ' 1 ' 1 10 20 30 40 50 60 70 80 90 100 Known age (days after hatching) Figure 1 Relationship between the number of increments and the known age for delta smelt {Hypomesus transpacificicus). Dotted line is the 1:1 ratio line. Solid line is the linear regression line. not consider these slight deviations to have a significant effect on further residual analyses. Growth and ontogenetic effects and size back-calculations Correlations of age-independent effects and growth-rate effects are shown in Figure 3, A and B. The strong cor- relation between standard length-on-age residuals and otolith radius-on-age residuals may indicate that otolith size is proportional to fish size. The age-independent variability in the OS-FS relationship was accounted for by examining the unexplained variability in the residual analysis of otolith and fish size-on-age. Only 11% of the unexplained variability could be associated with age- independent effects. The Pearson correlation coefficient for the residual of standard length-on-age and otolith 34 Fishery Bulletin 105(1) 10 20 30 40 50 60 70 80 90 100 -| C Age (d) 90- " X 80- ^^ ^^ ^ yM 70- X CD (0 ■o 60- 50- OOJ& CD GD m/^OD CD 40- 0.05) between the two study sites. For this reason, the data from both sites were pooled. The index was highly variable for all species (Fig. 2). For blue rock- fish, the index varied from 181 fish/min in 1987 to 0.26 fish/min in 1992. For yellowtail rockfish, the index ranged from 162 fish/min in 1985 to 0.03 fish/min in 1994. Abundance of black rockfish peaked in 1999 at 22 fish/min and was lowest in 1998 at 0.01 fish/min. ■• NOAA (National Oceanic and Atmospheric Administration). Environmental Research Division. Southwest Fisheries Science Center. NOAA/NMFS/SWFSC. 1352 Lighthouse Ave., Pacific Grove, CA 93950-2097. Website: http://www. pfeg.noaa.gov (accessed 15 March 2006). ^ CALCOM (California Cooperative Survey). Commercial landings sampling program maintained by California Depart- ment of Fish and Game, 350 Harbor Blvd., Belmont, CA 94002; Pacific States Marine Fisheries Commission, 350 Harbor Blvd., Belmont, CA 94002; and Fisheries Ecology Division, SWFSC, NMFS, NOAA, 110 Shaffer Rd., Santa Cruz. CA 95060. Website: 128.114.3.187 (accessed on 15 March 2006). 42 Fishery Bulletin 105(1) Table 1 Number of dives and total number of one-minute surveys of juveni e rockfishes by year at each dive site. Data were collected | between 1 July and 15 September, 1983- 2003. Year Number of dives Number of 1-minute surveys Dark Gulch Sa mon Point Total Dark Gulch Salmon Point Total 1983 12 2 14 270 60 330 1984 8 3 11 185 99 284 1985 8 4 12 94 35 129 1986 9 2 11 124 30 154 1987 9 3 12 94 18 112 1988 4 3 7 62 38 100 1989 5 3 8 119 68 187 1990 4 2 6 71 33 104 1991 3 3 6 60 60 120 1992 3 3 6 60 60 120 1993 2 2 4 40 40 80 1994 3 3 6 60 60 120 1995 4 3 7 80 40 120 1996 4 1 5 88 22 110 1997 5 3 8 124 71 195 1998 7 2 9 188 40 228 1999 6 5 11 133 89 222 2000 4 3 7 82 84 166 2001 3 3 6 62 63 125 2002 4 4 8 85 87 172 2003 6 2 8 84 41 125 Total 113 59 172 2165 1168 3333 Average 5.4 2.8 dives/y ear 8.2 103.1 55.6 surveys/year 158.7 Standard deviation 2,6 0.9 2.7 54.7 22.8 63.9 Year-to-year variability was generally synchronous among the three species (Fig. 2). Abundance was below average for the three species in 1983-84, and 1989-98. Above average abundance in all three species occurred in 1985, 1987, 1988, and in 2001. Black rockfish at- tained relatively greater numbers in 1986, 1999, 2000. and 2003 compared to the other two species. In addition, blue rockfish experienced higher than average abun- dances in 2001 and 2002 compared to the other species. Generally, the index for black rockfish was lower than for the other two species, but black rockfish abundance was extremely high in 1999, but the abundance of the other two species was below average. Indices for blue and yellowtail rockfish were significantly correlated (P<0.001, r=0.91), and the index for black rockfish was not significantly correlated with either of the other two species. Using the log-transformed index, we found that blue, yellowtail, and black rockfish were all significantly correlated (P<0.01, r=0.76) with each other. We used the log-transformed index for the remaining analyses. In evaluating seasonal oceanographic variables, sea level anomaly and nearshore temperature were sig- nificantly and positively correlated (P<0.05, 7-=0.53), nearshore temperature and offshore Ekman transport were significantly and negatively correlated (P<0.05, r=-0.69), and sea level anomaly and offshore Ekman transport were not significantly correlated. There was no significant correlation between the seasonal oceano- graphic variables and the index for any species. With PCA, 65% of the variability in the monthly oceanographic (q) data sets was explained by the first eigenvector (PCIq). PCIq was characterized by the contrast between Ekman transport and the other two variables (Table 2). The second eigenvector (PC2q) ex- plained 30% of the variability and was associated with high sea level anomaly. For the rockfish abundance (p) indices, 77% of the variability was explained in the first eigenvector (PClp), which was associated with the abundance time series of all three species. The second eigenvector (PC2p) explained 15% of the variability and was associated with the contrast between blue rockfish abundance and the other two species, especially late in the time series. Although PC3p explained only 8% of the variability in these time series, it was associated Laidig et al.; Relationship between abundance of juvenile Sebastes spp. and environmental variables 43 Table 2 Eigenvectors determined from the principal components analysis (PCAl for the monthly oceanographic variables (q) and log-transformed annual juvenile rockfish abun- dance ( p) index by species PClo PC2o PC3o % variance explained 64.5 29.5 6.0 Nearshore temperature 0.68 -0.04 0.73 Sea level anomaly 0.46 0.80 -0.39 Ekman transport -0.57 0.40 0.56 PClp PC2p PC3p 9c variance explained 77.1 14.6 8.3 Blue rockfish 0.55 0.83 0.13 Yellowtail rockfish 0.60 -0.28 -0.75 Black rockfish 0.5S -0.49 0.64 with the unique pattern in abundance of black rockfish, especially in contrast to yellowtail rockfish. Correlations varied between the log-transformed abundance indices and the monthly means of the oceanographic variables (Table 3). All three species were significantly and negatively correlated with sea level anomaly during some of the months from Janu- ary to June. For nearshore monthly temperature, there was a similar pattern of negative correlation with in- dex for all species during January to June. Only blue rockfish were significantly correlated (P<0.001, r=0.58) with offshore Ekman Transport, and only in Febru- ary (Table 3). Months of high nearshore temperature or sea level anomaly resulted in low juvenile rockfish abundances. For example, when the abundance index for blue rockfish was compared with sea level anomaly in February (Fig. 3), years of highest abundance of blue rockfish were always years of low sea level anomaly (i.e., 1985, 1987, 1988, 2001, and 2002), but years of low sea level anomaly did not always lead to particularly high rockfish abundance (i.e., 1989). However, years of highest sea level anomaly (e.g., 1983, 1992, and 1998) resulted in the lowest blue rockfish abundance. The results of the CCA also indicate that low annual abundance was associated with years of high nearshore temperatures and high sea level anomaly. The first canonical correlation was 0.57, which was significantly different from zero (P<0.0001), and explained 97 % of the covariance between the two data sets. The remain- ing two canonical correlations were not significant and explained little of the variability. Blue, yellowtail, and black rockfish were negatively correlated with the first canonical variable for the oceanographic data set (-0.43, -0.45, and -0.55, respectively; Fig. 4). This indicates that fish abundances were low when temperatures and sea level anomalies were high. Blue rockfish annual abundance x=25-1 SD=45.2 = = Jl I S3 M es B6 87 90 9I9293W9S96 97 989900 0I Yellowtail rockfish annual abundance "" x=188 180 . SD=41.8 c 160 E " QJ 120 E 100 T C 80 1 l± !0 5 60 40 : Mi 20 :- J.II ^"=ii ■■=iil.- S3 B4 85 ee a? 88 S9 90 31 9£ 93 Black rockfish annual abundance x=3.5 SD=5.0 jiil i.E Itail 83 B4 as 86 87 88 89 90 91 92 93 W 95 96 97 98 99 00 01 02 03 Year Figure 2 Mean annual juvenile rockfish abundance index for all one-minute surveys in two kelp beds for blue iSebastes mystinus), yellowtail (S. flavidus), and black (S. melanops) rockfish by year. Note the different scale for black rockfish. Error bar=one standard error. The 21-year mean (represented by a dashed line) and the standard deviation for each species are given on each plot. The sample size for each year is reported in Table 1. The data for year-class strength of commercially caught adult yellowtail rockfish corresponded with the annual abundance index for juvenile yellowtail rockfish (Table 4), and were significantly correlated for all ports combined for 1997 (P<0.01, n= 29, r=0.48) and 1999 (P<0.04, n = 31, r=0.36), but not for 1998 {P<0.1, n=21, r=0.37). For 1997, juvenile rockfish abundance was sig- nificantly correlated with adult fish numbers at Bodega 44 Fishery Bulletin 105(1) Bay (P<0.02, R = 0.43, Fig. 5A), Ft. Bragg (P<0.001, r=0.62), and Eureka (P<0.01, r=0.78). In 1998, Bodega Bay was the only port where adult fish numbers were significantly correlated (P<0.05, r=0.38. Fig. 5B) with juvenile abundance. In 1999, Eureka was the only port where adult fish numbers were significantly correlated tP<0.05, i? = 0.35) with juvenile abundance. From the 1998 adult numbers in Bodega, 1985 was the largest year class of adult yellowtail rockfish, which corre- 6.0 r 5.0 i 4.0 o 2 3.0 0) ^ 2.0 o I 10 c g' 0.0 _l -1.0 86 88 02 -2.0 -200 83 98 -100 100 200 Sea level anomaly in February 300 Figure 3 An example of the log-transformed annual abundance index for blue rockfish tSebastes mystinus) and average sea level anomaly in February for each year of the survey (198.3-2003). Numbers represent individual years. Dashed lines represent the zero line for both axes. 1 Ekman 0.8 Transport , correlation o o 4^ d ' S 0.2 'c ' S o Blue 1 Sea Level ■D c -0.2 o 0! cn -0.4 1 anomaly 1 Temperature -0.6 -0.6 -0.4 -0.2 0.2 0.4 0.6 08 1 First canonical correlation Figure 4 The first and second canonical correlations for the annual abundance index (INDEX) for each rockfish species and the oceanographic fac- tors with the axes defined by the oceanographic data set. Dashed lines represent zero line for both axes. sponded to the highest index for juvenile yellowtail rockfish (Fig. 5B). Discussion The 21-year time series of juvenile rockfish abundance allowed us to examine long-term change in recruitment of commercially and recreationally important species off the coast of California. The year-to-year variability in recruitment likely relates to variability in year-class strength of the population entering the fisheries. The syn- chrony in recruitment variability among the three rockfish species indicates that similar environmental processes affect the abundance of all three species. By exam- ining oceanographic variables, we deter- mined that sea level anomaly and nearshore temperatures in February and March were important influences on juvenile rockfish abundance. Year-to-year variability in young rock- fish abundance has been documented in other studies off the west coast of the United States. Yoklavich et al. (1996) found a twenty-fold increase in the abun- dance of pelagic larval rockfishes off cen- tral California in 1993 compared to num- bers obtained during a similar time period in 1992. They attributed this difference to increased offshore transport and pos- sibly lower predation rates. Moser et al. (2000) observed large fluctuations in annu- al larval rockfish abundance off southern California from 1951 to 1998, which was attributed to the reproductive output of each species and oceanographic variables. Mearns et al. (1980) determined that the variability in recruitment of juveniles was the major source of seasonal and annual fluctuations in rockfish catches for strip- etail and calico rockfish (S. dallii). Mat- thews (1989) observed that recruitment levels varied between years for three spe- cies of rockfishes recruiting to nearshore habitats. Ainley et al. (1993) discovered a three-fold difference in pelagic juvenile rockfish abundance in seabird diets in cen- tral California between similar periods in 1985 and 1986, and they attributed this to cross shelf advection of larvae in January and February. Ralston and lanelli (1998) reported a large variability in juvenile bo- caccio abundance over a 13-year period and attributed some of this variability to El Niiio events. Year-class strength was likely estab- lished in the period from February through March during the larval stage of the three Laidig et al : Relationship between abundance of juvenile Sebostes spp, and environmental variables 45 Table 3 Correlations between the log- transformed annual juvenile rockfish abundance inde.x by species and the oceanographic vari- ables by month. * = P<0.05; *' = P<0.01; + = P<0.1. Jan Feb Mar Apr May Jun Sea level anomaly Blue rockfish -0.36 -0.71** -0.63** -0.51* -0.37+ -0.40+ Yellowtail rockfish -0.48* -0.53* -0.57** -0.55** -0.55** -0.35 Black rockfish -0.64** -0.53* -0.52* -0.42+ -0.65** -0.55** Temperature Blue rockfish -0.22 -0.53* -0.52* -0.41+ -0.19 -0.21 Yellowtail rockfish -0.32 -0.54* -0.53* -0.48* -0.46* -0.32 Black rockfish -0.54* -0.62** -0.55** -0.47* -0.65** -0.48* Offshore Ekman transport Blue rockfish 0.05 0.58** 0.42+ 0.01 -0.02 0.03 Yellowtail rockfish 0.27 0.27 0.39 + 0.21 0.17 -0.01 Black rockfish 0.29 0.25 0.19 0.29 0.40+ 0.01 Correlation coefficients between the log-trans tail rockfish (Sebastes flavidus) commercial **= P<0.01; + = P<0.1. Listed in parentheses yellowtail rockfish otoliths that were aged (no Table 4 formed annual juvenile rockfish abundance index and year-specific adult yellow- landings, by year class, at three ports closest to the study area. * = P<0.05; is the number of year classes determined in each year and the number of adult of year classes, number of otoliths aged). All ports Bodega Bragg Eureka 1997 1998 1999 0.48(10, 172)** 0.37 (9, 141)+ 0.36 (12, 198)* 0.43(10,16)* 0.38(6,30)* 0.32 (9, 84)+ 0.62(10,84)** 0.21(9,89) 0.25(11,37) 0.78(9,72)** 0.32(7,22)+ 0.35(12,77)* species of rockfishes in our study. The abundance of all three species were significantly and negatively correlat- ed with sea level anomaly and nearshore temperature during this time period. Ralston and Howard (1995) also argued that year-class strength was set in the lar- val period for rockfishes with winter parturition. They analyzed data from midwater trawls in May and June and compared them with data from nearshore surveys in summer. Because there was a strong correlation be- tween the two data sets, they postulated that the year- class was set earlier in the year than May, probably during the larval stage. VenTresca et al.^ also reported evidence of the establishment of year-class strength in the larval stage. In 1992, they found large concentra- tions of larval rockfishes in January, but three to four months later very few juveniles appeared in midwater trawls. They surmised that the El Nino conditions of elevated water temperatures and reduced upwelling resulted in poor survival. Synchrony in juvenile abundance among rockfish spe- cies has been observed in other studies. Ralston and Howard (1995) ascertained that trends in abundance for juvenile blue and yellowtail rockfish from midwater trawls were highly correlated over the 10 years of their study. Ammann (2001) discovered a comparable pattern in recruitment of juvenile yellowtail and black rockfish to the kelp bed environment in 1999 and 2000. Ste- phens et al. (1984) reported that juvenile abundance of both blue and olive rockfish (S. serranoides), dropped to virtually zero during the years 1978-81. In our study, blue, yellowtail, and black rockfish covaried over a pe- riod of 21 years. Although offshore Ekman transport, or upwelling, has been suggested as a predictor of year-class strength, we found it to have little correlation with the abundance of juvenile rockfishes from northern California. Maximum upwelling off the central California coast occurs in late spring and summer (Rosenfeld et al., 1994; Yoklavich et al., 1996). This upwelling occurred after the larval period for winter-spawning Sebastes spp., and therefore after the timing of year-class determination. Larson et al. (1994) found that larger pelagic juveniles were often close to shore even when upwelling was strong, indicating that later-stage pelagic juveniles were not directly affected by upwelling. If water movement on- shore and offshore influences the population size of juvenile rockfishes, perhaps this effect occurs during the early stages. 46 Fishery Bulletin 105(1) Temperature and sea level anomaly proved to be im- portant correlates with year-class strength for juvenile rockfishes. Of the three environmental variables ex- amined, nearshore temperature and sea level anomaly were significantly and negatively correlated with the abundance of all three species. Moser et al. (2000) reported a similar relationship, with reduced abun- dance of larval bocaccio and cowcod during periods of high temperature off southern California. Ralston and Howard (1995) reported a negative correlation between recruited juvenile blue and yellowtail rockfish year- class strength and sea surface temperature in January through March. Stephens et al. (1984) surmised that warm water was the limiting factor in the low recruit- ment of juvenile blue and olive rockfish during 1978-81 4.000 A 1997 84 3,000 88 85^^^ 2,000 : 90 ^ ^^''^^87 1,000 - 83 ^ ""'^ n ^ |92, 1 89 1 1 86 91 1 . 1 1 1 1 > 1 -1 10 000 000 000 ,000 ,000 ,000 ,000 ,000 ,000 ,000 03 1998 85 / - 84/"^ - 83 86 . . 1 .r- 1 ''^ 89 90 1 1 1 1 1 1 1 1 ..- J — . — 1 -3 -2 -1 1 2 Log (index) Figure 5 Year-specific catch-at-age data from commercial trawl-caught adult yellowtail rockfish (Sebastes flavidus) from Bodega Bay, CA, and the log-transformed annual abundance index for yellowtail rockfish. (A) 1997 landings; (B) 1998 landings. Numbers represent the year class of aged fish. The solid line represents predicted values from the relationship of the catch-at-age data and the log-transformed index. off southern California. Ainley et al. (1993) determined that sea level during February and sea surface tempera- ture in March were negatively correlated with pelagic rockfish abundance for the period of 1973-90 off central California. Temperature affects the growth rates of rockfishes, which, in turn, affects species abundance. Johnson et al. (2001) determined that growth rates declined during months of high temperatures for juveniles of three rockfish species. However, Boehlert and Yoklavich (1983) found increased growth rates forjuvenile black rockfish with increased temperature in the laboratory, except under starvation conditions. Therefore, reduc- tion in growth rates may be due to lower prey avail- ability during El Nino conditions (Mullin and Conversi, 1989). This lower growth rate of juvenile rockfishes during periods of high tempera- tures may lead to reduced survival and lower year-class strength. Although growth can vary directly with temperature, temperature may also have an indirect effect on rockfish abundance. Strong El Nifio events, associated with unusually high water temperatures last- ing from a few months to over a year, have occurred off California in 1982-83, 1991-92, and 1997-98 (Fedorov and Phi- lander, 2000; Rebstock, 2001). VenTresca et al. (1995) observed reduced condition factor and gonadal indices for blue rock- fish during the 1983 and 1992 El Niiios off central California when water tem- peratures were elevated. These reduc- tions could lead to fewer larvae being produced and hence ultimately to a lower abundance of juveniles. El Nifio events can also lead to changes in the strength and timing of the annual phytoplank- ton bloom, both of which can reduce the distribution and abundance of the zoo- plankton on which the juvenile rockfishes feed (Lenarz et al., 1995). This reduction in food availability could lead to lower growth or survival of the juvenile rock- fishes. Keister et al. (2005) found sev- eral warm-water species of euphausiids, chaetognaths, and copepods in Oregon waters during the 1997-98 El Nifio. Reb- stock (2001) observed that during periods of high temperatures in 1983, 1992, and 1998, the species richness of copepods was lower than a 49-year average. Co- pepods are a preferred prey item for ju- venile rockfishes (Reilly et al., 1992). A change in species richness may reflect a change in species composition to less desirable food sources (e.g., prey is less nutritious or less available to young rock- fishes). Although the exact mechanism is not clear, we observed that the three Laldig et al Relationship between abundance of luvenile Sebastes spp, and environmental variables 47 years of lowest abundance for juvenile rockfishes oc- curred during El Nino events. The primary factors that lead to annual fluctuations in abundance were similar for blue, yellowtail, and black rockfish. These three species are found in similar areas at the planktonic stage (Lenarz et al., 1991; Lar- son et al., 1994). Therefore, it is expected that changes in ocean conditions affect all three species similarly. Nearshore temperature and sea level anomaly had high negative correlations with abundance for all three spe- cies, whereas offshore Ekman transport was not corre- lated with abudance. This finding implies that poor re- cruitment occurs during years of high temperature and strong, positive sea level anomaly (poleward flow), and vice versa. Our results indicate that recruitment is poor during periods of strong, positive sea -level anomaly and that recruitment is strong only during years of negative sea level anomaly (equatorward flow). However, poor recruitment also occurs in some years with negative sea level anomaly. Therefore, other factors are prob- ably involved in the process and can affect year-class strength in rockfishes. Some of the other factors that have been suggested to have at least some influence on rockfish recruitment include adult spawning biomass (Mason, 1998), increased predation by siphonophores and chaetognaths on larval stages during years of high sea temperature (Yoklavich et al., 1996), turbulence (Ainley et al., 1993), and diet of juvenile rockfishes (Reilly et al., 1992). Large-scale multiyear oceanographic events (e.g.. Pa- cific Decadal Oscillation and the El Nino-Southern Os- cillation indices) also appear to affect juvenile rockfish abundance. Large changes in indices reflect regime shifts in ocean conditions, such as those occurring in 1977, 1989. and 1998 (Hare and Mantua, 2000; Ben- son and Trites, 2002). Although the mechanisms that affect or cause changes in abundance are unclear, our time series of juvenile rockfish abundance reflects these large-scale shifts in ocean conditions by the generally high recruitment prior to 1989, the much reduced re- cruitment from 1989 to 1998, and the generally higher recruitment after 1998. We have continued our juvenile rockfish surveys to present, and have had the opportunity to determine the usefulness of our abundance index as a predictor of rockfish year-class strength. In 2005, average monthly temperature was elevated from January to June (as much as two degrees above average as measured by our temperature monitors). Interestingly, the abundance index of all three species of juvenile rockfishes from our surveys in 2005 was very low. Black and yellowtail rockfish abundances, in particular, were at the third lowest level estimated during what is now a 23-year time-series of recruitment. Therefore, our results dem- onstrate our ability to predict annual levels of abun- dance for these species of juvenile rockfishes. Juvenile yellowtail rockfish abundance in our study reflected adult yellowtail rockfish abundance in the fishery. Mearns et al. (1980) also reported a relation- ship between juvenile abundance and subsequent adult biomass for stripetail and calico rockfishes by following yearly cohorts from seasonal trawls in southern Califor- nia over nine years. The high recruitment for bocaccio in 1985 was manifested in the recreational fishery in Monterey Bay, CA in subsequent years (Mason, 1998). Similar trends were observed with large year-classes of chilipepper, S. goodei, and yellowtail rockfish (Mason, 1998). Ralston and lanelli (1998) also found that the abundance of juvenile bocaccio was an indicator of year- class strength in the fishery. Accordingly, the study of juvenile rockfish abundance can help predict good and bad year classes entering a fishery. These data can then be incorporated into fisheries models (see stock assess- ments for widow rockfish, S. entomelas, [He et al.*"]) to better manage the stocks. Acknowledgments We first thank Edmund (Ted) Hobson for his initial conception of this project and his many years of data collection and direction. We also thank all the divers who helped with juvenile rockfish surveys, especially Kelly Silberberg for his many years of dedication to this project. We thank Peter Adams, Steve Ralston, Susan Sogard, Ralph Larson, Mary Yoklavich, and four anony- mous reviewers for helpful comments on drafts of this manuscript. Lastly, we thank Craig Syms and Brian Wells for their generous help with statistical matters. Literature cited Ainley, D. G., W. J. Sydeman, R. J. Parrish, and W. H. Lenarz. 1993. Oceanic factors influencing distribution of young rockfish (Sebastes) in central California: A predator's perspective. Calif. Coop. Oceanic Fish. Invest. Rep. 34:133-139. Ammann, A. 2001. Evaluation of a standard monitoring unit for the recruitment offish in central California. M.Sci. thesis, 92 p. Univ. California, Santa Cruz, CA. Anderson, T. W. 1983. Identification and development of nearshore juve- nile rockfishes (genus Sebastes) in central California kelp forests. M.Sci. thesis, 216 p. California State Univ., Fresno, CA. Benson, A. J., and A. W. Trites. 2002. Ecological effects of regime shifts in the Bering Sea and eastern North Pacific Ocean. Fish and Fish- eries 3:95-113. Boehlert, G. W., and M. M. Yoklavich. 1983. Effects of temperature, ration, and fish size on growth of juvenile black rockfish, Sebastes melanops. En- viron. Biol. Fish. 8:17-28. 6 He, X., D. E. Pearson, E. J. Dick, J. C. Field, S. Ralston, and A. D. MacCall. 2005. Status of the widow rockfish resource in 2005, 119 p. Pacific Fisheries Management Council, 7700 NE Ambassador Place, Suite 200, Portland, Oregon 97220. 48 Fishery Bulletin 105(1) Butler, J. L., L. D. Jacobson, J. T. Barnes, and H. G. Moser. 2003. Biology and population dynamics of cowcod (Sebastes tevis) in the southern California Bight. Fish. Bull. 101:260-280. Fedorov, A. V., and S. G. Philander. 2000. Is El Nifio changing? Science 288:1997-2002. Hare, S. R., and N. J. Mantua. 2000. Empirical evidence for North Pacific regime shifts in 1977 and 1989. Prog. Oceanogr. 47:103-145. 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. 42:153-166. Keister, J. E., T. B. Johnson, C. A. Morgan, and W. T. Peterson. 2005. Biological indicators of the timing and direction of warm-water advection during the 1997/1998 El Nino off the central Oregon coast, USA. Mar. Ecol. Prog. Ser. 295:43-48. Larson, R. J., W. H. Lenarz, and S. Ralston. 1994. The distribution of pelagic juvenile rockfish of the genus Sebastes in the upwelling region off central California. Calif. Coop. Oceanic Fish. Invest. Rep. 35:175-221. Lenarz, W. H., R. J. Larson, and S. Ralston. 1991. Depth distributions of late larvae and pelagic juve- niles of some fishes of the California Current. Calif. Coop. Oceanic Fish. Invest. Rep. 32:41-46. Lenarz, W. H., F. B. Schwing, D. A. VenTresca, F. Chavez, and W. M. Graham. 1995. Explorations of El Nifio events and associated biological population dynamics off central California. Calif Coop. Oceanic Fish. Invest. Rep. 36:106-119 Love, M. S., M. Yoklavich, and L. Thorsteinson. 2002. The rockfishes of the northeast Pacific, 405 p. Univ. California Press, Berkeley, CA. Mason, J. 1998. Declining rockfish lengths in the Monterey Bay, California, recreational fishery, 1959-94. Mar. Fish. Rev. 60:15-28. Matthews, K. R. 1989. 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. Mearns, A. J., M. J. Allen, M. D. Moore, and M. J. Sherwood. 1980. Distribution, abundance, and recruitment of soft- bottom rockfishes (Scorpaenidae: Sebastes) on the south- ern California mainland shelf Calif Coop. Oceanic Fish. Invest. Rep. 21:180-190. Moser, H. G., R. L. Charter, W. Watson, D. A. Ambrose, J. L. Butler, S. R. Charter, and E. M. Sandknop. 2000. Abundance and distribution of rockfish iSebastes) larvae in the Southern California Bight in relation to environmental conditions and fishery exploitation. Calif Coop. Oceanic Fish. Invest. Rep. 41:132-147. MuUin, M. M., and A. Conversi. 1989. Biomasses of euphausiids and smaller zooplankton in the California Current — geographic and interannual comparisons relative to the Pacific whiting, Merluccius productus, fishery. Fish. Bull. 87:633-644. Fasten, G. P., S. Katayama, and M. Omori. 2003. Timing of parturition, planktonic duration, and settlement patterns of the black rockfish, Sebastes inermis. Environ. Biol. Fish. 68:229-239. Ralston, S., and D. F. Howard. 1995. On the development of year-class strength and cohort variability in two northern California rock- fishes. Fish. Bull. 93:710-720. Ralston, S.. and J. N. lanelli. 1998. When lengths are better than ages: the complex case of bocaccio. In Fishery stock assessment models: proceedings of the international symposium on fishery stock assessment models for the 21st century (October 8-11, 1997, Anchorage, Alaska) (F. Funk, T. J. Quinn II, J. Heifetz, J. N. lanelli, J. E. Powers, J. F. Schweigert, P. J. Sullivan, and C.-I. Zhang, eds.), p. 451-468. Lowell Wakefield Fisheries Symposium No. 15. Univ. Alaska Sea Grant College Program AK-SG-98-01, Fairbanks, AK. Rebstock, G. A. 2001. Long-term stability of species composition in cala- noid copepods off southern California. Mar. Ecol. Prog. Ser. 215:213-224. Reilly, C. A., T. Wyllie Echeverria, and S. Ralston. 1992. Interannual variation and overlap in the diets of pelagic juvenile rockfish (genus:Se6as/cs) off central California. Fish. Bull. 90:505-515. Rosenfeld, L. K., F. B. Schwing, N. Garfield, and D. E. Tracy. 1994. Bifurcated flow from an upwelling center: a cold water source for Monterey Bay. Continental Shelf Res. 14:931-964. Shaffer, J. A., D. C. Doty, R. M. Buckley, and J. E. West. 1995. Crustacean community composition and trophic use of the drift vegetation habitat by juvenile split- nose rockfish, Sebastes diploproa. Mar. Ecol. Prog. Ser. 123:13-21. Stephens, J. S. Jr., P. A. Morris, K. Zerba. and M. Love. 1984. Factors affecting fish diversity on a temper- ate reef; the fish assemblage of Palos Verde Point, 1974-1981. Environ. Biol. Fish. 11:259-275. VenTresca, D. A., R. H. Parrish, J. L. Houk, M. L. Gingras, S. D. Short, N. L. Crane. 1995. El Nifio effects on the somatic and reproductive condition of blue rockfish, Sebastes mystinus. Calif. Coop. Oceanic Fish. Invest. Rep. 36:167-174. Yoklavich, M. M., V. J. Loeb, M. Nishimoto, and B. Daly. 1996. Nearshore assemblages of larval rockfishes and their physical environment off central California during an extended El Nifio event, 1991-1993. Fish. Bull. 94: 766-782. 49 Abstract — Skipjack (Katsuwonus pelamis), yellowfin (Thunnus alba- cares), and bigeye iThunnus obesus) tunas are caught by purse-seine vessels in the eastern Pacific Ocean (EPO). Although there is no evidence to indicate that current levels of fish- ing-induced mortality will affect the sustainability of skipjack or yellowfin tunas, fishing mortality on juvenile (younger than 5 years of age) bigeye tuna has increased, and overall fish- ing mortality is greater than that necessary to produce the maximum sustainable yield of this species. We investigated whether time-area clo- sures have the potential to reduce purse-seine bigeye catches with- out significantly reducing skipjack catches. Using catch and effort data for 1995-2002. we identified regions where the ratio of bigeye to skipjack tuna catches was high and applied simple closed-area models to investi- gate the possible benefits of time-area closures. We estimated that the most optimistic and operationally feasible 3-month closures, covering the equa- torial region of the EPO during the third quarter of the year, could reduce bigeye catches by ll.S^r, while reduc- ing skipjack tuna catches by 4.3%. Because this level of bigeye tuna catch reduction is insufficient to address sustainability concerns, and larger and longer closures would reduce catches of this species signficantly, we recommend that future research be directed toward gear technology solutions because these have been successful in many other fisheries. In particular, because over 50% of purse-seine catches of bigeye tuna are taken in sets in which bigeye tuna are the dominant species, methods to allow the determination of the species composition of aggregations around floating objects may be important. The potential use of time-area closures to reduce catches of bigeye tuna (Thunnus obesus) in the purse-seine fishery of the eastern Pacific Ocean Shelton J. Harley (contact author) Jenny M. Suter Inter-Amencan Tropical Tuna Commission 8604 La Jolla Shores Drive La Jolla, California, 92037-1508 Present address lor S. J. Harley; Ministry of Fisheries PC Box 1020 Wellington, New Zealand Email address for S. I. Harley: harleysSfish govt.nz Manuscript submitted 2 April 2004 to the Scientific Editor's Office. Manuscript approved for publication 5 April 2006 by the Scientific Editor. Fish. Bull. 105:49-61 (2007). The Inter-American Tropical Tuna Commission (lATTC) was established by an international convention in 1950 and is responsible for the conservation of tunas and management of fisheries for tunas and other species taken by tuna-fishing vessels in the eastern Pacific Ocean (EPO). Such conser- vation and management is accom- plished by measures imposed by the nations participating in the fishery in response to recommendations by the scientific staff of the lATTC. Cur- rently, the lATTC has adopted two measures to ensure the conservation of bigeye tuna in the EPO (lATTCM: catch limits for each longline fleet (based on their 2001 catch levels) and a series of closures for the purse-seine fleet. In this article, we examine the use of the temporary closure of a given area, referred to as a "time-area clo- sure," for management of the purse- seine fishery. Since the early 1990s, considerable purse-seine fishing effort in the EPO has been directed at tunas associated with floating objects, including man- made fish-aggregating devices (Lenne- rt-Cody and Hall, 2000). The predom- inant species captured are skipjack {Katsuwonus pelamis), bigeye {Thun- nus obesus), and yellowfin (Thunnus albacares) tunas. The floating-object (FOB) fishery has had no noticeable affect on skipjack tuna abundance (Maunder, 2002a) and little effect on yellowfin tuna because the catches of yellowfin tuna from the floating object fishery are small compared to the catches from other purse-seine fisheries (Maunder, 2002b). However, the FOB has led to a considerable in- crease in fishing mortality on juvenile bigeye tuna (Maunder and Harley, 2002; Harley et al., 2005). The most recent bigeye tuna stock assessment (lATTC, 2004) has in- dicated that overall fishing effort should be reduced by at least 38% to allow the stock to produce the maxi- mum sustainable yield (MSY). This assessment is based on a single EPO stock with no net migration between the eastern and western Pacific; how- ever, a "Pacific-wide" assessment has provided a very similar picture of low movement rates for bigeye tuna in the EPO (Hampton et al.^). Since the expansion of the FOB fishery, catches of bigeye tuna from the purse-seine fishery have exceed- ed those from the longline fishery in some years (Table 1). The bigeye tuna ' lATTC (Inter-American Tropical Tuna Commission). 2003. Resolution on the conservation of tuna in the eastern Pacific Ocean, 3 p. Resolution C-03-12, lATTC, 8604 La Jolla Shores Drive, La Jolla, California 92037. - Hampton, J., P. Kleiber, Y. Takeuchi, H. Kurota, and M. Maunder. 2003. Stock assessment of bigeye tuna in the western and central Pacific Ocean, with compari- sons to the entire Pacific Ocean, 81 p. SCTB16 BET-1. Sixteenth meeting of the standing committee on tuna and billfish, Mooloolaba, Queensland, Australia; 9-16 July 2003. 50 Fishery Bulletin 105(1) Table 1 Annual catches (metric tons) of b by set type (FOB=floating object ated with dolphins) and longline igeye (Thiinnus obesus) and skipjack (Katsuwonus pelamis) tuna from purse-seine (PS) fisheries associated school, UNA=tuna school unassociated with dolphins, and DOL=tuna school associ- fisheries from the eastern Pacific Ocean, east of 150°W, as used in the stock assessments. Year Bigeye tuna Skipjack tuna FOB UNA DOL PS total Longline total Annual total FOB UNA DOL Longline PS total total Annual total 1990 3360 1351 4711 98,990 103,700 34,980 35,788 867 71,635 42 71,677 1991 1963 1739 38 3740 104,159 107,874 37,655 22,958 786 61,399 33 61,432 1992 1154 4343 5497 84,396 89,893 45,556 35,333 869 81,758 24 81,782 1993 6274 4724 134 11,132 72,351 80,420 48,144 34,865 714 83,723 63 83,786 1994 37,901 2624 40,525 71,360 100,734 47,992 22,916 516 71,424 69 71,493 1995 45,204 6291 51,495 58,076 95,403 81,253 50,715 1032 133,000 74 133,074 1996 66,568 4280 70,848 46,771 98,124 74,260 34,635 729 109.624 40 109,664 1997 69,293 1868 48 71,209 52,078 103,693 123,002 29,510 6004 158,516 94 158,610 1998 43,226 5183 91 48,500 45,632 80,787 115,370 25,108 2879 143,357 65 143,422 1999 49,452 6574 56,026 32,565 73,176 178,824 84,036 1214 264,074 94 264,168 2000 83,489 3266 86,755 46,424 116,579 116,508 81,551 440 198,499 29 211,049 2001 56,753 1273 14 58,040 60,572 103,421 115,571 20,163 1218 136,952 61 144,949 2002 61,230 1166 62,396 68,195 103.394 118,485 32,471 2093 153,048 145 157,593 caught in the longhne fishery are larger (110-160 cm) and considerably more valuable than the smaller bigeye tuna (50-80 cm) caught mostly by the purse-seine fishery. Improving the long-term sustainability of the bigeye tuna fisheries could be achieved by reducing the fish- ing mortality of the smaller individuals that are caught predominantly in the FOB fishery. Annual catches of skipjack tuna from the purse-seine fishery in the EPO are larger and more economically important than those of bigeye tuna (Table 1). Furthermore, there are no concerns regarding sustainability of the skipjack tuna population in the EPO (Maunder, 2002a). Thus, bigeye tuna caught by the FOB fishery are essentially bycatch of the targeted skipjack tuna fishery; thus determining a mechanism by which the catches of bigeye tuna are reduced while minimizing losses in the catches of skip- jack tuna is an important management issue. Hall (1996) argued that to understand and solve by- catch problems it is important to classify the problem by a number of factors (e.g., time, space, and the level of control that fishermen have). With this information, there are many potential tools that can be used by fisheries managers to reduce fishing mortality, e.g., gear regulations, catch limits, closed seasons, and closed areas (Beverton and Holt, 1957). Time and area closures (time-area closures) are recommended as a means to reduce catches of sharks (Baum et al., 2003), protect billfishes from exploitation by the longline fish- ery (Goodyear, 1999), and protect biodiversity hotspots (Worm et al., 2003). Although time-area closures are not particularly appropriate for fisheries managed un- der quota systems, they may be beneficial for effort- managed fisheries (Horwood et al., 1998) or fisheries targeting multispecies (Hilborn et al., 2004), such as those for tunas in the EPO. In this study, we investigated the potential of time- area closures to reduce bigeye tuna catches while min- imizing impacts on the catches of skipjack tuna. In contrast to common closure-strategy studies, i.e., those studies devoted to fisheries targeting a single species, we investigated the potential impacts of time-area clo- sures on two species: a large and highly productive skipjack tuna stock, and a considerably smaller and less productive bigeye tuna stock. We used catch and effort data from the purse-seine fishery to search for potential time-area hotspots for bigeye catches and then applied simple "in-sample" closed-area models to predict the potential impact of closures of these areas. We discuss the likely use of such closures in the light of our findings, alternative management actions that could possibly reduce bigeye tuna catches, and finally, the strengths and weaknesses of the approach used for the closed-area models. Although yellowfin tuna form an important part of the purse-seine fishery in the EPO, where annual catches are greater than those for bigeye and skip- jack tuna combined, we did not consider them in our analysis. Within the EPO purse-seine fishery there are essentially two fleets: one targets yellowfin tuna schools associated with dolphins or schools not associ- ated dolphins and the other targets mainly skipjack tuna associated with floating objects. In our study, we focused on the second fleet and there are many reasons to believe that effort could not be transferred from one fleet to the other, e.g. markets, technological differences (the vessels require different equipment), geographical Harley and Suter; Potential use of time-area closures to reduce the catches of Thunnus obesus in eastern Pacific Ocean 51 e 50001 Floating-object sets Dolphin-associated sets Unassociated sets <=5t Hx\ o -30° 10° - 10° - 20° - o O G O O O o o o . .-- V ^ QOGOoGOGG^o© .-' o o o o o o o o O G o © 0:-^ °°ooooOOQQ° ° ooooOOOOOe» x»-o" 000000° eastern Pacific Ocean ■10° -20° 160' 150° 100° 90° 80° 140° 130° 120° 110° Figure 1 Average annual distribution of the purse-seine catches of bigeye tuna [Thunnus obesus), by set type and 5-degree latitude by 5-degree longitude area, in the eastern Pacific Ocean, 1995-2002. The size of the circles is proportional to the catch in each area. (the fisheries have limited spatial overlap), and restric- tions on dolphin mortality limits. In addition, only a small proportion (about 10%) of the purse-seine catches of yellowfin tuna are taken in floating-object sets. Later we discuss extensions to our analysis to include not only yellowfin tuna, but a range of bycatch species taken in the different purse-seine fisheries. Materials and methods Data We used set -by-set catch and effort data from purse-seine vessels that operate in the EPO. The majority of the data was obtained by scientific observers. In the absence of observer data, we used records from the logbooks of the vessels. Data were grouped by 5-degree latitude by 5- degree longitude areas (hereon referred to as 5°x5°areas) by seasonal quarter. The FOB fishery, which is respon- sible for over 90% of the purse-seine catches of bigeye tuna, was in an expansion phase during 1992-94; there- fore we restricted our attention to data for 1995-2002 (Table 1). Because very small amounts of bigeye and skipjack tuna are caught in dolphin-associated (DOL) sets, we excluded these from the analysis and instead focused on sets of tuna associated with floating objects and sets on schools not associated (UNA) with dolphins. For 1995-2002, these two set types were responsible for over 99% of bigeye and skipjack tuna catches from the purse-seine fishery (lATTC, 2004). These two set types were combined in the closed-area model because it was possible to switch effort between those two types of sets. The spatial distribution of catches by set type for bigeye and skipjack tuna are provided in Figures 1 and 2. Definition of "hotspot" In defining the spatial and temporal extent of the bigeye catches, we looked for areas where the ratio of bigeye to 52 Fishery Bulletin 105(1) 60° 100' L. 90° 80° 30° 20° 10° - 30° A— ^ 5000 t I I Floating-object sets ^^1 Dolptiln-associated sets Unassociated sets X <=5t O O o Q 9 x^ r i\ • • • (3 '' ^ G G GQGGGQ©© ® ''^' Q G G G G G G OOGGGGGQ Q( ^'0 o o O O G O G G G G G oooooooooG C y v^ eastern Pacific Ocean '■> *;^ • V ->v^ 5"- •30° ■20° ■10° -0° •30° 140° 130° 120° 110° 100° 90" 80° Figure 2 Average annual distribution of the purse-seine catches of skipjack tuna (Katsuwonus pelamis), by set type, in the eastern Pacific Ocean, 1995-2002. skipjack tuna catches was high, rather than just areas of high bigeye tuna catches, because we wanted areas where the losses in skipjack tuna catches would be mini- mized. We chose 5°x5° areas by quarter of the year as the scale for the individual hotspots. We defined an index for each time-area strata for each year. The index was a ratio of bigeye tuna catch to skipjack tuna catch that was robust to annual fluc- tuations in the abundance of either species. We then summed the annual indices over the time period to find areas that consistently resulted in high bigeye to skipjack tuna ratios. The indices were calculated sepa- rately for each year so that they were not dominated by data from years with exceptionally high or low catches of either species. The data used for this and the closed-area analysis were the following: B bigeye catch in quarter i in area/ in year t; skipjack catch in quarter /' in area^ in year t. We standardized catches within a year on the basis of the median catch of the year. s,^^ = S,^,/median(S ,). The location of hotspots did not differ noticeably if we standardized by the mean or total catch for each year, rather than the median. Using the standardized catches, we defined the an- nual index for a single 5°x5°area by quarter, 0,^, as e.„,. where 1 where f(b, j,,s, J,) otherwise b,,,=0 ».,.,.= (1) b, , , / s, , , - min(6, , , / s, , , ) where f(b, , ,s, ,. ,) = '-'■' '/■' .'■■'■' '-'/ (2) ■•'•' ''J-' max{b,j,/s,j,}-mm(b,j,/s,j,) Harley and Suter: Potential use of time-area closures to reduce the catches of Thunnus obesus in eastern Pacific Ocean 53 Note that the index is scaled to be between and 1, and the larger values within this range were associ- ated with greater bigeye-skipjack ratios. To obtain an overview of the hotspots over the 1995-2002 period we summed the annual indices (=2002 (=1995 ^j= I ^,r (3) We defined hotspots as those time-area regions where the summed index was in the top 20% of the values. Closed-area model The basic model is summarized in four steps: The new catch for each time-area closure was es- timated as the new effort multiplied by the original CPUE: "i.j,tlx,y ~ ^iJMx.y '^ i.j.t and '-'ij.tlx.y " '^i.jfix.: u i.J.t (8) (9) because it was assumed that CPUE in an area will not change when additional effort is added with closure. The summary statistic for each simulated closure was the percentage change in bigeye and skipjack tuna catches, compared to the catches observed in the ab- sence of a closure. 1 Choose an area to close in a given time period. 2 Re-allocate effort from the chosen area during the period of the closure to other areas in proportion to the effort in each area. Leave the effort outside the closure period unchanged. 3 Calculate the new catch of each species expected in each area based on the new effort and catch per unit of effort (CPUE) for each species in each area. 4 Compare new annual catches to original catches. The possible consequences of these assumptions and alternative modeling approaches are detailed in the discussion. The data used for the closed-area analysis were simi- lar to those used in the hotspot analysis. The definitions of catches remained the same (e.g., B, ^,1, although the spatial strata reflected by j differed, depending on the closure considered. We incorporated effort in terms of the number of sets, £, ,, and defined the CPUE of bigeye and skipjack tuna in tons per set, '■■■ "'.j.i and Uf ;j.t "'■j.i "ij,( (5) We allocated effort from a time-area closure (/=.<• and j=y) to the remaining areas in that time period on the basis of proportion of effort in each area (P,jt) (exclud- ing the closed area) e.g., E, ^'.j.i i,j,tit=x,j^y (6) For each time-area closure we determined the new effort allocation, E, ,,i,„, as E. ij.t\x,y ' E,,j,t+E^y, Pij^t/x.y ■'i.j.t where i = x andj = y where i = x andj ^ y. in. where i = x Z^i.j'^i.j.' (10) and AS. la,j^,.J.t\x.y Z^.j^l.j.t xlOO. (11) We repeated the calculations for the catch and effort data in each year (1995-2002) to consider the potential variability in the effect of a closure due to interannual variation in the spatial distribution of fish and fishing effort. In addition to the model described above, we also considered a "two set-type" model in which FOB and UNA sets were redistributed separately (i.e., we did not allow switching between set types). Although this model gave very similar results, it was probably less realistic; therefore the results are not presented here. Simulated closures We compared the performance of two closed areas for each quarter and year. The first closed area corresponded to the hotspots (those 5°x5° areas for a quarter for which 0^ I was in the top 20%) associated with each quarter. A closure of the hotspots should be optimal in the sense of reducing bigeye tuna catch with minimal impact on skipjack tuna catch but may not be practical from a management perspective because the 5°x5° areas are not continuous. The second closed area approximated the hotspot closure, but it was a more practical, continuous region. It extended from 5°N-10°S, to 90°-120°W. The total area of this closure was the same as the total area of the hotspot regions. We refer to this as the practical closure. In each case, effort during the closure period was redistributed between two areas, one north and one south of the equator, in proportion to the effort in each open area. Summaries of the effort and CPUE data, stratified by the areas that we used in the practical closure analysis, are provided in Table 2. 54 Fishery Bulletin 105(1) B Figure 3 Hotspots of bigeye tuna iThunnus obesus) catches in relation to skipjack tuna {Katsuwonus pelamis} catches by seasonal quarter and 5°x5'" region for 1995-2002 as measured by our hotspot index (Equations 1-3). Plots A-D represent quarters 1-4. Shades are used for different percentiles of the index, and darker regions indicate areas of higher bycatch rates. The black regions represent those time and area strata for which the hotspot index was in the top 20%. Results Hotspots The hotspots were not evenly spread over the year; the third seasonal quarter contained more 5°x5° hot- spots (24) than the other quarters (15-18 each) (Fig. 3). During quarters 1 and 4, most of the hotspots were located between 5°N and 10°S, whereas during quarters 2 and 3, the hotspots extended south to 15°S. Over all time-area strata, 90% of the hotspots were west of 90°W and east of 135°W, and over 95% were between 5°N and 15°S — indicating that the hotspots are found within a fairly restricted area. When we compared the hotspots to the practical clo- sure, 75% of the hotspots were found within the prac- Harley and Suter: Potential use of time-area closures to reduce the catches of Thunnus obesus in eastern Pacific Ocean 55 Quarter 3 D Quarter 4 Figure 3 (continued) tical closure area. Of the remaining 25'7f of hotspots, most were west of 120°W. Time-area simulations Over all years and quarters, the predicted decrease in bigeye tuna catches associated with the hotspot closure ranged from 2.8% to 23.7%, whereas the change in skipjack tuna catches ranged from a 0.9% increase to a 14.1% reduction (Fig. 4). The greatest reductions in bigeye tuna catch were associated with second- and third-quarter closures (mean reduction = 14.6%). The mean reductions in skipjack tuna catches did not vary much across quarters (means ranged from 2.8% to 3.7%). For several years, there was little or no predicted reduc- tion in skipjack tuna catch associated with a hotspot closure. Based on the median of the ratios, the greatest contrast between bigeye and skipjack tuna catch reduc- tions was associated with a third quarter closure; the average percentage reduction in bigeye tuna catch was 14.6%, versus 2.8% for skipjack tuna. The performance of second-quarter closures was similar to that of third- quarter closures, but the former was much more variable across years. The performance of the practical closure was generally similar to that of the hotspot closure. Over all years and 56 Fishery Bulletin 105(1) Quarter 1 Quarter 3 Skipjack 10 - S B S S B B B S S 10 - B B g B B 20 - 30 - -I 1 1 1 1 r 1 997 1 999 2001 Year Quarter 2 10 - s S S S S S 10 - B B B S S B 20 - B B B B -30 - 1999 Year 10 - S S S S S a s 10 - B B B B B B 20 - B B 30 - T 1 1 1 1 1 1 r 1996 1997 1999 2001 Year Quarter 4 10 - R S B i s B 10 - S B B B S B B 20 - 30 - 1999 Year BsSQ "1 1 r~ 2 3 4 Quarter Bigeye Figure 4 Predicted changes in annual purse-seine catches of bigeye iThuninis obesus) and skipjack {Katsuwonus pelamis) tuna associated with a closure of the hotspot areas from Figure 1 for each quarter of the year. The left and middle panels indicate the change in annual catch for bigeye (B) and skipjack (S) tuna estimated to occur in each year of the 1995-2002 period if that area was closed in the 1st, 2nd, 3rd, and 4th quarter, respectively. The plots on the right summarize the predicted changes in bigeye and skipjack tuna catch over the eight years for a closure in each seasonal quarter. The white bar indicates the median change, the dark rectangle indicates the interquartile range (25"'-75th percentiles), and the outer lines indicate the extremes. The horizontal dashed line represents zero (i.e., no change in catch). Table 2 Summary of annual purse-seine effort and catch per unit of effort for bigeye [Thunnus obesus) and skipjack (Katsuwonus pelamis) tuna by set type (FOB=tuna school associated with floating object! s); UNA=tuna school not unassociated with floating object(s)) for the three areas modeled in the "practical" closure. A "practical" was a more practical, continuous region. It extended from 5°N-10°S, to 90°-120°W. The total area of this closure was the same as the total area of the hotspot regions. The means and standard deviations (SD) were calculated from annual values for 1995-2002. Number of sets Bigeye catch per set Skipjack catch per set Area FOB UNA FOB UNA FOB UNA Practice 1 closure Mean 532 229 14.75 SD 228 295 7 North Mean 474 882 4.4 SD 301 517 2.53 South Mean 404 296 10.43 SD 377 278 6.35 0.9 21.64 9.97 1.27 9.9 8.61 0.1 19.42 5.65 0.12 10.19 4.17 0.44 17.43 5.48 0.53 7.4 5.14 Harley and Suter: Potential use of time-area closures to reduce the catches of Thunnus obesus in eastern Pacific Ocean 57 Quarter 1 Quarter 3 1997 1999 Year Quarter 2 ~i 1 1 1 1 1 1 r^ 1995 1997 1999 2001 Year n 1 1 1 1 1 1 r 1995 1997 1999 2001 Year Quarter 4 -I 1 \ 1 1 1 1 r 1995 1997 1999 2001 Year Skipjack 10 - —I— — ' — , 10 - 20 - 30 - • 2 3 Quarter Bigeye 10 - - , ' 1 , ' 10 - ! '' J- 30 - -^ 1 2 3 Quarter Figure 5 Predicted changes in annual purse-seine catches of bigeye {Thunnus obesus) and skipjack tKatsuwonus pelamis) tuna associated with a closure of the area between 5'N-IO^S and 90'-120^"W for each quarter of the year. See Figure 4 for further description of the panels. quarters, the predicted reductions in bigeye tuna catches associated with the practical closure ranged from 0.3% to 24.5% (777 metric tons [t] to 20,206 t), and the change in skipjack tuna catches ranged from a 1.1% increase to a 17.0% reduction (1204 t to 32,773 t) (Fig. 5). The extreme values of skipjack tuna catch were associated with first- quarter closures. As with the hotspot closure, the great- est reductions in bigeye tuna catch were associated with second- and third-quarter closures (average reductions of 13.4% and 11.5%, respectively, across years) and again the mean reduction in skipjack tuna catches did not vary greatly across quarters (mean reductions ranging from 3.8% to 4.9%). Based on the median of the ratios, the greatest contrast between bigeye and skipjack tuna reductions was associated with a second-quarter closure; the average percentage reduction in bigeye tuna catches was 13.4%, versus 4.9% for skipjack tuna catches. Overall, the hotspot closure predicted slightly greater reductions in bigeye tuna catches and slightly lesser reductions in skipjack tuna catches than did the prac- tical closure, but the difference in the median of the ratios (5.0 times for the hotspot closure and 3.8 times for the practical closure) is probably not significant. Results for both closures may indicate that a closure during the second or third quarters is optimal. Because the predicted variability in performance was less for a third-quarter closure than a second-quarter closure (in both analyses), the former was preferred as a manage- ment tool. Discussion Time-area closures are one of a number of fisheries man- agement options (Hilborn et al., 2004). In our study we investigated, using simulations that use historical catch and effort data, whether time-area closures could be a useful tool to reduce bigeye tuna catches in the purse- seine fishery without leading to large reductions in the catches of skipjack tuna. 58 Fishery Bulletin 105(1) In the remainder of this article we discuss our find- ings in terms of the recent stock assessment recommen- dations for reductions in fishing effort — more specifi- cally whether reductions predicted in our study would be sufficient to reach management objectives. We also discuss alternative measures for reducing bigeye tuna catches in the purse-seine fisheries of the EPO and de- scribe potential improvements for our time-area closure modeling approach that may lead to a more accurate analysis of the likely performance of closures that may be considered in the future. Predicting performance of time-area closures Following Hall (1996), we looked for time-area strata in which there were high bigeye tuna to skipjack tuna ratios. These areas were relatively confined geograph- ically and did not vary greatly by quarter. For this reason, the hotspot and practical closures predicted similar results. Simulation of a practical closure (and one that able to be implemented) indicated that moderate average re- ductions in bigeye tuna catch (11.5%) could be achieved with lesser average reductions in skipjack tuna catches (4.9%). When we considered these reductions in terms of total catch by weight, the annual bigeye tuna catch reductions ranged up to 20,206 t (average 5722 t) and up to 32,773 t (average 6,807 t) for skipjack tuna. Based on the current mix of fishing gears in the bigeye tuna fisheries in the EPO, and the estimated maximum sustainable yield (MSY) of about 77,000 t (lATTC, 2004), the purse-seine share of the MSY was around 40,000 t (S. J. Harley, unpubl. data). Consider- ing current purse-seine catches of over 60,000 t, and the 11.5% reduction predicted for the practical closure, we believe that these closures alone are unlikely to yield the required reductions in bigeye tuna catches from the purse-seine fishery. The closures investigated in our study were based on strata where the ratio of bigeye tuna to skipjack tuna Table 3 Proportion of the annual floating object fishery catches of bigeye ( Th innus obesus) and skipjack iKatsuwonus pelamis) | tuna that are caught in sets with or without other species. Year Bigeye tuna Skipjack tuna Without With Without With 1995 0.05 0.95 0.31 0.69 1996 0.06 0.94 0.25 0.75 1997 0.05 0.95 0.25 0.75 1998 0.04 0.96 0.27 0.73 1999 0.02 0.98 0..39 0.61 2000 0.08 0.92 0.38 0.62 2001 0.11 0.89 0.44 0.56 2002 0.08 0.92 0.29 0.71 Average 0.06 0.94 0.32 0.68 catches was the greatest. For these closures, the reduc- tion in catches (in metric tons) is about the same for bigeye and skipjack tuna, but if a closure is larger or longer, the losses in skipjack catches would quickly out- weigh the reductions in bigeye tuna catches. Therefore, although we did not examine larger or longer closures in our study, it is unlikely that these closures could lead to the necessary reductions in bigeye tuna catches without unacceptable losses in skipjack tuna catches. The lack of effectiveness of the time-area closures is related to the extent of the interaction between skip- jack and bigeye tunas. For the 1995-2002 period, 94% of the bigeye tuna caught by purse-seiners was taken in sets that also caught skipjack tuna (Table 3). This percentage is greater than the proportion of skipjack tuna catch that was taken in association with bigeye tuna (68%). Given this fact, it is not surprising that time-area closures are insufficient. Management alternatives to reduce catches of bigeye tuna We have shown that time-area closures alone are unlikely to result in the necessary reductions in fishing mortality for bigeye tuna; therefore alternative or supplementary management actions would be appropriate. In many instances, studies of fish behavior (Wardle, 1983) and gear technology (Larsen and Isaksen, 1993) have led to changes in gear configurations and deployment, result- ing in significant reductions of catches of unwanted species. A good example of this type of change is the reduction of dolphin catch from tuna-dolphin aggrega- tions in the EPO (NRC, 1997). In the 1970s, many thousands of dolphins (mostly Stenella sp. and Delphinus sp.) were caught and killed by purse-seine vessels that set on dolphins in order to catch the yellowfin tuna that were associated with them (NRC, 1997). Through the introduction of fine-mesh net panels, use of a "back-down" procedure, and the avoid- ance of areas where oceanographic conditions could lead to net collapse, this mortality was reduced dramatically by the 1990s (NRC, 1997). It is also possible to exploit behavioral differences among fish species. Through examination of the differ- ential behavior of cod (Gadus morhua) and haddock (Me- lanogrammus aeglefinus], it was found that it was pos- sible to configure bottom trawl nets to catch the target species and allow the other species to escape through larger meshes (Cotter et al., 1997). Sorting grids have also been used to allow the escape of unwanted species (Larsen and Isaksen, 1993; Misund and Beltestad^; lATTC^). Unless studies of bigeye and skipjack tuna be- 3 Misund, O. A., and A. K. Beltestad. 1994. Size-selection of mackerel and saithe in purse seine. International Council for the Exploration of the Sea Council Meeting, 1994/B:28. ^ lATTC (Inter-American Tropical Tuna Commission). 1999. Report of the bycatch working group, 25 p. 63rd Meeting of the lATTC; June 8-10, 1999. lATTC, 8604 La JoUa Shore Drive, La Jolla, California 92037. Harley and Suter: Potential use of time-area closures to reduce the catches of Thunnus obesus in eastern Pacific Ocean 59 havior determine a mechanism by which bigeye, but not skipjack tuna, can escape through a sorting grid in a purse-seine net, sorting grids are more likely to be useful for overall reductions in catches of small tunas than as a mechanism for reducing bigeye tuna catches without re- ducing skipjack tuna catches. Lennert-Cody and Hall (2000) used a range of statistical models to determine factors (e.g., area, season, characteris- tics of the floating object and the purse- seine net) that were associated with high- er catches of bigeye and skipjack tuna. Unfortunately, many factors were con- founded because the fishing practices of the fleet often differ in time and space, making it difficult to determine which gear characteristics may be important. Thus, it appears unlikely that analysis of fishery-collected data will lead to techni- cal measures with the potential to reduce catches of bigeye tuna. Although it may be difficult to deter- mine important factors relating to bigeye tuna catch rates, fisheries data can be used to examine the nature of the catch- es of this species. For example, we found that 94% of bigeye tuna are caught in sets that also caught skipjack tuna. We were interested in how the bigeye tuna catches were distributed; were they predominantly from a small number of sets with high catches or from a large number of sets with small catches? Our analysis of this question, based on data for 1995-2002, is presented in Figure 6. It shows than only 57c of bigeye tuna were caught in single-species sets, but that about 50% of bigeye tuna came from sets that contained at least 60% of this species. These sets are responsible for only 7% of the skipjack tuna catch from the floating-object fishery and a smaller proportion of the overall skipjack catch given that about 30% of skipjack tuna catch is still taken from schools unas- sociated with dolphins (lATTC, 2004). The analysis of the catch composition of purse-seine sets described above indicates that if fishing captains can determine, at least roughly, the species composition of an aggregation prior to setting (i.e., which species is dominant), large reductions in bigeye tuna catches could be achieved by not setting on bigeye-tuna-dominated aggregations. Such a measure would have little impact on overall skipjack tuna catches and would not require the fleet to be restricted in its activity by time-area closures. Schaefer and Fuller (2005) used a range of electronic tags, supplemented with sonar images of fish aggrega- tions around floating objects, to describe differences in the behavior of skipjack and bigeye tunas around float- ing objects. Exploitation of these differences, combined with the potential ability of fisherman to identify large aggregations of bigeye tuna around floating objects, 1.0- *—____ w ' ^^^ OJ ' '^ , Bioeve ig-ob|ect s o CD \^ i^ . Skipiack ra x. o 0.6- *\ E X o i \, \ .c \ a 0.4- \ o o "' ^. 1 0.2- o o. '■■•■-.. X o ~ ~ A - ^^ 0.0- --'-- — --.---1 1 1 1 1 ' 0.0 2 4 6 8 10 Proportion of bigeye in the set Figure 6 Proportion of bigeye {Thunnus obesus) and skipjack (Katsuwomus pela- mis) tuna taken in sets where at least a given proportion of skipjack tuna were taken, for 1995-2002. For example, just over 60% of bigeye tuna came from sets where 50% or more of the tuna in the set was bigeye tuna, while only 15% of skipjack tuna were taken in these bigeye tuna dominated sets. may lead to the development of fishing practices that can reduce bigeye tuna catches with minimal impact on skipjack tuna catches. Critical to this approach will be the establishment of incentives 1) to encourage both the identification of the schools and 2) not to set on bigeye-tuna-dominated aggregations. Similar to the dolphin mortality limits currently applied by the lATTC, it could also be possible to have individual vessel limits for bigeye tuna and let fishermen determine how best to modify their fishing operations in order to achieve a given limit. Harley et al.^ used historical catch-by-vessel data and found that indi- vidual vessel limits of about 350 t would be sufficient to reduce purse-seine catches of bigeye tuna in the EPO by 50% in most years. Independent studies of fish behavior, coupled with experimental work investigating modifica- tions in fishing practices and gear, could be fruitful. Modeling potential effects of time-area closures We applied simple closed-area models that used his- torical catch and effort data. Several assumptions are implicit in these models. First, we assumed that the fishing fleet has the flexibility to reallocate effort out- Harley, S. J., P. K. Tomlinson, and J. M. Suter. 2004. Pos- sible utility of catch limits for individual purse-seine vessels to reduce fishing mortality on bigeye tuna in the eastern Pacific Ocean, 8 p. Inter-American Tropical Tuna Commis- sion, 5'*" working group on stock assessments, 11-13 May 2004, Document SAR-5-05 BET A. lATTC, 8604 La Jolla Shore Drive, La Jolla, California 92037. 60 Fishery Bulletin 105(1) side the closed area. We redistributed effort during the closure to other areas in proportion to historical effort within the same seasonal quarter. Previous studies have chosen not to redistribute effort (Goodyear, 1999), to redistribute effort in proportion to target catch (Worm et al., 2003), or to redistribute effort with the assumption that catch, rather than effort, is a limiting factor (Baum et al., 2003). A good understanding of fleet dynamics is necessary to determine appropriate models for effort redistribution. Second, we assumed that redistributed effort would yield the same CPUE as previous effort in the area. Redistributed effort assumes that CPUE will remain unchanged when more fish are removed. It is likely that CPUE would decline with abundance as a result of in- creased effort, therefore it is possible that our analysis overestimates the catches during the closure. Similarly, our model assumes that CPUE is constant within each area, i.e. regardless of where one fishes within the area, one achieves the same CPUE. In reality, it is possible that fishermen could fish close to the edge of the closed area and potentially undermine the effectiveness of a closure. Related to these first two points is the case of switch- ing between fishing modes. By grouping FOB and UNA sets in our model, we allowed for switching between set types when fishing outside the closed area. Harley et al."* showed that the purse-seine vessels that catch the majority of the bigeye tuna, fish almost exclusively on floating-objects (over 90% of the sets). Even with this information, we still believe that the implicit assump- tion of grouping the two set types is acceptable. We did not consider dolphin-associated sets (that catch almost exclusively yellowfin tuna). We consider it much less likely that effort would be shifted towards dolphin-as- sociated schools for several factors, including politics, market pressure, technological and gear differences, and the inexperience that many skippers who partici- pate in the FOB fishery would have with this alterna- tive mode of fishing. Finally, we implicitly assumed in our model that fish not caught as a result of the closure could not be caught later in the year. This assumption could lead us to underestimate catches outside of the closure. Thus, we have two potential biases in opposite directions that could affect our conclusions. The best way to quan- tify these biases would involve a model that integrated population and fisheries dynamics. A dynamic approach to modeling closed areas could take into account the abundance of fish in different areas and the movement of fish between areas dur- ing the year. Modeling the relationship between effort and catches in different areas should include account- ing for abundance (e.g., through the use of the catch equation). Tagging data are necessary to estimate stock param- eters, such as residence times within a closed area and fish movement rates between the open and closed areas. In addition to conventional tagging data, information from electronic tagging of bigeye tuna (Schaefer and Fuller, 2002) could provide a basis for describing move- ment by means of simple movement models (e.g., those of Adam et al. [2003]). Because the vessels catch bigeye and skipjack tunas together, the model must include the movement patterns of both species. This approach is extremely data demanding, and many of the data for this approach are not yet avail- able. Notwithstanding these problems, future analysis of time-area closures should include consideration of im- portant biological factors such as those described above, as well as socioeconomic data that may be important for predicting fleet dynamics. Another extension of the modeling approach in our study is to consider additional target and bycatch spe- cies. Worm et al. (2003) considered bycatch from the United States swordfish and tuna longline fisheries in the Atlantic when modeling closed areas. With this approach it would be useful to include not only yel- lowfin tuna and dolphin sets in the model, but also the bycatch species that are taken in the different areas and fisheries. Conclusions Time-area closures are one of the many management actions available for the regulation of fisheries. Because of the strong interactions between bigeye and skipjack tunas, we have shown that time-area closures alone are unlikely to be sufficient to address concerns regard- ing the sustainability of bigeye tuna because it may not be possible to achieve the necessary reductions in bigeye tuna catches without large losses in skipjack tuna catches. We suggest that it will be important to investi- gate aspects offish behavior to determine measures that could be used either in conjunction with, or instead of, closures to help reduce mortality on juvenile bigeye tuna while sustaining the important skipjack fishery. Acknowledgments We thank R. Allen, W. Bayliff, R. Deriso, and three anon- ymous reviewers for comments on this manuscript, and M. Hall, C. Lennert-Cody, M. Maunder, and K. Schaefer for useful discussions of closed areas and management options for bigeye tuna. Literature cited Adam, M. S., J. Sibert, and D. Itano. 2003. Dynamics of bigeye [Thunnus obesus) and yel- lowfin (T. albacares) tuna in Hawaii's pelagic fisheries: analysis of tagging data with a bulk transfer model incorporating size-specific attrition. Fish. Bull. 101: 215-228. Baum, J. K., D. Kehler, R. A. Myers, B. Worm, S. J. Harley, and P. A. Doherty. 2003. Collapse and conservation of shark populations in the Northwest Atlantic. Science 299:389-392. Harley and Suter: Potential use of time-area closures to reduce the catches of Thunnus obesus in eastern Pacific Ocean 61 Beverton, R. J., and S. J. Holt. 1957. On the dynamics of exploited fish populations. Fish. Invest. London 19:1-533. Cotter, A. J. R., T. W. Boon, and C. G. Brown. 1997. Statistical aspects of trials of a separator trawl using a twin rig trawler. Fish. Res. 29:25-32. Goodyear, C. P. 1999. An analysis of the possible utility of time-area closures to minimize billfish bycatch by U.S. pelagic longlines. Fish. Bull. 97:243-255. Hall, M. A. 1996. On bycatches. Rev. Fish Biol. Fisheries 6:319- 352. Harley, S. J., M. N. Maunder, and R. B. Deriso. 2005. Assessment of bigeye tuna {Thunnus obesus) in the eastern Pacific Ocean. Col. Vol. Sci. Pap. ICCAT, 57(2):218-241. Hilborn, R., K. Stokes, J. J. Maguire, T. Smith, L. W. Botsford, M. Mangel, J. Orensanz, A. Parma, J. Rice, J. Bell, K. L. Cochrane, S. Garcia, S. J. Hall, G. P. Kirkwood, K. Sainsbury, G. Stefansson, and C. Walters. 2004. When can marine reserves improve fisheries management? Ocean Coast. Manag. 47:197-205. Horwood, J. W., J. H. Nichols, and S. Milligan. 1998. Evaluation of closed areas for fish stock con- servation. J. Appl. Ecol. 35:893-903. lATTC (Inter-American Tropical Tuna Commission). 2004. Tunas and billfishes in the eastern Pacific Ocean in 2003. Fishery Status Report 2, 114 p. lATTC, La Jolla, California. Larsen, R. B., and B. Isaksen. 1993. Size selectivity of rigid sorting grids in bottom trawls for Atlantic cod (Gadus morhua) and haddock [Melanogrammus aeglefinus). ICES Mar. Sci. Symp. 196: 178-182. Lennert-Cody, C. E., and M. A. Hall. 2000. The development of the purse seine fishery on drifting fish aggregating devices in the eastern Pacific Ocean: 1992-1998. In Tuna fishing and fish aggregating devices, p. 78-107. [Title translated into English. The paper is in English, but the book is in French], IFREMER (French Research Institute for Exploitation of the Sea), Issy-les-Moulineaux Cedex, France. Maunder, M. N. 2002a. Status of skipjack tuna in the eastern Pacific Ocean in 2001 and outlook for 2002. In Stock assess- ment report 3: status of the tuna and billfish stocks in 2001, p. 135-200. lATTC, La Jolla, CA. 2002b. Status of yellowfin tuna in the eastern Pacific Ocean in 2001 and outlook for 2002. In Stock assess- ment report 3: status of the tuna and billfish stocks in 2001, p. 47-134. lATTC, La Jolla, CA. Maunder, M. N., and S. J. Harley. 2002. Status of bigeye tuna in the eastern Pacific Ocean in 2001 and outlook for 2002. In Stock assessment report 3: status of the tuna and billfish stocks in 2001, p. 201-311. lATTC, La Jolla, CA. NRC (National Research Council). 1997. Dolphins and the tuna industry, 176 p. National Academy Press, Washington D.C. Schaefer, K. M., and D. W. Fuller. 2002. Movements, behavior, and habitat selection of bigeye tuna (Thunnus obesus) in the eastern equato- rial Pacific, ascertained through archival tags. Fish. Bull. 100:765-788. Schaefer, K. M., and D. W. Fuller. 2005. Behavior of bigeye {Thunnus obesus) and skipjack iKatsuwonus pelamis) tunas within aggregations asso- ciated with floating objects in the equatorial eastern Pacific. Mar. Biol. 146:781-792. Wardle, C. S. 1983. Fish reactions to towed fishing gears. In Experi- mental biology at sea (A. MacDonald, and I. G. Priede, eds.), p. 167-195. Academic Press, New York, NY. Worm, B., H. K. Lotze, and R. A. Myers. 2003. Predator diversity hotspots in the blue ocean. Proc. Natl. Acad. Sci. USA 100:9884-9888. 62 Abstract — Oceanic incidence and spawning frequency of Chesapeake Bay striped bass iMorone saxatilis) were estimated by using microchemi- cal analysis of strontium in otoliths. Otoliths from 40 males and 82 females sampled from Maryland's portion of the Chesapeake Bay were analyzed for seasonal and age-specific patterns in strontium and calcium levels. The pro- portion of oceanic females increased from 509c to 75% between ages seven to 13; the proportion of oceanic males increased from 20% to -50% between ages four to 13. Contrary to an earlier model of Chesapeake Bay striped bass migration, results indicated that a substantial number of males under- took oceanic migrations. Further, we observed no mass emigration of females from three to four years of age from the Chesapeake Bay. Seasonal patterns of estuarine habitat use were consistent with annual spawning runs by striped bass of mature age classes, but with noteworthy exceptions for newly mature females. Evidence of an early oceanic presence indicated that Chesapeake Bay yearlings move into coastal regions — a pattern observed also for Hudson River striped bass. Otolith microchemical analyses revealed two types of behaviors (estua- rine and oceanic) that confirm migra- tory behaviors recently determined for other populations of striped bass and diadromous species (e.g., Ameri- can eels [Anguilla rostrata] American shad [Alosa sapidissima] and white perch [Morone Americana]). Oceanic migration rates of Upper Chesapeake Bay striped bass (Morone saxatilis), determined by otolith microchemical analysis* David H. Secor (contact author)^ Philip M. Piccoii^ Email for D H Secor: secori@cbl.umces.edu ' Chesapeake Biological Laboratory University of Maryland Center for Environmental Science Solomons, Maryland 20688 ^ University of Maryland Department of Geology College Park, Maryland 20742 Manuscript submitted '23 July 2004 to the Scientific Editor's Office. Manuscript approved for publication 17 April 2006 by the Scientific Editor. Fish. Bull 105:62-73 (2007). As an estuarine-dependent species, striped bass (Morone saxatilis) dem- onstrate large plasticity in migration patterns (Secor and Piccoli, 1996). Striped bass in Chesapeake Bay are partial migrants; only a fraction of individuals will leave estuarine habi- tats for oceanic waters (Kohlenstein, 1981). Never the less, Chesapeake Bay striped bass are the major con- tributors to interjurisdictional ocean fisheries (Merriman, 1941; Wirgin et al., 1993). Rates of contributions by Chesapeake Bay striped bass to those fisheries are determined by lifetime patterns of habitat use, ontogenetic rates of egress from Chesapeake Bay, and regional rates of exploita- tion and natural mortality (Dorazio et al., 1994). Rates of oceanic resi- dence have been shown to vary by sex and increase with age. Analyzing striped bass tagged in the Potomac River, Kohlenstein (1981) advanced the working hypothesis that young striped bass remain in or near the tributary in which they were spawned for two or three years. At this point a substantial proportion of immature females (ca. 50%) emigrate from the Bay, remaining in ocean waters until sexually mature (age 5-7 years). In contrast, males are mature by age 2 but remain in the Bay throughout their lives. This hypothesis remains untested, although substantial devi- ation from this proposed pattern is indicated by tagging studies and catch records. From recaptured striped bass tagged on Chesapeake Bay spawning grounds, Dorazio et al. (1994) esti- mated that by 800 mm total length (TL), approximately half of the popu- lation (males and females combined) used ocean habitats. This length would correspond to an age of sev- en to 10 years (Secor et al., 1995b). Tagging studies comprise quasilon- gitudinal analyses, which could pro- vide estimates of age-specific egress rates if tagged and recaptured fish are representative of the population. However, striped bass are moderately long lived and show migration behav- iors that vary substantially with sex and age. Therefore, tagging studies often do not comprise sufficient spa- tial and temporal scales to provide the precise information needed to pre- dict how Chesapeake Bay striped bass contribute to coastal fisheries. Striped bass longevity exceeds 30 years (Merriman, 1941; Secor et al., 1995b). Life table analysis has indicated that maximum reproduc- tive rate occurs relatively late in life (10-12 years) (Secor, 2002) and that accumulation of adult biomass (repro- ductive potential) represents an im- portant "storage mechanism" (Warner and Chesson, 1995), improving the ' Contribution 4041 of the University of Maryland Center for Environmental Sci- ence, Chesapeake Biological Laboratory, Solomons, Maryland. Secor and Piccoli: Oceanic migration rates of Morone saxatilis. determined by otolith microchemlcal analysis 63 odds of recruitment over the lifetime of a fish (Secor. 2000a, 2000b, in press). Life-table-based models (e.g.. Goodyear, 1984) depend upon the assumption that an- nual spawning occurs, which remains unsubstantiated for this species. If spawning frequency declines with age, for instance, then generation time and age-at- maximum reproductive value will be substantially over- estimated, which in turn will affect biological reference points (Marshall et al., 2003). Electron probe micro-analysis (EPMA) of Sr has been developed as a method to reconstruct individual patterns of migration and habitat use by anadromous populations of striped bass (Secor, 1992). In estuarine environments, strontium is often a reliable tracer of salinity; higher marine concentrations (7 ppm) become diluted in estuarine environments by freshwater inputs when freshwater Sr:Ca end-members (where "end-mem- bers" are the source of a Sr:Ca ratio) are low (Ingram and Sloan, 1992). Kraus and Secor (2004a) surveyed available data and determined that 83% of estuaries have low freshwater end-members, indicating that the ratio of Sr to Ca should be positively (but not neces- sarily linearly) related to salinity in most estuaries. In our study we applied EPMA to examine the frac- tion of Chesapeaka Bay striped bass that migrate to ocean waters and the frequency at which females and males undertake spawning runs. We previously used this method to chart age- and sex-specific patterns of Hudson River striped bass (Secor and Piccoli, 1996; Zlo- kovitz and Secor, 1999; Secor et al., 2001). Expectations for ontogenetic rates of emigration (i.e., Kohlenstein, 1981) and annual spawning were tested. In addition, we sought evidence for contingent groups ( subpopulation groups with similar lifetime migration patterns; Secor, 1999), which we observed previously in Hudson River striped bass. Material and methods Samples Samples collected during spawning runs present the best opportunity to collect a representative sample of mixed age classes, sexes, and migratory behaviors. These sam- ples comprise mostly those ages that have fully recruited to the mature population. We note in our study that this sample incompletely represents migratory behav- iors for those females that have not yet become mature and are not participating in the years spawning run. During the period 15 April-30 May 2000, we obtained samples of 247 male and 122 female striped bass from the upper Chesapeake Bay (N. of 39°00'; n=21), mid- Bay (N. of 37='53WS. of 39°00'; r?=76); Choptank River (/! = 199), Patuxent River (« = 33), and Potomac River (71 = 28). Capture methods were diverse and included the use of gill- and pound-nets (Maryland Department of Natural Resources monitoring), electro-shocking (National Marine Fisheries Service Northeast Center and University of Maryland scientific collections), and charter boat angling. All fish were measured (fork length [FL] and weight [g]), sex and diet were determined, and otoliths and scales were collected. Fork lengths ranged from 685 to 1110 mm for females and from 320 to 1029 mm for males. Otolith Sr:Ca measures To conduct EPMA analyses, otoliths (sagittae) were extracted, soaked in \% sodium hypochlorite solution, rinsed with deionized water, and embedded within a resin (Secor et al., 1992). Transverse sections, approxi- mately 1 mm thick, were cut through the otolith cores with a metallurgical wafering saw. The sections were mounted on glass slides, polished on wetted 600-grain sandpaper, and polished again on a slurry of 0.3-^m alumina until their surfaces were free of pits and abrasions, which can cause artifacts in microprobe analysis (Kalish, 1990). Annuli were enumerated based upon standard criteria under optical microscopy (Secor et al., 1995b). Before analysis, otoliths were cleaned ultrasonically and carbon-coated in a high-vacuum evaporator. X-ray intensities for Sr and Ca were quantified by using a JEOL 8900 electron probe microanalyzer (Cen- ter for Microscopy and Microanalysis, Univ. Maryland, College Park, MB). Calcite (CaCOg) and strontianite (SrCOg) were used as reference standards and the protocol was checked by using secondary standards containing both Ca and Sr. The details of this analy- sis can be found elsewhere (Secor and Piccoli, 1996). Detection limits for Sr were approximately 230 ppm (±2 standard errors). Four slides, each containing four otolith sections, were loaded into the specimen chamber of the microanalyzer. After initial calibration to Sr and Ca standards (at programmed settings and intervals), transect assignments were made for up to 16 otolith sections. Transects comprised a series of point mea- surements from young to old ages across the sectioned otolith. X-ray maps of otolith structure were collected by using wavelength spectrometers. Sr was expressed as a ratio of Ca (Sr:Ca) because of expected competitive interactions between the isotopic species (Kraus and Secor, 2004a). Further Sr:Ca records were converted to salinity exposure profiles according to the model (Secor et al., 1995a): ''Salinity inhabitance" (sic) (psu) 40.3 (l-h56.3 e -1523(Sr:Cal )-l. r2=0.94 ; n = 54 where "salinity inhabitance" (sic) is the salinity level (practical salinity units, psu) in the otoliths for the period of time represented for each Sr:Ca datum. Oceanic incidence of striped bass For our analysis, a subsample of 122 fish (40 males and 82 females) was drawn from the upper Bay (n = lQ], mid-Bay (/?=46), and Choptank River (?i = 66). Esti- mated salinity records for the last year of life (recent 64 Fishery Bulletin 105(1) habitat use) were determined from at least five point measurements taken across the last completely-formed annulus (i.e., the last full year of life prior to winter). Opaque zone formation on the otolith occurs just prior to the spawning season (Zlokovitz et al., 2003); therefore measurements were taken between the penultimate and most peripheral (recently formed) opaque zones. We selected the record of maximum salinity, because at least one point in this series can be influenced by the previous year's spawning run. Oceanic habitat use was defined as salinities >29, and individual fish were classified accordingly. Because of unequal sampling among ages, we analyzed four age classes by sex: 4-6, 7-9, 10-12, and 13-18 years of age. Two-way classifi- cation tables were constructed to evaluate differences between age classes and sexes in probability of recent oceanic residence. Life history transects and spawning frequency Life history transects of salinity exposure, a series of Sr:Ca ratios from the juvenile period to the end of life, were constructed from EPMA measurements from 30 female and 10 male striped bass. This subsample was drawn from the upper Bay (n=2), mid-Bay {n=27), and Choptank River («=11). To weight seasonal data among ages, time series were selected so that four or five analyzed points were included for each annulus. Data were standardized (Z score= {transect datum - tran- sect mean)ltransect standard deviation) and plotted to examine variations about the transect mean (Sokol and Rohlf, 1981). Time-series data represented by the life history tran- sects were expected to show autocorrelation across sea- sonal points and ages. An appropriate method of data analysis that shows interdependence among repeated measures on the same individual is repeated measures multivariate analysis (RM-MANOVA) (Chambers and Miller, 1995). This analysis simultaneously fits several dependent variables to independent factors of interest (SAS, Statistical Analysis System, SAS Institute, Inc., Gary, NC) and evaluates the matrix equation, S, = Sex/3 + E, where S, = salinities at seasonal points; and t = relative distance between successive opaque zones. For each fish, S^ is arrayed in Ji rows, Sex contains two treatment levels for each factor (male vs. female) arrayed in n rows, and E is the matrix of model residu- als. Degrees of freedom in the analysis depend upon n, which represents the number of individual fish. To avoid the problem of interdependence of seasonal data for combined ages, separate MANOVAs were performed for each age class. To conduct the RM-MANOVA, it was necessary to have equal numbers of seasonal points (S,) for each age class. Therefore, narrow annuli that had fewer than three seasonal (interannual) points were omitted from analyses. This excluded analyses of some of the old- est age classes, which typically exhibit narrow annuli (Secor, 1992). For years sampled with more than four or five seasonal points, the extra points were omitted from our analysis. Selection of points to be included was based upon their proximity to the axial distances of the prescribed intervals (either [0, 0.25, 0.5, 0.75] or [0, 0.2, 0.4, 0.6, 0.8]). Individual probe points within transects were separated by 10 to 35 microns. Results Demographics Ages among the sampled Maryland Chesapeake Bay striped bass ranged from three to 18 years and sizes ranged 320 to 1110 mm FL. Females were significantly older and larger than males (ANOVA; P<0.01) and grew at a faster rate. Males were more heavily represented by ages <10 years than females in age-frequency distribu- tions, although fish with ages >15 years were observed for both sexes. Relatively strong year class contributions within the sample occurred for 1982, 1989, 1993, and 1996 and coincided with high young-of-the-year juvenile abundances observed in those years (Secor, 2000a). Oceanic incidence of striped bass Female fish, more often than male fish, were classified as having a recent period of oceanic residence based upon the analysis of the last fully formed annulus, but this difference was not significant (/-, P>0.1). For individual age classes with sample sizes >5, oceanic incidence ranged from 60% to 75% for females and from 17% to 50% for males (Table 1). There was an indication that the proportion offish of both sexes with oceanic residence increased with age. Oceanic inci- dence was observed consistently for >50% of the females. For males, oceanic incidence was 8-32% less than for females within each age class. Error bars, based upon a binomial probability distribution, indicated a fairly well- estimated oceanic classification rate for the age class 10-12 years due to a relatively high sample size. For this age class, oceanic incidence was estimated at 59% and 50% for females and males, respectively. Conversely, use of oceanic habitat for males at ages <10 was poorly esti- mated because of low sampling size. Indeed, estimated ratios for this group could not be statistically resolved from zero. With increased size, oceanic incidence (sexes combined) tended to increase (Table 1), although there was a decline in the proportion of fish with evidence of oceanic residence from 65.8% for the size class 900- 999 mm FL to 46.1% for the size class >1000 mm FL. Life history transects Life history transects showed considerable variability (Fig. 1). It is noteworthy that some males exhibited Secor and Piccoli: Oceanic migration rates of Mo/one soxotilis, determined by otolith microchemical analysis 65 Age class (yr) 4-6 7-9 10-12 13-18 some degree of oceanic incidence (fish identification [ID] number=98, 198, 260) throughout portions of their lives and many females exhibited a pattern of estuarine use (e.g., ID = 197, 263, 271, 272, 280, 281). A single instance of freshwater residency was observed for a fairly long-lived female (ID=271; age = ll years; FL = 875 mm). Mean lifetime salinity exposure differed sig- nificantly between males and females (Kruskal-Wallis ANOVA; P=0.01), with females exhibiting an average 10% higher use of high-salinity habi- tat during the mature portion (age>6) of their lives (Fig. 2). The male sample, albeit small (7! =10), did not exhibit age-dependent patterns in salinity exposure (Fig. 2). In part, high variance in older age classes (8-12) obscured any pattern of salinity exposure at ages <6 years. Females showed a strong and nearly linear trend of increased salinity ex- posure with age (Fig. 2). Modal salin- ity increased from a range of 20-25 to a range of 25-30 for ages 2 and 7, respectively. Interestingly, both males and females showed that a polyha- line (salinity>18) habitat was used during the period between age 1 and 2 years. Thus, yearlings may be preferentially using polyhaline regions, followed by a return of some individuals to lower salinity regions (because slightly depressed salinities were observed for ages 2 and 3 compared to age 1 yr in Fig. 2). We have observed a similar pattern in Hudson River striped bass (Zlokovitz et al., 2003). The ontogenetic trend of increased salinity exposure with age in females could be related to maturation (ages 6-8 years) — a pattern observed in two-thirds of female fish (Fig. 3). Males also showed a rise in salinity exposure with age, albeit less consistently (Fig. 4). Spawning frequency Life history transects for the sample of ten males and thirty females gave evidence of strong intra-annual pat- terns in the salinity levels of their habitat (Figs. 3 and 4). Intra-annual trends often showed nadirs at or near the opaque zone of the otolith, a pattern occurring in both males (e.g., ID = 124, 196, 327) and females (e.g., ID = 192, 297, 298). In some instances, seasonal cycles indicated either less than annual (Fig. 3: ID = 320; Fig. 4: ID=298, 300) or greater than annual (Fig. 3: ID = 99: Fig. 4: ID = 295, 300) cycles in patterns of salinity exposure. Significant seasonal effects on patterns of salinity exposure were observed for ages 6 and 8 and for pooled ages 6-11 (Table 2), indicating that across individuals there was a seasonal pattern in salinity exposure for Table 1 Degree of oceanic incidence of Chesapeake Bay striped bass collected in 2000. Data is presented for age classes by sex, and for size-class pooled sexes (the latter to permit comparison with results of Dorazio et al., 1994). Lower (LCL) and upper (UCL) confidence limits are presented for oceanic incidence by age class. Oceanic incidence (% ) Males in; LCL-UCL) Females (n; LCL-UCL) 20.0(5:0-41.3) 25.0(8:0-50.2) 50.0(18:37.4-62.5) 44.4(9:34.0-77.4) 55.5(18:42.5-68.1) 58.8(51:56.8-62.5) 76.9(13:60.8-93.6) Oceanic incidence (%) FL size class (mm) This study (h ) Dorai 300-599 20.0(8) <5.0 600-699 — -8.0 700-799 42.9(14) -35.0 800-899 59.0(44) -80.0 900-999 65.8 (41) -90.0 >1000 46.1(13) >95.0 Dorazio et al. (1994) mature age classes. The other analysis of five, rath- er than four, seasonal points of measurement in the otolith did not show strong evidence for seasonality, although pooled age classes 6-11 and 7-11 showed marginal significance at P=0.06. In the selection of individuals that contained five seasonal points, sample sizes were substantially reduced. This reduced sample size in turn would have resulted in less statistical sensitivity. Despite high variances among individuals for each seasonal measurement, there was a trend for a seasonal nadir in salinity exposure near the opaque zone on the otolith (seasonal interval=0) (Fig. 5). This trend was especially apparent in males, and in females >6 years. There was a slight indication of sex-related differ- ences in seasonality of salinity exposure. Significant between-sex differences occurred for Sg at ages 8 and 11, and all mature age class groupings (Table 2). Sig- nificant differences occurred for Sg 25 at ages 8 and 9 and mature age groupings 7-11 and 8-11; and for Sg 5 at age 8 and all mature age groupings. Males were found at lower salinities at these seasonal points than were females (Fig. 5). X-ray maps of Sr within otolith sections showed clear concentric patterns of alternating high and low regions of Sr in association with annuli (Fig. 6). X-ray maps confirmed the cyclical patterns observed in the life history transects (Figs. 3 and 4), but with greater ap- parent difference between peak and nadir levels of Sr. In one of the X-ray maps (Fig. 6, bottom panel), a high level of Sr occurred within the first annulus). 66 Fishery Bulletin 105(1) 40 r 10 Males L _L J I I L _L ^'i^ # ^'P ri? r,^ ^T? ^n^ ID 30 1 10 Females ^(^filnffi B M I I T 1 I 1 I I 1 1 I I I 1 I I I I I 1 I 1 1 I 1 I I ,.i^,o,1^fc\6V\1VV\*\*\A*V^A"5VVV^''VV^^\^VVV^A•i^^^^^^ ID Figure 1 Box and whisker plots of salinity records experienced over a lifetime by individual male (A) and (B) female Chesapeake Bay striped bass (Morone saxatilis) collected in 2000. Median values are given within each box, which represents the first two quartiles of data about the median. Asterisks and circles indicate near and far outliers in relation to the first two quartiles, respectively. Salinity is given in practical salinity units. ID indicates the identification number assigned to each fish. Discussion Oceanic incidence of striped bass Our analysis and that of Dorazio et al. (1994) does not support Kohlenstein's (1981) model of mass egress of female striped bass from Chesapeake Bay after ages two or three. Rather, our life history transects indi- cated a fairly gradual shift to use of ocean habitats — a shift associated with maturation at ages five to eight (Table 1; Fig. 2). For mature age classes, evidence of oceanic residence was observed for 50-75% of the female sample. Also, in contrast to previous expectations, otolith microanalysis indicated that a large fraction of males leave Chesapeake Bay, albeit at rates <50'7f . Mirroring the results of Dorazio et al.'s (1994) tag- ging experiment, our results showed a trend of in- creasing oceanic residence with fish size, but found Secor and Piccoli; Oceanic migration rates of Morone saxatilis. determined by otolith microchemical analysis 67 10 11 12 1 2 3 4 5 6 7 8 9 10 11 12 Age (y) Figure 2 Box whisker plots of age-specific salinities experienced by Chesapeake Bay striped bass iMorone saxatilis) collected in 2000. Salinity data for each fish include quarterly seasonal salinity records. Median values are given within each box, which represents the first two quartiles of data about the median. Asterisks and circles indicate near and far outliers in relation to the first two quartiles, respectively. Salinity is given in practical salinity units. ID indicates the iden- tification number assigned to each fish. 124 196 10 12 Age (yr) Figure 3 Standardized life history transects for male Chesapeake Bay striped bass {Morone saxatilis) collected in 2000. ID, an identification number assigned to each fish, is presented above each panel. Z-score indi- cates standardized salinity records for each fish where Z = (record-mean)/standard deviation. Dashed line demarcates the mean. 68 Fishery Bulletin 105(1) 192 295 297 10 1 1 298 T — I — I — I — r 1 9 10 Figure 4 Standardized life history transects for female Chesapeake Bay striped bass (Morone saxatilis) collected in 2000. ID, an identification number assigned to each fish, is presented above each panel. Z-score indicates standardized salinity records for each fish where Z = (record-mean)/standard deviation. Dashed line demarcates the mean. two important differences. First, we did not observe >90% oceanic incidence at >900 mm TL (877 mm FL) (Table 1). Rather, a substantial fraction of striped bass remained resident in Chesapeake Bay throughout their lives regardless of age or size. At ages >12 years, 25% of the female sample was estimated to have be resident in Chesapeake Bay. As an extreme example, one individual female resided in freshwater during its entire 11-yr lifespan. Secondly, early rates of oceanic migrations at sizes <700 mm TL (<685 mm FL) were substantially higher than rates indicated in the Dorazio et al. (1994) model, which predicted <5% of individuals migrate to ocean waters. Several factors may have contributed to the different results. Dorazio et al. (1994) predicted the degree to which Chesapeake Bay striped bass mi- grate to coastal regions north of Cape May New Jersey. Thus, their recapture sample represents only a subset of possible coastal fish. This bias would tend to under- estimate oceanic residence; yet the Dorazio et al. (1994) estimates tend to be higher for larger mature striped bass. Recapture and reporting rates probably varied between coastal and Chesapeake regions because of more restrictive fishing regulations in Chesapeake Bay that contributed to an overestimate of migrant fish to northern ocean habitats. In comparing our results to those of past tagging studies, we must also carefully consider limitations to the otolith microchemistry approach. We have sought to overcome some of the past hurdles regarding low sample size and resolution, yet these remain principal concerns. Despite large improvements in microprobe technology, otolith microchemistry remains a very expensive and time-consuming procedure to evaluate population-spe- cific patterns in fish migration. Sample sizes, while larger than in many past projects, remain modest. Also, ages and sizes were not uniformly distributed in the populations sampled because of strong year classes, and this lack of uniform distribution would curtail generalizations across size and age classes. This strong year class phenomenon is common in striped bass, and it is likely that migration patterns in Chesapeake Bay striped bass typically will be influenced by dominant year classes (Merriman, 1941). A central assumption in using strontium as a tracer of salinity levels is that the ratio of strontium to cal- cium (Sr:Ca) in the otolith can accurately distinguish between oceanic (salinity >29) and estuarine (salinity <30) habitat use. First, the designation, between oce- anic and estuarine water is somewhat arbitary, par- ticularly considering that the mouth of Chesapeake Secor and PIccoli: Oceanic migration rates of Morone saxatilis, determined by otolith microchemlcal analysis 69 Table 2 Repeated measures MANOVAs for effects of sex on seasonal salinities experienced by Chesapeake Bay striped bass {Morone saxatilis). S, = salinity at seasonal point t, where t = relative distances between successive opaque zones in otoliths. "Season" is defined by two interval 3 on the otolith as a proportion of otolith increment width. Univariate F tests (P-values reported) for 1 contrasts bet ween males and females are given for each seasonal point. Differences among seasonal salinities ( "Seasonality") within each age were eva uated with a Wilk's statistic (Chambers and Miller, 1995). The interaction between season and sex | tested whether sex affected patterns of seasonality in salinity levels in otolith s. NS = not signi ficant. Age (yr) ;; Contrast between males and females Seasonality Season x Sex So So.25 So.5 So.75 1 11 NS NS NS NS NS NS 2 19 NS NS NS NS NS NS 3 21 NS NS NS NS NS NS 4 29 NS NS NS NS NS NS 5 33 0.09 NS NS NS NS NS 6 35 NS NS NS NS 0.02 0.04 7 32 NS NS NS NS NS NS 8 22 0.02 0.01 0.04 NS 0.04 NS 9 22 NS 0.03 NS NS NS NS 10 20 NS NS NS NS NS NS 11 7 0.04 NS 0.07 NS 0.07 NS 6-11 138 0.04 NS 0.02 NS 0.04 NS 7-11 103 0.04 0.01 0.03 NS 0.09 NS 8-11 71 0.03 0.002 0.01 NS NS NS Age n Contrast between males and females Seasonality Season x Sex So So,2 So.4 So.6 Sq.s 5 31 0.09 NS NS NS NS NS NS 6 31 NS NS NS 0.03 NS NS NS 7 22 NS 0.003 NS NS NS NS NS 8 19 0.02 0.07 0.02 NS NS NS NS 9 13 NS 0.01 0.02 NS NS NS NS 10 11 NS NS 0.02 NS NS NS NS 11 5 NS NS NS NS NS NS NS 6-11 101 0.001 0.007 0.001 0.01 0.05 0.06 NS 7-11 70 0.001 0.001 0.001 NS NS 0.06 NS 8-11 48 0.005 0.01 0.001 0.08 NS NS NS Bay averages about 25-30 psu during summer and fall months. Still, it was necessary to have a designation for assessing habitat use, and the above designations may have resulted in a liberal estimate of oceanic residence. A second issue is the resolution of the relationship of otolith Sr-Ca ratio to salinity. Resolution level esti- mated by experimental work of Secor et al. (1995a) was a salinity of 6 psu, which would support the contention that estimates presented here are fairly precise if er- ror is unbiased. Still, the relationship between otolith Sr:Ca and salinity was logistic and very rapid changes in Sr:Ca were predicted to occur with small changes in salinity at salinities between 25 and 35 psu. These rapid changes at high salinities could indicate higher unexplained variability at salinities >24 psu. Finally, our samples were unequally weighted across subpopulations of striped bass. For instance, no lower Chesapeake Bay subpopulations (those spawning in the James, York, and Rappahannock systems) were represented in our sample). Recent tagging studies have either focused more narrowly on the Potomac River (Kohlenstein, 1981) or have drawn a larger and more representative sample from the Maryland sec- tion of Chesapeake Bay (Dorazio et al., 1994). Studies on Virginia subpopulations of striped bass have his- torically shown low rates of oceanic residence (<5%; Vladykov and Wallace, 1952; Massman and Pacheco, 1961). Kohlenstein (1981) effectively argued that these and other early tagging studies (i.e., Mansueti, 1961) were not appropriately stratified to provide evidence of an increased likelihood of oceanic residence by larger size fish. In sum, we believe that our otolith microchemistry results indicate higher rates of early oceanic residency in females and overall higher rates of oceanic migra- tions by males than were observed in previous tag- 70 Fishery Bulletin 105(1) ra 26 24 22 20 18 16 26 24 22 20 18 16 Male < 7 years 0.25 0.5 0.75 Female < 7 years 30 26 22 0.25 0.5 0.75 33 29 25 21 Male > 6 years 0.25 0.5 0.75 Female > 6 years 0.25 0.5 0.75 Interval Figure 5 Seasonal patterns in salinity exposure as recorded in the otoliths of Chesapeake Bay striped bass iMorone saxatilis) collected in 2000. Mean salinities and standard errors are shown for age classes <7 and >6 years. Seasonal interval indicates the relative distances from the opaque zones in the otolith, which are formed in early spring. Intervals represent the seasons as a proportion of annual otolith increment width. ging studies (Kohlenstein, 1981; Dorazio et al., 1994). Further, increasing trends in oceanic habitat use with age observed in our study are consistent with the two previous tagging studies. The increased incidence of males in ocean environments shown in our study and previous ones could reflect a true increased likelihood of emigration, perhaps driven by increased striped bass density, or by poorer habitat conditions in Chesapeake Bay. This view would be consistent with the near his- torically high abundances of Chesapeake Bay striped bass and the increased incidence of summertime hy- poxia in Chesapeake Bay during the past two decades (Hagy et al., 2004). However, error due to the otolith microchemistry approach in classifying fish to oceanic or estuarine habitat use must be acknowledged and caution should precede application of the estimates of oceanic residence provided in this study. Spawning frequency In general, otolith microchemical analyses gave evidence for the view that most mature striped bass undertake annual spawning runs. For females, immature age classes did not show significant seasonality in changes in salinity, but many mature age classes did. Further, cycles in salinity were largely defined by a nadir that occurs during early spring as evidenced by changes in the chemistry of the opaque zone of the otolith. Thus, evidence of the use of low-salinity habitat was recorded near the opaque zone, consistent with the view of an annual up-estuary migration to low-salinity spawning habitats. Where such nadirs were not observed in the otolith microchemistry, two interpretations are plausible: 1) no spawning migration occurred, or 2) the otolith microchemistry method had insufficient resolution to allow us to detect the spawning migration. The resolu- tion issue relates to two problems. First, the spacing of the microprobe assays could have been such that a spawning run event was missed. Second, spawning- run striped bass occur in low-salinity regions for short periods during which they are not growing and thus incorporating Sr material into their otoliths. In this instance, there would be an insufficient signal for oto- lith microanalysis of Sr to detect. Despite these likely sources of error, we were still able to detect a dominant annular cycle in otolith Sr for mature age classes of males and females. Therefore, we believe that the otolith microchemistry analysis supports annual spawning for the majority of mature Chesapeake Bay striped bass. Alternatively, spawning in striped bass may occur less than once a year. Less than an annual spawning, once thought to be specific to relatively few taxa (e.g.. Secor and Piccoll: Oceanic migration rates of Morone soxot/l/s, determined by otolith microchemical analysis f^::3^^^^-^; «^-,it>'^^i^ : . Figure 6 X-ray composition maps of srontium (Sr) seen in otolith for two female striped bass ^Morone saxatilis) (17 and 11 years old) collected in Chesapeake Bay in 2000. Arrows demarcate annuli. The "warmer" colors (red, orange) represent regions with higher Sr concentrations whereas the "cooler" colors (yellow, green) represent regions with lower Sr concentrations. Composition maps show the coincidence of stron- tium banding patterns with banding patterns of annuli, indicative of annual anadromous migrations from high to low salinity waters. sturgeons; Gross et al., 2002) could in fact be common in some longer-lived species (Rideout et al., 2005). For instance, data-archiving electronic tags inserted on At- lantic bluefin tuna (Thunnus thynnus) have definitively shown that many adults are found outside spawning habitats for an entire annual spawning season (Block et al., 2005). Further evidence is provided by the re- productive behavior of Atlantic cod (Gadus morhua; Jorgensen et al., 2006), a species that does not always spawn each year because of density dependence or other environmental limits on its ability to provision gonads. Interestingly, evidence for this can be observed in two female life history transects in our study of striped bass (Fig. 4: ID=295, 300), where no nadir was observed at age seven following a clear nadir at age six (female striped bass typically mature between five and seven 72 Fishery Bulletin 105(1) years of age). We also should expect that, for similar energetic reasons, not every female, which undertakes a springtime up-estuary migration, will actually spawn. Some mature females in a spawning run will not spawn; rather, they will reabsorb final-stage oocytes (first au- thor, personal observ. ). Therefore, accurate measurement of spawning frequency depends on both the probability of successful spawning in the field and the frequency of up-estuary migratory runs. X-ray maps confirmed an annual cycling in otolith strontium, but also showed cycles during the immature period of females, contrary to patterns observed from life history transects. Further, X-ray mapping and life history transects indicated that many yearlings move into oceanic regions — a pattern observed for Hudson River striped bass (Zlokovitz et al., 2003) but not yet described for Chesapeake Bay striped bass. The pos- sibility that young-of-the-year or yearling striped bass are present in ocean environments deserves additional research in the Chesapeake Bay and elsewhere. Contingent migration behavior In past research on Hudson River striped bass, we observed modalities in lifetime migration behaviors (Secor et al., 2001); groups of individuals that share similar migration behaviors with some, but not all, members of their population are termed "contingents" (Hjort, 19141; Gilbert 19172; Secor, 1999). In particular, one contingent comprising a small fraction of the Hudson River springtime sample was resident to freshwater and oligohaline regions and was heavily contaminated by polychlorinated biphenyls — an apparent consequence of this lifetime migration behavior (Zlokovitz and Secor, 1996; Ashley et al., 2000). A similar contingent migra- tion behavior has been reported for the Steweiacke River population of striped bass in Nova Scotia (Morris et al., 2003). In contrast, only a single Chesapeake Bay striped bass (of 40 analyzed) exhibited freshwater resident behavior. Small sample sizes could indicate considerable error in estimates of the frequency of this behavior among populations if these estimates are based on research to date, but the fact that three distinct populations exhibited this behavior in our study indi- cates that contingent migration structuring is common to Chesepeake Bay striped bass. Contingent migration structure has been observed across diverse taxa, such as American eels (Anguilla rostrata), American shad (Alosa sapidissima), white perch (Morone americana), bluefish (Pomatoinus salta- ' Hjort, J. 1914. Fluctuations in the great fisheries of north- ern Europe. Rapports Conseil Permanent International Pour L'exploration de la Mer, 20:1-228. Library, Chesapeake Biological Laboratory, P.O. Box 38, Solomons, MD 20688. - Gilbert, C. H. 1917. Contributions to the life-history of the sockeye salmon. Paper No. 4, report to the Commissioner of Fisheries. British Columbia Fisheries Department, 48 p. + plates. Library, Pacific Biological Station, Fisheries and Oceans Canada, 3190 Hammond Bay Road, Nanaimo, V9R 5K6 British Columbia. Canada. trix), and Atlantic bluefin tuna (Secor, 1999, in press; Fromentin and Powers, 2005). Here, a nursery or forag- ing habitat associated with one contingent migration behavior may make a small contribution in a given year, but over a decade may contribute significantly to spawning stock biomass. Thus, over generation-long time scales, we should expect that minority lifetime migration behaviors can contribute significantly to sustained recruitment. Further research is needed to determine the proximate cause of contingent struc- ture, but based upon research on the sympatric white perch, we advance the hypothesis that Chesapeake Bay striped bass contingent migration structuring re- sults from divergent early growth rates and dispersal behaviors associated with early growth (Kraus and Secor, 2004b). Acknowledgments We acknowledge NOAA MARFIN (Marine Fisheries Initiative, Northeast Center) support for this research. S. McGuire assisted with otolith preparation and data analyses. Personnel at Maryland Department of Natural Resources, Horn Point Environmental Laboratory, and the U. S. Fish and Wildlife Service assisted in collection of striped bass. The electron probe microanalyzer was purchased, in part, by a grant from the National Science Foundation (EAR 98-1244). Literature cited Ashley, J. T. F, D. H. Secor, E. Zlokovitz, J. E. Baker, and S. Q. Wales. 2000. Linking habitat use of Hudson River striped bass to accumulation of polychorinated biphenyl congeners. En- viron. Sci. Techn. 34:1023-1029. Block, B. A., L. L. H. Teo, A, Walli, A. Boustany, M. J. W. Sokesbury, C. J. Farwell, K. C. Weng, H. Dewar, and T. D. Williams. 200.5. Electronic tagging and population structure of Atlantic bluefin tuna. Nature 434:1121-1127. Chambers, R. C, and T.J. Miller. 1995. Evaluating fish growth by means of otolith incre- ment analysis: Special properties of individual-level longitudinal data. In Recent developments in fish otolith research (D. H. Secor, S. E. Campana, and J. M. Dean, eds.), p. 155-175. Belle W. Baruch Library in Marine Sciences Number 19. Univ. South Carolina Press, Columbia, SC. Dorazio, R. M., K. A. Hattala, C. B. McCollough, and J. E. Skjeveland. 1994. Tag recovery estimates of migration of striped bass from spawning areas of the Chesapeake Bay. Trans. Am. Fish. Soc. 123:950-963. Fromentin, J. M., and P. E. Powers. 2005. Atlantic bluefin tuna: population dynamics, ecol- ogy, fisheries, and management. Fish and Fisheries 6:281-306. Goodyear, C. P. 1984. Analysis of potential yield per recruit for striped Secor and Piccoli: Oceanic migration rates of Morone saxatilis, determined by otolith microchemical analysis 73 bass produced in Chesapeake Bay. N. Am. J. Fish. Manage. 4:488-496. Gross, M. R., J. Repka, C. T. Robertson, D. H. Secor, and W. Van Winkle. 2002. Sturgeon conservation: insights from elasticity analysis. Am. Fish. Soc. Symp. 28:3-10. Hagy, J. D., W. R. Boynton, C. W. Wood, and K. V. Wood. 2004. Hypoxia in Chesapeake Bay, 1950-2001: long- term changes in relation to nutrient loading and river flow. Estuaries 27:634-658. Ingram, B. L., and D. Sloan. 1992. Strontium isotopic composition of estuarine sedi- ments as paleosalinity-paleoclimate indicator. Science. 255:68-72. Jergensen, C, B. Ernande, O. Fiksen, and U. Dieckmann. 2006. The logic of skipped spawning in fish. Can. J. Fish. Aquat. Sci. 63:200-211. Kalish, J. M. 1990. Use of otolith microchemistry to distinguish prog- eny of sympatric anadromous and non-anadromous salmonids. Fish. Bull. 88:657-666. Kohlenstein, L. C. 1981. On the proportion of the Chesapeake stock of striped bass that migrates into the coastal fishery. Trans. Am. Fish. Soc. 110:168-179. Kraus, R. T., and D. H. Secor. 2004a. Partitioning of strontium in otoliths of estuarine fishes: experimentation in a model system, white perch Morone americana (Gmelin). J. Exp. Mar. Biol. Ecol. 302:85-106. 2004b. The dynamics of white perch (Morone ameri- cana Gmelin) population contingents in the Patuxent River Estuary, Maryland USA. Mar. Ecol. Prog. Ser. 279:247-259. Mansueti, R. J. 1961. Age, growth, and movements of the striped bass, Roccus saxatilis, taken in size selective fishing gear in Maryland. Chesapeake Sci. 2:9-36. Marshall, C. T., L. O'Brien, J. Tomkiewicz, G. Marteinsdbttir, M. J. Morgan, F. Saborido-Rey, F. Koster, J. L. Blanchard, D. H. Secor, G. Kraus, P. Wright, N. V. Mukhina, and H. Bjornsson. 2003. Developing alternative indices of reproductive potential for use in fisheries management: Case stud- ies for stocks spanning an information gradient. J. Northw. Atl. Fish. Sci. 33161-190. Massman, W. H., and A. L. Pacheco. 1961. Movements of striped bass tagged in Virginia waters of Chesapeake Bay. Chesapeake Sci. 2:37-44. Merriman, D. 1941. Studies of the striped bass {Roccus saxatilis) of the Atlantic coast. USFWS Fish. Bull. 50:1-77. Morris, J. A., R. A. Rulifson, and L. H. Toburen. 2003. Life history strategies of striped bass, Morone saxatilis, populations inferred from otolith micro- chemistry. Fish. Res. 62:53-63. Rideout, R. M., G. A. Rose, and M. P. M. Burton. 2005. Skipped spawning in female iteroparous fishes. Fish and Fisheries 6:50-72. Secor, D. H. 1992. Application of otolith microchemistry analysis to investigate anadromy in Chesapeake Bay striped bass. Fish. Bull. 90:798-806 1999. Specifying divergent migration patterns in the concept of stock: the contingent hypothesis. Fish. Res. 43: 13-34. 2000a. Longevity and resilience of Chesapeake Bay striped bass. ICES J. Mar. Sci. 57:808-815. 2000b. Spawning in the nick of time? Effect of adult demographics on spawning behavior and recruitment of Chesapeake Bay striped bass. ICES J. Mar. Sci. 57:403-411. 2002. Estuarine dependency and life history evolution in temperate sea basses. Fish. Sci. 68lsuppl. 1): 178- 181. In press. The year-class phenomenon and the storage effect in marine fishes. J. Sea Res. Secor, D. H., J. M. Dean, and E. H. Laban. 1992. Otolith removal and preparation for microstruc- tural examination. Chapter 3 in Otolith microstruc- ture examination and analysis iD. Stephenson and S. Campana, eds.), p. 17-59. Special Publication, Can. J. Fish. Aquat. Sci. 117. Secor, D. H., A. Henderson-Arzapalo, and P. M. Piccoli. 1995a. Can otolith microchemistry chart patterns of migration and habitat utilization in anadromous fishes? J. Exp. Mar. Biol. Ecol. 192:15-33. Secor, D. H., and P. M. Piccoli. 1996. Age- and sex-dependent migrations of the Hudson River striped bass population determined from otolith microanalysis. Estuaries 19:778-793. Secor, D. H., J. R. Rooker, E. Zlokovitz, and V. S. Zdanowicz. 2001. Identification of riverine, estuarine, and coastal contingents of Hudson River striped bass based upon otolith elemental fingerprints. Mar. Ecol. Progr. Ser. 211:245-253 Secor, D. H., T. M. Trice, and H. T. Hornick. 1995b. Validation of otolith-based ageing and a compari- son of otolith and scale-based ageing in mark-recaptured Chesapeake Bay striped bass, Morone saxatilis. Fish. Bull. 93:186-190. Sokol, R. R., and F. J. Rohlf 1981. Biometry, 2"'* ed., 859 p. W. H. Freeman and Co., New York, NY. Vladykov, V. D., and D. H. Wallace. 1952. Studies of the striped bass, Roccus saxatilis (Wal- baum), with special reference to the Chesapeake Bay region during 1936-1938. Bull. Bingham Oceanogr. Coll. 14:132-177. Warner R. R., and P. L. Chesson. 1995. Coexistence mediated by recruitment fluctua- tions — a field guide to the storage effect. Am. Natu- ralist 125:769-787. Wirgin, I., L. Maceda, J. R. Waldman, and R. N. Crittenden. 1993. LTse of mitochondrial DNA polymorphisms to esti- mate the relative contributions of the Hudson River and Chesapeake Bay striped bass to the mixed fishery on the Atlantic Coast. Trans. Am. Fish. Soc. 122: 669-684. Zlokovitz, E. R., and D. H. Secor. 1999. Effect of habitat use on PCB body burden in Hudson River striped bass (Morone saxatilis). Can. J. Fish. Aquat. Sci. 56 (suppl.l):86-93. Zlokovitz, E. R., D. H. Secor, and P. M. Piccoli. 2003. Patterns of migration in Hudson River striped bass as determined by otolith microchemistry. Fish. Res. 63:245-259. 74 Abstract— From 2001 to 2004 in the eastern Aleutian Islands, Alaska, killer whales (Orcinus orca) were encountered 250 times during 421 days of surveys that covered a total of 22,491 miles. Three killer whale groups (resident, transient, and off- shore) were identified acoustically and genetically. Resident killer whales were found 12 times more frequently than transient killer whales, and offshore killer whales were encoun- tered only once. A minimum of 901 photographically identified resident whales used the region during our study. A total of 165 mammal-eating transient killer whales were identi- fied, and the majority (70%) were encountered during spring (May and June). The diet of transient killer whales in spring was primarily gray whales iEschrichtius robustus), and in summer primarily northern fur seals (Callorhinus ursinus}. Steller sea lions (Eumetopias jubatus) did not appear to be a preferred prey or major prey item during spring and summer. The majority of killer whales in the eastern Aleutian Islands are the resident ecotype, which does not consume marine mammals. Ecotypic variation and predatory behavior among killer whales {Orcinus orca) off the eastern Aleutian Islands, Alaska Craig O. Matkin (contact author)' Lance G. Barrett-Lennard^ Harald Yurk^ David Ellifrit^ Andrew W. Trites'' Email address for C. O, Matkin: cmatkin(5iacsalaska,net ' North Gulf Oceanic Society 3430 Mam St. Bl Homer, Alaska 99603 ^ Vancouver Aquarium Marine Science Center University of British Columbia 845 Avison Way Vancouver, B.C , Canada V6G 3E2 ^ Center for Whale Research 355 Smugglers Cove Friday Harbor, Washington 98250 " Marine Mammal Research Unit, Fisheries Centre University of British Columbia Vancouver, B.C., Canada V6T IZ4 Manuscript submitted 5 December 2005 to the Scientific Editor's Office. Manuscript approved for publication 18 April 2006 by the Scientific Editor. Fish. Bull. 105;74-87 (2007). In 1992, flipper tags from fourteen Steller sea lions (Eumetopias jubatus} were found in the stomach of a killer whale (Orcinus orca) that had died in Prince William Sound (Heise et al., 2003). This discovery prompted considerable interest and specula- tion about the role that killer whales may have played in the decline and lack of recovery of Steller sea lions in western Alaska (Barrett-Lennard et al., 1995). Since the late 1970s, Steller sea lions in the Gulf of Alaska and Aleutian Islands have declined by over 80% (Merrick et al., 1987; Trites and Larkin, 1996; Loughlin and York, 2000; Winship and Trites, 2006). Similar sharp declines have also occurred among some popula- tions of harbor seals (Phoca vitulitia), northern fur seals (Callorhinus ursi- nus), and sea otters (Enhydra lutris) (York, 1987; Pitcher, 1990; Trites, 1992; Estes et al., 1998). Whether or not these declines are related to killer whales is currently the sub- ject of considerable scientific debate (Springer et al., 2003; Trites et al.. 2006; DeMaster et al., 2006; Mizroch and Rice, 2006 ). Most knowledge about killer whales in the North Pacific has been gathered between California and the northern Gulf of Alaska, where three distinct lineages of killer whales have been identified: fish-eat- ing "resident" killer whales, which appear predictably in large groups from Washington to Alaska; marine mammal-eating "transient" killer whales, which appear infrequently and in smaller groups; and "offshore" killer whales, whose feeding habits are poorly known, but are thought to eat fish, including sharks (Mat- kin et al., 1999a; Barrett-Lennard, 2000; Ford et al., 2000; Saulitis et al., 2000). These groups are geneti- cally and behaviorally distinct, but have overlapping geographic ranges and are considered as ecotypes be- cause of their differences in diet. However, prior to our study, it was not known whether these lineage and ecotype distinctions extended to the northwestern Gulf of Alaska and the Matkin et al.: Ecotypic variation and predatory behavior among Orcinus orca off Ihe eastern Aleutian Islands 75 Aleutian archipelago, nor was much known about the extent to which killer whales prey on Steller sea lions and other species of marine mammals in these regions. The goals of our study were 1) to determine whether the eastern Aleutian Islands are also home to the three lineages and ecotypes of killer whales that have been identified else- where in the northeastern Pa- cific; 2) to derive estimates of killer whale numbers for this region; and 3) to document the behaviour of killer whales foraging on marine mammals. Obtaining such information about killer whale numbers, diets, and hunting behavior is critical for resolving the role that killer whales may have played in the decline and lack of recovery of Steller sea lions and other species of marine mammals in western Alaska. Materials and methods Vessel survey tracklines Summer survey area '. Spring survey area False Pass Alaska Peninsula BERING SEA Akutan Unalaska island Island 54°N - PACIFIC OCEAN 166°W I 15 30 60 Nautical Miles !_] Alaska Figure 1 Tracks of the vessels during the surveys for killer whales {Orcinus orca) for the years 2001-2004. The vessels tended to return to areas that produced encounters with killer whales. The spring False Pass-Unimak Island surveys are distinguished from the summer Unimak Pass-Umnak Island surveys by the dashed line. Following the research method used to collect informa- tion on killer whales in other regions (Matkin et al., 1999a; Ford et al., 20001, five types of data were gath- ered: photo-identification pictures of individuals and groups, acoustic recordings of killer whale calls, skin tissue samples for genetic analysis, prey samples, and accounts of predation. Additionally, we documented the presence of potential marine mammal prey. Analysis of these data allowed the determination of killer whale eco- types and a description of killer whale feeding habits. Field methods Boat-based surveys over a wide geographic range occurred during June-September 2002-2004 from Unimak Pass to Samalga Pass, and surveys over a relatively small range occurred in May and early June 2003-2004 in the False Pass-Unimak Island region. Surveys in the broader region traversed 19,686 nauti- cal miles and were focused in the Bering Sea within twenty miles of the shoreline between Unimak Pass and eastern Umnak Island (Fig. 1). The 2003-2004 False Pass-Unimak surveys traversed 1970 miles in Ikatan Bay and along the Pacific shore of Unimak Island. We operated in areas of suspected high killer whale density according to information provided by local fishermen and researchers from the National Marine Mammal Labora- tory (NMML) (Dahlheim, 1997) during previous transect surveys that covered broader regions. We modified our surveys each season to cover the areas that were most productive in providing killer whale encounters. The research was conducted from aluminum-hull fish- ing vessels (powered by diesel inboard engines) ranging from 10 to 14 meters in length. Survey effort varied by year with a total of 372 days from 2001 to 2003 from the Unimak Pass to Samalga Pass in summer and a total of 49 survey days in 2003 and 2004 in the False Pass-Unimak Island region in spring (Table 1). All sightings of marine mammals during vessel surveys were recorded and the number of individuals was esti- mated to determine the relative abundance of potential prey items. Photographs of the left side of dorsal fins and saddle patches of killer whales were taken with a Nikon F-lOO camera (B and H Photo, New York, NY) equipped with either fixed 300-mm lenses or 100-300 zoom lenses and loaded with Fuji Neopan ASA1600 black and white film (B and H Photo New York, NY). These photographs were checked against existing photo-catalogues of Alaskan killer whales (Dahlheim, 1997; Matkin et al., 1999a) and other unpublished photographs. Tissue samples of at least one whale in each group were collected for genetic analysis and biopsy when weather and behavior of the whales permitted close approach. These samples were collected by using lightweight darts and an air- powered rifle (Barrett-Lennard et al., 1996). The outer skin portion of the samples was used for genetic analy- sis, and the underlying blubber portion was used for 76 Fishery Bulletin 105(1) Table 1 Survey effort (in days) and number of encounters viiith resident and transient killer whales iOrcinus orca ) in the eastern Aleutians (2001-2004). "Survey days" are the days spent looking for whales; "Miles" are given in nautical miles. Year Dates False Pass-Unimak Island region (single vessel 2003-2004) Dates (Single Eastern Aleutians vessel 2001, two vessels 2002-04) Survey days Miles Encounters Survey days Miles Encounters Resident Transient Resident Transient 2001 — — 19 Jun- 18 Aug 16 835 13 1 2002 — — — 17 Jun- 24 Aug 188 6599 57 4 2003 16 May- 3 Jun 18 642 13 10 Jun- 31 Aug 108 6321 49 4 2004 4 May- 3 Jun 31 1328 32 7 Jun- 9 Sep 130 6766 70 7 Total 49 1970 45 372 20,521 189 16 contaminant analysis, lipid and fatty acid, and stable isotope analysis (see Herman et al., 2005). Genetic analysis involved sequencing the entire mitochondrial control region (see Barrett-Lennard [2000] for details). Acoustic recordings were made when whales were vocal- izing and ambient noise levels permitted using an Off- shore Acoustics^*^ (Offshore Acoustics, Nanaimo, British Columbia) hydrophone with a built-in preamplifier and a Sony WM-D6C (B and H Photo, New York, NY) cas- sette recorder. This system had a frequency response of 10 Hz to 8 KHz (±3 dB). Single, continuous observation periods with killer whales were termed encounters. During these vessel- based observation periods, the location of the killer whales was plotted at approximately 5-min intervals by using a global positioning system (GPS) linked to a computer with Nobletec^"' (Nobletech, Beaverton, OR) navigational software. Time spent in different behav- ioral states (e.g., feeding, socializing, resting, traveling) was recorded on data sheets. Whales were observed con- tinuously during encounters, and any signs of possible predation were recorded. During behavioral observa- tions, marine mammal kills were confirmed only when marine mammal parts were observed in the mouths of the whales, or when bits of blubber, skin, or viscera, hair were collected, or blood or oil was observed on the surface of the water. Predation on fish was confirmed by observations of fish in the mouths of whales or by collecting and inspecting floating parts. To document the potential marine mammal prey in the region during the period of the study, we recorded the time, location, and number of all marine mammals sighted. Analytical methods Photo-identification All photographic negatives were examined over a light table with an 8.0 power Peak^^ (B and H Photo, New York, NY) magnification loop. Identifi- able individuals were recorded and assigned a unique alphanumeric name in order to be tracked throughout the study. Whales that could not be positively re-identi- fied were not assigned a name. From this photographic database, the actual number and identity of individual killer whales and groups of whales present for each encounter were determined. Because some of the pho- tographs were of poor quality, these photographs were rejected from further analyses; thus not all the whales encountered were identifiable. Acoustics We inspected acoustic recordings for the presence of discrete calls by listening to tapes and monitoring real-time spectrograms using Cool Edit 2000™ (Syntrillicum Software Corp., Phoenix, AZ) sound manipulation software. Calls were analyzed fol- lowing the protocol of Ford (1991) and Yurk (2005). Killer whales produce a variety of different types of vocalizations that can be described as clicks, whistles, and calls (Ford, 1989). Calls are the most common type of vocalization, occurring in over 90% of all encounters with vocalizing killer whales. These pulsed vocalizations occur as either signals of a frequently repeated acous- tic pattern or as signals of variable acoustic pattern (Ford, 1991). Discrete calls were chosen for this analysis because they retain their recognizable acoustic structure for many years and likely for many generations (Yurk, 2005). Recognized calls were digitized at a 44.1-kHz sampling rate with a 16-bit sample size and further analysed spectrographically using Canary 1.2.4 sound analysis software (Cornell Laboratory of Ornithology, Ithaca, NY). The spectrographic analysis was done by using fast-Fourier transformations (FFT) of time series of the recorded sound pressure waves with sizes of 1024 points for each analyzed time series. The FFT identifies the composing sine waves in sound pressure waves of Matkin et al,: Ecotypic variation and predatory behavior among Orcinus orca off the eastern Aleutian Islands 77 acoustic signals and allows spectrographic representa- tions of the sound frequency versus time and pressure. Spectrograms were produced with an 87.5% overlap of the analyzed time series. Resulting spectrograms had a time resolution of 2.9 milliseconds and a frequency resolution of 43 Hz. We categorized calls by ear and by visual inspec- tion of distinct upper and lower frequency components of the sound spectrum (UFC and LFC, respectively), as described by Miller and Bain (2000) and Yurk et al. (2002). When categorizing the calls as distinct, particular attention was given to 1) the existence and contour shapes of UFCs; 2) LFC contour shapes; 3) LFC segmentation (elements separated by silent inter- vals); and 4) the component structure (elements within the LFC arising from abrupt shifts in contour and not separated by silent intervals) of the LFCs (Ford, 1991; Yurk, 2005). The three known ecotypes of killer whales inhabit- ing waters off British Columbia and southern Alaska (resident, transient, and offshore) are acoustically dis- tinguishable by 1) vocalization rate; 2) the occurrence of different discrete calls; 3) the syllables used in calls; and 4) the production rate and characteristics of echolo- cation clicks. Transient killer whales, which appear to rely on passive listening to catch their marine mammal prey, vocalize less frequently than resident killer whales (Deecke et al., 2005). Transients rarely use echolocation clicks, in contrast to resident and offshore killer whales (Deecke et al., 2005). All calls of transient killer whales are distinct from the calls of resident whales by 1) an audible quavering of the fundamental sound frequencies (instead of a crisp appearance of these sound frequen- cies that is typical of calls from resident killer whales), and 2 ) a distinctively lower amount of different call syl- lables and a distinct order of these syllables compared to those in calls of resident killer whales (Yurk, 2005). Transient and resident killer whales are distinguish- able from offshore killer whales by their use of unique call types (Yurk, 2005). We determined whether the encountered whales fell into discrete acoustic groups and, if so, whether those acoustic groups were similar to any of the acoustic groups observed in British Columbia and southern Alaska. Analysis was completed by Yurk (2005), independent of knowledge of genetic differences and social associations among groups. Call rates were estimated from field estimates of killer whale group sizes for each encounter. Genetics DNA was extracted from the skin portion of the biopsies using proteinase K digestion, phenol and chloroform purification, and ethanol precipitation using standard procedures (Sambrook et al., 1989) We obtained mtDNA sequences using the following procedure; 1) the entire D-loop region was PCR-ampli- fied by using custom-designed primers that annealed to the flanking tRNA-Thr and 12s-rRNA regions (Barrett- Lennard, 2000); 2) the PCR product was purified with QIAQuick® spin columns (Qiagen, Valencia, CA) follow- ing protocols supplied by Qiagen, Ltd. (Valencia, CA); 3) a sequencing reaction was performed with Fs-Taq® (Ap- plied Biosystems, Foster City, CA) system reagents and protocols supplied by Applied Biosystems, Ltd. (Foster City, CA); and 4) the sequence was resolved on an Ap- plied Biosystems 377 (Applied Biosystems, Foster City, CA) automated DNA sequencer. Because the sequence was too long (950 bases) to be entirely resolved in one direction, sequencing reactions were run from each end of the amplified fragment. We visually checked the output graphs from the automated sequencer and cor- rected the computer-generated sequences accordingly. We also used the approximately 400-base overlap in the sequences of opposite directions to check for errors. As a final check of accuracy, we overlaid each output graph with a reference graph on a transparent sheet, and scanned the two graphs for differences. We then aligned unique sequences using the program CLUSTAL-W (Eu- ropean Bioinformatics, Cambridge, UK) (Thompson et al., 1994). Results Summary of survey effort and encounters with killer whales On 250 occasions, groups of one or more killer whales were encountered during the surveys that covered a total of 22,491 miles in 421 days in the eastern Aleutians and False Pass-Unimak Island area (Table 1). The majority of survey effort and encounters occurred west of Unimak Pass during summer; surveys in False Pass-Unimak Island area were not initiated until 2003. From approxi- mately half of our encounters with groups of killer whales in both regions, we obtained genetic samples or acoustic recordings (Table 1). Killer whales of the offshore ecotype were encountered only once (in 2003) and both acoustic and genetic samples were obtained during this encounter. Use of acoustic data, genetic analysis, and group association to infer lineage Genetic and acoustic analyses revealed the presence of three killer whale populations. As described in more detail below, one population clustered genetically and acoustically with resident killer whales ranging from Puget Sound, Washington to Kenai Fjords, Alaska, and a second population clustered with transient killer whales from the same general area. Accordingly, those two groups were provisionally classified as a resident killer whale group and a transient killer whale group, respectively. The third population clustered genetically with offshore killer whales sampled off British Columbia, and were provisionally classified as an offshore killer whale group. Acoustic comparison was not possible in the case of offshore killer whales because of a scarcity of recordings. Resident, transient, and offshore killer whales have never been observed interacting socially in the ar- 78 Fishery Bulletin 105(1) eas where they were previously identified and stud- ied (Puget Sound to Kenai Fjords), and no interaction was observed in our study between the genetically or acoustically distinguished groups. Therefore, it was possible to infer the population status from the group- association patterns of individuals for which there was no genetic or acoustic data. Animals observed in asso- ciation with whales of known genetic or acoustic type were assumed to be of that same type. We did not use diet as a criterion for classification to avoid circular reasoning (evidence of dietary differences between popu- lations becomes tautological if diet is used to define populations). Acoustic analysis During 31 of 39 encounters in which we recorded killer whale vocalizations and did not collect genetic samples, the use of distinct calls, use of echo- location clicks, and the call rate were consistent with attributes of resident killer whale vocalizations from other regions of the Northeast Pacific (Table 2) (Yurk, 2005). All encounters had average call rates of three calls or more per minute, and strings of echolocation clicks were abundant across encounters. During these encounters, 23 structurally distinct calls were identi- fied. Seven calls showed no obvious similarities to calls recorded elsewhere in the northeast Pacific and 15 had structural similarities, sharing some call components with calls used by killer whales that regularly occur in the northern Gulf of Alaska. However, no call from these encounters was identical to any call of the known killer whale call repertoires. The resident-type killer whales encountered in west- ern Alaska possibly belong to groups that are distinct from the groups of resident killer whales in other re- gions of Alaska because no call syllables or call pat- terns (sequence of syllables) between groups were found to match. Resident killer whales learn their distinct call structures in their maternal group, in which they remain for life, and call structures remain stable and group-specific for more than one generation (Ford, 1991; Yurk et al., 2002). However, because we do not know the complete call repertoires of killer whales in western Table 2 Location and number of encounters that produced record- ings of killer whales (Orcinus orca) used in our acoustic analyses from 2001 through 2004 in the Eastern Aleu- tians and False Pass, Alaska. Year Location Number of encounters 2001 2002 2003 2004 Total Unalaska Unalaska, Akutan Island, Umnak Island False Pass False Pass 31 2 3 39 Alaska we cannot be sure that we will not find complete call-type matches in the future. In eight killer whale encounters that did not yield genetic samples (five from False Pass and three from other areas in the eastern Aleutians), vocal activity or average call rate was considerably lower than the three calls per minute that are typical for residents, and was closer to one or less than one call per minute, which is typical for transient killer whales from other areas of the northeast Pacific (Deecke et al., 2005; Saulitis et al., 2005). Furthermore, all recorded calls showed typi- cal characteristics of transient type calls, such as the quavering of the fundamental sound frequencies and the lower number of call syllables compared to those in calls from resident killer whales. The three encounters that were not from the False Pass region (Table 2) contained three distinct calls that were structurally similar but did not show identi- cal order of syllables or identical syntax to calls used by members of the ATI transient community. The ATI transient community is thought to be limited to the Prince William Sound and Kenai Fjords region and to use a distinct call repertoire (Saulitis et al., 2005). In the recordings made during five encounters in the False Pass region in 2003 and 2004 (Table 2), 14 distinct calls were identified in more than one of the recording sessions. Thirteen of these 14 distinct calls were identified from recordings made during two en- counters in 2003. Ten of these 13 calls were also re- corded during 3 encounters with killer whales in May 2004 in the same area. Thus, although the majority of calls recorded in 2004 were already identified in 2003, one new distinct call was found. This high number of same distinct-type calls is typical for transient killer whales (Deecke, 2003). Call repertoires of resident killer whales are generally much larger, and this larger rep- ertoire is likely responsible for the detection of several new calls from newly encountered whales in recordings from consecutive years (Ford, 1991; Yurk, 2005). All call types recorded in the False Pass region appeared to be distinct from calls recorded from transient killer whales in other regions of the North Pacific. However, some structural similarity (in the form of matching call components) was found for some of the 14 calls recorded in our study and for the calls recorded from a transient community that inhabits waters along the west coast of North America. These results may indicate that the transient killer whales we encountered in the eastern Aleutians comprise one or more unique populations or communities that show some acoustic similarity with transient killer whales found in other regions of the Pacific. Genetic analysis A total of 93 skin samples were col- lected from 2001 through 2004 by using biopsy darting techniques. Separation of ecotypes based on mtDNA haplotype (Barrett-Lennard, 2000) showed that 47 of the 93 samples were of transient-type lineage, 42 were of the resident-type lineage, and 4 were of the offshore-type lineage. Preliminary classifications of lineages based Matkin et al,: Ecotypic variation and predatory behavior among Orcinus orca off the eastern Aleutian Islands 79 on morphology and behavior and determined from field observations and photographs were consistent with the genetic analysis. For the 26 encounters that yielded both genetic and acoustic data, the two kinds of data provided identical classifications of lineage. All 35 killer whales sampled in the False Pass- Unimak Island region had transient haplotypes. Eight of the nine samples collected in the area in 2003 had the GATl haplotype, which was first identified in transient killer whales from the northern Gulf of Alaska area in or near Kenai Fjords and Prince William Sound. The re- maining sample contained the ATI haplotype, formerly sequenced only in members of the ATI transient popula- tion of the Prince William Sound area. In 2004, 14 of 26 killer whales sampled in the False Pass-Unimak Island area had the GATl haplotype, and the remainder had a GAT2 haplotype, a similar but not identical haplotype known to exist at a low frequency in the Gulf of Alaska transient killer whale population (Barrett-Lennard, 2000). Eleven of the 12 transient whales sampled in the summer months during 2001-2004 in the eastern Aleutians had the GATl haplotype, and the remaining one had the GAT2 haplotype. Ecotypic parameters Resident killer whales During a majority of our encoun- ters, resident killer whales tended to be found and to travel along or near the 200-meter depth contour (Fig. 2A). This contour corresponds to a steep drop-off from the coastal shelf. Approximately 929c of the encounters with killer whales during our summer surveys from Unimak Pass to Umnak Island were with whales determined by ge- netics, acoustics, or group association to be of the resi- dent ecotype; however, this ecotype was not encountered in the spring surveys east of Unimak Pass (Table 1). A minimum of 901 resident whales used the Eastern Aleutians during the study; this count was based on individuals identified from photographs taken from 2001 through 2004. Of these individuals, 143 were seen only once during the study, and the remainder were repeatedly identified. The number of new, previously un-photographed whales observed each year declined from 534 whales in 2002 to 211 whales in 2003, to 156 whales in 2004 (Table 3). The decline in new whale sightings each year may indicate that we have identi- fied the majority of whales that use this area during the summer months; however, there may be hundreds of whales that occasionally use the area but have not been encountered. The study area is likely only a portion of the range of the identified resident whales; several whales were matched with individuals seen in photo- graphs taken in the Pribilof Islands over 200 miles to the west. The numbers of individuals that could be positively identified in each resident ecotype encounter ranged from 4 to 109. A total of 347 whales were placed in 82 tentative matrilines which consisted of a reproductive female and her offspring of both sexes. These matrilines Table 3 Number of sightings of individual resident killer whales ( Orcinus orca 1 in the eastern Aleutians from June to Sep- tember, 2001 through 2004. Year Previously identified whales New whales identified Total whales identified 2001 38 38 2002 38 496 534 2003 268 211 479 2004 334 156 490 Total 901 were determined from repeated association of individu- als in both photographs and field observations. This method of determining matrilines was demonstrated effective in other population studies of resident killer whales (Bigg et al., 1990; Matkin et al., 1999b). Most of the matrilines comprised two generations (mother and offspring), although some included a probable grand- mother. All matrilines were of consistent composition and maintained their structure over the course of the study, which has been the case in all other resident populations studied to date (Matkin et al., 1999a; Ford et al., 2000). The structure of the population was in- ferred from 41 groups of one or more matrilines that appeared to be longer-term associations. These groups could be considered as tentative pods (as defined by Bigg et al., 1990). Twenty-one of these groups, contain- ing 266 whales, were sighted frequently enough that basic age and sex classes could be determined. These groups contained 65 adult males (2A A^c), 105 females or immature males (39.5%), and 96 juveniles and calves (36.1%). These proportions of males, females and imma- ture males, and juveniles and calves are comparable to those observed in other resident populations in Alaska and British Columbia (Leatherwood et al., 1990). There was no evidence that resident killer whales consumed marine mammals. Whales belonging to the resident ecotype were observed consuming fish only during infrequent observations of predation (halibut were identified from samples, and salmon were probable from visual observations only). Much of the predation by resident killer whales was not visible at the sur- face and therefore prey samples could not be obtained. Resident whales were the only killer whales observed removing fish from the lines of commercial fishermen and observed following and feeding on fish discards from trawlers. Transient killer whales A total of 165 individual killer whales were determined to be transients from encoun- ters during 2001-2004 (Table 4). A majority of these whales (114) were photographed during the May-June field work in the False Pass-Unimak Island region in 80 Fishery Bulletin 105(1) A — Resident killer whale encounters V & BERING SEA w p. Jiil^^^^M^^HwS^^^^ Akutan ^<'^^ M Island Ji^^m ^ >Tih^ j""^ -t^j^T^ Unalaska Island Umnak ^_^k 54'=N- island ^^^HB' 10 20 168°W ' ' 1 ' 1 1 ' ^ Nautical Miles A 40 ° Alaska B — Transient killer whale encounters BERING SEA 4 ' Uninnak island Unimak Pass Bogoslof Island Umnak Island Akutan Unalaska Island PACIFIC OCEAN 54°N - 10 20 40 1BR°W ' I I I 1—1 I— I '°° ^^ Nautical Miles Figure 2 Tracks of the vessels during times that the vessels accompanied groups of killer whales (Orcinus orca) 2002-2004. (A) tracks of vessels following resident killer whales in the Eastern Aleutians during summer; (B) tracks of vessels following transient killer whales in the Eastern Aleutians during summer; (C) tracks of transient killer whales near False Pass- Unimak Island during spring. 2003 and 2004 (Fig. 2B). The remaining 51 individuals were photographed during the summer field season when encounters with transient killer whales were relatively infrequent in the eastern Aleutians from Unimak Pass west to Umnak Island (Fig. 2C). There were only six transient whales (less than 4% of the total identified) Matkln et al.: Ecotypic variation and predatory behavior among Orcinus orco off the eastern Aleutian Islands 81 — Transient killer whale encounters BERING SEA y *«i -^'#L, Alaska Peninsula / PACIFIC OCEAN 164°W L_ 5 10 20 ' ' ' I I I 1..^ Nautical Miles •^^■^ Alaska Figure 2 (continued) Table 4 Number of individual transient killer whales (Orcinus orca) identified by region and by year. And the overlap of individuals between regions. Regions False Pass Eastern Aleutians (May-June) (July-September) Year Total whales New whales Total whales New whales between regions 2001 _ _ 5 5 _ 2002 _ _ 18 18 — 2003 84 84 25 18 2 2004 75 30 22 16 4 Total 114 51 6 that were common to both regions and time periods, although the regions are geographically adjacent (Figs. 1 and 2). In the second year, 2004, in False Pass-Unimak Island, 74 whales were identified and 45 (60%) had been photographed the previous year. In the Unimak Pass-Umnak Island surveys in both 2003 and 2004, only about 28% of the whales identified had been previously photographed in the region. Offshore killer whales Only one encounter, which occurred in the eastern Aleutians (10 July 2003), was with killer whales identified by genetic and acoustic data as the offshore ecotype. We photographed 54 offshore killer whales in this encounter, although not all whales present were photographed. A total of 44 of these offshore whales had been previously photographed off British Columbia, Washington State, and Kenai Fjords, Alaska, and 10 had not been previously photographed (Ellis'). Description of the marine mammal prey base Although we did not measure the actual abundance of potential marine mammal prey, we recorded sightings of all marine mammals and calculated average group ^ Ellis, G. 2005. Unpubl. data. Fisheries and Oceans Canada, Pacific Biological Station, 3190 Hammond Bay Road, Nanaimo, British Columbia, Canada V9T 6A7. 82 Fishery Bulletin 105(1) Table 5 Number of sightings, harassments, and observed kills of known marine mammal prey species of killer whales (Orcinus orca). Prey species Total no. Average No. of No. of No. observed Prey species of sightings group size harassments observed kills to have escaped Gray whale (May-June only) 18' 1.4 19 18 1 Ball's porpoise 521 3.9 3 3 Steller sea lion 153 53.4 1 1 Humpback whale 834 4.1 1 Northern fur seal 388 4.2 5 4 1 Minke whale 42 1.2 2 2 Harbor porpoise 18 4.9 Harbor seal 82 9.6 Sea otter 94 3.7 California sea lion 2 1.0 Fin whale 15 2.2 Sperm whale 7 1.6 Baird's beaked whale 10 9.4 Total 31 26 5 ' This number comprises only live whales; most kills were already dead w hen observed and were not included in these sight ngs. size as an indication of the relative availability of poten- tial prey (Table 5). During the 346 survey days in the eastern Aleutians in June-September 2002-2004, the largest number of pinniped sightings was of northern fur seals (375 sightings), which were frequently encountered as single individuals resting at the surface, or were observed on the rookery at Bogoslof Island. Steller sea lions (124 sightings) were counted during our repeated monitoring of rookeries and haulouts in the region, resulting in a relatively large average group size (53.4 sea lions) but were encountered only occasionally off the rookeries. Other pinnipeds included harbor seals (73 sightings), which were primarily observed hauled out in small groups of less than 30 individuals, and two sightings of individual California sea lions (Zalophus californianus). Sea otters were observed on 81 occa- sions. The most frequently encountered cetacean was the humpback whale (Megaptera novaeangliae; 834 sight- ings), followed by Ball's porpoise (Phocoenoides dalli; 521 sightings). Less frequently observed species were minke whales (Baleanoptera acutorostrata; 42 sightings), fin whales (Baleanoptera physalus; 15 sightings), sperm whales (Physeter macrocephalus; 7 sightings), and Bairds beaked whales (Berardius bairdii; 10 sightings). During 49 survey days in the False Pass-Unimak Island region in May and early June 2003-2004, Steller sea lions were the most frequently ncountered pinniped (29 sightings), although some fur seals (13 sightings) and harbor seals (9 sightings) were also observed. Sea otters were also present (13 sightings). The most fre- quently encountered cetaceans were gray whales (18 sightings) and harbor porpoises (Phocoena phocoena; 18 sightings). Other cetaceans sighted in this region were minke whales (3 sightings) and humpback whales (1 sighting). Predation on marine mammals Predation on or harassment of marine mammals was observed on 31 occasions and was attributed solely to the transient killer whale ecotype. Attacks that did not yield proof of a kill (i.e., tissue, blood, or prey in the mouth) were considered harassments. Gray whales (Eschrichtius robustus) were the most frequently taken species (Table 5), with 19 apparent harassments, of which 18 resulted in kills, observed in False Pass in 2003 and 2004 (May to early June). In all but one case, the gray whale was already dead and was being consumed when we found the whale. The only other predatory event during this period was a single harassment of Steller sea lions that were hauled out. During the summer season in the east- ern Aleutians (Unimak Pass-Umnak Island), northern fur seals were the most frequently harassed prey; four of the five observed harassments resulted in kills. Other species included Dall's porpoises (harassed on three different occasions), two minke whales (harassed and killed), one Steller sea lion (harassed and killed), and one humpback whale (harassed). Discussion Our analysis indicates that the three killer whale eco- types in the eastern North Pacific also are found in the Matkin et al,: Ecotypic variation and predatory behavior among Orcinus orca off the eastern Aleutian Islands 83 eastern Aleutian Islands. All killer whales examined by acoustic or genetic analysis could be placed unequivo- cally in the resident, transient, or offshore lineage. Resident killer whales A vast majority of the whales observed in the eastern Aleutians during summer were determined to be of the resident ecotype. Our minimum estimate of 901 whales is as high as any density of resident killer whales in any region of the eastern North Pacific studied to date (Matkin et al., 1999a; Ford et al., 2000). Our data indicate that the eastern Aleutian resi- dent killer whales comprise a distinct population, but evidence is equivocal at this time. No repeated asso- ciations have been recorded between eastern Aleutian residents and those photographed off Kodiak Island north and eastward, despite extensive field effort and examination of photographic databases for matches. However, at least one group of resident killer whales has been photographed in both regions (Durban-j. Acoustic analysis indicates that resident whales sam- pled in the eastern Aleutians have call repertoires distinct from other well-known resident populations de- scribed from Kenai Fjords through Washington State. However, on the basis of structural similarities among calls from these regions, it can not be ruled out that some social contact occurs or that these whales share a recent common ancestry. The structure of some call syllables appears to change quickly in a climate of di- minishing social contact (Deecke et al., 2000), whereas the overall syllable type and the syntax of syllables in calls remains stable for a longer period of time (Ford, 1991; Yurk, 2005). Genetic samples taken in the east- ern Aleutians revealed only the NR haplotype, whereas those from Kodiak Island waters and in Kenai Fjords yielded a mixture of NR and southern resident hap- lotypes (SR haplotypes) (Barrett-Lennard, 2000). An examination of nuclear alleles is needed to clarify the relationship between eastern Aleutian residents and other resident killer whales in other regions of the North Pacific. Offshore killer whales Only one group of whales was determined to be of the offshore ecotype, with 54 individuals identified in a single encounter. Most (44) of the individuals identified in that encounter had been identified in other regions, including southern British Columbia and Kenai Fjords. Alaska, indicating that there is a single wide-ranging population in the eastern North Pacific. This ecotype is not known to consume marine mammals and the only reported stomach contents are salmonid bones, crab shell, sculpin, and eelgrass (Heise et al., 2003). Transient killer whales Transient killer whales, as determined in our analyses, were the only whales observed consuming marine mam- mals and were not seen feeding on fish or engaging in behaviors associated with fish-feeding in other areas (Ford and Ellis, 1999; Saulitis et al., 2000 ). This obser- vation supports findings in other regions that indicate transient killer whales are a distinct ecotype specializ- ing in marine mammal prey and comprise a subset of the total whales found in any region (Matkin et al., 1999b; Ford et al., 2000). Most of the 165 transient individu- als identified in our study were present only in spring and early summer when gray whales were migrating. We documented only 51 different transient individuals in late summer, at which time their appearance was sporadic and they seemed to leave the region for periods of weeks or longer. Transient killer whales in the eastern Aleutians display a unique call repertoire that is distinct from from the repertoire of transient killer whales in other regions. Therefore, the eastern Aleutian group may rep- resent a separate population. Other than three whales photographed near the Barren Islands by NMML in 2001 (Durban-), and resighted east of Unalaska Island in 2002; no other transient whales from this area have been photographed north and east of the Shumagin Islands. The results of mitochondrial DNA analysis are equivocal because the three haplotypes we identified all occur in waters of the northern Gulf of Alaska. Again, extensive examination of nuclear alleles and comparison with those from other regions will be needed to clarify population structure. For example, more detailed genetic analysis of eastern Aleutian transient killer whales exhibiting the ATI haplotype has shown that they have dissimilar nuclear alleles from those of the threatened ATI population of Prince William Sound and Kenai Fjords (Barrett-Len- nard'^). This finding indicates that the similarity of their haplotypes reflects historical lineage sorting rather than a recent descent from a common maternal ances- tor. Because haplotypes reflect maternal lineages, the co-occurrence of two haplotypes in the transient whale aggregations encountered in the False Pass-Unimak Island area during spring supports the idea that dis- tinct matrilines that may not associate at other times of year join to form these aggregations. The small overlap (3.6% of the individuals) between transient killer whales encountered west of Unimak Pass in summer and transient killer whales observed in spring in the False Pass-Unimak Islands area indicates there is further seasonal and spatial structuring in the population. The large percentage of new transient killer whales encountered in each year of summer studies, compared to the lower percentage of resighted individu- ^ Durban, J. 2005. Unpubl. data. National Marine Mammal Lab, National Marine Fisheries Service, 7600 Sand Point Way NE, Seattle, WA 98115. ^ Barrett-Lennard L. 2005. Unpubl. data. Vancouver Aquarium, 845 Aviso Rd. Vancouver, BC, Canada.V6G 3E2 84 Fishery Bulletin 105(1) als in the spring, indicates that a smaller percentage of the summer whales have been identified. A survey of nearshore waters from the Gulf of Alaska to the Aleutian Islands revealed that the highest densities of transient killer whales were from the Shumagin Islands through the eastern Aleutian Islands, and an estimated abundance of 226 (CV=0.45) transient killer whales were present west of the Shumagin Islands in summer (Zerbini et al., 2006). Transient killer whales near False Pass in May were concentrated on the Pacific Ocean side of Unimak Is- land and in Ikatan Bay, where gray whales pass along a shallow shelf and water depth rarely exceeds 70 me- ters (Fig. 1). Reports from mariners and pilots have indicated that other areas around Unimak Island and along the Bering Sea coast (e.g.. Cape Lutke and the coastline near Nelson Lagoon) may also be points of interception of gray whales by killer whales. Our own surveys (Fig. 2B) indicate that transient killer whales inhabit a wide area around Unimak Island and the tip of the Alaska Peninsula, where we recorded kills of gray whales at Deer Island (110 km northeast of False Pass) and Cape Lutke (140 km southwest of False Pass). Despite uncertainties regarding the range of tran- sient killer whales, it is evident that they are numerous, concentrated, and consistently present in the spring from Unimak Pass eastward. Gray whales have been previously reported as killer whale prey (Matkin and Saulitis, 1994); however, the extent to which transient killer whales were focused on gray whale predation during May-June around Unimak Island has not been previously described. Although subsequent surveys in these areas during summer (C. O. Matkin, unpubl. data; Durban'-.) have identified some of the same whales as those identified in the spring, most of the whales do not remain in these nearshore waters. It is not known whether these transient whales move offshore and dis- perse, follow the gray whales into the Bering Sea, or move into other unstudied regions. The distribution of most, if not all, transient killer whales that we identified undoubtedly extends well beyond our survey area. Technical advances in satellite and radio tagging procedures that could be applied to killer whales would aid considerably in understanding the movements and range of transients in this region. Without a better understanding of the range of these whales, it is impossible to fully assess their impact on prey populations. Northern fur seals appear to be an important prey for killer whales from late June to September west of Unimak Pass. This finding is based on observed kills compared to kills of other species. A substantial number of the fur seals sighted in summer were likely associated with the recently established and expanding fur seal rookery on Bogoslof Island. This population increased rapidly from 898 pups in 1992 to 5096 pups in 1999 (Angliss and Lodge, 2004). Additionally, peak numbers of migrating fur seals pass through Unimak Pass into the Bering Sea in June on their way to the Pribilof Islands and then migrate back to the Pacific in peak numbers during October-November (Bigg, 1990). Northern fur seals have long been indicated as an important prey for killer whales in the Pribilof Region (Hanna, 1923; Zenkovich, 1938; Tomilin, 1957). How- ever, their importance as prey in the eastern Aleutians has not been previously documented and may have de- veloped with the growth of the Bogoslof rookery. This geographic region presents an opportunity to examine the effects of killer whale predation on an apparently stable or increasing population of fur seals — a species that is declining in other areas. Minke whales made up a substantial proportion of summer predation despite the relative low frequency with which they were sighted, and the apparent dif- ficulty that killer whales have in capturing this fast swimming species in open water (Ford et al., 2005). Minke whales appear to be a minor part of the diet of killer whales from Washington State to northern south- eastern Alaska (Ford et al., 2005). We observed the harassment of a humpback whale by killer whales once; during the attack, other humpback whales rapidly converged on the attackers and appeared to drive the killer whales away. No injuries were appar- ent. Harassments of humpbacks have been reported in other regions of Alaska (Saulitis et al., 2000), but did not result in a kill or apparent injury. Photographs of scars indicate that most killer whale attacks on baleen whales target young animals, probably calves on their first migration from low-latitude breeding and calving areas to high-latitude feeding grounds (Mehta^). Although none of the attacks that we observed on Dall's porpoises resulted in confirmed kills, Dall's por- poises could be a significant prey as has been indicated in other regions (Ford et al., 1998; Saulitis et al., 2000); however, more observations are needed. Harbor seals were conspicuously absent from our prey observations despite also being an important prey in other regions of Alaska (Saulitis et al., 2000; Matkin et al., in press). Harbor seals are found in relatively low numbers in the eastern Aleutians. Although Steller sea lions were observed as prey on one occasion and were harassed on another, they did not appear to be a primary target of the transient killer whales we observed during our spring and summer surveys. Whether or not they are an important prey during other seasons (fall and winter) is not known and will require additional study or the application of other methods to be fully assessed. Although our study was limited by a small sample size in the summer, it provided significant information on the distribution of transient killer whale prey and the importance of fur seals in the killer whale diet dur- ing summer west of Unimak Pass and the importance of grey whales in killer whale diet during spring from Unimak Pass east. An increase in sample size of ob- ^ Mehta, A. 2005. Unpubl. data. Woods Hole Oceanographic Institute, Woods Hole, MA 02543-1050. Matkin et al.: Ecotypic variation and predatory behavior among Orcinus orca off tfie eastern Aleutian Islands 85 served kills is therefore important to develop greater confidence and detail in estimating the composition of killer whale diets. Our seasonal bias towards spring and summer leaves uncertainty about killer whale diets during fall and winter. In this regard, analytical techniques that in- clude identification of fatty acids, stable isotopes, and contaminants may prove useful when coupled with field observations to obtain a more complete picture of the feeding habits of killer whales during these seasons (Herman et al., 2005). Conclusions Our work underscores the importance of determining lineages and ecotypes of killer whales before making assumptions regarding feeding habits and potential impact of killer whales on prey populations. Although there may be well over 100 marine mammal-eating tran- sient killer whales that aggregate in False Pass-Unimak Island region to feed on gray whales in spring, the major- ity of the killer whales present in summer are fish-eating residents. In the summer, marine-mammal-eating tran- sients are far less abundant than in spring. Our study indicates that the diet of transient killer whales off the eastern Aleutian Islands contrasts with the diets of transient killer whales in other parts of the North Pacific. In British Columbia, for example, transient killer whale diet is composed primarily of harbor seals (Ford et al., 1998), whereas both harbor seals and harbor porpoise are the primary prey of killer whales in northern Glacier Bay and Icy Strait region of southeastern Alaska (Matkin et al., 2005). Further north, in Prince William Sound and Kenai Fjords, the dominant prey of the ATI transient killer whales are harbor seals and Ball's porpoises (Saulitis et al., 2000). Only in the Gulf of Alaska (Kenai Fjords) has predation by some Gulf of Alaska transient killer whales appar- ently focused on sea lions (Matkin et al., 2005). Killer whale feeding behavior needs to be examined on a region-by-region basis, as well as seasonally. Expe- rience in other regions of the North Pacific has shown that estimated population sizes, life history param- eters, and dietary information can be obtained with a concerted long-term research effort. Our study has demonstrated that the eastern Aleutians also support the presence of three killer whale ecotypes, as has been previously described along the Pacific Coast of North America. It also has developed minimum estimates of the numbers of transient and resident killer whales that use the region and has provided information that may indicate that grey whales and northern fur seals are important prey items in this region at certain times and in certain areas. Steller sea lions were not a primary prey during our spring and summer surveys. Whether or not killer whales are impeding population recovery of Steller sea lions in the eastern Aleutian Islands cannot be answered decisively, nor can the effect that killer whales may be having on other species in this region as yet be ascertained. Answers to these and other ques- tions are expected to become clearer as the observation- al database for killer whales is expanded and the data for regions within the North Pacific are compared. Acknowledgments Financial support was provided by the Cooperative Insti- tute for Arctic Research (CIFAR), the National Marine Mammal Laboratory, the National Oceanic and Atmo- spheric Administration, the Steller Sea Lion Research Initiative (SSLRI), the North Pacific Universities Marine Mammal Research Consortium (NPUMMRC), the North Pacific Marine Science Foundation, and the Alaska Sea Life Center (ASLC). Field biologists included P. Nielson, T. Markowitz, D. Power, and L. Mazzuca and vessel operations were directed by M. Brittain. Vessel operators were B. Schauff, P. Thompson, B. Laukitis, D. Graves, T. Peterson, A. Bartlett, and G. Marshall. Assistance in the field was provided by L.A. Holmes, R. Brewer, V. Vergara, T. Ciosek, G. Calef, C. Bianchini, A. G. MacHutchon, K. Leonard, F. Nicklin, K. Walker, D. Masters, and R. Blancato. Office and logistic support were provided by P. Rosenbaum and J. Rose. Literature cited Angliss, R. P.. and K. L. Lodge. 2004. Alaska marine mammal stock assessments, 2003. NOAA Technical Memo. NMFS-AFSC-144, 237 p. Barrett-Lennard, L. G. 2000. Population structure and mating patterns of kil- ler whales (Orcinus orca) as revealed by DNA anal- ysis. Ph.D. diss., 92 p. Univ. British Columbia, Van- couver, B.C., Canada. Barrett-Lennard, L. G., K. A. Heise, E. Saulitis, G. Ellis, and C. Matkin 1995. The impact of killer whale predation on Steller sea lion populations in British Columbia and Alaska. Re- port of the Marine Mammal Research Consortium, Univ. British Columbia, 6248 Biological Sciences Road, Van- couver, BC, V6T-1Z4, Canada. Barrett-Lennard, L. G., T G. Smith, and G. M. Ellis. 1996. A cetacean biopsy system using lightweight pneu- matic darts, and its effect on the behavior of killer whales. Mar. Mamm. Sci. 12:14-27. Bigg. M. A. 1990. Migration of northern fur seals (Callorhinus ursi- nus) off western North America. Canadian Technical Report of Fisheries and Aquatic Sciences no. 1764, 64 p. Bigg, M. A., P F. Olesiuk, G. M. Ellis. J. K. B. Ford, and K. C. Balcomb IIL 1990. Social organization and geneaology of resident killer whales (Orcinus orca) in the coastal waters of British Columbia and Washington State. Report of the International Whaling Commission, (special issue) 12:386-406. Dahlheim, M. E. 1997. A Photographic catalog of killer whales, Orcinus orca, from the central Gulf of Alaska to the southeastern Bering Sea. NOAA Tech. Rep. NMFS-131, 54 p. 86 Fishery Bulletin 105(1) Deecke, V. B. 2003. The vocal behaviour of transient killer whales (Orcinus orca) communicating with costly calls. Ph.D. diss. 138 p. Univ. St. Andrews, St. Andrews, Scotland. Deecke, V. B., J. K. B. Ford, P. J. B. Slater. 2005. The vocal behavior of mammal-eating killer whales: communicating with costly calls. Anim. Behav. 69(2):395-405. Deecke, V. B., J. K. B. Ford, and P. Spong. 2000. Dialect change in resident killer whales: implica- tions for vocal learning and cultural transmission. Anim. Behav. 60(51:629-638. DeMaster, D. P., A. W. Trites, P. Clapham, S. Mizroch, P. Wade, R. J. Small, and J. Ver Hoef 2006. The sequential megafaunal collapse hypothesis: testing with existing data. Prog. Ocean. 68:329-342 Estes, J. A., M. T. Tinker, T. M. Williams, and D. F. Doak. 1998. Killer whale predation on sea otters links oceanic and nearshore ecosystems. Science 282:473-476. Ford, J. K. B. 1991. Vocal traditions among resident killer whales {Orci- nus orca) in coastal waters of British Columbia. Can. J. Zool. 69:1454-1483. 1989. Acoustic behaviour of resident killer whales {Orci- nus orca ) off Vancouver Island, British Columbia. Can J. Zool. 67:727-745. Ford, J. K. B., G. M. Ellis, D. R. Matkin, K. C. Balcomb, D. Briggs, and A. B. Morton. 2005. Killer whale attacks on minke whales: prey cap- ture and antipredator tactics. Mar. Mamm. Sci. 21: 603-618. Ford, J. K. B., and G. M. Ellis. 1999. Transients: Mammal-hunting killer whales, 96 p. Univ. British Columbia Press, Vancouver, Canada. Ford, J. K. B., G. M. Ellis, and K. C. Balcomb. 2000. Killer whales, 2"'' ed., 104 p. Univ. British Colum- bia Press, Vancouver, Canada. Ford J. K. B., G. M. Ellis, L. G. Barrett-Lennard, A. B. Morton, and K. C. Balcomb III. 1998. Dietary specialization in two sympatric popula- tions of killer whales (Orcinus orca) in coastal British Columbia and adjacent waters. Can. J. Zool. 76:1456- 1471. Hanna, G. D. 1923. Rare mammals of the Pribilof Islands, Alaska. J. Mammal. 4:209-215. Heise, K., L. G. Barrett-Lennard, E. Saulitis, C. Matkin, and D. Bain. 2003. Examining the evidence for killer whale pre- dation on Steller sea lions in British Columbia and Alaska. Aquat. Mamm. 29:325-334. Herman, D. P., D. G. Burrows, P. R. Wade, J. W. Durban, C. O. Matkin, R. G. LeDuc, L. G. Barrett-Lennard, and M. M. Krahn. 2005. Feeding ecology of eastern North Pacific killer whales Orcinus orca from fatty acid, stable isotope, and organochlorine analyses of blubber biopsies. Mar Ecol. Prog. Ser. 302:275-291. Leatherwood, S., C. O. Matkin, J. D. Hall, and G. M. Ellis. 1990. Killer whales, Orcinus orca, photo-identified in Prince William Sound, Alaska, 1976 through 1987. Can. Field Nat. 104:362-371 Loughlin, T. R., and A. E. York. 2000. An accounting of the sources of Steller sea lion mortality. Mar. Fish. Rev. 62(4):40-45. Matkin, C. O., G. M. Ellis, P. Olesiuk, and E. L. Saulitis. 1999a. Association patterns and genealogies of resident killer whales (Orcinus orca) in Prince William Sound, Alaska. Fish. Bull. 97:900-919. Matkin, C. O., G. M. Ellis, E. L. Saulitis, L. G. Barrett-Lennard. and D. Matkin. 1999b. Killer whales of zouthern Alaska. North Gulf Oceanic Society, Homer, AK. Matkin, C. O., and E. Saulitis. 1994. Killer whale (Orcinus orca): biology and m.anagement in Alaska. U.S. Mar. Mamm. Comm., Washington D.C., 80 p. [Contract T75135023.1 Matkin, C. O., E. Saulitis, D. Maldini, J. Maniscalco, and L. Mazzuca. 2005. Steller sea lion predation by killer whales in Kenai Fjords/Prince William Sound, Alaska. In Syn- opsis of research on Steller sea lions: 2001-2005 (T. R. Loughlin, S. K. Atkinson, and D. G. Calkins, eds.), p. 212-226. Alaska Sea Life Center's Steller Sea Lion Research Program, Seward, AK. Matkin, D. R., J. M. Straley, and C. M. Gabriele. In press. Killer whale feeding ecology and non-predatory interactions with other marine mammals. In Proceed- ings of the fourth Glacier Bay science symposium, 2004 (J. F. Piatt, and S. M. Gende, eds.). U.S. Geological Survey, Information and Technology Report USGS/BRD/ ITR-2005. Washington, D.C. Merrick, R. L., T R. Loughlin, and D. G. Calkins. 1987. Decline in abundance of the northern sea lion, Eume- topias jubatus. in 1956-86. Fish. Bull. 85:351-365. Miller, P. J. O., and D. E. Bain. 2000. Within-pod variation in the sound production of a pod of killer whales, Orcinus orca. Anim. Behav. 60(5):617-628. Mizroch, S. A., and D. W. Rice. 2006. Have North Pacific killer whales switched prey species in response to depletion of the great whale populations? Mar. Ecol. Prog. Ser. 310:235-246. Pitcher, K. W. 1990. Major decline in number of harbor seals, Phoca vitu- lina richardsi, on Tugidak Island, Gulf of Alaska. Mar. Mamm. Sci. 6:121-134. Sambrook, J., E. F. Fritsch, and T. Maniatis. 1989. Molecular cloning: a laboratory manual. 3'''' ed. Cold Spring Harbor, NY. Saulitis, E., C. O. Matkin, and F. H. Fay. 2005. Vocal repertoire and acoustic behavior of the isolated ATI killer whale subpopulation in southoin Alaska. Can. J. Zool. 83:1015-1029. Saulitis, E. L., C. O.Matkin, K. Heise, L. Barrett Lennard, and G. M. Ellis. 2000. Foraging strategies of sympatric killer whale (Orcinus orca) populations in Prince William Sound, Alaska. Mar. Mamm. Sci. 6:94-109. Springer, A. M., J. A. Estes, G. B van Vliet, T. M. Williams, D. F. Doak, E. M. Danner, K. A. Forney, and B. Pfister. 2003. Sequential megafaunal collapse in the North Pacific Ocean: an ongoing legacy of industrial whaling? Publ. Nat. Acad. Sci. 100:12,223-12,228. Thompson, J. D., D. G. Higgins, and J. Gibson. 1994. Clustal W: improving the sensitivity of progressive multiple sequence alignment through sequence weight- Matkin et al.: Ecotypic variation and predatory behavior among Orcinus orca off tfie eastern Aleutian Islands 87 ing position specific gap penalties and weight matrix choice. Nucl. Acids Res. 22:4673-4680 Tomilin, A. G. 1957. Cetacea. Vol. 9 of Mammals of the U.S.S.R. and adjacent countries (V. G. Heptner, ed.l, 756 p. Izd. Akad. Nauk SSSR, Moscow. [Translated from Russian by Israel Prog. Sci Transl., 1967.] Trites, A. W. 1992. Northern fur seals: why have they declined? Aquat. Mamm. 18:3-18. Trites, A. W., V. B. Deecke. E. J. Gregr, J. K. B. Ford, and P. F. Olesiuk. 2006. Killer whales, whaling and sequential megafaunal collapse in the North Pacific: a comparative analysis of the dynamics of marine mammals in Alaska and British Columbia following commercial whaling. Mar. Mamm. Sci. 22. Trites, A. W., and P. A. Larkin. 1996. Changes in the abundance of Steller sea lions (Eumetopias jubatus) in Alaska from 1956 to 1992: How many were there? Aquat. Mamm. 153-166. Winship, A. J., and A. W. Trites. 2006. Risk of extirpation of Steller sea lions in the Gulf of Alaska and Aleutian Islands: a population viability analysis based on alternative hypotheses for why sea lions declined in western Alaska. Mar. Mamm. Sci. 22:124-155 York, A. E. 1987. Northern fur seal, Callorhmus ursinus, eastern Pacific population (Pribilof Islands, Alaska, and San Miguel Island, California). In Status, biology, and ecol- ogy of fur seals (J. P. Croxall, and R. L. Gentry, eds.), p. 9-21. NOAA Tech. Rept. NMFS 51. Yurk, H. 2005. Vocal culture and social stability in resident killer whales (Orcinus orca) of the northeastern Pacific. Ph. D. diss., 126 p. Univ. British Columbia, Vancouver, British Columbia, Canada. Yurk, H., L. G. Barrett-Lennard, J. K. B Ford, C. O. Matkin. 2002. Cultural transmission within maternal lineages: vocal clans in resident killer whales in southern Alaska. Anim. Behav. 63(6):1103-1119. Zenkovich, B. A. 1938. On the grampus or killer whale (Grampus orca Lin.). Priroda 4:109-112. [Translated from Russian by L. G. Robbins. U.S. Geological Survey S(570). Zerbini, A. N., J. M. Waite, J. W. Durban, R. LeDuc, M. E. Dahlheim, and P. R. Wade. 2006. Estimating abundance of killer whales in the nearshore waters of the Gulf of Alaska and Aleutian Islands using line transect sampling. Mar. Biol. DOI 10.1007/S0027006-0347-8. Abstract— Abundance indices derived from fishery-independent surveys typically exhibit much higher inter- annual variability than is consistent with the within-survey variance or the life history of a species. This extra variability is essentially observation noise (i.e. measurement error); it prob- ably reflects environmentally driven factors that affect catchability over time. Unfortunately, high observa- tion noise reduces the ability to detect important changes in the underlying population abundance. In our study, a noise-reduction technique for uncor- related observation noise that is based on autoregressive integrated moving average (ARIMA) time series mod- eling is investigated. The approach is applied to 18 time series of fin- fish abundance, which were derived from trawl survey data from the U.S. northeast continental shelf. Although the a priori assumption of a random- walk-plus-uncorrelated-noise model generally yielded a smoothed result that is pleasing to the eye, we rec- ommend that the most appropriate ARIMA model be identified for the observed time series if the smoothed time series will be used for further analysis of the population dynamics of a species. Removing observational noise from fisheries-independent time series data using ARIMA models William T. Stockhausen (contact author) Michael J. Fogarty Email address for W. T, Stockhausen; William. Stockhausenta'noaa.gov National Ocean and Atmospheric Administration National Marine Fisheries Service Northeast Fisheries Science Center 166 Water Street Woods Hole, MA 02543 Present address for corresponding author; National Marine Fisheries Service Alaska Fisheries Science Center 7600 Sand Point Way NE Seattle, Washington 98115 Manuscript submitted 12 November 2004 to the Scientific Editor's Office. Manuscript approved for publication 18 April 2006 by the Scientific Editor. Fish. Bull. 107:88-101 (2007). Time series of species abundance from fishery-independent surveys, such as bottom trawl or acoustic surveys, are important in monitoring temporal change in the abundance of marine populations. For commercially impor- tant species, catch and effort data from the commercial fishery may be available, allowing estimation of tem- poral trends of the stock population by means of stock assessment models (e.g., virtual population analysis). However, such records are not avail- able for many species, especially those with little commercial (but perhaps significant ecological) value. Fishery- independent surveys may thus con- stitute the only source of information for assessing temporal changes in the abundance of these species (Penning- ton, 1985; Helser and Hayes, 1995). Annual estimates of abundance de- rived from fisheries-independent sur- veys are typically regarded as provid- ing a relative measure of population abundance (i.e., they are indices of abundance, not true estimates of to- tal population size) (Grosslein, 1969; Clark, 1979). Thus, the expected value of the abundance index (e.g., mean catch-per-tow for trawl surveys) is regarded as proportional to the size of the actual population, although the constant of proportionality (the catch- ability) is unknown. As such, rela- tive changes in an abundance index should reflect similar relative changes in the actual population, and trends in the time series of such an index should reflect similar trends in the corresponding population. Unfortunately, abundance indices derived from large-scale fishery-in- dependent surveys typically exhibit interannual variability much higher than one would expect from within- survey variance (Byrne et al., 1981; Pennington, 1985). Part of the vari- ability in such indices is presumably due to the variability in the underly- ing population — a variability that is caused by population-dynamic pro- cesses such as recruitment. However, part of the variability is due to ob- servation noise that arises from both within-survey sampling variability because of the heterogeneous distri- bution of many fish stocks (Byrne et al., 1981), and because of environmen- tally driven factors that affect catch- ability over time (Byrne et al., 1981; Collie and Sissenwine, 1983). Low signal-to-noise ratios in abundance indices that are due to high observa- tion noise reduce chances of detecting important changes or trends in actual population abundance. Variability due to within-survey sampling can be re- duced (before the fact) by adding more stations to a survey, but additional stations will not reduce variability caused by changes in catchability. Time series modeling using autore- gressive integrated moving average (ARIMA; Box and Jenkins, 1976) models provides an approach to re- Stockhausen and Fogarty: Removing observational noise from time series data using ARIMA models 89 moving observation noise from abundance estimates. ARIMA models are frequently used in economic fore- casting (Enders, 2004) and are becoming more common in fisheries research. Recent applications of ARIMA models to other fisheries problems include forecasting monthly landings in the Mediterranean (Lloret et al., 2000), testing theories of population dynamics (Becerra- Munoz et al., 1999), and modeling nutrient dynamics in an upwelling system (Nogueira et al., 1998). In the context of reducing the influence of observa- tional noise in time series data, Cleveland and Tiao (1976) first developed a noise-reduction and smoothing algorithm for processes that could be described by an ARIMA time series model. Their approach requires that the ARIMA model for the unobserved, underlying process be known. This known model, in turn, uniquely determines the ARIMA model for the observed time se- ries contaminated by observation noise and allows one to estimate the variance of the observation noise. Unfor- tunately, although the ARIMA model for an unobserved, underlying process may be known in some instances (from theory, perhaps), in many cases the model for the unobserved process will be unknown. Box et al. (1978) extended Cleveland and Tiao's (1976) ideas and developed a noise-reduction algorithm based on the ARIMA model for the observed time series. How- ever, the ARIMA model for the observed time series merely constrains, but does not determine, the model for the unobserved, underlying process; it provides only an upper bound for the observation error variance. Con- sequently, this approach generally requires an external estimate of the observation error variance to determine the appropriate level of noise reduction. Pennington (1985) first applied these ARIMA-based time series modeling techniques to smoothing abun- dance indices derived from trawl survey data. He as- sumed that an observed abundance time series reflected a combination of the underlying population abundance and independent, uncorrelated, and multiplicative ob- servation noise (the latter arising perhaps from envi- ronmentally driven changes in catchabilityl. He further assumed that both the (log-transformed) observed time series and unobserved population process could be rep- resented by ARIMA models. Pennington (1985) then developed an alternative algorithm to that of Box et al. (1978); his derivation allowed particular simplification in the case where the underlying population process could be modeled as a random walk. In this simple case, the resulting noise reduction filter is an exponentially- weighted average of the observed time series for the endpoint of the time series (Pennington, 1985). More importantly, the observation error variance can be eas- ily estimated from the ARIMA model parameters and an external estimate is unnecessary. Thus, for the case where a random walk model for the underlying process is valid, the appropriate level of smoothing is objectively determined. As a demonstration, Pennington (1985) applied his noise reduction algorithm to groundfish trawl survey data for haddock (Melanogramnius aeglefinus) from the northeastern Atlantic coast of the United States. He found that the variances of the smoothed indices were "considerably lower" than those of the originals. How- ever, this demonstration used an ARIMA model derived from a much longer time series that had been generated from a stock assessment based on commercial catch data. Pennington (1985) assumed that this model rep- resented the underlying population and therefore did not develop models based on the observed time series. Although this assumption was perfectly reasonable, given that such alternative data (the stock assessment) were available, it cannot be applied to situations when only survey data are available to fishery analysts. The ARIMA model Pennington (1985) derived from stock assessment results was a random walk model; therefore the appropriate level of noise reduction for the corresponding survey data could be objectively deter- mined from the model parameters. Pennington's (1985) method was later used to apply random walk mod- els to survey data (Fogarty et al.^; Pennington, 1986; Anonymous, 1988, 1993). Pennington (1986) found that random walk models were appropriate for the survey time series considered in his study. However, random walk models were assumed a priori in the remaining three references (Fogarty et al.^; Anonymous, 1988, 1993) to generate smoothed abundance trajectories; because less than 25 observations for each time series were considered in these references, reliable identifica- tion of the model structure for each time series was considered problematic and random walk models were used as "null" models. When it is an appropriate description of the underly- ing process, a random walk model yields an objective determination of the degree of noise reduction appro- priate to an observed time series. However, an a priori adoption of this model should be viewed with some skep- ticism. Additionally, if a random walk model is not an appropriate description of the underlying process, the resulting smoothed time series may seem reasonable, but the result no longer has support as the unobserved, underlying process. In this circumstance, we regard the effect of the ARIMA algorithm as merely smoothing, and not necessarily as noise reducing. As such, we feel that the utility of ARIMA-based approaches to noise reduction for abundance indices derived from survey data has not been adequately ex- plored to date. In addition, substantially longer time series (e.g., 40 observations) are now available with which to test this concept. In our study, we test the utility of the ARIMA time series noise reduction ap- proach propounded by Pennington (1985), using time series of abundance indices from fishery-independent trawl survey data for nine finfish species (Table 1) during two seasons on Georges Bank. We first review the original methods developed by Cleveland and Tiao 1 Fogarty, M. J., J. S. Idoine, F. P. Almeida, and M. Pen- nington. 1986. Modeling trends in abundance based on research vessel surveys. ICES CM (council meeting) 1986/G, p. 92. ICES, Copenhagen, Denmark. 90 Fishery Bulletin 105(1) Table 1 Time series of abundance indices for the following finfish species on Georges Bank were derived from a fisheries-independent trawl survey and used to test the ARIMA-based smoothing algorithm. Common name Scientific name Type Winter skate Leucoraja ocellata Elasmobranch Little skate Leucoraja erinacea Elasmobranch Silver hake Merluccius bilineahs Groundfish Atlantic cod Gadus morhua Groundfish Haddock Melanogrammus aeglefinus Groundfish Winter flounder Pseudopleurenectes ainericanus Flatfish Yellowtail flounder Lima/ida ferruginea Flatfish Atlantic herring Clupea harengus Pelagic schooling fish Atlantic mackerel Sco/nber scombrus Pelagic schooling fish (1976) and Box et al. (1978). Framing the problem in terms of power spectra, we also offer some additional new insights into this noise reduction approach. Next, we apply the ARIMA-based noise reduction approach to the time series data and present the results. We have implemented Box et al.'s (1978) algorithm, not Pennington's (1985). Finally, we discuss our perceptions of the utility of this approach in light of our results and overall experience with it. Materials and methods General characteristics of ARIMA models In this section, we first briefly review ARIMA models for stochastic processes. Then we review the approach of Box et al. (1978) for obtaining maximum likelihood estimates for an underlying ARIMA time series from a time series of observations with independent and identi- cally distributed (IID) observation noise. ARIMA models are parsimonious models that can adequately represent many stochastic time series (Box and Jenkins, 1976). Stochastic time series that can be represented by ARIMA models are essentially the out- put of a linear filter applied to an input time series of white noise (Box and Jenkins, 1976). We will refer to such time series as ARIMA processes. For a zero-mean stochastic time series |2,| that can be expressed as an ARIMA model, we denote the model (using the notation of Box and Jenkins, 1976) as (p(B)z, =a(B)ai, (1) where z (p(B) the value of the time series at time t; the generalized autoregressive (AR) opera- tor; a(B} = the moving average (MA) operator; B = the backward shift operator; and a, = IID normally distributed random variables with mean zero and variance o~. The backward shift operator B has the property that B 2f = 2, p hence B'" 2, = 2,_,„. The operators q>{B) and a(B) are polynomials in B of order p+d and q (respec- tively) such that (p(B) = l-Y,'(=3,+e,. (4) where the z/s represent the unobserved, underlying process and the e/s are IID normal variables with vari- ance Op2 (i.e., e~N(0,a^^)) that are also independent of the £,'s (i.e., < e 2^> = for allj,/j). The goal is to estimate the unobserved time series |2,| by using the observed time series |,y,|. For the analysis of fishery-independent time series, it seems reasonable to assume that only the ARIMA model for the observed time series |y,| is known (it can be estimated using standard techniques). In particular, this assumption means that the model for |2,| is unknown within constraints implied by the observation equation. However, to develop the approach used here it is helpful to start as though the model for the unobserved process \z,\ were known. Thus, we assume that the time series {2,1 can be rep- resented by an ARIMA (p,d.q) process: In this equation, a^, (j(B),and (piB) are known from the ARIMA model for the observed process, whereas o^, a'^, and a(B) are unknown. In general, many combinations of a]-, a~, and a(B) will satisfy the equality. Defining an "acceptable model" for the unobserved process as one that, given the model for the observed process, a{B) satisfies the previous equation and its zeros are on or outside the unit circle, Box et al. (1978) show that 1) for every given model of an observed process, at least one acceptable model for the unobserved process exists; 2) for a given model of an observed process, the possible values of o^^ are bounded; and 3) for a given model, every o^^ between and the upper bound (K . say) determines a unique acceptable model. The upper bound on the observation error variance, K\ is determined from the constraint that, for a model of the unobserved process to be accept- able, a^-a(B)a(F)2.Q everywhere on the unit circle (i.e., the power spectrum of the corresponding MA process is non-negative definite). Then, from equation 8, K is given by (p{B)z,=a{B)c,, (5) where the c,'s are IID and c, ~N(0,a;). Substituting in y-e, for 2, and rearranging, one obtains (p{B)y,=a{B)c,+(p(B)e,, (6) which can be expressed as an ARIMA model for |y,) of the form (p(B)y, = ri{B)d,. (7) where the c//s are IID, c?,~N(0,a^,) and the MA opera- tor is i]{B). Thus, the generalized AR operator for the observed |y,| is identical to that for the unobserved {2,). Furthermore, because a(B) has order q and (p{B) ^. . la^riiBjnlF] K = nun < —^ |B|=i (p{B)(p{F) (9) and is completely determined by the ARIMA model for the observed process. When a^,- = K , the variance of the added white noise is maximal, as will be the smoothing of the observed time series. It is instructive to interpret Equations 8 and 9 in terms of constraints on the power spectra of (2,}, [y,], and le,l, although this interpretation is strictly correct only when {2,) and |y,| are stationary. Let p^{f), p^lf), p^(f) denote the power spectra for [z,], (y,), and |e,}, respectively. Recalling the definition of the power spec- trum (Eq. 3), Equation 8 on the unit circle can be easily recast (multiply both sides by 2l[(p{B)(p{F)]} as 92 Fishery Bulletin 105(1) PAf) = Py(f)-pM) because the power spectrum for white noise is constant with frequency: p^{F)=2o~. Because power spectra are nonnegative definite and p/f) does not depend on f, the maximum possible observation noise variance K corre- sponds to the minimum of p^,{f) over /"(see Fig. 1). Thus, Equation 9 can be recast as (10) 2p which we denote as Elzjy), is a symmetric moving average filter of {y,| (Cleveland and Tiao, 1976; Box et al., 1978): E{z,\y) = (0(B}y„ K'= min {pAf)]/2. 02 exhibited substantially less smoothing than models with a MA order ^2. Models that were of MA order 1 generally resulted in the greatest smoothing. Models that were of MA order did not (and could not) occur. Of the 18 time series we considered (Tables 2 and 3, Figs. 2 and 3), only half were adequately represented as random-walk-plus-uncorrelated-noise (RWPUN) models. The ARIMA models we developed were varied in structure, ranging from a simple MA(1) model to rather complicated models with multiple parameters. Thus, our results provide evidence against the appro- priateness of assuming a particular model structure a priori when the objective of the analysis is to identify the underlying dynamic structure of the population. This evidence is further strengthened by the results of Becerra-Munoz et al. (1999), who found only 9 of 52 abundance time series for finfish species from the NMFS/NEFSC bottom trawl survey that corresponded to random walk models. As an exercise, we also attempted to smooth the nine data sets that were not adequately described as RW- PUN models, using this model structure as an a priori assumption, even though our analysis indicated that other models were more appropriate. We were not able to estimate convergent models for three species: At- lantic cod (fall), winter flounder (spring), and Atlantic mackerel (spring). For the remaining six time series, the smoothed results appeared to be quite reasonable (Fig. 4), although we obtained little noise reduction when we employed the "correct" ARIMA model. The RWPUN-smoothed time series for haddock (Fig. 4, C and F) were similar to that for spawning biomass de- rived from virtual population analysis (see Brodziak et al.'-), but the smoothed time series for silver hake (Fig. 4E) exhibited higher frequency variability than that found for total biomass with a production model (see Brodziak et al.'^). From the standpoint of estimat- ing the unobserved underlying process, these smoothed results should be viewed with some skepticism: the use of the RWPUN model is rather arbitrary in this situation and it may impose artificial structure on the smoothed results. However, it may be that these time series do not meet one of the key assumptions of the noise reduction method: namely that the observation noise is uncorrelated. The ARIMA models for all six time series had MA orders a3, and one effect of cor- related observation noise could be to increase the MA - Brodziak, J., M. Traver and L. Col. 2005. Georges Bank haddock. In Assessment of 19 northeast groundfish stocks through 2004 (R. K. Mayo, and M. Terceiro, eds.), section 2, p. 30-80. 2005 groundfish assessment review meeting. Northeast Fisheries Science Center, Woods Hole, Massachu- setts; 15-19 August 2005. NEFSC Ref Doc. 05-13. NEFSC, 166 Water Street, Woods Hole, MA 02543. 3 Brodziak, J. K. T., E. M. Holmes, K. A. Sosebee, and R. K. Mayo. 2001. Assessment of the silver hake resource in the Northwest Atlantic in 2000, 134 p. NEFSC Ref Doc. 01- 03. NEFSC, 166 Water Street, Woods Hole, MA 02543. 98 Fishery Bulletin 105(1) Fall A Winter skate Spring D Little skate 1 1 1 1 1 1 1 Ij^i I if 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 B Little skate h 1 1 1 1 1 1 1 1 ijj^l iTi«l 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 E Silver hake U I I Ij^l I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I F Haddock 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 141 1 1*1 1 1 1 1 1 1 1 1 1 1965 1970 1975 1980 1985 1990 1995 2000 HI 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 4 1965 1970 1975 1980 1985 1990 1995 2000 -♦- Original ^ Smoothed Year Figure 4 Comparison of original time series of estimated total biomass on Georges Bank from the fall (left column) and spring (right column) bottom trawl survey (filled circles) and inverse-transformed, ARIMA-smoothed time series (open triangles). For these time series, an ARIMA (0,1,1) model (i.e., a random-walk- plus-uncorrelated-noise model) was used, although this was not the model selected in our time series analysis. order of the observed time series beyond that expected for uncorrelated noise. Of the nine species considered, only three (little skate, Atlantic herring, and flounder) had models that exhib- ited the same ARIMA order for both the fall and spring surveys. Taken at face value, this would indicate that the other six species exhibited substantially different dynamical processes during the fall and spring that influenced abundance on Georges Bank. One potential mechanism for this could be differential seasonal migra- tion patterns that result in changes in catchability that have different autocorrelation structures. For example, cod are distributed across the bank during the spring survey, but are found only in deeper waters on the pe- riphery of the bank during the fall where they may be less available to the survey. Assuming that all Georges Bank cod are available to the spring survey, if the frac- tion of cod available to the survey in the fall is density dependent or is driven by autocorrelated environmental conditions, then the fall survey abundance will exhibit dynamical behavior different from that of the spring survey abundance (note: if this were the case, it would be inappropriate to smooth the fall time series for cod with the ARIMA noise reduction approach applied in our study because, as noted previously, one of the basic assumptions with this approach is that the observation noise is uncorrelated). Other plausible mechanisms can be developed, as well. However, we feel it more likely that the inconsistency in ARIMA order between spring and fall surveys for the same species is an estimation problem and indicates that even a 40-year time series may not be long enough to reduce the variability inher- Stockhausen and Fogarty: Removing observational noise from time series data using ARIMA models 99 ent in ARIMA model estimation to reasonable levels for most species. Alternative methods, such as locally weighted scat- terplot smoothing (LOESS), moving average filters, exponential smoothing filters. Kalman filters, and fre- quency-domain approaches can be applied to time series to achieve smoother results (e.g., Cleveland and Grosse, 1991; Hamilton, 1994). These approaches typically em- ploy at least one user-determined parameter that can be used to change the amount of smoothing that an algorithm achieves. Generally, one "fiddles" with the adjustable parameters until a "nice," smoothed fit is achieved. However, we think it important to distinguish between these smoothing algorithms and the ARIMA- based noise reduction algorithms. It is quite possible to smooth out real fluctuations in the underlying popula- tion process. The principal advantage that we see for the ARIMA-based noise reduction algorithms (used with an appropriate model! over alternative methods is that the former provide a more objective approach to determining an appropriate level of smoothing. As noted previously, Pennington (1985) showed that when a RWPUN model is appropriate, the ARIMA smoothing approach is completely determined by the ARIMA model for the observed time series because it is possible to de- termine the observation noise variance from the model parameters. In the more general case. Box et al.'s (1978) algorithm at least yields a maximum value for the vari- ance of the observational noise and thus sets an upper limit to the amount of noise reduction and smoothing that can be achieved. For trawl survey data, our results from nine time series where RWPUN models were ap- propriate (and we can consequently estimate the actual observation noise variance) indicate that smoothing at ~90% of the maximum possible noise reduction level is not an unreasonable default percentage (Table 4). One drawback to the greater application of ARIMA- based noise reduction methods to time series data is the lack of an integrated software package that allows a user 1) to quickly evaluate an appropriate ARIMA model for a given time series, and 2) to calculate the smoothed time series. We used SAS for the first step and MATLAB for the second, but we found this ar- rangement rather awkward and burdensome. However, econometrically oriented software packages such as ForecastPro'' or AutoBox^ that automate model selection may substantially simplify the first step even if they don't address the second step. On the whole, ARIMA-based time series models ap- pear to provide the basis for a more objective approach to reducing observation noise in time series data, in- cluding time series of fishery abundance indices derived from trawl survey data, than do more conventional '' Business Forecast Systems, Inc. 2006. Website: http://www. forecastpro.com/products/fpfamily/index.html (accessed on 29 March 2006). ^ Automatic Forecasting Systems. 2003. Website: http:// www.autobox.com/autoboxdesc.htm (accessed on 29 March 29 2006). smoothing approaches. In the absence of additional information regarding the level of observation noise, we recommend smoothing trawl survey data at 909c of the maximum possible noise reduction level. We also suggest that development of an integrated software package for implementing ARIMA-based noise reduction will facilitate future use of this method. Finally, if a smoothed time series is desired (e.g., for graphical presentation only), then use of a RWPUN model in lieu of a model-fitting exercise will generally yield a curve pleasing to the eye. Alternatively, other methods such as LOESS could be employed to generate the smoothed results. However, if the resulting time series is to be used for further analysis of the dynami- cal behavior of the fish stock, we strongly recommend that a model-fitting approach be used to identify the most appropriate ARIMA model for the observed time series, from which the time series for the unobserved, underlying process can be computed. Otherwise, real fluctuations in the underlying process may be over- smoothed, resulting in an apparent dynamical behavior that displays little variability. This oversmoothing, in turn, may lead to erroneous conclusions being drawn regarding, for example, the resiliency of a stock to ex- ploitation or environmental change, and to perhaps concomitant errors being propagated in advice provided to fishery managers. Acknowledgments We would like to thank M. Pennington for his insight- ful comments on an early version of this manuscript. The comments and suggestions from two anonymous reviewers were also very helpful and are much appreci- ated. This work was supported by the National Research Council, which provided funding support for one of us (Stockhausen) as a postdoctoral fellow at the Northeast Fisheries Science Center. Literature cited Akaike, H. 1973. Information theory as an extension of the maximum likelihood principal. In Second international symposium on information theory (B. N. Petrov, and F. Csaki, eds.), 267-281. Akademiai Kiado, Budapest, Hungary. Anonymous. 1988. An evaluation of the bottom trawl survey program of the Northeast Fisheries Center. NOAA Tech. Memo. NMFS-F/NEC-52, 83 p. Northeast Fish. Center. 166 Water St., Woods Hole, MA 02543. 1993. Status of fishery resources off the northeastern United States for 1993. NOAA Tech. Memo. NMFS- F/NEC-101, 140 p. Northeast Fish. Center, 166 Water St., Woods Hole, MA 02543. Azarovitz, T. R. 1981. A brief historical review of the Woods Hole Labora- tory trawl survey time series. In Bottom trawl surveys (W. G. Doubleday, and D. Rivard, eds.), 62-67. Can. Spec. Pub. Fish. Aquat. Sci., vol. 58. 100 Fishery Bulletin 105(1) Becerra-Munoz, S., D. B. Hayes, and W. W. Taylor. 1999. Stationarity and rate of damping of modeled indi- ces of fish abundance in relation to their exploitation status in the Northwest Atlantic Ocean. Ecol. Mod. 117:225-238. Box, G. E. P., S. C. Hillmer, and G. C. Tiao. 1978. Analysis of seasonal time series. In Seasonal analysis of economic time series (A. Zellner, ed.), 309- 334. U.S. Dep. Comm., Washington, DC. Box, G. E. P., and G. M. Jenkins. 1976. Time series analysis: forecasting and control, 575 p. Holden-Day, Oakland, CA. Byrne, C. J., T. R. Azarovitz, and M. P. Sissenwine. 1981. Factors affecting variability of research trawl surveys. Can. Spec. Pub. Fish. Aquat. Sci. 58:238-273. Clark, S. H. 1979. Application of bottom trawl survey data to fish stock assessment. Fisheries 4:9-15. Cleveland, W. S., and E. Grosse. 1991. Computational methods for local regression. Stats, and Comp. 1:47-62. Cleveland. W. P, and G. C. Tiao. 1976. Decomposition of season time series: a model for the Census X-11 program. J. Am. Stat. Assoc. 71:581-587. Collie, J. S., and M. P. Sissenwine. 1983. Estimating population size from relative abundance data measured with error. Can. J. Fish. Aquat. Sci. 40:1971-1983. Edwards, R. L. 1968. Fishery resources of the North Atlantic area. In The future of the fishing industry in the United States (D. W. Gilbert, ed.), p. 52-60. Univ. Washington, Seattle, WA. Enders, W. 2004. Applied econometric time series, 460 p. John Wiley and Sons, Inc., Hoboken, NJ. Grosslein, M. D. 1969. Groundfish survey program of BCF Woods Hole. Comm. Fish. Rev. 31:22-30. Hamilton, J. 1994. Time series analysis, 820 p. Princeton Univ. Press, Princeton, NJ. Harley, S. H., and R, A. Myers. 2001. Hierarchical Bayesian models of length-specific catchability of research trawl surveys. Can. J. Fish. Aquat. Sci. 58:1569-1584. Helser, T. E., and D. B. Hayes. 1995. Providing quantitative management advice from stock abundance indices based on research surveys. Fish. Bull. 93:290-298. Lloret. J., J. Lleonart, and I. Sole. 2000. Time series modeling of landings in northwest Mediterranean Sea. ICES J. Mar. Sci. 57:171-184. Nogueira, E., F. F. Perez, and A. F. Rios. 1998. Modelling nutrients and chlorophyll a time series in an estuarine upwelling ecosystem (Ria de Vigo: NW Spain) using the Box-Jenkins approach. Estuar. Coast. Shelf Sci. 46:267-286. Pennington, M. 1985. Estimating the relative abundance offish from a series of trawl surveys. Biometrics 41:197-202. 1986. Some statistical techniques for estimating abun- dance indices from trawl surveys. Fish. Bull. 84: 519-525. Reid, R. N., F. P. Almeida, and C. A. Zetlin. 1999. Fishery-independent surveys, data sources, and methods. NOAA Tech. Memo. NMFS-NE-122. North- east Fish. Sci. Center, 166 Water St., Woods Hole, MA 02543. Appendix For the convenience of the reader, we summarize here Box et al.'s (1978) algorithm to calculate the coefficients of the smoothing polynomial loiB). Recall from Equation 14 that coiB)-- ct; P+D. First, define CiB)^^^;CiB} = l + aB'+C,B'~ + .... (A.3) ri(B) ^ ^ One can solve for the coefficients of C using an iterative process by recognizing that the coefficients of each power of B in the following expression must be zero: = (p{B)-n(B)C(B). Consequently, one obtains c, = ii^-Q) can be computed recursively using the relation C,=IC,_,'7,. (A.5) Next, define X(B,F)^(B} . (A.IO) For />Q, the X^'s are computed recursively by using the relation ^.=1^.-1 a- (A.ll) 1=1 Finally, the coefficients of w(B} are given by «o = i — |-^o; «j= — |-^,'i = i'- (A.12) 1 -'?! -1q -1q -"Iq -Iq 1 ll -nq-i -n, 1 102 Abstract — The eastern Stellar sea lion {Eumetopias jubatus) population comprises animals that breed along the west coast of North America between California and southeast- ern Alaska. There are currently 13 major rookeries (>50 pups): five in southeastern Alaska, three in British Columbia, two in Oregon, and three in California. Overall abundance has increased at an average annual rate of 3.1% since the 1970s. These increases can largely be attributed to popula- tion recovery from predator-control kills and commercial harvests, and abundance is now probably as high as it has been in the last century. The number of rookeries has remained fairly constant {n = ll to 13) over the past 80 years, but there has been a northward shift in distribution of both rookeries and numbers of animals. Based on the number of pups counted in a population-wide survey in 2002, total pup production was estimated to be about 11,000 (82% in south- eastern Alaska and British Colum- bia), representing a total population size as approximately 46,000-58,000 animals. Abundance and distribution of the eastern North Pacific Steller sea lion (Eutnetopias jubatus) population Kenneth W. Pitcher^ Peter F. Olesiuk^ Robin F. Brown^ Mark S. Lowry^ Steven J. Jeffries^ John L. Sease^ Wayne L. Perryman^ Charles E. Stinchcomb^ Lloyd F. Lowry^ Email address for K. W, Pitcher: ken_pitcher@fishgame.state.ak.us ' Division ol Wildlife Conservation Alaska Department of Fish and Game 525 West 67"^ Avenue Anchorage, Alaska 99518 2 Pacific Biological Station Department of Fisheries and Oceans Canada Nanaimo, BC V9T 6N7, Canada 3 Marine Mammals Research Program Oregon Department of Fish and Wildlife 7118 NE Vandenberg Avenue Corvallis, Oregon 97330 '' Southwest Fisheries Science Center National Marine Fisheries Service, NOAA 8604 La Jolla Shores Drive La Jolla, California 92037 5 Marine Mammal Investigations Washington Department of Fish and Wildlife 7801 Phillips Road SW Lakewood, Washington 98498 ' National Marine Mammal Laboratory National Marine Fisheries Service, NOAA 7600 Sand Point Way, NE Seattle, Washington 98115 ^ 73-4388 Paiaha Street Kailua-Kona, Hawaii 96740 Manuscript submitted 28 March 2006 to the Scientific Editor. Manuscript approved for publication 21 June 2006 by the Scientific Editor. Fish. Bull. 107:102-115 (2007). The Steller sea lion [Eumetopias juba- tus) rs the largest of the Otariidae and inhabits the North Pacific Rim from California to Japan. Individuals breeding at rookeries^ located along the west coast of North America from California northward through south- eastern Alaska (Fig. 1) to 144°W longitude form a distinct population segment, generally referred to as the eastern population. Historically, exchange of reproductive females with the Steller sea lion population to the north and west of 144°W longitude has been extremely low as shown by genetic studies (Bickham et al., 1996) and resightings of marked animals (Raum-Suryan et al., 2002). This indi- cates that population changes have been driven by birth and death rates within each population because immi- gration and emigration of breeding females among populations were too infrequent to affect population dynam- ics. More recent genetic analyses have confirmed the ancient divergence of the eastern and western populations. However, two new rookeries (White Sisters and Graves Rocks, Fig. 1) at the northern end of the range of the eastern population appear to have been colonized by females from both populations (O'Corry-Crowe et al., 2005). The number of western female immigrants to the eastern population has been small (in the 100s) to date, has not had a major impact on the growth dynamics of the overall east- ern population and has been limited to the extreme northern range of the eastern population. However, the pres- ence of breeding female immigrants from the western population within the range of the eastern population indicates that our prior assumption that population dynamics of the east- ern population was completely driven by internal rates of reproduction and survival was incorrect for the past several years. ' For purposes of this paper, rookeries are arbitrarily defined as traditional, ter- restrial sites where >50 pups are born annually. Other terrestrial sites used by sea lions are referred to as haulouts. Small numbers of pups are also born on haulouts, but probably constitute <1% of the total <100 in the eastern population. Pitcher et at. Abundance and distribution of Eumetopias /ubotus 103 500 In recent years, attention has focused on the western Alaskan population because of a precipitous decline since the 1970s (Lough- lin et al., 1992; Trites and Larkin; 1996) resulting in an "endangered" classification under the U.S. Endangered Species Act. The eastern population is currently classified as "threatened." Abundance from southern Or- egon through southeastern Alaska has gen- erally shown an increasing trend (Calkins et al., 1999; Brown et al.-; DFO, 2003), whereas numbers in southcentral California have de- clined substantially (Le Boeuf et al.-^; Hast- ings and Sydeman, 2002). This is the first detailed population-wide status evaluation of abundance, trend, and distribution with a historical perspective for the eastern popula- tion. We also present the results of the first population-wide census of pup production con- ducted in 2002 and apply life-table analysis to estimate total population size. In our study, we reviewed records of Steller sea lion abundance, with particular emphasis on data collected at rookeries. Some counts date back to the early 1900s, but early sur- veys were not systematic and methods lacked standardization, and some of the counts may have been affected by culling and hunting activities. Although these earlier survey methods preclude formal statistical analyses, the historical data provide a general sense of gross changes in abundance and distribu- tion. Systematic surveys began in most regions along the west coast in the 1970s, but counting techniques varied among the researchers and agencies conducting the sur- veys, and surveys were not coordinated between jurisdic- tions. Nevertheless, these time series indicate changes in relative abundance within each geopolitical region. In recent years, there has been an effort to compare and calibrate counting techniques, especially for pups (Snyder et al., 2001; P. F. Olesiuk, unpubl. data), and to synthesize survey results (Loughlin et al., 1992). Materials and methods Count data used to estimate population trends between the late 1970s and 2004 were of two types: 1) counts of pups obtained between late June and early July (at the end of the pupping season) when most pups are <1 month of age, and 2) counts of juveniles and adults al year of White Sisters Hazy Is. Scott Is. t: Noilh Danger Rocks Orford Reef Rogue Reef Sugarloaf Is., Cape Mendocino Ano Nuevo Is 500 1000 Kilometers 65=N 35° 145°W 125° 115° 135° Figure 1 Geographic range of the eastern Steller sea lion [Eumetopms jubalus) population showing locations of major (>50 pups born) breeding rookeries. 2 Brown, R. F., S. D. Riemer. and B. E. Wright. 2002. Pop- ulation status and food habits of Steller sea lions in Oregon. Report from Oregon Dept. of Fish and Wildlife to Oregon State Univ. Contract F0225A-01, 17 p. Oregon Department of Fish and Wildlife, Marine Mammal Research Program, 7118 NE Vandenberg Ave.. Corvallis, OR 97330. 3 Le Boeuf, B. J., K. Ono, and J. Reiter. 1991. History of the Steller sea lion population at Ano Nuevo Island, 1961-1991. NOAA Admin. Report NMFS-SWFSC LJ-91-45C, 9 p. age (i.e., nonpups) obtained from mid June to early July (mid to late in the breeding season). Steller sea lions normally give birth between late May and early July and breed between late May and mid July, although timing of these events varies somewhat geographically (Pitcher et al., 2001). Counts of pups are the preferred index to popu- lation size for many species of pinnipeds (Berkson and DeMaster, 1985). For the Steller sea lion, the vast major- ity of births occur at traditional rookeries, and because pups are confined to land for the first month of life, sur- veys of rookeries at the end of the pupping season provide a nearly complete estimate of annual pup production. Pups are more difficult to count than nonpups be- cause of their small size and dark color. This disad- vantage is especially pronounced for counts made at oblique angles from aircraft circling rookeries or from vessels adjacent to the sites. From the mid 1970s to the late 1990s, pups were usually counted by placing people on rookeries, herding nonpups into the water, and tallying the number of pups while walking through the rookery (Calkins and Pitcher, 1982). However, the methods of obtaining such counts are disruptive to sea lions (Lewis, 1987), and counts may not be possible where rookeries are protected in parks or ecological and nature reserves. More recently, vertical 126-mm format aerial photography has been shown to be as accurate and far less disruptive (Snyder et al., 2001) for counting pups. Depending on the physical size, 104 Fishery Bulletin 105(1) substrate, and topography of rookeries, high-quality oblique 35-mm photographs can sometimes provide counts of pups with an acceptable accuracy (P. F. Ole- siuk, unpubl. data). In 2002, vertical 126-mm format photography was used at all rookeries within the range of the eastern population to obtain the first estimate of total pup production (pup numbers at some rookeries had been reported previously but not for all rookeries in a single year). We have included additional counts of pups made at some sites between 2003 and 2005 for trend analyses within geographic subareas. However, only counts from the complete population-wide survey in 2002 were used to estimate total population abun- dance in order to provide an estimate for a single point in time. Table 1 Counts of pups and nonpups for each rookery and for all haulout sites combined by region for the population-wide survey of the eastern Steller sea lion (Eumeto- pias jubatus) population in 2002. Pup counts were made from vertical 126-mm format images, and nonpup counts from either vertical 126-mm format images or oblique 35-mm photographs. Nonpup counts included counts of pups at the indi- cated number of major sites (used by >50 animals on a regular basis during the breeding season), as well as counts of pups at numerous minor sites and counts of a few scattered animals. Site Pups Nonpups Southeastern Alaska Graves Rocks White Sisters Biali Rocks Hazy Islands Forrester Island Haulout sites (20 major sites) Southeastern Alaska total British Columbia North Danger Rocks Cape St. James Scott Islands Haulout sites (24 major sites) British Columbia total Washington Haulout sites (2 major sites) Oregon Orford Reef Rogue Reef Haulout sites (7 major sites) Oregon total California Saint George Reef Sugarloaf Island-Cape Mendocino Ano Nuevo Island Haulout sites ( 6 major sites) California total Eastern population 98 1001 403 1156 59 625 1257 2050 3060 3699 9 6752 4886 (49%) 15,283(43%) 207 592 655 982 2451 3865 5 6681 3318 (33%) 12,120 (34%) (0%) 651 (2%) 382 1178 746 1264 8 1727 1136(11%) 4169(12%) 367 716 150 588 189 255 7 1543 713 (7%) 3102 (9%) 10,053 Few reliable counts of pups were available before the 1970s, but counts of non-pups on rookeries have dated back to the early 1990s. Non-pups are easier to count, and there tends to be a high degree of correlation for counts of non-pups between oblique 35-mm format and vertical 126-mm format images (Fritz and Stincomb, 2005). However, some Steller sea lions, particularly juveniles, range widely (Raum-Suryan et al., 2002); therefore counts at haulouts within a particular geo- graphic area may not necessarily represent the number of animals supported by local rookeries, although breed- ing animals show a higher degree of site fidelity. The number and proportion of various sex and age classes of non-pups that are hauled out varies with season, time of day, and (in some cases) with tide (Winthrow, 1982; Calkins et al., 1999). Counts from the 2002 population- wide survey (Table 1) indicated a fairly tight relationship between the number of pups and nonpups counted on rookeries (Fig. 2). A similar pattern was noted for rookeries in British Columbia and the relationship persisted over the three decades concurrent pup and nonpup counts were available (P. F. Olesiuk, unpubl. data). The historical counts of nonpups (or total animals where pups and nonpups were not distinguished) on rookeries thus likely provide a general index of the size of the breeding population associated with each rookery. Systematic surveys have been conducted to monitor trends of the eastern Steller sea lion population, but methods and schedules have varied depending on the agency conducting the surveys. In south- eastern Alaska, the Alaska Depart- ment of Fish and Game periodically conducted ground counts of pups on rookeries from 1979 through 1998, and used vertical 126-mm format photography to count pups since 1998. In British Columbia, the De- partment of Fisheries and Oceans has conducted province-wide aerial surveys of rookeries and haulout sites at 2-5 year intervals since the early 1970s, using oblique 35- mm format photography to count both pups and nonpups. In 1998 and 2002, both pups and nonpups were counted at British Columbia rookeries with the use of vertical 126-mm format photography. There are no Steller sea lion rookeries in Washington, but the Washington 35,325 Pitcher et al : Abundance and distribution of Eumetopias /ubalus 105 Department of Fish and Wildlife has conducted numerous aerial surveys of haulout sites dur- ing the breeding season using oblique 35-mm format photography since 1978. In Oregon, the Oregon Department of Fish and Wildlife has conducted state-wide aerial surveys of nonpups on rookeries and haulouts using oblique 35- mm format photography on a nearly annual basis since the mid-1970s and has periodically obtained ground, or more recently vertical 126- mm format or high-resolution digital 35-mm format, pup counts. In California, the National Marine Fisheries Service, Southwest Fisheries Science Center, conducted statewide surveys during early July beginning in 1996 using vertical 126-mm format photography to count pups and nonpups at all rookeries and haulout sites. Time series of counts that were obtained with assorted methods were also available for some rookeries in California dating back to the 1970s. Although these surveys provide reliable information on changes in relative abundance within each region or at a particular rookery, they are difficult to synthesize into a popula- tion-wide assessment because of uncoordinated survey schedules and methods. Given the consistency within, but inconsistency between, these geo-politi- cal jurisdictions, we assessed trends in abundance by region (southeastern Alaska, British Columbia, Wash- ington, Oregon, and California). Counts for each re- gion were converted to natural logarithms and then regressed on year to determine average annual popula- tion growth rates. We estimated the total population size in 2002 from the predicted ratio of pups to nonpups in the population (Calkins and Pitcher, 1982; Trites and Larkin, 1996). From life tables for a stable sea lion population in the Gulf of Alaska, Calkins and Pitcher (1982) estimated total population size to be about 4.5 times the number of pups born. In order to apply this approach to the eastern population, which was not stable but increas- ing (see "Results" section), we conducted sensitivity analyses to determine how this multiplier varies with population growth rate (A) by incrementally chang- ing each of the life history parameters that affect it, namely juvenile mortality rates, adult mortality rates, age at maturation, and fecundity rates (Lotka, 1907; Cole, 1954) . We also reviewed historical records of Steller sea lion abundance in an attempt to relate current popu- lation size with abundance prior to the initiation of standardized surveys. Although these records provide insights into relative population levels, caution must be used because the older counts were obtained by a variety of methods and the seasonal timing of counts was inconsistent. In most cases the counts were made by professional biologists or naturalists hired by govern- ment agencies to conduct sea lion investigations, and special trips were made to rookeries to obtain first-hand counts; therefore it is unlikely numbers were grossly in- 3000 • 2000 ^-^ c 3 O O ^^ Q. 3 D. 1000 V^' c 1000 2000 3000 4000 1 Nonpup count Figure 2 Relationship between number of pup and nonpup Steller sea lions iEumetopias jubatus) counted on rookeries during the population- wide survey in 2002 (r- = 0.90; n = U: P<0.001). accurate. Because of the ad hoc nature of these counts, it was difficult to synthesize them into even a regional estimate of abundance, or to conduct statistical analy- ses; therefore these counts were generally examined on a rookery-by-rookery basis (Appendix). Results Southeastern Alaska Counts of Steller sea lion pups in southeastern Alaska increased from 2219 in 1979 to 5510 in 2005 (Fig. 3A), representing an average annual rate of increase of 3.2% (r2=0.91; n = 10; P<0.001). Prior to the early 1980s, the only rookery in southeastern Alaska was the Forrester Island complex. Only 50-100 animals were recorded when the site was first noted in the 1920s, and 350 animals were recorded when the site was revisited in 1945, and there was no mention of pupping in either case (Rowley, 1929; Imler and Sarber, 1947). Thus, although count data are extremely limited, it appears that Steller sea lion abundance was probably quite low in south- eastern Alaska during the first half of the 20"' century. Counts are not available, but the Forrester Island rook- ery must have grown dramatically through the 1950s and 1960s (Fig. 4A). By the time the first aerial survey was conducted in 1961, Forrester Island had grown to about one-third its current size in terms of both the numbers of pups and nonpups (Bigg, 1985). However, increases at Forrester Island appear to have slowed since the late 1970s, showing only a slight increase in pup production (0.6% per year; r2=0.40; n=13; P=0.021) and no discernible increase in the number of nonpups (r2 = 0.22;;i=12;P=0.125). 106 Fishery Bulletin 105(1) With the slowing of growth on Forrester Island, sev- eral new rookeries were established in southeastern Alaska (Calkins et al., 1999) (Appendix I). Hazy Islands were a substantial haulout in the 1950s (Mathisen and Lopp, 1963), but pup counts increased after they were first observed in 1979 (13% per year, ^2 = 0.76; n = ll; P<0.001). White Sisters developed into a rookery in the early 1990s and counts of pups also increased rapidly (16% per year, r2=0.87; /!=10; P<0.001). In recent years. Graves Rocks and Biali Rocks appear to be developing into rookeries; 175 and 100 pups were counted respec- 6000 5000 4000 3000 2500 2C00 A Southeastern Alaska 7000 "n 6000 UJ 5000 3 o o m 4000 3000 2500 b 2000 ' I ' ' ' ' I ' ' ' I I ' ' ' ' I B British Columbia C Oregon 3000 2500 2000 1500 1000 — 1 — I — I— 1 — r— 1 — I — p— 1 — r—i — I— 1 — — 1—1 — >— I — >— 1 — I — 1970 1975 1960 1985 1990 1995 2000 2005 Year Figure 3 Recent trends in counts of Stellar sea lion iEumetopias jubatus) pups (O) and nonpups (•) on rookeries in (A) Southeastern Alaska, (B) British Columbia, and (C) Oregon. These areas combined account for over 90% of pup production in the eastern population. Survey tech- niques were standardized within each region, but differed among regions. The slopes are all statistically significant (P<0.001), and none differed significantly from the overall rate of increase of 3.1%. tively at the two sites in 2005. Growth of these four new rookeries accounted for about 48% of the increase in total pup production in southeastern Alaska during the 1980s, and for about 74% of the total increase since 1990. In addition to the five rookeries, sea lions use about 20 major haulout sites (>50 animals) and several small- er sites in southeastern Alaska on a regular basis dur- ing the breeding season, as well as numerous other sites during the nonbreeding season. During the 2002 survey, a total of 6752 nonpups were counted at haulout sites and another 8531 nonpups were counted at rooker- ies (Table 1). British Columbia There are currently three Steller sea lion rookeries in British Columbia: the Scott Island complex (Triangle, Beresford-Maggot, and Sartine Islands), Cape St. James, and North Danger Rocks. Counts of pups from oblique 35-mm format photographs increased from 941 in 1971 to 3276 in 2002 (Fig. 3B), representing an average annual rate of increase of 3.2% (r2=0.71; 7i=9; P=0.005), similar to the overall rate observed in southeastern Alaska. However, piecewise regressions provide a better fit to the time series of pup counts, indicating that most of this increase has occurred since the 1980's (/•2 = 0.85; n = 9; P= 0.002). Significant increases in pup production (P<0.005) were evident at all three rookeries (Appendix), but mean rates varied among sites (3.7% at Scott Islands, 2.0% at Cape St. James, and 2.7% on North Danger Rocks). Numbers of nonpups on rookeries also increased significantly (7-2 = 0.89; n=9; P<0.001), paralleling the increases in pup production (Fig. 3B). Counts on rookeries in British Columbia date back to 1913 (Newcombe and Newcombe, 1914) and indicate breeding populations were historically large (Fig. 4B). Extensive sea lion reduction programs were conducted in British Columbia from 1912 through 1966, and attempts were made to commercially harvest sea lions during the 1960s. One major rookery, the Sea Otter Group, was eradicated by intensive control efforts during the 1920s and 1930s. The site was visited each year toward the end of the pupping season and all pups and as many nonpups as possible were killed, and by about 1940 it was no longer used as a rookery. Predator-control kills and commercial harvests in British Columbia continued into the 1960s and impacted all rookeries, and the breeding population was reduced to about 30% of peak levels by the late 1960s (Bigg, 1985). It appears that numbers at Scott Islands have fully recovered from these kills, but numbers at the two other rookeries are still below his- torical peak levels (Appendix). Sea lions also currently use 24 major haulout sites (>50 animals) in British Columbia on a regular basis during the breeding season, up from 18 sites when systematic province-wide surveys were initiated in the early 1970s (Bigg, 1985). Numbers of animals counted on these sites increased at rate of 4.0% since the early 1970s (r~=Q.&2; «=9; P<0.001), which is not significantly different from Pitcher et al.; Abundance and distribution of Eumetopios /ubatus 107 the rate of growth observed on rookeries. During the 2002 survey, 6681 nonpups were counted on haulout sites, and another 5439 on rookeries (Table 1). Washington There are no rookeries in Washington, but Stellar sea lions are found along the coast throughout the year. Four haulouts, including two major sites (>50 animals), are regularly used during the breeding season. Since 1989, surveys have been conducted almost annually, and numbers of sea lions counted have increased at an average annual rate of 9.2% (r'-=0.38; n = 37\ P<0.001). These animals are assumed to be immature animals and nonbreeding adults associated with rookeries from other areas. Juvenile sea lions branded as pups on For- rester Island in southeastern Alaska (Raum-Suryan et al., 2002) and on Rogue Reef in Oregon (R. F. Brown, unpubl. data) have been observed in Washington. Older records indicate that current abundance on the Washington coast is reduced from historical lev- els (Fig. 4C). Between 2000 and 3000 Steller sea li- ons were reported to be present during August and September of 1914, 1915, and 1916 on Jagged Island (Kenyon and Scheffer, 1959), compared with a maxi- mum statewide breeding season count of 847 during 1978-2001. Washington State Department of Fisher- ies offered a bounty of $8.00 for sea lions between 1944-48, but in 1949 this was reduced to $3.00 and limited to inside waters because aerial patrols indi- cated that the main coastal haulouts at Jagged Island and Split Rock had been reduced from 600 sea lions in the 1930s to fewer than 100 by 1949 (Scheffer, 1950). Only sporadic counts were available for individual sites during the 1950s and 1960s, but they indicate that few sea lions (<100 animals) were present during the breeding season and that total abundance did not exceed 500 during any season by the 1950s (Scheffer, 1950; Kenyon and Scheffer, 1959). Oregon Steller sea lions breed and pup at two rookeries, located at Rogue Reef and Orford Reef, and occupy seven major haulout sites in Oregon during the breed- ing season. The total number of nonpup sea lions on rookeries increased from 1186 in 1977 to 2442 in 2002 (Fig. 3C), representing an average annual rate of increase of 2.5% (r2 = 0.49; n=26: P<0.001). Although not as well documented, pup numbers also appear to have increased. In 1990, 492 and 298 pups were observed during ground counts at Rouge Reef and Orford Reef respectively, compared with 746 and 382 pups on 126 mm format images in 2002 (2.3% average annual rate of increase). During the 2002 population- wide survey, an additional 1727 nonpups were counted at haulout sites in Oregon (Table 2). Historical data on Steller sea lion abundance in Oregon are few (Fig. 4D). Pearson and Verts (1970) counted 862 animals (including some pups) during a state-wide aerial 0000 8000 A Southeastern Alaska 6000 4000 • 2000 • • S o 15000 B British Columbia 12000 • A 9000 .• A A A 6000 3000 • ° . O A * * • •• oo o A A • • o o m 5000 C Washington 4000 3000 A- - 2000 1000 A A A ,t^, 5000 D Oregon 4000 J 4,000 killed lOfbounly dunng 1925-1929 JUOO . ^ /•. 1 2000 1000 A 1 1 1 A t *. • ° o° 6000 1 1 T ' E California • T 5000 • • • - 4000 • - 3000 2000 1000 0^ 1905 1915 1925 1935 1945 1955 1965 1975 1985 1995 2005 Year Figure 4 Historical counts over the last century of Steller sea lion {Eumetopias jubatus) pups (O), non-pups on rookeries (•), and total nonpups on rookeries and haulouts (A) for (A) Southeastern Alaska, (B) British Columbia, (C) Washington, (D) Oregon, and (E) California. survey in June 1968, somewhat lower than the 1977 nonpup count of 1461 animals. The largest rookery was Orford Reef, where 475 animals, including pups, were counted. Interestingly, only 125 animals were reported at Rogue Reef, which is currently the largest rookery in Oregon, and Pearson and Verts (1970) suggested that it was no longer used as a rookery. Earlier counts are lacking, but the population was presumably substantially larger in the 1920s because about 4000 sea lions were 108 Fishery Bulletin 105(1) Table 2 Results of life-table sensitivity analys es showing the potential change in ratio of total population size to pups for a population increasing at 3.1% per annum. The vital rates in Calkins and Pitcher 's (1982) life tables' for a stable population of Steller sea lions (Eumetopias jubatus) were incrementally adjusted until a population growth rate, A, of 3.1% was attained The correspond- ing stable sex- and age-distributions were calculated by using Cole's (1954) finite approximations of Lotka's (1907) population | equations. Parameter that changed Relative change Population growth rate (A.) Pup multiplier 4 Mortality all ages -15% 3.1% 5.0 A Juvenile mortality -27% 3.1% 5.2 A Adult mortality -33% 3.1% 4.7 A Fecundity -h32% 3.1% 4.2 A Age at maturation -1.6 years 3.1% 4.2 ' Calkins, D. G., and K. W. Pitcher. 1982. Population assessment, ecology and trophic relationships of Steller sea ions in the GulfofAlaska. In Environmental assessment of the Alaskan continental shelf p. 447-546. U.S. Department of Commerce and U.S Department of Interior, Final | Report of Principal Investigators 19:1-565. killed for bounty on the Oregon coast during 1925-29 (Pearson and Verts, 1970), although some of these may have been nonbreeding animals associated with rookeries in California, British Columbia, and Alaska. California Steller sea lions historically have used six rookeries in California (San Miguel Island, Ano Nuevo Island, the Farallon Islands, Seal Rocks off San Francisco, Sug- arloaf Island-Cape Mendocino, and Saint George Reef). San Miguel Island and Seal Rocks are no longer used by Steller sea lions and only a few pups have been born on the Farallon Islands each year since the 1980s. There may have also been several additional small rookeries south of Ano Nuevo (Bonnot, 1928; Rowley, 1929). Statewide surveys, with the use of vertical 126-mm for- mat aerial photography, were implemented in California in 1996. From 1996 through 2004 there was no discern- ible statewide trend for nonpups on rookeries (7-'- = 0.408; n=7; P=0.123), however, pup production increased at an average annual rate of 8% (^2=0.68; «=8; P=0.012). Although there has been a long and intermittent time series of counts for rookeries in California over the last 75 years (Bonnot, 1928, 1929; Bonnot and Ripley, 1948; Bartholomew and Boolootian, 1960; Orr and Poulter, 1967; LeBoeuf et al., 1991; Stewart et al., 1993), caution is warranted when attempting to evaluate population trends from the older data because they are drawn from a variety of sources where different survey methods were used. Statewide, total counts of nonpups at the six rook- eries during the first half of the 20**" century were on the order of 3900-5600. The 2004 count at these same six sites was 1578 nonpups and 818 pups — indicating that, perhaps, only about a third as many animals are cur- rently present in the state (Fig. 4E). Population trends differed markedly among sites (Appendix). Historically, Steller sea lions extended south to the Channel Islands in southern California, and San Miguel Island was considered to have been the southernmost rookery (Bonnot, 1928, 1929). It appears that Steller sea lion were once more abundant than California sea lions (Zalophus caUfornianus) in that area (Bartholomew, 1967). Steller sea lions were reported to breed there in small numbers; Bonnot (1929) counted 50 pups in 1928. Abundance of nonpups in the Channel Islands peaked at about 2000 in the late 1930s (Appendix), although hunt- ing and harassment could have resulted in fewer animals being present during the surveys (Bonnot and Ripley, 1948; Stewart et al., 1993). Numbers subsequently de- clined — the main declines occurring between the late 1930s and 1950s (Bartholomew and Boolootian, 1960; Bartholomew, 1967). No births have been recorded since 1982 and no adults have been seen since 1983 (Stewart et al., 1993). In central California, Steller sea lion abundance at Ano Nuevo and the Farallon Islands is currently only about 20% of the levels reported between the 1920s and 1960s (Appendix). Steller sea lions had deserted the rookery at Seal Rocks near the entrance to San Francisco Bay by the late 1920s, purportedly as a result of persistent harassment by fishermen (Rowley, 1929). During the 1920s, Alio Nuevo Island and the Farallon Islands were identified as the most important rookeries in California, with 625 and 400 pups counted at each site, respectively, in 1922 (Bonnot, 1929). On Ano Nuevo, numbers re- mained at high levels until the early 1960s, then declined thru the mid-1990s (Orr and Poulter, 1967; Le Boeuf et al., 1991) (Appendix). Since 1996, both pup produc- tion (^2 = 0. 035; « = 8; P=0.656), and nonpup numbers (r2=0.018; n=8, P=0.755) have been stable. Fewer counts are available for the Farallon Islands, but the pattern appears to be similar (Appendix); abundance was at high levels from the 1920s to early 1960s and then declined sharply during the 1960s or early 1970s (Hastings and Sydeman, 2002). Pup production on the Farallons has been low since at least 1974 (Appendix). An average of only nine pups was counted between 1996 and 2004 and the site presently does not meet our criteria for a rook- ery (>50 pups). Nonpup numbers were stable (r'^=0.173; Pitcher et al.: Abundance and distribution of Eumetopias jubatus 109 1920's total count for rookeries 13SW 13QW 125 W 120 W 1970's total count for rookeries 135 W 130 W 125 W 120 W t \ 2002 nonpups on rookeries 135 W 130 W 125 W 120 W 2002 pup production 35 W 13QW 125 W 1 20 W %. Figure S Map showing the shift in distribution and relative importance of rookeries in the eastern Steller sea lion (Eumetopias jubatus) population. Circles represent the proportion associated with each rookery of the total estimated abundance in the 1920s (1913-17 for British Columbia) and 1970s; and the proportion of nonpups and pups associated with each rookery during the 2002 range-wide survey. The horizontal lines indicate the center of the distribution (the latitude of each rookery weighted by the number of animals on it). For the 1920s, rookery counts in California represent minimum abundance because pups were not always included. Historic counts were unavailable for Oregon; therefore the minimum abundance was taken as the number killed for bounty during 1925-29. Because Oregon lies near the estimated center of the breeding distribution, the center of distribution is insensitive to the numbers assumed on Oregon rookeries (halving or doubling the Oregon figures shifts the center by less than 0.5° of latitude). 73 = 15; P=0.123) at low levels (Appendix) between 1974 and 2004. Steller sea lions have been counted only sporadically at the Sugarloaf-Cape Mendocino and Saint George Reef rookeries in northern California until recent years (Ap- pendix). Numbers of nonpups have been relatively stable since 1996 at both Sugarloaf-Cape Mendocino (r- = 0.106; 7j = 8; P=0.431) and Saint George Reef (r2 = 0.128; « = 9; P=0.345). A comparison of counts made during the 1927-47 period with recent counts (Appendix) indicates that current abundance is probably only slightly reduced from historical levels. The Sugarloaf-Cape Mendocino rookery is small; counts of pups increased from 62 in 1996 to 131 in 2004, representing an average annual increase of 13% (r2=0.725; n=8; P=0.007). For the Saint George Reef rookery, located near the California-Oregon border, counts of pups increased from 243 in 1996 to 444 in 2004, representing an average annual rate of 10% (r2=0.70; n=8; P=0.009). Over the same period, counts of nonpups showed no discernible trend (r2=0.11; n=12; P=0.431). Steller sea lions use about six major (>50 animals) haulout sites along the California coast between Saint George Reef and Ano Nuevo Island, as well as numer- ous smaller sites, during the breeding season. In 2002, a total of 1543 nonpups were counted at haulouts, in addition to the 1559 nonpups counted on rookeries. At least 12 former Steller sea lion haulout sites and per- haps a few rookeries between the Channel Islands and Ano Nuevo Island (Bartholomew and Boolootian, 1960; Bonnot, 1928; Bonnot and Ripley, 1948; Rowley, 1929) have been abandoned. Overall trend for the eastern North Pacific Steller sea lion population The eastern North Pacific Steller sea lion population has exhibited significant and similar annual rates of growth in all three regions that support the largest rookeries: 3.2% in southeastern Alaska, 3.2% in British Columbia, and 2.5%- in Oregon (Fig. 3). Combining the trend trajec- tories for these three regions, which currently account for over 90% of total pup production in the eastern popu- lation, overall abundance is estimated to have increased by about 215% over the last 25 years, representing an annual rate of increase of 3.1%. The time series for California is shorter; however pup production increased significantly at 7% per year between 1996 and 2004. no Fishery Bulletin 105(1) With the exception of the southernmost rookery at Ano Nuevo Island and the (former) Farallon Islands rookery, both greatly reduced from historical levels, pup produc- tion has increased consistently throughout the range of the eastern population over the past 25+ years. The total population-wide pup count in 2002 was 10,053 pups, of which 49% were found in southeastern Alaska, 33% in British Columbia, 11% in Oregon, and 7% in California (Table 1). This represents minimum pup production because some pups may have died and disap- peared from rookeries prior to the survey, or were born after the census. Following Trites and Larkin (1996), we applied an arbitrary adjustment of 10%- to account for pups that had been missed during our survey, giving a pup production estimate of 11,060. Using life tables. Calkins and Pitcher (1982) estimated the ratio of total animals to pups in a stationary population would be about 4.5:1. Our sensitivity analyses indicated that for a population increasing at 3.1%, the ratio could be as low as 4.2:1 if the growth were due to increased fecundity, or as high as 5.2:1 if the growth was due to reduced juvenile mortality (Table 2). The eastern population is thus estimated to have numbered about 46,000-58,000 animals in 2002. During the 2002 survey, we actually counted 45,378 animals (10,053 pups and 35,325 non- pups) on rookeries and at haulouts. This count represents an absolute minimum population size because not every site was surveyed and some animals were absent from rookeries and haulouts during the surveys and therefore were not counted. The general sparseness and lack of standardization of the pre-1970 counts prevents a rigorous comparison of current and historical population levels; however several clear patterns emerge (Appendix). In south- eastern Alaska abundance was apparently quite low during the first half of the 20"^ century, but numbers have increased consistently since that time. We have no explanation for the low numbers during the early 1900s because we are not aware of large-scale hunting or predator control efforts. Numbers were high in Brit- ish Columbia in the early 1900s but were then reduced by about 70% by predator control and hunting. They have since recovered to levels approximately two-thirds of those of the early 1900s. Numbers on haulouts in Washington State were severely reduced by bounty hunting in the early to mid-1900s. Although there has been substantial recovery, peak numbers still appear to be only about half of levels of 1915. There are no count data available for Oregon prior to 1968, but the fact that about 4000 sea lions were killed for bounty during 1925-29 would indicate a sizable population at that time. There has been a substantial recovery since the 1968 surveys. The California population was apparently large during the early 1900s. Sites in south- ern California began declining in the late 1930s and that portion of the range was abandoned by the 1980s. Numbers in central California remained high into the 1960s, then declined to low levels, and stabilized dur- ing the 1990s. In northern California numbers were likely reduced during the mid 1900s, but now appear to be approaching levels of the early 1900s. Overall, the eastern population currently appears to be similar in size to historical levels of the early 1900s; the large population increase in southeastern Alaska balances out the declines in the southern portion of the range. Although the number of rookeries used by the east- ern Steller sea lion population has remained relatively constant (range 10-13), their distribution has shifted (Fig. 5). In the 2002 survey, the breeding population was centered (the latitude of each rookery weighted by the number of animals on it) at about 51.5°N (central British Columbia coast). Just over half of the rookeries (7 of 13) and births (57%) occurred north of that latitude, with the northernmost rookery at 58.2°N. For the 2002 popula- tion-wide survey, the pattern was similar for both pups and total numbers (pups and nonpups), suggesting they both provided an index of breeding distribution. In com- parison, during the 1970s the breeding population was centered at roughly 49.9°N (central Vancouver Island), with the northernmost rookery at 54.8°N, representing a northward shift of 0.5° of latitude or 65 km per decade. In the 1920s, the breeding population was probably cen- tered somewhere around 46.0°N (Washington-Oregon border); only two small rookeries accounted for about 13% of total abundance situated north of 51.5°N (the current center of pupping). At the southern end of their range, the declines of Steller sea lions appear to have begun in southern California (San Miguel) between the late 1930s and 1950s, and were followed by declines in central California between 1960 and 1990; however the two northernmost sites in California exhibited relative stability. Conversely, at the northern end of their range, Steller sea lions probably began breeding in significant numbers in southern southeastern Alaska (Forrester Island) in the late 1940s or 1950s and extended their breeding range to central southeastern Alaska (Hazy Islands) in the early 1980s, and northern southeastern Alaska (White Sisters) in the 1990s. Overall, the south- ern end of the breeding range contracted by about 3° latitude (330 km), and the northern limit was extended by about 5° latitude (550 km). Discussion The population increases observed in recent years over most of the range of eastern North Pacific Steller sea lion population almost certainly represent recovery from the impacts of prior predator-control programs, harvest- ing, and indiscriminate killing that took place prior to protection under the Canadian Fisheries Act of 1970 and implementation of the U.S. Marine Mammal Pro- tection Act in 1972. The overall annual rate of increase of 3.1% was widespread (from Oregon to southeastern Alaska) and has been underway for at least 25 years, and there is no evidence of it slowing with increasing sea lion densities. The consistent, long-term observed rate of increase of 3.1% throughout most of the range of the eastern population is well below the theoreti- cal maximum intrinsic rate of increase for pinnipeds Pitcher et al,: Abundance and distribution of Eumetopias /ubatus 111 (Wade, 1998; Harkonen et al.. 2002). This annual rate of increase indicates that either some factor or factors are still limiting the growth rate of this population or that the growth potential of this otariid is less than the theoretical maximum, which was derived from phocid population growth rates. We have observed Steller sea lions that have been shot or entangled in marine debris, and this undocumented mortality could be preventing the population from increasing at a higher rate. In addi- tion, the Steller sea lion tends to have a longer period of maternal investment and a lower reproductive rate than most phocids (Pitcher et al., 1998), both of which may limit the growth potential of populations. Although the three geographic regions supporting the largest rookeries all increased at about the same rate, individual rookeries often exhibited different population growth rates or temporal changes in growth rates. At the northern end of the range, Forrester Island accounted for essentially all of the population growth until the 1970s; however the observed rate of change has slowed since the 1980s. At the same time, some of the rookeries to the south of Forrester Island in British Columbia and to the north of it in central-northern southeastern Alaska have exhibited higher-than-average growth rates since the 1980s. The mechanism causing these geographic patterns is unknown, but could involve 1) dispersal of breeding animals between rookeries, 2) differences in local condi- tions that affect reproduction and survival, or 3) a shift in distribution of prey resources. Some dispersal of breed- ing females from their natal rookeries has been shown to occur. Six of 31 females that were marked as pups on the Forrester Island rookery were subsequently observed to have given birth on other rookeries (Raum-Suryan et al., 2002). The authors of that study concluded that the Steller sea lion generally conformed to the metapopula- tion concept as depicted by Hanski and Simberloff (1997), in that local breeding populations (rookeries) and move- ments among these local populations have the potential of affecting local dynamics. For our assessment of long-term historic population trends, we relied mainly on counts of non-pups (or oc- casionally pups and nonpups combined) on rookeries, as few reliable pup counts were available prior to the 1970s. The 2002 population-wide survey (Fig. 2) and the last 30 years of counts in British Columbia indicated there is a relationship between the numbers of nonpups and pups on rookeries. However, departures from this relation- ship can occur, especially where existing rookeries are being abandoned or new rookeries are being formed. For example, the Farallon Islands, which no longer meet our definition of a rookery, now serves largely as a haulout site (Le Boeuf et al., 1991). The historical rookery on the Sea Otter Group in British Columbia, the only rookery known to have been extirpated by control efforts, is also still used during the breeding season as a haulout by nonbreeding animals. Conversely, in southeastern Alas- ka, the new rookeries were established at sites previously used as major haulouts by nonbreeding animals. The lack of accurate pup counts may, thus, have influenced our historical interpretation of historical data and our depiction of the exact breeding range, but there is a gen- eral consensus that the breeding range has shifted. Pup production in southern California has disappeared and in central California has dropped to less than one-fifth of what it was in the 1920s. Few, if any, pups were born in southeastern Alaska in the early 1900s, whereas this area now accounts for nearly half of total pup production in the eastern North Pacific population. Control programs and harvesting clearly depleted the eastern Steller sea lion population and may have con- tributed to its redistribution, but the kills cannot fully explain the shift in the distribution. For example, while control efforts were underway in British Columbia dur- ing the 1950s and 1960s, animals may have taken ref- uge just north of the British Columbia-Alaska border at Forrester Island, or animals breeding on Forrester Island may have benefited from reduced competition as a result of the reductions on British Columbia rookeries. However, the northward expansion of the breeding range in southeastern Alaska continued through the 1980s and 1990s, even though killing of sea lions in British Columbia ceased in the 1960s. At the southern end of their range, sea lions were apparently very abundant in California before the 1860s, but were depleted during the 1870s because of intense hunts of sea lions for oil and hides (Bonnot, 1929). The last organized kills were made in 1909, although hunting, especially of bulls for trim- mings (genitals, lips with whiskers, and gall bladders) continued into the 1930s. Nevertheless, the population declines in southern California began in the late 1930s, and in central California began in the late 1960s and early 1970s, well after major kills by humans had ended (Hastings and Sydeman, 2002). The reason for the northward shift in the overall breeding distribution is unknown, and different factors may have been in play at the southern and northern ends of the range. In the south, competition with in- creasing populations of other pinnipeds may have been a factor in range constriction (Stewart et al., 1993). In particular, the number of California sea lions breeding in California increased from at most a few thousand in the 1920s (Bonnot, 1928) to about 240,000 in 2000 (Lowry and Maravilla-Chavez, 2005). It is likely that California sea lions and Steller sea lions compete with each other because 1) their ranges overlap, 2) they share the same haulout sites, and 3) they probably consume many of the same prey species. On San Miguel Island and the Farallon Islands, where Steller sea lions used to predominate (Bartholomew and Boolootian, 1960; Ripley et al., 1962; Stewart et al., 1993), the decHnes in Steller sea lions coincided with large increases in numbers of California sea lions (Stewart et al., 1993; Hasting and Sydeman, 2002). For unknown reasons, southeastern Alaska represents the only area throughout the range of the eastern North Pacific population where new Steller sea lion rooker- ies have been established. Steller sea lion rookeries are normally located on remote, offshore islands or reefs and require adequate areas above high water levels where young pups can survive most weather conditions. There 112 Fishery Bulletin 105(1) must also be adequate prey on a consistent basis within the foraging range of lactating females. Perhaps the lim- ited availability of such sites has restricted the establish- ment of new rookeries at other locations. Changes in the ocean environment, particularly to- wards warmer water temperatures (Field et al., 2006), have also been proposed as a factor that has favored the California sea lion and other pinnipeds over the Steller sea lion in the southern part of their range (Bartholomew and Boolootian, 1960). Environmental conditions can affect sea lion populations directly or indirectly. Tem- perature could directly affect the survival of animals and such effects would be expected to be most evident at the latitudinal extremes of the range. The ocean environment can also act indirectly by affecting marine food webs, and thus the quantity and quality of prey available to sea lions. Unfortunately, with historical survey data being so scant, and with sea lions having been artificially reduced below natural levels, one can only speculate about the long-term effects of environmental conditions on the east- ern Steller sea lion population, but conditions currently appear to be favorable through much of their range. A somewhat similar change in Steller sea lion distribu- tion and the establishment of new rookeries have been noted along the Asian coast. There the southern range limit has moved northward by 500-900 km over the past 50 years and several new rookeries have been established (Burkanov and Loughlin, in press). Based on the population-wide survey in 2002, pup production for the eastern population is currently esti- mated to be about 11,000, and total abundance on the order of 46,000-58,000. It should be emphasized that this should be regarded as a "general" estimate because several factors can affect the accuracy of pup counts and correction factors. Following Trites and Larkin (1996), we added 10% to pup counts to estimate pup production (i.e., actual number of births), which seems reasonable, but the adjustment is subjective and arbitrary, and in reality the adjustment probably varies from site-to-site and year-to-year. The sex and age structure of popula- tions, and hence the ratio of pups to nonpups, may differ between populations and change with population status in ways we do not understand. We attempted to delineate the possible range of changes in the correction factors by using sensitivity analyses, which showed the multi- plier could either decrease if population productivity is controlled by fecundity or age at maturation, or increase if population productivity is controlled by mortality. As- sessments for the western North Pacific population have indicate that the population declines were primarily due to poor juvenile survival (York, 1994), and if this is in fact the main determinant of population growth, the pup multiplier and estimated abundance of the eastern popu- lation may lie toward the high end of our range. During the 2002 population-wide survey, a surpris- ingly large number of nonpups were observed (75-100% of the number expected based on our life table analy- sis). Because one would expect appreciable numbers of juveniles and adults to be dispersed at sea and missed during surveys, the actual size of the eastern population may be near the upper end of our estimated range. On the other hand, 2002 may merely have been an excep- tional year for pup production, although the more recent pup counts available for California (2003 and 2004) and southeastern Alaska (2005) indicate that pup numbers have continued to increase. The apparent surplus of non- pups observed during the 2002 survey could also be indicative of the presence of nonbreeding animals asso- ciated with the western population in our survey area. Studies (where sea lions have been branded) have shown there is some overlap in the nonbreeding range of the two populations (Raum-Suryan et al., 2002), although there is no reason to expect a higher degree of movement from west to east. Moreover, the observed ratios of total counts to pup counts was uniformly high over the entire range of the eastern population (4.1 in southeastern Alaska, 4.7 in British Columbia, 4.7 in Oregon, and 5.4 in Cali- fornia), and if anything decreased slightly towards the north where one would expect the greatest overlap with the western population. The high nonpup to pup ratios indicate that high survival rather than high fecundity may be the primary mechanism responsible for popula- tion growth. Steller sea lions in the eastern population currently breed at 13 major rookeries (>50 pups born), and the highest concentration of breeding animals is in south- eastern Alaska, northern British Columbia, and near the Oregon-California border. Currently there is a large gap (993 km) between the Scott Islands rookery off north- western Vancouver Island and the Orford and Rogue Reef rookeries in southern Oregon. There are no records of rookeries along this coastline, and natives hunting sea lions along the Washington coast had no knowledge of rookeries in that state (Scheffer, 1950). However, it would not be surprising to see new rookeries founded or re-established at haulout sites along this gap, as has occurred in southeastern Alaska, if the eastern popula- tion continues to increase in the northern part of its range. Nonbreeding animals use approximately 59 major haulout sites (>50 animals during) during the breeding season, plus numerous smaller sites and many seasonal haulout sites. The major haulouts are widely distributed from Cape Fairweather (58.8°N, 137.9°W) to Ano Nuevo Island (37.1°N, 122. 3°W), providing Steller sea lions with access to coastline spanning about 22° of latitude or 2400 km. During the 1970s the eastern population represented only about 10% of the total number of Steller sea lions along the North American coast. With the large decline in the western population in conjunction with the in- crease in the east, this percentage has changed dramati- cally; about 55% of pup production in North America now occurs in the eastern population. We anticipate that con- tinued monitoring and comparisons of the growing east- ern population with the western population will provide insight into factors that ultimately regulate Steller sea lion populations, and we hope this synthesis for the east- ern population will contribute toward better coordination of surveys and standardization of counting methods over the distribution range of the species. Pitcher et al,: Abundance and distribution of Eumelopios /ubatus 113 Acknowledgments We thank S. Riemer and B. Wright of the Oregon Depart- ment of Fish & Wildlife for their extensive contributions to survey work, counting, and database management. D. McAllister of the Alaska Department of Fish and Game conducted many of the surveys in southeastern Alaska. We thank K. Raum-Suryan for drafting Figure 1. R. DeLong and R. Small reviewed and provided useful com- ments on an earlier draft of this manuscript. We appre- ciate the timely comments and suggestions provided by three anonymous reviewers and by the editorial staff of Fishery Bulletin. Surveys in the U. S. were conducted under research permits from the Office of Protected Resources of the National Marine Fisheries Service and in Canada with authorization of Fisheries and Oceans Canada. Surveys in Oregon were conducted under spe- cial use permits granted by the U.S. Fish & Wildlife Service, Oregon Coastal Refuge Complex, and surveys at protected sites in B.C. were authorized by the Ecological Reserves Unit of B.C. Parks and Parks Canada. Literature cited Bartholomew, G. A. 1967. Seal and sea lion populations of the Channel Islands. In Proceedings of the symposium on the biol- ogy of the California Channel Islands (R. N. Philbrick, ed.), p. 229-244. Santa Barbara Botanical Garden, Santa Barbara, CA. Bartholomew, G. A., and R. A. Boolootian. 1960. Numbers and population structure of pinnipeds on the California Channel Islands. J. Mammal. 41:366-375. Berkson. J. M., and D. R DeMaster 1985. Use of pup counts in indexing population changes in pinnipeds. Can. J. Fish. Aquat. Sci. 42: 873-879. Bickham, J. W., J. C. Patton, and T. R. Loughlin. 1996. High variability for control-region sequences in a marine mammal: implications for conservation and bio- geography of Steller sea lions (Eumetopias jubatus). J. Mammal. 77:95-108. Bigg, M. A. 1985. Status of Steller sea lion (Eumetopias jubatus) and California sea lion {Zalophus californianus) in British Columbia. Can. Spec. Pub. Fish. Aquat. Sci. 77:1-20. Bonnot, P. 1928. The sea lions of California. Calif Fish and Game. 14:1-16. 1929. Report on the seals and sea lions of California. California Division of Fish and Game. Fish. Bull. No. 14, 61 p. Bonnot, P., and W. E. Ripley. 1948. The California sea lion census for 1947. Calif. Fish and Game 34:89-92. Burkanov, V. N., and T. R. Loughlin. In press. Distribution and abundance of Steller sea lions on the Asian coast, 1720s-2005. Mar. Fish. Rev. Calkins, D. G., D. C. McAllister, K. W. Pitcher, and G. W. Pendleton. 1999. Steller sea lion status and trend in Southeast Alaska: 1979-1997. Mar. Mamm. Sci. 152:462-477. Cole, L. C. 1954. The population consequences of life history phenomena. Quarterly Rev. Biol. 29:103-137. DFO (Department of Fisheries and Oceans, Canada). 2003. Steller sea lion tEumetopias jubatus). DFO, Can. Sci. Advis. Sec. Stock Status Rep. 2003/037, 9 p. Field, D. B., T. R. Baumgartner, C. D. Charles, V. Ferreira-Bar- tina, and M. D. Ohman. 2006. Planktonic foraminfera of the California Current reflect 20"' century warming. Science 311:63-66. Fritz, L. W., and C. Stinchcomb. 2005. Aerial, ship, and land-based surveys of Steller sea lions (Eumetopias jubatus) in the wetern stock in Alaska, June and July 2003 and 2004. NOAA Tech. Memo. NMFS-AFSC-153. 56 p. Hanski, I., and D. Simberloff 1997. The metapopulation approach, its history, con- ceptual domain and application to conservation. In Metapopulation biology: ecology, genetics and evolution (Hanski and M. Gilpin, eds.), p. 5-26. Academic Press, London. England. Harkonen, T., K. C. Harding, and M. Heide-Jorgensen. 2002. Rates of increase in age-structured populations: a lesson from the European harbour seal. Can. J. Zool. 80:1498-1510. Hastings, K. K., and W. J. Sydeman. 2002. Population status, seasonal variation in abundance, and long-term population trends of Steller sea lions (Eumetopias jubatus) at the South Farallon Islands, California. Fish. Bull. 100:51-62. Imler, R. H., and H. R. Sarber 1947. Harbor seals and sea lions in Alaska. U.S. Fish Wildl. Serv., Spec. Sci. Rep. 28, 23 p. Kenyon, K. W., and V. B. Scheffer. 1959. Wildlife surveys along the northwest coast of Washington. Murrelet 42:29-37. Lewis, J. P. 1987. An evaluation of a census-related disturbance of Steller sea lions. M.S. thesis, 89 p. Univ. Alaska, Fairbanks, Alaska. Lotka, A. J. 1907. Relation between birth rates and death rates. Sci- ence 26: 21-22. Loughlin, T. R., A. S, Perlov, and V. A. Vladimirov. 1992. Range-wide survey and estimation of total number of Steller sea lions in 1989. Mar. Mamm. Sci. 8:220-239. Lowry, M. S., and O. Maravilla-Chavez. 2005. Recent abundance of California sea lions in western Baja California, Mexico and the United States. In Pro- ceedings 6th California Islands symposium, Ventura, Cal- ifornia, U.S., December 1-3, 2003, p. 485-497. National Park Service Tech Publ. CHIS-05-01, Institute for Wild- life Studies, Areata, CA. Mathisen, O. A., and R. J. Lopp. 1963. Photographic census of the Steller sea lion herds in Alaska, 1956-58. U.S. Fish Wildl. Serv., Spec. Sci. Rep. Fish. No. 424, 20 p. Newcombe, C. F., and W. A. Newcombe. 1914. Sea lions on the coast of British Columbia. Ann. Rep. British Columbia Comm. Fish for 1913, p. 131-145. O'Corry-Crowe, G., T. Gelatt, K. Pitcher, and B. Taylor. 2005. Crossing significant boundaries: evidence of mixed-stock origins of new Steller sea lion, Eumetopias jubatus, rookeries in Southeast Alaska (Abstract). In Fishery Bulletin 105(1) The 16'*' biennial conference on the biology of ma- rine mammals (book of abstracts); December 12-16, 2005, San Diego, California, 330 p. Society for Marine Mammology. Orr, R. T., and T. C. Poulter. 1967. Some observations on reproduction, growth, and social behavior in the Steller sea lion. Proc. Calif. Acad. Sci. 32:377-404. Pearson, J. P.. and J. P. Verts. 1970. Abundance and distribution of harbor seals and northern sea lions in Oregon. Murrelet 51:1-5. Pitcher, K. W., V. N. Burkanov, D. G. Calkins, B. J. Le Boeuf, E. G. Mamaev, R. L. Merrick, G. W. Pendleton. 2001. Spatial and temporal variation in the timing of births of Steller sea lions. J. Mammal. 82:1047-1053. Pitcher, K. W., D. G. Calkins, and G. W. Pendleton. 1998. Reproductive performance of female Steller sea lions: an energetics-based reproductive strategy? Can. J. Zool. 76:2075-2083. Raum-Suryan, K. L., K. W. Pitcher, D. G. Calkins, J. L. Sease, and T. R. Loughlin. 2002. Dispersal, rookery fidelity, and metapopulation structure of Steller sea lions [Eumetopias jubatus) in an increasing and a decreasing population in Alaska. Mar. Mamm. Sci. 18:746-764. Ripley, W. E., K. W. Cox, and J. L. Baxter. 1962. California sea lion census for 1958, 1960 and 1961. Calif Fish. Game 48:228-231. Rowley, J. 1929. Life history of the sea-lions on the California coast. J. Mammal. 10:1-39. Scheffer, V. B. 1950. Mammals of the Olympic National Park and vicinity. Northwest Fauna no. 2, p. 192-225. Snyder, G. M., K. W. Pitcher, W. L. Ferryman, and M. S. Lynn. 2001. Counting Steller sea lion pups in Alaska: an evalu- ation of medium-format, color, aerial photography. Mar. Mamm. Sci. 17:136-146. Stewart, B. S., P. K. Yokum, R. L. DeLong, and G. A. Antonelis. 1993. Trends in abundance and status of pinnipeds on the southern California Channel Islands. In Third California Islands symposium: recent advances in research on the California islands (E. Hochberg, ed.), p. 501-516. Santa Barbara Museum of Natural His- tory, Santa Barbara, CA. Trites, A. W., and P. A. Larkin. 1996. Changes in the abundance of Steller sea lions (Eumetopias jubatus) in Alaska from 1956 to 1992: how many were there? Aquat. Mamm. 22:153-166. Wade, P. R. 1998. Calculating limits to the allowable human-cause mortality of cetaceans and pinnipeds. Mar. Mamm. Sci. 14:1-37. Withrow, D. E. 1982. Using aerial surveys, ground truth methodology, and haulout behavior to census Steller sea lions, Eume- topiasjubatus. M.S. thesis, 102 p. Univ. Washington, Seattle, WA. York, A. E. 1994. The population dynamics of northern sea lions, 1975-1985. Mar. Mamm. Sci. 10:38-51. Appendix 2500 2000 1500 "O 0} — 1000 13 o O 500 c ra o A San Miguel Island -'-• — * 1 t-tmmmmm 4000 • B Ano Nuevo Island - 3000 • 4» * 2000 • • V 1000 m • ^^*fc, 1905 1915 1925 1935 1945 1955 1965 1975 1985 1995 2005 2000 1 1 1 1 1 • C Farallon Islands 1500 * 1000 • • • 500 • • 1 1 1 1 1 — .^1 1 1 1 1 1'i'iTi in [ mirf Kmto — D Sugarloaf / Cape Mendocino —[ 1 r- • •• 1905 1915 1925 1935 1945 1965 1965 1975 1985 1995 2005 Year Appendix Figure Historic counts made over the last century of Steller sea lion iEumetopias jubatus) pups (O), nonpups (•), and total number of animals (♦) at each breeding rookery within the range of the eastern Steller sea lion population. Pitcher et al,: Abundance and distribution of Eumetopias /ubatus 115 2000 E St. George Reef 3000 ' K Nortfi Danger Rocks • 2500 - 1500 • ^ » 2000 - 1000 • */ 1500 . .• • *,, •\ 1000 . 500 • • . %v- 500 • ."^S 2 J 3000 F Rogue Reef - 10000 L Forrester Island Complex « 2500 - - 8000 2000 . : ♦.♦ • 6000 ♦* * . 1500 1000 ocSP 4000 s • 500 2000 <^ oo O 1 1 1 1 1 , , • , , ., , , 1 1 1 1 1 2000 G Ortord Reef 4000 M Hazy Islands 1500 • ^ . 3000 Is counted 8 § • •••/ V • 2000 1000 , e ° (0 — 8000 H Scott Islands I 2000 N Biali Rock - 0) -| 6000 1500 - 3 ■z. 4000 . 1000 - : • ♦♦:•:• •%; A: 2000 6 - 0.4 - 0.2 0,0 /WlVCl/=0,98{n/c)-°59 r2 = 0,98 10 20 30 40 F Sample size per length class (n/c) Figure 1 The mean weighted coefficient of variation (MWCV) for 596 subsamples was closely related to the sample size (n ) divided by the number of per length classes in the sample (c). A good fit was obtained for the power function indicated by the solid line; its parameters are given at the top of the plot. The dashed line indicates the theoretical maximum MWCV (Eq. 4.). The histograms show the distribution of the samples on both axes. excessively large sample sizes. The range of sample sizes was between 2.2 and 24.7 times the number of length classes (2.5% and 97.5% quantiles), resulting in a range of MWCVs between 0.14 and 0.61. With a minor increase in effort, the sample size might be increased to 10 per length class for each subsample, resulting in an MWCV of around 0.25 for all samples. Considering that the precision deteriorates very rap- idly for sample sizes of less than 10 per length class, a minimum sample size of 10 times the number length classes in the sample is suggested as a rule-of-thumb in the present case. The previous analysis shows that, in order to ob- tain the same level of precision for all subsamples, the sample size should be directly proportional to the number of size classes. In absence of specific guid- ance on the sample size during the 2005 survey, the chosen sample size was only weakly correlated to the number of length classes in the sample of poor cod and haddock, whereas no significant correlation was found for blue whiting and Norway pout (Fig. 2). The same figure also shows that the MWCV in subsamples tended to increase with the mean length of the fish in the sample. This increase indicates that samples with a large mean size tended to be sampled with lower precision than samples of smaller fish of the same species. Discussion Length distributions that result from combining a number of different samples exhibit greater variation than predicted under the multinomial model given in Equation 1 (Smith and Maguire, 1983). Fish populations are usually not uniformly mixed; therefore individual samples are not random samples from the population (Pennington et al., 2002). The simple multinomial model does not take account of the between-sample variability and will therefore underestimate the total variance. However, Equation 1 does provide an unbiased estimate of the variability within each sample, which is the vari- ability that would occur if one could repeatedly take a random sample at the same location and time and measure these without error. This is the variability that is of interest when deciding whether the sample size is large enough to estimate the length distribution from a particular haul with a certain precision. Therefore, the MWCV is a suitable measure for this exercise. In order to obtain a precise population estimate, it is important to maximize the number of sampling lo- cations because of the considerable between-sample variability that is usually present (Pennington et al., 2002). Pennington et al. (2002) suggested maximiz- ing the number of sampling locations at the expense of the number of fish measured. However, the number of hauls is often limited by practical considerations, and length measurements can be obtained quickly and cheaply. Therefore, it seems prudent to sample enough fish from each haul to obtain a length distribution that is representative of that catch at that particular loca- tion. Detailed information on the length distribution at each station can be valuable for exploratory data analysis, such as investigating the spatial structure in the data. Nevertheless, this level of sampling may not be strictly necessary for a precise population estimate of the length-frequency distribution for an age- or length- based assessment. The samples in Figure 1 included a large range of species and size categories offish, but the variability in the MWCV was small after taking account the sample sizes. This small amount of variability indicates that the MWCV is not very sensitive to the exact shape of the distribution and can be predicted with high preci- sion, at least within the range of distributions encoun- tered on the survey. A minimum sample size of 10 times the number of length classes in the sample appears to be a reasonable compromise between effort and preci- sion in the present case. The current analysis has focused on subsampling during surveys; however the same principles can be applied to any process of collecting data for which the shape of the distribution is of interest. The desired pre- cision level for these cases will depend on a number of factors. For certain species that are of little commercial or scientific interest, but which may span across a large number of length classes, the suggested sample size of 10 per length class may be excessive. Likewise, as the MWCV is directly proportional to the number of length Gerritsen and McGrath: Precision estimates and suggested sample sizes for length-frequency data 119 Haddock 300- r2=0-27: P<0,001 / / (11 N 200- •■ ''l^^ (1) •X/^ n ^J^' E 100- ; >V\ • en jA** '• ' * 0- / Blue whiting Norway pout Poor cod 1 — I — I — I — \ — \ — r 5 10 20 30 r==0 00; P=0,88 1 I \ I I I T 1 — I — I — I — I — I — r 5 10 20 30 5 10 20 r==0.06; P=0.03 Number of length classes > o Haddock 0.6- r-'=0 67. P<0 001 4- •J>^^^ 02- i:;>r^-— 0.0- 1 1 1 1 1 Blue whiting Norway pout 1 — I — I — I — I — I — r 30 5 10 20 30 Poor cod r-=0 29. P<0 001 15 20 25 30 35 14 "1 — I — r 18 22 26 10 12 14 16 1 Mean length (cm) Figure 2 The sample sizes of subsamples taken on the survey were correlated with the number of length classes in the samples of haddock iMelanogrammus aeglefinus) and poor cod iTrisopterus minutus), but not signifi- cantly so for blue whiting {Micromesistius poutassou) and Norway pout iTrisopterus esmarkii) (top row). There was considerable variation in the mean weighted coefficient of variation (MWCV), which correlated with the mean length offish in the samples (bottom row). The solid lines represent linear regressions and the dashed lines indicate the sample sizes and MWCV that would have resulted from a sampling scheme where the sample size was chosen to be 10 times the number of length classes in the distribution. The coefficients of determination, r-, are given together with their P-values. classes in the sample, the choice of the interval of the length classes will determine the precision. Although increasing the size of length intervals will reduce the MWCV, this action will result in a loss of information which is undesirable. The cost of sampling, the detail required, and the purpose of the data collection need to be considered before the required precision level can be determined for applications other than the present example. Without formal guidance on the appropriate sample size, the sample sizes chosen were, at best, weakly correlated with the number of size classes in the sam- ples. It appears that the samplers under-estimated the required sample size for samples with large fish, whereas samples of smaller fish of the same species were over-sampled. This tendency to under-estimate the sample size may be related to the fact that the volume of a sample increases with the cube of its mean length; therefore a sample size of large fish may appear to be larger than the same number of small fish. In addition, samples with large fish tend to be spread out over a larger number of size classes, thus requiring higher sample numbers. In practice, it will be difficult for a sampler to esti- mate both the number of size classes and the number of fish in a sample. Therefore, the Marine Institute in Ireland is developing a software application that al- lows samplers to examine the length frequencies of the samples directly after they have been measured. The software estimates the weight of the suggested sample size for each distribution. Because size distributions tend to be similar on consecutive hauls, the sampler can gain an insight into the required weight of an appropri- ate sample for each species and size category. The information contained in a length-frequency dis- tribution is largely a function of sample size. The pres- ent method allows the amount of information contained in a length-frequency distribtuion to be quantified in terms of precision, allowing samplers to make informed decisions on the sample size that is required to obtain an adequate estimate of the length-frequency distribu- tion of a particular catch. 120 Fishery Bulletin 105(1) Acknowledgments Free R-software has been used for this work and the authors would like to thank the R development core team and all contributors to the R project (http://www. R-project.org). We also thank Colm Lordan and two anonymous reviewers for their valuable comments. Literature cited Anderson, R. O., and R. M. Neumann. 1996. Length, weight, and associated structural indices. In Fisheries techniques, 2""* ed. (B. R. Murphy and D. W. Willis, eds.), p. 447-482. Am. Fish. Soc, Bethesda, MD. Erzini, K. 1990. Sample size and grouping of data for length-fre- quency analysis. Fish. Res. 9:355-366. Pennington, M., L. M. Burmeister, and V. Hjellvik. 2002. Assessing the precision of frequency distribu- tions estimated from trawl-survey samples. Fish. Bull. 100:74-80. Smith, S. J., and J. J. Maguire. 1983. Estimating the variance of length composition samples. In Sampling commercial catches of marine fish and invertebrates (W. G. Doubleday and D. Rivard, eds.), p. 165-170. Can Spec. Pub. Fish. Aquat. Sci. 66. Ottawa, Canada. Thompson, S. K. 1987. Sample size for estimating multinomial propor- tions. Am. Statist. 41:42-46. Vokoun, J. C, C. F. Rabeni, and J. S. Stanovick. 2001. Sample-size requirements for evaluating popula- tion size structure. N. Am. J. Fish. Manage. 21:660- 665. Zar, J. H. 1999. Biostatistical analyis, 4'*' ed., 663 p. Prentice- Hall, Inc., Englewood cliffs, NJ. 121 Abstract — We tested the hypothesis that larger juvenile sockeye salmon [Oncorhynchus nerka) in Bristol Bay, Alaska, have higher marine- stage survival rates than smaller juvenile salmon. We used scales from returning adults (33 years of data) and trawl samples of juveniles (?i = 3572) collected along the east- ern Bering Sea shelf during August through September 2000-02. The size of juvenile sockeye salmon mir- rored indices of their marine-stage survival rate (e.g., smaller fish had lower indices of marine-stage survival rate). However, there was no relationship between the size of sockeye salmon after their first year at sea, as estimated from archived scales, and brood-year survival size was relatively uniform over the time series, possibly indicating size-selec- tive mortality on smaller individuals during their marine residence. Varia- tion in size, relative abundance, and marine-stage survival rate of juvenile sockeye salmon is likely related to ocean conditions affecting their early marine migratory pathways along the eastern Bering Sea shelf. Early marine growth in relation to marine-stage survival rates for Alaska sockeye salmon iOncorhynchus nerka) Edward V. Farley Jr^ James M. Murphy* Milo D. Adkison^ Lisa B. Eisner* John H. Helle' Jamai H. Moss' Jennifer Nielsen^ Email address for E V Farley: Ed.Farley@noaa.gov ' NOAA, Auke Bay Laboratory National Marine Fisheries Service, NOAA 1 1305 Glacier Highway Juneau, Alaska 99801 ^ Juneau Center, School of Fisheries and Ocean Sciences, University of Alaska Fairbanks 11120 Glacier Highway Juneau, Alaska 99801 ^ U.S. Geological Survey, Alaska Science Center 1011 East Tudor Road Anchorage, Alaska 99503 Manuscript submitted 27 March 2006 to the Scientific Editor. Manuscript approved for publication 5 July by the Scientific Editor. Fish. Bull. 105:121-130 (2007). Pacific salmon {Oncorhynchus spp.) experience relatively high mortality rates during the first few months at sea (Hartt, 1980), and it is believed that size plays an important role in survival (Parker, 1968; Pearcy, 1992). Size-dependent mortality of juvenile salmon may be concentrated during two specific life-history stages. The first stage is thought to occur just after juvenile salmon enter the marine environment, where smaller individu- als are believed to experience higher size-selective predation (Parker, 1968; Willette et al., 1999). The second stage is thought to occur after the first summer at sea, when smaller indi- viduals may not have sufficient energy reserves to survive late fall and winter (Beamish and Mahnken, 2001). Thus, larger individuals likely have a higher probability of survival during both of these stages, and size and growth while salmon reside in the estuary and during their first summer at sea may be important for survival. Previous studies indicate that scale radius length is proportional to fish body length (Francis, 1990; Ricker, 1992) and, in particular, incremental increases in sockeye salmon (O. nerka) scale radius are strongly correlated with somatic growth (Fukuwaka and Kaeriyama, 1997). In our study, scales from adult Bristol Bay sockeye salmon were examined to determine the rela- tionship between size after their first year at sea and survival to adulthood. We compared the time series (1965-97) of brood-year returns per spawner with scale growth measurements taken from adult sockeye salmon re- turning to the Egegik and Kvichak River systems in Bristol Bay, Alaska. Juvenile sockeye salmon enter the marine waters of the eastern Bering Sea during May and June (Burgner, 1991) and migrate through Bristol Bay to the Bering Sea and North Pacific during the summer and early fall months (Straty, 1981; Farley et al., 2005). Two differing models of seaward migration are believed to ex- ist for juvenile Bristol Bay sockeye salmon: in some years juvenile sock- eye salmon migrate along the coastal waters of the eastern Bering Sea near the Alaska Peninsula, and in other 122 Fishery Bulletin 105(1) years their migration is farther offshore (Farley et al., 2005). We also compared the size of juvenile Bristol Bay sockeye salmon collected during late summer and early fall (2000-02) trawl surveys along the eastern Bering Sea shelf with indices of their abundance, marine stage survival rate after our survey, and returns per spawner from these cohorts. Interannual differences in the size and growth rates of juvenile sockeye salmon were also compared to their early marine distribution and ocean conditions. The specific objectives of this study were to determine whether larger, presumably faster growing, juvenile sockeye salmon in fact had higher survival rates than smaller, presumably slower growing indi- viduals, and what aspects of the marine environment might influence these growth rates. Materials and methods Data Our research focused on Bristol Bay sockeye salmon because this region has the largest returns and com- mercial harvest of sockeye salmon in the world. Scales from adult sockeye salmon, their fork lengths, and data on brood-year return per spawner for the Egegik and Kvichak Rivers in Bristol Bay, Alaska, as well as annual totals of the number of adult Bristol Bay sock- eye salmon returns and spawners, were obtained from the Alaska Department of Fish and Game (ADF&G). Salmon scales are collected annually by ADF&G to estimate the age composition of adult sockeye salmon for fishery management. Age was designated by the European notation, i.e., 0.6, where a = the number of winters spent in freshwater prior to going to sea and b = the number of winters spent in the ocean (Koo, 1962a). Salmon scale collections and brood year return per spawner data were available for the dominant freshwater and ocean age groups of sockeye salmon sampled in the Kvichak River (brood year returns for ages 1.2, 1.3, 2.2, 2.3; 1965-97) and the Egegik River (brood year returns for ages 1.3, 2.2, 2.3; 1965-97). Scales were selected for measurement by following the procedures described in Ruggerone et al. (2005). Brief- ly, scales were selected when our age determination matched that previously made by ADF&G, the shape of the scale indicated that the scale was from the "pre- ferred area" (below the dorsal fin and above the lateral line — see Koo, 1962b I, and the circuli and annuli were clearly defined and not affected by scale regeneration or significant resorption along the measurement axis. The number of scale samples for each river system and age group are provided in Table 1. Scales from adult sockeye salmon were digitized fol- lowing procedures described by Hagen et al.' and Rug- Table 1 The total number of scale samples for sockeye salmon iOncorynchus nerka) for brood years 1965-97 from the Egegik and Kvichak River systems in Bristol Bay, Alaska. Age groups are 1.2, 1.3, 2.2, and 2.3. River Age group 1.2 1.3 2.2 2.3 Egegik Kvichak 1563 1265 1441 1592 1582 1581 1246 1 Hagen, P. T., D. S. Oxman, and B. A. Agler 2001. Devel- oping and deploying a high resolution imaging approach for scale analysis. Doc. 567, p. 11. North Pacific Anadromous Fish Commission, 889 Pender Street, Vancouver, Canada. gerone et al. (2005). The scale measurement axis was determined by a perpendicular line drawn from a line intersecting each end of the first saltwater annulus. Distance (mm) between the focus and the outer edge of the scale was designated as the total scale length. The relationship between total scale length and adult fork length was linear for both the Egegik River (F-test, P<0.001; r- = 0.41) and Kvichak River (F-test, P<0.001; r^ = 0.36) sockeye salmon samples. For an index of total growth through the first year at sea, we measured the distance from the focus to the outer edge of the first saltwater growth zone for each fish. A time series of annual means of the individual growth during the first year at sea (MSWl^ ^ ,) estimated for each adult fresh- water age group (o represents 1 or 2) within a river system (i represents Egegik, Kvichak) was used as an index of size that sockeye salmon would have attained after their first year it) at sea. The total number of fish caught and the fork lengths (mm) of juvenile sockeye salmon within each trawl haul were recorded during the Bering-Aleutian Salmon In- ternational Survey (BASIS) research cruises along the eastern Bering Sea shelf during fall (August-September) 2000 to 2002 (Fig. 1). The surveys were conducted over a broad area of the shelf and over major oceanograph- ic domains (coastal and middle domains; Kinder and Schumacher, 1981) along the eastern Bering Sea shelf. In addition, the surveys were designed to sample the en- tire population of juvenile sockeye salmon from Bristol Bay lake systems to reduce the chance of sample vari- ability that could affect one's ability to interpret results from these samples. Recent descriptions of juvenile salmon migration pathways along the eastern Bering Sea shelf (Farley et al., 2005) and genetic stock com- position indicate that juvenile sockeye salmon collected during the surveys were primarily from Bristol Bay. Fish were collected by using a mid water rope trawl (see Farley et al., 2005 for description) rigged to sample the top 15 m of the water column. We attempted to collect scales from juvenile sockeye salmon during all three years of the survey; however, sample sizes of scales (from the preferred location on fish) were too small for statistical analyses because of descaling of the juvenile salmon by our mid-water rope trawl. Data Farley et aL: Early marine growth In relation to marine-stage survival rates for Oncorhynchus nerka 123 collected during each trawl included the trawl speed obtained with a global positioning system and the height and width of the net opening obtained with a Simrad FS900 (Simrad, Lynnwood, WA) net sounder. The mean date of collection of juvenile sockeye salmon sampled for length differed slightly between years (i.e., 26 August during 2000; 5 Septem- ber during 2001; 1 September during 2002); lengths were adjusted to account for these differences. Survival and early marine-stage growth rates inferred from adult scales For each freshwater age group, we calcu- lated an index of survival rate that nor- malized the data and removed possible density-dependent effects (i.e., Peterman et al., 1998; Mueter et al., 2002). Specifi- cally, our index of survival rate was the time series of residuals from a Ricker model defined by In "■i.a.2.t+2 "'"■"i,a.3,(+3 (1) = a. ■P.A..- where t = the first ocean year for sock- eye salmon; S = the total number of spawners within river system / (i rep- resents Egegik or Kvichak); R = the total return (catch-i- spawners) for each fresh- water age group a (a repre- sents freshwater age 1 or 2 ) within river system /; a and [i = model parameters represent- ing the number of recruits per spawner at low numbers of spawners and the level of density dependence (Quinn and Deriso, 1999); and £,^ , = the normally distributed residuals of the model. For our analysis, partitioning salmon brood-year produc- tivity by freshwater age group was necessary to directly compare our index of survival with our time series of MSWl,^, growth. Analysis of covariance (ANCOVA) was used to ex- amine the effect of MSWl, ^^ ^ on our indices of survival (see Fig. 2, A-D for scatter plots of f, „ , and MSWl, „ , and the addition of river system, age group and year were used as factors in the model. The results indi- cated that the year factor was highly significant (F-test, P<0.001) and that MSWl, „ , was not significant (F-test, ■60°0'0"N ■55°0'0"N 165°0'0"W 160°0'0"W Figure 1 Survey area of the annual August-September (2000-2002) Bering- Aleutian Salmon International Survey (BASIS) within the coastal and middle domains of the eastern Bering Sea. P=0.18). It was possible that during some years all fish could have had excellent growth and attained a large size, but the ANCOVA model would have attributed the large size to the highly significant year factor. However, when we removed the year factor from the ANCOVA model. MSWl, ^, was less significant (F-test, P=0.27). In addition, the residuals from these models contained significant positive autocorrelation. Because our data contained significant autocorrelation and showed a time series character, we created univari- ate time series models (Wei, 1990) for both MSWl, ^ , and f, Q , to determine whether autoregressive or moving average components were present. The univariate mod- els were developed by examining the sample autocor- relation and partial autocorrelation functions for each time series. Time series data were considered white noise processes, i.e., uncorrelated random variables 124 Fishery Bulletin 105(1) 3-1 A 2- 1 0- -1 - • -2- Residuals 5 1.6 1.7 1.8 19 B 2- • 1 0- ■1 • -2- -3 J 1.5 1.6 1.7 1.8 1.9 Msw^ Figure 2 The relationship between the index of marine-stage survival rate (residuals) and growth after the first year at sea iMSWX) for Egegik freshwater age groups 1 (A) and 2 iB) and Kvichak freshwater age groups 1 (C) and 2 (D) sockeye salmon (Oncorhynchus nerka). with constant mean and variance, when none of the components of the sample autocorrelation and partial autocorrelation functions differed significantly (P=0.01) from zero (Wei, 1990). Multivariate time series models in the form of linear transfer function (LTF) models (Liu and Hudak, 1992) were developed to describe the relationship between MSWl^ ^, I and our index of survival rate for each fresh- water age group and river system. All the univariate time series of survival rate indices contained signifi- cant positive first-order autoregressive parameters (see Table 2). Therefore, we included the first order autore- gressive parameter in the LTF models. The models were defined as f,.a.,='-,..+^,.aMSWl,,,+ (l-0i.,.oB) -N. (2) where t, i, a are described in Equation 1; MSWl = the early marine growth index for sockeye salmon during their first year at sea; the autoregressive lag 1 parameter; the symbol for the backshift operator (i.e.. B c, g and A, ^ = parameters within the model; and N^^i = a sequence of random errors that are independently and identically distributed with a normal distribution. Parameters within the univariate and LTF models were deemed significant when their lvalue was greater than 2.0(P<0.05). Autocorrelation analysis was used to examine wheth- er the model residuals were white noise. Univariate and LTF models were compared by using Schwartz's Bayes- ian criterion (SBC; Wei, 1990) to determine if the inclu- sion of MSWl in Equation 2 improved the model fit. Analyses of data from juveniles collected by trawling Next, we developed an index of relative survival rate of adult Bristol Bay sockeye salmon for 2000-02, and indices of abundance and marine-stage survival rate of juvenile sockeye salmon collected during 2000-02 to compare with mean lengths of juvenile sockeye salmon collected during those years. Relative survival was defined as the number of returning adult sockeye salmon from brood-year escapements that contributed to the Farley et al.: Early marine growth in relation to marine-stage survival rates for Oncorhynchus nerka 125 3- c 2- 1 0- -1 -2- * • .♦♦ . Ul 13 1.5 1.6 1.7 1.8 1.9 3-| D 2- 1 - 0- -1 • • • • -2- • • ' ' ' 1.5 1.6 1.7 MSW^ Figure 2 (continued) 1.8 1.9 Table 2 Univariate (Univ) and linear transfer function (LTF) models indicating the effect of early marine growth index (MSWl) on the index of marine stage survival rates for each freshwater age group of sockeye salmon [Oncorhynchus nerka) returning to the Egegik and Kvichak rivers. Other variables include the constant [Const) and the autoregressive parameter ^^. Model statistics included the number of effective observations)??), the coefficient of determination (r-), residual standard error (i?S£), the number of parameters (M), and Schwartz's Bayesian criterion {SBC). River Age (yr) Model n r2 RSE M SBC Model coefficients ^value Const. MSWl 4>, Const. MSWl •P, Egegik 1 Univ 32 0.34 0.958 2 4.19 0.13 0.48 0.40 — 3.37 LTF 32 0.41 0.908 3 4.22 -10.38 6.36 0.30 -2.22 2.25 2.07 2 Univ 32 0.17 0.630 2 -22.64 0.02 — 0.42 0.10 — 2.64 LTF 32 0.25 0.601 3 -22.19 -6.61 3.69 0.41 -1.74 1.75 2.59 Kvichak 1 Univ 32 0.29 1.075 2 11.56 -0.15 — 0.54 -0.36 — 3.60 LTF 32 0.29 1.074 3 14.97 1.06 -0.73 0.54 0.21 -0.24 3.62 2 Univ 32 0.35 0.986 2 6.00 -0.32 — 0.64 -0.64 — 3.99 LTF 32 0.35 0.985 3 9.43 0.71 -0.59 0.64 0.18 -0.26 3.98 juvenile sockeye salmon in the trawl samples taken during years 2000, 2001, and 2002. For instance, juve- nile sockeye salmon contributing to the early marine population during 2000 comprised age-2.0 fish from the 1997 and age-1.0 fish from the 1998 brood-year escape- ments, and these fish would have returned as adults during 2002 and 2003. Relative marine-stage survival rate (RS) was thus calculated as 126 Fishery Bulletin 105(1) RS, 2 I a=l S*-^a.2.(+2+-Ra.3./+3) (S,.2 + S,.3)/2 (3) where t = the year juvenile sockeye salmon were sam- pled (f=2000, 2001, 2002); a = the freshwater age (1.0 or 2.0); R = is the total number of returning adult sock- eye salmon to Bristol Bay after year t; and S = the total number of spawners in Bristol Bay that contributed to the juvenile salmon population during year t. For instance, freshwater juvenile sockeye salmon (ages 1.0 and 2.0) sampled during 2000 came from cohorts spawning during 1998 (age 1.0) and 1999 (age 2.0) and returned to Bristol Bay during 2002 as adult salmon at age 1.2 and 2.2 in 2002 and at age 1.3 and 2.3 in 2003. The numbers of returning adult and spawning Bristol Bay sockeye salmon were estimated from brood-year return information provided by ADF&G. Annual indices of juvenile sockeye salmon abundance (lA) were defined as IA,=[SA/ina)C,, (4) where SA = the estimated survey area (189,000 km^); rna = the mean area sampled by a trawl haul during the survey (distance traveled during the tow multiplied by the width of the net); and the mean number of juvenile sockeye salmon caught during year t ('?=2000, 2001, 2002). C, This formula would give the abundance of juvenile sock- eye salmon in the survey area if we assumed that catch- ability with our midwater trawl was 1, (i.e., all fish in front of the net were caught). Because this is unlikely, we treat our estimates as an index rather than as actual abundance. In fact, our juvenile sockeye salmon abun- dance indices were less than the resultant adult returns in some years, indicating that catchability of the net was much less than 1. One study (Shuntov et al., 1993) where larger surface trawl gear was used to sample juvenile salmon indicated that catchability of juvenile salmon was 0.3. We therefore divided our abundance indices by 0.3, although we still considered these values to be indices. An index of juvenile sockeye salmon marine-stage survival rate (IMS) was estimated by sampled (^=2000, 2001, 2002), and a is freshwater age (age 1.0 or 2.0). These survival rate indices were correlated with the mean length of juvenile fish collected during the cor- responding first year at sea. Because the mean date for juvenile sockeye salmon sampled for length differed among years, we adjusted fish lengths to provide a standardized length using September 1 as the standard date. Adjusted mean fish lengths were calculated by as- suming three different daily growth rates: 1) mm/day, representing no daily growth at sea; 2) 0.3 mm/day, the lower end of published growth-rate ranges for juvenile Pacific salmon; and 3) 1.7 mm/day, representing the upper end of the ranges (see Fisher and Pearcy, 1988, 1990; Fukuwaka and Kaeriyama, 1994; Orsi et al., 2000 for daily growth-rate ranges for juvenile Pacific salmon). Results Analyses of adult scale data Examination of the autocorrelation and partial auto- correlation functions for Kvichak River freshwater age groups 1 and 2 and the Egegik River freshwater age- group-1 MSWl univariate time series indicated that these time series had a constant mean and variance. For the Egegik River freshwater age-group-2 MSWl growth index, the sample autocorrelation and partial autocorrelation functions indicated that a lag-1 autore- gressive parameter was appropriate and the estimate of the parameter was significant (^test, P<0.01). Coef- ficients of variation were less than 4% for the MSWl growth-rate indices for each freshwater age group, and thus confirmed the univariate model results that these time series varied little over time. By comparison, the coefficients of variation for the time series of returns per spawner for each freshwater age group were between 70% and 135%. The MSWl growth index was not significantly related to survival in any of the LTF models except for Egegik freshwater age group 1 (Table 2). Parsimonious univari- ate models were reasonable explanations of survival for both river systems and age groups, having values of SBC nearly as low as the "best" models. The sample autocorrelation and partial autocorrelation functions indicated that a lag-1 autoregressive parameter was appropriate for the all of the univariate survival rate time series models. The estimates of the lag-1 autore- gressive parameter were positive for all of the univari- ate models. IMS, = s^ X 100, lA, (5) where R is defined above in Equation 3, /A, is defined in Equation 4, t is the year juvenile sockeye salmon were Analyses of data from juveniles collected by trawling The distribution of juvenile sockeye salmon along the eastern Bering Sea varied among years (Fig. 3). During 2000 and 2001, 75% of the total catch of juvenile sock- eye salmon occurred south of 56°N, within the middle domain and south within the stratified waters near the Farley et al.: Early marine growth in relation to marine-stage survival rates for Oncorhynchus nerka 127 100-. 90- 80- //' / 2000 o 50- // / 2001 1 40- £ 30- ^ 20- .'' / 2002 10- j/ ^y . / ^^^^ 54.5 55 55.5 56 56.5 57 57.5 58 58.5 59 59.5 60 Latitude =N Figure 3 The percentage of total catch of juvenile sockeye sal- mon (Oncorhynchus nerka) in relation to latitude (°N) along the eastern Bering Sea shelf during August- September 2000. 2001, and 2002. coastal domain along the Alaska Peninsula. During 2002. 75% of the total catch of juvenile sockeye salmon occurred north of 57°N, with 50% of the total catch occurring north of 58°N within the shallow stratified waters near the northern coastal domain. Average fork length of juvenile sockeye salmon was significantly smaller during 2001 than during 2000 and 2002 for growth rates greater than 0.3 mm/day (^-test; P<0.01) and not significantly different from 2000 (Mest; P=0.05) for growth rates equal to mm/day (Table 3). The rank order of juvenile sockeye salmon fork lengths was the same for all growth rates, and the largest fish taken in 2002 and the smallest, in 2001. For all three growth rates, average fork length was significantly larger during 2002 than during 2000 and 2001 (^test; P<0.01). The marine-stage survival rate and abundance in the indices mirrored the observed variation in fish fork length; they were highest during 2002 and at or near their lowest during 2001 (Table 4). In addition, the nearshore distribution of juvenile sockeye salmon (2001; Fig. 3) appeared to coincide with lower indices of abun- dance and marine-stage survival rate, whereas fish distributed in the northern area of our survey (2002; Fig. 3) exhibited higher marine-stage survival rate and abundance. Discussion Our study indicates that the size of Bristol Bay sockeye salmon after their first year at sea is not directly related to their survival, when size is measured from growth rings on the scales of adults returning to the Egegik and Kvichak rivers. Analyses of the MSWl growth index indicated that most of the time series had a constant mean and variance. Similar studies where adult scales Table 3 Average fork length of juvenile sockeye salmon (Onco- rhynchus nerka) collected along the eastern Bering Sea during 2000, 2001. and 2002. Daily growth rate ( mm ) was assumed to be 0, 0.30, and 1.7. Statistics include sample size (n), average fork lengths, and standard deviation (SDl of the original length data. 2000 2001 2002 834 802 1936 174.77 171.91 197.96 176.53 170.61 197.94 184.76 19.99 164.53 35.55 197.83 34.74 Table 4 Indices of abunc ance (M), marine-stage survival rate (IMS) and relative marine-stage survival rate (RS) for | juveni le sockeye salmon (Oncorhyncus nerka) collected along the eastern Bering Sea during 2000, 2001, and | 2002. Year M IMS RS 2000 130 21?^ 3.8 2001 137 15% 1.9 2002 180 34% 6.0 from Atlantic (Salmo salar L.; Crozier and Kennedy, 19991, coho (O. kisutch; Briscoe, 2004), and chum (O. keta\ Helle, 1979) salmon were used to measure growth (size) of salmon during their first year at sea revealed that the survival rate of a cohort was statistically unre- lated to variation in growth (size) of the salmon. The relative uniformity in the size of salmon after their first year at sea and the lack of a relationship between size and survival rate is contrary to the prevailing paradigm that the size achieved by fish after their first summer at sea is important to survival (Beamish and Mahn- ken, 2001). However, these results do not necessarily invalidate this paradigm; the adult scale samples avail- able for analysis may reflect only those juvenile salmon that had attained sufficient size in order to survive to adulthood, and not those that died at sea (Crozier and Kennedy, 1999). In support of this possibilty, when we directly mea- sured the fork length of juvenile sockeye salmon (Tables 3 and 4) during late summer and early fall surveys along the eastern Bering Sea shelf (2000-02), smaller fish had lower indices of marine-stage survival rate. This result is consistent with that from other studies of teleost fish, where larger individuals gained a survival advantage over smaller conspecifics during the juvenile life-history stage (Parker, 1968; Healey, 1982; Holtby et 128 Fishery Bulletin 105(1) al., 1990; Pearcy, 1992; Sogard, 1997; Mortensen et al., 2000; Beamish and Mahnken, 2001; Moss et al, 2005). This result is also in accord with the critical-size and critical-period hypothesis in which brood-year survival is determined by the number of juvenile salmon that have reached a critical size by the end of their first ma- rine summer (Beamish and Mahnken, 2001; Beamish et al., 2004). The assumption with this hypothesis is that fish that do not reach a critical size after their first summer at sea will die because they are unable to meet minimum metabolic requirements during late fall and winter (Beamish and Mahnken, 2001). Although our re- sults indicate that larger juvenile sockeye salmon have higher relative marine-stage survival rate after their first year at sea, it is difficult to directly address when the mortality would occur because sockeye salmon can spend an average of 2 to 3 years at sea. However, the overwhelming evidence from field and laboratory studies of juvenile stages of teleost fishes seems to indicate that size-selective mortality occurs during winter because larger members of a cohort are better than smaller members at tolerating physical extremes and enduring longer periods without food (Sogard, 1997). One other test of the critical-size and critical-pe- riod hypothesis is that mortality rates after this period should be large in relation to other sources of early ma- rine mortality (Beamish et al., 2004). To interpret our indices of marine-stage survival rate as the actual post- survey marine survival rate requires making a variety of questionable assumptions (e.g., that the vulnerability of juvenile salmon to our gear is known). However, if our estimates are close to correct, they would indicate that marine-stage mortality rates of juvenile sockeye salmon may be greater than 70% (Table 4) after our late-summer-early-fall surveys. These marine-stage mortality rates are substantial and approach late fall and winter mortality rates of greater than 90% found for other Pacific salmon (Beamish et al., 2004). Lengths of juvenile sockeye salmon differed signifi- cantly among years if we assumed daily growth rates of 0.3 mm and greater. Differences in fork length of juvenile sockeye salmon could reflect annual differ- ences in early marine growth rates or may also reflect annual differences in the size of smolt leaving Bristol Bay lake systems. However, limited surveys of sockeye salmon smolt from the Kvichak and Ugashik Rivers during 2000 through 2002 (Egegik River sampling was not undertaken in 2002) by ADF&G indicate that dif- ferences in smolt length among years and within age classes and river systems were less than 9%. In ad- dition, the smallest average smolt size among these three years was seen during 2002, the year with the largest juvenile sockeye salmon size. Thus, it is likely that annual differences in length observed during our survey were due to differences in marine growth rates between years. The annual variability in juvenile sockeye salmon size and in indices of marine-stage survival rates may be linked to the early marine migration of these salmon along the eastern Bering Sea shelf. Although we had only three years of data, size and survival indices of Bristol Bay sockeye salmon were lowest when juvenile sockeye salmon were distributed nearshore along the Alaska Peninsula (i.e., the coastal migration pathway) and highest when they were distributed farther north and offshore. In support of this theory, the coastal mi- gration pathway of juvenile Bristol Bay sockeye salmon observed by Straty (1981) during the late 1960s and early 1970s coincided with a significantly lower produc- tion of Bristol Bay sockeye salmon that occurred before the mid 1970s (Adkison et al., 1996). The annual variability in seaward migration path- ways is likely related to ocean conditions on the shelf during spring and summer. Recent studies indicate that sea surface temperatures along the eastern Bering Sea in summer, the period when juvenile sockeye salmon are present on the shelf, is positively correlated with Bristol Bay sockeye salmon survival rates (Mueter et al., 2002). It is possible that the effect of sea surface temperatures on survival rates of juvenile Bristol Bay sockeye salmon is a result of its influence on early marine distribution of juvenile sockeye salmon. For example, during the late 1960s and early 1970s, the nearshore migration of juvenile Bristol Bay sockeye salmon was thought to be a result of sockeye salmon using the warmer near- shore waters rather than the colder sea surface tem- peratures offshore in order to maximize their growth (Straty, 1981). Depth-averaged sea temperatures from an oceanographic mooring along the eastern Bering Sea middle shelf domain from mid-July to mid-September were consistently warmer during 2001 through 2002 than during 1995 through 1997 (Overland and Sta- beno, 2004). Presumably, the warmer sea temperatures during 2001 would have been conducive to offshore migration of juvenile sockeye salmon during that year. Although sea temperatures were warmer during 2001 through 2002, sea temperatures along the shelf were 1° to 2°C cooler from late June to September during 2001 than during 2002 (Overland and Stabeno, 2004). Thus, it may be that warmer sea temperatures during the time juvenile sockeye salmon first are present over the eastern Bering Sea shelf (beginning in June) provide a conduit for rapid offshore migration (and possibly higher survival) and that cooler sea temperatures delay offshore migration. Our results indicate that after the first summer in the Bering Sea, larger juvenile sockeye salmon may gain a survival advantage over smaller individuals. This result, coupled with previous findings of reduced juvenile-to-adult survival for pink (Moss et al., 2005) and coho (Beamish et al., 2004) salmon that spend their first summer in the coastal waters of the Gulf of Alaska and Strait of Georgia, indicates that reduced growth of Pacific salmon during their first year at sea may lead to substantial salmon mortality, presumably during their first winter at sea. This phenomenon may not be seen if size of the salmon after their first year at sea is inferred from the scale growth increments of returning adults, because these individuals could be a biased sample from the faster-growing portion of the Farley et al.: Early marine growth in relation to marine-stage survival rates for Oncorhynchus nerka 129 population (Crozier and Kennedy, 1999). We suggest that annual variability in the size of sockeye salmon may be related to summer sea surface temperatures along the eastern Bering Sea shelf temperatures that appear to influence the spatial distribution and early marine migration pathways of this species. Acknowledgments P. Hagen, formally with the Alaska Department of Fish and Game in Juneau. Alaska, was instrumental in aging and creating digital pictures of juvenile sockeye salmon scales. We gratefully appreciate the help of the captains (M. Cavanaugh and S. Brandstitter) and the crew of the FV Sea Storm and captain C. Bronson and the crew of the FV Great Pacific for their fine efforts and technical assistance in all aspects of the field surveys. We thank A. Wertheimer, G. Kruse, and three anonymous review- ers for suggestions during this study and comments on the manuscript. Literature cited Adkison, M. D., R. M. Peterman, M. F. Lapointe, D. M. Gillis, and J. Korman. 1996. Alternative models of climatic effects on sockeye salmon, Oncorhynchus nerka, productivity in Bristol Bay. Alaska, and the Fraser River, British Columbia. Fish. Oceanogr. 5:137-152. Beamish, R. J., and C. Mahnken. 2001. A critical size and period hypothesis to explain natural regulation of salmon abundance and the link- age to climate and climate change. Prog. Oceanogr. 49:423-437. Beamish, R. J., C. Mahnken, and C. M. Neville. 2004. Evidence that reduced early marine growth is asso- ciated with lower marine survival of coho salmon. Trans. Am. Fish. Soc. 133:26-33. Burgner, R. L. 1991. Life history of sockeye salmon (Oncorhynchus nerka). In Pacific salmon life histories (C. Groot and L. Margolis, eds.), p. 3-117. Univ. British Columbia Press, Vancouver, Canada. Briscoe, R. J. 2004. Factors affecting marine growth and survival of Auke Creek, Alaska coho salmon {Oncorhynchus kisutch). M.S. thesis. 57 p. Univ. Alaska Fairbanks. Fairbanks, AK. Crozier. W. W., and J. A. Kennedy. 1999. Relationships between marine growth and marine survival of one sea winter Atlantic salmon. Salmo salar L.. from the River Bush. Northern Ireland. Fish. Manage. Ecol. 6:89-96. Farley, E. V., Jr., J. M. Murphy, B. W. Wing, J. H. Moss, and A. Middleton. 2005. Distribution, migration pathways, and size of west- ern Alaska juvenile salmon along the eastern Bering Sea shelf Alaska Fish. Res. Bull. ll(l):15-26. Fisher, J. P., and W. G. Pearcy. 1988. Growth of juvenile coho salmon (Oncorhynchus kisutch) off Oregon and Washington, USA, in years of differing coastal upwelling. Can. J. Fish. Aquat. Sci. 45:1036-1044. 1990. Spacing of scale circuli versus growth rate in young coho salmon. Fish. Bull. 88:637-643. Francis, R. I. 1990. Back-calculation of fish length: a critical review. J. Fish Biol. 36:883-902. Fukuwaka, M., and M. Kaeriyama. 1994. A back-calculation method for estimating indi- vidual growth of juvenile chum salmon by scale analysis. Scientific Reports of the Hokkaido Salmon Hatchery 48:1-9. 1997. Scale analyses to estimate somatic growth in sock- eye salmon, Oncorhynchus nerka. Can. J. Fish. Aquat. Sci. 54:631-636. Hartt, A. C. 1980. Juvenile salmonids in the oceanic ecosystem; the critical first summer. In Salmonid ecosystems of the North Pacific (W. J. McNeil and D. C. Him- swortb, eds.), p. 25-57. Oregon State Univ. Press, Corvallis, OR. Healey, M. C. 1982. Timing and relative intensity of size-selective mortality of juvenile chum salmon (Oncorhynchus keta) during early sea life. Can. J. Fish. Aquat. Sci. 39:952-957. Helle, J. H. 1979. Influence of marine environment on age and size at maturity, growth, and abundance of chum salmon, (Oncorhynchus keta) from Olsen Creek, Prince William Sound, Alaska. Ph.D. diss., 118 p. Oregon State Univ., Corvallis, OR. Holtby, L. B., B. C. Andersen, and R. K. Kadowaki. 1990. Importance of smolt size and early ocean growth to interannual variability in marine survival of coho salmon (Oncorhynchus kisutch). Can. J. Fish. Aquat. Sci. 47:2181-2194. Kinder, T. H., and J. D. Schumacher. 1981. Hydrographic structure over the continental shelf of the southeastern Bering Sea. In The eastern Bering Sea shelf: oceanography and resources, vol. 1 (D. W. Hood and J. A. Calder, eds.), p. 31-52. Univ. Washington Press, Seattle, WA. Koo, T. S. Y. 1962a. Age designation in salmon, /n Studies of Alaska red salmon (T. S. Y. Koo, ed.), p. 41-48. Publications in Fisheries, New Series 1, Univ. Washington, Seattle, WA. 1962b. Age and growth studies of red salmon scales by graphical means. In Studies of Alaska red salmon (T. S. Y. Koo, ed.), p. 49-121. Publications in Fisheries, New Series 1, Univ. Washington, Seattle, WA. Liu, L., and G. B. Hudak. 1992. Forecasting and time series analysis using the SCA statistical system, 435 p. Scientific Computing Associates Corporation, Oak Brook, IL. Mortensen, D., A. Wertheimer, S. Taylor, and J. Landingham. 2000. The relation between early marine growth of pink salmon, Oncorhynchus gorbuscha, and marine water temperature, secondary production, and survival to adulthood. Fish. Bull. 98:319-335. Moss, J. H., D. A. Beauchamp, A. D. Cross, K. Myers, E. V. Farley Jr, J. M. Murphy, and J. H. Helle. 2005. Higher marine survival associated with faster growth for pink salmon (Oncorhynchus gorbuscha). Trans. Am. Fish. Soc. 134:1313-1322. 130 Fishery Bulletin 105(1) Mueter, F. J., D. M. Ware, and R. M. Peterman. 2002. Spatial correlation patterns in coastal environmen- tal variables and survival rates of salmon in the north- east Pacific Ocean. Fish. Oceanogr. 11(4):205-218. Orsi, J. A., M. V. Sturdevant, J. M. Murphy, D. G. Mortensen, and B. L. Wing. 2000. Seasonal habitat use and early marine ecology of juvenile Pacific salmon in southeastern Alaska. N. Pac. Anad. Fish Comm. Bull. 2:111-122. Overland, J. E., and P. J. Stabeno. 2004. Is the climate of the Bering Sea warming and affecting the ecosystem? EOS, Trans. Am. Geophys. Union. 85(33):309-312. Parker, R. R. 1968. Marine mortality schedules of pink salmon of the Bella Coola River, Central British Columbia. J. Fish. Res. Board Can. 25:757-794. Pearcy, W. G. 1992. Ocean ecology of the North Pacific salmonids, p. 179. Univ. Washington Press, Seattle, WA. Peterman, R. M., B. J. Pyper, M. F. Lapointe, and M. D. Adkison. 1998. Patterns of covariation in survival rates of British Columbian and Alaskan sockeye salmon (Oncorhynchus nerka) stocks. Can. J. Fish. Aquat. Sci. 55:2503-2517. Quinn, T. J., II, and R. B. Deriso. 1999. Quantitative fish dynamics, 542 p. Oxford Univ. Press, New York, NY. Ricker, W. E. 1992. Back-calculation of fish lengths based on pro- portionality between scale and length increments. Can. J. Fish. Aquat. Sci. 49:1018-1026. Ruggerone, G. T., E. V. Farley Jr., J. Nielsen, and P. Hagen. 2005. Seasonal marine growth of Bristol Bay sockeye salmon {Oncorhynchus nerka) in relation to competition with Asian pink salmon (O. gorbuscha) and the 1977 ocean regime shift. Fish. Bull. 103:355-370. Shuntov, V. P., V. I. Radchenko, V. V. Lapko, and Y. N. Poltev. 1993. The distribution of the Pacific salmons in the Sakhalin-Kuril region at a period of anadromous migra- tion (in Russian). J. Ichthy. Moscow 33:348-358. Sogard, S. M. 1997. Size-selective mortality in the juvenile stage of teleost fishes: a review. Bull. Mar. Sci. 60(3):1129- 1157. Straty, R. R. 1981. Ecology and behavior of juvenile sockeye salmon (Oncorhynchus nerka) in Bristol Bay and the eastern Bering Sea. In The eastern Bering Sea shelf: ocean- ography and resources, vol. 2 (D. W. Hood and J. A. Calder, eds.), p. 285-320. Univ. Washington Press, Seattle, WA. Wei, W. W. S. 1990. Time series analysis: univariate and multivari- ate methods, 478 p. Addison-Wesley Publ. Company, Redwood City, CA. Willette, T. M., R. T. Cooney, and K. Hyer. 1999. Predator foraging mode shifts affecting mortality of juvenile fishes during the subartic spring bloom. Can. J. Fish. Aquat. Sci. 56:364-376. 131 Abstract — Inaccuracy in the aging of postovulatory follicles (POFs) and in estimating the effect of temperature on the resorption rate of POFs may introduce bias in the determination of the daily spawning age classes with the daily egg production method (DEPM). To explore the above two bias problems with field-collected European pilchard iSardina pilchar- dus, known regionally as the Iberian sardine), a method was developed in which the time elapsed from spawning {POF age) was estimated from the size of POFs (i.e., from the cross-sec- tional area in histological sections). The potential effect of the preserva- tive type and embedding material on POF size and the effect of ambient water temperature on POF resorp- tion rate are taken into account with this method. A highly significant log- linear relationship was found between POF area and age; POF area shrank by approximately 50% per day. POFs were also shown to shrink faster at higher temperatures (approximately 3'7c per degree), but this temperature effect is unlikely to be an important source of bias in the assignment of females to daily spawning classes. The embedding material was also shown to influence the size of POFs, the latter being significantly larger in resin than in paraffin sections. In conclusion, the size of POFs provides an indirect, reliable estimation of the time elapsed from spawning and may thus be used to test both the validity of POF staging criteria for identify- ing daily classes of spawners and the effect of other factors (such as tem- perature and laboratory processing) in applications of the DEPM to S. pilchardus and other fish species. Degeneration of postovulatory follicles in the Iberian sardine Sardina pilchardus: structural changes and factors affecting resorption Konstantinos Ganias^-^ Cristina Nunes^ Yorgos Stratoudakis^ Email for K Ganias: kganias@bio.auth.gr ' School of Biology Laboratory ol Ichthyology Aristotle University of Thessaloniki 54 124 Thessaloniki, Greece ^ Institute Nacional de Investigacao Agraria e das Pescas Institute de Investigacao das Pescas e do Mar Avenida de Brasilia s/n, 1449-006 Lisbon, Portugal Manuscript submitted 12 May 2006 to the Scientific Editor. Manuscript approved for publication 5 July 2006 by the Scientific Editor Fish. Bull. 105:131-139 (2007). The postovulatory follicle (POF) consists of the follicular layers that remain in the ovary of fish after the release of the ovum during spawning (Saidapur, 1982). Initially, the POF is a distinct structure, but it rapidly deteriorates and becomes undetectable within a few days (Hunter and Gold- berg, 1980). The study of POF degen- eration is important in fishery studies because it permits the assignment of spawning females to daily classes to provide estimates of the daily frac- tion of spawning fish in the population (POF method, Hunter and Goldberg, 1980). The most common application of the POF method is in the daily egg production method (DEPM; Parker, 1980), where spawning fraction, together with other adult parame- ters, are used to estimate daily spe- cific fecundity (Picquelle and Stauffer, 1985) for the fisheries-independent estimation of spawning biomass. A prerequisite for such applications is the existence of an accurate aging key that describes the time course of POF degeneration (Hunter and Macewicz, 1985). In most DEPMs, the degeneration of POFs is described by a small num- ber of histomorphological stages (see "Materials and methods" section) that are usually assumed to correspond to distinct daily classes (see review by Stratoudakis et al., 2006). How- ever, because the process of POF de- generation is continuous and DEPM samples are usually obtained oppor- tunistically throughout the day, the direct assignment of POF stages to daily classes of spawning fish can be imprecise. Also, morphological stages are often attributed to daily classes without prior validation and thus can lead to biased estimates of the spawning fraction. Validation is best performed in the laboratory by sacri- ficing female spawning fish at known time intervals after ovulation (e.g.. Hunter and Goldberg, 1980; Perez et al., 1992). Alternatively, in fish with daily spawning synchronicity, such as with the Iberian sardine Sardina pilchardus (also known as the Euro- pean pilchard, FAO, 1985) (Bernal et al., 2001; Zwolinski et al., 2001; Ganias et al., 2003), validation can be performed indirectly through the examination of field samples collected at different hours of the day (Gold- berg et al., 1984). Another source of potential bias in the POF method is the duration of follicular degeneration, which may be temperature-dependent (Hunter and Macewicz, 1985), because the meta- bolic rates of poikilotherms, like fish, may be directly affected by ambient temperature. As a result, POF de- generation may be faster at higher temperatures and may lead to biased 132 Fishery Bulletin 105(1) LNl?"(!- 'm Figure 1 (A) Microphotograph of a postovulatory follicle (POF) situated at the epithelium of an ovarian lamella (L) in the Iberian sardine Sardina pilchardus; (B) the same POF magnified, indicating the perimeter of its cross-sectional area (XSA) and its diameter (D). Scale bar = 0.1 mm. estimates of the spawning fraction and biomass when spatial variation in ambient water temperature is large within a survey area. This dependence can be assessed under laboratory conditions through inspection of speci- mens spawning under different temperature regimes (Fitzhugh and Hettler, 1995). Alternatively, the effect of temperature on POF degeneration can be studied in the wild by the examination of individuals caught at variable environmental conditions (Ganias et al., 2003; Roumillat and Brouwer, 2004). Because of the above unresolved issues of the POF method (that can be species specific), the spawning fraction remains the most poorly estimated DEPM pa- rameter (Hunter and Lo, 1997; Stratoudakis et al., 2006). The main objective of our study was to devise a method whereby one can test indirectly for potential sources of bias in the attribution of stage and age to the POFs of the field-collected sardine S. pilchardus. We used the size of POFs (cross-sectional area and dia- meter) as an index of POF age and, together with other histomorphological characteristics (follicle shape, state of the granulosa layer), we refined existing criteria for determining the stage and age of POFs in sardine. Then, we modeled the size of POFs as a function of POF age, ambient temperature, type of preservative, and type of embedding material. The analysis allowed us to test whether our staging criteria were valid and to examine whether temperature and laboratory process- ing introduced bias in the process. Materials and methods Adult sardines were collected off Portugal and the Gulf of Cadiz in 1997, 1999, and 2005 within the framework of DEPM surveys for the estimation of the spawning biomass of the Atlanto-Iberian sardine stock (ICES, 2004). Sampling was conducted during peak spawning months for sardine (January-March), either on board the RV Noruega or from the commercial purse-seine fleet. During the surveys, sea surface temperature (SST) was recorded on an extensive grid representing hydrographic casts that covered the whole sampling area during the ichthyoplankton surveys (ICES, 2004). Fish gonads were immediately removed after capture and placed in jars either with AFA (65 parts by volume of 50% alcohol, 32 parts formalin, and 13 parts glacial acetic acid) solution [1997, 1999] or with 4% formalin [2005]). In the laboratory, ovaries were embedded ei- ther in paraffin (1997, 2005) or in resin (1999), and histological sections (paraffin: 5 f> , where a and b are parameters. (1) In case of isometry, (i.e., 6 = 2), shrinkage is supposed to occur evenly in all dimensions, and the shape of the regressing POFs does not un- dergo significant changes. In any other case (fe>2, or 6<2), the shape is altered along resorp- tion, and thus its intermediate phases may be used in the staging of POFs. Existing histological criteria for the stag- ing of sardine POFs were refined and based on two hauls from the 2005 survey that contained more than 90 histological specimens each. Giv- en that individuals in each sample were caught at the same time and temperature, POFs in each daily cohort should have been at the same stage and have had the same age, maximizing the morphological contrast among daily classes. Furthermore, one of these hauls was performed in the evening, just before the average daily spawning hour for the Iberian sardine (20:00; Zwolinski et al.. 2001; ICES, 2004) and thus provided information on the final histological state of each daily class of POFs. After refining the staging criteria and clas- sifying the POFs from all surveys into daily classes, we used the time of capture and daily spawning hour to estimate the exact age of POFs, i.e., the time in hours elapsed between spawning and sampling. The effect of POF age and other factors on POF cross-sectional area were tested with a generalized linear model (GLM) with an overdispersed Poisson distribu- tion and a logarithmic link function. Apart from POF age, the effect of temperature, sampling year, and the one-way interaction of year with age were considered in the model. The significance of the relationship between POF age and size indicated whether our staging and aging criteria were accurate. Furthermore, because the laboratory processing of the ovaries differed from year to year, the year effect and its interaction with age was used to test the effect of preservation medium (AFA, formalin) and embedding material (paraffin, resin) on the size of POFs. Residual inspection plots revealed the adequacy of the fitted model. Results A total of 249 ovaries with POFs were detected and used in our analysis (1997: 65, 1999: 104, 2005: 80). The two hauls from the 2005 survey that were used for refining POF staging criteria contained many females with POFs at various stages and sizes, facilitating the distinction between successive daily classes of spawners. The frequency distribution of POF cross-sectional areas in the two hauls displayed three size modes, which were considered to correspond to different age classes (Fig. 2). 0.005 0.009 0.013 0.017 0.021 0.005 0.009 0.013 0.017 0.021 POF area (mm^) Figure 2 Frequency distribution of postovulatory follicle (POF) cross-sec- tional areas from two samples of the Iberian sardine Sardina pilchardus from the 2005 survey. (A) Sample collected at 12:00, /! = 48; (Bl sample collected at 17:30. n=32. Arrows indicate POF size modes in each sample. The most advanced size mode in the sample hauled at 12:00 (Fig. 2A) consisted of larger POFs compared to the POFs in the sample hauled at 17:30 (Fig. 2B). Re- examination and comparison of POF size modes in each sample showed that, apart from size, they differed in shape and in the fine histological characteristics of the follicular layers (Fig. 3). In particular, large POFs had an irregular shape and contained a large, convoluted, and thick granulosa layer (Fig. 3A). In the intermediate size mode, the shape of the follicle changed to semirect- angular and the granulosa layer tended to lose its con- voluted appearance and to form a single layer (Fig. 3B). Finally, in the smaller size mode, all POFs displayed a triangular shape (Fig. 3, C-E). However, more detailed examination showed that these triangular POFs could be further separated into 1) a group of slightly larger POFs with a thin layer of the granulosa (Fig. 3C), and 2) a group of very small POFs that contained only granu- losa remnants in the form of residual vacuoles (Fig. 3, D and E). POF diameter increased significantly with POF cross- sectional area (Fig. 4) and the relationship was not significantly different between the paraffin samples from 1997 and 2005 (P>0.05). The allometric coefficient b of Equation 1 differed significantly from the square 134 Fishery Bulletin 105(1) Figure 3 (A-El Microphotographs of postovulatory follicles (POFs) at consecutive phases of deterioration in the Iberian sardine Sardina pilchardus, showing the degeneration of POF size and shape, and the state of the granulosa (same scale for all images). Scale bar = 0.1 mm. (P<0.001), indicating that POF resorption in the Iberian sardine is not isometric, i.e., the shape of POFs changes throughout degeneration. More specifically, b was esti- mated to be 1.5 (standard error=0.13), indicating that the diameter of POFs along the lamellar epithelium diminishes at a lower rate than that for the overall POF area. This allometric pattern of POF resorption confirms the shape differences along POF degeneration described above and shown in Figure 3. The differences in the dimensional characteristics and the morphological state of the granulosa (Table 1) were used to assign POFs from all surveys into four daily classes. The reliability of these aging criteria was confirmed by a very good relationship between POF age and POF size in all years of the study (Table 2; Fig. 5A). The rate of resorption, i.e., the relationship of 0.05 n 0.04- E E_ 0.03 - LL O Q- 0.02 - 0.01 - 0.00 0.05 0.15 0.2 0.25 POF diameter (mm) 0.3 Figure 4 Allometric relationship between postovulatory follicle (POF) diameter and POF cross-sectional area for the Iberian sardine Sardina pilchardus. the slope of the POF cross-sectional area to POF age was not found to differ significantly between the three years, indicating that estimates of resorption rate are not biased either by the preservation medium (AFA or formalin) or by the embedding material (paraffin or resin). In all study years, POFs were shown to shrink exponentially with time in a way that their cross-sec- tional area decreased daily by almost 50% (Fig. 5A). The relationship of the intercept of the POF cross- sectional area on POF age in the GLM was similar for the two preservation mediums (no significant difference between 1997 and 2005), but differed significantly be- tween the two embedding materials (significant differ- ence between 1999 and the other two years) (Table 2). POF area at any given time was significantly higher for resin (Fig. 5A), indicating that processing in paraffin wax leads to a higher shrinkage of all cellular struc- tures in the gonad. This higher rate of shrinkage was also evident by differences in the histological appear- ance of POFs between the two embedding materials, especially at earlier phases of degeneration (Fig. 6). The structure of POFs in resin was more compact and the cellular organization of the granulosa was clearly vis- ible (Fig. 6A). On the other hand, in paraffin sections, the cells of the granulosa layer were hardly detectable and the follicular folds were usually detached from the surrounding theca (Fig. 6B). During the entire survey period, sea surface tempera- ture ranged between 11.6° and 19.3°C, and there were marked interannual differences (2005 being the coldest [mean SST: 14°C ±2.1] and 1997 being the warmest year [mean SST: 16.2°C ±2.2°]). The fitted GLM showed that ambient temperature had a significant effect on the rate of POF degradation (Table 2; Fig. 5B). However, this effect appeared to be limited because an increase of 1°C in ambient temperature accelerated the rate of POF resorption by only 3% (compared to the reduction of POF area by almost 50% per day since spawning; Table 2). This finding indicates that the maximum dif- Ganias et al : Degeneration of postovulartory follicles in Sordino pilchardus 135 Table 1 Summary of dimensional (shape, cross -sectional area) and fine histological (state of the granulosa layer) characteristics of differ- ent daily classes of postovulatory follicles (POFs) in the two embedding materials (P=paraffin; R=resin) for the Iberian sardine | Sardina pilchardus. Daily POF class Shape State of granulosa Cross-sectional area (mm^ ) <1 Irregular Thick and looped 0.0164 ±0.0004 (?) 0.0355 ±0.0001 (R) 1-2 Rectangular One well-formulated layer 0.0093 ±0.0002 ( P) 0.0176 ±0.0005 (R) 2-3 Triangular A thin receding layer 0.0059 ±0.0002 (P) 0.0126 ±0.0003 (R) >3 Triangular Resorption almost completed, some residual vacuoles 0.0042 ±0.0002 (P) 0.0072 ±0.0004 (R) ference in POF duration for the Iberian sardine across the study years could never exceed 0.5 days, thus re- ducing the concerns in relation to the potential bias introduced by varying ambient temperatures in the estimation of the spawning fraction. Discussion The incorrect attribution of age to POFs and the effect of temperature on POF resorption rate constitute major sources of potential bias in the determination of daily spawning classes in routine applications of the DEPM (Stratoudakis et al., 20061. Our method was devised as an indirect way to test the above issues together with the effect of laboratory processing, i.e., type of preserva- tive and embedding material. The method is based on the assumption that in species with diel spawning synchronicity, such as in S. pilchardus, daily classes of spawning fish have, at any given time, POFs of similar size. The first task in our analysis was to refine existing staging criteria for the POFs of Sardina pilchardus. This task was mainly attempted through analyzing the frequency distribu- tion of POF cross-sectional areas in females caught simultaneously and by inspecting POFs in each size-age mode for differences in the cytomor- phological characteristics. The main cytomorphological changes during the degeneration of POFs in sardine ovaries are the gradual deterioration of the structure of the follicular layers and alterations in the dimensional characteristics (shape and size), ac- companied by a decrease in their overall number in the slide. Histo- 0.04 0,03 u. 0,02- O Q. Table 2 Summary statistics of the general inear model fitted to the postovulatory follicle cross-sectional area for the sar- | dine Sardina pilchardus. All parameter estimates were tabulated at the scale of the linear predictor (the inter- | cept for resin as an increment). SE = = standard error. Variable Estimate SE ^value P Age (hour) -0.021 0.0004 -46.78 <0.001 Temp ( ' C ) -0.030 0.012 -2.44 0.016 Intercept -3.344 0.197 -16.98 <0.001 (paraffin) Intercept 0.710 0.027 25.90 <0.001 (resin) 01 - 0,04 0.02 ff 0-00- ^ -0,02- -0,04 20 40 60 POF age (fi) 14 15 16 17 18 Sea temperature (°C) Figure 5 (A) Degeneration of the postovulatory follicle (POF) cross-sectional area (POF area) with time elapsed from spawning (POF age) for the Iberian sardine Sardina pilchardus; 0=resin ; • = paraffin. (B) Effect of ambient temperature on the sardine POF cross-sectional area. 136 Fishery Bulletin 105(1) Figure 6 Microphotographs of very early-stage postovulatory follicles from the female Iberian sardine Sardina pilchardus, captured during the daily spawning period. Embedding materials used in the experiment were (A) resin; (B) paraffin. logical changes in the granulosa seem to follow the general pattern of degeneration described for other populations of sardine (Japanese sardine [Sardinops melanostictus]: Murayama et al., 1994; Mediterranean sardine [S. pilchardus]: Ganias et al., 2003; Pacific sardine [Sardinops sagax]: Goldberg et al., 1984; South African sardine [S. sagax]: Akkers et al., 1996) and oth- er fish species (see interspecific comparison in Hunter and Macewicz, 1985). The information provided in our study mostly concerns the changes in the dimensional characteristics of POFs and how these may be used in the aging of these POFs. The evolution in the shape of POFs is allometric be- cause the POF surface along the lamellar epithelium decreases at a lower rate than the overall area of the follicles. As a result, throughout degeneration, POFs passed consecutively from an irregular to a semirectan- gular and finally a triangular shape, providing a useful additional morphological criterion for determining the stage of the POF. POFs remain for the whole of their "life" on the epithelium of the lamellae, where they oc- cupy approximately the space of an oocyte at the yolk vesicle stage (early POFs) or of a primary oocyte (late POFs). For an indeterminate spawner like sardine, where recruitment of new spawning batches of oocytes occurs continuously and directly from the oogonia, fast resorption is necessary because the aggregation of old POFs would restrict the space available for the devel- opment of new oocytes. On the other hand, late atretic pre-ovulatory follicles separate from the epithelium and concentrate medially in the lamellae — a pattern that has also been observed in other fish species such as striped mullet (Mugil cephalus) (McDonough et al., 2005). Late atretic follicles remain in fish ovaries for long periods, which might extend up to the next spawn- ing season (Hunter and Lo, 1997; Miranda et al., 1999). Therefore, their separation from the lamellar epithelium possibly constitutes a mechanism for managing space availability for the newly recruited spawning batches. The aforementioned morphological phases and the different histological characteristics of the granulosa layer were used, together with follicle size, in the aging of POFs. Four daily classes were identified, implying that full POF resorption in the Iberian sardine ex- ceeded 72 hours. The duration of POF degeneration is variable among fish species, ranging from less than 1 day in the skipjack tuna {Katsuwonus pelainis) (Hunter et al., 1986) to more than 7 days in the piau-jejo (L. taeniatus) (Santos et al., 2005). However, there are reports of intraspecific variability in the duration of POF resorption, both under laboratory conditions (e.g., Atlantic menhaden, Brevoortia tyrannus; Fitzhugh and Hettler, 1995) and in the field (e.g., spotted seatrout [Cynoscion nebulosus]: Roumillat and Brouwer, 2004). In some cases the duration of POF resorption is under- estimated because late POFs can be confused with late atretic stages (e.g., northern anchovy [Engraulis mor- dax] Hunter and Macewicz, 1980). However, given that surveys for refining the DEPM are undertaken at peak spawning months (Stratoudakis et al., 2006), POFs would so greatly outnumber atretic follicles in fish ova- ries that there would be little confusion in distinguish- ing the two and thus would lead to a very minor bias in the duration of POF resorption. In addition, POFs in S. pilchardus are effectively distinguished from all types of atresia, and this distinction has been confirmed by the high degree of consistency in the scoring of post- ovulatory and atretic follicles by different observers. At Ganias et al,: Degeneration of postovulartory follicles in Sordino pikhordus 137 Figure 7 Consecutive stages of oocyte atresia in the Iberian sardine Sardina pilchardiis from very new (A), to intermediate (B), to very late (C, D). T=theca layer; G = granulosa layer; arrow indicates the aggregation of very late atretic oocytes in the central area of an ovarian lamella. Scale bar: 0.1 mm. all stages, atretic follicles constitute enclosed cellular structures (Fig. 7), whereas POFs always maintain an opening towards the ovarian lumen (Fig. 3). In addition, as previously reported, late atretic oocytes separate from the epithelium of the lamellae (Fig. 7D), whereas POFs remain on the epithelium until full resorption (Fig. 3). Sardine POFs shrank exponentially with time, reduc- ing to almost half their size daily. Therefore, after the exponential relationship of POF resorption with time is estimated, ages may be estimated by applying the data for the cross-sectional areas of POFs and times of capture to the model. The measurement of POF cross- sectional areas is not very time consuming and could be merged into the routine of histological analysis for DEPM analyses. Finally, the attribution of ages may be finely tuned by inspecting histomorphological char- acteristics, especially from specimens that lie outside the fitting curve. Before applying the above aging procedure to fish populations, attention should be drawn to other fac- tors that may affect POF cross-sectional area, such as the laboratory treatment of ovaries and the envi- ronmental conditions at sampling, e.g., water tempera- ture. In our study, the use of different preservation media (AFA solution and formalin) did not appear to affect the size of the POFs. On the other hand, the embedding material significantly affected the follicle's size, for POFs in resin were almost double those in paraffin. Resin is known to maintain cellu- lar organization in tissues slightly affected, whereas paraffin causes significant shrinkage (Casotti, 2001; Dorph-Petersen et al., 2001). Besides size, the mor- phological characteristics of the follicles also differed between the two materials. Although the whole POF structure remained compact in resin, follicular folds in paraffin had shrunk and were detached from the thecal layer, even in young POFs. These differences, in combination to the thinner slices that are achieved with resin, make resin a much better material for de- tailed histological observations. However, for accuracy in the staging of POFs, both resin and paraffin seem to provide similar results. Water temperature may affect the rate of POF resorp- tion in teleosts both in the field (Ganias et al., 2003; Roumillat and Brouwer, 2004) and laboratory studies (Fitzhugh and Hettler, 1995). However, this effect has never been precisely quantified and thus temperature 138 Fishery Bulletin 105(1) could never be introduced as an auxiliary factor in the aging of POFs. In the present study, temperature had a significant effect on the rate of POF resorption. This effect appeared to be limited because an increase of 1°C in ambient temperature was estimated to accelerate the rate of POF resorption by almost 3%. Nevertheless, the maximum range of 4-5°C that can be observed during sardine DEPM surveys off the Iberian Peninsula cor- responds to a 12-15% maximum difference in the rate of resorption and thus to a maximum of 8 hours lag in the degree of POF degeneration. The results of this analysis imply that in each DEPM survey, temperature differences between the subareas of the survey area are not expected to introduce serious bias in the cor- rect classification of POFs and to subsequently affect estimates of their ages. Finally, given that the late daily classes of POFs are usually excluded from the estimation of spawning fraction, the maximum effect between the most extreme temperatures is even less important (<10%). Tests, such as described in our study, should be per- formed at least once for each species or population to as- sess bias in the criteria used to determine POF stages and ages. Moreover, the test would provide compara- tive information on POF resorption rates, the impact of embedding material, and the effects of temperature and other environmental parameters on the estimates. In routine DEPM analyses, the measurement of POF cross-sectional areas could increase technical work, but not necessarily the precision in the estimates of the spawning fraction because females are again broken down into spawning nights as they were with the his- tological staging method. However, in cases where such relationships of POF age on POF size are already avail- able, correspondence of POF sizes to spawning nights would be much more realistic than simple histological staging, which strongly depends on the experience of the observer and the quality of the slides. Acknowledgments This work was supported by the program PELAGICOS (PLE/13/00) funded by the Portuguese Ministry of Sci- ence and the National Sampling Plan for DEPM sur- veys funded by the European Union. The contribution of K. Ganias was funded by a postdoctoral scholarship in Portugal (FCT-BPD/17488/2004). We thank four anonymous reviewers for helpful recommendations. In particular, we thank all IPIMAR (Instituto de Inves- tigafao das Pescas e do Mar) staff that contributed to the collection of DEPM samples and the histological preparations. Literature cited Akkers, T. R., Y. C. Melo, and W. Veith. 1996. Gonad development and spawning frequency of the South African pilchard Sardinops sagax during the 1993-1994 spawning season. S. Afr. J. Mar. Sci. 17:183-193. Bernal, M., D. Borchers, L. Valdes, A. Lago de Lanzos, and S. T. Buckland. 2001. A new aging method for eggs of fish species with daily spawning synchronicity. Can. J. Fish. Aquat. Sci. 58:2330-2340. Casotti, G. 2001. Effects of season on kidney morphology in house sparrows. J. Exp. Biol. 204:1201-1206. Dorph-Petersen, K., J. Nyengaard, and H. Gundersen. 2001. Tissue shrinkage and unbiased stereological estimation of particle number and size. J. Microsc. 204:232-246. Fitzhugh, R. G., and F. W. Hettler. 1995. Temperature influence on postovulatory fol- licle degeneration in Atlantic menhaden, Brevoortia tyrannus. Fish. Bull. 93:568-572. Ganias, K., S. Somarakis, A. Machias, and A. Theodorou. 2003. Evaluation of spawning frequency in a Medi- terranean sardine population (Sardina pilchardus sardina). Mar. Biol. 142:1169-1179. Goldberg, S. R., V. Alarcon, and J. Alheit. 1984. Postovulatory follicle histology of the pacific sardine, Sardinops sagax, from Peru. Fish. Bull. 82: 443-445. Hunter, J. R., and S. R. Goldberg. 1980. Spawning incidence and batch fecundity in north- ern anchovy, Engraulis mordax. Fish. Bull. 77:641- 652. Hunter, J. R., and N. C. H. Lo. 1997. The daily egg production method of biomass estimation: some problems and potential improve- ments. Ozeanografika 2:41-69. Hunter, J. R., and B. J. Macewicz. 1980. Sexual maturity, batch fecundity, spawning frequency, and temporal pattern of spawning for the northern anchovy, Engraulis mordax, during the 1979 spawning season. Calif. Coop. Oceanic Fish. Invest. Rep. 21:139-149. 1985. Measurement of spawning frequency in multiple spawning fishes. In an egg production method for esti- mating spawning biomass of pelagic fish: application to the northern anchovy, Engraulis mordax (R. Lasker, ed.), p. 79-93. NOAA Tech. Rep. NMFS 36. Hunter, J. R., B. J. Macewicz, and H. R. Sibert. 1986. The spawning frequency of skipjack tuna, Kat- suwonus pelamis, from the South Pacific. Fish. Bull. 84:895-903. ICES (International Council for the Exploration of the Sea). 2004. The DEPM estimation of spawning-stock biomass for sardine and anchovy. ICES Coop. Res. Rep., 91p. McDonough, C. J., W. A. Roumillat, and C. A. Wenner 2005. Sexual differentiation and gonad development in striped mullet (Mugil cephalus L.) from South Carolina estuaries. Fish. Bull. 103:601-619. Miranda, A. C. L., N. Bazzoli, E. Rizzo, and Y. Sato. 1999. Ovarian follicular atresia in two teleost species: a histological and ultrastructural study. Tissue and Cell 31:480-488. Murayama, T, M. Shirashi, and I. Aoki. 1994. Changes in ovarian development and plasma levels of sex hormones in the wild female Japanese sardine (Sardinops melanostictus) during the spawn- ing period. J. Fish Biol. 45:235-245. Ganias et al.: Degeneration of postovulartory follicles in Sordino pilchardus 139 Parker, K. 1980. A direct method for estimating northern anchovy, Engraulis mordax, spawning biomass. Fish. Bull. 78:541-544. Perez, N., I. Figueiredo, and B. J. Macewicz. 1992. The spawning frequency of sardine, Sardina pil- chardus (Walb.), off the Atlantic Iberian coast. Bol. Inst. Esp. Oceanogr. 8:175-189. Picquelle, S. J., and G. Stauffer. 1985. Parameter estimation for an egg production method of anchovy biomass assessment. In An egg production method for estimating spawning biomass of pelagic fish: application to the northern anchovy, Engraulis mordax (R. Lasker, ed.), p. 7-16. NOAA Tech. Rep. NMFS 36. Roumillat, W. A., and M. C. Brouwer 2004. Reproductive dynamics of female spotted seatrout (Cynoscion nebulosus) in South Carolina. Fish. Bull. 102:473-487. Saidapur, K. S. 1982. Structure and function of postovulatory folli- cles (corpora luteal in the ovaries of nonmammalian vertebrates. Int. Rev. Cytol. 75:243-285. Santos, B., E. Rizzo, N. Bazzoli, Y. Sato, and L. Moro. 2005. Ovarian regression and apoptosis in the South American teleost Leporinus taeniatus Lutken (Characi- formes, Anostomidae) from the Sao Francisco Basin. J. Fish Biol. 67:1446-1459. Stratoudakis, Y., M. Bernal, K. Ganias, and A. Uriarte. 2006. The daily egg production method (DEPM): recent advances, current applications, and future chal- lenges. Fish and Fisheries 7:35-57. Whitehead, P. J. P., Wongratana, T., and G. J. Nelson. 1985. Clupeid fishes of the world (Suborder Clupeoidei): an annotated and illustrated catalogue of the herrings, sardines, pilchards, sprats, shads, anchovies, and wolf- herrings. FAO species catalogue, vol. 7, part 1, 303 p. FAO, Rome. Zwolinski, J., Y. Stratoudakis, and E. Soares. 2001. Intra-annual variation in the batch fecundity of sardine off Portugal. J. Fish Biol. 58:1633-1645. 140 Use of genetic data to assess the uncertainty in stock assessments due to the assumed stock structure: the case of albacore {Thunnus alalungd) from the Atlantic Ocean Haritz Arrizabalaga^ (contact author) Victoria Lopez-Rodas^ Eduardo Costas^ Alberto Gonzalez-Garces^ Email address for H. Arrizabalaga: harri@pas.azti.es ' AZTI Tecnaha Herrera Kaia Portualdea z/g 20110 Pasaia (Gipuzkoa) Spam 2 Universidad Complutense Genetica, Facultad de Vetennaria Avda. Puerta de Hierro s/n 28080 Madrid. Spain 3 lEO Cabo Estay, Canido Apdo 1552 Vigo (Pontevedra), Spain Stock assessments can be problematic because of uncertainties associated with the data or because of simpli- fied assumptions made when model- ing biological processes (Rosenberg and Restrepo, 1995). For example, the common assumption in stock assessments that stocks are homo- geneous and discrete (i.e., there is no migration between the stocks) is not necessarily true (Kell et al., 2004a, 2004b). On the other hand, it is essential that the stock structure assumed during the assessment and manage- ment process corresponds to the real population structure of the resource. Otherwise, fishery management be- comes inefficient (less productive populations may be overfished and collapse, while more productive popu- lations may be underexploited [Allen- dorf et al., 1987; Begg et al., 1999]) and may affect biological attributes, such as growth, productivity, or ge- netic diversity (Ricker, 1981). In spite of this problem, current regulations on several fisheries are based on spa- tial schemes that do not necessarily reflect the real biological structure of the populations (Pawson and Jen- nings, 1996; Stephenson, 1999; Ward, 2000). In these cases, the results of stock assessments may be biased and, in general, an important level of uncertainty exists in stock assess- ments (NEC, 1994; Turner, 1998) due to the assumed stock structure. An assessment of the magnitude of this uncertainty is important so as to increase confidence in the as- sessment itself. Moreover, quantify- ing the uncertainty allows the evalu- ation of the relative effect of stock structure assumptions with respect to other assumptions about biologi- cal, fishery, or modeling parameters in the assessment. Knowing the rela- tive importance of the effect of these underlying assumptions will allow management scientists to prioritize the types of research needed to bet- ter ground the stock assessments with real information. In this note, we suggest a way to assess uncertainty in stock assess- ments that is due to assumptions of stock structure. The assessment is essentially based on a sensitivity analysis conducted by testing alter- native stock structure hypotheses derived from available genetic, fish- ery, and biological information. The method is illustrated with albacore (Thunnus alalunga, Bonn. 1788) in the Atlantic Ocean. Albacore is a highly migratory spe- cies distributed between latitudes 45°N and 45°S. Studies of albacore reproduction in the Atlantic Ocean have shown different spawning pe- riods and areas in both hemispheres (Beardsley, 1969; Koto, 1969). Shio- hama (1971) and Uozumi (1996), based on Japanese longline distribu- tion studies, described an adult con- centration area in each hemisphere. These findings, along with studies of larval concentration areas (Ueyanagi, 1971), support the existence of two separate populations, one in each hemisphere. Based on these studies, it is assumed within the Interna- tional Commission for the Conserva- tion of Atlantic Tunas (ICCAT) that there are two albacore management units in the Atlantic, separated by parallel 5°N. However, various au- thors have suggested the possibil- ity that albacore move between the north and south Atlantic (reviewed in Gonzalez-Garces, 1997). Moreover, the continuous spatial distribution of catches around the equator also sug- gests this possibility (Fig. 1). Recent studies have shown ge- netic differences between north and south Atlantic albacore (Takagi et al., 2001; Arrizabalaga et al., 2004), but it is still unclear whether the limit between both populations is at latitude 5°N or somewhere else. In fact, results from Arrizabalaga et al. (2004) are not concordant with the limit at latitude 5°N because a sample from the Gulf of Guinea (1°N, 15-16°W) was genetically more like the sample from the north Atlantic than the one from the south Atlan- tic. This observation may indicate that either the limit between both stocks may be located farther south than that currently assumed or that Manuscript submitted 15 June 2005 to the Scientific Editor's Office. Manuscript approved for publication 14 December 2005 by the Scientific Editor. Fish. 105:140-146(2007). NOTE Arrizabalaga et al : Uncertainity In slock assessments of Thunnus alalunga due to assumed stock structure 141 there may be some interchange be- tween individuals of both stocks. An earlier statistical comparison of blood group frequencies in albacore found in the Gulf of Guinea (lat. 0°-9°S, long. 0°-8°W), northwest Atlantic (lat. 23°-31°N, long. 60°-70°W) and middle-north Atlantic (lat. 1-34°N, long. 11°-40°W) in an earlier study (Suzuki, 1962) did not show differ- ences between them, again indicating that the fish present in the Gulf of Guinea may belong to the northern population. Materials and methods Taking into account the above find- ings, we assessed the uncertainty in north and south Atlantic albacore stock assessments by means of a sen- sitivity analysis. This analysis con- sisted in assessing both stocks, either under alternative stock boundaries or by assuming certain migration rates between them. Stock assessment under the assumption of alternative boundaries between stocks 100 w 70 W 40 W low 20 E 60 N 30 N ON 30 S 60S 60 N 30 N ON 30 S 60S 100 W 70 W 40 W low 20 E Figure 1 Spatial distribution of Atlantic and Mediterranean albacore {Thun- nus alalunga) catches (ICCAT, 2001). The size of the circles is propor- tional to the square root of the total catch at each location. Two alternative boundaries between albacore stocks were considered: at lat. 0°N and lat. 5°S. The catch-at-age within lat. 5°N-0°N and lat. 5°N-5°S was removed from the southern catch-at-age matrix and added to the northern one, by using available catch (ICCAT'), size, and growth information (Bard, 1981; Sarralde et al., 2002). For each boundary, abundance and fishing mortality rates were estimated separately for each stock by virtual population analysis (VPA) by using the VPA-2box, vers. 3.0 program (Porch et al., 2001). This program assesses the abundance and mortality of one or two intermixing stocks by fitting age-structured population equations to fishery data. All stock assess- ment options were maintained as in the ICCAT 2001 report (ICCAT-) and variance of estimated parameters was computed by performing 400 nonparametric boot- straps of the abundance indices. ' ICCAT (International Commission for the Conservation of Atlantic Tunas). Website: http;//www.iccat.int/ (accessed 31 June 2005). 2 ICCAT (International Commission for the Conservation of Atlantic Tunas). 2001. Report of the ICCAT SCRS albacore stock assessment session (Madrid, Spain; Octo- ber 9 to 1.5, 2000). Collect. Vol, Sci. Pap. ICCAT, 52, p. 1283-1390. International Commission for the Conserva- tion of Atlantic Tunas, Corazon de Maria 8, 28002 Madrid, Spain. Stock assessment under the assumption that there is migration between stocks Blood group frequency data in Arrizabalaga et al. (2004) were used, with the assumption that the sample in the Gulf of Guinea was within the currently accepted range limit for both stocks, and therefore that it may be made up of a mixture of albacore belonging to the northern and southern populations. The proportion that each pop- ulation would contribute to the mixture was calculated according to Cavalli-Sforza and Bodmer (1981) as x,„ = mx^ +(l-m)xg (1) where m = the fraction of population A in the mixture; x,,, = the allelic frequency in the mixture; and Xj^ and Xg = the allelic frequencies in populations A and B, respectively. The variance of m is given by 9 1 f 9 2'>/i \''2'I (x^-Xb) (2) where, a^,„, o^^ and aj,^ are the variances of the allelic frequencies in the mixture and populations A and B, 142 Fishery Bulletin 105(1) respectively. When several diallelic loci are analyzed, the fraction of population A in the mixture can be computed as a weighted average: m (3) Table 1 Proportion of albacore (Thunniis alalunga). with a given blood group in the North Atlantic, Gulf of Guinea, and South Atlantic determined with three different lectins (Con A: Concanavaline A\ WGA: Triticum viilgare: EGA: Eritrina cristagally; from Arrizabalaga et al., 2004) and estimated mixing proportions in the Gulf of Guinea. OT=proportion of northern origin fish in the Gulf of Guinea sample; m=weighted average proportion. Standard deviation of m and m is given in parentheses. Lectin Con A WGA EGA North Atlantic Gulf of Guinea South Atlantic Using three lectins, for which the positive lectin binding proportion in the Gulf of Guinea was intermedi- ate between the proportions for the northern and southern populations as described in Arrizabalaga et al. (2004), we obtained /n values shown in Table 1. The weighted mean pro- portion indicated that 79% of the fish present in the Gulf of Guinea would belong to the northern Atlantic population, and this result was used to formulate plausible migration hypotheses for the two stocks. The mean historical (1975-99) catch around the equator (between lat. 5°N and lat. 5°S) has been 1218 metric tons (t) per year, and therefore 974 t would belong to fish from the North Atlantic population and 244 t to fish from the South Atlantic population. In reference to the average total catch in each stock (38,960 t and 27,111 t in the northern and southern stocks, respec- tively), these quantities would imply that about 2.5% of the fish from the North Atlantic and 0.9% from the South Atlantic are present in the Gulf of Guinea every year. Assuming that albacore in this area are migrat- ing from one stock to the other, these percentages would, in broad terms, represent the yearly transfer rates between stocks. Several scenarios were established and tested. Sce- nario number 1 reflects the above situation (2.5% and 0.9% annual migration rates from north to south and south to north, respectively). However, high variances for mixing proportions were obtained because no diag- nostic loci was detected, and those precision estimates could, in fact, be overestimated because not all fish were sampled from different schools in the study of Arrizabalaga et al. (2004). This overestimation may indicate that annual migration rates vary considerably from those in scenario 1. Thus, a range of alterna- tive migration scenarios were explored in which ad- ditional biological or fishery aspects were taken into account. Scenarios 2, 3, 7, and 8, reflected the situ- ation in which migration occurs only in one direction (5% yearly from north to south and south to north in scenarios 2 and 3, respectively, and 10% from north to south and south to north in scenarios 7 and 8, respec- tively). Because no fishing effort targeting albacore exists in the equatorial area, the real migration rate may be higher than the one inferred from catches in that area. Accordingly, in scenario 4, twice the migra- tion rates of scenario 1 (5% from north to south and 0.2500 0.2174 0.0357 0,8478(0.5553) 0.4500 0.3913 0.2857 0,6427(0,7794) 0,7900(0.4232) 0.0500 0.0435 0,8695(1,2006) 1.8%. from south to north) were adopted. In scenario 5, migration was considered to be limited to the adult fraction of the stock (ages 5-8-I-), as size distributions in this area indicated, and finally in scenarios 6 and 9, high rates of migration (5% and 10%, respectively) in both directions were chosen. Although these scenarios are believed to be representative of the true nature of mixing between the stocks, it should be stressed that they represent only some of many different possible mixing scenarios. All scenarios were tested by assuming an overlap migration model (fish return back to the area of origin for spawning) and using the VPA-2box program (Porch et al., 2001). No diffusive migration was considered because it is not consistent with observed genetic dif- ferentiation. Results for all scenarios were compared (in terms of spawning stock biomass trends and the small sample bias-adjusted version of the Akaike information criteria ([AICc, Hurvich and Tsai, 1995]) with the base case where no migration was assumed to occur between stocks. Results Stock assessment under the assumption of alternative boundaries between stocks Best fits for northern and southern stocks were obtained by assuming different stock boundaries, at lat. 5°S and lat. 5°N, respectively. However, estimated abundance and fishing mortalities, with the assump- tion of any of the alternative stock limits, showed minor differences with respect to the base case (Table 2). The effect of considering the limit in lat. 0°N or in lat. 5°S was practically the same because most of the catch in the equatorial area happens in the Northern Hemisphere (between lat. 5°N and lat. 0°N). All coef- ficients of variation (CV) were below 15%, except for the ■F'5+**^'^® in the south Atlantic, which were between 15% and 30%. NOTE Arrizabalaga et a\: Uncertainlty In stock assessments of Thunnus alalunga due to assumed stock structure 143 Table 2 Results of model fits for alternative boundaries between stocks of albacore (Thunnus albacore). Instantaneous fishing mor- tality (F) and abundance iN. in millions of individuals! estimates are averaged by age groups (subscripts) and time peri- ods (superscripts). Corresponding mean coefficients of variation are given within parentheses, n = number of data points; p = number of estimated parameters; AICc = adjusted Akaike information criteria iHurvich and Tsai, 1995). North Atlantic albacore South Atlantic albacore Limit 5°N = N 5^8 5°N 0°N 5=S -logL 66.68 66.74 66.52 80.85 81.98 82.20 Deviance 117.04 117.01 116.92 61.95 61.97 62.02 n 117.00 117.00 117.00 62.00 62.00 62.00 P 14.00 14.00 14.00 12.00 12.00 12.00 AICc 149.15 149.12 149.04 92.31 92.34 92.39 p 75-86 0.11(0.2%) 0.11(0.23%) 0.11(0.19%) 0.00(0.75%) 0.00 (0.86%) 0.00(0.92%) P 87-96 0.16(3.87%) 0.16(4.27%) 0.16(3.97%) 0.01(6.84%) 0.01 (6.87%) 0.01 (6.63%) P 75-86 •'^2-4 0.39 (0.19%) 0.37 (0.22%) 0.37(0.18%) 0.11 (0.8%) 0.11 (1.02%) 0.11 (1.12%) P 87-99 ■'^2-4 0.44 (10.44%) 0.41 (10.91%) 0.4(9.9%) 0.2 (8.44%) 0.2(8.95%) 0.21 (8.46%) p 75-86 0.3(0.21%) 0.3 (0.24%) 0.3(0.19%) 0.16(0.86%) 0.15(1.12%) 0.15(1.24%) p 87-99 5+ 0.2(11.48%) 0.21(13.97%) 0.22(11.17%) 0.3(18.46%) 0.26(29.39%) 0.25(22.42%) ^^75-86 10.37 (0.19%) 10.6(0.22%) 10.67 (0.18%) 8.23(0.87%) 8.17(1.05%) 8.16(1.14%) Ar^87-96 8.76 (3.94%) 9.02(4.33%) 9.12(4.16%) 7.55(8.6%) 7.37 (8.52%) 7.3 (9%) \T 75-86 ■''2-4 12.51(0.15%) 12.87(0.18%) 12.97(0.15%) 13.64(0.78%) 13.68(0.99%) 13.69(1.09%) fa 87-99 "2-4 9.07 (7.97%) 9.47(8.5%) 9.62(7.99%) 11.96(7.97%) 11.56(8.14%) 11.61 (8.33%) NJ5-SS 3.29 (0.17%) 3.35 (0.2%) 3.36(0.16%) 4.31(0.83%) 4.37(1.07%) 4.41(1.19%) N^Jl-99 1.65 (7.87%) 1.81(8.4%) 1.85(7.67%) 3.57(10.04%) 3.57(10.19%) 3.59(10.12%) Stock assessment under the assumption of migration between stocks Fits under several migration scenarios were more par- simonious than under the assumption of no migra- tion between stocks (Table 3). Scenarios 1 (p^g = 0.025, Ps^=0.009), 2 (pi,,s = 0.05.psN=0l 3 (pvs = 0' Psiv=0-05), 4 (/)^,s = 0.05, p>,.^=0.018), 6 (p^s = 0.05, Ps^=0.05), and 7 (p,^g=0.1, pg^=0) showed lower AICc values than in the base case (BO. Spawning stock biomass (SSB) values and trends under scenarios 1, 2, 3, 4, 5, 6, and 7 were similar to the ones observed in the base case, especially in the sec- ond half of the study period (except for the SSB of the southern stock over the last two years, which showed more variability. Fig. 2). In contrast, scenarios 8 and 9 showed a very different pattern in the last half of the series. In the north Atlantic, after the decline of the SSB during the beginning of the 1980s, the recovery was much more effective under these two scenarios, reaching higher values at the end of the 1990s than in the 1970s. Meanwhile, the SSB values for the southern stock were only slightly lower than those under the assumption of no migration, and under scenario 9 they were higher than in the base case over the last two years (Fig. 2). Table 3 Considered overlap migration scenarios and associated | results of model fits for albacore (Thunnus a lalunga). n = 179 ;p = 25; AICc = adjusted Akaike information cri- 1 teri a (Hurvich and Tsai, 1995); BC = base case; p„,, = 1 annual mi gration rate from north to south; pt;;^, = annual migration rate from south to north. See nario Pns PsN Age range (yr) AICc BC 1-8-1- 18.76 1 0.025 0.009 1-8-I- 13.74 2 0.05 1-8-1- 15.12 3 0.05 1-8+ 18.03 4 0.05 0.018 1-8+ 13.10 5 0.05 0.018 5-8+ 19.41 6 0.05 0.05 1-8+ 17.67 7 0.1 1-8+ 12.77 8 0.1 1-8+ 37.58 9 0.1 0.1 1-8+ 46.81 Discussion The method used to assess the uncertainty in the stock assessments that is due to the assumed stock structure 144 Fishery Bulletin 105(1) 120,000-, North Atlantic 100,000- f\ 80,000- 1 60,000- ^^\. r^ 40,000- ^\ >^V uJ 20,000- \^ Xi4^ CD LDr^Oi-'-nLnr--CT)t-coini^o> h-h-h-cocoaaoooocno^CTjOcn CT>010i01CDO>CBCT)Cr)CJ)01010 in 120,000t South Atlantic 100.000- 80,000- 60,000- A 40,000- i 20,000- 0- h-N-h-COCOCOCOCX305Cr>CT>CT)0^ BC ° 1 Figure 2 Historical evolution of north and south Atlantic albacore iThuunus alalunga) spawning stock biomass (SSB) estimated under different overlapping migration scenarios (see Table 3 for details). Vertical bars indicate 95% confidence intervals for the spawning stock biomass in the base case, BC = base case. is based on a simple sensitivity analysis conducted by testing alternative stock structure hypotheses. When plausible hypotheses are generated from the cumulated knowledge on the biology and fisheries of the species, the effect of these hypotheses on the assessment results can be studied. However, it is also important to consider a range of alternative hypotheses so that they produce significantly different results in the assessment, in refer- ence to the base case. The wide range of results allows us to discern the level of migration that is of concern for assessment purposes and shows the level of migra- tion that is not likely to be realistic, given the available catch and effort data. However, several other sources of uncertainty should be taken into account at this stage. The analysis could be extended to future biomass pro- jections under different management strategies in order to indicate those management strategies that are more robust to violation of stock structure assumptions. This kind of study should run parallel to other studies where additional sources of uncertainty (e.g., in biological parameters or fishery data) are quantified because this approach would show their relative importance so that research could be prioritized with the goal of providing improved information on stock status. In the case of Atlantic albacore, recent genetic stud- ies indicate that, either the limit between both popula- NOTE Arrizabalaga et al.: Uncertainity in stock assessments of Thunnus alalunga due to assumed stock structure 145 tions is not at lat. 5°N, but farther south, or that some amount of migration exists between them. Because no diagnostic loci were found by Arrizabalaga et al. (2004), the estimated proportions from each stock in the Gulf of Guinea sample could not be precisely determined and the true nature of mixing between north and south At- lantic albacore has yet to be fully determined. In spite of this uncertainty, the present exercise has made it possible to explore the response of biomass trends in different plausible discrete-stock scenarios and stock- mixing scenarios. Although it is not possible to determine where the real line lies between populations, we can conclude from our knowledge about the current low level of reported equitorial catch and the size structure of this catch and if we assume a limit of latitude 0° north or latitude 5°S that our current perceptions of the stock structure of this species are probably accurate. On the other hand, several migration scenarios fitted the observed catch-at- age and abundance indices better than the scenario of no migration. In all these scenarios, SSB trends were very similar and values did not differ significantly from the ones in the base case; therefore it can be concluded that, although some rate of migration between stocks likely exists, the perception that we have about stock status, assuming there is no migration, is quite realistic. In other words, uncertainty in northern and southern Atlantic albacore stock assessments associated with the assumed stock structure does not seem to be important, given current biological knowledge and fishery data. The highest variations in SSB were observed for northern Atlantic albacore in scenarios with high mi- gration rates from south to north, showing high levels of SSB at the end of the study period. However, the ob- served difference in SSB levels with respect to the base case in the South Atlantic was not that pronounced, showing that northern Atlantic albacore biomass is more sensitive to biomass input from the south than vice versa. This result occurs because the minimum level of SSB in the north Atlantic in 1987 coincides in time with the maximum SSB in the southern stock, which is an order of magnitude higher. In this case, a migration rate of 10% from south to north would imply the input of approximately half the biomass present in the north at that moment, leading to a more rapid re- covery of historic levels than under the null migration assumption. However, the existence of such important migration rates from south to north seems unlikely given the observed catch-at-age and abundance indices for both stocks. The present analysis allows for the increase in con- fidence levels regarding stock assessment results for northern and southern Atlantic albacore obtained with- in ICCAT, assuming that stocks are separated at lat. 5°N and that there is no migration between them. This information is essential in order that the catch- and effort-related management measures that are in force for Atlantic albacore remain effective. Nevertheless, ad- ditional hypotheses, such as migration between North Atlantic and Mediterranean albacore, or between South Atlantic and Indian Ocean albacore, should be inves- tigated further as future research findings are made available. Moreover, it should be noted that migration between stocks could vary among years, and a yearly based assessment of genetic mixture, based on DNA analysis, would be more useful for quantitative stock assessments. Acknowledgments We are grateful to Clay Porch for his help in the use of VPA-2box and statistical considerations regarding the interpretation of the results. Mauricio Ortiz, Hilario Murua. Dorleta Garcia, two anonymous referees, and the scientific editor also made helpful suggestions. We are also grateful to the Spanish "Comision de Ciencia y Tecnologia" (CICYT) and the "Departamento de Agricul- tura y Pesca" of the Basque Government for the funding of the MAR98-0233 and RP2002023 research projects, respectively. Literature cited Allendorf, F. W., N. Ryman, and F. M. Utter. 1987. Genetics and fishery management: past, present and future. In Population genetics and fishery manage- ment (N. Ryman and F. M. Utter, eds.), p. 1-19. Univ. Washington Press, Seattle and London. Arrizabalaga, H., E. Costas, J. Juste, A. Gonzalez-Garces, B. Nieto, and V. Lopez-Rodas. 2004. Population structure of albacore Thunnus ala- lunga inferred from blood groups and tag-recapture analyses. Mar. Ecol. Prog. Ser. 282:245-252. Bard, F. X. 1981. Atlantic albacore tuna {Thunnus alalunga Bonna- terre 17881. From population dynamics to demographic strategy. Ph.D. diss, 333 p. [In French.] Universite Pierre et Marie Curie, Paris, France. Beardsley, G. L. 1969. Proposed migrations for albacore, Thunnus ala- lunga, in the Atlantic Ocean. Trans. Am. Fish. Soc. 98:589-598. Begg, G. A., K. D. Friedland, and J. B. Pearce. 1999. Stock identification and its role in stock assess- ment and fisheries management: an overview. Fish. Res. 43:1-8. Cavalli-Sforza, L. L., and W. F, Bodmer. 1981. The genetics of human populations, 942 p. Free- man and Com., San Francisco, CA. Gonzalez-Garces, A. 1997. Contribution to knowledge about north Atlantic albacore iThunnus alalunga, Bonn. 1788) population dynamics. Ph.D. diss., 200 p. [In Spanish.] Univer- sidad Complutense de Madrid, Ciudad Universitaria, Madrid, Spain. Hurvich, C. M., and C. L. Tsai. 1995. Relative rates of convergence for efficient model selection criteria in linear regression. Biometrika 35:418-425. Kell, L. T, W. W. Crozier, and C. M. Legault. 2004a. Mixed and multi-stock fisheries: Introduc- tion. ICES J. Mar. Sci. 61:1330. 146 Fishery Bulletin 105(1) Kell, L. T., R. Scott, and E. Hunter. 2004b. Implications for current management advice for North Sea plaice: Part I. Migration between the North Sea and English Channel. J. Sea Res. 51:287-299. Koto, T. 1969. Studies on the albacore — XIV. Distribution and movement of the albacore in the Indian and the Atlan- tic Oceans based on the catch statistics of Japanese tuna longline fishery. Bull. Far Seas Fish. Res. Lab.. 1:115-129. NRC (National Research Council). 1994. An assessment of Atlantic bluefin tuna, 166 p. Na- tional Academy Press, Washingon D.C. Pawson, M. G., and S. Jennings. 1996. A critique of methods for stock identification in marine capture fisheries. Fish. Res. 25:203-217. Porch, C. E.. S. C. Turner, and J. E. Powers. 2001. Virtual population analyses of Atlantic bluefin tuna with alternative models of transatlantic migration: 1970-1997. Collect. Vol. Sci. Pap. ICCAT 52:1022-1045. Ricker, W. E. 1981. Changes in the average size and average age of Pacific salmon. Can. J. Fish. Aquat. Sci. 38: 1636- 1656. Rosenberg, A. A., and V. R. Restrepo. 1995. Uncertainty and risk evaluation in stock assess- ment advice for U.S. marine fisheries. Can. J. Fish. Aquat. Sci. 51: 2715-2720. Sarralde, R., F. X. Bard, and A. Herve. 2002. Update of tropical east Atlantic albacore iTIuin- nus alalunga) purse seine bycatch information. |In Spanish,] Collect. Vol. Sci. Pap. ICCAT 55:245-250. Shiohama, T. 1971. Studies on measuring changes in the character of the fishing effort of the tuna longline fishery. Con- centrations of the fishing effort to particular areas and species in the Japanese Atlantic fishery. Bull. Far Seas Fish. Res. Lab. 5:107-130. Stephenson, R. L. 1999. Stock complexity in fisheries management: a per- spective of emerging issues related to population sub- units. Fish. Res. 43:247-249. Suzuki, A. 1962. Serological studies of the races of tuna. VI. Bigeye- 3 antigen occurred in the albacore. Rep. Nankai Reg. Fish. Res. Lab. 16:67-70. Takagi, M., T. Okamura, S. Chow, and N. Taniguchi. 2001. Preliminary study of albacore tThunnus alalunga) stock identification inferred from microsatellite DNA analysis. Fish. Bull. 99:697-701. Turner, S. C. 1998. Stock structure and mixing. Collect. Vol. Sci. Pap. ICCAT 50:17-23. Ueyanagi, S. 1971. Larval distribution of tunas and billfishes in the Atlantic Ocean. FAO Fisheries report 71:297-305. Uozumi, Y. 1996. A historical review of Japanese longline fishery and albacore catch in the Atlanic ocean. Collect. Vol. Sci. Pap. ICCAT 43:261-268. Ward. R. D. 2000. Genetics in fisheries management. Hydrobiologia 420:191-201. 147 New information from fish diets on the importance of glassy flying squid iHyaloteuthis pelagica) (Teuthoidea: Ommastrephidae) In the epipelagic cephalopod community of the tropical Atlantic Ocean Yves Cherel (contact author)^ Richard Sabatie^ Michel Potier^ Francis Marsac^ Frederic Menard* ' Centre d'Etudes Biologlques de Chize UPR 1934 du Centre National de la Recherche Scientifique BP 14 79360 Villiers-en-Bois, France Email address for Y Cherel cherel@cebc.cnrs.fr ^ Pole Halieutique Laboratoire d'Ecologie Halieutique, Agrocampus-Rennes 65 rue de Saint Brieuc 35042 Rennes Cedex, France 3 Institut de Recherche pour le Developpement Centre de La Reunion UR 109 Thetis, BP 172 97492 Sainte Clotilde Cedex, Isle de La Reunion, France ■* Institut de Recherche pour le Developpement Centre de Recherche Halieutique Mediterraneenne et Tropicale UR 109 Thetis, BP 171 34203 Sete Cedex, France Squids of the family Ommastrephi- dae are a vital part of marine food webs and support major fisheries around the world. They are widely distributed in the open ocean, where they are among the most abundant in number and biomass of nektonic epipelagic organisms. In turn, seven of the 11 genera of this family (Dosi- dicus. Illex, Martialia, Nototodariis. Ommastrephes, Sthenoteuthis, and Todarodes) are heavily preyed upon by top marine predators, i.e., birds, mammals, and fish, and currently support fisheries in both neritic and oceanic waters (Roper and Sweeney, 1984; Rodhouse, 1997). Their com- mercial importance has made the large ommastrephids the target of many scientific investigations and their biology is consequently reason- ably well-known (NigmatuUin et al., 2001; Zuyev et al., 2002; Bower and Ichii, 2005). In contrast, much less information is available on the biol- ogy and ecological role of the smaller, unexploited species of ommastrephids (e.g., Eucleoteuthis, Hyaloteuthis, Ornithoteuthis, and Todaropsis). Hyaloteuthis pelagica (Bosc, 1802), the glassy flying squid, is the smallest ommastrephid, reaching a maximum mantle length of 90 mm (Nesis, 1987). It appears to be an epipelagic species that is probably distributed in all tropical and sub- tropical oceans (Nesis and Nigmatul- lin, 1979; Wormuth, 1998). Hyaloteu- this pelagica is rarely captured, but was caught in large numbers during a cruise off Brazil, where it was the dominant ommastrephid captured in nets ( Warneke-Cremer, 1986). Almost nothing is known about its trophic relationships, either as prey or predator (Nesis and NigmatuUin, 1979). Numerous remains of//, pe- lagica — from a few intact squids to a fairly large number of accumulated beaks — were found in the stomachs of large predatory fishes during re- search cruises in the central Atlan- tic Ocean in autumn 2000. In this note, we describe the importance of H. pelagica in fish diets, thus add- ing new information about the abun- dance and trophic role of a poorly known ommastrephid species. Materials and methods Fieldwork was carried out in the central Atlantic Ocean during three cruises of the Japanese RV Shoyo Maru in October-December 2000 (Fig. 1). Cruise I took place in tem- perate waters of the north equato- rial current (between 8-21°N and 42-29°W) and cruises II and III took place in tropical waters of the south equatorial divergence (between 2N-10°S and 13-26°W, and between 7-9°S and 9-24°W). Cruises were a part of the Bigeye Tuna Year Pro- gram (BETYP) that was undertaken under the auspices of the Interna- tional Commission for the Conserva- tion of Atlantic Tunas (ICCAT). The purpose of the cruises was to tag live tunas caught by longlines in order to investigate their migration pattern and behavior in relation to fish aggre- gating devices. Fish were measured (eye-fork length for billfishes and fork length for other species) and dissected onboard. In the laboratory, each fish stomach was thawed, opened, and both accu- mulated (cephalopod beaks with no flesh attached) and fresh items were sorted. Fresh remains were divided into broad prey classes (fish, cepha- lopods, crustaceans, and others), and weighed to calculate their proportion by mass in the diet. Identification of cephalopod prey relied on the exter- nal morphological features of either intact specimens or beaks. Beaks Manuscript submitted 2 March 2006 to the Scientific Editor's Office. Manuscript approved for publication 24 May 2006 by the Scientific Editor. Fish. Bull. 105:147-152 (2007). 148 Fishery Bulletin 105(1) 20° N (both lower and upper) were identified by reference to features given by Clarke (1986) and by comparison with material held in our own reference collection. Well-preserved specimens of H. pelagica were identified from the special arrange- ment of luminous spots on the ventral side of the mantle (Nesis, 1987). Beaks from those specimens identified from the spots (reference beaks) allowed us to identify almost all ommastrephid beaks found in fish samples as belonging to H. pelagica. Importantly, wings of the lower beaks darkened at a small size, thus precluding misidentification with beaks of other om- mastrephid species that darken at larger sizes, e.g., Sthenoteuthis pteropus (Clarke, 1986). Lower rostral length (LRL) of beaks was measured to 0.1 mm with a vernier caliper and the allometric equations given by Clarke (1986) were used to estimate dorsal mantle length (ML) and whole wet mass (M) from LRL. Specimens of H. pe- lagica were assumed to be adults at 46 mm ML and larger (Dunning and Brandt, 1985). Dietary data are presented by using two calculation techniques, namely the frequency of occurrence and the percentage by number of each prey type. Data were sta- tistically analyzed by using SYSTAT 9 (SPSS, Chicago, IL). Values given are means (±SD). Results Most (97%) of the fish caught on longlines belonged to 10 different species of large oceanic predatory fishes, including longnose lancetfish {Alepisaurus ferox), four scombrids (wahoo [Acanthocybium solandri], albacore [Thunnus alalunga], yellowfin tuna [T. albacares], and bigeye tuna [T. obesus]), swordfish (Xiphias gladius), and four istiophorids (sailfish [Istiophorus albicans], blue marlin [Makaira nigricans], and white marlin [Tet- rapturus albidus], and longbill spearfish [T. pfluegeri]). Most of the fish (93%) contained fresh remains in their stomachs. Fish prey dominated the diet by mass (>50%) of eight predator species (Table 1). Fish and cephalopod items were almost equally important in the diet of white marlin, whereas fish, cephalopods, and crustaceans were the main food sources of albacore. Cephalopods amounted to slightly more than 50% of the diet by mass in one fish species (white marlin) only. They were an important prey class (>10%) in five other species and were still a minor, but significant (a5-10%), portion of the food of the four remaining fishes (Table 1). Overall, cephalopods (both fresh and accumulated items) were found in most of the individu- als (76%) and a total of 2701 cephalopod beaks were identified from the stomach of 105 fish. Hyaloteuthis pelagica was by far the most important cephalopod prey of the community of large predatory fishes, amounting • Cruise I A Cruise I Cruise I Figure 1 Locations of longline sets carried out in the central Atlantic Ocean during three cruises of the RV Shoyo Maru between October and December 2000. to more than 50% of the total number of cephalopods (up to 93%) in six species (Table 2). Indeed, it was found to be the main cephalopod prey in all the fishes, except in bigeye tuna and lancetfish where it ranked second and third, respectively. Hyaloteuthis pelagica was much more abundant in the diet of fish caught in tropical wa- ters (cruises II and III) than in the diet of individuals fished in temperate waters (cruise I) (?!=1937 and 15 beaks, respectively). Other important cephalopod prey (>10% by number) included the small onychoteuthid squid Walvisteiithis (= Onykia) rancureli in the diets of bigeye tuna and albacore, and the pelagic octopuses Japetella diaphana and common blanket octopus Tremoctopus violaceus in those of yellowfin and bigeye tunas, respectively. Three other ommastrephid squids were identified from fish stomach contents; they were two rare prey species, the Atlantic bird squid Ornithoteuthis antillarum and the orangeback flying squid Sthenoteuthis pteropus, as well as the bait, Argentine shortfin squid Illex argentinus. All fishes fed upon the same size range of H. pe- lagica, including both juvenile and adult squids (Ta- ble 3, Fig. 2), but overall they segregated by prey- ing on squids of different sizes (ANOVA on LRL, F,g 9071 = 16.36, P<0.0001). Post hoc Tukey multiple comparison tests showed three groups of predators: yellowfin tuna and sailfish fed on smaller squids (51 mm and 3.9 g on average), bigeye tuna, white marlin, and longbill spearfish fed on larger individuals (60, 59, and 58 mm; 6.6, 6.2, and 5.7 g, respectively), and albacore and blue marlin fed on squids of intermedi- ate sizes (54 mm and 4.7-4.8 g). Accordingly, bigeye tuna, white marlin, and longbill spearfish fed more on adult squids (89%, 92%, and 93% of the total number of//, pelagica, respectively) than did albacore and blue marlin (85% and 77%) and yellowfin tuna and sailfish (62%. and 73%) (Table 3). NOTE Cherel et al.: Importance of Hyaloteuthis pelagica in the eplpelagic cephalopod community of tfie tropical Atlantic Ocean 149 Table 1 Frequency of occurrence (FO) and proper tion by mass (%) of four broad prey classes (fish, cephalopods, crustaceans, and others) recovered from the stomach contents of 10 species of predatory fishes sampled between October and December 2000 in the central | Atlantic Ocean. Length No. of Fish Cephalopods Crustaceans Others No. of specimens ±SD (cm) stomachs with fresh remains Species FO Mass FO Mass FO Mass FO Mass Alepisauridae Alepisaurus ferox 29 120 ±15 26 18 67.8 9 9.8 19 13.6 9 8.8 Scombridae Acanthocybium solandri 7 136 ±18 7 6 63.5 2 7.5 0.0 6 29.0 Thunnus alalunga 16 109 ±6 15 11 31.5 9 31.7 15 29.7 5 7.1 Thunnus albacares 6 147 ±7 6 6 77.2 5 5.8 5 12.1 2 4.9 Thunnus obesus 24 107 ±22 24 24 77.0 17 20.2 13 2.7 2 0.1 Xiphiidae Xiphias gladius 8 126 ±32 6 6 89.3 3 10.3 2 0.4 0.0 Isiophoridae Istiophorus albicans 4 152 ±13 4 3 85.7 4 13.1 0.0 1 1.2 Makaira nigricans 8 180 ±25 5 3 95.2 2 4.7 0.0 3 0.2 Tetrapturus albidus 7 130 ±10 7 7 48.2 6 51.2 0.0 4 0.6 Tetrapturus pfluegeri 30 143 ±6 29 25 86.8 23 13.0 4 0.0 4 0.1 Table 2 Number (and '?f composition by number) of the main cephalopod prey species found in the diet of Scombridae and Istiophoridae from the tropical Atlantic Ocean between October and December 2000. Only prey species contributing more than 59c by number are reported, n = number of stomachs examined. Thunnus Thunnus Thunnus Istiophorus Makaira Tetrapturus Tetrapturus alalunga albacares obesus albicans nigricans albidus pfluegeri Species n=15 n=5 n=19 n = 4 n = 5 n = 7 n=29 Ommastrephidae Hyaloteuthis pelagica 194(51.5) 133(61.6) 45(19.8) 78(80.4) 101 (75.4) 481(92.9) 897(85.8) Onychoteuthidae Onychoteuthis banksi 21(5.6) 4(1.9) 2(0.9) 0(0.0) 2(1.5) 1(0.2) 12(1.1) Walvisteuthis rancureli 63(16.7) 5(2.3) 62(27.3) 1(1.0) 9(6.7) 0(0.0) 7(0.7) Grimalditeuthidae Grimalditeuthis bonpland 4(1.1) 2(0.9) 17(7.5) 0(0.0) 0(0.0) 2(0.4) 0(0.0) Tremoctopodidae Tremoctopus violaceus 3(0.8) 35(16.2) 3(1.3) 8(8.2) 7(5.2) 11(2.1) 19(1.8) Argonautidae Argonauta argo 6(1.6) 15(6.9) 2(0.9) 8(8.2) 4(3.0) 21(4.1) 82(7.8) Bolitaneidae Japetella diaphana 34(9.0) 5(2.3) 28(12.3) 0(0.0) 1(0.7) 0(0.0) 7(0.7) Other cephalopods 52(13.8) 17(7.9) 68(30.0) 2(2.1) 10(7.5) 2(0.4) 21(2.0) Total 377(100.0) 216(100.0) 227(100.0) 97(100.0) 134(100.0) 518(100.0) 1045(100.0) Discussion This study is the first, to our knowledge, to point out the abundance of H. pelagica in the tropical pelagic ecosystem. In the central Atlantic Ocean, H. pelagica was found as a prey of all the fish species that were investigated. When looking at both the proportion by mass of cephalopods in the fish diet (Table 1) and the proportion by number of H. pelagica in their cephalopod diet (Table 2), H. pelagica was a major prey of white 150 Fishery Bulletin 105(1) ...1 Thunnus alalunga n=95 lllll. . Tetraplurus pfluegeri n=430 25 LRL(mm) 1.0 1.5 2.0 2.5 Figure 2 Frequency distribution of lower rostral lengths (LRL) (mm) of Hyalo- teuthis pelagica glassy flying squid eaten by yellowfin tuna iThunnus alba- cares), albacore (Thunnus alalunga), white marlin (Tetraplurus albidus), and longbill spearfish (Tetraplurus pfluegeri) in the tropical Atlantic Ocean. Table 3 Characteristics of Hyaloteuthis pelagica eaten by large predatory fish from the ti opical Atlantic Ocean. Va ues given are means (±SD) with ranges in parentheses. LRL=lower rostral length, ML=mantle length, M=body mass. Measured Estimated Estimated Adults Number LRL ML M (ML>46mm) Species (n) (mm) (mm) (gl (%) Alepisauridae Alepisaurus ferox 2 1.6-1.8 60-64 6.1-7.3 100.0 Scombridae Acanthocybium solandri 3 2.1*0,2(1.9-2.3) 71+5(67-77) 9.7 ±2.0 (8.5-12.0) 100.0 Thunnus alalunga 95 1.4+0.3(0.5-2.5) 54 ±8 (34-81) 4.7 ±2.1 (0.8-14.0) 85.2 Thunnus albacares 60 1.2 ±0.3 (0.8-1.9) 51 ±7 (39-67) 3.9+1.8(1.5-8.5) 61.7 Thunnus obesus 18 1.6 ±0.5 (0.8-2.3) 60 ±11 (39-77) 6.6 ±3.2 (1.5-12.0) 88.9 Xiphiidae Xiphias gladius 5 1.9 ±0.2 (1.6-2.2) 67 ±6 (59-75) 8.4 ±2.0 (5.9-11.3) 100.0 Istiophoridae Istiophorus albicans 33 1.3+0.2(0.9-1.8) 51 ±5 (42-64) 3.9 ±1.3 (2.0-7.3) 72.7 Makaira nigricans 39 1.4 ±0.4 (0.5-2.0) 54 ±9 (34-69) 4.8 ±2.2 (0.8-9.2) 76.9 Tetraplurus albidus 239 1.6+0.3(0.7-2.3) 59 ±8 (38-78) 6.2 ±2.4 (1.4-12.4) 92.1 Tetraplurus pfluegeri 430 1.5 ±0.3 (0.7-2.4) 58 ±8 (37-79) 5.7 ±2.2 (1.2-12.8) 92.6 marlin, a common food item of albacore, longbill spear- fish, and sailfish and a minor prey for the remaining fishes. However, more information is needed to assess the spatiotemporal importance of H. pelagica in the fish community, because 1) all fish were caught during a relatively short period of time, and 2) a medium to low number of specimens per fish species were collected during the cruises. NOTE Cherel et al.: Importance of Hyaloteuthis pelagico in the epipelagic cephalopod community of the tropical Atlantic Ocean The ommastrephid Sthenoteuthis pteropus, usually abundant in the tropical Atlantic Ocean, was surpris- ingly absent in fish diets in the present study. Two hypotheses may account for that apparent absence: either fish selected H. pelagica rather than S. pteropus, or S. pteropus was not an important and available nek- tonic prey organism at the time of sampling. The latter hypothesis is likely to be the best explanation because tunas and billfishes are known to be opportunistic predators. Moreover, the geographical distribution of S. pteropus shows that juvenile squids are not abundant in the central Atlantic Ocean where cruises of the present investigation took place (Warneke-Cremer, 1986; Zuev and Nikolsky, 1993). Instead, our study underlines the numerical importance of H. pelagica. together with O. antillarum (Vaske et al., 2004), in the area, and our numbers are in agreement with the large catches of the species with nets between 20°S and 31°S off Brazil during 1966 and 1968 (Warneke-Cremer, 1986). The present study documents the largest number of H. pelagica ever reported, thus emphasizing the usefulness of marine predators to gain valuable in- formation on the biology of their prey (Clarke, 1980; Cherel et al., 2004). Other ommastrephid species are important food items of various fishes, seabirds, and marine mammals (Clarke, 1996; Cherel and Klages, 1998), but H. pelagica was previously found only as a rare prey of squids (Shchetinnikov, 1992), fishes (Matthews et al., 1977; Okutani and Tsukada, 1988; Vaske et al., 2004). birds (Harrison et al., 1983), and cetaceans (Robertson and Chivers, 1997). In the same way, the squid Grimalditeuthis bonplandi and the pe- lagic octopods T. violaceus and J. diaphana were rarely found in significant numbers in the diet of cephalopod predators (Okutani and Tsukada, 1988; Le Corre et al., 2003), but we commonly found them as fish prey. Consequently, our study shows that these poorly known cephalopods, together with adults of//, pelagica, con- stitute a link in the transfer of energy from lower trophic levels (most likely mesozooplankton) to higher trophic levels (including tunas and billfishes) in the tropical Atlantic Ocean). Acknowledgments The authors thank P. Borsa, P. Dewals, and O. Maury for their help to collect scientific samples on board, and the captain and crew of the RV Shoyo-Maru. Literature cited Bower, J. R., and T. Ichii. 2005. The red flying squid iOmmastrephes bartramii): a review of recent research and the fishery in Japan. Fish. Res. 76:39-55. Cherel, Y., G. Duhamel, and N. Gasco. 2004. Cephalopod fauna of subantarctic islands: new information from predators. Mar, Ecol. Prog. Ser. 266:143-156. Cherel., Y., and N. Klages. 1998. A review of the food of albatrosses. In Albatross biology and conservation (G. Robertson, and R. Gales, eds.), p 113-136. Surrey Beatty and Sons, Chipping Norton, Australia. Clarke, M. R. 1980. Cephalopoda in the diet of sperm whales of the Southern Hemisphere and their bearing on sperm whale biology. Discovery Rep. 37:1-324. 1986. A handbook for the identification of cephalopod beaks, 273 p. Clarendon Press, Oxford, England. 1996. The role of cephalopods in the world's oceans. Phil. Trans. R. Soc. Lond. B 351:977-1112. Dunning, M., and S. B. Brandt. 1985. Distribution and life history of deep-water squid of commercial interest from Australia. Aust. J. Mar. Freshw. Res. 36:343-359. Harrison, C. S., T. S. Hida, and M. P. Seki. 1983. Hawaiian seabird ecology. Wildl. Monogr. 85: 1-71. Le Corre, M., Y. Cherel, F. Lagarde, H. Lormee, and P. Jouventin. 2003. Seasonal and interannual variation in the feed- ing ecology of a tropical oceanic seabird, the red-tailed tropicbird Phaethon rubricauda. Mar. Ecol. Prog. Ser. 255:289-301. Matthews, F. D., D. M. Damkaer, L. W. Knapp, and B. B. Collette. 1977. Food of western North Atlantic tunas iThunnus) and lancetfishes (Alepisaurus). NOAA Tech. Rep. NMSF SSRF-706, 19 p. Nesis, K. N. 1987. Cephalopods of the world. Squids, cuttlefishes, octopuses, and allies, 351 p. TFH Pubis., Neptune City, NJ. Nesis, K. N., and C. M. Nigmatullin. 1979. The distribution and biology of the genus Orni- thoteuthis Okada, 1927 and Hyaloteuthis Gray, 1849 (Cephalopoda: Oegopsida). Bull. Moscow Soc. Nat. 84:50-63. [In Russian.] Nigmatullin, C. M., K. N. Nesis, and A. I. Arkhipkin. 2001. A review of the biology of the jumbo squid Dosidi- cus gigas (Cephalopoda;Ommastrephidae). Fish. Res. 54:9-19. Okutani, T, and S. Tsukada. 1988. Squids eaten by lancetfish and tunas in the tropical Indo-Pacific Ocean. J. Tokyo Univ. Fish. 75:1-44. Robertson, K. M., and S. J. Chivers. 1997. Prey occurrence in pantropical spotted dol- phins, Stenella attenuata, from the eastern tropical Pacific. Fish. Bull. 95:334-348. Rodhouse, P. G. 1997. Large and meso-scale distribution of the ommas- trephid squid Martialia hyadesi in the Southern Ocean: a synthesis of information relevant to fishery forecasting and management. Korean J. Polar Res. 8:145-154. Roper, F. E., and M. J. Sweeney. 1984. Cephalopods of the world. An annotated and illus- trated catalogue of species of interest to fisheries. FAG Species Catalogue 3:1-277. Shchetinnikov, A. S. 1992. Feeding spectrum of squid Sthenoteuthis oualani- ensis (Oegopsida) in the eastern Pacific. J. Mar. Biol. Assoc. U.K. 72:849-860. Vaske Jr., T, C. M. Vooren. and R. P. Lessa. 2004. Feeding habits of four species of Istiophoridae (Pisces: Perciformes) from northeastern Brazil. En- viron. Biol. Fishes 70:293-304. 152 Fishery Bulletin 105(1) Warneke-Cremer, C. 1986. Contributions to the systematics of ommastrephid squid (Mollusca, Cephalopoda, Teuthoidea) and their distribution in the Atlantic, based on the catches of FFS "Walter Herwig" made during 1966 and 1968. Mitteil. Inst. Seefisch. Hamburg 40:1-116. lln German.] Wormuth, J. H. 1998. Workshop deliberations on the Ommastrephidae: a brief history of their systematics and a review of the systematics, distribution, and biology of the Genera Martialia Rochebrune and Mabille, 1889, Todaropsis Girard, 1890, Dosidicus Steenstrup, 1857, Hyaloteuthis Gray, 1849, and Eucleoteuthis Berry, 1916. Smithson- ian Contrib. Zool. 586:373-383. Zuev, G. v., and V. N. Nikolsky 1993. Ecological mechanisms related to intraspecific structure of the nektonic squid Sthenoteuthis pteropus (Steenstrup). In Recent advances in fisheries biology (T. Okutani, R. K. O'Dor, and T. Kubodera, eds.), p. 653-664. Tokai Univ. Press, Tokyo, Japan. Zuyev, G., Ch. Nigmatullin, M. Chesalin, and K. Nesis. 2002. Main results of long-term worldwide studies on tropical nektonic oceanic squid genus Sthenoteuthis: an overview of the Soviet investigations. Bull. Mar. Sci. 71:1019-1060. 153 Measurements of total scattering spectra from bocaccio iSebastes paucispinis) Stephane G. Conti (contact author)^ Benjamin D. Maurer^ Mark A. Drawbridge^ David A. Demer' Email address for S. G. Conti: sconti@ucsd.edu ' Southwest Fishenes Science Center 8604 La Jolla Shores Drive La Jolla, California 92037 Present address for S. G. Conti; Marine Physical Laboratory Scripps Institute of Oceanography University of California San Diego 9500 Gilman Drive La Jolla, California 92093-0238 ^ Hubbs-SeaWorld Research Institute 2595 Ingraham St. San Diego, California 92109 Marine sportfishing in southern California is a huge industry with annual revenues totaling many bil- lions of dollars. However, the stocks of lingcod and six rockfish species have been declared overfished by the Pacific Fisheries Management Coun- cil. As part of a multifaceted fisheries management plan, marine conser- vation areas, covering many million square nautical miles, have been mandated. To monitor the recovery of the rockfish stocks in these areas, scientists are faced with the follow- ing challenges: 1) multiple species of rockfish exist in these areas; 2) the species reside near or on the bottom at depths of 80 to 300 m; and 3) they are low in numerical density. To meet these challenges, multifrequency echosounders, multibeam sonar, and cameras mounted on remotely oper- ated vehicles are frequently used (Reynolds et al., 2001). The accuracy and precision of these echosounder results are largely dependent upon the accuracy of the species classifica- tion and target strength estimation (MacLennan and Simmonds, 1992). Broad bandwidth characteriza- tion of sound scatter from marine organisms has some potential for re- motely classifying fish species (Conti and Demer, 2003), shapes and sizes (Conti et al., 2005), behaviors (Conti et al., 2006b), and to validate models for target strength estimation (De- mer and Conti, 2003). All of these studies have employed variants of a new method for measuring the broad bandwidth total scattering cross sec- tion ia-p) of animals moving in a re- verberant tank. With the new method, the total scattering cross section (oj) of live animals in tanks is obtained from a comparison of the coherent and inco- herent acoustical intensities reverber- ated in a tank (de Rosny and Roux, 2001, 2003). The accuracy of this measurement technique was shown by using standard metal spheres (De- mer et al., 2003). This technique was successfully used on krill (Demer and Conti, 2003; Conti et al., 2006a), fish (Conti and Demer, 2003), and humans (Conti et al., 2004). In our study, we explored the potential and limitations of the method to char- acterize the broad bandwidth sound scattering from bocaccio (Sebastes paucispinis). Materials and methods The total scattering cross section, o-p, of bocaccio was measured over acous- tic frequencies ranging from 10 to 150 kHz with a group of fish {n=20) swimming freely in a large, insulated fiberglass tank at Hubbs-SeaWorld Research Institute, San Diego, CA, on 1 and 2 July 2004. The tank had 5.1 cm of foam insulation on the exterior, measured 2.44 m in diameter, and was filled with seawater to a depth of 1.37 m (V [volume] = 6.4 m^). The pool was thermostated at approximately 12°C. The acoustic measurement technique and a variety of its applications have been well documented (de Rosny and Roux, 2001; Conti and Demer, 2003; Demer and Conti, 2003; Demer et al., 2003; Conti et al., 2004). However, the general procedure and details of these experiments are presented here for convenience and clarity. Each of the 20 fish was handled one time, a week prior to the experi- ment, to measure their weight (W) and total length (L). These data were summarized and plotted in graphs (Table 1, Fig. 1). An emitter transmitted M acousti- cal pulses into the tank every other second {dT=2s). The corresponding reverberation time-series hf,{t) were simultaneously recorded on multiple receivers while the fish were swim- ming between consecutive shots. The boundaries, volume, as well as the positions of the emitter and the re- ceivers in the tank remained identi- cal during the measurements. The time series hi,(t) were composed of echoes from the boundaries of the tank and the fish. For two consecu- tive time series h/^it) and h^^j(t), the contributions from the boundaries of the tank were identical, whereas the contributions from the fish were not. The coherent '" k=i and incoherent 1 M Manuscript submitted 19 July 2005 to the Scientific Editor. Manuscript approved for publication 7 July 2006. Fish. Bull. 105:153-157 (2007). 154 Fishery Bullelin 105(1 3 2,5 O) ^ 1.5 1^=15.74 1^-0.27 t-'' A=23% • ^,-''' *-• -- • ---' ^ ,' - » » 1 1 J 1 1 J 1 04 06 08 01 012 014 016 018 02 L3 (m3) Figure 1 Weight (W) of each boccacio (Sehaates paticispinis) versus length-to-the-third power (L'^) (dark dots), and linear least-squares fit (dashed line), and the mean deviation A is shown below the equation. intensities in the tank were estimated from the M recorde(i time series. The coherent component repre- sents the acoustical intensity reverberated by the fixed Table 1 Total length (L) and weight ( W). and the mean dard deviation of each,, for the 20 boccacio and stan- iSebastes paiicispinis) used to measure the total scattering cross section over acoustic frequencies from 10 to 150 kHz in a laboratory tank at Hubbs SeaWorld Research Institute, San Diego, CA. Fish number Total length (L, mm) Weight (W, kgl 1 467 0.99 2 444 0.85 3 480 1.03 4 467 0.97 5 453 1.18 6 442 0.77 7 519 1.4 8 439 0.73 9 450 1.13 10 500 1.95 11 448 1.25 12 510 2.08 13 480 1.54 14 508 1.84 15 534 2.47 16 563 2.73 17 491 1.88 18 394 0.95 19 382 0.95 20 405 1.23 Mean 468 1.37 Standard deviation 45.5 0.57 boundaries of the tank. The incoherent component also accounted for the acoustical intensity scattered by the fish. When the positions of the fish were uncorrelated between consecutive pulses, the ratio Sit) of the coher- ent to the incoherent intensities decreased exponentially with the scattering mean free path l^ of the fish (de Rosny and Roux, 2001): S(t)-- S,.(f) SM) = exp l ( cNor^ where the bracketing [ ] designates the average for mul- tiple receivers. The scattering mean free path is related to the total scattering cross section of a single fish in the tank ia-,,), the sound speed (c), the number of fish (N). and the volume (V). Multiple receivers may be used simultane- ously to reduce the heterogeneities of the acoustical field on the coherent and incoherent intensities in the tank. Knowing N, c, and V, Oj (normalized to a single fish) was estimated from the exponential decay of Sit). Thus, Oj, averaged over 10 to 150 kHz was estimated for a single bocaccio, and its total scattering spectrum was similarly estimated after filtering the recorded time series h/St) into twenty narrow frequency bands. Each band corresponded to the bandwidth of the transmitted chirp, divided by twenty. The signal acquisition system (Fig. 2) consisted of a function generator CompuGen 1100 (GageApplied, Montreal, Canada) internally clocked with two 16-bit CompuScope 1610 (GageApplied, Montreal, Canada) dual-channel acquisition boards in a portable computer. The internal clocking of the function generator and the acquisition boards allowed perfect timing between the emitted and the recorded signals. Ensembles of M=100 chirps were transmitted over 50 ms every other second for three frequency bandwidths from 10 to 40 kHz with an ITCIOOIB emitter if,=25 kHz); 30 to 70 kHz with an ITC1032 emitter (/;=56kHz); and 60 to 150 kHz with NOTE Conti et al.; Measurements of the total scattering spectra from Sebasfes paucispinis) 155 Figure 2 Diagram of the experimental setup: a computer with integrated function generator (CompuGen 1100, GageApplied, Montreal, Canada) and analog to digital acquisition system (CompuScope 1610, GageApplied, Montreal, Canada) connected to the emitters in black (ITCIOOIB, ITC1032, ITC1042; ITC Transducers Company, Santa Barbara, CA) and receivers in gray (one ITClOOl, one ITC1032, and two ITC1042; ITC Transducers Company, Santa Barbara, CA) in the tank. The transmitted signal is amplified with a power amplifier (Krohn-Hite 7500, Krohn-Hite Corporation, Brockton, MA). an ITC1042 emitter (/; = 105 kHz) (ITC Transducers Company, Santa Barbara, CA). The transducers were inserted from the top of the aquarium. The amplitude of the chirps from the function generator was 1 volt peak- to-peak, amplified 100 times (40 dB) with a Krohn-Hite 7500 amplifier (Krohn-Hite Corporation, Brockton, MA). The corresponding reverberation time series were re- corded at a 500-kHz sampling rate, for at least 90 ms on a four transducer array consisting of one ITClOOl, one ITC1032, and two ITC1042 transducers. To increase the signal-to-noise ratio, the recorded reverberation time series were cross correlated with the transmitted signal to obtain the impulse responses hi^it). The measurements for the lowest (f^=25 kHz) and highest (/"j.=105 kHz) frequencies bands were repeated 11 times. The measurements for the center frequency {f,=50 kHz) band were repeated 20 times. For f^=25, 50, and 105 kHz, the total scattering cross sections were estimated from the signals on the ITClOOl, ITC1032, and two ITC1042 receivers, respectively. Results The mean weight (W) and total length (L) of the 20 bocaccio were 1.37 kg (ranging from 0.73 to 2.73 kg; standard deviation (SD) = 0.57 kg), and 468 mm (rang- ing from 382 to 563 mm; SD = 45.5 cm), respectively (Table 1). Fish masses did not correlate well to fish lengths (Fig. 1). The mean deviation was about 25%, for fish weight to fish length-to-the-third-power. This fit may indicate heterogeneity in the shapes (i.e., long and thin versus short and wide) and could be observed from visual inspection of the fish. Because the fish were not moving very actively during the experiments, the time between consecutive shots ST had to be increased to assure uncorrelated positions of the fish. This was effectively achieved by considering the shots k and ^-h20 instead of ^ and ^-i-l to estimate -S,(t), resulting in ^^,= 40 s between the shots. By in- creasing the time between shots, the measured total scattering cross section reached a stable plateau at Of 0.01 m~ for each of the considered frequency bands (Fig. 3). This plateau indicated that the positions of the fish were uncorrelated between shots, and the mea- sured total scattering cross section was not biased by the correlation of fish positions (Conti et al., 2006a). The measurement variance was conspicuously higher in the lowest and highest frequency bands because of decreased signal-to-noise ratios. The differences in sig- nal-to-noise ratio with frequency was due to the experi- mental setup at high frequency, and the lack of acoustic modes propagating in the tank at low frequencies. The total scattering cross section was equivalent to the one expected for a rigid sphere of diameter 90 mm in water (Fig. 4; Faran, 1951), which is of the order of the size of the swimbladder for the fish in the tank (Foote, 1979). The largest discrepancies between the measurements and expectations were at the lowest and highest fre- quencies, again because of lower signal-to-noise ratios. The total scattering cross-sectional area was virtu- ally the same for each of the three frequency bands because all three measurements were in the geometric scattering domain for these fish. In other words, in all three frequency bands the radius of the equivalent scattering sphere was greater than half the wavelength. In the geometric domain, the scattering cross-section increased with the square of the frequency, in contrast 156 Fishery Bulletin 105(1) to the frequency to the fourth power in the Rayleigh domain. For these fish, the transition from the Rayleigh to geometric regimes occurred at frequencies around 15 kHz (Fig. 4). The spectra of Oj- are shown for over 60 narrow band- widths (Fig. 4) after we averaged the results from all of the experiments. Each of the three frequency bands were resolved into 20 narrow bandwidths. Further nar- rowing of the filters on the reverberation time series served only to increase the standard deviation of the measurements without providing more information on the spectra. The mean a^. was in agreement with the previous broad bandwidth average. The a-j, and its standard deviation increased at the lowest and highest frequencies, likely because of a decrease in signal-to- noise ratios. The back scattering target strength TS=101ogiQ(a^,) for these fish can be compared to the target strength from a rigid sphere in water of radius a for which the resonances are damped by the surrounding flesh. For this type of scatterer, the target strength at high ka : 2nf is equal to \Q\og^Q(mi^) diameter 90 mm, lOlogjQ For a rigid sphere in water of (TO2)=-22dB. The swimbladder 0.01 N -I -in 005 -CM ^^^^^ 10 15 20 67(5) 25 30 0,015 N STI C II 0.01 0.005 10 15 20 bT{s) 25 30 0.015 N ,1 0.01 0,005 —I 1— 40 20 87(5) Figure 3 Total scattering cross section a-j., normalized to one bocaccio iSebastes pau- cispinis), versus time between shots, 6T, for three frequency bands. The mea- surements (dotted lines) for the lowest and highest frequencies band were repeated 11 times, and 20 times for the center frequencies band, and averaged (dashed dark lines). For short time between shots, Oj. was biased because of slow fish motion. The measurements stabilized with increasing time between shots,6r. is gas filled and can be considered a hard scatterer and its resonances are damped by the surrounding flesh. As a first-order approximation, the theoretical predictions for the air bubble can be used to estimate the target strength of the fish by adjusting the radius of the rigid sphere to the size of the fish swimbladder. Discussion The mean Oj from 10 to 150 kHz was measured from boccacio with a mean length of 468 mm. It is roughly equivalent to that from a 90-mm-diameter rigid sphere in water. This result is in agreement with the approxi- mate size of their swimbladder. Thus, the spectrum of the total scattering cross section for boccacio can be measured with this technique. Despite these positive results, the following should be considered for future experiments. These measurements were made against the frequency from a heterogeneous- size group of a single species of fish. As such, the re- sults do not permit comparisons of Oj. with frequency, and animal species and morphological features (e.g., size, shape, length, sex, etc.). Measurements should be made of individual fish. It should also be noted that the bocaccio in these experiments moved very slowly. Simply increasing the time to achieve incoherence be- tween pulses may itself introduce systematic and random measure- ment error because of instabili- ties in the medium. That is, some additional incoherence can result from bubbles and fluctuations in the water temperature, sound speed, volume, and water surface. The magnitude of this incoherence could eclipse the differences in Oj. between individual fish. In these measurements, bubbles and motion of the air-water interface caused by breaking bubbles and fish motion were noticeable visually but had only minor effects on the data. The most significant shortcom- ing of these experiments was a generally low signal-to-noise ra- tio because of the tank size and material properties. This low ratio caused appreciable measurement uncertainty at the lowest and highest frequencies. The water volume was large for the projected signal intensities, and the fiber- glass boundaries were not very reflective in comparison to other materials such as stainless steel and glass used in previous experi- ments. Therefore, to obtain and compare measurements of Oj. from 35 40 "T r 35 40 NOTE Conti et al ; Measurements of the total scattering spectra from Sebastes paucispinis) 157 rockfish of different species and sizes, fu- ture measurements should be made of in- dividual fish in tank volumes appropriate to the size and the frequency range of the fish, insuring high signal-to-noise ratios and reverberation times. Because of fish handling constraints and access to only a single species of rockfish, it was not pos- sible to do to include these measurements in the present study. Acknowledgments We are grateful to P. Sylvia and the rest of the staff at Hubbs-SeaWorld Research Institute for hosting the experiments, and to Chevron Corporation for sponsoring the research. Literature cited Geometric domain Rayleigh domain Conti, S. G., and D. A. Demer. 2003. Wide -bandwidth acoustical charac- terization of anchovy and sardine from reverberation measurements in an echoic tank. ICES J. Mar. Sci. 60(3):617- 624. Conti, S. G., D. A. Demer, and A. S. Brierley. 2005. Broadbandwidth sound scattering and absorption from krill (Meganyctiph- ones norvegica), mysids {Praunus flexu- ousus and Neomysis integer) and shrimp (Crangon crangonl. ICES J. Mar. Sci. 62{5):956-965. Conti, S. G., J. de Rosny, P. Roux, and D. A. Demer . 2006a. Scatterer motion characterization and estima- tion in a reverberant medium. J. Acoust. Soc. Am. 119(2):769-776. Conti, S. G., P. Roux, D. A. Demer, and J. de Rosny. 2004. Measurement of the scattering and absorption cross-section of the human body. App. Phy. Let. 84(5):819-821. Conti, S. G., P. Roux. C. Fauvel, B. D. Maurer, and D. A. Demer. 2006b. Acoustic monitoring of fish density, behavior, and growth rate in a tank. Aquaculture 251(2-4):314- 323. Demer, D. A., and S. G. Conti. 2003. Reconciling theoretical versus empirical target strengths of krill; effects of phase variability on the distorted wave Born approximation. ICES J. Mar. Sci. 60i2):429-434. Demer, D. A., S. Conti, J. De Rosny, P. Roux. 2003. Absolute measurements of total target strength from reverberation in a cavity. J. Acoust. Soc. Am. 113:1387-1394. de Rosny, J., and P. Roux. 2001. Multiple scattering in a reflecting cavity: appli- cation to fish counting in a tank. J. Ac. Soc. Am. 109:2587-2597. Frequency (kHz) Figure 4 Total scattering spectra Oj. normalized to one boccacio ^Sebastes paucispinis), over the full frequency band, low to high frequency bands (solid lines from dark to light), and one standard deviation error bounds for these measurements (dashed lines). The spectra were obtained after filtering the data in 20 narrow bands bins for each of the three frequency bands. Theoretical predictions for a 90-mm-diameter rigid sphere in water (Faran, 1951; solid line from to 150 kHz), showing the transition between the Rayleigh and the geometric regimes. de Rosny, J., P. Roux, M. Fink, and J. H. Page. 2003. Field fluctuation spectroscopy in a reverber- ant cavity with moving scatterers. Phy. Rev. Let. 90:9-4302. Faran. J. J. 1951. Sound scattering by solid cylinders and spheres. J. Ac. Soc. Am. 23(4):405-418. Foote, K. G. 1979. Fish target-strength-to-length regressions for application in fisheries research. Proceedings of the Ultrasonic International 19, Graz, Austria, 327-333 p. IPC Science and Technology Press Ltd., Guilford, Eng- land. MacLennan, D. N., and E. J. Simmonds. 1992. Fisheries acoustics, 325 p. Chapman and Hall, London. Reynolds, J. R., R. C. Highsmith, B. Konar, C. G. Wheat, and D. Doudna. 2001. Fisheries and fisheries habitat investigations using undersea tehnologies, OCEANS. MTS/IEEE Confer- ence and Exhibition. In Proceedings of the Oceans 2001 conference; 5-8 November 2000, Honolulu, HI. Hawaiian Sea Grant Program, Univ. Hawaii at Manoa, Manoa, HI. 158 Fishery Bulletin Guidelines for authors Content of manuscripts Contributions published in Fishery Bulletin describe original research in marine fishery science, fishery engi- neering and economics, as well as the areas of marine environmental and ecological sciences (including model- ing). Although all contributions are subject to peer review, responsibility for the contents of papers rests upon the authors and not upon the editor or publisher. Submission of an article implies that the article is original and is not being considered for publication elsewhere. Manuscripts must be written in English. Authors whose native lan- guage is not English are strongly advised to have their manuscripts checked by English-speaking colleagues prior to submission. 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. Manuscript preparation Title page should include authors' full names and mailing addresses and the senior authors telephone, fax number, and e-mail address, and a list of key words to describe the contents of the manuscript. Abstract should be limited to 150 words (one-half 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 cir- culated by abstracting agencies, it is important that they represent the research clearly and concisely. Text must be typed in 12 point Times New Roman font throughout. A brief introduction should convey the broad significance of the paper; the remainder of the paper should be divided into the following sections: Materials and methods, Results, Discussion (or 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, abbre- viations, and technical terms should be written out in full the first time they are used. Include FAO common names for species in the list of keywords and in the introduction. Regional common names may be used throughout the rest of the text if they are different 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 cur- rent 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 Novem- ber 2000. Measurements should be expressed in metric units, e.g., 58 metric tons (t); if other units of measure- ment are used, please make this fact explicit to the reader. Write out the numbers zero through nine unless they form part of measurement units (e.g., nine fish but 9 mm). Text footnotes should be inserted in 9-point font at the bottom of the page that displays the first citation of the footnote. Footnotes should be formatted in the same manner as citations. Footnote all personal communica- tions, unpublished data, and unpublished manuscripts with full address of the communicator or author, or, as in the case of unpublished data, where the data are on file. Authors are advised to avoid references to nonstandard (gray) literature (such as internal, project, processed, or administrative reports, ICES Council Minutes, IWC Minutes or Working Papers, any "research" or "work- ing" documents, laboratory or contract reports. Man- agement Council reports, and manuscripts in review) wherever possible. If these references are used, present them as footnotes and list whether they are available at NTIS (National Technical Information Service) or at some other public depository. 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). Literature cited comprises published works and those accepted for publication in peer-reviewed litera- ture (in press). Follow the name and year system for cita- tion format in the "Literature cited" section. If there is a sequence of citations in the text, list chronologically: ( Smith, 1932 ; Green, 1947; Smith and Jones, 1985 ). Abbre- viations 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. If the authorship for a sequence of citations is identical, list works chronologically. Tables and figures — general format • Zeros should precede all decimal points for values less than one. • Sample size, n, should be italicized. • Capitalize the first letter of the first word in all labels within figures. • Do not use overly large font sizes in maps and for units of measurements along axes in figures. • Do not use bold fonts or bold lines in figures. • Do not place outline rules around graphs. • Do not use horizontal lines in graphs to indicate mea- surement units on axes. • Use a comma in numbers of five digits or more (e.g. 13,000 but 3000). • Maps should have a North arrow (or compass sign) and degrees latitude-longitude (e.g., 170°E) Fishery Bulletin 105(1) 159 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 footnote markers. Use asterisks to indicate probability in statistical data. Do not type table legends on a separate page; place them on the same page as the table data. 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 will aid in the comprehension of the text, but they should be limited to presenting patterns rather than raw data. Figures are costly to print and should be limited to six. Figures must be labeled with author's name and number of figure. Avoid placing labels vertically (except on 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 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). Reprints are available free of charge to the senior author (50 copies] and to his or her laboratory (50 copies). Submission The Scientific Editorial Office encourages authors to submit their manuscripts as a single PDF (pre- ferred), or Word (zipped) document by e-mail to Fish- ery.Bulletin@noaa.gov. Please use the subject heading "Fishery Bulletin manuscript submission." Do not send encrypted files. For further details on electronic submis- sion, please contact the Scientific Editorial Office directly (see address below). Or you may send your manuscript on compact disc in one of the above formats along with four printed copies (one original plus three copies [stapled]) to the Scientific Editor, at the address shown below. Dr. Adam Moles Scientific Editor, Fishery Bulletin 11305 Glacier Hwy Juneau, AK 99801-8626 Once the manuscript has been accepted for publication, you will be asked to submit a final software 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 as EPS files. Send a copy of figures in original software if conversion to any of these formats yields a degraded version. MBI. WHOI I.IBKARY UH nFl U 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) (Daytiine phone including area code) (Purchase Order No. ) Charge your order. Please Choose Method of Payment: I I Check Payable to the Superintendent of Documents □ GPO Deposit Account I I I I I I I | — fl I I VISA or MasterCard Account (Credit card expiration date) (Authorizing Signature) Mail To: Superintendent of Documents P.O. Box 371954, Pittsburgh, PA 15250-7954 To fax your orders (202) 512-2250 Thank you for your order! Also available online at http;//bookstore. gpo.gov/collections/fishery-bulletin U.S. Department of Commerce Volume 105 Number 2 April 2007 U.S. Department of Commerce Carlos M. Gutierrez Secretary National Oceanic and Atmosplieric Administration Vice Admiral Conrad C. Lautenbacher Jr., USN (ret.) Under Secretary for Oceans and Atmosphere National Marine Fisheries Service William T. Hogarth Assistant Administrator for Fisfieries Scientiric Editor Adam Moles, Ph.D. Associate Editor Elizabeth Calvert * National Marine Fisheries Service, NOAA 11305 Glacier Highway Juneau, Alaska 99801-8626 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 C 15700, Seattle, WA 98115-0070. Periodicals postage is paid af 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: $28.00 domestic and $35.00 foreign. See back for order form. Editorial Committee Thomas Shirley Texas A&M University David Somerton National Marine Fisheries Service Mark Terceiro National Marine Fisheries Service Fishery Bulletin web site: 'WAVw.fishbull.noaa.gov The Fishery Bulletin carries original research reports and technical notes on investigations in fisher\' 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 sub.scription 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 105 Number 2 April 2007 Fishery Bulletin Contents Articles MBLWHO! Librarv JUN 1 9 20U7 WOODS HOLE Massachusetts 02543 The conclusions and opinions ex- pressed in Fisher}' Bulletin are solely those of the authors and do not represent the official position of the National Marine Fisher-ies Ser- vice (NOAA) or any other agency or institution. The National Marine Fisheries Service (NMFS) does not approve, recommend, or endorse any proprie- tar\- 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. 161-167 Centner, Brad Sensitivity of angler benefit estimates from a model of recreational demand to ttie definition of tfie substitute sites considered by the angler 168-179 Anderson, Tara J., and Mary M. Yoklavich Multiscale habitat associations of deepwater demersal fishes off central California 180-188 McClelland, Gary, and Jason Melendy Use of endoparasitic helminths as tags in delineating stocks of American plaice iHippog/ossoides platessoides) from the southern Gulf of St. Lawrence and Cape Breton Shelf 189-196 Hamilton, Judy, and Brenda Konar Implications of substrate complexity and kelp diversity for south-central Alaskan nearshore fish communities 197-206 Ward, Rocky, Kevin Bowers, Rebecca Hensley, Brandon Mobley, and Ed Belouski Genetic variability in spotted seatrout (Cynoscion nebulosus), determined with microsatellite DNA markers 207-233 Smith-Vaniz, William F,, and Kent E. Carpenter Review of the crevalle lacks, Caranx hippos complex (Teleostei; Carangidae), with a description of a new species from West Africa Fishery Bulletin 105(2) 234-248 Trites, Andrew W., Donald G. Calkins, and Arliss J. Winship Diets of Stellar sea lions (Eumetopias jubatus) in Southeast Alaska, 1993-1999 249-265 Margulies, Daniel, Jenny M. Suter, Sharon L. Hunt, Robert J. Olson, Vernon P. Scholey, Jeanne B. Wexler, and Akio Nakazawa Spawning and early development of captive yellowfin tuna (Thunnus albacares) 266-277 Zeller, Dirk, Shawn Booth, Gerald Davis, and Daniel Pauly Re-estimation of small-scale fishery catches for US. flag-associated island areas in the western Pacific: the last 50 years 278-291 Somerton, David A., Peter T. Munro, and Kenneth L. Weinberg Whole-gear efficiency of a benthic survey trawl for flatfish Notes 292-295 Strange, Rex M., and Carol A. Stepien Yellow iPerca flavescens) and Eurasian (P. fluviatilis) perch distinguished in fried fish samples by DNA analysis 296-304 Cooper, Daniel W., Katherine P. Maslenikov, and Donald R. Gunderson Natural mortality rate, annual fecundity, and maturity at length for Greenland halibut (Reinhardtius h/ppoglossoides) from the northeastern Pacific Ocean 305-309 Roemer, Marja E., and Kenneth Ollveira Validation of back-calculation equations for juvenile bluefish (Pomatomus saltatnx) with the use of tetracycline-marked otoliths 310-311 Guidelines for authors 312 Best Paper Awards 161 Abstract — Fishery managers are mandated to understand the effects that environmental damage, fishery regulations, and habitat improve- ment projects have on the net benefits that recreational anglers derive from their sport. Since 1994, the National Marine Fisheries Service (NMFS) has worked to develop a consistent method for estimating net benefits through site choice models of recreational trip demand. In estimating net benefits with these models, there is a tradeoff between computational efficiency and angler behavior in reality. This article examines this tradeoff by consider- ing the sensitivity of angler-welfare estimates for an increase in striped bass (Morone saxatalis) angling quality across choice sets with five travel distance cut-offs and compares those estimates to a model with an unrestricted choice set. This article shows that 95'7( confidence intervals for welfare estimates of an increase in the striped bass catch and keep rate overlap for all distance-based choice sets specified here. Sensitivity of angler benefit estimates from a model of recreational demand to the definition of the substitute sites considered by the angler Brad Centner NMFS Office of Science and Technology Fisheries Statistics and Economics Division 1315 East West Highway Sliver Spring, Maryland 20910 Email address: brad genter@noaa.gov Manuscript submitted 24 August 2005 to the Scientific Editor. Manuscript approved for publication 7 August 2006 by the Scientific Editor. Fish. Bull. 105:161-167 (2007). Recreational angling is the second most popular outdoor sport nationwide when measured by number of partici- pants. In 2004, 10.2 million anglers took 73.8 million recreational trips in the United States, exclusive of Alaska, Hawaii, and Texas (NMFS^). In addi- tion to participation, anglers spend $20.4 billion dollars annually on trip- related and durable expenditures to pursue saltwater gamefish (Gentner et al., 2001), producing $30.5 billion in economic impacts and supporting nearly 350,000 jobs (Steinback et. al, 2004). Recreational fishing is an economically important activity and the National Marine Fisheries Service (NMFS) is mandated by law to exam- ine changes in net benefits to anglers after the impact of environmental damage (oil spills, algal blooms, etc.), fishery regulations (bag limits, size limits, seasonal closures), and habitat improvement projects (damn removal, water quality improvements, etc.). Calculation of net benefits involves an examination of angler behavior when they make choices about taking recre- ational fishing trips. Modeling angler trip demand in- volves observing anglers making rec- reation site choices and using a site choice model to estimate a recreation- al trip demand function. Site-choice models are typically estimated by us- ing a random utility model (RUM). RUMs are used to estimate net ben- efits by looking at the cost of travel- ing to the site that anglers selected and comparing that cost to the cost of traveling to other sites in their choice set (set of sites considered by the an- gler). Without any other information about the site, these models allow one to estimate the net benefits of access to that site which can be used to ex- amine closures due to environmental damages or regulation. If site-qual- ity information is available, such as catch rates or other measures of en- vironmental quality, the net benefits of those ecosystem services can be estimated as well. Since 1994, it has been the goal of the NMFS to develop a consistent method for estimating recreational site-choice models to increase the speed and efficiency of meeting legal mandates. To this end, NMFS has sponsored a good deal of research into RUMs of recreational site choice to value site closures and angling qual- ity (the quality of the angling experi- ence as measured by catch and keep rates) (Haab and Hicks, 1999, Jones and Lupi, 1999, Parsons et. al, 1999). From this, and other work, the com- position of an individual's choice set can impact net benefit estimates, giv- ing rise to several difficulties when modeling angler net benefits. First, NMFS's RUM models concentrate on only single day trips, because it is difficult to disentangle the value of angling for anglers on trips that have ^ NMFS (National Marine Fisheries Service). 2006. Fisheries Statistics and Eco- nomics Division. Marine Recreational Fisheries Statistical Survey Real Time Data Queries. Website: http://www. st.nmfs.gov/stl/recreational/database/ queries/index. html (accessed on 13 August 2006). 162 Fishery Bulletin 105(2) multiple purposes. Because focus is strictly on single day trips, it would be incorrect to include sites in an angler's choice set if those sites are "too far" for the angler to consider when choosing a site for a single day trip. Second, a large number of sites in each individual's choice set can be computationally costly, particularly when a nested choice structure is appropriate, and in- crease the time it takes to bring policy analyses to the table. This problem may indicate that there is a tradeoff between computational efficiency and angler behavior in reality; a balance that will be examined here. There is literature on the specification of choice sets based on many factors including distance. Parsons and Hauber (1998) estimated a freshwater recreational an- gler site choice model and found that there is little difference in the magnitude of welfare effects as one reduces the spatial scope of choice sets until a threshold of 1.6 hours one-way travel time is reached. This spa- tial scope translates into 32 mile and 80 mile distance thresholds, if one assumes a 20 mile per hour (mph) urban travel speed and a 50 mph highway travel speed, respectively. Below that threshold, welfare estimates inflate as the constraint tightens. Whitehead and Haab (1999) estimated a site choice model using a range of choice sets constructed with distance and site-quality metrics. They found that there is very little difference in the trip cost coefficients across distance-based choice sets that eliminate between 13% and 82% of the avail- able sites. Hicks and Strand (2000) found that because the probability of choosing a site depends on the choice set, the likelihood function is also dependent on the choice set. If the choice set is incorrect, biased param- eter estimates could be a consequence. The welfare es- timates derived in the "Materials and methods" section below explicitly include the choice set and demonstrate this interaction. This analysis will examine the sensitivity of wel- fare estimates in a RUM model of recreational demand across six distance-based definitions of site choice. This analysis will focus on a single species, striped bass (Morone saxatalis), from a single mode (the private rental boat mode) to avoid a nested choice structure. A simulation approach will be used to derive confidence intervals around these estimates in order to examine the significance of any differences found and to ex- pand the literature that has previously been focused on only on the magnitude of the differences in welfare estimates. Materials and methods An angler chooses a fishing site from the set of all alter- native sites if the utility of visiting that site is greater than the utility of visiting any other site in the global choice set. Denoting the set of all alternatives faced by any angler by S = 11, . . . , A^^l as the choice set, the indi- rect utility of visiting site 7 is where C/ = an individuals utility; V = the deterministic portion of utility; y = income; p = the cost of angling at site j; q = a. vector of characteristics of sitey; and £ = the unobservable portion of indirect utility. In the RUM framework, an angler will choose site j from S if V,(9,,y-p,) + £,>n(g„y-p,) + £„7eS,VAeS, (2) where the indirect utility of visiting site j is greater than the indirect utility of visiting site k for all k in the global choice set, S. The random portion of the random utility model stems from the unobservable portion of indirect utility, cap- tured here in the error term f^. If this error term is assumed to be distributed in a type-I extreme value dis- tribution, the above site choice framework can be mod- eled with the conditional logit model. Maddala (1983) has provided a complete derivation of the conditional logit model. Within this framework, the probability that i visits site 7 is given by P,(j)^P(j\jeS) ,v/i,-y-pj> s... Vj,( V^( (5) Because the goal of the present study is to examine the sensitivity of welfare estimates of a quality change to the specification of choice sets, it is necessary to show how the choice set enters the calculation of compensating variation (CV), or the level of income required to keep the angler at the same level of expected utility after the quality change. The following expression for CV is taken from the work of Bockstael et al. (1991), who examined the value of quality improvements in the demand for recreation, where j3^, is the travel cost parameter. Uj(.qj,y-Pj,£j) = Vj(qj,y-Pj) + ej, (1) CV ln(l..,^'"^-1-l"l"(l..,^"""" (6) Centner: Sensitivity of angler benefit estimates to tfie definition of substitute sites considered by tfie angler 163 The summation of the indirect utilities is across the choice set facing each individual, S,, and not the global choice set, S. Since 1979, data have been collected on marine recre- ational angling during the Marine Recreational Fishery Statistics Survey (MRFSS). The MRFSS consists of two independent but complementary surveys: a field survey and a telephone survey, conducted annually in six two-month "waves." The field survey is an intercept survey of anglers conducted at fishing access sites and is designed to obtain a random sample of recreational trips for computing catch per unit of effort. Fish re- tained by interviewed anglers are sampled for length and weight. Fish not retained by the angler are not observed, but count data on this unobserved catch are collected. The data on harvest provide a picture of the size distribution of the kept fish from the stock. If a fishery is regulated by a minimum size limit, a catch- and-keep rate calculated from these data indicates the catchability of fish large enough to keep. As such, it is the observed rate at which anglers can catch and keep fish from a stock. The intercept sample is stratified by state, wave, fish- ing mode, fishing area, catch type, and species. Specific data elements collected during the intercept survey include state, county, and zip code of angler's residence, hours fished, primary area fished, target species, gear used, and days fished in the last two and 12 months. During the intercept portion of the survey, data are collected on the length and weight of all fish species retained by the angler and the species and condition of all catch not retained by the angler. Upon completion of the base MRFSS, anglers in the Northeast (NE) (Maine, New Hampshire, Massachusetts, Rhode Island, Con- necticut, New York, New Jersey, Delaware, Maryland, and Virginia) were asked to complete a short add-on questionnaire in 2000. This questionnaire provided information on whether or not the trip was a single-day or longer trip and, if it was a multiple-day trip, whether fishing was the primary purpose of the trip. Data were also collected on the angler's saltwater fishing experi- ence (in number of years), boat ownership (whether owned or not), and whether or not the individual took time off without pay to take the fishing trip. If the in- dividuals responded in the affirmative to the later, they were asked the number of hours in their work week and their personal income. The survey instrument is avail- able at the NMFS web site (NMFS^). In order to reduce the complexity of the modeling ef- fort, the angler's choice to fish rather than participate in some other recreational activity, the angler's choice to fish in a private or rental boat mode, and the angler's decision regarding a species target are exogenous to the model. Because the area fished is not documented ^ NMFS (National Marine Fisheries Service). 2006. Fisher- ies Statistics and Economics Division. Survey Instruments. Website: http://www.st.nmfs.gov/stl/econ/surveys/survey_ timeline.html (accessed on 13 August 2006). in the MRFSS, a fishing site is defined as the point of fishing access. As mentioned previously, the treatment of all substitute sites can be quite costly from a data standpoint for a number of reasons. Because thousands of individual sites in the North East (NE) region are recognized in the MRFSS, estimation can be a lengthy process, particularly with nested models. In addition, not all species are sampled in all survey waves at all sites in all modes; therefore the calculation of historic catch rates at the individual site level results in many empty cells. To speed estimation and to fill some of these empty cells, all sites within a coastal county were aggregated into one site that represented that county. Across the NE, there are roughly 63 coastal counties, and therefore 63 sites. In order to examine whether this aggregation strategy induces any bias into the estima- tion of the conditional logit model, a variable im) was created that represents the number of MRFSS sites aggregated into each new site. The rule that a county equals a site was not strictly followed in all cases. Some geographically diverse counties (i.e., those counties with both ocean frontage and bay frontage) were separated into two sites because of the different opportunities provided by these different types of water. Both the historic five-year average catch rate (catch rate) and catch-and-keep rate (KRATE) were calculated for the boat mode for each wave and site combination. KRATE measures the catchability of a striped bass large enough to keep, incorporating the five-year aver- age probability of catching a striped bass large enough to keep. The distinction between the catch rate and KRATE is particularly important for striped bass be- cause this species is heavily regulated. Historic KRATE was used in the model because it represents the portion of the catch that an angler would be able to keep, not just the increase in overall catch. It is also the mea- sure of angler quality used in the Whitehead and Haab (1999) study. Even after the site aggregation, some counties did not contain enough data points on striped bass catch from the boat mode over the 5-year period. Whitehead and Haab (1999) replaced missing catch rates using the catch rate from the nearest neighboring site in some cases and with zero values, in other cases depending more or less on mode. Hicks et al. (1999) recognized this approach to be ad hoc and estimated his model using both nearest neighbor and zero value assignment, another ad hoc approach, and found that the treatment of missing values did not significantly affect the welfare estimates. He concluded that the zero assignment is perhaps less arbitrary because the empty cells actually convey information. That is, if there are no observations of average catch within a particular wave-Hmode+site-i-species combination, the site is not very productive over that combination. As a result, zero assignment requires less judgment by the researcher; therefore that is the approach used here. Estimating any demand equation requires a price variable. Because recreational fishing experiences are not openly traded in markets, travel cost (both the ac- tual cost of travel plus the opportunity cost of time) is 164 Fishery Bulletin 105(2) used as the price. Round-trip travel cost ittc) is calcu- lated as the following; ttc = ($0.33 X distance x 2) + ( [^^ffS'H^E^^ + ^rsf \ 40 xlost _ income x w. (7) where distance = the one-way distance from the anglers home zip code to the zip code, or lati- tude/longitude, of the intercept site. This distance is multiplied by the Federal Travel Reg- ulations reimbursement rate for private transportation ($0.33) and includes both the fixed and variable costs of operating an automobile. The variable lost_income is a dummy that takes the value of 1.0 if the individual did take time off work without pay to go fishing. If the individual lost income, their wage rate («>) is multiplied by the travel time plus the time on site (hrsf) and this amount is added to the travel cost (40 miles per hour is used as the average travel speed). Therefore, the opportunity cost of onsite time and travel time is only included if an individual took time off work to par- ticipate in fishing on a given day. If the individual is not losing income for the trip, his travel cost is simply round-trip distance multiplied by the fixed and vari- able costs of operating an automobile. As is typical for these MRFSS data sets, very few anglers (3.34%) report having foregone income to take the trip. To account for the opportunity cost of time for those anglers not los- ing income, travel time is used as a measure of time cost for those individuals. In order not to double count those that lost income taking a trip, the expression for travel time itt) is tt = ( (distance X 2) 40 )- lost _ income) . (8) Keeping only those anglers that have targeted or caught striped bass from the boat mode on a single day trip leaves 3630 usable observations. With an aggregation strategy in place and the vari- ables defined, the estimation of the conditional logit model follows. As a reminder, the choice of whether or not to take a fishing trip, which mode to fish in, and what species to pursue are made outside of this model. The angler then chooses the site that maximizes in- direct utility from his or her set of substitutes. Every model carries a set of implicit assumptions. Angler behavior within this model is defined on a trip-by-trip basis and the angler is not allowed to modify the num- ber of trips taken each season. Therefore, each choice is independent of the next, and unobservable utility, f^, is independent of any other trips. Additionally, the MRFSS intercept survey is assumed to approximate a random sample of trips. The author acknowledges these contentions with choice-based sampling in the MRFSS data, and this is an area of research that this author and NMFS scientists continue to explore. Variables in the deterministic portion of indirect util- ity include travel cost ittc), travel time itt), log of the number of MRFSS intercept sites aggregated into the county site used in the model (m), and historic KRATE per trip for striped bass at site 7 (q ). Indirect utility is I5^ttc,j + p^q, + p„tt,, + p,„ ln( m^ ) + E^. (9) With this expression for indirect utility, the probability that angler i selects site 7 is P.ij)-- (10) and the expression for the change in compensating varia- tion for a change in the historic catch and keep rate, after assuming a constant marginal utility of income, is the following: (11) where q° = the historic KRATE; and q^ = the KRATE after the environmental or policy change. Table 1 provides descriptive statistics for all the vari- ables to be used in the analysis and some angler-specific attributes in order to give the reader some background on these anglers. Throughout the range of this data col- lection, the bag limit for striped bass is two fish per day. On average, anglers catch far less than the limit. In fact, the base catch rate for anglers targeting or catch- ing striped bass from the boat mode is less than one fish per trip. What is readily apparent is that there are some irrational anglers in this group, at least concern- ing travel time. The maximum travel time translates into a 798 mile one-way travel distance, which does not seem feasible for a one-day trip. Even after eliminating those anglers that admit to taking an "overnight" trip, there are obviously anglers that are away from home longer than 24 hours. One explanation is that these anglers live in the local area seasonally and have given the zip code of their permanent address, which is used to calculate travel distance. Another explanation arising from the author's experience in the field is that some of these anglers drive incredibly long distances and fish for 24 or more hours. They do not consider their trip to be an overnight trip because they are not staying in a hotel even though their round trip travel distance indicates that they were away from home for more than 24 hours. There were only 3 individuals in the data set with one-way travel distances greater than 500 miles and the results were not sensitive to leaving these out- liers in the model. As a result they remain in the data set. Other statistics of note include the variable that Gentner: Sensitivity of angler benefit estimates to the definition of substitute sites considered by tfie angler 165 Table 1 Descriptive statistics for selected variables describing angler and trip characteristics from the on ly data set used m this study from the 2000 Marine Recreational Fisheries Statistical Survey economic add-on survey conducted in Maine New Hampshire, Massachusetts, Rhode Island, Connecticut, New York, New Jersey, Del aware, Maryland, and Virg nia. Variable Mean Standard deviation Minimum Maximum Used in model Travel cost (^/c) ( US $) 30.39 38.36 0.0 778.73 Travel time itt) (hours) 1.99 2.15 0.0 39.93 Aggregation variable im) (number of sites) 43.85 37.88 5.0 138.00 Catch-and-keep rate (q) 0.19 0.40 0.0 2.00 One-way travel distance (distance) (milesOmiles) 41.54 43.13 0.1 798 Not used in model Boat ownership (%) 0.73 0.44 0.0 1.00 Hours fished 4.20 1.88 0.5 22.00 Years fished 23.61 14.76 0.0 70.00 Catch rate (no. offish) 0.89 1.64 0.0 10.67 looks at the aggregation of sites. On average there are almost 44 MRFSS intercept sites aggregated into the definition of a "site" used in the present study and a maximum of 138 sites and a minimum of five sites in a county. On average, anglers spend 4.2 hours on the wa- ter and 73% own the boat they are fishing from. Finally, this is a fairly experienced group, with an average of almost 24 years of saltwater fishing experience. The final portion of this analysis yet to be discussed is the definition of distance-based choice sets. Haab and Hicks (1997) and Parsons and Hauber (1998) found that welfare estimates change little beyond a certain threshold. Up to a point, limiting an angler's choice set by using a distance-based metric only increases the realism of the choice that anglers consider in reality. Hicks et al. (1999) used such a designation in their analyses. Both included all sites within 150 miles, if the angler lived within 30 miles of the site selected, and all sites within 400 miles otherwise. The extreme rational limit for a one-day trip is probably 400 miles one-way. That distance translates into a 10 hour or 6 hour and 40 minute one-way travel time at 40 and 60 miles per hour, respectively. Whitehead and Haab (1999) used definitions of distance-based choice sets that ranged from 180 miles one-way to 360 miles one-way (3-6 hours one-way at 60 miles per hour), realizing that the 360 mile cut-off is likely not very realistic. These defini- tions probably drive the small difference in parameters across specifications, because eliminating sites outside of what anglers are really considering should have little effect (Whitehead and Haab, 1999). To examine the definition of choice sets, this study examined much smaller cut-offs than those found in the previous literature that focused on saltwater angling. The cut-offs included the following: a full unrestricted choice set, and 300-, 250-, 200-, 150-, and 100-mile dis- tance cut-offs. Initially, a 50-mile one-way distance cut- off was included, but, because of the site aggregation strategy used, the only substitutes left in the choice set at the 50-mile cut-off were closer than the site that was chosen, for most anglers. If it is possible to estimate the model on an individual site basis, it would be possible to run smaller distance cutoffs without encountering this problem. In order not to also drop observations when applying these cut-offs, if an individual was observed to make a choice outside of the cut-off, that observation is retained. If all substitute sites for that individual are also outside of the cut-off, the next nearest site is included in that angler's choice set. Therefore, anglers have at least one substitute left in their choice set, no matter how restrictive the cut-off becomes. The aver- age number of sites in each choice set is 63, 37.2, 30.5, 23.3, 16.9, and 10.6 for the six choice sets, from least restrictive to most restrictive, respectively. In percent- age terms, these restrictions on the choice set eliminate between 409f and 937c of the available sites. Estimation of the confidence intervals around these welfare estimates is calculated by taking 1000 random draws from a multivariate normal distribution parame- terized by the vector of estimated parameters and their covariance matrix. Sorting these draws from highest to lowest and removing the upper and lower 2.5% and 59c, respectively, construct 95% and 90% confidence intervals (Krinsky and Robb, 1986). Results Table 2 contains the parameter estimates and standard errors of the six conditional logit models. All six models strongly reject the hypothesis that the coefficients are simultaneously equal to zero. Also, all coefficients in all models are statistically significant at the 99% level or better. In general, anglers prefer sites that are closer to home, both in terms of the cost of driving and the time cost (travel time multiplied by the individual's 166 Fishery Bulletin 105(2) Table 2 Parameter estimates for the six distance-based conditional logit models (standard errors in parantheses . Variable Choice set Full 300-mile 250-mile 200-mile 150-mile 100-mile Travel cost («c) (US$) 0.06242 -(0.00573) -0.06240 (0.00573) -0.06232 (0.00575) -0.06176 (0.00582) -0.05827 (0.00606) -0.04150 (0.00682) Travel time («) (hours) -0.23226 (0.07753) -0.23154 (0.07759) -0.23216 (0.07782) -0.23771 (0.07873) -0.27057 (0.08186) -0.36472 (0.09201) Log of aggregation variable (m) (no. of sites) 0.68125 (0.02482) 0.68220 (0.02483) 0.68262 (0.02484) 0.68293 (0.02484) 0.68187 (0.02484) 0.67143 (0.02479) Catch-and-keep rate iq) (no. offish) 0.67653 (0.03792) 0.67535 (0.03794) 0.67584 (0.03795) 0.67506 (0.03795) 0.66885 (0.03794) 0.65325 (0.03820) LR' 17,245.31 13,192.55 11,695.28 9695.16 7350.95 3891.39 R-' 0.573 0.507 0.523 0.430 0.364 0.236 ' Value of the likelihood ratio (LR) statistic testing the hypothesis that all betas = 0. ' McFadden's psuedo R^. wage rate). Additionally, anglers prefer sites that offer the possibility of catching and keeping more striped bass. Finally, with regard to the number of MRFSS sites aggregated into a site as defined in this study, anglers preferred to visit counties that contain more sites. Table 3 gives the mean CV for a one-fish increase in the catch-and-keep rate. A one-fish increase in the catch-and-keep rate is equivalent to the net benefits of an improvement in angling quality large enough to increase the keep rate by one fish or a regulation that allows increasing keep rates. A quality change significant enough to change KRATE by one fish would be unrealistic in the short term, considering the striped bass stock size distribution inferred from the catch rate and KRATE estimates (Table 1). Because KRATE incorporates the five-year average probability of catch- ing a striped bass large enough to keep, this one-fish increase in KRATE models an angler's willingness to pay for a one-fish increase in the bag limit or an angler's willingness to pay for a special license allowing the retention of one striped bass more than the current two-fish limit. This result supports Parson and Hauber's (1998) and Whitehead and Haab's (1999) results that there is indeed little difference in the definition of choice sets with the use of a distance metric. To examine the significance of the difference in welfare estimates, and not just the magnitude, 95% confidence intervals were calculated around each welfare measure (Krinsky and Robb, 19861. In fact, the mean of the smallest choice set is almost entirely contained within the 95% confidence interval of the next smallest choice set, and the entire lower bound for the smallest choice set is contained in the next smallest choice set. This is demonstrated graphically in Figure 1. From Figure 1, however, it Table 3 Mean increase in angler benefits, measured by compen- sating variation (with 95% and 90% confidence intervals [CIs], for a one-fish increase in he catch-and-keep rate by | distance-based choice set). 95% CI 90% CI Mean Choice set increase Upper Lower Upper Lower Full $10.84 $13.56 $8.89 $13.05 $9.15 300-mile $10.82 $13.57 $8.86 $13.04 $9.13 250-mile $10.84 $13.61 $8.90 $13.29 $9.18 200-mile $10.93 $13.79 $8.95 $13.30 $9.23 150-mile $11.48 $14.73 $9.28 $14.22 $9.57 100-mile $15.74 $23.57 $11.65 $22.05 $12.23 appears that as the choice sets are truncated past the 150-mile threshold, welfare estimates rise — a similar result to that of Parson and Hauber (1998). Unfortu- nately, the aggregation strategy necessary when using the MRFSS data precludes an examination of a dis- tance-based cut-off as small as that used in Parsons and Hauber's study (1998). Conclusions In general, as choice sets are restricted, the coefficient on cost goes up, its absolute value goes down, and its stan- dard error goes up, but only slightly, until the point is reached where the aggregation strategy begins to impose an artificial restriction on the choice set with this data Centner: Sensitivity of angler benefit estimates to the definition of substitute sites considered by tfie angler 167 E o O set. It would be interesting to examine the effect of the aggregation strategy by using the individual MRFSS intercept sites. It is possible that using average catch-and-keep rates calculated over a longer time series would result in far fewer empty cells, which are the main hurdle to using the individual MRFSS sites. Mean one-way travel distance in the data set is 41.5 miles (Table 1). All of the choice sets at and above the 150 mile cut-off have an almost equal proportion of sites in the choice set and sites chosen at the cut-off point. That is, the percentage of substitutes within the cut-off and the percentage of sites chosen within the cut-off are equal (near QO'/f for both) for the full, 300-, 250-, and 200-mile choice sets. At the 150 mile cut-off this equality begins to fail and the percentage of chosen sites inside the cut-off fall to 98% and 94% for the 150- and 100-mile cut-offs, respectively. This fall is being driven partially by the aggregation strategy. Although this result has not been examined by the author, it is likely that the average distance for an angler to travel outside his county of residence is somewhere between 100 and 150 miles. Again, if the historic catch rate could be calculated to examine individual MRFSS sites, this aggregation restriction could be examined to determine the overall sensitivity of welfare estimates to the designation of distance-based choice sets. In conclusion, when estimating the net benefits of quality changes for recreational anglers with the MRFSS data, it matters little how restrictive the choice sets become with a distance metric, as long as the re- searcher does not ask more of the aggregation strategy than it can provide. This result quantifies the signifi- cance of the difference in welfare estimates across ag- gregation strategies and indicates the strengths and the weaknesses of a nationwide data set on marine angling in estimating net benefits and thus makes policy analy- sis quicker and easier. Literature cited Bockstael, N. E., K. E. McConnell, and I. E. Strand. 1991. Recreation. Iii Measuring the demand for envi- ronmental quality (J. Braden, and C. Kolstad, eds.), 227-268 p. North-Holland Pubis., Amsterdam, The Netherlands. Gentner, B. J., S. Steinback, and M. Price. 2001. Marine angler expenditures in the Pacific coast region, 2000. NOAA Tech. Memo. NMFS-F/SPO-49, 66 p. Haab, T., and R. Hicks. 1997. Accounting for choice set endogeneity in random $25.00n I $20. DO- CT $15.00- § $10.00- S5.00 $0.00 \ \ \ Unrestricted 300 Mile -250 Mile A 200 Mile XI 50 Mile • 100 Mile Cfioice sets Figure 1 Mean benefits as measured by compensating variation (with 9.5 '?( confidence intervals) for a one-fish increase in the catch-and-keep rate by distance-based choice set. utility models of site choice. J. Environ. Econ. Manage. 34:127-47. Haab, T, and R. Hicks. 1999. Choice set considerations in models of recreation demand. Mar. Res. Econ. 14:271-281. Hicks, R., A. B. Gautam, D. VanVoorhees, M. Osborn, and B. Gentner. 1999. Thalassorama: an introduction to the NMFS marine recreational fisheries statistical survey with an emphasis on economic valuation. Mar. Res. Econ. 14:375-385. Hicks, R., and I. E. Strand. 2000. The extent of information: Its relevance for random utility models. Land Econ. 76:374-385. Jones, C. A., and F. Lupi. 1999. The effects of modeling substitute activities on recreational benefit estimates. Mar. Res. Econ. 14:357-374. Krinsky, I., and A. L. Robb. 1986. On approximating the statistical properties of elasticities. Rev. Econ. Stats. 68:715-719 Maddala, G. S. 1983. Limited dependent and qualitative variables in econometrics, 401 p. Economic Soc. Monographs. Cam- bridge Univ. Press, New York, NY. Parsons, G. R., D. M. Massey, and T. Tomasi. 1999. Familiar and favorite sites in a random utility model of beach recreation. Mar. Res. Econ. 14:299-315. Parsons, G. R., and A. B. Hauber 1998. Spatial boundaries and choice set definition in a random utility model of recreation demand. Land Econ. 74:32-48. Steinback, S., B. Gentner, and J. Castle. 2004. Economic impacts of marine recreational angling in the United States. U.S. Dep. Commer., NOAA Prof. Pap. NMFS 2, 169 p. Whitehead, J. C, and T C. Haab. 1999. Distance and catch based choice sets. Mar. Res. Econ. 14:283-299. 168 Abstract — Fish-habitat associations were examined at three spatial scales in Monterey Bay, California, to deter- mine how benthic habitats and land- scape configuration have structured deepwater demersal fish assemblages. Fish counts and habitat variables were quantified by using observer and video data collected from a submersible. Fish responded to benthic habitats at scales ranging from em's to km's. At broad-scales (km's), habitat strata classified from acoustic maps were a strong predictor of fish assemblage composition. At intermediate-scales (m's-100 m's), fish species were asso- ciated with specific substratum patch types. At fine-scales (<1 m), micro- habitat associations revealed differing degrees of microhabitat specificity, and for some species revealed niche separation within patches. The use of habitat characteristics in ecosystem- based management, particularly as a surrogate for species distributions, will depend on resolving fish-habitat associations and habitat complexity over multiple scales. Multiscale habitat associations of deepwater demersal fishes off central California Tara J. Anderson (contact author) Mary M. Yoklavich Email address lorT. J. Anderson: t.anderson@alms.gov.au Fisheries Ecology Division Southwest Fisheries Science Center National Marine Fisheries Service 110 Shaffer Road, Santa Cruz, California 95060 Present address (for T J Anderson); Australian Institute of Marine Science PMB 3 Townsville MC Queensland 4810, Australia Manuscript submitted: 16 June 2006 to the Scientific Editor. Manuscript approved for publication 17 August 2006 by the Scientific Editor. Fish. Bull. 105:168-179 (2007). Measuring fish-habitat associations on a number of spatial scales is essential in determining the relative importance of habitat types and landscape configu- ration in structuring fish assemblages and populations. Many benthic habi- tat characteristics (e.g., substratum type, depth, relief) are important in explaining the local distribution and abundance patterns of demersal fishes (e.g., Jones and Syms 1998; Stephens et al., 2006). An organism's use of habitat may also change as a function of scale (Wiens, 1989). For example, fishes may make a considerable range of choices about their occupancy of specific habitats and may sample their environment at a range of spatial and temporal scales (Ault and Johnson, 1998; Syms and Jones, 1999). Habitat types (abiotic and biotic), however, are found within a large spatial domain (landscape) in which the configuration and connectivity between neighbor- ing habitat areas may contribute to population structure (Forman, 1995). For example, landscape configura- tion and the degree of habitat patchi- ness may modify the distribution and movement of an organism, and the interactions among species (Addicott et al., 1987). On the U. S. West Coast, demersal fishes, particularly rockfishes iSe- hasfes) are a dominant feature of the benthic ecosystem (Love and Yoklav- ich, 2006) and are important for both commercial and recreational fisheries (Love, 2006). At broad spatial scales, traditional trawl surveys have docu- mented a range of biogeographical and depth patterns for harvested de- mersal species (Gunderson and Sam- ple, 1980; Weinberg, 1994; Williams and Ralston, 2002). Less research has been done on the role that benthic habitat variables, such as substra- tum type and relief, play in explain- ing the distribution and abundance of either commercial or noncommercial species. Strong relationships between demersal fish species and a range of habitat characteristics, particularly substratum type and abundance of giant kelp, have been identified in shallow (<30 m) coastal waters (Ste- phens et al., 2006). However, many demersal species in this system, par- ticularly rockfish species, are found over extensive depth ranges beyond those that can safely be investigated with SCUBA (Love et al., 2002). The use of submersibles and remotely op- erated vehicles (ROVs), with sampling protocols similar to those of nearshore surveys (Stein et al., 1992; Adams et al., 1995; Yoklavich et al., 2000), provides the capabilities to make quantitative in situ observations of fish-habitat associations in deepwater (>30 m). Studies in which these tools are employed are also beginning to demonstrate characteristic habitat as- sociations for deepwater demersal fish species (Love and Yoklavich, 2006). The importance of spatial scale in un- Anderson and Yoklavich: Habitat association of deepwater demersal fishes off central California 169 1000m 500ni 200m 100m 4 km > 137 l^oltalian ' Ledge 3123 3121: M20 '"-" "^/'' W Pt Pinos m Reef-.. ,/ "^■f Portuguese ^ Ledge :tt :^ /-^ San'grancisco m\ I ^^ \~^^^ \ Monterey) California Bay ^ . Santa Cruz National — if Manne ^Monterey Sanctuary ^^ \ I21^57'46"W Figure 1 Seafloor map of the continental shelf in southern Monterey Bay, central California, depicting the three acoustically derived broad-scale strata and Delta submersible sampling locations (dive num- bers 3120-3141). Hard substratum (i.e., complex outcrops) is depicted as dark gray areas. Mixed substratum (areas of hard mixed with soft) is depicted as medium gray areas. Soft substratum (i.e., areas of contiguous soft sediments) is depicted as light gray areas. White areas were not surveyed; box = 10x12 km study area. Bottom insert is an example of the observed intermedi- ate-scale substratum types recorded within the three transects of dive 3121 (depicted by the three rectangles Tl=transect 1, T2 = transect 2, and T3=transect 3) in relation to the seafloor map of that area. Transects sampled in hard substratum (Tl and T2) were heterogeneous and were composed of mixed patches dominated by rock (dark gray), boulders (diagonal hatching) and cobbles (light gray checks). In contrast, transects run within soft substratum (T3) were more homogeneous, composed of either sand (white) or mud (light gray). derstanding these associations, however, has received much less attention (Langton et al., 1995). In this study, we examined the relationship between deepwater demersal fishes and benthic habitat vari- ables at three spatial scales in an area encompassing a proposed marine protected area (MPA) in southern Monterey Bay, California. At the broad spatial scale of km's, habitat strata were identified from acoustic seafloor maps. Within these strata we conducted sub- mersible transects and recorded both benthic habitat variables — such as substratum type, depth, relief, and habitat patchiness — and fish abundance and size. At the intermediate scale of 10-100's of meters, within-tran- sect habitat measures, in combination with fish counts, provided measures of habitat patchiness and fish use of these patches. Finally, we assessed fine-scale or micro- habitat (<1 m) fish-habitat associations by recording the habitat type located directly beneath each fish. These multiple spatial scales of habitat association were in- tegrated to examine multiscale habitat and landscape requirements of these species. Material and methods Survey of fish habitat To quantitatively sample demersal fishes and benthic habitats on the continental shelf in southern Monterey Bay (36°E, 121°S) (Fig. 1), in situ counts and habitat characterizations were made from the two-person Delta submersible. The submersible survey was conducted 170 Fishery Bulletin 105(2) from 9 through 12 October 1993 (boreal fall), between the hours of 07:30 (1 hour after sunrise) and 17:00 (1 hour before sunset). Thirty-three strip-transects (2 m wide by 10 minutes in duration) were surveyed within a 10 X 12 km study area (Fig. 1). During each transect, the scientific observer made observations from the central starboard porthole while the pilot drove the submersible about 1 m above the seafloor at a speed of 0.4-0.9 knots depending on currents and topography. Three broad- scale strata (hard, mixed, soft substratum), which had been identified from seafloor maps by using geophysical data (Eittreim et al., 2002; Anderson et al., 2005), were sampled at depths ranging from 72 to 252 m. Within each 10-minute transect, all demersal fishes within 2 m of the submersible were identified to the lowest taxon, measured (total length was visually es- timated to 5-cm size classes), and counted vocally by the scientific observer. An external starboard mounted Hi-8 video camera simultaneously recorded the seafloor along each transect and the scientific observers' vocal commentary on the audio track. A hand-held sonar gun was used to gauge transect width, and paired lasers, set 20 cm apart and projected into the observers' field of view, were used to gauge fish size. A Pisces Video Plus 11 data-logger {Pisces Design, San Diego, Califor- nia) superimposed time, date, depth, and altitude of the submersible onto the video image. Final fish sizes and counts were derived from the videotape, by using the audio commentary as supporting information. All video analyses were conducted by the same person to reduce between-observer variability. Individual fish that could not be distinguished to species were as- signed to a taxonomic group, for example: to subgenus (e.g., young-of-year Sebastes spp. [YOY], Sebastomus spp. [rosy-like rockfish species]), genera (e.g., Cithar- ichthys spp. [sanddabs], Zaniolepis spp. [combfishes]), family (e.g., Agonidae [poachers], Cottidae [sculpins]) or order (e.g., Pleuronectiformes [flatfishes]). Benthic habitat characteristics within each transect (intermediate scale) were categorized and delineated from the videotape. Substratum composition (rocks, boulders [>25.5 cm], cobbles [6.5-25.5 cm], sand, and mud) within a patch was categorized by using the dominant (primary=>50'%) and subdominant (second- ary=>20%) percentages of substratum cover used by Stein et al. (1992) and Yoklavich et al. (2000). For example, a patch comprising >50% rock and >20% boulders was classified as rock-boulder (RB); a patch comprising >70% rock was classified as rock-rock (RR). Patches were delineated from videotape where patch duration exceeded 3 seconds of elapsed video time (i.e., where patch size >1.7 m). Habitat relief within each patch was categorized as flat (0-5°), low (5-30°), or high (>30°). These methods adequately defined interme- diate scale habitat composition and patchiness within transects (i.e., m's-100's m), yet logistically enabled long transects (max. 585 m) to be quantified. To de- scribe fine-scale (<1 m) microhabitat use by demersal fish species, we recorded the type of substratum (rock, boulders, cobbles, sand, or mud) directly beneath each fish. This multiscale approach enabled habitat asso- ciations at each scale to be recorded independently of associations at other scales. Transect length, independent of submersible speed, was estimated by using the known distance between the lasers (i.e., 20 cm) as a ruler, by counting the number of lengths that occurred sequentially over a 15-s duration within each minute of videotape, and then multiplying by transect duration (i.e., 10 min). Patch lengths were calculated by using the same method but were multi- plied by patch duration (elapsed time per patch). Analysis The categorical measures of substratum type were recoded as semiquantitative variables. Primary and secondary categories were recoded so that each sub- stratum type within a patch was given a percent cover value of 0%, 20%, 50%, or 70%. For example, rock-rock (RR) was recoded as 70%- rock (50%-i-20%) while all other substratum types scored a value of 0%; similarly boul- der-cobble (BC) was recoded as 50% boulder, 20% cobble and all other types scored a value of 0%. Habitat relief was recategorized as an ordinal variable with values of 1, 2, and 3 that corresponded with flat, low, and high relief. The mean and standard error for substratum types and relief, and median depth were then calculated for each transect (broad-scale) and patch (intermedi- ate-scale). Habitat patchiness at the broad-scale was represented by "patch number" — the number of patches within each transect, and "patch size" — calculated as the \ogipatch length) within each transect. Benthic habitat variables, with the exception of patch number and patch size, were .r"^ transformed to improve data normality and linearity between variables. Principal components analysis (PCA) was run on the correlation matrix of the transformed transect-level data to evaluate the validity of the broad-scale strata classifications and to describe the relationship between benthic habitat variables over broad spatial scales. To examine the relationship between fish and habi- tat, total abundance and species richness were calcu- lated for all fish species and rockfish species at both transect (transect length x 2 m width) and patch (patch length X 2 m) scales: fish densities were then expressed as numbers per 1000 m- (transects), and 200 m^ (patch- es). To examine the fish assemblage in relation to har- vest potential, we classified species as either commer- cial (e.g., Sebastes paucispinis [bocaccio], S. ruberrimus [yelloweye rockfish], S. flavidus [yellowtail rockfish], Ophiodon elongatus [lingcod], and Microstomus pacificus [Dover sole]) or noncommercial (e.g., S. wilsoni [pygmy rockfish], Rhinogobiops nicholsii [blackeyed goby], and Zaniolepis spp.). We also categorized fishes as small (s20 cm) or large (>20 cm). Individual species and taxon groups were included in analyses when they were pres- ent in more than 5% of all patches. Consequently, 21 taxa (15 species and six groups) from nine families were retained for analyses. The data on fish distributions were examined by using histograms and Taylor power Anderson and Yoklavich Habitat association of deepwater demersal fishes off central California 171 plots (i.e., \og(variance) versus \og{mean)). Data were generally right-skewed and had a positive variance- mean relationship. The slope of the Taylor power plot was used to optimally decouple variance from mean by raising the data to the power of ((2-slope)/2) (McArdle et al., 1990). Consequently, species abundance data were ^y■o ir.^ transformed, total abundance was transformed by loglO(.r-i-l), and a square root (.v" ''I transformation was applied to species richness. To examine broad-scale relationships between fish species and benthic habitat variables, we ran a canonical correlation analysis on the transect-level data matrix and then plotted the total structure coefficients of the fish in habitat space. The standardized redundancy output values of the model were used to measure the amount of variation for both fish species and benthic habitat variables. To examine intermediate-scale relationships between fish species and benthic habitat variables, densities of fishes per patch types were examined. However, be- cause all patch types were not equally available, we also standardized patch-use relative to habitat avail- ability (patch selectivity) by subtracting proportional occurrence of each patch type from the proportional abundance for each species. Here, a positive association with a patch type revealed that more individuals were found in that patch type than would be expected given random habitat use (i.e., no selectivity). Conversely, a negative association revealed that fewer individuals were found in that patch type than would be expected by random habitat use. Finally, because microhabitat availability was not measured independently of fish presence, microhabitat use by fishes was restricted to graphical presentation. Results Seafloor composition We sampled 11.15 linear km of seafloor within the 12 x 10 km survey region, using submersible strip-transect methods. At broad-scales, benthic habitat variables were grouped a posteriori in order to reliably distinguish hard, mixed, and soft strata (Fig. 2). Hard stratum comprised patchy "high-relief outcrops" of rock, boul- ders, and sand. In contrast, mixed stratum comprised "low-relief outcrops" of cobbles and mud. Soft stratum comprised "homogeneous mud." The three broad-scale habitat strata also varied in their depth distribution, and strata and depth were strongly collinear. High-relief outcrops were generally shallower (60-100 m) than low- relief outcrops (90-150 m), and although homogeneous mud occurred in most depth ranges, it was the only stratum surveyed in deep offshore locations (80-260 m). Benthic habitat variables within each of the three strata were also strongly collinear. For example, rock always co-occurred with boulders and sand, forming complex high-relief outcrops in shallower water (i.e., <100 m). Therefore, if a species was correlated at broad spatial scales with high-relief outcrops, differentiating the rela- A Bentfiic variables Patch size 0.4- Sand Boulders 0.0- nppth — ' Rnlirf Cobbles , Rock Mud -0.4- 5. Patch I number 8 c -0,0 g -0.8 -0.4 0.0 0.4 0. Q. E ^ B strata Q. 5 o c i 4 o H Hard stratum A Mixed stratum 3 ' 2 - 1 - - -1 - -2 - A o O A ^ O Soft stratum Ji m^ " %< ""■■ -3- A -4- -5-4-3-2-10 1 2 3 4 5 Principal component 1 (45%) Figure 2 Principal components analysis of broad-scale benthic habitat characteristics: (A) correlation matrix of ben- thic variables; (B) projection of strata categories in ordination space. tive importance of substratum composition, depth, or some corequisite would be problematic. Variability in intermediate-scale habitat also was dis- cernible (Fig. 3). Five substrata (rock, boulders, cobbles, sand, and mud) were recorded during this survey, which at intermediate scales were present in 21 of 25 possible paired "substratum patch types" fall but mud-sand, cobble-sand, sand-cobble, or sand-mud patches types were recorded). However, the proportional availabil- ity of these patch types differed between strata. For example, hard strata contained the highest number 172 Fishery Bulletin 105(2) of substratum types (n = 19), where rock and boulders types were the most abundant (Fig. 3A). Mixed strata also contained a variety of patch types («=10), but were devoid of rock and contained higher proportions of mud (Fig. 3B). Soft strata contained the fewest patch types (n = 3), composed primarily of homogeneous mud, and small amounts of homogeneous sand and mud-cobble patch types (Fig. 3C). Structure of fish assemblages and broad-scale fish-habitat associations Sixty-two species of demersal fishes (from 21 fami- lies) totalling 21,184 fishes were recorded during this survey. Rockfishes were the most abundant portion of the demersal fish assemblage, representing 93% of all fish sampled (i.e., 24 rockfish species, totalling 19,668 rock- fishes). Most fishes recorded (96%) were small (TL s20 cm) noncommercial species, dominated by small-bodied 1" Substratum II Rock 1; ;:■ ;l Boulders 1 1 Cobbles ^M Sand ^m Mud OC CQ O "5 S CC IT DC tr q; Substratum patch type Figure 3 Intermediate-scale habitat characteristics: substratum patch composition within the three broad-scale habitat strata: (A) hard stratum, (B) mixed stratum, and (C) soft stratum. Substratum types recorded were R = rock, B = boulders, C = cobbles, S = sand, and M=mud. The first and second letters of each patch type (e.g., RR, RB, RC, to MM) represent primary (50%) and secondary (20%) substratum types, respectively. (dwarf) rockfishes, such as S. wilsoni (n = 5857, 28% of all fish sampled), S. semicinctus (halfbanded rockfish) (m = 5247, 25%), and S. hopkinsi (squarespot rockfish) (n=2747, 13%). In comparison, both small (TL ^20 cm) and large (TL >20 cm) fishes of commercial species and large noncommercial species were uncommon (462 small- size commercial fish (2%); 295 large-size commercial fish (1%); and 79 large-size noncommercial fish (0.4%)). Fish density and species richness varied between the three broad-scale strata. Hard stratum had the highest density of fish (1357 fishes per 1000 m^), fol- lowed by mixed stratum (862 fishes per 1000 m'-^), and both strata were dominated by rockfishes (90%, 98% respectively). Inversely, soft stratum had com- paratively few fish (130 fishes per 1000 m-), domi- nated by nonrockfish species (63%). Small-size fishes accounted for the majority of demersal fishes within hard (98% of all fish sampled), mixed (99%), and soft (79%) strata. In comparison, large demersal fishes (TL>20 cm) were relatively uncommon in all three substrata; however, the hard stratum had higher densities (27 per 1000 m^ [2% of all fishes in hard substratum]) than the soft (21 per 1000 m^ [16%]), or mixed (4 per 1000 m^ [0.5%]) strata. The mixed stratum had the highest number of species (44 species), where 64% of the species composition comprised non- rockfish species. The hard stratum had slight- ly fewer species (41 species) but comprised a more even mix of rockfish (54%) and nonrock- fish (46%) species. Soft stratum had the few- est species of all three strata (19 species), of which most were nonrockfish species (74%). The number of commercially important species decreased as habitat complexity decreased: 18 commercial species (15 rockfish species) were recorded from hard substratum, com- pared with 16 (10 rockfish species) in mixed substratum, and 11 (5 rockfish species) in soft substratum. Assemblage composition varied between the three broad-scale strata (Fig. 4). High-relief outcrops (hard stratum) were characterized by schools of small-bodied rockfishes (S. hopkinsi, S. wilsoni, and YOY), a suite of large-bodied rockfish (e.g., S. paucispinis, S. flavidus, S. rubrivinctus [flag rockfish], S. rosaceus [rosy rockfish], and Sebasfomus spp.), and a few non- rockfish species (e.g., R. nicholsii and O. elon- gatus). Low-relief outcrops (mixed stratum), in contrast, were characterized by schools of the small-bodied rockfish, S. semicinctus, two large- bodied rockfishes (S. chlorostictus [greenspot- ted rockfish] and S. elongatus [greenstriped rockfish]), and a variety of nonrockfish species (e.g., Citharichthys spp., Zalembius rosaceus [pink seaperch], Zaniolepis spp. [combfishes], Argentina sialis [Pacific argentine], and O. elon- gatus). Homogeneous mud areas (soft stratum) differed from high-relief and low-relief outcrops Anderson and Yoklavich: Habitat association of deepwater demersal fishes off central California 173 by the characteristic presence of Pleuronecti- formes and Agonidae. Intermediate- and fine-scale fish-habitat associations At the level of the individual fish species, a range of benthic habitat variables and spatial scales were important in explaining species-spe- cific distributions. Intermediate-scale informa- tion on patch use, patch selectivity, along with fine-scale microhabitat use, revealed four types of species-specific groups (Fig. 5-8). The first group, rock and boulder associates (e.g., S. hopkinsi, S. flavidus, and S. paucispi- nis) were species that at the intermediate-scale were strongly associated with patches of rock or boulders (or both) (Fig. 5). At the fine-scale, these three species were found on or above rocks (69%, 76%, and 30%, respectively) or boulders (28%f , 24%, and 18%, respectively); S. paucispi- nis also used mud microhabitats (52%). The second group, generalists (e.g., S. wilsoni, S. rosaceus, and O. elongatus) were species that at the intermediate-scale were associated with a variety of patch types (Fig. 6). However, when standardized by habitat availability, these spe- cies were strongly associated with patches of boulders, cobbles, and to a lesser extent, rock, and were negatively associated with patches of homogeneous mud. At the fine-scale, these species were also found on or above all possible microhabitat types and showed a flexibility in habitat use at all three spatial scales. Onto- genetic shifts in habitat use also were indi- cated. For example, small O. elongatus, (<25 cm; «=54) were more abundant in patches with mud or cobbles (e.g., 74% in mud-mud [MM], cobble- boulder [CB], and mud-cobble [MC]), whereas medium- size O. elongatus (25-50 cm; 7) = 57) were found more fre- quently in patches with boulders (40%) and rock (32%). Larger individuals (>50 cm; n=&), on the other hand, were found in patches of rock (83%), indicating that O. elongatus move from mixed mud and cobble habitats to more complex rocky outcrops as they grow. The third group, cobble-mud associates (e.g., S. semi- cinctus, S. chlorostictus, and S. elongatus) were species that at the intermediate-scale were found in patches containing various mixtures of cobbles, mud, and to a lesser extent, boulders (Fig. 7). At the fine-scale, these species were found over mud (66%, 54%, and 81%, respectively) or low-relief cobbles and boulders (pooled 33%, 47%, and 16%, respectively) indicating that mud habitats adjacent to or within mixed cobble- mud areas had inherent properties above either habitat in isolation. Finally, the fourth group, soft-sediment associates (e.g., Pleuronectiformes, Agonidae, Citharichthys spp., and R. nicholsii) were species that at the intermediate- scale were strongly associated with patches containing 1.0 0.5 -0.5 -1.0 Sand / . J //" y nibrivincti Cobbles Z rosiici'us Cilharichtlixs spp S seiniciiicnis •JPatch number ^ wiho thliTi Z/m;t(/i/l\ Relief S rosaceus YOY Rock^^,,,,,,.^,^^. /™{<.m/ Boulders nicholsii S hopkinsi tus S. puucispinis A sicilis / Ziiiiiolcpis spp Z latipinnis^ S elongatus/.- Musi- \ / Agonidae \ Agonidae Pleuronectiformes 4 Mud Depth -1.0 -0.5 0.0 Canonical variate 1 (54.5°' 0.5 1.0 Figure 4 Broad-scale associations of the demersal fish assemblages with benthic habitat variables as discerned from canonical correla- tion analysis. Circles depict the three broad-scale strata (dark gray=hard, white = mixed, light gray=soft sediment), and are presented to assist in the visual association of species and ben- thic habitat variables. Vectors are the eigenvectors of the benthic habitat variables. YOY = young-of-year Sebastes spp., "Stomus" = Sebastomus spp.; S = Sebastes; R = Rhinogobius; O = Ophiodon; Z = Zaniolepis: and A = Argentina. mud or sand (Fig. 8). Pleuronectiformes, Agonidae, and Citharichthys spp. were all associated with homoge- neous soft sediments at all spatial scales (Fig. 8, A-C). In contrast, R. nicholsii were found in a range of soft- sediment patch types (e.g., sand-boulder [SB], sand- sand [SS], mud-rock [MR], mud-boulder [MB], etc.) and microhabitats. However, homogeneous soft-sediment areas had few or no R. nicholsii (Fig. 8D), indicating that, for this species, sediment gaps within a rocky outcrop matrix had inherent properties above either rock or sediment habitats in isolation. Discussion The composition, complexity, and configuration of the seafloor at multiple scales allowed us to predict assem- blage structure and species distributions across the continental shelf within southern Monterey Bay. Broad- scale habitat strata, which are routinely mapped by acoustic methods, showed clear distinctions in assem- blage structure. Hard stratum, composed of high-relief outcrops, was occupied by a diverse range of demer- 174 Fishery Bulletin 105(2) Patch use A S. hopkinsi Patch selectivity Microhabitat use ^! f 1 1 1 100% 50% B S. flavidus Cs pauaspinis 1° Substratum Rock y/.^ Boulders r~1 Cobbles Sand Mud «3 100% 50% — I r— R B C S M -60-40-20 20 40 Patch type Selectivity index Figure 5 Intermediate and fine-scale habitat use by rock and boulder associates: (A) squarespot rockfish (Sesbastes hopkinsi) (B) yellowtail rockfish (S. flavi- dus), and (C) bocaccio (S. paucispinis). At the intermediate-scale, patch types (i.e.. R=rock, B = boulders, C = cobbles, S = sand, and M = mud) are represented by primary (tick labels and shading) and secondary (sequence of ticks within each primary category R, B, C, S, and M) substratum categories and are ordered from hard ( left = rock-rock [RR], rock-boulder [RE], rock-cobble [RC]...) to soft (right = ... mud-cobble [MC], mud-sand [MS], mud-mud [MM]) substratum types. "Patch use" depicts the mean number offish plus standard errors (SE) found in each patch type. "Patch selectivity" depicts the relative patch use by fish, standardized by patch availability: graphs indicate positive (right- hand side of the plot) or negative (left-hand side of the plot) associations with patch types (ordered from hard (top = RR, KB, RC.) to soft (bottom= ... MC, MS, MM) substratum types) and the relative strengths of these associations. Fine-scale microhabitat use is represented by the proportion of fish found on or above a particular substratum type. sal fish species dominated by small rockfish species. Although hard stratum was the least common of the three strata (Anderson et al., 2005), it supported the highest overall densities of fish, including more com- mercial species, than either mixed or soft strata. High fish densities, a dominance of small rockfish species, and the presence of large commercial species over high- relief outcrops have been recorded in other submersible surveys in California (Yoklavich et al., 2000, 2002), Oregon (Stein et al., 1992), Washington (Jagielo et al., 2003), British Columbia (e.g., Murie et al., 1994), and Alaska (O'Connell and Carlile, 1993). For example, Yoklavich et al. (2000) found high numbers of large commercially important rockfish species (e.g., S. pau- cispinis, S. ruberrimus, S. levis [cowcod]) associated with discrete rocky outcrops in a submarine canyon Anderson and Yoklavlch: Habitat association of deepwater demersal fishes off central California 175 Patcfi use 800 600 400 200 ^ 50 + e 40 ^ 30 r 20 o E i Patcfi selectivity Microtiabitat use A S. wilsoni 1" Substratum kXxjj Boulders EI] Cobbles Sand Mud Bs 1 k. i„ ffivfv iT^iT' r r I =33 1 t 1 1 — ' C O. elongatus I h 1 3 1 100% 50% 0% 100% 50% 0% 100% 50% 0% B C S Patcfi type -40-20 20 40 Selectivity index Figure 6 Intermediate and fine-scale habitat use by species that are habitat general- ists: (A) pygmy rockfish iSehastes wilsoni), (B) rosy rockfish (S. rosaceus), and iC) lingcod (O. elongates). Symbols and interpretation are given in Figure 5. off central California. Jagielo et al. (2003) compared trawlable and untrawlable habitats off Washington and found rockfishes iSebastes helvomaculatus [rosethorn rockfish], S. rubberimus, S. flavidus, Sebastes nigro- cinctus [tiger rockfish], and Sebastes spp.) were three times more abundant in untrawlable habitats. In more complex habitat systems. Stein et al. (1992) found high densities of juvenile Sebastes spp. and S. flavidus on the tops of high-relief rocky pinnacles on Heceta Bank, Oregon, whereas in the Gulf of Alaska, O'Connell and Carlile (1993) found the commercially important S. rub- berimus in highest densities in complex habitats. Mixed stratum, characterized by lower complexity and relief than areas of hard stratum, also comprised a distinctive demersal fish assemblage with high numbers of species. High diversity in these areas resulted from a combination of species unique to the mixed stratum (e.g., S. semicinctus), and species characteristic of both hard (e.g., S. wilsoni, O. elongatus, S. rosaceus) and soft (e.g., Pleuronectiformes and Agonidae) strata. In addition to high diversity, some species (e.g., S. chlo- rostictus, S. elongatus, and Z. frenata) were also more abundant in the mixed stratum, indicating that some inherent property of heterogeneous habitats (e.g., mul- tiple resource needs, higher levels of habitat fragmen- tation, and interface zones) may be important to these species. Similar findings have been reported in other submersible surveys. Stein et al. (1992), for example, found more species and higher densities of these species (e.g., S. chlorostictus, S. wilsoni) in patches with either "mud and boulder" or "mud and cobble" than in patches with mud, boulders, or cobbles in isolation. Species use of interface regions can also be inferred from previous studies even though habitat use at the microscale was not explicitly measured. For example, both Richards (1986) and Pearcy et al. (1989) reported higher num- bers of S. elongatus in soft sediment areas adjacent to rocks. Similarly, Yoklavich et al. (2002) found that 176 Fishery Bulletin 105(2) Patch use A S. semicinctus Patch selectivity Microhabitat use ] \ E k . -,....-,.... .....:..! B S. chlorostictus ^^^tS"T^ 1° Substratum ^Rocl< m^i Boulders ED Cobbles Sand IVIud C S. elongatus I I I B C S Patch type 100% 50% 0% 1 00% 50% 0% 100% 50% 0% -40 -20 20 40 Selectivity index Figure 7 Intermediate and fine-scale habitat use by cobble-mud associates: (A) halfbanded rockfish {Sebastes semicinctus), (B) greenspotted rockfish (S. chlorostictus), and (C) greenstriped rockfish (S. elongates). Symbols and interpretation are given in Figure 5. S. chlorostictus, along with other species, used habitats comprising a combination of rock and mud. Soft substratum had the lowest habitat complexity and the lowest diversity and density of fishes of all three strata, although many of these species, particu- larly the Pleuronectiformes, are important commer- cial species. Stein et al. (1992), Yoklavich et al. (2000, 2002), and Jagielo et al. (2003) also recorded similar demersal fish assemblages in flat mud habitats (i.e., Pleuronectidae, namely M. pacificus, Glyptocephalus zachirus [rex sole], and Lyopsetta exilis [slender sole]), Agonidae, Sebastes saxicola [stripetail rockfish], Zo- arcidae [eelpouts], and Sebastolobus spp. [thornyhead species]). Although demersal fish assemblages over trawlable habitats have been well documented by tra- ditional fishery methods (e.g., Weinberg et al., 2002), biases in catchability between strata (because trawls may snag in complex habitats) mean that differences in fish assemblage structure between soft, mixed, and hard strata have been difficult to identify. Although in situ submersible surveys facilitate these types of com- parisons, some biases may still be present. For example, soft-sediment habitats reported in submersible studies (e.g., Stein et al., 1992; O'Connell and Carlile, 1993; Yoklavich et al., 2002) are often adjacent to, at the base of, or in the general vicinity of rock outcrops. As a result, it is unclear how the proximity of hard structure influences demersal fish composition and abundance, or whether these habitats are representative of soft-sedi- ment areas where rock outcrops are not present. The analysis of submersible transects in relation to distance from rocks, or alternatively trawl surveys that include video or acoustic images of the benthos, may help to clarify these patterns. All three spatial scales provided valuable informa- tion on how demersal fish species use benthic habitats. Anderson and Yoklavich: Habitat association of deepwater demersal fishes off central California 177 E Patcti use A Pleuronectiformes Patcfi selectivity 2.0 B Agonldae 1.5- 1.0- 0.5- 0.0 1° Substratum ^^Rock t-:-:-M Boulders L_J Cobbles ^Sand iHIMud I , I l ^' l . , C Citharichthys spp Patcfi type Microtiabitat use I i E [ I ^ E c J □ : [ 1311 U -20 20 40 60 Selectivity index 100% 50% 0% 100% 50% 0% 100% 50% 0% 100% 50% 0% Figure 8 Intermediate and fine-scale habitat use by soft-sediment associates; (A) flat- fishes, Pleuronectiformes, (B) poachers, Agonidae, iC) sanddabs, Citharichthys spp., and the blackeyed goby (Rhinogobiops nicholsii). Symbols and interpreta- tion are given in Figure 5. For example, broad-scale strata supported characteris- tic fish assemblages. However, at intermediate scales (within a strata), species distribution varied by patch composition, patch size, and the neighborhood of sur- rounding patches. At fine scales, microhabitat use by fishes indicated which portions of habitat-patches were actually used (e.g., species A in cobbles and species B in mud, where both species were present within the same cobble-mud patch). A vital aspect of using a multiscaled approach, however, was that information from each spa- tial scale could then be integrated to examine the rela- tive importance of habitat types and their structural configuration, and this information also indicated that for some species the landscape context was important. For example, R. nicholsii was mainly found on sand or at the interface between sand and rock (microhabitat use), but these microhabitats were located within a range of rock and sediment patch types (intermediate- scale), which in turn were located within the complex hard stratum (broad-scale). This structure indicated 178 Fishery Bulletin 105(2) that for R. nicholsii sediment gaps within or adjacent to a rocky landscape were required. On the other hand, S. chlorostictus and S. elongatus were both more abundant in the mixed stratum than in the hard stratum (broad scale) and were present together within mixed boulder, cobble, and mud sub- strata (intermediate scale). At fine scales, however, microhabitat use by these species differed; S. elongatus was common in the mud portion of these patches and S. chlorostictus was common over boulders and cobbles. These findings indicated that both species were in- terface associates, but within these interface regions different substratum types were used. The inclusion of microhabitat information within this multiscale ap- proach provided a more comprehensive understanding of how demersal fish use benthic substrata. However, recording microhabitat use for each fish (« =21,184 fishes) was time consuming and therefore would likely negate its use in some studies. A recommended alter- native method for recording microhabitat use might be to measure microhabitat use for a subset of fish per species, where subsamples are selected unbiasedly from the overall sample pool. The ability to describe and predict fish-habitat re- lationships, as identified in this study, can be used to address area-based management concerns in several ways. For example, species captured by benthic trawl and long-line gear could be used to infer the presence of seafloor substratum types. Although this form of information is not novel, our study provides detailed quantitative species-habitat associations that validate this approach. For example, a benthic trawl that cap- tures Pleuronectidae, Agonidae, S. semicinctus, S. chlo- rostictus, and S. elongatus, would indicate that the area trawled encompassed multiple strata (e.g., one or more areas of low-relief outcrop and homogenous mud). How- ever, the proportions and spatial configuration of these strata would not be known unless a video camera, for example, was mounted on a benthic trawl (e.g., Abookire and Rose, 2005), or a seafloor substrata map was avail- able for the area (e.g.. Bellman et al., 2005). Conversely, habitat could be used to predict commu- nity structure and species distributions. In this study, substratum type was a good indicator of distribution and abundance of many commercial and noncommer- cial fish species. However, the spatial arrangement and degree of habitat patchiness, in addition to substra- tum type, also were important predictive variables. Consequently, although areal estimates of substrata are likely to be effective for modeling the abundance and distribution of certain species (e.g., S. rosaceus and S. flavidus), accurately estimating other species will require additional knowledge of the spatial ar- rangement of these substrata. For example, species associated with sediment-rock interfaces, such as S. chlorostictus, S. elongatus, and Z. frenata, are likely to be modeled more effectively by estimating the perimeter of either an outcrop or specific habitat type. Likewise, the ability to model gap-associate species, such as R. nicholsii, will require information on the availability of sediment-outcrop interfaces and sediment gaps within an outcrop matrix. For other species, such as young- of-year rockfish, a measure of habitat patchiness, in combination with areal estimates of substrata, may be required. The ability to map this level of habitat detail will depend to a large degree on a trade-off between data acquisition and resolution of the mapping tools used, and the amount of seafloor needed to be mapped (Anderson et al., 2005). In conclusion, the overall success of area-based man- agement strategies will reflect the ability of research- ers to accurately measure the functional relationships between organisms and their habitat. Multiscale in situ surveys, such as this one, undertaken in multi- ple locations, in combination with larger-scale fishery surveys can improve our understanding of the role of benthic habitats in structuring demersal fishes across the broader U. S. West Coast. These insights, in turn, improve our ability to characterize and map essential fish habitat, estimate habitat availability, and predict multispecies distributions and habitat associations within specified areas such as marine protected areas. Importantly, this study also provides a quantitative baseline of demersal fish assemblage structure for both commercial and noncommercial species, which is critical for future comparisons of spatiotemporal abundance, diversity, and habitat use. This baseline is also vital for assessing the effects and value of increased protection of West Coast shelf ecosystems. Acknowledgments We thank G. Cailliet, R. Lea, M. Love, G. Moreno, R. Parrish, P. Reilly, L. Snook, R. Starr, D. Sullivan, the Delta Oceanographies (R. Slater, D. Slater, and C. Ijames) for logistical support and biological data collec- tion. We thank M. Carr, C. Grimes, S. Ralston, C. Syms, B. Tissot, and W. Wakefield for their conversations and advice during the analysis and writing of this research, and Tim Simmonds and Guy Cochrane for graphical support. The manuscript was improved through the com- ments of M. Cappo, C. Grimes, T. Laidig, C. Syms, and three anonymous reviewers. This project was partially supported by NOAAs National Undersea Research Pro- gram, West Coast and Polar Undersea Research Center, University of Alaska, Fairbanks (Grants UAF-92-0063 and UAF-93-0036) and by a University of California, Santa Cruz postdoctoral fellowship to Tara Anderson, with joint funding from National Marine Fisheries Ser- vice, U. S. Geological Survey, and the National Marine Protected Areas Center Science Institute. Literature cited Abookire, A. A., and C. S. Rose. 200.5. Modifications to a plumb staff beam trawl for sampling uneven, complex habitats. Fish. Res. 71: 247-254. Anderson and Yoklavlch; Habitat association of deepwater demersal fishes off central California 179 Adams, P. B., J. L. Butler, C. H. Baxter. T. E. Laidig. K. A. Dahlin, and W. W. Wakefield. 1995. Population estimates of Pacific coast groundfishes from video transects and swept-area trawls. Fish. Bull. 9.3:446-455. Addicott, J. F., J. M. Aho, M. F. Antoun. D. K. Padilla, J. S. Richardson, and D. A. Soluk. 1987. Ecological neighbourhoods: scaling environmental patterns. Oikos 49:340-346. Anderson, T. J., M. M. Yoklavich, and S. L. Eittreim. 2005. Linking fine-scale groundfish distributions with large-scale seafloor maps: Issues and challenges of combining biological and geological data. In Benthic habitats and the effects of fishing (P. W. Barnes, and J. P. Thomas, eds.), p. 667-678. Am. Fish. Soc. Symp. 41. Bethesda, MD. Ault, T. R., and C. R. Johnson. 1998. Spatially and temporally predictable fish communi- ties on coral reefs. Ecol. Monogr. 68:25-50. Bellman, M. A., S. A. Heppell, and C. Goldfinger. 2005. Evaluation of a US west coast groundfish habi- tat conservation regulation via analysis of spatial and temporal patterns of trawl fishing effort. Can. J. Fish. Aquat. Sci. 62:2886-2900. Eittreim, S. L., R. J. Anima, and A. J. Stevenson. 2002. Seafloor geology of the Monterey Bay area conti- nental shelf. Mar. Geol. 181:3-34. Forman, R. T. T. 1995. Land mosaics: the ecology of landscapes and regions, 632 p. Cambridge Univ. Press, Cambridge, U.K. Gunderson, D. R., and T. M. Sample. 1980. Distribution and abundance of rockfish off Wash- ington, Oregon, and California during 1977. Mar. Fish. Rev. 42:2-16. Jagielo, T., A. Hoffmann, J. Tagart, and M. Zimmermann. 2003. Demersal groundfish densities in trawlable and untrawlable habitats off Washington: implications for the estimation of habitat bias in trawl surveys. Fish. Bull. 101:545-565. •Jones, G. P., and C. Syms. 1998. Disturbance, habitat structure and the ecology of fishes on coral reefs. Aust. J. Ecol. 23:287-297. Langton, R. W., P. J. Auster, and D. C. Schneider. 1995. A spatial and temporal perspective on research and management of groundfish in the Northwest Atlantic. Rev. Fish Sci. 3:201-29. Love, M. S. 2006. Subsistence, commercial, and recreational fisheries, /n The ecology of marine fishes: California and adjacent waters (L. G. Allen, D. J. Pondella, and M. H. Horn, eds.), p. 567-594. Univ. California Press, Berkeley, CA. Love, M. S., and M. Yoklavich. 2006. Deep rock habitats. In The ecology of marine fishes: California and adjacent waters (L. G. Allen, D. J. Pondella. and M. H. Horn, eds.), p. 411-427. Univ. California Press, Berkeley, CA. Love, M. S., M. Yoklavich. and L. Thorsteinson. 2002. The rockfishes of the northeast Pacific, 405 p. Univ. California Press, Berkeley and Los Angeles, CA. McArdle, B. H., K. J. Gaston, and J. H. Lawton 1990. Variation in the size of animal populations: pat- terns, problems and artifacts. J. Anim. Ecol. 59: 439-454. Murie, D. J., D. C. Parkyn, B. G. Clapp, and G. G. Krause. 1994. Observations on the distribution and activities of rockfish, Sehastes spp., in Saanich Inlet, British Columbia, from the Pisces IV submersible. Fish. Bull. 92:313-323. O'Connell. V. M., and D. W. Carlile. 1993. Habitat-specific density of adult yelloweye rockfish Sebastes ruberrimus in the eastern Gulf of Alaska. Fish. Bull. 91:304-309. Pearcy, W. G., D. L. Stein, M. A. Hixon, E. K. Pikitch, W. H. Barss, and R. M. Starr. 1989. Submersible observations of deep-reef fishes of Heceta Bank, Oregon. Fish. Bull. 87:955-965. Richards, L. J. 1986. Depth and habitat distributions of three species of rockfish iSebastes) in British Columbia: observa- tions from the submersible PISCES IV. Environ. Biol. Fishes 17:13-21. Stein, D. L., B. N. Tissot, M. A. Hixon, and W. H. Barss. 1992. Fish-habitat associations on a deep reef at the edge of the Oregon continental shelf. Fish. Bull. 90:540-551. Stephens, J. S. Jr., R. J. Larson, and D. J. I. Podella. 2006. Rocky reefs and kelp beds. In The ecology of marine fishes: California and adjacent waters (L. G. Allen, D. J. Pondella, and M. H. Horn, eds.), p. 227-252. Univ. California Press, Berkeley, CA. Syms, C, and G. P. Jones. 1999. Scale of disturbance and the structure of a tem- perate fish guild. Ecology 80:921-940 Weinberg, K. L. 1994. Rockfish assemblages of the middle shelf and upper slope off Oregon and Washington. Fish. Bull. 92:620-632. 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, 140 p. Wiens, J. A. 1989. Spatial scaling in ecology. Funct. Ecol. 3:385- 397. 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., G. M. Cailliet, R. N. Lea, H. G. Greene, R. M. Starr, J. de Marignac, and J. Field. 2002. Deepwater habitat and fish resources associated with the Big Creek Marine Ecological Reserve. CalCOFI Rep. 43:120-140. Yoklavich, M. M., H. G. Greene, G. M. Cailliet, D. E. Sullivan, R. N. Lea, and M. S. Love. 2000. Habitat association of deep-water rockfishes in a submarine canyon: an example of a natural refuge. Fish. Bull. 98:625-641. 180 Abstract — Endoparasitic helminths were inventoried in 483 American plaice iHippoglossoidesjjlatessoides) collected from the southern Gulf of St. Lawrence, NAFO (North Atlantic Fisheries Organization) division 4T, and Cape Breton Shelf (NAFO sub- division 4Vn) in September 2004 and May 2003, respectively. Forward step- wise discriminant function analysis (DFA) of the 4T samples indicated that abundances of the acanthocephalans Echinorhynchus gadi and Corynosoma strumosum were significant in the classification of plaice to western or eastern 4T. Cross validation yielded a correct classification rate of 79% over- all, thereby supporting the findings of earlier mark-recapture studies which have indicated that 4T plaice comprise two discrete stocks: a western and an eastern stock. Further analyses including 4Vn samples, however, indi- cated that endoparasitic helminths may have little value as tags in the classification of plaice overwinter- ing in Laurentian Channel waters of the Cabot Strait and Cape Breton Shelf, where mixing of 4T and 4Vn fish may occur. Use of endoparasitic helminths as tags in delineating stocks of American plaice iHippoglossoides platessoides) from the southern Gulf of St. Lawrence and Cape Breton Shelf Gary McClelland (contact author) Jason Melendy Fisheries and Oceans Canada Gulf Fisheries Centre 343 University Avenue Moncton, New Brunswick E1C9B6 Canada Email address for G. McClelland: mcdeliandgcSdfo-mpo gc.ca Manuscript submitted 17 May 2006 to the Scientific Editor's Office. Manuscript approved for publication 25 August 2006 by the Scientific Editor. Fish. Bull. 105:180-188(2007). American plaice iHippoglossoides platessoides) are found in Northwest Atlantic waters along the continental shelf and upper continental slope from west Greenland to Rhode Island, favor- ing intermediate depths (90-250 m), cold waters (<0-1.5°C), and fine sand or mud bottom (Scott and Scott, 1988). Amercian plaice has ranked second in importance to Atlantic cod (Gadiis morhua) among groundfish landed in the southern Gulf of St. Lawrence (North Atlantic Fisheries Organiza- tion [NAFO] division 4T) over the past four decades, but commercial landings, which had ranged from 4907 to 11,780 t between 1965 and 1992 (Morin et al.i), fell to 401 t by 2004 (Fisheries and Oceans Canada'-). Commercial and research survey data show that declines in abundance of southern Gulf plaice since 1991 have occurred primar- ily in western 4T between the Gaspe Peninsula and the Magdalen Islands (see Fig. 1), and abundances east of the Magdalens have remained stable. The eastern and western 4T stocks monitored in recent assessments are consistent with the meristically indis- tinguishable "Miscou-Magdalen" and "Cape Breton" groups which Powles (1965) delineated through mark-recap- ture experiments. Powles (1965) sug- gested that the meristic uniformity of these two groups may be a conse- quence of larval drift because mature fish from the respective groups seldom mix on their summer feeding grounds. More recently, Stott et al. (1992) found no evidence of population subdivision of Southern Gulf plaice in respect to allozyme variation and restriction fragment length polymorphisms in mitochondrial DNA. As evident from commercial catch records and tag returns, 4T plaice, with the exception of some immature fish that remain in the shoals year round, migrate from summer feeding grounds on the Magdalen Shallows to deeper waters of the Laurentian Channel in winter (Powles, 1965). The winter distribution of plaice in the Laurentian Channel is continuous from the Gaspe to the Cabot Strait (Clay, 1991) and eastward along the Cape Breton Shelf (NAFO subdivision ' Morin, R., I. Forest, and G. Poirier. 2001. Status of NAFO Division 4T American plaice, February 2001. Cana- dian Science Advisory Secretariat Research Document. 2001/023, 70 p. Fisheries and Oceans Canada, Sci- ence Branch, Marine Fish Division, Gulf Region, P.O. Box 5030, Moncton, New Brunswick, Canada. - Fisheries and Oceans Canada. 2005. American plaice in the southern Gulf of St. Lawrence (Division 4T). Fisheries and Oceans Canadian Science Advisory Secretariat Science. Advisory Report. 2005/008, 5 p. Maritime Provinces, Regional Advisory Process, Fisheries and Oceans Canada, P.O. Box 1006, Dartmouth, Nova Scotia, Canada. McClelland and Melendy: Heminths as tags in delineating stock of Hippoglossoides platessoides 181 49°N 48'N 47°N - 46 N - 65 W 64 W 63 W 62"W 61 "W 60W 59'W 58=W Figure 1 Map of eastern Canada indicating sites where American plaice {Hippoglossoides plates- soides} were sampled in the southwestern (•) and southeastern (■) Gulf of St. Lawrence, and on or near the Cape Breton Shelf (♦). The depth contour ( ) is at 100 m. The pale gray lines demarcate North Atlantic Fisheries Organization divisions (4T, 4S. 4R) and subdivisions (3Pn, 3Ps, and 4Vn). 4Vn) (Swain et al., 1998). Clay (1991), noting the abun- dance of plaice in the Gulf portion of the Channel in January, surmised that southern Gulf plaice, unlike 4T cod, with which they are often closely associated, do not migrate through the Cabot Strait to over-winter in 4Vn. Length-at-age data from commercial landings indicate that plaice taken near the Cabot Strait late in the year are primarily from the slower growing eastern 4T (Cape Breton) stock (Tallman, 1991). A few older plaice tagged on Bradelle Bank, northwest of the Magdalen Islands, however, have been recovered in the Cabot Strait area in fall and winter (Powles, 1965). Clearly, tagging data and information on seasonal distributions from research surveys and commercial fisheries have provided little insight into the movements of 4T plaice stocks in the Laurentian Channel in winter and have not dispelled the possibility that 4T stocks may be exploited during the winter fishery within 4Vn. Parasites as biological tags have often proven advan- tageous over more costly mark-recapture methods in studies offish stock structure and migration (Williams et al, 1992) and are useful as markers in surveys for allocating quotas and combating illegal landings (Pow- er et al., 2005). Parasite markers have shown potential in stock delineation of numerous demersal fish species (Marcogliese et al., 2003; MacKenzie and Abuanza, 2005; Melendy et al., 2005), including flatfish species such as Greenland halibut (Reinhardtius hippoglossoi- des) (Arthur and Albert, 1993; Boje et al, 1997), Pacific halibut (Hippoglossus stenolepis) (Blaylock et al., 2003), and winter flounder (Pleiironectes americanus) (McClel- land et al., 2005). Although there have been no prior attempts to use parasite assemblages in describing the stock structure of American plaice, parasitological sur- veys of plaice in the Northwest Atlantic have identified a number of potential parasite tags. Scott (1975), for example, concluded that the enteric digeneans Sterin- gotrema ovacutum, Zoogonoides viviparous, and Fel- lodistomum furcigerum might prove useful as tags for plaice stocks from the southern Gulf of St. Lawrence, Scotian Shelf, and northeastern Gulf of Maine. Zub- chenko (1985) remarked that species compositions and infection parameters of 20 protozoan and metazoan parasites of plaice from the Grand Bank, Flemish Cap, and northeastern Newfoundland and Labrador are peculiar to their geographic origin. Finally, studies of the temporal and geographic distributions of larval sealworm {Pseudoterranova decipiens) (Nematoda) in Atlantic Canadian groundfish have revealed significant disparities in prevalence and abundance of sealworm in neighboring plaice stocks (McClelland et al., 2000; McClelland and Martell, 2001b). 182 Fishery Bulletin 105(2) The primary objective of the present study was to in- vestigate the possibility of using parasite markers to de- lineate the western and eastern 4T plaice stocks, evident in earlier mark-recapture experiments (Powles, 1965). In light of potential mixing of southern Gulf and Cape Breton Shelf plaice over-wintering in the Laurentian Channel, parasite tags were also employed in an effort to examine the discreteness of 4T and 4Vn stocks. Materials and methods Sampling of the host American plaice, 31 to 40 cm in total length, were col- lected from two locations in the southern Gulf of St. Lawrence (NAFO division 4T) and from three locations on or near the Cape Breton Shelf (NAFO subdivision 4Vn) (Fig. 1). Samples from the southern Gulf were collected from the Canadian Coast Guard Ship (CCGS) Alfred Needier and the CCGS Teleost during a Fisher- ies and Oceans Canada (DFO) demersal fish survey in September of 2004, and plaice from the Cape Breton Shelf were sampled from the Alfred Needier in May 2003 during a dedicated survey of parasites and diseases in 4Vn groundfish. Fish were caught with a Western IIA otter trawl which was towed for 30 minutes at each sta- tion. A total of 483 plaice were collected, including 137 plaice taken at depths of 43 to 88 m in the western 4T division and 95 plaice sampled at depths of 49 to 70 m in eastern 4T division. In the 4Vn survey, 128 plaice were sampled from the eastern slope of the Smokey Channel, 46 along the eastern edge of the Cape Breton Shelf, and 77 in the Louisbourg Hole. "Channel," "Edge," and "Hole" samples were taken at depths ranging from 77 to 124 m, 72 to 158 m, and 77 to 135 m, respectively. Samples were frozen onboard in a walk-in freezer at -17°C and transferred on landing to a -20° C walk-in freezer at the Gulf Fisheries Centre (GFC), Moncton, New Brunswick, where they were stored for future examination. Individual fish were thawed at room temperature and measured lengthwise to the nearest centimeter. External surfaces and gills were inspected by eye for signs of trauma and other disease conditions. Viscera were examined for endoparasitic helminths with a dis- secting microscope, and fillets were removed and sliced into thin sections (McClelland and Martell, 2001a) for detection of larval digeneans and nematodes. All hel- minth parasites were counted with the exception of Stephanostomum baccatum metecercariae. Because the latter were often too numerous and too widely distrib- uted in host tissues to be counted within a reasonable time frame, only their presence was noted. Statistical analysis Prevalence (?) and abundance (A) of individual parasite species were calculated according to the methods of Bush et al. (1997), with prevalence being the numbers of infected fish in a sample divided by the total number of fish in a sample, expressed as a percentage, and abundance being the total number of parasites recov- ered from a sample divided by the total number of fish in a sample. Individual fish were coded 1 (infected), or (uninfected) (Li, 1964; Neter et al., 1985) for analyses of prevalence, and a log in+l) transformation was used to bring the distributions of parasite counts closer to normality (Sokal and Rohlf, 1969; Piatt, 1975). Parasite infection parameters that best distinguished between sampling locations were selected by forward stepwise discriminant function analysis (DFA). The Kappa sta- tistic iK) was used to determine the improvement over chance of DFA (Titus et al., 1984). Misclassification rates of DFAs were calculated by employing cross validation procedures described by Arthur and Albert (1993) and Boje et al. (1997). All statistical procedures were per- formed with Systat for windows, vers. 7.0 (SPSS Inc., Chicago, IL). Because the survey was confined to plaice in a narrow length range (31-40 cm TL) (McClelland et al., 2000), effects of host size on parasite infection parameters were not investigated. Plaice from the edge of the Cape Breton Shelf were not included in the DFA analysis because of the small sample size (« = 46) and the fact that they were collected from two widely sepa- rated groups of trawl sets (Fig. 1); the northern (n=21) and southern {n=25) portions of the sample were more distant from each other than were the samples from Smokey Channel and Louisbourg Hole. Results Plaice sampled from the southern Gulf of St. Lawrence (NAFO Division 4T) and the Cape Breton Shelf and vicinity (NAFO subdivision 4Vn) ranged from 31 to 40 cm in total length (TL), but the great majority fell within the lower half of this range. Mean length of fish (±standard error) was 32.34 ±0.180 cm (n=137} for plaice from the western 4T sample, and 32.31 ±0.22 cm (/! = 95) for the eastern 4T sample. Mean lengths of 4Vn plaice were 33.17 ±0.26 cm {n=128) and 32.88 ±0.23 cm (n=n) for Smokey Channel and Louisbourg Hole fish, respectively. Eleven species of endoparasitic helminths were identi- fied during routine parasitological examinations of 483 plaice (Table 1), but the larval digeneans Otodistomum sp., plerocercoid larvae of an unknown cestode species, and the nematode Hysterothylaciuin aduncum were ob- served too infrequently ( IN lO CO t> »n o o O CO ^ ^ ■-H CJ) 3 o o II d d d d d d d d d < +1 +1 +1 +1 +1 +1 +1 +1 +1 -a c CD -5 o rt Tf CD I> IM O CD C^ ^. |^i rt cfi — CD d d ci (^ ci <^ o 3 ^ o o H -^ hJ k c CU CD O Ir- CO ^ .-H CO Tf O CO 00 IM CO lO .-1 CO c 5 Oi [M CO C^ CD C^ d) TO — O O 03 O O CO CD CO II 0) c: < d d +1 ti d +1 d <6 <6 +1 +1 +1 d +1 O (>] _cO o o^ o ^ o bc i m o CO "S .-H .-H Tj^ CD ^3 ^ d d T-H d d --i •-H S ~2 W "a) o rG ^J 0) tfl m :^ t*-. x: M o Cu -* Ol Oi ^ o lo o en o O CO -^ CO (M CO rH 1X1 CO 6 S5 .= cu vj -= '^ cx ^ in ^ , I c:^ ■^ i-H I:- 00 lO QC o o lO o O O ^ .-< O .-H ..^ d d d d d d d d d d c >. II ^5 < +1 +1 +1 CD _C0 +1 T-H +1 +1 +1 +1 CO c^ ^ t- +1 +1 CTl 00 ^ £f --1 o C^ "c 1— t IM o C-: t-. O Tf S 3 o — d d d d d o .-H d d II E S M g -2 3 a, o hJ CO J2 Dh CO --I lO CO CO IM (M C35 00 ^ lO ■y. hJ Tf CO OJ CO CD IM :|^ o to 8 <«' o ^ -i: 0) _ OOnC •^ P M CO 00 IM CO Od CO in Table reton o CO o O O .-1 ^ CO rH < d d d <6 o d d d d +1 +1 +1 +1 +1 +1 +1 +1 +1 CD ■= CO CO IM q 00 "^ •-< "^ CO in S pa O 1 — 1 c O O CO CO rA d I ^ OJ II d d d d d .-4 a. a CO c^ M 03 OJ Cu CD O CD CO ,_, 00 -* lo r- o o o x: rsi CO ,-1 CD in CO 1^ < be ,bX t- o T-H T. ^ C^l t^ CO O 00 IM CO CD 1 s o o O O r-l ,-H O O d d d d d d d d d ■^ ca < +1 •fi +1 +1 +1 +1 +1 +1 +1 S j: ^ == 0^ _co CO m r^ ^ in "qj CJ o 00 "c o i-i O CD ^ i-H ,— 1 J= >. 2 1; d d d d d d '-i d d o OJ CO — 5^ S Cfi o, ^ o Oi lO CO O 1-H CD CO C^ 00 CO . IM ^ (M c~ a dj " 01 < g s^ Is "s S ., •^cc 3 lit "s o 3 -iS .Q, 3 -Q i2 CO 3 -o O c ~ CO H - 'S' i.* c — u s 1 § B ^ 3 d s CO c; E c2 s o 1 -a ^ 1 § 1 ^ ^' ^ t CO CO *« 3 §--2 « J3 ^, c c — CO CO .2 ?^ OJ C! ^ ■*- g §C)0 MX Q o CO s Cfi o CO S ^ S CO 3 -a -a o -3 o ^_» C +J CO LD CO •z. CO CO ■;::: p ^ .£3 ■n -2 ts ^ g o j^ 'i; tc o a, -g < 2 < c t; w to O C3 .2 m o 15 > CO 184 Fishery Bulletin 105(2) Discussion Forward stepwise DFA of parasite tags has been suc- cessfully employed in describing the stock structure of various flatfish species. Arthur and Albert (1993) used Table 2 Results of forward stepwise discriminant function analy- sis of samples of American plaice iHippoglossoides plates- soides) from the southern Gulf of St. Lawrence (North Atlantic Fisheries Organization Division 4T) showing the relative importance of abundances of two species of acanthocephalan in the classification of individual fish to western and eastern 4T stocks. Standardized coefficient of first canonical Variable variable df F score Eigen- value Echinorhynchus 0.862 a Corynosoma 0.499 strumosum 230 71.48 0.46 229 20.84 Table 3 Cross-validation of American plaice iHippoglossoides pla- tessoides) from the western and eastern North Atlantic Fisheries Organization division 4T (southern Gulf of St. Lawrence) determined by discriminant function analysis of abundances of two species of acanthocephalans. Location Western 4T Eastern 4T 9c Correct Western 4T Eastern 4T Totals 117 28 145 20 67 87 85 71 79 Table 4 Cross-validation of American plaice iHippoglossoides platessoides) from the Smokey Channel and Louisbourg Hole, North Atlantic Fisheries Organization subdivision 4Vn, as revealed by discriminant function analyses of the prevalence or abundance of four helminth species (preva- lence of Stephanostomum baccatum, and abundances of Fellodistomum sp., Corynosoma strumosum. and Pseudo- terranova decipiens). Location Smokey Channel Louisbourg Hole Correct Smokey Channel Louisbourg Hole Totals 86 20 105 42 57 99 67 74 70 five species of larval helminth to classify widely sepa- rated stocks of Greenland halibut from the Saguenay fiord, St. Lawrence estuary, Labrador coast, and Baffin Island, amd Boje et al. (1997) used six helminth species to gain insight into the stock structure of Greenland halibut from northeastern Newfoundland and Green- land. The overall correct classification rate was >99% in the former, and 77% in the latter study. Blaylock et al. (2003) used eight helminth species to describe the stock structure of Pacific halibut along the Pacific coast of North America from the Bering Sea to northern Cali- fornia and achieved an overall correct classification rate of 83%. In all three studies, accuracy of classification increased when the numbers of geographical categories were reduced by pooling sampling locations. Although our survey was confined to a much smaller geographic area than those surveyed in the studies above, the accuracy of classification of plaice stocks from the southern Gulf of St. Lawrence (NAFO division 4T) is remarkably high (79%) (Table 3). DFA with only two markers, namely the abundances of the acantho- cephalans E. gadi and C. strumosum, indicated that southern Gulf plaice consist of discrete western and eastern stocks on their summer feeding grounds; this result supports the results of earlier mark-recapture studies (Powles, 1965). Because the acanthocephalans are acquired passively in prey, and 4T plaice feed pri- marily in the Magdalen Shoals, spring through fall, while fasting in winter (Swain et al., 1998), these markers may also be employed in studies of seasonal migrations and mixing of the two stocks. Analysis of infection parameters of four helminth taxa (abundances of C stru!7iosum. the digenean Fellodistomum sp.. the larval anisakine nematode Pseudoterranova decipiens, and prevalence of the larval digenean Stephanostomum baccatum) reveals a somewhat lower degree of discrete- ness (70% correct classification) between plaice from Smokey Channel and Louisbourg Hole, within NAFO subdivision 4Vn. Abundances of Fellodistomum sp., P. decipiens, E. gadi, and C. strumosum, and prevalence of S. baccatum are significant factors in a DFA of plaice from all four locations, although the overall accuracy of classification falls to only 48% (Table 5). Although an improvement over random classification (25%), this latter result of- fers little encouragement for the use of parasites tags in future investigations of movements and possible mix- ing of 4T and 4Vn plaice overwintering in Laurentian Channel waters of the 4Vn subdivisioin. DFA of eastern 4T and Smokey Channel (4Vn) samples similarly seems to indicate that helminth tags would have little value in studies of the winter composition of plaice stocks in the Cabot Strait area. In this analysis, with the use of E. gadi and the larval nematodes Anisakis simplex and Contracaecum osculatujn as markers, 98% of the 4Vn plaice were classified correctly, but 52% of the plaice from southeastern 4T were misclassified as 4Vn fish. However, given the differences in infection parameters of E. gadi in eastern 4T (prevalence [P] = 50%, abun- dance [A]=1.30) and Smokey Channel samples (P=4%, McClelland and Melendy: Heminths as tags in delineating stock o[ Hippoglossoides platessoides 185 Table 5 Cross-validation of American plaice (Hippoglossoides platessoides) from the western and eastern North Atlantic Fisheries Orga- nization (NAFO) division 4T and from Smokey Channel and Louisbourg Hole in NAFO subdivision 4Vn, as revealed by dis- criminant function analyses of the prevalence or abundance of five species of parasitic helminth (prevalence oi Stephanostomum baccatum, and abundances oi Fellodistomum sp., Corynosoma strumosum, Echinorhynchus gadi, and Pseudoterranova decipi- ens). Location Western 4T Eastern 4T Smokey Channel Louisbourg Hole % Correct Western 4T 57 10 37 33 42 Eastern 4T 14 48 15 18 51 Smokey Channel 28 5 64 31 50 Louisbourg Hole 18 8 11 40 52 Totals 117 71 127 122 48 A=0.09) (Table 1), this acanthocephalan may yet prove to be a useful marker for the detection of southeastern 4T migrants among plaice overwintering in Cabot Strait and 4Vn. Parasites that contributed to the DFAs above meet MacKenzie's (1987) criteria for biological tags. None of the helminths are known to reproduce directly on or in plaice. Each of the markers is abundant in fish from at least one of the sampling areas, and there are significant geographical variations in the infection pa- rameters of each species within the survey area. The third criterion, that the parasite must be long-lived in the host, is clearly met by five species of larval hel- minth that contribute significantly to classification. Metacercariae of S. baccatum. the larval anisakines A. simplex, C. osculatum. and P. decipiens, and cystacanths of C. strumosum, found variously in the body cavity and musculature, are believed to survive indefinitely in the fish host and have been used as markers in other stud- ies of flatfish stock structure (Arthur and Albert, 1993; Boje et al.,1997; Blaylock et al., 2003). Although the life spans of enteric helminths (Fellodis- tomum sp. and E. gadi, herein) in marine fish remain largely unknown, it is possible that species infecting cold-water hosts may persist for several months or even years (Margolis and Boyce, 1969) — sufficient time to meet MacKenzie's (1987) criterion for tag longevity. Echinorhynchus gadi infecting a relict population of Atlantic cod in Lake Mogil'noye, Russia, for example, recruit in the fall and die off in late summer and early fall of the following year (Kulachkova and Timofeyeva. 1993). Similarly, E. lageniformis survive for about a year in the intestines of starry flounder (Platichthys stellatus) in the coastal waters of Oregon (Olson and Pratt, 1971). In any event, parasites need not be long lived in order to be suitable as markers, and there are numerous precedents for use of enteric helminths as markers in studies of host stock structure (Williams et al., 1992). Khan and Tuck (1995) identified E. gadi as an important indicator of the discreteness of New- foundland cod stocks, and Power et al. (2005) employed abundances of enteric digeneans in classifying bogue Table 6 Cross-validation of American plaice (Hippoglossoides platessoides) to eastern North Atlantic Fisheries Organi- zation (NAFO) division 4T and Smokey Channel (NAFO subdivision 4Vn) as revealed bv discriminant function analyses of the abundances of three helminth species (abundances oi Echinorhynchus gadi. Anisakis simplex. and Contracaecum osculatum). Smokey % Location Eastern 4T Channel Correct Eastern 4T 46 49 48 Smokey Channel 3 125 98 Totals 49 174 77 (Boops boops) (Sparidae), a demersal species from Span- ish fisheries. Abundances of passively transmitted parasites are ultimately a function of host feeding behavior. Hence, geographical differences in intermediate-host abun- dance and frequency in plaice diets would manifest themselves as disparities in infection parameters of passively transmitted helminths in 4T and 4Vn plaice stocks. Digeneans, which mature in the digestive tract of marine fish, are usually acquired through the con- sumption of invertebrates (crustaceans, molluscs, poly- chaetes, and echinoderms, among other taxa) which harbor encysted metacercariae (Rohde, 2005). Brittle stars (Ophiuroidea), which are frequently exploited by both 4Tand 4Vn plaice (Powles, 1965; Minet, 1975), are host to the metacercariae of Fellodistomum sp. (Koie, 1980), an important influence in our DFAs involving 4Vn plaice. Larval anisakine nematodes (A. simplex, C. osculatum. and P. decipiens), which proved significant in stock delineation herein, are acquired through predation on the parasite's invertebrate hosts, usually crustaceans (Rohde, 2005). Although the life cycle of A. simplex is largely pelagic, the larvae may be transmitted to de- 186 Fishery Bulletin 105(2) mersal fish, such as plaice, by diurnal vertical migrants (euphausiids, shrimp, etc.) which feed pelagically at night but are found near the seafloor by day (McClel- land, 1990). Larval P. decipiens infections are acquired through consumption of various benthic invertebrates, including mysids, isopods, amphipods, decapods, and polychaetes (McClelland, 2002) The acanthocephalan E. gadi, which was the most significant species in the classification of 4T plaice stocks in this study, reach maturity in the intestines of dozens of species of marine fish in the North Atlantic and use gammaridean and caprellid amphipods and mysids as intermediate hosts (Marcogliese, 1994). Among influences on digenean infection parameters in fish are the distributions and abundances of mol- luscan intermediate hosts, where the parasites perform one or more generations of asexual reproduction. Brittle stars, which transmit Fellodistomum metacercariae to plaice, become infected by feeding on cercariae which develop in bivalve moUusks (Koie, 1980). In contrast to the other helminths used as markers here, the digenean S. baccatum, an important component of DFAs involving 4Vn plaice, is transmitted actively, through penetration of the skin by cercariae (Wolfgang, 1955). MoUusks that host the asexual reproduction of S. baccatum to the cercarial stage are whelks, especially the common or waved whelk (Buccinum undatum), which is widely distributed and commercially exploited in eastern Ca- nadian waters. Infection parameters of larval helminths in 4T-4Vn plaice may also be influenced by temporal and spatial distributions of the final hosts. The digenean S. bac- catum matures and reproduces in the intestines of large piscivorous fish such as sea raven (Hemitripterus americanus) and Atlantic halibut {Hippoglossus hippo- glossus) (Wolfgang, 1955), and the anisakine nematode A. simplex matures and reproduces in the stomachs of cetaceans (Rohde, 2005). Adults of the anisakines Contracaecum osculatum and P. decipiens, and the acanthocephalan C. strumosum occur, respectively, in stomachs and intestines of seals. Surveys of various demersals in waters off Nova Scotia, Canada (Marco- gliese and McClelland, 1992; McClelland et al., 2000; McClelland and Martell, 2001a&b) revealed that in- fection parameters of P. decipiens and Corynosoma wegeneri increased with proximity to Sable Island, site of the largest grey seal (Halichoerus grypus) colony in the Northwest Atlantic. Spatial disparities in prevalences and abundances of parasitic helminths in fish may also be traced to variation of physical parameters (temperature, salin- ity, depth, and bottom habitat) that influence distribu- tions of the invertebrate precursor hosts (Williams and Jones, 1994). Our 4T plaice samples were collected at relatively uniform depths (43-88 m), and the substrates in both sampling areas ranged from sandy pelite to sandy gravel and had outcroppings of sandstone bed- rock (see Loring and Nota^). The mean near-bottom temperature for sampling stations in eastern 4T was 1.62 (0.49-3. 23)°C (tt = 5), but only 0.21 (-0.01-0.77)°C (« = 9) for stations in western 4T. Near-bottom tempera- tures prevalent on the Magdalen Shoals are extremely low throughout the year (Swain et al., 1998), and small variations in temperature may have dramatic effects on the developmental and transmission rates of helminth parasites, as well as on the distributions and devel- opmental rates of their poikilothermic intermediate hosts. Hence, the fact that acanthocephalan infections in southeastern Gulf plaice were much heavier than those found in plaice from the northwestern Magdalen Shoals may, to some extent, reflect the relative warmth of waters occupied in eastern 4T. In summary, DFAs of abundances of the acantho- cephalans E. gadi and C. strumosum support the find- ings of earlier mark-recapture studies (Powles, 1965), which indicate the presence of distinct northwestern (Miscou-Magdalen) and southeastern (Cape Breton) plaice stocks in the southern Gulf of St. Lawrence. Moreover, both parasite markers could be employed in future studies of migration and mixing of the two 4T stocks within 4T, and infection parameters of E. gadi alone may prove useful for detecting the presence of southeastern 4T migrants among stocks overwinter- ing in Laurentian Channel waters of the Cabot Strait and 4Vn. The strength of our conclusions may be miti- gated, however, by the fact that 4T and 4Vn plaice were sampled only during September and May, respectively, and samples from the two areas were taken more than a year apart. Hence, the possibility of seasonal or lon- ger term variations in infection parameters of enteric helminths, e.g., Fellodistomum sp. and E. gadi, could not be investigated. Finally, given problems inherent in stepwise procedures (Power et al., 2005), statistical procedures employed in the present study, and in simi- lar studies, could be improved upon. Power et al. (2005), for example, adopted an "all possible subsets" approach to selection of indicator parasites used in linear and quadratic discriminant analyses and in nonparametric classification of bogue landed at Spanish fishing ports. Efforts will be made to address these shortcomings in future surveys. Acknowledgments The authors are grateful to T. Hurlbut, Fisheries and Oceans Canada, Moncton, for supervising the collection of plaice samples during the 2004 groundfish cruise in the southern Gulf of St. Lawrence. We also thank R. Morin, Fisheries and Oceans Canada, Moncton, and four anonymous referees for their helpful comments and advice on earlier drafts of the manuscript. •' Loring, D. H., and J. G. Nota. 1973. Morphology and sedi- ments of the Gulf of St. Lawrence. Fisheries Research Board of Canada Bulletin, 182, 147 p. 116 Lisgart Street, Ottawa, Canada, KIA 0H3. McClelland and Melendy, Heminths as tags In delineating stock of Hippog/ossoides platessoides 187 Literature cited Arthur, J. R., and E. Albert. 1993. Use of parasites for separating stocks of Greenland halibut {Reinhardtius hippoglossoides) in the Cana- dian northwest Atlantic. Can. J. Fish. Aquat. Sci. 50:2175-2181. Blaylock, R. B., L. Margolis, and J. C. Holmes. 2003. The use of parasites in discriminating stocks of Pacific halibut (Hippoglossus stenolepis) in the northeast Pacific. Fish. Bull. 101:1-9. Boje, J., F. Riget, and M. Koie. 1997. Helminth parasites as biological tags in population studies of Greenland halibut [Reinhardtius hippoglos- soides (Walbaum)), in the northwest Atlantic. ICES J. Mar. Sci. 54:886-895. 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. Clay, D. 1991. Seasonal distribution of demersal fish (Osteich- thyes) and skates (Chondrichthyes) in the southeastern Gulf of St. Lawrence. In The Gulf of St. Lawrence: small ocean or big estuary? (J. C. Therriault, ed.), p. 241-259. Can. Spec. Pub. Fish. Aquat. Sci. 113. Khan R. A., and C. Tuck. 1995. Parasites as indicators of stocks of Atlantic cod (Gadus morhua) off Newfoundland, Canada. Can. J. Fish. Aquat. Sci. 52:195-201. Koie, M. 1980. On the morphology and life-history of Steringo- trenia pagelli (Van Beneden, 1871) Ohdner, 1911 and Fellodistomum fellis (Olsson, 1868) Nicoll, 1909 [syn. S. ovacutum (Lebour, 1908) Yamaguti, 1953] (Trematoda, Fellodistomidae). Ophelia 19:215-236. Kulachkova, V. G., and T. A. Timofeyeva. 1993. The acanthocephalan Echinorhynchus gadi (Zoega) from relict cod of Lake Mogil'noye. Can. Transl. Fish. Aquat. Sci. no. 5602. Li, J. C. R. 1964. Statistical inference. Vol. 1, 658 p. Edwards Bros., Ann Arbor, ML MacKenzie, K. 1987. Parasites as indicators of host populations. Int. J. Parasit. 17:345-352. MacKenzie, K., and P. Abaunza. 1998. Parasites as biological tags for stock discrimination of marine fish: a guide to procedures and methods. Fish. Res. 38:45-56. 2005. Parasites as biological tags. In Stock identifica- tion methods. Applications in fisheries science. (S. X. Cadrin, K. D. Friedland, and J. R. Waldman, eds.), p. 211-226. Elsevier Academic Press, San Diego, CA. Marcogliese, D. J. 1994. Aeginina longocornis (Amphipoda: Caprellidea), new intermediate host for Echinorhynchus gadi (Acanthoceph- ala: Echinorhynchidae). J. Parasitol. 80:1043-1045. Marcogliese, D. J., E. Albert, P. Gagnon, and J. M. Sevigny. 2003. Use of parasites in stock identification of the deepwater redfish iSebastes mentella) in the Northwest Atlantic. Fish. Bull. 101:183-188. Marcogliese, D. J., and G. McClelland. 1992. Corynosoma wegeneri (Acanthocephala: Poly- morphida) and Pseudoterranova decipiens (Nematoda: Ascaridoidea) larvae in Scotian Shelf groundfish. Can. J. Fish. Aquat. Sci. 49:2062-2069. Margolis, L., and N. P. Boyce. 1969. Life span, maturation and growth of two hemiurid trematodes, Turbulovesicula lindbergi and Lecithaster gibbosus, in Pacific salmon (Genus Oncorhynchus). J. Fish. Res. Board Can. 26:893-907. McClelland, G. 1990. Larval anisakine nematodes in various fish species from Sable Island Bank and vicinity. In Population biol- ogy of sealworm i Pseudoterranova decipiens) in relation to its intermediate and seal hosts (W. D. Bowen, ed.), p. 83-118, Can. Bull. Fish. Aquat. Sci. 222. 2002. The trouble with sealworms {Pseudoterranova decip- iens, species complex, Nematoda): a review. Parasitol. 124(suppl.):183-203. McClelland, G., and D. J. Martell. 2001a. Surveys of larval seal-worm, Pseudoterranova decipiens infection in various fish species sampled from Nova Scotian waters between 1988 and 1996, with an assessment of examination procedures. In Sealworms in the North Atlantic: ecology and population dynamics (G. Desportes, and G. McClelland, eds. ), p. 57-76. North Atlantic Marine Mammal Commission Scientific Pub- lications, Troms0, Norway. 2001b. Spatial and temporal distributions of larval seal- worm, Pseudoterranova decipiens (Nematoda: Anisaki- nae) in Hippoglossoides platessoides (Pleuronectidae) in the Canadian Maritime Region from 1993 to 1999. In Sealworms in the North Atlantic: ecology and population dynamics (G. Desportes, and G. McClelland, eds.), p. 77-94. North Atlantic Marine Mammal Commission Scientific Publications, Troms0, Norway. McClelland, G., J. Melendy, J. Osborne, D. Reid, and S. Douglas. 2005. Use of parasite and genetic markers in delineat- ing populations of winter flounder from the central and south-west Scotian Shelf and north-east Gulf of Maine. J. Fish Biol. 66:1082-1100. McClelland, G., R. K. Misra, and D. J. Martell. 2000. Spatial and temporal distributions of larval seal- worm [Pseudoterranova decipiens, Nematoda: Anisaki- nae), in Hippoglossoides platessoides (Pleuronectidae) in eastern Canada from 1980 to 1990. ICES J. Mar. Sci. 57:69-88. Melendy, J., G. McClelland, and T. Hurlbut. 2005. Use of parasite tags in delineating stocks of white hake (Urophycis tenuis) from the southern Gulf of St. Lawrence and Cape Breton Shelf. Fish. Res. 76: 392-400. Minet, J-P. 1975. Data on the biology of American plaice from the southern bank of Newfoundland and from the Cape Breton shelf Rev. Trav. Inst. Peches Marit. Nantes. 38: 343-434. Neter, J., W. Wassermann, and M. H. Kutner. 1985. Applied linear statistical models: regression, analysis of variance, and experimental design, 1127 p. Richard Irwin Inc., Homewood, IL. Olson, R. E., and I. Pratt. 1971. The life cycle and larval development of Echino- rhynchus lageniformis Ekbaum, 1938 (Acanthocephala: Echinorhynchidae). J. Parasitol. 57:143-149. Piatt, N. E. 1975. Infestation of cod [Gadus morhua L.) with the larvae of codworm iTerranova decipiens) and herringworm, Anisakis sp. (Nematoda: Ascaridata) in North Atlantic and Arctic waters. J. Appl. Ecol. 12:437-450. 188 Fishery Bulletin 105(2) Power, A. M., J. A. Balbuena, and J. A. Raga. 2005. Parasite infracommunities as predictors of harvest location of bogue (Boops hoops L.): a pilot study using statistical classifiers. Fish. Res. 72:229-239. Powles, P. 1965. Life history and ecology of American plaice, (Hippo- glossoides platessoides F. ) in the Magdalen Shallows. J. Fish. Res. Board Can. 22:565-596. Rohde, K. 2005. Marine Parasitology, 565 p. Commonwealth Sci- entific and Industrial Research Organisation Publish- ing, Collingwood, Victoria, Australia. Scott, J. S. 1975. Geographic variation in incidence of trematode parasites of American plaice ( Hippoglossoides plates- soides) in the Northwest Atlantic. J. Fish. Res. Board Can. 32:547-550. Scott, W. B., and M. G. Scott. 1988. Atlantic fishes of Canada. Can. Bull. Fish. Aquat. Sci. 219:1-731. Sokal, R. R., and F. J Rohlf 1969. Biometry, 797 p. W.H. Freeman, San Francisco, CA. Stott, W., M. M. Ferguson, and R. F. Tallman. 1992. Genetic population structure of American plaice (Hippoglossoides platessoides) from the Gulf of St. Lawrence, Canada. Can. J. Fish. Aquat. Sci. 49:2538-2545. Swain, D. P., G. A. Chouinard, R. Morin, and K. F. Drinkwater. 1998. Seasonal variation in the habitat associations of Atlantic cod (Gadus morhua) and American plaice {Hippoglossoides platessoides) from the southern Gulf of St. Lawrence. Can. J. Fish. Aquat. Sci. 55: 2548-2561. Tallman, R. F. 1991. Causes of within season decline in the size at age of American plaice, Hippoglossoides platessoides. In The Gulf of St. Lawrence: small ocean or big Estuary? (J. C. Therriault, ed.), p. 269-275. Can. Spec. Pub. Fish. Aquat. Sci. 113. Titus, K., J. A. Mosher, and B. K. Williams. 1984. Chance corrected classification for use in discrimi- nant analysis: ecological applications. Am. Midi. Nat. 111:1-7. Williams, H., and A. Jones. 1994. Parasitic worms of fish, 593 p. Taylor and Francis Ltd., London, England. Williams, H. H., K. MacKenzie, and A. M. McCarthy. 1992. Parasites as biological indicators of the popu- lation biology, migrations, diet, and phylogenetics of fish. Rev. Biol. Fish. 2:144-176. Wolfgang, R. W. 1955. Studies of the trematodes Stephanostomum bac- catum (NicoU, 1907) IIL Its life cycle. Can. J. Zool. 33:113-128. Zubchenko, A. V. 1985. Parasitic fauna of American plaice (Hippoglossoides platessoides) from the Northwest Atlantic. J. Northw. Atl. Fish. Sci. 6: 165-171. 189 Abstract — Understanding the in- teractions between kelp beds and nearshore fish is essential because anthropogenic changes and natural variability in these beds may affect available habitat for fishes. In this study fish communities were inves- tigated in south-central Alaska kelp beds characterized by a range of sub- strate complexity and varying den- sities of both perennial understory kelps and annual canopy kelps. Many of the observed fish species, as well as understory and canopy kelps, were positively associated with structurally complex substratum. Targeted canopy and understory kelp beds supported seasonal populations of adult and juvenile Pacific cod (Gadus macro- cephalus), rockfishes (Sebastes spp.), and year-round populations of green- lings {Hexagrammos spp. I. Monthly changes in kelp and fish communi- ties reflected seasonal changes; the densities of some species were great- est during periods with higher tem- peratures. This work illustrates the importance of structurally complex kelp beds with persistent understory kelp populations as important fish habitat for several commercially and recreationally important fishes. Implications of substrate complexity and kelp variability for south-central Alaskan nearshore fish communities Judy Hamilton Kachemak Bay National Estuanne Research Reserve 95 Sterling Highway, Suite 2 Homer, Alaska 99603 Email address: judy_hamilton@fishgame.state.ak.us Brenda Konar School of Fisheries and Ocean Sciences University of Alaska Fairbanks 245 O'Neill Building Fairbanks, Alaska 99775 Manuscript submitted 10 September 2004 to the Scientific Editor. Manuscript approved for publication 29 August 2006 by the Scientific Editor. Fish. Bull. 105:189-196 2007). Marine macroalgal communities in the shallow, rocky, nearshore zones are among the most productive aquatic biomes on earth and provide impor- tant habitat for invertebrates, fishes, and marine mammals (Steneck et al., 2002). Although the importance of kelp bed variability (including kelp density, distribution, and species com- position) to fishes has been demon- strated in Alaskan waters (Dean et al., 2000; Hegwer, 2003; Hamilton, 2004; Calvert, 2005), the role of sea- sonality in these habitats is poorly understood, particularly in regions with seasonal extremes, such as the subarctic (but see Calvert, 2005). Fur- thermore, although the persistence and stability of kelp beds are at least partly determined by suitable space and substratum type (Dayton, 1985), the importance of overall habitat com- plexity (i.e., kelp cover and substrate topography) to kelp-associated fish species has not been investigated in Alaska. Previous studies in Alaskan kelp beds have shown positive correla- tions between the presence of fishes and the density (or biomass) of un- derstory algae (Dean et al., 2000; Hegwer, 2003; Hamilton, 2004). Re- searchers elsewhere have also agreed with these findings (Dayton, 1985) but have demonstrated relationships between fish density or biomass and the relative abundance of the canopy kelp Macrocystis pyrifera (Bodkin, 1986; Carr, 1994). Macrocystis forms dense stands that are generally stable and provide persistent habitat; beds composed of this perennial species exhibit relatively little seasonal and annual variation in structure (Day- ton, 1985; Steneck et al., 2002). In contrast, northern Pacific canopy kelps are annuals (Alaria fistulosa and Nereocystis luetkeana [hereafter Nereocystis]) that afford much less midwater structure. As a result, ver- tical relief in northern Pacific kelp beds is often seasonally restricted to, and is more consistently provided by, physical structure of the seafloor and the perennial understory kelp species. The importance of physical structure (described by the measures of rugos- ity, substrate size, and verticality) to temperate and tropical reef fish assemblages has been documented (Aburto-Oropeza and Balart, 2001; Garcia-Charton and Perez-Ruzafa, 2001), but little is known about the importance of the physical structure of kelp and substrate to fishes in the northern Pacific rocky nearshore zones. The objectives of this study were to assess relationships of fish to habitat structure and to seasonal variability in kelp communities. We determined the relationship of fishes to habitat structure (classified according to ru- gosity, size of substrate, and vertical- ity) and kelp densities. Because north Pacific macroalgal communities vary 190 Fishery Bulletin 105(2) seasonally with the growth and senescence of annual kelps, changes in the kelp com- munities were correlated with associated fish abundances. Materials and methods Study sites This study was conducted in Kachemak Bay, the southernmost inlet on the western shore of the Kenai Peninsula, in south-cen- tral Alaska. Ten sites were chosen based on their structural characteristics and the presence of kelp communities (Fig. 1). All sites contained understory kelp, providing varying degrees of macroalgal cover, and five sites contained the canopy-forming kelp Nereocystis. A distance of at least 200 m (predominantly sandy bottom) separated all sites from each other. Sites were situated at a water depth of approximately 7 m. Study design At each site, transects (77 = 3) were surveyed monthly to quantify kelp densities and fish presence between May 2002 and Septem- ber 2003. A haphazard starting point was selected for each transect from which a random direction was taken. Although visibility varied among sampling periods, transects were surveyed when visibility was at least a transect width (2 m on each side) or more and therefore such visibility was not included in the analyses. Because of turbidity and poor visibility at this site, MacDonald Spit was not sampled in July and November 2002 and Anisom Point was not sampled in October 2002. Each survey had two components (a kelp and a fish survey), which were conducted concur- rently by two separate divers. Kachemak Bay TUT I SPI JAK LJA LTL^ " j,^(,3 e^j, N A OJA * ' ^-Little Tutka Bay ' Uttle Jakolof Bay < lakolof Bay Figure 1 Location of Kachemak Bay and study sites. Sites characterized by high structural complexity of the substratum are denoted with a circle and low structural complexity sites are denoted with a square. Study sites are abbreviated as follows: ANI=Anisom Point; HER = Herring Islands; HES = Hesketh Island; JAK= Jakolof Bay; LJA=Little Jakolof Bay; LTU = Little Tutka Bay; OJA=Outside Jakolof Bay; SAD = Sadie Cove; SPI = MacDonald Spit; and TUT=Tutka Bay. Physical habitat variables Physical habitat variables (rugosity, substrate size, and verticality) were measured once for each site in September 2003. Rugosity and substrate size were measured for every quadrat at all sites during Sep- tember 2003. Rugosity provides a measure of habitat complexity on a small spatial scale and is defined as the ratio of the true distance contour along the bottom to a one-meter horizontal distance (Leum and Cheat, 1980). Rugosity was measured by using a 1-m bar with a series of 5-mm links attached at one end. The bar was held horizontally with the link end resting on the substrate. The links were then draped along the substrate beneath the bar. These links were counted and a rugosity measure was calculated for each quad- rat and averaged per transect. Substrate size was determined by measuring the diameter of samples of the bottom relief (e.g., sand, cobble, bedrock) that composed greater than 50% of the quadrat (Garcia- Charton and Perez-Ruzafa, 2001). When no substrate type dominated, the percentages and sizes of each substrate type were noted. These measurements were categorized from one (sand or silt) to five (bedrock) and an average value was calculated for each site. Verti- cality, a subjective measure ranging from one (for low structural relief) to five (high), was assigned to each site (Bodkin, 1986). Monthly water temperature was also measured at each site. Hamilton and Konar: Implications of substrate complexity and kelp variability for Alaskan nearshore fish communities 191 Surveys of kelp and fish For surveys of kelp, randomly placed 0.25-m2 quadrats {n = 10) were examined per 120-m- transect. All under- story kelps in each quadrat were counted and identified to species. Because all understory kelp species were structurally similar (in size and overall shape), they were grouped as "annual" iCostaria costata, Cymathere triplicata, and Laminaria saccharina,) or "perennial" {Agarum clathratum, L. bongardiana, and L. yezoensis) understory for statistical analyses. All data collected in May 2002 were omitted from the analyses involving kelp because understory data for that month were incomplete. Analyses were conducted on the average understory kelp densities per transect to enable comparison with the relatively sparse densities of the canopy kelp and fish communities. Because Nereocystis was relatively rare, all individuals were counted within each 120-m- band transect. For fish surveys, all fishes observed within each tran- sect and within one meter of the bottom (30 mx4 mx 1 m=120 m-^) were enumerated and identified to species whenever possible. Because few fishes were observed, the three most abundant families (Hexagrammidae, Scorpaenidae. and Gadidae) were analyzed by family group. All other fishes were rarely observed and were grouped as "other fishes" for the analyses. Statistical analyses Statistical analyses were performed by using multivari- ate approaches and linear models with STATISTICA vers. 6 (StatSoft, Tulsa, OKI. Cluster analyses were used to examine site variability in the kelp and fish communities and how this variability relates to structural complex- ity. Averages of all data were calculated by site across month and year for the ordination analyses. Kelp and fish densities were considered by species with the physi- cal variables of rugosity, substrate size, and verticality. The Bray and Curtis dissimilarity coefficient (Bray and Curtis, 1957) was used and the Euclidean distance was calculated for physical variables, kelp, and fish. Water temperature did not vary among sites within months and was not used in our analysis. A one-way ANOVA was used for temporal variation of water temperature. Partial correlation analysis (with Pearson's correlation coefficient, r) was performed between kelp groups (based on average annual understory, perennial understory, and canopy kelp densities [no./120 m'-]) and physical habitat data (rugosity, substrate size, and verticality: average values per 120 m-; water temperature: °C per month) while controlling for potentially intercorrelating variables. These results were considered significant at a<0.05. Because of the low number offish observed, fish counts were converted to presence or absence data and logistic regression was applied. Independent variables were the four log-transformed physical variables (rugos- ity, substrate size, and verticality: logjg [average values per 120 m-]; temperature: logjQ [value per month],) and the three log-transformed kelp groups (annual under- story kelp, perennial understory kelp, and canopy kelp: logjg [number of kelps per 120 m^]). Analyses were conducted separately for each fish family that composed at least 20% of total abundance (Hexagrammidae, Scor- paenidae, and Gadidae). Results Physical habitat variables Substrate in the study sites varied from complex (rocky outcrops, large boulders, and bedrock) to homogeneous (small cobble and sand). The ten sites were partitioned by clustering techniques into two general structural complexity groups based on dissimilarities among the three measured structural characteristics (rugosity, substrate size, and verticality; Fig. 2A). Water tempera- ture varied significantly among months (Fjg 45g=1983.2, P<0.001). Temperature ranged from 1.8°C in winter to 11.0°C in summer and was the only physical variable that did not vary among sites. Water temperatures also differed significantly between years (Fj 45g=1028.6, P<0.001), and were higher in 2003 than 2002. Surveys of kelp and fish A comparison of cluster dendrograms revealed patterns of spatial variation among the kelp and fish groups that mirrored the substrate trends. When all biological data (understory and canopy kelp densities and fish pres- ence) were averaged across months and years, five sites grouped with higher counts of kelp and fish exhibited the greatest structural complexity (Fig. 2, B [kelp] and C [fish]). Similarly, three of the structurally homogenous sites were grouped consistently with lower values for both kelp and fish. Two sites (LJA and OJA) showed inconsistencies in these groupings. Little Jakolof Bay (with a lower complexity designation) was in the higher macroalgal count group but in the lower fish abundance group. Outside Jakolof Bay (with a higher complexity designation) was in the lower density groups for both kelp and fish. Kelp communities were variable in species composi- tion and density over space and time and understory kelp communities were considerably denser than the canopy kelp. Understory kelps were present every month and perennial kelp dominated in all months except late October 2002 (Fig. 3). The annual understory kelp C. costata contributed at most 2% to the annual kelp rela- tive abundance in any month, whereas L. saccharina composed at least 75%. Annual understory kelps were found in greatest densities during periods with warmer water temperatures; perennial kelps, however, were not significantly correlated with temperature (Table 1). Pe- rennial understory kelps were found on all transects; an overall equal contribution was made by A. clathratum and the perennial Laminaria species (L. bongardiana and L. yezoensis. densities lumped together). Both an- nual and perennial understory kelps were found in 192 Fishery Bulletin 105(2) greatest densities in sites with higher values for rugos- ity and verticality, although, curiously, not for substrate size (Table 1). Five sites (ANI, HER, HES, OJA, and SPI) contained Nereocystis in 2002, compared to two 80 A o o -- 80 P i 100 60 sad Ija Itu |ak OJA HER TUT HES SPI ANI B L-| — •— sad OJA Itu lak SPI HER TUT l|a HES ANI c sad Itu l|a OJA lak HER TUT HES SPI ANI Figure 2 Cluster dendrograms based on percent dissimilari- ties of variables among sites. (A) Structural habitat descriptors (rugosity, substrate size, and verticality); (B) Kelp communities; and (C) Fish communities. Site abbreviations indicate relative structural complexity (lower case = low complexity; upper case = high complex- ity. Study sites are abbreviated as follows: ANI=Anisom Point; HER = Herring Islands; HES = Hesketh Island; JAK= Jakolof Bay; LJA=Little Jakolof Bay; LTU=Little Tutka Bay; OJA=Outside Jakolof Bay; SAD = Sadie Cove; SPUMacDonald Spit; and TUT=Tutka Bay. sites (HER and SPI) in 2003. Canopy kelp persisted throughout the winter at one site (HER). The great- est number of canopy kelp individuals was observed in October 2002, the lowest densities (fewer than 5 Nereo- cystis/120 m- transect) were observed from November through April, and no canopy kelp was observed in May 2003 (Fig. 4). Canopy kelp was more abundant during months with higher water temperatures and in sites with larger substrate and greater vertical relief, but, unlike the understory kelp, was negatively correlated with rugosity (Table 1). The presence of some fishes was associated with season, year, physical habitat characteristics, and kelp. Four hundred twenty-two fishes representing 15 spe- cies from eight families were sighted on 34% (n = 171) of transects surveyed. Three families (Hexagrammidae [greenlings], Scorpaenidae [rockfishes], and Gadidae [codfishes]) each composed at least 20% of the total abundance and together accounted for more than 80% of all fishes sighted. Infrequently sighted fishes in- cluded those in the families Pholidae (gunnels, 6%), Cottidae (sculpins, 3%), Pleuronectidae (flatfishes, 2%), and others (including ronquils, searchers, and un- identified fishes, 5%). Fish presence varied over time (Table 2); more fishes were sighted in 2003 (2.59 ±6.92 fish/transect) than in 2002 (0.63 ±0.96 fish/transect). More fish (considering all fish species pooled across months, years, and sites) were seen during periods with higher temperatures and in sites characterized by larger substrate and greater densities of annual understory and canopy kelp (Table 2). Greenlings (pri- marily kelp greenling [Hexagrammos decagrammus]) accounted for the majority of sightings (35% of total abundance) and their presence did not differ among months (Table 2). Greenlings were most commonly seen in sites with low rugosity values and larger sub- strate (i.e., boulder to bedrock), and during periods with warmer water temperatures and higher densities of annual understory kelps. Schooling species, such as rockfishes (primarily the black rockfish [S. mela?iops]) and adult Pacific cod (Gadus macrocephalus), were observed infrequently. However, these groups accounted for the greatest num- ber of fish seen on any one transect and exhibited the greatest variability in sightings per month in the fish groups, primarily during summer 2003 (Fig. 5). There was no difference in the presence of rockfishes among months, but there were significant temporal differences for codfishes (Table 2). Considering the major families observed in this study, only the presence of rockfish- es showed significant annual variability (Table 2) — more in 2003 (1.23 ±5.83 fish/transect) than in 2002 (0.08 ±0.42 fish/transect). Both rockfishes and cod- fishes were most commonly seen during periods with higher water temperatures (Table 2). Although these results are based only on sightings of adult fishes, large schools (thousands of individuals) of juvenile codfishes (predominately G. macrocephalus) were ob- served at all sites during August and September 2002. The juvenile codfish schools observed in summer 2003 Hamilton and Konar: Implications of substrate complexity and kelp variability for Alaskan nearsfiore fisfi communities 193 Table 1 Partial correlation analysis (Pearson's correlation coefficient, r) between physical variables and kelp groups (n=473 transects). Values are considered significant at P < 0.05. Rugosity provides a small-scale measure of habitat complexity and is the ratio of the true distance contour along the bottom to a one-meter horizontal distance. Substrate size is based on the average diameter of the substrate comprising the majority of a 0.25-m2 quadrat. Verticality was assigned to each site on the basis of overall vertical relief. Kelp group Canopy Annual understory kelp Perennial understory kelp Temperature Rugosity of substrate Substrate size Verticality of su 0.18 -0.32 0.29 0.09 P<0.001 P<0.001 P<0.001 P<0.001 0.17 0.16 0.01 0.16 P= 0.001 P<0.001 P= 0.790 P<0.001 0.02 0.27 -0.05 0.26 P=0.685 P<0.001 P=0.325 P<0.001 were composed of much fewer individuals (at most, tens of individuals/school). These juvenile codfishes were not included in any analyses in our study because of difficulties in accurately quantifying them. Discussion Structural habitat complexity is important to both fish and kelps in that greater physical habitat complexity is associated with greater overall densities of fish in these communities. In particular, greenlings associated most consis- tently with kelp beds that had a predominately rocky (i.e., large cobble/bedrock) and structur- ally complex bottom habitat. The association of the dominant fish species in our study with relatively larger substrate and higher rock relief may indicate that south-central Alaska kelp-bed fishes follow the same trend (of a strong associa- tion offish with kelp and rocky substrate relief) documented elsewhere. For example, in Cali- fornia kelp beds, significant correlations exist between fish density and bottom relief (Ebeling et al„ 1980; Bodkin. 1986). However, rockfishes in Puget Sound inhabited low-relief rocky kelp beds during summer (Matthews. 1990) and in Prince William Sound were positively associ- ated with relief (Dean et al., 2000). Rockfishes and codfishes in the present study were never associated with any bottom structure, perhaps because of the sporadic sightings of these species. The lack of association of rockfishes and codfishes with any physical habitat variables or kelp may reflect the tran- sient nature and seasonal association of these fishes with kelp. However, the rarity of any fishes observed higher than one meter above the substratum in the present study indicates that bottom structure may be important to the fishes observed in our study, if perhaps indirectly, by also being appropriate substratum for kelp habitat. The variability of Alaska kelp communities and as- sociated fish populations may be partially attributable Figure 3 Monthly variation in the density (no. of plants/m^ ±standard error) of understory kelp groups, n = 503 transects. Because surveys were conducted during alternate neap tide cycles throughout the 17 month study, two surveys occurred, at least partly, during October 2002. to the extreme seasonal nature of the northern environ- ment. Increased sightings of fish during periods charac- terized by warmer water indicate seasonal variability in fish communities associated with kelp communities. For example, although greenlings were observed in the shallow, rocky nearshore sites during every month in the present study, rockfishes and codfishes were rarely observed except in summer. Healthy understory kelp populations exist in this area on rocky substratum at depths of up to 16 m (first author, personal observ.) and fish populations may shift seasonally to similar 194 Fishery Bulletin 105(2) ^.i°^.<*^^o^vvv"l*vv^-^"■'^»'V^^'^'VVvv' Figure 4 Monthly mean Nereocystis luetkeana density (no. of Nereocys- tis/120 m^ ±standard error). The mean was averaged across the five sites initially containing canopy, n = 248 transects. habitats in adjacent, deeper water. Puget Sound rockfishes tend to move to shallower water in summer and deeper water in winter (Moulton and MillerM, possibly avoiding increased storm surge during winter months. Fish communities inhabiting seasonally and annually variable kelp beds in the north Pacific must be capable of enduring a wider variety of environmental variables over the course of a season, year, or lifetime than those occupying the relatively stable, perennial canopy-dominated kelp beds of more temperate zones. The importance and magnitude of seasonal cues vary among kelp and fish species throughout their ranges and the cues include temperature, photoperiod, tur- bidity, increased frequency of storms and surge in winter, and prey and nutrient availability (Ebeling et al., 1980, Dayton, 1985). However, the thresholds of many environmental factors are to some extent temperature-dependent (Dayton, 1985), providing an easily quantified surrogate variable for seasonality in the present study. Kelp beds (and associated fishes) at the Miller, B. S., and L. L. Moulton. 1988. Charac- terization of Puget Sound fishes for the EPA Bay Program, p. 77-84. In Proceedings, First Annual Meeting on Puget Sound Research, vol. 1, Seattle, WA, March 18-19, 1988. Puget Sound Action Team, Office of the Governor, P.O. Box 40900, Olympia WA 98504-0900. Table 2 Logistic regression results on the presence of fishes by families and by total abundance. Only families composing greater than 207c of the total fish abundance are included. Independent variables are time (n = 503 transects for tests of "month" or n=208 transects for test of "year," the comparison of June to September of years 2002 and 2003 ), kelp densities (canopy kelp and annual and perennial understory kelp groups; /! = 473 transects), and physical habitat variables (water temperature, rugosity, substrate size, and verticality). Only significant values (P<0.05) are reported. Analysis group Independent variable Parameter estimate Standard error Wald's x^ P-value Hexagrammidae Scorpaenidae Gadidae Total fish Annual kelp understory Water temperature Rugosity Substrate size Year Water temperature Month Water temperature Month Year Canopy Annual understory Water temperature Substrate size 0.62 4.26 -2.48 1.90 0.64 9.00 0.27 4.28 0.08 0.28 0.66 0.21 5.12 0.80 0.17 0.91 1.15 0.47 0.29 3.00 0.08 1.62 0.03 0.13 0.32 0.10 0.78 0.38 12.65 <0.001 22.15 <0.001 4.66 0.031 16.11 <0.001 4.67 0.031 13.00 < 0.001 12.19 0.001 6.96 0.008 10.19 0.001 4.61 0.032 4.37 0.037 3.99 0.046 43.13 <0.001 4.49 0.034 Hamilton and Konar: Implications of substrate complexity and kelp variability for Alaskan nearshore fisfi communities 195 northern edge of their range are subject to wide fluctuations in all of these factors, as well as wide inter-annual variation in intensity and duration of seasonal factors. It is these extremes that make studying these habitats difficult or impossible dur- ing all but summer months and result in the paucity of consistent seasonal data in northern Pacific systems. Aerial surveys of kelp beds in Kache- mak Bay during three consecutive sum- mers (2000-2002) showed significant dif- ferences in size, location, and presence of Nereocystis canopy (Schoch and Chenelot, 2004), illustrating great interannual vari- ability that may be apparent on relatively short temporal scales. Such variability was also observed in the present study, in that three of five sites originally contain- ing canopy kelp did not recruit Nereocystis juveniles the second summer. Although this canopy kelp species is considered an annual, we observed Nereocystis individu- als reproducing into a second summer. Our findings of relatively more fishes inhabiting Nereocystis beds than under- story-only kelp beds indicates that areas characterized by enhanced Nereocystis growth may have greater fish densities. In one northern California study, densities of kelp greenling in Nereocystis beds were four times greater than in the present study (Bodkin, 1986). Nereocystis beds in California were similarly more important to rockfishes (Bodkin, 1986; Love et al., 1991) than understory kelp alone. How- ever, the existence of high understory kelp densities in the canopy-containing sites may be of greater importance to fishes than the canopy itself. In our study sites, both the greenlings and grouped fishes were positively associated with densities of annual understory. In addi- tion, fishes in the present study were usually observed in close association with the understory and substratum as has been typical in other south-central Alaska stud- ies (Rosenthal-, Dean et al., 2000). The perennial-domi- nated understory of south-central Alaska kelp beds may provide a degree of habitat stability for some fishes for at least part of the year. It is difficult to account for all factors influencing a natural system, particularly without knowing the re- cent history of the community. Because little is known 667 5 1 4 5 1 |BTotalfisfi □Hexagrammidae ■Scorpaenidae DGadidae 4 3 5 3 T E o JJ 25 o Z 20- 1 5 ■r 1 1 r T : 5 I c 1 i ^ It X j ■| I Lm 1 fl I L y Rii 1 c] L 1 ^* %<>* /VVVVVVV% A*^\.W\^*V Figure 5 Monthly variation of major fish families (no. offish/120 m^ ±standard error) averaged across sites. Only families comprising greater than 20% of total density were included, n = 503 transects. ^ Rosenthal, R. J. 1979. A preliminary assessment of compo- sition and food webs for demersal fish assemblages in several shallow subtidal habitats in lower Cook Inlet, Alaska, 58 p. In Final report by Dames and Moore, Inc., 800 Cordova Street, Suite 101, Anchorage, AK 99501. for Alaska Depart- ment of Fish and Game. Commercial Fisheries Division, 211 Mission Rd., Kodiak, Alaska 99615. about interactions between kelp and fish communities and their natural variability in south-central Alaska, investigation of more obvious, small-scale processes over an entire year is necessary. Physical factors, such as size of the kelp beds and related edge effects of the habitat, salinity fluctuations and freshwater runoff, degree and direction of exposure to light and tidal cur- rents, and the frequency of storm events may play a sig- nificant role in structuring these dynamic communities. In addition, biological factors that may influence algal community structure include inter- and intra-species competition and herbivory. A growing body of evidence points to the importance of temporal and spatial scales in ecological processes (i.e., Dayton and Tegner, 1984; Wiens and Addicott, 1986; Foster, 1990). The structur- ally complex kelp beds surveyed in the present study appear to provide critical habitat throughout the year for greenling species. However, this habitat is also sea- sonally important to rockfishes and codfishes (both adult and juvenile). 196 Fishery Bulletin 105(2) This work provides a description and baseline infor- mation on the structural characteristics of south-cen- tral Alaska's nearshore kelp beds and associated fish communities, and provides insight on the importance of seasonality. These findings may enable managers to identify potentially important nearshore fish habitat from relatively easily quantified structural habitat variables. The identification of critical habitat areas for juvenile and adult fishes is essential for sustainable management, and the importance of habitat structure in and influence of seasonality on these habitats has been further illuminated by this work. Acknowledgments This publication is the result of research sponsored by Alaska Sea Grant with funds from the National Oceanic and Atmospheric Administration Office of Sea Grant, Department of Commerce, under grant no. NA 16RG2321 (project number R/31-09) and from the University of Alaska with funds appropriated by the state. We wish to thank S. Hills, G. Haas, and A. Blanchard, as well as numerous reviewers, divers, and logistical supporters for their priceless assistance. Literature cited Aburto-Oropeza, O., and E. F. Balart. 2001. Community structure of reef fish in several habi- tats of a rocky reef in the Gulf of California. Mar. Ecol. 22:283-305. Bodkin, J. L. 1986. Fish assemblages in Macrocystis and Nereocys- tis kelp forests off central California. Fish. Bull. 84:799-807. Bray, J. R., and J. T. Curtis. 1957. An ordination of the upland forest communities of southern Wisconsin. Ecol. Mono. 27:325-329. Calvert, E. L. 2005. Kelp beds as fish and invertebrate habitat in southeastern Alaska. M.S. thesis, 113 p. Univ. Alaska Fairbanks, Juneau, AK. Carr, M. H. 1994. Effects of macroalgal dynamics on recruitment of a temperate reef fish. Ecology 75:1320-1333. Dayton, P. K., and M. J. Tegner. 1984. Catastrophic storms. El Nino, and patch stabil- ity in a southern California kelp community. Science 224:283-285. Dayton, P. K. 1985. Ecology of kelp communities. Annu. Rev. Ecol. Sys. 16:215-245. Dean, T. A., L. Haldorson, D. R. Laur, S. C. Jewett, and A. Blanchard. 2000. The distribution of nearshore fishes in kelp and eelgrass communities in Prince William Sound, Alaska: associations with vegetation and physical habitat characteristics. Environ. Biol. Fishes 57:271-287, Ebeling, A. W., R. J. Larson, W. S. Alevizon, and R. N. Bray. 1980. Annual variability of reef fish assemblages in kelp forests off Santa Barbara, California. Fish. Bull. 78:361-367. Foster, M.S. 1990. Organization of macroalgal assemblages in the northeast Pacific: the assumption of homogeneity and the illusion of generality. Hydrobiologia 192:21-33. Garcia-Charton, J. A., and A. Perez-Ruzafa. 2001. Spatial pattern and the habitat structure of a Mediterranean rocky reef fish local assemblage. Mar. Biol. 138:917-934. Hamilton, J. A. 2004. Kelp bed variability and fish population dynamics in Kachemak Bay, Alaska. M.S. thesis, 57 p. Univ. Alaska Fairbanks, Fairbanks, AK. Hegwer, C. L. 2003. Seasonal abundance and diversity of nearshore fishes around Steller sea lion haulouts of Kodiak Island. M.S. thesis, 76 p. Univ. Alaska Fairbanks, Fairbanks, AK. Leum, L. L., and J. H. Choat. 1980. Density and distribution patterns of the temper- ate marine fish Cheliodactylus spectabilis in a reef environment. Mar. Biol. 57:327-337. 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. 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. Schoeh, G. C, and H. Chenelot. 2004. The role of estuarine hydrodynamics in the dis- tribution of kelp forests in Kachemak Bay, Alaska. J. Coast. Res. 45:179-194. Steneck, R. S., M. H. Graham, B. J. Bourque, D. Corbett, J. M. Erlandson, J. A. Estes, and M. J. Tegner. 2002. Kelp forest ecosystems: biodiversity, stability, resil- ience, and future. Environ. Cons. 29:436-459. Wiens, J. A., and J. F. Addicott. 1986. Overview of the importance of spatial and temporal scale in ecological investigations. In Community ecology (J. Diamond, and T J. Case, eds.), p. 145-153. Harper and Row, New York, NY. 197 Abstract — Variation in the allele frequencies of five microsatellite loci was surveyed in 1256 individual spot- ted seatrout (Cynoscion nebulosus) obtained from 12 bays and estuaries from Laguna Madre. Texas, to Char- lotte Harbor. Florida, to St. John's River on the Florida Atlantic Coast. Texas and Louisiana collection sites were resampled each year for two to four years (1998-2001). Genetic dif- ferentiation was observed. Spotted seatrout from Florida waters were strongly differentiated from spotted seatrout collected in Louisiana and Texas. The greatest genetic disconti- nuity was observed between Tampa Bay and Charlotte Harbor, and Char- lotte Harbor seatrout were most simi- lar to Atlantic Coast spotted seatrout. Texas and Louisiana samples were not strongly structured within the north- western Gulf of Mexico and there was little evidence of temporal differentia- tion within bays. These findings are contrary to those of earlier analyses with allozymes and mitochondrial DNA (mtDNA) where evidence of spatial differentiation was found for spotted seatrout resident on the Texas coast. The differences in genetic struc- ture observed among these markers may reflect differences in response to selective pressure, or may be due to differences in underlying genetic processes. Genetic variability in spotted seatrout iCynoscion nebulosus), determined with microsatellite DNA markers Rocky Ward (contact author) Kevin Bowers Rebecca Hensley Brandon Mobley Ed Belouski Email address for R. Ward: rward@usgs.gov Texas Parks and Wildlife Department Coastal Fisheries Division 4200 Smitti School Road Austin, Texas 78744 Present address for R. Ward Northern Appalachian Research Laboratory USGS/Leetown Science Center 176 Straight Run Rd. Wellsboro, Pennsylvania 16901 Manuscript submitted 23 February 2006 to the Scientific Editor's Office. Manuscript approved for publication 11 September 2006 by the Scientific Editor. Fish. Bull. 105:197-206 (2007). Spotted seatrout (Cynoscion nebulo- sus) support an important recreational fishery in the northern Gulf of Mexico and along the U.S. Atlantic Coast. Management of this fishery is multi- jurisdictional, employing a variety of strategies including reduction or elim- ination of commercial exploitation, adjustments of recreational fish-size limits and bag limits, closed seasons, and artificial spawning and stock- ing of fish (Vanderkooy and Muller, 2003). Effective management requires an understanding of the ecology, life history, and genetic structuring of a species. Understanding genetic popu- lation structure is important to every aspect of fishery management but is especially critical when stocking fish is the chosen management strategy. Stocking without regard for existing genetic variability within and among populations places the genetic integ- rity of the targeted species at risk (Allendorf et al., 1986). Genetic popu- lation structuring may be evidence of adaptation to past environmental differences, whereas genetic variabil- ity may enable a population to meet future environmental challenges. Evidence of population structuring in spotted seatrout has been gained through morphological, physiological, and genetic examinations. Regional differences have been found in oto- lith and scale structure, growth rate (Iverson and Tabb, 1962; however, see Murphy and Taylor, 1994 for a differ- ent interpretation), and in reproduc- tive physiology (Brown-Peterson et al., 2002). Each of these studies found evidence of biologically significant re- gional differentiation — a finding con- sistent with observations of limited movement within and between bays (Music, 1981; Overstreet, 1983; Baker and Matlock, 1993) and between bays and adjacent nearshore waters (Baker et al., 1986). Studies of genetic markers general- ly support the existence of population structuring among spotted seatrout in the northern Gulf of Mexico. Stud- ies examining protein variation found evidence of weak (Ramsey and Wake- man, 1987; King and Pate, 1992) to strong (Weinstein and Yerger, 1976) population subdivision. The King and Pate (1992) study found indi- cations of clinal variation in mean heterozygosity and in alleles of the aspartate aminotransferase-2 locus, indicating possible adaptation to en- vironmental gradients on the Texas coast (King and Zimmerman, 1993). In each of the allozyme studies indi- cations of isolation by distance were found, as was found in a survey in which mitochondrial DNA was used (mtDNA; Gold et al., 1999). In the 198 Fishery Bulletin 105(2) mtDNA study, significant heterogeneity was found in haplotype frequencies among collection sites, indicating that spotted seatrout were spatially differentiated. In contrast, Gold et al. (2003) found no significant differ- ences in microsatellite DNA allele frequency among spotted seatrout inhabiting Texas bays. A similar study of microsatellite variation among spotted seatrout of the U.S. Atlantic and Florida Gulf coasts (employing different loci from those of Gold et al., 2003) found extensive spatial differences that coincided with known zoogeographic barriers (Wiley and Chapman, 2003). Surprisingly, the Indian River spotted seatrout sample on Florida's Atlantic coast was genetically more similar to the Choctawhatchee sample from the Florida Pan- handle than to more northerly Atlantic Coast samples. The overall pattern that emerges from applications of molecular markers to the examination of spatial genetic variability in spotted seatrout is mixed. Most studies have found limited gene flow between adjacent bays and moderate population subdivision, and pat- terns of differentiation that may be described as "isola- tion by distance." The failure of the Gold et al. (2003) to find statistically significant genetic subdivision is surprising, given that studies employing allozymes and mtDNA have been successful in discerning genetic structuring and given that microsatellites are consid- ered to be among the most capable genetic markers at resolving population level differentiation (Wright and Bentzen, 1994). Gold et al. (2003) focused on the northwestern Gulf of Mexico, and their geographically limited examination may have restricted the ability of the marker to discern patterns in the spatial variability of spotted seatrout. It is also possible that underlying genetic properties of the markers accounted for the differences in observed variation, or that some mark- ers (i.e., allozymes) were able to detect genetic adapta- tion to environmental gradients which microsatellites, assumed to be selectively neutral, were not. The present study is an attempt to re-examine genetic variability in spotted seatrout in the northern Gulf of Mexico. As did Gold et al. (2003) (who also examined a portion of the present data), we used microsatellite markers. Samples from Louisiana and the Gulf and Atlantic coasts of Florida were added in an attempt to give a greater spatial perspective that may be useful in evaluating the observed genetic variability among populations. In addition, samples from multiple years were included for Texas and Louisiana bays, allowing for evaluation of the stability and robustness of detected genetic structure. Materials and methods A total of 1256 individuals were examined. Sample collection sites are shown in Figure 1. All Texas sites were sampled in three consecutive years, 1998-2000, except Corpus Christi Bay which was sampled across four years, 1998-2001. The Louisiana site was sampled in 1999 and in 2000, and the Florida sites were sampled only in 2000. Texas and Florida samples were obtained through routine resource sampling efforts of Texas Parks and Wildlife Department and Florida Department of Natural Resources personnel, respectively, and Loui- siana samples were donated by licensed recreational anglers. Soft dorsal-fin tissue was removed from the fish, placed in 95% ethanol, and stored at room temperature until processed. Genomic DNA was extracted by using the PureGene DNA isolation kit and protocols (Centra Systems, Inc., Minneapolis, MN). Primers designed to amplify three microsatellites (Socl2, Soc50. and Soc243) originally developed for red drum {Sciaenops ocellatus) by Turner et al. (1998) were employed. Two additional primer pairs (C?!el33 and C7!el33') were designed by sequencing products of the Socl33 (Turner et al., 1998) forward and reverse primers and then identifying internal primer sequences that amplified separate repeat regions within the origi- nal iSocl33 amplicon. Primer sequences for Cnel33 and C7!el33' and protocols for amplification and interpreta- tion are discussed in Gold et al. (2003). Summary statistics were generated by using the Mi- crosoft Excel add-on Microsatellite Toolkit (Park, 2001) and ARLEQUIN, version 2.001 (Schneider et al., 2000), which was also employed to test frequencies for devia- tions from Hardy-Weinberg equilibrium by using exact tests performed with Markov-chain randomization (Guo and Thompson, 1992). Permutations with 1000 resam- plings (Manly, 1991) were used to generate probability values (P) for each test of Hardy-Weinberg equilibrium for each microsatellite locus in each sample. ARLE- QUIN was also used to test for linkage between micro- satellite loci, and significance of P-values was estimated by 1000 resamplings. Critical values for interpreting significance levels for simultaneous inferential compari- sons were adjusted by using the sequential Bonferroni approach (Rice, 1989). Allelic distribution homogeneity of each microsatellite was assessed with exact tests implemented with the statistical package GENEPOP, version 3.4 (Raymond and Rousset, 1995), and significance was estimated by permutation with 1000 resamplings for each com- parison. Population subdivision was estimated with Weir and Cockerham's (1984) theta (B) as generated in FSTAT, version 2.9.3 (Goudet, 1995), and a bootstrap procedure in FSTAT was employed to calculate a 95% confidence interval (CI). Although the use of theta is contigent on the assumption of an infinite-alleles mu- tation model (Kimura and Crow, 1964), it has been shown to compare favorably with other measures of genetic subdivision when employed with microsatellite data (Ruzzante, 1998). The significance of population differentiation across all loci was estimated as the com- bined probability of P-values for Fisher's exact tests for individual loci. Separate analyses were made for data sets that comprised 13 samples defined by site and col- lection date and 13 samples defined by site alone, with date combined across all year classes. A hierarchical analysis of gene diversity was per- formed by using the analysis of molecular variance Ward et al Genetic variability in Cynoscion nebulosus, determined with DNA markers 199 model (AMOVA; Michalakis and Ex- coffier, 1996) in ARLEQUIN (Excof- fier et al., 1992). The components of genetic diversity attributable to vari- ance between regions (Atlantic ver- sus Gulf of Mexico), variance among sampling sites within regions, tem- poral variance among years within sampling sites, and variance among individuals within samples were estimated. The significance of each variance component was tested with nonparametric permutation proce- dures (-1000 permutations; Excof- fier et al., 1992). In addition, genetic differentiation among all collection sites for each sampling year and be- tween pairs of populations within sampling years was estimated by using the theta statistic of Weir and Cockerham (1984) accessed on FSTAT (Goudet, 1995). Cavalli-Sforza and Edwards' chord distance (Df.; Cavalli-Sforza and Edwards. 1967) was used to re- construct phylogenetic relationships among collection sites. Estimations of D(. were obtained with the statis- tical package NJBPOP (Cornuet et al., 1999). Takezaki and Nei (1996) found D(, to be a better estimate of genetic divergence with microsatel- lite DNA data than with measures based on the step-wise mutation model. This estimate is not based on the assumption of a constant population size or a constant mutation rate among loci (Takezaki and Nei, 1996) and appears to accurately resolve closely related populations (Paetkau et al., 1997; Angers and Bernatchez, 1998). A phenogram was generated from the chord-distance matrix with the neighbor-joining (N-J) algorhithm. Robustness of each node was evalu- ated by bootstrapping over loci for 2000 replications (Hedges, 1992) with the SEQBOOT program on PHY- LIP, version 3.5c (Felsenstein, 1995). The PHYLIP program CONSENSE then was used to generate a consensus tree which was drawn with the program TREEVIEW (Page, 1996). Results The number of alleles per sample (Table 1) exceeded those reported for the same loci in red drum (Sciaenops ocellatus), the species of origin for the markers (Turner et al., 1998). Mean observed heterozygosity (Hq) ranged from 0.21 to 0.39, and there were no statistically sig- nificant deviations from Hardy-Weinberg expectations at any locus and sample combination after Bonferroni adjustment. Without the Bonferroni correction, allele frequencies at 16 of 165 comparisons would have failed to Figure 1 Sampling localities for spotted seatrout ^Cynoscion nebulosus) examined from the northern Gulf of Mexico and the Atlantic Coast of Florida. LL = lower Laguna Madre; UL = upper Laguna Madre; CC = Corpus Christi Bay; AB = Aransas Bay; SA = San Antonio Bay; MB = Matagorda Bay; EM = east Matagorda Bay; GB = Galveston Bay; SL = Sabine Lake; LA = Grand Isle, Louisiana; FT = Tampa Bay, Florida; FC = Charlotte Harbor. Florida; FS = St. John's River, Florida. meet expectations. This included all loci except Soc243, which was observed to be in Hardy-Weinberg equilib- rium in all samples. Soc012 most often failed to meet Hardy-Weinberg expectations; six of 33 samples were out of equilibrium before Bonferroni adjustment. Observed heterozygosity was lower than expected in 14 of the 16 locus-and-sample combinations that failed to meet Hardy-Weinberg expectation before adjustment. Sta- tistically significant linkage disequilibrium was noted for one pair of loci in one sample {Cnel33' with Socl2 in sample GB99) after Bonferroni adjustments. If the unadjusted critical value (a=0.05) was applied, 15 of 330 comparisons were statistically significant. Interestingly, the associated loci Cnel33 and Cnel33' did not exhibit linkage disequilibrium for any sample. After Bonferroni adjustment, exact tests for allele distribution homogeneity across all 33 samples dem- onstrated statistically significant differentiation for all microsatellite loci except Soc50, which approached statistical significance (P=0.05). All theta estimates were significantly greater than zero after Bonferroni adjustments, as was the overall theta of 0.116 (95% CI, 0.007-0.073; P<0.001), indicating significant genetic differentiation across both the spatial and temporal di- mensions sampled in this study. When spatial samples were collapsed to form 13 spatial samples, all loci except Soc50 exhibited statistically significant deviations from 200 Fishery Bulletin 105(2) Table 1 Summary statistics for spotted seatrout {Cynoscion nebulosus) samples included in a survey of microsatellite variation, n number of individuals in sample; n^ = number of alleles; P^ ferri-corrected critical value of a = 0.0003. : probability of meeting Hardy-Weinberg expectations (with Bon- Sample location Sample acronym St. Johns River, Florida FSOO 26 2000 Charlotte Harbor, Florida FCOO 22 2000 Tampa Bay, Florida FTOO 40 2000 Grand Isle, Louisiana LA99 60 1999 Grand Isle, Louisiana LAOO 39 2000 Sabine Lake, Texas SL98 23 1998 Sabine Lake, Texas SL99 36 1999 Sabine Lake, Texas SLOO 30 2000 Galveston Bay, Texas GB98 28 1998 Galveston Bay, Texas GB99 37 1999 Galveston Bay, Texas GBOO 40 2000 E. Matagorda Bay, Texas EM98 33 1998 E. Matagorda Bay, Texas EM99 40 1999 E. Matagorda Bay, Texas EMOO 40 2000 Matagorda Bay, Texas MB98 40 1998 Matagorda Bay, Texas MB99 40 1999 Matagorda Bay, Texas MBOO 40 2000 San Antonio Bay, Texas SA98 40 1998 San Antonio Bay, Texas SA99 40 1999 San Antonio Bay, Texas SAOO 40 2000 Aransas Bay, Texas AB98 40 1998 Aransas Bay, Texas AB99 39 1999 Aransas Bay, Texas ABOO 40 2000 Corpus Christi Bay, Texas CC98 40 1998 Corpus Christi Bay, Texas CC99 50 1999 "a "hw "a 'HW "a "a Phw '\ "a Phw '^a ^HW "<■ Phw "a Phw 40 40 'HV/ 40 "o "a "a "a Phw ' HW Soc Soc Soc Cne Cne 12 50 243 133 133' 3 2 3 3 3 0.32 0.19 0.13 1.00 0.11 3 2 2 3 2 0.18 1.00 1.00 1.00 1.00 3 2 2 4 2 0.18 1.00 1.00 0.23 0.25 3 2 3 4 3 0.46 1.00 0.86 1.00 0.05 3 3 2 3 2 0.69 0.21 0.51 1.00 1.00 3 2 2 3 2 0.157 1.00 1.00 0.04 1.00 3 3 3 3 2 0.61 0.07 1.00 0.30 1.00 3 3 3 3 2 1.00 0.22 0.34 1.00 1.00 4 2 2 3 2 0.03 1.00 1.00 1.00 1.00 3 3 3 3 2 0.61 0.07 1.00 0.30 1.00 3 5 3 3 3 0.03 0.32 0.39 0.80 1.00 4 2 2 3 2 0.91 0.29 0.72 0.27 1.00 4 2 2 3 2 0.47 0.02 0.15 0.01 1.00 3 2 2 3 2 0.33 1.00 0.72 0.21 0.25 3 3 3 3 2 0.22 0.01 0.67 1.00 1.00 3 2 2 3 2 0.36 0.25 0.51 1.00 1.00 3 2 3 4 2 0.36 1.00 0.56 1.00 1.00 3 3 3 3 2 0.03 0.12 0.37 1.00 1.00 3 3 3 3 3 0.84 0.02 0.39 0.68 1.00 4 4 2 3 2 0.01 0.39 1.00 1.00 1.00 3 2 3 3 3 1.00 0.18 0.67 0.16 1.00 4 3 3 4 2 0.43 1.00 0.67 0.57 1.00 3 2 3 4 2 0.04 0.06 0.39 0.19 1.00 3 2 3 3 2 0.61 1.00 0.57 1.00 1.00 3 4 3 3 2 0.04 <0.01 0.81 0.17 1.00 continued Ward et al Genetic variability in Cynoscion nebu/osus, determined with DNA markers 201 Table 1 (continued) Sample location Sample acronym n Soc 12 Soc 50 Soc 243 Cne 133 Cne 133' Corpus Christ! Bay, Texas 2000 CCOO 40 "a Phw 3 0.10 2 0.33 4 0.25 3 1.00 3 0.19 Corpus Christi Bay, Texas 2001 CCOl 35 3 0.59 4 0.02 4 0.71 4 0.74 3 1.00 Upper Laguna Madre, Texas 1998 UL98 40 3 0.05 2 1,00 3 0.33 3 1.00 2 1.00 Upper Laguna Madre, Texas 1999 UL99 40 "a 3 0.07 3 0.05 4 0.16 3 0.04 3 1.00 Upper Laguna Madre. Texas 2000 ULOO 40 "a 3 0.31 4 1.00 4 0.36 4 1.00 2 1.00 Lower Laguna Madre, Texas 1998 LL98 38 "a 4 0.06 3 0.13 2 1.00 3 0.79 2 0.01 Lower Laguna Madre, Texas 1999 LL99 40 "a Phw 3 0.55 3 1.00 5 1.00 4 0.04 2 1.00 Lower Laguna Madre, Texas 2000 LLOO 40 Phw 3 0.84 2 1.00 3 0.31 3 1.00 2 0.39 homogeneous allele distributions and the overall test was significant i.P^^^^,<0.001). Estimated theta values were statistically significant after Bonferroni adjust- ment, except Cnel33'. The overall estimate of theta (6=0.057, 95% CI, 0.005-0.062) was lower than the estimate including both spatial and temporal dimen- sions (0=0.116), indicating that temporal differences likely contributed to overall population differentiation in spotted seatrout. Analysis of molecular variance indicated a statisti- cally significant 4.11% of the among-sample genetic variance was attributable to differences between Gulf and Atlantic samples (P=0.04). Statistically signifi- cant variance was also detected among sampling sites within the Gulf of Mexico (0.49% of the total variance, P<0.001). No significant genetic variance attributable to temporal differences was found within bays (P=0.42). An overall Fgj of 0.046 was estimated, which was sta- tistically significant (P<0.001). We found no significant differences among Texas bays in the 1998 collections (Table 2). In 1999, spotted seatrout from Texas and Louisiana were not differenti- ated except for the Galveston Bay samples that differed significantly from all other samples. Analyses of year 2000 collections, which included samples from Texas, Louisiana, and Florida, revealed differences between Florida samples and all samples from Louisiana and Texas. The Louisiana sample differed from most Texas samples, and within Texas most samples were geneti- cally undifferentiated, except for samples Galveston Bay which were significantly different from those of all bays, except Sabine Lake and Corpus Christi. Fi- nally, the upper and lower Laguna Madre samples were statistically different from each other. Temporally, the only statistically significant differences among years were seen among Louisiana and Galveston Bay sam- ples. The overall theta within sampling years ranged from less than 0.006 in 1998 to 0.080 in 2000, perhaps reflecting the increased genetic variability introduced by the Florida samples in the 2000 data set. The sta- tistically insignificant differentiation among sampling years within bays, with the exception of the Louisiana and Galveston Bay, supports the notion that temporal differentiation, at least on the limited scale reported in our study, may be ignored and temporal samples can be collapsed within bays. The genetic structure of spotted seatrout in Galveston Bay is highly dif- ferentiated, both temporally and spatially, showing significant differences in 1999 and 2000 with almost all other bays. The topology of the neighbor-joining tree based on D(. for 34 site and year groups (Fig. 2) was poorly sup- ported by bootstrap replications and demonstrated little correspondence to geographic and temporal patterns. An exception was the separation of samples FCOO and FSOO (Charlotte Bay and St. John's River, FL, respectively; acronyms are defined in Table 1) from samples collected in the northern and western Gulf. When groupings were collapsed across years (Fig. 3), the distinctiveness of the Florida samples from the southern Gulf Coast and the Atlantic Coast continued to be supported. Spotted seatrout from Florida's Tampa Bay (FTOO) were found to be more closely related to spotted seatrout from Louisiana and Texas than to spotted seatrout in the other Florida samples. Texas and Louisiana samples formed a well-differentiated grouping; however, within that grouping, there was little correspondence between genetic differentiation and geographic location. One exception was the south coast of Texas (Corpus Christi Bay and the upper and 202 Fishery Bulletin 105(2) lower Laguna Madre), which formed a weakly sup- ported clade. A statistically significant (P<0.001) correlation be- tween D^. and geographic distance was observed (r=0.90) and thus supported the isolation-by-distance hypothesis. The greatest geographic distance between adjacent col- lection sites was observed between Charlotte Harbor and St. John's River; however, when the St. John's River sample was excluded from the analysis, a significant positive relationship was still evident (r=0.80, P<0.001). The correlation between D^. and geographic distance within the western Gulf of Mexico (Texas and Louisi- ana) remained positive (r=0.42j and statistically sig- nificant (P=0.05). Discussion Spotted seatrout inhabiting a series of sites from the lower Laguna Madre of Texas to St. John's River on the Atlantic Coast of Florida were genetically differentiated by analyses of allele frequencies of five microsatellite markers. Samples from Florida waters were strongly differentiated from spotted seatrout from Louisiana and Texas, and the statistically significant correlation between geographic distance and genetic distance was due primarily to these differences. Within Florida, the Charlotte Harbor spotted seatrout is genetically more similar to the Atlantic Coast spotted seatrout from St. John's River than to the neighboring Tampa Bay spotted seatrout, which is genetically more similar to Texas and Louisiana than to other Florida fish of the same species. Differences in allele frequency between Charlotte Harbor and Tampa Bay samples represented the greatest discontinuity observed in this study. This putative population structure is congruent with Wiley and Chapman's (2003) findings of distinct population subdivision among Atlantic Coast spotted seatrout, although details of that structure may be difficult to reconcile. Wiley and Chapman (2003) found spotted Table 2 Results of tests for homogeneity in allele distributions among samples of spotted seatrout iCynoscion nebulosus). Pairwise-9 estimates for 1998 samples (above diagonal), 1999 samples (below diagonal), and 2000 samples (lower matrix), fl^is 9 calculated across all sampling periods within a bay. * indicates statistical significance (o=0.05) after adjustment for multiple comparisons (Rice, 1989). Acronyms for within-bay samples collapsed across years are the two letters (e.g., LA99+LA00 = LA). Definitions for abbreviations for sample locations (LA, SL, etc.) are given in Table 1. Sample location LA SL GB EM MB SA AB CC UL LL SL -0.007 0.000 0.026 -0.011 -0.010 -0.003 0.011 0.015 0.012 0.002 GB 0.092* 0.070* 0.000 0.019 0.010 0.009 0.006 0.009 0.004 0.029 EM 0.009 0.002 0.053* 0.000 -0.007 0.006 -0.003 0.019 0.018 0.020 MB 0.007 0.007 0.083* 0.000 0.000 -0.008 -0.001 0.002 -0.001 0.005 SA -0.004 -0.006 0.078* -0.001 -0.006 0.000 0.008 -0.002 -0.005 0.000 AB 0.008 0.001 0.089* -0.004 -0.000 -0.001 0.000 0.009 0.008 0.024 CC -0.006 -0.008 0.079* 0.001 -0.004 -0.011 -0.001 0.000 0.000 0.010 UL 0.003 -0.002 0.081* -0.001 0.011 -0.002 0.000 0.001 0.000 0.003 LL 0.00.5 0.002 0.072* -0.009 -0.006 -0.003 -0.006 -0.003 0.000 0.000 FC FT LA SL GB EM MB SA AB CC UL LL FS 0.007 0.326* 0.108* 0.061* 0.110* 0.039* 0.077* 0.053* 0.053* 0.0532- 0.031* 0.081* FC 0.341* 0.136* 0.077* 0.145* 0.031* 0.085* 0.054* 0.059* 0.068* 0.043* 0.081* FT 0.245* 0.185* 0.257* 0.263* 0.195* 0.229* 0.219* 0.236* 0.288* 0.136* LA 0.057 0.011 0.084* 0.067* 0.058* 0.039 0.029 0.093* 0.063* SL 0.065 0.006 0.012 -0.006 0.003 -0.001 0.016 0.003 GB 0.105* 0.104* 0.083* 0.072* 0.050 0.115* 0.083* EM 0.011 -0.003 0.004 0.004 -0.007 0.019 MB 0.009 -0.004 0.013 0.016 0.001 SA -0.006 -0.005 0.004 0.013 AB -0.004 0.008 0.008 CC 0.013 0.016 UL 0.034* if 0.065* 0.004 0.049* <0.001 -0.004 -0.002 <0.001 0.013 0.001 0.013 Ward et al Genetic variability in Cynoscion nebulosus, determined with DNA markers 203 seatrout from Indian River. Florida (which is near St. John's River), to be genetically more similar to spot- ted seatrout from Choctawhatchee Bay in the Florida Panhandle than to other Atlantic Coast fish of the same species. The analyses of the two studies, taken together, indicate at least two distribution breaks in the eastern Gulf and the Atlantic, the first between Georgia and the upper Atlantic Coast of Florida and a second between Charlotte Harbor and Tampa Bay. The clustering of populations observed by Wiley and Chap- man (2003) between Indian River and Choctawhatchee Bay may, in light of our finding of a genetic discontinuity between the intervening Charlotte Harbor and Tampa Bay, reflect relative differences in genetic affinity dis- cerned by the two data sets. Resolution of this possible incongruence will require examination of numerous sampling sites collected from both the Atlantic and Gulf coasts of Florida. Spotted seatrout inhabiting the northwestern Gulf of Mexico from the Laguna Madre to Grand Isle, Louisi- ana, were not found to be subdivided into discernible stocks or populations and there was little indication of temporal differentiation within bays. Exceptions to this lack of temporal differentiation were seen in Galveston Bay and in Louisiana. Temporal differences in these two sites may be due to sampling error (although the n for each year's sample in the two sites appear to be adequate), or these two large regions may harbor populations that are temporally or spatially genetically structured. Some indications of geographically coherent spatial patterns were observed among spotted seatrout in the northwestern Gulf of Mexico. For example, there was an indication that this species on the lower coast is genetically differentiated, albeit weakly, from conspe- cifics inhabiting bays on the middle and upper Texas coast. This finding is similar to that found in allozyme (King and Pate, 1992) and mtDNA data (Gold et al., 1999) where differences between Laguna Madre sam- ples and more northerly bays were observed. Galveston Bay spotted seatrout were also found to be genetically divergent, being genetically distinct from spotted seat- rout in all other bays in 1999 and most bays in 2000. Galveston Bay was, in addition, one of two sites where the genetic structure of spotted seatrout was found to be temporally heterogeneous. Gold et al. (2003) found the upper Texas Coast to be a region of genetic transi- tion; a notable shift in allele frequencies of the Soc201 locus was evident between Matagorda Bay and Sabine Pass — a span that includes Galveston Bay. The lack of genetic population subdivision in the northwestern Gulf is consistent with the observed de- crease in heterozygosity in relation to Hardy-Weinberg expectations. Similar heterozygote deficiencies in white seabream (Diplodus sargus) (Lenfant and Planes, 2002) were hypothesized to represent mixing of genetically disparate individuals during some stage of recruit- ment (the Wahlund effect). This phenomenon may be characteristic of many marine fishes, especially those with local populations recruited from a larval stage with highly dispersive capabilities. Spotted seatrout. SL99 SL97 0.01 SLOG SA99 CC99 SL98 M98 r-l I MB99 LAOO GB98 scTCm SAOO MB98 -LLOO ,LL99 -CC01 "CC98 50 EMOO LA99 -GB99 ABOO SA98 70 pCCOO UL99 -EM99 GBOO -UL9 — MBOO LL98 FTO .mi - AB98 - ULOO - AB99 -FCOO 74 FSOO Figure 2 Cavalli-Sforza and Edwards (1967) chord distance neighbor-joining phenogram calculated for 33 spatial and temporal samples of spotted seatrout ^Cynoscion nebulosus). Values along branches indicate bootstrap values as percentages of replicates, based on locus (above line) or individual (below linel. Letter codes are defined in Table 1. due to its unique life history characteristics, is not an obvious example of a marine species expected to exhibit high gene flow. Spotted seatrout are confined to nearshore waters, spend most of their life within an estuarine habitat, and spawn and select their nursery area within the estuary (McMichael and Peters, 1989). Results of tagging studies support the hypothesis of a natal bay affinity based on spotted seatrout life his- tory (e.g.. Baker and Matlock. 1993). Stock structure among spotted seatrout was detected by the studies of morphology, physiology, (Iverson and Tabb, 1962), and genetics (Gold et al., 1999, and references therein). This population structure was not detected in the Gold et al. (2003) or the present analyses of variation in microsatellite DNA loci — markers expected to yield a high-resolution analysis of population-level genetic vari- ability. The differences observed between microsatellite variability and that seen in allozymes (Weinstein and Yerger, 1976; Ramsey and Wakeman, 1987; King and Pate. 1992) and mtDNA (Gold et al., 1999) may reflect different evolutionary processes (Gold et al., 2003). 204 Fishery Bulletin 105(2) 0.01 59 64 93 92 SL cc LL UL SA — LA GB MB EM AB FT 65 86 FC FS Figure 3 Cavalli-Sforza and Edwards (1967) chord distance neighbor-joining phe- nogram calculated for 13 spatial samples of spotted seatrout iCynoscion nebulosus). Values along branches indicate bootstrap values as percent- ages of replicates, based on locus (above line) or individual (below line). Letter codes are defined in Table 1. Microsatellites are assumecl to be selectively neutral, whereas the allozyme and mtDNA markers used in earlier studies are potentially subject to selection and thus may present different patterns for the same region (Hellberg et al., 2002). King and Zimmerman (1993) suggested the cline in AAT-2 observed by King and Pate (1992) may reflect adaptation to temperature or salinity gradients along the Texas Coast. Microsatellite markers would not, in the absence of linkage to loci affected by selection, be subject to such processes. It is also possible, as Gold et al. (2003) suggested, that the earlier allozyme and mtDNA studies provided evidence, not of genetic differentiation of populations inhabiting neighboring bays, but rather of a general confirmation of the isolation-by-distance model, where greatest ge- netic differences are found between the most peripheral sampling sites. Currently, about 5 million fingerling spotted seat- rout are stocked per year into Texas bays and estuar- ies. Neither Florida nor the other states of the eastern Gulf of Mexico have implemented large-scale spotted seatrout stocking programs; however such efforts are being considered. The genetic population structure ob- served in studies of allozymes (Weinstein and Yerger, 1976; Ramsey and Wakeman, 1987; King and Pate, 1992) and mtDNA (Gold et al., 1999) argue for a cau- tious policy concerning the stocking of spotted seatrout. Gold et al. (2003) suggested the gene flow observed in microsatellite markers argued against the current Texas policy of stocking only into the bay from which broodfish were procured. Allowing stocking into both the bay of broodfish origin and into adjacent bays would meet this suggestion of simulated gene flow and still protect the putative population subdivision detected by the earlier studies. Should stocking programs in Florida or elsewhere in the northeastern Gulf be implemented, it is critical, considering the level of population subdi- vision observed in the present study and that of Wiley and Chapman (2003), that fine-scale genetic surveys in the eastern Gulf be accomplished. It is also obvious that inter-regional transfers of spotted seatrout should be strictly avoided. Acknowledgments This research was partially funded by the Federal Aid to Sportfish Restoration Program, project F-36-R. Sam- ples were provided by staff of the Coastal Fisheries Division of Texas Parks and Wildlife Department, the Florida Fish and Wildlife Conservation Commission, and A. Landry and W. Dailey of Texas A&M University, Galveston. I. Blandon, W. Karel, and J. Burr provided laboratory assistance. A debt of gratitude is owed L. McEachron and R. Colura of Texas Parks and Wildlife Department, who supported and guided this work. Ward et al. Genetic variability in Cynoscion nebulosus, determined with DNA marl200 mm FL). Total lengths (TL) are given when that was the only length measurement reported in cited references. All measurements are in mm unless specified as cm. Measurements expressed ^ Eschmeyer, W. N. Catalog of fishes, on-line edition. Web- site: http://www.calacademy.org/research/ichthyology/catalog/ (accessed June 2006). in percent fork length or head length, are given only in the description of the new species Caranx fischeri. Fork length is measured from the front of the upper lip to the tip of shortest median caudal-fin ray. Body depths are measured from the anterior base of the spinous dorsal fin (DIO) to the origin of the pelvic fin (P20) and from the anterior base of the spine at the origin of the dorsal-fin lobe (D20) to the anterior base of the anal-fin spine at the origin of the anal-fin lobe (A20). Lengths of the dorsal- (D2) and anal-fin (A2) bases are straight-line measurements from either the D20 or A20 to the posterior base of the terminal fin ray of the respective fin. Head length is measured from the front of the upper lip to the posterior end of the opercular flap. Snout length is measured from the anterior end of the upper lip to the anterior edge of the eye. Eye diameter is the greatest bony diameter. Upper jaw length is taken from the anterior end of the upper lip to the posterior end of the maxilla. The curved part of the lateral line is measured as a chord (straight-line distance) of the arch extending from the upper edge of the opercle to its junction with the straight part; the straight part of the lateral line is measured from its junction with the curved part to its termination on the caudal-fin base (end of last scute). Scutes are defined as scales that have a raised horizontal ridge or a small to moderate projecting spine on the posterior margin ending in a point not exceeding a 120° angle; for de- tailed description and illustrations of scute formation and development in Caranx crysos (Mitchill) see Berry (I960). All scutes were counted, including those extend- ing onto the caudal-fin base. Pectoral-fin ray counts do not include the dorsal-most spine-like element. Gill raker counts are from the first gill arch (usually on the right side), and the raker at the angle is included in the lower-limb count; rudimentary gill rakers, with the diameter of their bases greater than their height, are defined as tubercles or short rakers. The anterior dorsal-fin pterygiophore formula indicates the inter- digitation pattern of supraneurals and pterygiophores within interneural spaces; neural spines are indicated by slashes, supraneural (predorsal) bones by an "S," pterygiophores by "2" (pterygiophores with two super- numerary rays and a serially associated ray) or "1" (no supernumerary ray and one serially associated ray). Results Taxonomy and distributions Some recent authors (Amezcua-Linares, 1996; Randall, 1996; McBride and McKown, 2000) still follow Briggs (1960) in erroneously reporting a worldwide distribu- tion in tropical and subtropical latitudes for Caranx hippos, although Nichols (1920) had correctly concluded that records of the species from the Indian and west- ern Pacific oceans were based on misidentifications. Other authors (Talwar and Kacker. 1984; Krishnan and Mishra, 1994; Mishra et al., 1999; Khan, 2003; 210 Fishery Bulletin 105(2) 120°W 105°W 90°W 75°W 60°W 45°W 30° W 15°W 15"E 45°N 30°N 15°N 15°S 30°S ?'., £r "'^~^ ^.Jii" • • 1> J^ * « tJ ^ ^-^ ^ D b Caranx hippos complex ■d C caninus • C. hippos a C. fischeri • N I i 45°N 15°N - 15°S 30°S 120°W 105°W 90°W 75°W 60°W 45°W 30°W 15°W 15°E Figure 2 Distribution of members of the crevalle jack iCaranx hippos) complex: (Mediterranean locality records for longfin crevalle jack (C. fischeri) are based solely on literature reports; discussion of geographic distribution appears in individual species accounts). Mishra and Krishnan, 2003) reported C. hippos as C. carangus (Bloch) from the Indian Ocean based on misidentifications of Caranx heberi (Bennett). What was once considered to be a single widespread species is herein recognized as consisting of three species (Fig. 2). For almost a century, most ichthyologists and fishery biologists who have worked on West African crevalle jacks have failed to distinguish the new species Cai-anx fischeri described herein from C. hippos, although both species are commonly taken together. Adults of Caranx hippos from opposite sides of the Atlantic Ocean are indistinguishable externally but ex- hibit consistent differences in the degree of development of the hyperostosis in the first dorsal-fin pterygiophore and neural spines of some of the anterior vertebrae (see "Geographic variation" in C. hippos species account). Although we consider these predictable ontogenetic and consistent site-specific patterns obvious evidence of genetic divergence associated with bone metabolism, an important consideration is the unknown functional significance of hyperostosis. In light of this, we believe it would be premature to recognize the eastern Atlantic population of C. hippos as taxonomically distinct. No formal change in classification should be made in the absence of collaborative molecular data. The Caranx hippos complex The C. hippos species complex can be diagnosed by the following combination of characters: a pair of strong symphyseal dentary canines (Fig. 3); breast naked ven- trally except for a small oblong patch of prepelvic scales (Fig. 4) which forms at about 30 mm FL; rounded black blotch on the lower rays of the pectoral fin in adults; large black opercular spot; and vertebrae 10 precaudal -I- 14 caudal. Only the black blotch on the pectoral fin is unique to these species. Adults of the horse-eye jack, Caranx latus Agassiz, occasionally have a somewhat sim- ilarly placed dusky blotch on the pectoral fin (although the dark area is different in character and never as well defined as in C. hippos), and this similarity in appear- ance has occasionally resulted in field misidentifications, especially by scuba divers unfamiliar with both species. The typical breast squamation pattern of the C. hippos species complex is not duplicated in any other Atlantic or eastern Pacific species of Caranx, although it occurs in three Indo-west Pacific species: commonly in C. ignobilis (Forsskal) and C. papuensis Alleyne and Macleay, and less frequently in C heberi. Dentition has been used as an important diagnostic character of carangid genera, but comparison of the dentition of a large number of Smith-Vaniz and Carpenter; Review of the Caronx hippos complex with a description of a new species from West Africa 211 Figure 3 Dentition of crevalle jack iCara/ix hippos), UF 69645, 180 mm FL, Florida, Gulf of Mexico; premaxilla (abovej, dentary (below); scale bar = 5 mm. carangid species reveals an almost complete continuum of dentition types that in some cases does not agree with traditional generic assignments. In all members of the C. hippos complex the upper jaw has an outer row of strong canines (widely spaced in adults) and an inner band of small villiform teeth that is widest anteriorly. The lower jaw has a single row of strong conical teeth that are smaller anteriorly, and one or two pairs of noticeably enlarged inner symphyseal canines. Enlarged symphy- seal dentary canines are absent in the following species of Caranx: C. crysos. C. caballus Giinther, C. melampy- gus Cuvier, C. papuensis. and C. senegallus Cuvier. Gill (1862) proposed the genus Paratractus for Ca?-a?ix pis- quetus Cuvier, a junior synonym of C. crysos, primarily because of the absence of symphyseal canines. Some recent authors follow Randall (1996) in assign- ing several common Atlantic carangids to the genus Carangoides Bleeker, but we maintain traditional usage for reasons given by Smith-Vaniz et al. (1999, p. 237). Figure 4 Breast squamation (naked areas shaded) of long- fin crevalle jack iCaranx fischeri), ANSP 158495, 154 mm FL, Nigeria. Hyperostosis In Caranx species Hyperostosis appears to have been an integral part of the evolutionary history of the Caranx hippos com- plex, but the pattern of expression is surprisingly different in each species (Table II. Hyperostosis involves the expansion or swelling of certain bones into globose, gall-like structures characterized by cellular bone foci and bone-resorbing osteoclasts. In most carangids the condition is usually ap- parent only in relatively large individuals (but can be detected histologically in smaller individuals) and the onset in different bone foci is typically sequential rather than simultaneous. A large number and size range of individuals of each species usually must be examined before the ontogenetic pattern can be precisely deter- mined. Although Smith-Vaniz et al. (1995) were unable to determine the functional significance of hyperostosis, they found no histological evidence of hyperostosis as a pathologic condition and concluded that the intraspe- cific predictability and site-specificity of hyperostosis in a taxonomically diverse group of marine fishes was indicative of genetic control. A detailed description of hyperostosis in Caranx is beyond the scope of this article, but to appreciate the context of its site-specificity and distribution in the C. hippos complex we briefly discuss its known occurrence in the genus. We found no evidence of hyperostosis in adults of six species: C. heberi, C. ignobilis, C. lugubris 212 Fishery Bulletin 105(2) Table 1 Comparison of hyperostosis in the Caranx hipp ally developed in all individuals. 9S species complex. Sizes are minimum fork length at which hyperostosis is usu- Hyperostotic bones C. fischeri (E.Atlantic) C. hippos (E.Atlantic) C. hippos (W.Atlantic) C. caninus (E.Pacific) Posttemporal yes >20 cm. Fig. 5 none none none Cleithrum none well-developed, >35 cm Fig. 10 well-developed, >35 cm Figs. 9, B-C none Neural spines none slight, 56 cm well-developed, >40 cm Figs. 9, B-C, 13 none Pleural ribs well-developed >34 cm (usually ribs 5-7), Figs. 5, B-C slight, 56 cm (ribs 6-8) well-developed, >38 cm (usually ribs 6-8), Figs. 9, B-C none or 5th rib only. Fig. 14A (in 6 of 16 spec. >34cm) Pelvic girdle none well-developed, 56 cm well-developed, >50 cm well-developed, >45 cm 1st pterygiophore of dorsal fin none Figs. 5, 11 slight. Figs. 10, 11, 12A well-developed, Figs. 9, 11, 12B well-developed, Figs. 11, 14 1st pterygiophore of anal fin none not convex anteriorly convex anteriorly. none not convex anteriorly none not convex anteriorly. Fig. 15 well-developed, >40 cm Figs. 14, 15 Poey, C. melampygus, C. papuensis, and C. tille Cuvier. In addition to C. fischeri, hyperostotic posttemporal bones are present in large individuals of the blue run- ner (C. crysos) and green jack (C caballus) allopatric species that are possibly conspecific. Caranx hippos is exceptional in that the neural spines of at least verte- brae 6-12 are hyperostotic in large adults. The ventral end of the cleithrum is hyperossified in large C. hippos, C. latiis, and C. sexfasciatus, but the shape of the hyper- ossifcation is noticeably different (wider and shorter) in the latter two species, which also differ from C. hippos in having two separate regions of hyperostosis on the cleithrum. The pelvic bones are hyperossified only in C. hippos and C. caninus. In large adults of C. cajiinus and western Atlantic C. hippos the first pterygiophore of the spinous dorsal fin becomes so enlarged that it resembles an oblong swollen fruit; but see discussion of geographic variation associated with hyperostosis of this pterygiophore in C. hippos species account. Even in small individuals of both species (where no pterygi- ophore swelling is evident), this bone is noticeably wider in lateral profile than in similar size individuals of C. fischeri, a species that never develops hyperostosis in this pterygiophore. The only site of hyperostosis in C. senegallus (largest specimen examined was 30 cm FL) is the posterior part of the supraoccipital crest. Hy- perostosis is extensive in C. bucculentus Alleyne and Macleay and includes the entire supraoccipital crest, first supraneural, first pterygiophore of the dorsal and anal fins, and a pair of patches on the caudal fin near its base. The ribs on precaudal vertebrae 5-7 (C fisch- eri) or 6-8 (C. hippos) exhibit extensive hyperostosis in relatively large individuals, but C. fischeri differs in that only the distal half of each rib is hyperossified. The only apparent contradiction to the consistent site specificity of hyperostosis in the C. hippos group is the pattern of occurrence seen on ribs of C. caninus. Ribs of the fifth precaudal vertebra appear "normal" in nine specimens (359-643 mm FL), are strongly and uniquely hyperostotic in six others (335-431 mm FL), and in SIO 65-176A (670 mm FL) there is a slight but noticeable swelling only in the middle part of the rib. Even more unexpectedly, in two of six individuals only one rib of these bilaterally paired structures was strongly hyperostotic and its counterpart rib exhibited no hyperostosis. Caranx carangopsis Steindachner, described from mid-Miocene deposits near Vienna, Austria, also de- serves mention. Heckel (1852) recognized the distinc- tiveness of this fossil species and gave it a scientific name, but the subsequent description was prepared entirely by Steindachner (1859) who must be credited as author of the species. The original description is based on an incomplete series of disarticulated bones, some of which are clearly hyperostotic, from several individuals estimated to have been about 0.9 meters in length. The scientific name refers to the presumed close relationship of this fossil species to C. cara?igus ( = C. hippos) — a relationship based, in part, on the oc- currence of hyperostotic bones (including the ribs, some of the vertebrae, and the first dorsal-fin pterygiophore) in both species. The text descriptions and illustrations of the massively swollen first pterygiophore and pleural ribs of C. carangopsis agree reasonably well with those of western Atlantic C. hippos, but do not resemble that characteristic of eastern Atlantic C. hippos. Steindach- ner's (1859) accurate description (footnote on p. 690) of the swollen neural spines of the vertebrae in a 1220 mm TL C. carangus ( = C. hippos) also contrasts sharply with his illustrations (pi. 7, Figs. 1-3) of the very differ- Smith-Vaniz and Carpenter: Review of the Caranx hippos complex witti a description of a new species from West Africa 213 ent thickened vertebrae of C. carangopsis. These fossil vertebrae are similar to those of hyperostotic Trachurus trachurus Linnaeus (see Desse et al., 19811, suggest- ing that the original description of C. carangopsis is likely based on material (deposited at NMW) from two carangid genera. Biology, fisheries, fish size, and edibility Remarkably little information has been published on the biology of members of the Caranx hippos complex. Both Kwei (19781 and McBride and McKown (2000) discussed the importance of estuaries as nurseries for juvenile C hippos, and such importance undoubtedly also applies to the other species. The former work is a comprehensive reference on the biology and fisheries of the "crevalle jack" in West Africa; unfortunately no photographs or meristic data were included that can be used to confirm the identification of species. Caranx hippos may well have been the most abundant species in the study, but C. fischeri is also very common in the region and almost certainly was included in some of the samples. Noting the occurrence of smallest juveniles in his study. Berry (1959) stated that off the southeastern Atlantic coast of North America spawning probably oc- curred offshore from March to September. Kwei (1978) reported juveniles present in Ghanaian lagoons (Keta region) during every month of the year and re-enter- ing the sea at sizes 2I2 cm FL. Large shoals of Caranx entered Ghanaian inshore waters from September to December, spawning appeared to be protracted, and peak spawning activity (determined from limited data) occurred from October to late January. Low frequency of ripe fish from inshore waters indicated that spawning occurred offshore. Thompson and Munro (1983) reported collecting seven "ripe" C hippos, four males and three females, in the vicinity of Jamaica. The smallest ripe males and females were 55 and 66 cm FL, respectively. Adults were found occasionally in reef habitats and reproductively active fish were taken in May, July, and November. Hildebrand (1939) recorded seven females (67-98 cm TL) with large or developing roe and 11 males (69-88 cm TL), most with developed testes, dur- ing 20-24 February 1935 from Gatun Locks, Panama Canal. McBride and KcKown (2000) reported young- of-the-year C. hippos, <4.0 cm FL, present in subtropi- cal estuaries (North Carolina to Florida) from June to November and discussed literature indicating that 9°C was likely the lower lethal temperature for the species. Franke and Acero (1993) suggested that C. caninus spawns throughout the year, peaking in January-Feb- ruary and August. Examining 96 specimens, they re- ported a 1;1 sex ratio, and the smallest mature males and females were 67 and 65 cm TL, respectively. All species of the crevalle jack complex are major predators of small schooling fishes in coastal areas. In the western Atlantic (Florida, Louisiana, and Texas), Saloman and Naughton (1984) reported that small jacks fed primarily on clupeids and larger fish fed usu- ally on clupeids, carangids, and sparids, but penaenoid shrimps, crabs, and other invertebrates were also con- sumed. Clupeids [Sardinella and Engraulis) were also the dominant prey of C. hippos in the Gulf of Guinea, and juvenile shrimps contributed 50-80% of the diet of juvenile fish during the dry season (Kwei, 1978). Most commercial landings of crevalle jack in the western Atlantic are from Florida, and annual catch- es of 221 to 320 t (metric tons) were recorded during 2000-2004 (NMFS2). In the eastern Atlantic, where data for C. hippos and C. fisheri are combined under "crevalle jack," commercial landings are reported only from Angola, Ghana, Sao Tome, and Principe, and for years 1995-2004 ranged from 2233 to 10,054 t (FAO. 2006). In the Gulf of Guinea, beach seine and set net fisheries for crevalle jack historically supported a large dried or salted fish industry. Okera (1978) reported C. hippos (as C. carangus) to be one of the dominate pelag- ic species in the beach seine fishery at Lumley, Sierra Leone, and that 80-100 cm TL fish were most common during September-October. Catches from Ghana in the mid 1950s to early 1960s and from Angola in the 1970s exceeded 15,000 t during some years, but such large catches no longer occur (FAO statistical data in Froese and Pauly^). With regard to fighting ability of the crevalle jack, Shipp (1986) stated "there is no tougher game to be had in shallow coastal waters with light tackle than this species." Caranx hippos is more important in recreation- al fisheries in the United States (statistics based only on Atlantic Coast, Gulf of Mexico, excluding Texas and Puerto Rico) and for years 2000-2004 annual catches ranged from 409 to 1030 t (NMFS-). Recreational fish- ing also occurs in West Africa for both C hippos and C. fisheri (Sehratwieser"'). The IGFA All-Tackle world-record C. hippos, from Barra do Kwanza, Angola, was caught in December 2000, weighed 26.5 kg (58 lb 6 oz) and was 114 cm FL and 129 cm TL; several other fish almost as large have also been recorded from West Africa. One C. fischeri caught at Ozouri Zimbani, Gabon, in January 1989, weighed 20.9 kg, and was approximately 100 cm FL and 127 cm TL. An even larger one (see Fig. 8C), released without being measured or weighed (est. weight 26 kg) was caught in Loango National Park (Iguela Lagoon mouth), Gabon, in December 2005. The IGFA All-Tackle world-record C. caninus was caught at Playa Zancudo, Costa Rica, in March 1997, weighed 19.7 kg, and was 101.6 cm TL. Crevalle jacks are strong fast-swimming predators with large quantities of red muscle and consequently 2 NMFS (NationalMarine Fisheries Service). 2006. Fisher- ies Statistics Division. Website: http://www.st.nmfs.gov/stl/ (accessed August 2006). ^ Froese, R., and D. Pauly, eds. FishBase world wide web electronic publication. Website: http://www.fishbase.org version (07/2006) (accessed August 2006). ^ Sehratwieser, J. 2006. Personal commun. International Game Fish Association, 300 Gulf Stream Way, Dania Beach, Florida. .33004. 214 Fishery Bulletin 105(2) their flesh is generally considered coarse and relatively unpalatable. Small individuals are more flavorful and bleeding immediately after capture is recommended. According to Shipp (1986), some of the better seafood cooks make delectable marinated specialties using C. hippos. Caranx fischeri, new species Longfin crevalle jack (Figs. 1A, 4-7, 8, A-C, 11; Tables 1-4) Caranx hippos (not of Linnaeus): Clark, 1915:385 (listed; Ascension Island); Fowler, 1919:254 (brief description; not distinct from American examples); Norman and Irvine, 1947:140, Fig. 65 (biology; ar- tisanal fishery; Ghana); Tortonese, 1952:302, Figs. 11-12 (description; Mediterranean specimens and records); Franca, 1954:24, pi. 3 Figs. 2-3 (descrip- tion; Luanda, Angola); Poll, 1954:131, Fig. 37, pi. 4, Figs. 1 and 3 (description); Cadenat, 1960:1392 (compared with "C carangus"=C. hippos); Cadenat, 1961:240 (listed); Bauchot and Blanc, 1963:43, Fig. 2c (in part, composite description, also includes C. hip- . pas; distribution); Daget and Stauch, 1968:40 (listed; Congo );Williams, 1968:252 (maximum reported size 75 cm); Blache et al., 1970:313, Fig. 818 (identifica- tion key; distinguished from "C. carangus"=C. hip- pos); Daget and litis, 1965:238, Fig. 152 (description; Ivory Coast); Tortonese, 1975:156, Fig. 64 (descrip- tion; Mediterranean records, after Tortonese, 1952); Okera, 1978:85 (beach seine fishery, occasionally taken with "C. carangus"=C. hippos; Sierra Leone); Smith-Vaniz and Berry, 1981: unpaginated; Fig. (in part, composite description; distribution); Bianchi, 1986:49, Fig., color pi. II, Fig. 10 (habitat; biology; fisheries utilization; Angola); Smith-Vaniz, 1986:824, Fig. (in part, composite diagnosis; habitat; distribu- tion); Bellemans et al., 1988:46, Fig., color pi. 2, Fig. 10 (local names; habitat); Papaconstantinou, 1988:95 (compiled; Greek seas); Edwards and Glass, 1987:1377 (unconfirmed records, St. Helena); Edwards, 1990:97, Fig. 48 (compiled description; unconfirmed occur- rence at St. Helena); Smith-Vaniz et al., 1990:732 (in part, composite synonymy; distribution); Afonso et al., 1999:73 (listed; Gulf of Guinea); Bilecenoglu et al., 2002:84 (Aegean and Turkish seas; compiled); Edwards et al., 2003:2238 (J. R. Irvine's Ghanaian specimens). Caranx carangus (non Bloch); Ehrenbaum, 1915:65 (misidentification, in part, Fig. of C. hippos after Goode, 1984; description; Cameroon); Chabanaud and Monod, 1927:18, Fig. 24 (listed, rare; Port Etienne, Mauritania); Collignon et al., 1957:192, Fig. 47 (brief description). Holotype ANSP 140256 (328 mm FL), Cameroon, Douala, 22 Aug 1978, obtained by P. J. P. Whitehead. Paratypes One-hundred twenty-eight specimens (33-530 mm FL) from 56 collections. SENEGAL: IRSNB 829 (530), Dakar, Madeleine Island, 9 Nov 1949, G. Marlier; MNHN 1978-260 (313), coast of Sen- egal. GUINEA: ISH 163/62 (227), 9°45'N, 13°55'W, 17 m. SIERRA LEON: ANSP 158497 (237), Freetown, 8°29'24"N, 13°11'30"W, 6 m, hook and line, 9 Feb 1968, RV Undaunted Cr. 6801, G. Beardsley; ANSP 158498 (239), 7°07'N, ir57'30"W, 18-21 m, Guinea Trawling Survey I, RV La Rafale. Trans. 12, sta. 1, 13 Nov 1963; BMNH 1928.8.3.14 (164.5), "Sierra Leone," J. Hornell; USNM 279566 (2, 163-203), St. Anne Banana Islands, 15-25 m, Feb 1986, G. Naylor. LIBERIA: USNM 193784 (121), Mesurado River beach, 6°19'N, 10^48'W, 24 May 1952, G. C. Miller; USNM 193790 (148) and USNM 193792 (102), Mesurado River beach, 20 Jun 1952, G. C. Miller; USNM 193779 (2, 148-163), Monrovia, Free- port, 5 May 1953, G. C. Miller. IVORY COAST: MNHN 1978-200 (253). GHANA: BMNH 1930.8.26.49-50 (2, 91-139), Accra, Mar 1930, F. R. Irvine (Irvine 53); BMNH 1938.12.15.48 (114), Volta River, Amedica, May 1938, F. R. Irvine (Irvine 237); BMNH 1939.7.12.12 (271), Prampram, Sep 1938, F. R. Irvine (Irvine 316); CAS-SU 64645 (118), Volta River; CAS-SU 64648 (124), Lower Volta River, Jun-Jul 1963, W. Titiati; CAS-SU 64700 (69), Battor River, 2 Mar 1964, T. R. Roberts; CAS-SU 66674 (41), Voha River at Amedia, 8 Mar 1963, T. R. Roberts; CAS-SU 69861 (2, 35-40), mouth of Volta River at Little Ada, 12 Jan 1963, T. R. Roberts; USNM 373240 (17, 33-52), Volta River at Big Ada, 9 Mar 1960, G. W. Bane; USNM 300660 (113), Tema Nunga, 18 May 1962, G. W. Bane; USNM 373242 (22, 37-52), beach at Tema fishing harbor, 15 Dec 1959, G. W. Bane; USNM 365702 (16, 71-90). Dix Cove Amaful, 25 Jan 1961, G. W. Bane; USNM 373244 (56), Ahiado River W. of Aman- ful, Takoradi, 4 Feb 1961, Amegah; USNM 368973 (3, 49-88), Ashantee, Beyah River, 27 Nov 1889, W. H. Brown, U.S. Eclipse African Exped. TOGO: ZMH 14575 (155), lagoon near Anecto, "Dr. Liebl," summer 1909. BENIN: MNHN 1967-826 (198), 6°15'N, 2°38'E, 23 m, 27 Jul 1964, A. Crosnier and J. Marteau. NIGERIA: BMNH 1968.11.15.31-32 (2, 85-169), Lagos Lagoon, 1967, S. O. Fagade; ANSP 158495 (3, 154-163), 4°15'N, 6°49'E, 15 m, Guinea Trawling Survey II, RV Thierry, trans. 43, sta. 1, 5 Apr 1964; CAS 38373 (135), 6°21'N, 2°54'E, 15 m, Guinea Trawling Survey II, RV Thierry, trans. 36, sta. 1, 19 Mar 1964; CAS 38375 (2, 145-160), 4°28'N, 5°07'E, 19 m, Guinea Trawling Survey II, RV Thierry, trans. 41, sta. 7, 2 Apr 1964; CAS 38395 (2, 146-150), 5°15'N, 5°09'E, Guinea Trawling Survey II, RV Thierry, trans. 40, sta. 1, 30 Mar 1964; ISH 1147/64 (145), 5°09'N, 4°39'W, 20 m; ZMUC P.46362 (450), Bonny River, 22 Feb 1946, "Atlantidae" sta. Ill; MNHN 1896- 327 (153), Campagne Touree; BMNH 1956.9.6.68 (183), Lagos Tarkwe, F. Williams. CAMEROON: ANSP 140288 (156), Victoria, 23 Aug 1978, FAO; BMNH 1936.12. 29.7 (172), Victoria, Cross River; D. Tovey; CAS-SU 15883 (2, 41-44), Bwanjo, Bwanjo River, 15 Sep 1936, A. I. Good; CAS-SU 15884 (150), Kribi, 25 Oct 1938, A. I. Good; CAS-SU 15885 (171), Kribi, 23 Feb 1940, A. I. Good; CAS-SU 18221 (77), Mbode, 23 Dec 1940, A. I. Good; CAS-SU 18222 (82), Kribi, Kribi River, 24 Sep Smith-Vaniz and Carpenter: Review of the Caranx hippos complex with a description of a new species from West Africa 215 1940, A. I. Good; CAS-SU 64900 (70). Kirbi, Kirbi River, 23 Nov 1940, A. I. Good; MNHN 1978-336 (147), Depierre; MNHN 1982-1093 (117), Yabassi, Depierre, 1980. UF 142347 (2, 144-155), 2°28'N, 9°44'E, 15-16 m, Guinea Trawling Survey II, RV Thierry Trans. 49, sta. 1, 25 Apr 1968; USNM 304197 (81), S. Korup at coast, Rio del Rey, 10 Mar 1988, G. M. Reid; ZMH 14576 (135), Douala, J. V. Eitzen, 1912-1913. EQUATORIAL GUINEA: ANSP 158493 (4, 173-197), Bioko (Fernando Po) 3°35'N, 9°19'E, 30 m, Guinea Trawling Survey II, RV Thierry, trans. 47, sta. 2, 18 Apr 1964; UF 142348 (2, 52-64), Bioko (Fernando Po), fresh water pool on SE end of island, 25 Sep 1959, G. W. Bane. GABON: CAS 38376 (197), 0°21N9n5'E, 20 m, Guinea Trawling Survey II, RV Thierry, trans. 52, sta. 1, 7 May 1964; BMNH 1896.5.5.14 (102), Corisco Island, M.H. Kingsley. CONGO: BMNH 1899.2.20.3 (173), Manyanga; BMNH 1899.11.27.87 (348), Banana; M. Delhez; MRAC 36 (403). Banana. 1896, Lt. E. Wilverth; MRAC 87428 (367), Banana, 1952, Major Maree. ASCENSION ISLAND: BMNH 1927.12.7.49 (358), J. Simpson. Other material Centro Oceanografico de Canarias, Tenerife, uncataloged (310), Benin, trawled in 38 m, 28 Jul 2002. FAO; SAIAB 26541 (130), Gulf of Guinea between Cameroon and Bioko; SAIAB 26541 (7, 138-233), Gulf of Guinea; MNHN 1978-235 (317), coast of tropical French Africa. Diagnosis A member of the Caranx hippos com- plex with the following combination of characters: segmented dorsal-fin rays 21-23 (exceptionally 24); segmented anal-fin rays 17-19, usually 18; posttem- poral bones hyperossified in specimens larger than 20 cm FL (Fig. 5); cleithrum, first pterygiophore of dorsal and anal fins, and neural spines of verte- brae relatively slender and never hyperossified; in specimens >20 cm FL, heights of longest dorsal- and anal-fin rays both 0.7-1.3 in head length; in adults, anal-fin lobe white anteriorly and remainder of fin gray to brown. Description Total range of values given first, fol- lowed by values for holotype in parentheses: dorsal- fin rays VIII-I, 21-24 (22); anal-fin rays II-I, 17-19 (18); pectoral-fin rays 18-21 (21); vertebrae 10 pre- caudal -i- 14 caudal; curved lateral-line scales 50-73 (69); straight lateral-line scales 0-16 (4); straight lateral-line scutes 24-41 (35); total scales -i- scutes in straight lateral line 32-47 (39); developed gill rakers 2-7 (3) upper, 14-17 (14) lower, 16-24 (17) total; rudimentary gill rakers 0-4 (4) upper, 0-3 (3) lower, 4-8 (7) total; rudimentary -i- developed gill rakers 20-25 (24) total. Posttemporal bones distinctly hyperossified in speci- mens larger than 20 cm FL (Fig. 5); cleithrum, pelvic bone, first pterygiophore of dorsal and anal fins, and vertebral neural spines not hyperossified, the latter Figure 5 Radiographs of longfin crevallejack iCaranx fischeri) exhib- iting hyperostotic bones (pale areas of hyperostotic bones are slightly computer enhancedl: (A) BMNH 1939.7.12.12, 271 mm FL. Gold Coast: iB) BMNH 1899.11.27.87. 348 mm FL, Congo: (C) ZMUZ P.46362, 450 mm FL, Nigeria. relatively slender; in specimens >34 cm FL distal half of pleural ribs of vertebrae 5-7 hyperostotic; anterior dor- sal-fin pterygiophore formula S/S-S/2-i-l/l; supraneurals relatively robust proximally; first anal-fin pterygiophore elongated anteroventrally. 216 Fishery Bulletin 105(2) 350 • C fischen 300 • O C hippos 1 250 A C caninus lobe height o o o ^•-*"°''° CO ° 50- ^^tF^^ 50 100 150 200 250 300 Head length (mm) Figure 6 Dorsal-fin lobe height versus head length in members of the crevalle jack {Caranx hippos) complex. Body robust and compressed; head blunt — upper pro- file strongly convex, lower profile only slightly curved anteriorly; caudal peduncle slender. Breast naked ven- trally to origin of pelvic fins, except for a small oval or oblong patch of scales in front of pelvic fins (Fig. 4); lat- erally, naked area sometimes extending slightly behind pelvic fins as a narrow wedge and always separated from naked base of pectoral fin by a narrow-to-broad band of scales; maxilla, lacrimal and dorsum of head naked; cheeks, preopercle, and opercle covered with scales; bases of dorsal and anal fins have a narrow scaly sheath anteriorly. Junction of curved and straight parts of lateral line below segmented dorsal-fin rays 5-10 (9); length of curved lateral line 0.93-1.47 (0.97) in straight lateral line. Dorsal fins well separated; first spine of spinous dorsal fin very slender and closely ap- plied to second spine, posterior 1-4 spines partially or completely embedded in large adults; third spine longest and much shorter than height of second dorsal- fin lobe. Height of second dorsal-fin lobe 0.7-1.3 (1.2) in head length; height of anal-fin lobe 0.7-1.3 (1.1) in head length; heights of both fin lobes longer than head in large adults (Figs. 6-7). Pectoral fin of adults long and falcate, 0.8-0.9 (0.8) in head length. Upper jaw 2.2-2.3 (2.2) in head length, extending to or slightly behind posterior margin of eye; eye diam- eter 4.0-6.1 (4.3) in head length, and adipose eyelid well-developed, especially posteriorly, in adults. Upper jaw with an outer row of strong canines (widely spaced in adults) and an inner band of small villiform teeth that are widest anteriorly; lower jaw with a single row of strong conical teeth that are smaller anteriorly and one (occasionally two) pair of noticeably enlarged inner symphyseal canines. Vomerine tooth patch triangu- lar-shaped, without a median posterior extension, and sparsely covered with small teeth. Measurements of 14 paratypes, 203-530 mm, and the holotype as percentages of FL: snout to DIO 40.7-43.9 350 •g 300 £ 250 gi 200 - CD t 150 2 100 c\j Q 50 C fischen a C hippos A C caninus 150 350 550 750 950 B 350 4 C fischen 300 □ C hippos ?50 A C caninus 200 150 100 ^-^" =."A a 50 n • 1 1 1 r 1 — T > T 150 350 550 750 950 400 350 300 250 200 150 100 50 C fischen O C hippos A C- caninus b"^ n ° igOD D * HD A n a Tr-, 00 ,S^ 150 350 550 Fork length (mm) 750 950 Figure 7 Selected measurements versus fork length (FL) in mem- bers of the crevalle jack (Caranx hippos) complex: (A) dorsal-fin lobe height versus FL; (B) anal-fin lobe height versus FL; IC) body depth (D20 to A20) versus FL, (41.6); snout to D20 55.7-59.8 (57.0); snout to P20 29.8-34.3 (30.2); snout to A20 55.5-60.2 (58.8); DIO to P20 33.4-39.0 (35.5); D20 to A20 35.3-42.6 (40.2); D2 base 31.2-34.1 (32.2); A2 base 28.4-32.5 (28.7); curved lateral-line length 27.6-36.0 (34.3); straight lateral-line length 32.1-43.8 (33.3); height of dorsal-fin lobe 21.6-29.6 (25.3); height of anal-fin lobe 22.7-30.6 (26.8); pelvic-fin length 13.6-15.7 (13.9); pectoral-fin length 33.7-38.7 Smith-Vaniz and Carpenter: Review of the Coranx hippos complex with a description of a new species from West Africa 217 Figure 8 Longfin crevalle jack iCaranx fischeri) (A-C) and crevalle jack iCaranx hippos) (D and E): (A) 31 cm FL, Benin, FAO; (B) 20.9 kg and ca. 100 cm PL, Gabon, IGFA; (C) ca. 26 kg, Gabon, A. Choinier; (D) Florida Bay, Gulf of Mexico, J. P. Reid; E) Angola, IGFA. (36.4); head length 29.0-31.6 (29.9). As percentages of head length: postorbital head length 49.5-57.4 (50.5); snout length 24.4-28.4 (26.1); eye diameter 16.4-25.2 (23.1); upper jaw length 43.1-46.3 (45.2). Fresh coloration of adults (Fig. 8, A-C) olive to green- ish-blue dorsally, changing to silvery white on lower sides and ventrally; prominent black spot, approximate- ly diameter of pupil, posteriorly on opercle at level of eye; an oval black spot on lower pectoral-fin rays and in upper axil of pectoral fins; dorsal fin dark brown; anal-fin lobe mostly white, especially anteriorly, and remainder of fin brownish-yellow; pelvic fins white and caudal fin brownish-yellow. In preserved adults, the dark spot on the opercle and dark blotch on lower pectoral-fin rays are readily apparent, the latter on rays 6 or 7 to 14-16 (counting ventrally). The relatively pale anterior of the anal-fin lobe in comparison to the remainder of the fin is also evident. Small juveniles have five dusky bands on body; lack the dark blotch on the pectoral fin, have a heavily pigmented spinous dorsal fin and the dorsal-fin lobe is dark distally. Two juveniles, 35-40 mm FL (CAS-SU 69861) collected near the mouth of the Volta River, Ghana, had the identical pigmentation of 28 specimens of C. hippos (CAS-SU 64646) taken in the same col- lection, including some in the same size range. Berry 218 Fishery Bulletin 105(2) Table 2 Frequency distributions of segmented dorsal- and anal-fin rays in the Caranx hippos species complex. Dorsal-fin rays Anal-fin rays Species 19 20 21 22 23 24 n 16 17 18 19 C. fischeri 31 C. hippos (E. Atlantic) 28 35 C. hippos iV^. Atlantic) 18 130 13 C. caninus 28 66 6 93 10 135 21.9 63 19.6 161 20.0 100 19.8 17 54 9 103 58 69 31 104 14 135 18.0 63 16.1 161 16.4 100 16.3 Dorsal + anal rays Pectoral-fin rays Species 35 36 37 38 39 40 41 42 43 n 18 19 20 21 C. fischeri C. hippos (E. Atlantic) C. hippos (W. Atlantic) C. caninus 15 18 83 13 26 30 7 22 79 51 9 25 47 22 6 135 39.8 63 35.7 161 36.3 100 36.1 1 12 1 18 3 68 15 57 21 50 38 12 1 2 (1959) included excellent illustrations of young C. hip- pos and C. latus. As both he and Laroche et al. (2006) discussed, small juveniles of these two species can not be distinguished solely by pigmentation. Thus, as might be expected, juveniles of C. fischeri and hippos also ap- parently cannot be distinguished by color pattern. Comparisons and relationships The unique pigmen- tation of the pectoral fin in adults, pattern of breast squamation (a relatively small number of C hippos and C. caninus are atypical in having the naked area of the breast continue without interruption to the pectoral- fin base), and the relatively large symphyseal dentary canines, which are shared by all members of the hippos complex, indicate their common ancestry. Of the three extant species, C. fischeri is readily distinguished by typically having more dorsal- and anal-fin rays (Table 2), and in specimens >20 cm FL the anterior dorsal- and anal-fin rays are relatively longer, and the body is deeper. The pattern of bones that exhibit hyperostosis is mark- edly different from that of the other species (Table 1), in neither of which is the posttemporal bone hyperossi- fied. The anal-fin lobe is white anteriorly in adults of C. fischeri, in contrast to the uniformly lemon yellow lobe of C. hippos. Adults of C. hippos also differ in having the underside of the caudal peduncle bright yellow. The presence of hyperostosis in a particular bone is presumed to be a derived condition (and conversely, the absence of hyperostosis is uninformative). On the basis of shared character states 5-6 (Table 1), C. hippos and C. caninus are considered to be sister species, and the geologically recent (-3.1 mya) rise of the Panamanian Isthmus was the likely vicariant event leading to the isolation and subsequent speciation of C. caninus. The common ancestor of C. fischeri and C. hippos-caninus presumably originated in the proto-Atlantic Ocean; and the sympatric occurrence of both C. fischeri and C. hippos in the eastern Atlantic is likely indicative of an earlier phylogenetic origin. Distribution African coast from Mauritania south at least to Mo?amedes, southern Angola (Franca, 1954), and at least historically it was present in the Mediter- ranean Sea (Fig. 2). The collection of an adult C. fischeri from Ascension Island indicates at least the occasional vagrant occurrence at insular localities. Unconfirmed historical reports of C. hippos from both Ascension (Clark, 1915) and St. Helena (Edwards, 1990) are likely based on misidentifications, possibly of C. fischeri. Tortonese (1952) discussed historical Mediterranean specimens dating from the 1890s in the Giglioli Col- lection and Geneva Museum and he identified these specimens as Caranx hippos. Our efforts to locate these or recent Mediterranean specimens of C. hippos have been unsuccessful (see Tortonese, 1973, for status of historical fish collections in Italy). Data that Tortonese (1952) provided for two of his five specimens, as well as an accompanying photograph of one of them, confirm their identification as C. fischeri. We assume that all five specimens were conspecific, and all Tortonese's Mediterranean distributional records are plotted in Figure 2. Papaconstantinou (1988) and Bilecenoglu et al. (2002) cited a few additional unconfirmed literature records of C. hippos from the Mediterranean, which we presume were also based on misidentifications of C. fischeri; these records are not shown on the distribu- tion map (Fig. 21. See discussion of probable erroneous recent photographic record of C. hippos from the Medi- terranean in the following species account. This species is often found in brackish water, some- times ascending rivers. The paratype series includes collections, mostly of juveniles, from three different river drainages. In their account of C. hippos, Norman and Irvine (1947) quoted a secondary source as report- Smith-Vaniz and Carpenter: Review of the Caranx hippos complex with a description of a new species from West Africa 219 ing that local fishermen say that Afafa fish (probably C. fischeri) swim far up rivers to spawn. Etymology We take great pleasure in naming this new species Caranx fischeri in honor of our friend and colleague Dr. Walter Fischer (retired) for his vision and dedication in initiating the Species Identification and Data Programme of the Food and Agriculture Organization of the United Nations (Fischer, 1989). In numerous ways this program has been an invaluable resource for marine fisheries biologists and ichthyolo- gists generally. Caranx hippos (.Linnaeus, 1766) Crevalle jack (Figs. IB, 3, 6-7, 8, D-E, 9-13, 15; Tables 1-4) Scomber hippos Linnaeus, 1766:494 (original descrip- tion; Carolina; putative holotype Linn. Soc. Lond. 130 [Garden no. 16]); Wheeler, 1985:55 (type status). Scomber carangus Bloch, 1793:69, pi. 340 (original de- scription; Antilles; syntype ZMB 1542). Caranx erythrurus Lacepede, 1801:58, 68 (no locality stated; based on Caranx hippos Linnaeus and other sources). Caranx carangua Lacepede, 1801:59, 74 (original de- scription; Martinique, West Indies; no type, based on a drawing by Plumier). Caranx antilliarum Bennett, 1840:282 (unnecessary re- placement name for Scomber carangus Bloch 1793). Caranx defensor DeKay, 1842:120, pi. 24, Fig. 72 (orig- inal description; New York; type whereabouts un- known). Carangus esculentus Girard, 1858:168 (name only); Gi- rard, 1859:23, pi. 11, Figs. 1-3 (description; Brazos Santiago, Texas; apparently an unnecessary replace- ment name for Scomber carangus Bloch to avoid "Strickland tautonymy" when Girard provided the new genus name Carangus). Caranx hippos: Goode, 1884:323, pi. 99 (biology, ed- ibility, distribution); Devincenzi, 1924:215, pi. 232, Fig. 1 (description; Rio de la Plata, Uruguay); Hil- debrand, 1939:26 (sexual maturity; Panama Canal); Ginsburg, 1952:93, pi. 5, Fig. C (synonymy; descrip- tion; distribution; Gulf of Mexico); Berry, 1959:503, Figs. 81-85 (juvenile description); Postel, 1959:157 (listed; Mauritania); Bauchot and Blanc, 1963:43 (composite description, also includes C. fischeri; dis- tribution); Vergara, 1972 (osteology and relationships of Cuban Caranx spp.); Menezes and Figueiredo, 1980:4, Fig. 4 (brief description; Brazil); Smith-Vaniz and Berry, 1981:unpaginated (in part; composite description; distribution); Uyeno et al., 1983:332, color photo (description, Suriname); Shipp, 1986:118, Fig. 133 (habits; edibility; Gulf of Mexico); Scott and Scott, 1988:376 (Canadian occurrence); Smith-Vaniz et al., 1990:732 (composite synonymy; distribution); Cervigon, 1993:63, Figs. 24-25 (description; distribu- tion; Venezuela); Randall, 1996:142, Fig. 173 (brief description; Caribbean); Murdy et al., 1997:165, Fig. 151 (description distribution; ecology; Chesapeake Bay); Debelius, 1997:159, unnumbered color Fig. (Ba- learic Islands, Spain; locality probably erroneous); Smith-Vaniz et al., 1999:238 (erroneous occurrence records; Bermuda); McBride and McKown, 2000:528 (seasonal dispersal patterns of juveniles between subtropical and temperate habitats; east coast of North America); Brito et al., 2002:220 (misidentifi- cation of C. latus; Canary Islands); Klein-MacPhee, 2002:415, Fig. 222 (description; early life history; Gulf of Maine); Laroche et al, 2006:1462, Figs, (early stages; early postflexion larvae indistinguishable from C. latus). Carangus hippos: Jordan and Evermann. 1902:306, un- numbered photograph (color description; "everywhere a food-fish of considerable importance"). Caranx hippos tropicus Nichols, 1920:45 (original de- scription; Para, Brazil; holotype AMNH 3889). Caranx africanus (not of Steindachner): Poll, 1954: pi. 4, Fig. 4 (misidentification; Banana, Congo). Caranx carangus: Cuvier and Valenciennes, 1833:91, pi. 57, Fig. 2 (description); Dumeril, 1861:262 (listed; Goree); Steindachner, 1870:704 (Senegal); Peters, 1877:836 (listed; Congo); Pellegrin, 1907:90, Fig. 7 (Dakar); Monod, 1927:699, Figs. 16-22B (Camer- oon); Cadenat, 1950:171, Fig. 103 (Senegal); Cadenat, 1960:1392 (compared with "C. hippos"=C. fischeri; Ghana and Nigeria); Williams, 1968:252 (maximum reported size 120 cm); Blache et al, 1970:313, Fig. 819 (identification key; distinguished from "C. hippos"=C. fischeri); Okera, 1978:84 (abundance in beach seine fishery; Sierra Leone). Diagnosis This species is a member of the Caranx hippos complex and has the following combination of characters: segmented dorsal-fin rays 19-21; segmented anal-fin rays 16 or 17; posttemporal bones never hyperos- sified; cleithra hyperossified distally in adults ^35 cm FL (Figs. 9, 10); first pterygiophore of dorsal fin (Figs. 11, 12) and neural spines of some vertebrae (Fig. 13) notice- ably (western Atlantic) or slightly to moderately (east- ern Atlantic) hyperossfied in adults ^50 cm FL; first pterygiophore of anal fin not hyperossified in large adults; pleural ribs 6-8 hyperossified in large adults; in specimens >20 cm FL, heights of longest dorsal- and anal-fin rays 1.3-2.1 and 1.2-2.0, respectively, in head length; anal-fin lobe and underside of caudal peduncle bright yellow in adults. Remarks Nichols and Roemhild (1946) gave frequency counts of dorsal- and anal-fin rays for 42 specimens of C. hippos from the western Atlantic Ocean. Their counts of 15 anal soft rays (3 specimens) and 18 dorsal soft rays (2 specimens) were not duplicated (see Table 2) in our material that was based on a total of 161 western Atlantic and 63 eastern Atlantic specimens. Because Berry (1959, Table 21) recorded the same range of soft rays (based on 132 western Atlantic C. hippos) that we also recorded, we conclude that the outlier counts given in the earlier study are erroneous. 220 Fishery Bulletin 105(2) Figure 9 Radiographs of crevalle jack (Caranx hippos) exhibiting hyperostotic bones (pale areas of hyperostotic bones are slightly computer enhanced): (A) AMNH 58046, 274 mm FL, Brazil; (B) USNM 132964, 400 mm FL, Cuba; (C) USNM 114618, 557 mm FL, Guatemala. Figure 10 Radiographs of crevalle jack iCaranx hippos) exhibiting hyperostotic bones I pale areas of hyperostotic bones are slightly computer enhanced): (A) MNHN 1978-216, 331 mm FL, Mauritania; (Bl ZMUC 25, 565 mm FL. Senegal. Smith-Vaniz and Carpenter; Review of the Coronx hippos complex with a description of a new species from West Africa 221 Comparisons Although long confused with Caran.x fischeri. as discussed under "Comparisons and relation- ships" in the account of that species, C. hippos is easily distinguished. However, C. hippos and C. caninus are so similar externally that many authors considered them to be taxonomically identical or only subspecifically distinct. They have broadly overlapping mensural (Figs. 6-7) and meristic values (Tables 2-4), but the pattern of hyperostosis (Table 1) is surprisingly different in the two species. They differ in four character states (Table 1, characters 2-4, 6) and share three others (Table 1, characters 5-6), although even in one of these (Table 1, character 5), the relative degree of hyperostosis is different (Fig. 11), namely the expansion of the first dorsal-fin pterygiophore being more pronounced in C. caninus. The color of the anal fin of a living fish is lemon yellow in C. hippos and is either uniformly white or brownish-orange in C. caninus. The underside of the caudal peduncle in adults of C. hippos is mostly yellow, a trait that C. Caninus lacks. Distribution This species is found on both sides of the Atlantic Ocean but is largely restricted to continen- tal shelf areas (Fig. 2). In the western Atlantic it is found from Nova Scotia only as rare waifs (Scott and Scott, 1998) to Rio de la Plata, Uruguay (Devincenzi, 1924), but is absent from Bermuda (Smith-Vaniz et al., 1999) and most of the Lesser Antilles. Confirmed insular locality records based on museum specimens include those for Jamaica and the Bahamas (Andros Island), and we have photographic documentation for the Virgin Islands (St. Thomas) near the southern end of the shallow Puerto Rico shelf, where the species is relatively common. Caran.x hippos is a regular summer visitor as far north as Woods Hole, Massachusetts (Klein-MacPhee, 2002), and young-of-the-year inhabit temperate estuaries of New York and New Jersey from July to November. McBride and McKown (2000) presented data indicating that these juveniles are spawned in subtropical latitudes and, aided by the Gulf Stream, disperse northward to coastal nurser- ies. Although the species is incapable of surviving the winter north of Cape Hatteras, growth rates and sea- sonal changes in distribution of this species indicate that some individuals successfully migrate southward to suitable over-wintering habitat and retain their potential contribution to the spawning population. In the eastern Atlantic C. hippos is known from Mauritania to Angola, but historical records for the Mediterranean Sea (Tortonese, 1952, 1975) are based on misidentifications of C. fischeri, as presumably are additional unconfirmed records cited by Papacon- stantinou (1988) and Bilecenoglu et al. (2002). The photograph (Debelius. 1997, p. 159) of a large school of adult Caran.x. identified in the caption as C. hip- pos and stated to have been taken at the Balearic Is- lands, Spain, may have been a substitution and this locality record for the species could not be confirmed (Debelius^). Reports of C. hippos from the Canary Is- lands are based on misidentifications of C. latus; and records of the species from the Azores (Arruda, 1997), Cape Verde Islands (Osorio, 1911), and St. Helena (Melliss, 1875; Edwards and Glass, 1987; Edwards, 1990) are unreliable and can not been confirmed. Adults are found inshore and frequently in upstream brackish waters (Klein-MacPhee, 2002) but are most common in salinities higher than 30 ppt (Gunter, 1945). Juveniles appear to use estuaries as nurseries in both temperate and tropical areas. Most reports of the spe- cies from freshwater are unreliable because of previous confusion with C. fischeri or are misleading (Herald and Strickland, 1949) because Homossassa Springs, Florida, has high alkalinity from the ionic composition of bicarbonate spring effluents. Gunter (1945) recorded juveniles and adults from Texas in salinities from 4.8 to 36.4 ppt. Smith (1985) reported that crevalle jacks are common summer residents in the Lower Hudson River, New York, and "in 1982 they were especially abundant as far upstream as River Mile 68 in early October and were still present at River Mile 66 in early November." McBride and McKown (2000) observed individuals in the Hudson River during July-October 1986-1993 at the freshwater interface (about 1 ppt), about 90-100 km inland. Geographic variation Juveniles and adults of C. hippos from opposite sides of the Atlantic Ocean are virtu- ally identical externally, including life coloration (Fig. 8, D-E), but differ notably in relative development of hyperostosis of the first pterygiophore of the dorsal fin. This bone is much less robust in adults of eastern Atlantic C. hippos (Figs. 11, 12). There is some variation in relative development of this pterygiophore in large western Atlantic specimens, but in all those we have examined (including a number of partially articulated skeletons at the AMNH not listed below) it is dorsolater- ally expanded in marked contrast to the slender profile of the bone in eastern Atlantic specimens (Fig. 12). The basal halves of the neural spines of some of the anterior vertebrae (usually vertebrae 5-12) are also consistently and strongly expanded (Fig. 9, B-C) in large adults from the western Atlantic. The neural spines are only slightly hyperossified in a 56-cm-FL specimen (Figs. lOB, 13) from Senegal. They were more expanded (although much less so than in similar-size western Atlantic specimens) in a 90-cm specimen from Angola that had been partially dissected and photographed at our request so that we could ascertain the condition of the neural spines. Caran.x hippos has an essentially continental distri- bution (there are no confirmed records from any oce- anic island) and populations on opposite sides of the Atlantic presumably are isolated and have little ge- netic connectivity, thus some geographic differentiation might be expected. Caran.x senegallus and C. fischeri are both eastern Atlantic endemics, but other Atlantic species of Caran.x with amphi-Atlantic distributions (C. ■' Debelius, H. 2004. Personal commun. IKAN-Underwasser- arehiv, Waldschulstrasse 166, 65933, Frankfurt, Germany. 222 Fishery Bulletin 105(2) 200 ~~r~ 400 600 ■^fe?C5 100 200 300 400 500 600 700 Fork length (mm) Figure 11 First dorsal-fin pterygiophore in lateral view (anterior of pterygiophore on left). Outlines were traced from radiographs plotted against fork length of Pacific crevalle jack (Caranx caninus) (pterygiophores numbers 1-36), crevalle jack (C. hippos) (37-83, western Atlan- tic, 84-87 eastern Atlantic, in black), and longfin crevalle jack (C fischeri) (88-106). Numbers correspond to those given in Appendix 1 with associated catalog numbers, specimen sizes and localities. crysos, C. latus, and C. lugubris) commonly are found at insular localities, including Ascension Island on the mid-Atlantic ridge. Material examined Two hundred fifty-one specimens (29.5-1070 mm FL) from 103 collections (western Atlantic localities abbreviated). MASSACHUSETTS: ANSP 165909 (4, 152-189); USNM 10431 (4, 262-276); USNM 13656 (265); USNM 126812 (2, 54-69). RHODE ISLAND; ANSP 98280 (2, 175-179); USNM 21654 (274). NEW JERSEY: ANSP 97864 (3, 176-189); ANSP 121305 (2, 167-175); ANSP 105515 (173); ANSP 165911 (118); USNM 37022 (187); USNM 45120 (133); USNM 64053(186). DELAWARE: USNM 187280(5, 102-120), Indian River. VIRGINIA: ANSP 52647 (167). NORTH CAROLINA: UF 124148 (153) and UF 124400 (154), Onslow Bay. SOUTH CAROLINA: UF 124149 (122). GEORGIA: UF 126976 (2, 132-133). FLORIDA: CAS 216873 (1015); ANSP 33039 (188); ANSP 93821 (366); ANSP 151093 (6, 186-208); CAS 216873 (1015); USNM 12681 (3, 254-269); USNM 22855 (622); USNM 29986 (372; USNM 53335 (263); USNM 57225 (260); USNM 57294 (2, 176-187); USNM 57295 (2, 172-175); USNM 154847 (167), St. Johns River; USNM 184871 (366); USNM 332457 (2, 390-421), Swannee River, salinity 14 ppt; USNM 362541 (3, 430-490), Caloosahatchee River; USNM 62289 (219); USNM 163601 (227). ALA- BAMA: ANSP 162288 (9, 810-950); USNM 157710 (4, 138-154). TEXAS: ANSP 99176 (5, 96-105); USNM 708 (10, 30-93); USNM 118497 (122); USNM 144017 (153). BAHAMAS: ANSP 102112 (2, 293-297); ANSP 102762 (657). CUBA: USNM 9867 (177); USNM 19821 (2, 357-386); USNM 132964 (2, 303-400). JAMAICA: USNM 32080 (725). DOMINICAN REPUBLIC: ANSP 81949 (114). PUERTO RICO: ANSP 151589 (254); ANSP 151590 (182). MEXICO: ANSP 159674 (5. 655-717); ANSP 156991 (9, 159-174); USNM 39278 (311); USNM 50473 (281). GUATEMALA: USNM 114580 (333); USNM Smilh-Vaniz and Carpenter: Review of the Caronx hippos complex with a description of a new species from West Africa 223 Figure 12 First dorsal-fin pterygiophore of crevalle jack (Caranx hippos) in dorsal (above) and lateral (below) views, scale bar = 5 cm: (A) USNM 377462. Guinea-Bissau, from 107 cm FL specimen; (B) ANSP uncataloged, Atlantic Mexico, from a 36-cm-FL specimen. See discussion of geographic variation in C. hippos species account. 114594 (420), 2 mi above mouth of Rio Sarstoon; USNM 114618 (7, 326-557); USNM 134378 (92); USNM 157572 (196). HOUNDURAS; ANSP 158504 (5, 227-251). COSTA RICA: USNM 89073 (122); USNM 94155 (114). PANAMA; ANSP 45238 (2, 119-136); USNM 79965 (287); USNM 79981 (440); USNM 128657 (2, 350-364); USNM 128658 (334). COLOMBIA: USNM 94769 (2, 142-1581; USNM 290077 (247). CURACAO; USNM 34914 (263). VENEZUELA; ANSP 161642 (2, 202-209); USNM 121801 (9, 136-213). GUYANA: USNM 186190 (357). FRENCH GUIANA: ANSP 148238 (294). BRAZIL: AMNH 3889 (300), Para Mkt., holotype of Caranx hippos tropicus; AMNH 58046 (274); ANSP 121329 (102); CAS 11861 (178); CAS-SU 22133 (4, 223-235); CAS-SU 51828 (2, 216-235); CAS-SU 51830 (231); CAS- SU 53013 (300); CAS-SU 53015 (311); CAS-SU 53016 (3, 316-319); CAS-SU 53025 (233); CAS-SU 53026 (232); CAS-SU 53080 (2, 605-617); CAS-SU 53082 (2, 534-555). MAURITANIA: MNHN 1978-216 (331), Port Etienne. SENEGAL; BMNH 1900.6.28.302-303 (2, 95— 109), St. Louise, M. P. Delhez; ZMUC 25 (565), Senegal, Dakar, Dec 1927, H. Madsen. GUINEA-BISSAU; USNM 377462 (estimated 107 cm) anterior dorsal-fin pterygi- ophore. May 2004, P. Sebile. LIBERIA; ANSP 158494 (16, 29.5-35.0), 6°31-7°07N', 11°29-11°57'30"W, sur- face dip net, 12 Nov 1963, B. B. Collette, sta. BBC 888. GHANA; CAS-SU 64646 (28, 34.8-68.4), mouth of Volta River at Little Ada, 12 Jan 1963, T. R. Roberts; USNM 373239 (3, 33-52), Volta River at Big Ada, 9 Mar 1960, G. W. Bane; USNM 42228 (71), Ashantee, Beyah River, 27 Nov 1889, W. H. Brown; USNM 373241 (3, 48-57), beach at Tema fishing harbor, 15 Dec 1959, G. W. Bane; USNM 373247 (66), 0.4 km above mouth of Rio Hwini, Takoradi, 26 Nov 1959, G. W. Bane; USNM 300496 (78), Takoradi swimming pool, 10 Aug 1961, G. W. Bane; USNM 368825 (65), Takoradi Fisheries Station bay, 14 Aug 1961, G. W. Bane. NIGERIA: MNHN 1896-328 (150), Campagne Toutee; BMNH 1968.11.15.29-30 (2, 93-94), Lagos Lagoon, 1967, S. O. Fagade. EQUA- TORIAL GUINEA: MNHN 1893-14 (155), Pobeguin. CONGO: MNHN 1967-0286 (88), Tchitemo, May 1964, A. Stauch. WEST AFRICA; MNHN 1978-230 (274) "coast of tropical French Africa." Caranx caninus GiJnther, 1867 Pacific crevalle jack (Figs. 1C, 6-7, 11, 13-15; Tables 1-4) Caranx caninus Giinther, 1867:601 (original description; Panama; holotype BMNH 1863.12.16.19); Gunther, 1868:432 (expanded description); Walford, 1937:72, color pi. 51, Fig. A (diagnosis; comparison with C. hip- pos; habits); Walford, 1974:15 ("disagreement among ichthyologists as to whether species is distinct from C. hippos;" distribution); Eschmeyer and Herald, 1983, Fig. 40 (diagnosis; possible synonym of C. hip- pos; distribution); Allen and Robertson, 1994:126, pi. VIII-4 (color photograph; brief description); Franke and Acero, 1993:57 (size at sexual maturity; Colom- bia); Grove and Lavenberg, 1997:362, Figs. 37 (color), 192, 193 (brief description; Galapagos); Garrison, 2000:166, color photograph (uncommon; Costa Rico, Cocos Island); Lea and Rosenblatt, 2000:122 (occur- rence in San Diego Bay). 224 Fishery Bulletin 105(2) 450 500 700 750 Fork length (mm) Figure 13 Eighth precaudal vertebra in lateral view (outlines traced from radiographs, anterior to left, plotted against fork length) of crevalle jack iCaranx hippos) (vertetra numbers 1-17, western Atlantic; 18, eastern Atlantic, in black; see discussion of geographic variation in species account.) and Pacific crevalle jack (Caranx caninus) (19-30). Numbers correspond to those given in Appendix 2 where associated catalog numbers, specimen sizes, and localities are also provided. Caranx hippos (not of Linnaeus): Jordan and Gilbert, 1883:201 (synonymy; in part; C. caninus listed as synonym); Jordan, 1895:432 (misidentification in part; important food fish, occasionally entering estu- aries; specimens from west coast and Havana indis- tinguishable; Mazatlan); Gilbert and Starks, 1904:77 (misidentification in part; Pacific and Atlantic speci- mens compared and considered conspecific; Panama Bay); Nichols, 1920:44 (Gulf of California fish indis- tinguishable from those from Atlantic coast); Meek and Hildebrand, 1925:350 (misidentification in part; distribution "Panama, common on both coasts of tropical America"); Hildebrand, 1946:208 (descrip- tion; Peru); Fierstine, 1968:1, Figs. 1-5 (description of dorsal-fin pterygiophore hyperostosis in Mio- cene deposits and living Caranx); Berry, 1974:240 (eastern Pacific and western Atlantic specimens essentially identical); Amezcua-Linares, 1996:88, unnumbered Fig. (description; biology; Mexico); Cas- tro-Aquirre and Balart. 2002:166 (listed; Revilla- gigedo Islands). Caranx (Tricropterus) hippos: Hiyama, 1937:33, color pi. 12 ("often identified to C. caninus Giinther; reaches 2 feet, abundant, good food fish"). Caranx hippos caninus: Nichols, 1937:58 (specimens from Ecuador compared with Atlantic C. hippos); Hobson, 1968:63, fig. 25 (predatory behavior; Gulf of California). Diagnosis A member of the Caranx hippos complex with the following combination of characters: segmented dorsal-fin rays 19-21 (Table 2); segmented anal-fin rays 16 or 17; posttemporal bones, cleithra, and neural spines of vertebrae never hyperossified (Fig. 14); first pterygiophore of dorsal fin distinctly hyperossfied in adults >38 cm FL (Figs. 11, 14); first pterygiophore of anal fin distinctly hyperossified, and having a convex anterior profile, in adults >40 cm FL (Figs. 14, B-C, 15); either none or 5th pleural rib only hyperossified (Fig. 14A) in adults ^38 cm FL; in specimens >20 cm FL, heights of longest dorsal- and anal-fin rays 1.3-1.7 and 1.3-2.0 mm, respectively, in head length; and anal- fin lobe varying from entirely white to brownish-orange in adults. Comparisons Caranx caninus and C. hippos have identical or broadly overlapping mensural and mer- istic values (Tables 2-4), although C. caninus usu- Smith-Vaniz and Carpenter: Review of the Caranx hippos complex with a description of a new species from West Africa 225 Figure 14 Radiographs of Pacific crevalle jack iCaranx caninus) ex- hibiting hyperostotic bones (pale areas of hyperostotic bones are slightly computer enhanced): (Al USNM 100998, 399 mm FL, Mexico; iBl USNM 206999, 490 mm FL. Colombia; (C) CAS 216871, 538 mm FL, Tres Marias Island. ally has more lateral-line scutes. Differences in development of hyperostosis (Table 1) are the most useful distinguishing characters (see comparisons in account of C. hippos). The color of the anal fin also differs in these two species. In C. hippos the anal fin is consistently lemon yellow, fading in postmor- tem individuals to orange-yellow. The anal fin of C. caninus varies from uniformly white to brownish- orange, and often some of the interradial membranes are dark brown. According to photographs of angler- 226 Fishery Bulletin 105(2) 400 450 500 550 500 650 700 Fork length (mm) Figure 15 First pterygiophore of anal fin in lateral view (outlines traced from radiographs, anterior to left, plotted against fork length) of Pacific crevalle jack (Caranx caninus) (ptergygiophore numbers 1-8) and crevalle jack (C. hippos) (9-21). Numbers correspond to those given in Appendix 3 where associated catalog numbers, specimen sizes and localities are also provided. caught fish, the underside of the caudal peduncle of C. caninus is never bright yellow as in C. hippos, and in fish with uniformly white anal fins the caudal peduncle is also white. Remarks We have not had the opportunity to study C. caninus in the field although we have examined many color photographs of recently caught adults. The pro- nounced differences in color of the anal fin in this species (see above) indicate the possibility of sexual dichroma- tism but determining the sex of large Caranx is best done with freshly caught specimens. The striking and inconsistent occurrence of hyperostosis of the third rib in this species (see Table 1) is also puzzling and the possibility that its presence or absence in adults may be sex linked and should be investigated. Distribution This species is restricted to the eastern tropical Pacific (Fig. 2), ranging north to San Diego Bay, California, where its occurrence is associated with El Nifio events (Lea and Rosenblatt, 2000), and from Mexico south to Lobos de Tierra Island, Peru (6°27'S); also known from the Galapagos, Malpelo, Cocos, and Revillagigedo islands, but it is unrecorded from Clip- perton Atoll (Robertson and Allen, 1996). Meek and Hildebrand (1925) reported the species, as C. hippos. from tidal streams at Corozal and Balboa, Panama. Material examined One hundred ten specimens (59-670 mm FL) from 51 collections. CALIFORNIA; SIO 75-383 (643), San Diego Bay. TRES MARIAS IS.: CAS 216871 (538); MEXICO: ANSP 144417 (25, 98-176); ANSP 158506 (4, 248-313); CAS 66825 (155); CAS 11112 (329); CAS-SU 55737 (2, 180-182); CAS 216872 (421); SIO 62-61 (3, 372-420); SIO 62-2725 (377); SIO 65-176A (670); SIO 65-182 (431); USNM 28293 (304); USNM 29556 (152); USNM 29617 (156); USNM 47143 (185); USNM 47144 (213); USNM 47145 (191); USNM 100991 (381); USNM 100998 (399); USNM 101006 (261); USNM 205166 (307). GUATEMALA: USNM 114469 (8, 77-233); EL SAL- VADOR: ANSP 136539 (172); ANSP 144401 (220); ANSP 144406 (169); USNM 220728 (2, 240-248); USNM 367522 (185); USNM 367542 (156); USNM 367671 (81); USNM 367946 (8, 61-77); USNM 367968 (63); USNM 367990 (166). PANAMA: ANSP 144409 (163); CAS 42539 (333); CAS 66826 (177); CAS 89955 (201); USNM 82080 (7, 75-187); USNM 79984 (2, 335-348); USNM 128659 (520); USNM 226417 (4, 59-112); USNM 321987 (2, 67-84). COLOMBIA: ANSP 144413 (164); USNM 206999 (493). ECUADOR: ANSP 158998 (137); CAS 66938 (212). GALAPAGOS IS.: USNM 89751 (2, 109-121), Indefatigable Id. PERU: SIO 58-83 (588); USNM 127917 (158); USNM 127918 (359); USNM 127919 (360). Smilh-Vaniz and Carpenter: Review of the Caranx hippos complex with a description of a new species from West Africa 227 Table 3 Frequency distributions of lateral-line scales and scutes in the Caranx hippos species complex. Curved lateral-line scales 50 52 54 56 58 60 62 64 66 68 70 72 74 Species 51 53 55 57 59 61 63 65 67 69 71 73 75 n .V SD C. fischeri 1 — 3 6 6 6 11 8 9 3 1 54 62.1 4.5 C. hippos (E. Atlantic) 1 1 3 1 5 — 1 12 66.2 3.3 C. hippos (W. Atlantic) 1 1 3 7 7 12 14 13 10 4 1 73 66.0 4.1 C. caninus 1 1 — 5 3 9 10 8 4 2 1 44 61.8 4.1 Straight lateral-line scales 1 3 5 7 9 11 13 15 Species 2 4 6 8 10 12 14 16 n X SD C. fischeri 5 6 21 12 19 6 — — 1 70 5.5 2.8 C. hippos iE. Atlantic! 1 1 4 5 3 2 — — 1 17 5.9 3.4 C. hippos {W.Atlantic) 4 10 21 19 13 8 5 1 1 82 5.4 3.2 C. caninus 8 9 17 11 7 1 1 54 3.8 2.7 Straight lateral-line scutes 24 26 28 30 32 34 36 38 40 42 44 46 Species 25 27 29 31 33 35 37 39 41 43 45 47 n .r SD C. fischeri 3 3 15 8 19 17 2 2 1 70 31.8 3.3 C. hippos iE Atlantic) 4 1 — 6 3 1 2 17 29.9 4.2 C. hippos (W. Atlantic) 1 4 12 26 27 8 3 1 82 31.2 2.4 C. caninus 1 — — 8 15 16 7 5 1 1 54 38.1 2.9 Straight lateral-line scales -h scutes 28 30 32 34 36 38 40 42 44 46 48 Species 29 31 33 35 37 39 41 43 45 47 49 n .r SD C. fischeri 11 7 24 17 6 3 — 2 70 37.0 3.1 C. hippos lE Atlantic! 1 1 1 4 4 4 1 1 17 35.8 3.5 C. hippos (W. Atlantic) 1 2 10 18 17 18 12 3 1 82 36.7 3.1 C. caninus 1 — — 1 4 4 14 13 8 8 1 54 41.9 3.6 Acknowledgments The study would not have been possible without the gen- erous cooperation of the following individuals who vari- ously facilitated loans and access to specimens, helped with literature, provided accommodations and technical assistance during visits, and general encouragement: M. L. Bauchot, the late F. H. Berry, B. Brown, G. H. Burgess, D. Catania, B. B. Collette, W. E. Eschmeyer, P. C. Heemstra, G. Kelly, T. Iwamoto, R. N. Lea, G. Lenglet, J. G. Lundberg, J. Maclaine, J. G. Nielsen, S. G. Poss, P. Pruvost, R. H. Robins, R. H. Rosenblatt, M. H. Sabaj, J. Schratwieser, O. Schultz, D. G. Smith, V. G. Springer, the late P. J. P. Whitehead, H. Wilkens, and J. T. Williams. Color photographs of Caranx spp. were provided by P. Afonso, A. Edwards, G. Kelly (IGFA), J. P. Reid, R. N. Lea, M. Lambouf (FAO), L Nicholson, E. Truter, and P. Wirtz. We especially thank H. L. Jelks whose expertise with Photoshop and Adobe Illustrator programs greatly im- proved the quality of Figures 2, 8, 11-13, and 15, and S. Reardon for digital photography of Figure 12A. We especially thank P. Sebile for his courtesy, with coop- eration of the International Game Fish Association, in obtaining and sending us the anterior dorsal-fin spines and pterygiophores of a huge (131 cm TL) Caranx hippos from Guinea-Bissau. This material was critical in con- firming our preliminary impression that certain aspects of hyperostosis were consistently different in eastern and western Atlantic populations of this species. 228 Fishery Bulletin 105(2) Table 4 Frequency distributions for gill raker counts ii the Caranx hippos species complex. Species Upper limb gill rakers Rudiments Developed Total .Y 1 2 3 4 5 2 3 4 5 6 7 8 4 5 6 7 8 9 C. fischeri <10 cm FL 18 5 — 1 1 1 1 — 17 6 13 11 1 6.5 >10cmFL 9 13 7 10 6 3 14 7 4 16 1 1 5 22 17 6.2 C. hippos iE. At\.) <10 cm FL 8 12 — 3 4 5 2 1 12 7 4 21 2 6.9 >10cmFL 2 1 2 1 1 2 — 2 1 1 — 2 2 1 6.8 C. hippos iW.Ati.) <10cmFL 2 1 — 1 1 2 — — 1 2 1 4 6.8 >10 cm FL 31 28 14 18 9 7 4 11 14 14 26 34 4 10 63 34 7.2 C. caninus <10 cm FL 9 2 1 1 5 6 3 9 7.7 >10 cm FL 9 14 12 10 8 3 4 14 6 9 18 5 1 16 36 3 7.7 Species Lower limb gill rakers Rudiments Developed Total T 1 2 3 13 14 15 16 17 18 16 17 18 19 20 C fischeri <10cmFL 4 11 10 3 11 7 4 7 18 16.7 >10cmFL 2 35 8 14 26 5 10 28 7 17.0 C. hippos (E.Ati.) <10cmFL 13 11 3 1 11 14 1 24 3 16.1 >10cmFL 5 1 1 4 1 5 1 16.2 C. /iippos (W. All.) <10cmFL 1 2 2 2 2 1 1 3 1 17.0 >10cmFL 1 9 80 17 1 35 57 13 1 29 65 13 16.9 C. caninus <10 cm FL 4 8 2 6 4 3 9 17.8 >10cmFL 2 13 38 3 1 10 35 9 1 4 14 31 6 1 17.8 Species Total g 11 rakers Total developed Total developed + rudiments .V 16 17 18 19 20 21 22 23 24 25 26 20 21 22 23 24 25 26 27 28 C. fischeri <10cmFL 1 1 — — 11 6 4 2 5 10 10 23.2 >10 cm FL 2 6 13 4 5 11 3 1 2 2 9 14 15 3 23.0 C. hippos (E. At\.) <10cmFL 3 3 2 6 8 4 1 4 21 2 23.0 >10 cm FL 1 2 1 — 2 1 — 2 2 — 1 22.5 C. hippos iW. At\.) <10cmFL 1 1 — 1 — 1 1 2 3 23.6 >10 cm FL 4 6 11 14 12 25 25 9 1 7 15 55 24 6 24.1 C. caninus <10 cm FL 1 1 4 2 4 2 2 8 25.5 >10 cm FL 1 7 12 5 8 12 10 1 2 7 19 21 6 1 25.4 Smith-Vaniz and Carpenter: Review of the Coranx hippos complex with a description of a new species from West Africa 229 Literature cited Afonso, P., F. M. Porteiro, R. S. Santos, J. P. Barreiros, J. Worms, and P. Wirtz. 1999. Coastal marine fishes of Sao Tome Island (Gulf of Guinea). Bull. Univ. Azores 17A:65-92. Allen, G. R.. and D. R. Robertson. 1994. Fishes of the tropical eastern Pacific, 332 p. Craw- ford House Press, Bathurst, New South Wales, Australia. Amezcua-Linares, F. 1996. Peces demersales de la plataforma continental del Pacifico central de Mexico, 184 p. Instituto de ciencias del Mar y Limnologia, UNAM/Comision Nacional para el Conocimiento y Uso de la Biodiversidad, Mexico, D.F. Arruda, L. M. 1997. Checklist of the marine fishes of the Azores. Arq. Mus. Bocage (n. s.) 3 (2):13-164. Bauchot, M. L. 1992. Carangidae. In Faune des poisons d'eaux douces et saumatres d'Afrique de I'Ouest, vol. 2 (C. Leveque, D. Paugy, and G. G. Teugels, eds.), p. 671-875. ORSTOM, Paris. Bauchot, M. L., and M. Blanc. 1963. Poissons marins de I'Est Atlantique tropical II. Percoidei (Teleosteens Perciformes). 2e partie. Atlantide Rep. 7:37-61. Bauchot, M. L., and A. Pras. 1980. Guide des poissons marins d'Europe, 427 p. De- lachaux et Niestle, Lausanne-Paris, France. Bellemans, M., A. Sagna, W. Fischer, and N. Scialabba. 1988. Fiches FAO d'identification des especes pour les besoins de la peche. Guide des resources halieutiques du Senegal et de la Gamble (especes marines et d'eaux saumatres), 223 p. FAO, Rome. Bennett, F. D. 1840. Narrative of a whaling voyage round the globe, from the year 1833 to 1836, vol. 2., 395 p. London. Berry, F H. 1959. Young jack crevalles iCaranx species) off the south- eastern Atlantic coast of the United States. Fish. Bull. 59:417-535. 1960. Scale and scute development of carangid fish Caranx crysos (Mitchill). Quart. J. Fla. Acad. Sci. 23:59-66. 1974. Crevalle Caranx hippos. In McClane's new stan- dard fishing encyclopedia and international angling guide (A. J. McClane, ed.), p. 240-241. Holt. Rinehart, and Winston, New York, NY. Bianchi, G. 1986. Fichas FAO de identificafao de especies para propositos comerciais. Guia de campo para as aspe- cies comerciais marinhas e de aguas salobras de Angola, 184 p. FAO. Rome. Bilecenoglu, M., E. Taskavak, S. Mater, and M. Kaya. 2002. Checklist of the marine fishes of Turkey. Zootaxa 113:1-194. Bini. G. 1968. Atlante dei pesci delle coste Italiane, vol. 5. Ostetti. Mondo Sommerso, 175 p. Edizioni Calderini, Bologna, Italy. Blaber. S. J. M. 2002. 'Fish in hot water': the challenges facing fish and fisheries research in tropical estuaries. J. Fish Biol. 61 (suppl. A):l-20. Blache, J., J. Cadenat. and A. Stauch. 1970. Cles de determination des poissons de mer signales dans I'Atlantique oriental entre le 20e parallele nord et le 15e parallele sud. Faune Trop. 18:1-479. Bloch, M. E. 1793. Naturgeschichte der auslandischen Fische, vol. 7, 144 p. Berlin. Briggs, J. C. 1960. Fishes of worldwide (circumtropical) distri- bution. Copeia 1960:171-180. Brito, A., P. J. Pascual, J. M. Falcon, A. Sancho, and G. Gonzaalez. 2002. Peces de las Islas Canarias catalogo comentado e ilustrado, 419 p. Francisco Lemus, La Laguna, Tener- ife, Canary Is. Cadenat, J. 1950. Poissons de mer du Senegal. Init. afr. 3:1-345. 1960. Notes d'ichtyologie ouest-africaine. XXX. Poissons de mer ouest-africans observes du Senegal au Camer- oun et plus specialement au large des Cotes de Sierra Leone et du Ghana. Bull. Inst. fr. Afr. noire, (A) 22 (4):1358-1420. 1961. Notes d'ichtyologie ouest-africaine. XXXIV. Liste complementaire des especes de poissons de mer (prov- enant des cotes de I'Afrique occidentale) en collection a' la section de biologie marine de I'lFAN a' Goree. Bull. Inst. fr. Afr. noire, (A) 23( l):231-245. Castro-Aguirre, J. L., and E. F. Balart. 2002. La ictiofauna de las Islas Revillagigedos y sus relacciones zoogeograficas, con comentarios acerca de su origen y evolucioon. In Libro Jubilar en Honor al Dr Salvador Contreras Balderas (Lozano-Vilano M de Lourdes, ed.) p. 153-170. Universidad Autonoma de Nuevo Leon, Monterey, Mexico. Castro-Aguirre, J. L., H. S. Espinosa Perez, and J. J. Schmitter- Soto. 1999. Ictiofauna estuarino-Lagunar y vicaria de Mexico. Coleccion Textos Politecnicos. Serie Biotech- nologfas:l-711. Cervigon, F. 1993. Los peces marinos de Venezuela, vol. 2., 499 p. Fundacion Cientifica Los Roques, Caracas, Venezuela. Chabanaud, P., and T. Monod. 1927. Les poissons de Port-Etienne, contribution a la faune ichthyologique de la region du Cap Blanc (Mau- ritanie Franfaise). Bull. Com. Etud. Hist. Scient. Afr. Occid. Fr. [1926) 9:225-287. Clark, R. S. 1915. Scottish National Antarctic Expedition: "Scotia" collection of Atlantic fishes. Rep. scient. Results Scott, natn. Antarct. Exped. 4 (16):379-401. Collignon, J., M. Rossignol, and C. Roux. 1957. Mollusques, crustaces, poisons marins des cotes d'A.E.F. en collection au Centre d'Oceanographie de I'lnstitut d'Etudes Centrafricaines de Pointe-Noire, 396 p. Paris. Cuvier, G., and A. Valenciennes. 1833. Histoire naturelle des poissons, vol. 9. 519 p. F. G. Levrault, Paris. Daget, J., and A. litis. 1965. Poissons de Cote D'lvoire (eaux douces et saumatres). Mem. Inst. Fr. Afr. Noire, (74):385 p. Daget, J., and A. Stauch. 1968. Poissons d'eaux douces et saumatres de la region cotiere du Congo. Cah. ORSTOM, Hydrobiol. 2 (2):21-50. Debelius, H. 1997. Mediterranean and Atlantic fish guide, 305 p. IKAN, Frankfurt, Germany. 230 Fishery Bulletin 105(2) DeKay, J, E. 1842. Zoology of New-York, or the New-York fauna; com- prising detailed descriptions of all the animals hitherto observed within the state of New-York, with brief notices of those occasionally found near its borders, and accom- panied by appropriate illustrations. Part IV. Fishes, 415 p. W. & A. White & J. Visscher, Albany. NY. Desmarest, E. 1856. Reptiles et poissons. In Encyclopedie d'histoire naturelle; ou, traite complet de cette science d'apres les travaux des naturalistes les plus eeminents de toutes les epoques, etc. . . . par le Dr. Chenu. Encyclope- die d'histoire naturelle, vol. 19. (J. G. Chenu, ed.l, 360 p. Paris. Desse, G., F.-J. Meunier, M. Peron, and J. Laroche. 1981. Hyperostose vertebrale chez I'animal. Rhuma- tologie, .33 (21:105-119. Devincenzi, G. J. 1924. Feces del Uruguay. II. Nuestra fauna ictiologica seguneuestras colecciones. Anales del Museo N. de Montevideo, ser. 2, 1 (5):139-290. Dumeril, A. H. A. 1861. Reptiles et poissons de la Cote occidentale d'Afrique. Archs. Mus. Hist. Nat. (Paris I 10 [1858]: 241-268. Edwards, A. J. 1990. Fish and fisheries of Saint Helena Island, 152 p. Centre for Tropical Coastal Management Studies, Univ. Newcastle upon Tyne, Newcastle, England. Edwards, A. J., A. C. Gill, and P. O. Abohweyere. 2003. A revision of F. R. Irvine's Ghanaian marine fishes in the collections of the Natural History Museum, London. J. Fish Biol. 37:2213-2267. Edwards, A. J., and C. W. Glass. 1987. The fishes of Saint Helena Island, South Atlan- tic Ocean. II. The pelagic fishes. J. Nat. Hist. 21:1367-394. Ehrenbaum, E. 1915. IJber kiisten Fische von Westafrika, besonders von Kamerun, 85 p, Hamburg, Germany. Eschmeyer, W. N. 1998. Catalog of fishes, 2905 p. Calif. Acad. Sci., Spec. Publ. 1, San Francisco, California. Eschmeyer, W. N., and E. S. Herald. 1983. A field guide to Pacific coast fishes of North America, 336 p. Peterson Field Guide Series, No. 28. Houghton Mifflin Co., Boston, MA. FAO (Food and Agriculture Organization of the United Nations). 2006. Yearbook of fishery statistics — capture production 2004. 98(11:1-560. FAO, Rome. Fierstine, H. L. 1968. Swollen dorsal fin elements in living and fossil Caranx (Teleostei: Carangidae). Cont. Sci. Los Angeles Co. Mus. 137:1-10. Fischer, W. 1989. The significance of FAO's biosystematics pro- gram in the enhancement of world fisheries. Aquatic Sci. l(4):683-692. Fowler, H. "W. 1919. The fishes of the United States "Eclipse" Expedi- tion to West Africa. Proc. U. S. Natl. Mus. 56 (2294): 195-292. 1936. The marine fishes of West Africa. Bull. Am. Mus. Nat. Hist. 70:1493 p. Franca, P. da. 1954. Contribuifao para o conhecimrento dos Carangideos (Carangidae, Pomatomidae e Coryphaenidael de Angola. An. Junta Invest. Ultramar 9 (2):3-52. Franke, R., and A. Acero P. 1993. Feces Carangoideos del Parque Gorgona, Paci- fico Colombiano (Osteichthyes: Carangidae, Nematis- tiidae y Coryphaenidael. Rev. Biol. Mar., Valparaiso 28:51-73. Garrison, G. 2000. Feces de la Isla del Cocos (Isla del Cocos fishes], 393 p. INBio, Santo Domingo de Heredia, Costa Rica. Gilbert. C. H., and E. C. Starks. 1904. The fishes of Panama Bay. Mem. Calif. Acad. Sci. part 4:1-304. Gill, A. C, and J. M.Kemp. 2002. Widespread Indo-Pacific shore-fish species: a challenger for taxonomists, biogeographers, ecologists, and fishery and conservation managers. Environ. Biol. Fish. 65:165-174. Gill, T. N. 1862. Note on some genera of fishes of western North America. Proc. Acad. Nat. Sci. Phila. 14:329-332. Ginsburg, I. 1952. Fishes of the family Carangidae of the northern Gulf of Mexico and three related species. Publ. Inst. Mar. Sci. 2:41-117. Girard, C. F. 1858. Notes upon various new genera and new species of fishes, in the museum of the Smithsonian Insti- tution, and collected in connection with the United States and Mexican boundary survey: Major William Emory, Commissioner. Proc. Acad. Nat. Sci. Phila. 10:167-171. 1859. Ichthyology. In United States and Mexican bound- ary survey, under the order of Lieut. Col. W. H. Emory, Major First Cavalry and United States commissioner. U.S. and Mexican Boundary Serv. 2 (2):l-85, Fishes pis. 1-41. Goode, G. B. 1884. The food fishes of the United States. In The Fisheries and Fishery Industries of the United States. U.S. Comm. Fish Fish. Rept., Sect. 1 (3):163-682. Grove, J. S., and R. J. Lavenberg. 1997. The fishes of the Galapagos Islands, 863 p. Stan- ford Univ. Press, Palo Alto, California. Gunter, G. 1945. Studies on marine fishes of Texas. Publ. Inst. Mar. Sci. Univ. Texas. 1:1-100. Giinther, A. 1867. On the fishes of the states of Central America, founded upon specimens collected in fresh and marine waters of various parts of that country by Messrs. Salvin and Godman and Capt. J. M. Dow. Proc. Zool. Soc. Lond. 1866 (3):600-604. 1868. An account of the fishes of the states of Central America, based on collections made by Capt. J. M. Dow, F. Godman, Esq., and O. Salvin, Esq. Trans. Zool. Soc. Lond. 6(71:377-494. Heckel, J. J. 1852. Caranx carangopis und fossile Delphin-wirbel. Jah- rbuch Geol. Reichsanstalt, Vienna 3 (21:160-161. Herald, E. S., and R. R. Strickland. 1949. Annotated list of the fishes of Homosassa Springs, Florida. Quart. J. Fla. Acad. Sci. (1948) 11: 99-101. Hildebrand, S. F. 1939. The Panama Canal as a passageway for fishes, Smith-Vaniz and Carpenter: Review of the Caranx hippos complex with a description of a new species from West Africa 231 with lists and remarks on the fishes and invertebrates observed. Zoologica 24 (l):15-45. 1946. A descriptive catalog of the shore fishes of Peru. Bull. U. S. Natl. Mus. 189, 530 p. Hiyama, Y. 1937. Marine fishes of the Pacific coast of Mexico. (T. Kumada, ed.), 75 p. Nissan Fisheries Institute, Odawara, Japan. Hobson, E. S. 1968. Predatory behavior of some shore fishes in the Gulf of California. Fish Wildlife Serv. Res. Rep. 73: 1-92. Hureau, J. C, and E. Tortonese. 1973. Carangidae. In Check-list of the fishes of the north-eastern Atlantic and the Mediterranean (CLOF- NAM), vol. 1 (Hureau, J. C, and T. Monod, eds.), p. 373-384. UNESCO ( United Nations Educational, Sci- entific, and Cultural Organziation), Paris. IGFA (International Game Fish Association). 2006. World record game fishes, 2006 ed., 384 p. IGFA, Dania Beach, FL. Jordan, D. S. 1895. The fishes of Sinaloa. Proc. Calif. Acad. Sci. (ser. 2)5:377-514. Jordan, D. S., and B. W. Evermann. 1902. American food and game fishes, 573 p. Doubleday, Page & Company, New York, NY. Jordan, D. S., and C. H. Gilbert. 1883. A review of the American Caranginae. Proc. U. S. Natl. Mus. 6:188-207. Khan, R. A. 2003. Fish faunal resources of Sunderban estuarine system with special reference to the biology of some commercially important species. Rec. Zool. Surv. India, Misc. Publ., Occas. Pap. 209:1-150. Klein-MacPhee, G. 2002. Jacks. Family Carangidae. In Bigelow and Schro- eder's fishes of the Gulf of Maine, 3rd ed. (B. B. Collette, and G. Klein-MacPhee, eds.), p. 411-427. Smithsonian Press, Washington, D.C. Krishnan, S., and S. S. Mishra. 1994. On a collection of fish from middle and south Andaman group of islands. Rec. Zool. Surv. India 94 (2-4):265-306. Kwei, E. A. 1978. Food and spawning activity of Caranx hippos (L.) off the coast of Ghana. J. Nat. Hist. 12:195-215. Lacepede, B. G. E. 1801. Histoire naturelle des poissons, vol. 3, 558 p. Plassan, Paris. Laroche, W.A., J. G. Ditty, J. T. Lamkin, and S. R. Whitcraft. 2006. Carangidae: jacks. In Early stages of Atlan- tic fishes: an identification guide for the western central North Atlantic, vol. 2 (W. J. Richards, ed.), p.1439-1509. CRC Press, Boca Raton, FL. Lea, R. N., and R. H. Rosenblatt. 2000. Observations on fishes associated with the 1997-98 El Nifio off California. Calif Coop. Oceanic Fish. Invest. Rep. 41:117-129. Linnaeus, C. 1766. Systema naturae sive regna tria naturae, secundum classes, ordines, genera, species, cum characteribus, dif- ferentiis, synonymis, locis. Laurentii Salvii, Holmiae. 12th ed., l(l):l-532. McBride, R. S., and K. A. Mckown. 2000. Consequences of dispersal of subtropically spawned crevalles jacks, Caranx hippos, to temperate estuaries. Fish. Bull. 98:528-538. Meek, S. E., and S. F. Hildebrand. 1925. The marine fishes of Panama. Part II. Field Mus. Nat. Hist. Publ. Zool. Ser. 15 (226):331-707. Melliss, J. C. 1875. Class V. Pisces. In St. Helena: a physical, his- torical and topographical description of the island, including its geology, fauna, flora, and meteorology, p. 100-113. Reeve & Co., London. Menezes, N. A., and J. L. Figueiredo. 1980. Manual de peixes marinhos do sudeste do Brasil. IV. Teleostei (3):l-96. Mishra, S. S., R. Gouda, L. Nayak, and R. P. Panigrahy. 1999. A check list of the marine and estuarine fishes of south Orissa, east coast of India. Rec. Zool. Surv. India 97 (3):81-90. Mishra, S. S., and S. Krishnan. 2003. Marine fishes of Pondicherry and Karaikal. Rec. Zool. Surv. India, Occ. Pap. 216:1-52. Monod, T. 1927. Pisces I, pisces marini. In Contributions a I'etude de la faune du Cameroun, part 2 (T. Monod, ed.), p. 643-742. Soc. Edit. Geogr. Marit. Colon., Paris. Murdy, E. O., R. S. Birdsong, and J. A. Musick. 1997. Fishes of Chesapeake Bay, 324 p. Smithsonian Inst. Press, Washington, D.C. Nichols, J. T. 1920. On the range and geographic variation of Caranx hippos. Copeia (83):44-45. 1937. On Caranx hippos (Linnaeus) from Ecuador. Co- peia 1937:58-59. Nichols, J. T., and J. Roemhild. 1946. Fin-count variation in Caranx hippos (Linnaeus). Marine Life, Occas. Papers 1(5):15-17. Norman, J. R., and F. R. Irvine. 1947. Marine fishes. In F. R. Irvine, The fishes and fisheries of the Gold Coast, p. 81-220. Crown Agents, London. Okera, W. 1978. Fishes taken by the beach seines fishing at Lumley, Freetown (Sierra Leone). J. Fish Biol. 12:81-88. Osorio, B. 1911. Peixes colhido nas visinhanfas do archipelago de Cabo Verde. Mems Mus. Bocage 3:37-49. Papaconstantinou, C. 1988. Check-list of marine fishes of Greece, 257 p. Fauna Graeciae IV, Hellenic Zoological Society, Athens. Peters, W.(C.H.). 1877. Ubersicht der wahrend der von 1874 bis 1876 unter dem Commando des Hrn. Capitan z. S. Frei- herrn von Schleinitz ausgefuhrten Reise S. M. S. Gazelle gesammelten und von der Kaiserlichen Admiralitat der Koniglichen Akademie der Wissen- schaften iibersandten Fische. Monatsb. Akad. Wiss. Berlin 1876:831-854. Pellegrin, J. 1907. Mission des pecheries de la cote occidentale d'Afrique, 2e memoire. Act. Soc. Linn. Bordeaux 62:71-102. Poll, M. 1954. Poissons. IV. Teleosteens acanthopterygiens (lere partie). Result. Sclent. Exped. Oceanogr. Belg. Eaux cot. Afr. Atlant. Sud (1948-49) 4(3B):l-390. 232 Fishery Bulletin 105(2) Postel, E. 1959. Liste commentee des poissons signales dans I'Atlan- tique tropico-oriental nord, du Cap Spartel au Cap Roxo, suivie d'un bref apercu sur leur repartition bathyme- tique et geographique. Bull. Soc. Scient. Bretagne 34 (1/2):129-170. Randall, J. E. 1996. Caribbean reef fishes, 3rd ed., 368 p. TFH Pub- lications, Inc., Neptune, NJ. Robertson, D. R., and G. R. Allen. 1996. Zoogeography of the shorefish fauna of Clipperton Atoll. Coral Reefs (1996) 15:121-131. Saloman. C. H., and S. P. Naughton. 1984. Food of crevalle jack tCarnnx hippos) from Florida, Louisiana, and Texas. NOAA Tech. Memo. NMFS- SEFC-134, 34 p. Sanders, A. L. 1997. Alexander Garden (1730-1791), pioneer natu- ralist in colonial America. In Collection building in ichthyology and herpetology (T. W. Pietsch, and W. D. Anderson Jr., eds.), p. 409-437. Am. Soc. Ich. Herp., Spec. Publ. 3. Scott, W. B., and M. G. Scott. 1988. Atlantic fishes of Canada. Can. Bull. Fish. Aquat. Sci. 219. 731 p. Shipp, R. L. 1986. Dr. Bob Shipp's guide to fishes of the Gulf of Mex- ico, 256 p. 20"^ Century Printing Co., Mobile, AL. Smith, C. L. 1985. The inland fishes of New York State, 522 p. New York State Dept. Environ. Conserv., Albany, NY. Smith-Vaniz, W. F. 1986. Carangidae. In Fishes of the north-eastern Atlantic and Mediterranean (FNAM), vol. 2 (White- head, P. J. P., M.-L. Bauchot, J.-C. Hureau, J. Nielsen, and E. Tortonese, eds.), p. 815-844. UNESCO (United Nations Educational, Scientific, and Cultural Organi- zation) Paris. 2003. Carangidae. In FAO species identification sheets for fishery purposes (K. E. Carpenter, ed.), p. 1426-1468. Western Central Atlantic. FAO, Rome. Smith-Vaniz, W. F., and F. H. Berry. 1981. Carangidae. //; FAO species identification sheets for fishery purposes, vol. 1 (W. Fischer and G. Bian- chi, eds.), unnumbered p. 1-7-H74. Eastern Central Atlantic. FAO, Rome. Smith-Vaniz, W. F., B. B. Collette, and B. E. Luckhurst. 1999. Fishes of Bermuda: history, zoogeography, annotated checklist, and identification keys. Am. Soc. Ich. Herp., Spec. Publ. 4, 424 p. Smith-Vaniz, W. F., J. C. Quero, and M. Desoutter. 1990. Carangidae. In Check-list of the fishes of the eastern tropical Atlantic (CLOFETA), vol. 2 (Quero, J. C, J. C. Hureau, C. Karrer, A. Post, and L. Saldanha, eds.), p. 729-755. UNESCO. Paris. Smith-Vaniz. W. F, L. S. Kaufman, and J. Glowacki. 1995. Species-specific patterns of hyperostosis in ma- rine teleost fishes. Mar. Biol. 121:573-580. Steindachner, F. 1859. Beitrage zur Kenntniss der fossilen Fisch- Fauna Osterreichs. Sitzungsber. Akad. Wiss Wien 37:673-703. 1870. Zur Fischfauna des Senegal. Erste Abtheilung. Sitzungsber. Akad. Wiss Wien 60 (1) [18691:669-714. Talwar, P. K., and R. K. Kacker. 1984. Commercial sea fishes of India. 997 p. Zool. Surv. India, Calcutta. Thompson, R., and J. L. Munro. 1983. The biology, ecology, and bionomics of the jacks, Carangidae. In Caribbean Coral Reef Fishery Resources (J. L. Munro, ed.), p. 82-93. International Center for Living Aquatic Resources Management, Manila, Phil- ippines. Tortonese, E. 1952. Monografia dei Carangini viventi nel Mediter- raneo (Pisces Perciformes). Ann. Mus. Civ. Stor. Nat. Giacomo Doria 65:259-324. 1973. The ichthyological collections now existing in Italy. Ann. Mus. Civ. Stor. Nat. Giacomo Doria 79: 108-116. 1975. Osteichthyes (Pesci ossei). Parte seconda. Edizioni Calderini, Bologna. Fauna Ital. 11:1-636. Uyeno, T., K. Matsuura, and E. Fujii, eds. 1983. Fishes trawled off Suriname and French Guiana, 519 p. JAMARC (Japan Marine Fishery Resources Research Center), Toyko. Vergara, R. 1972. Analisis taxonomico y consideraciones filogeneticas sobre las especies Cubanas del genero Caranx (teleos- tel, Perciformes, Carangidae). Cent. Inst. Pesq. Cuba (34):1-137. Walford, L. A. 1937. Marine game fishes of the Pacific Coast from Alaska to the Equator, 205 p. Univ. Calif. Press, Berkeley, CA. 1974. Marine game fishes of the Pacific Coast from Alaska to the Equator, p. 1-19. [i-xxix+1-205], pis. 1-69. Reprint of 1937 edition for the Smithsonian Insti- tution by T.F.H. Publications, Inc., Neptune, NJ. Wheeler, A. 1985. The Linnaean fish collection in the Linnaean Society of London. Zool. J. Linn. Soc. 84:1-76. Williams, F. 1968. Report on the Guinean Trawling Survey. Vol.1: General report. Lagos:l-828. Appendix 1 Catalog numbers, localities, and sizes (mm FL) of speci- mens used for outline drawings of the first dorsal-fin pterygiophore in Figure 11. Numbers in bold correspond to the numbers for the pterygiophores illustrated in Figure 11. Caranx caninus: 1, ANSP 144406 (169) El Salvador; 2, USNM 114469 (178) Guatemala; 3, USNM 820080 (187) Panama; 4, USNM 47145 Gulf of California (192); 5, USNM 114469 Guatemala (197); 6, CAS 66938 Ecuador (212); 7, USNM 47144 Gulf of Cali- fornia (213); 8, ANSP 144401 El Salvador (220); 9, USNM 114469 Guatemala (227); 10. USNM 114469 Guatemala (233); 11, USNM 200728 El Salvador (240); 12, ANSP 158506 Mexico, Sinalosa (248); 13, ANSP 158506 Mexico, Sinalosa (258); 14, US- NM 101006 Mexico (261); 15, ANSP 158506 Mexi- co, Sinalosa (262); 16, USNM 28293 Mexico (304); 17, USNM 205166 Baja California (307); 18, ANSP 158506 Mexico, Sinalosa (313); 19, CAS 42539 3"53'N, 105"10'W (333); 20, CAS 11112 Mexico (329); 21, US- Smith-Vaniz and Carpenter: Review of the Caranx hippos complex with a description of a new species from West Africa 233 NM 79884 Panama (348); 22, USNM 127918 Peru (359); 23. USNM 127919 Peru (360); 24, SIO 62-61 Mexico, Isabel Island (371); 25, USNM 100991 Mexico (381); 26, USNM 100998 Mexico (399); 27, SIO 62- 61 Mexico, Isabel Island (419); 28, CAS 216872 Baja California (421); 29, SIO 65-182 (431); 30, SIO-62-61 Mexico, Isabel Island (438); 31, USNM 206999 Colom- bia (490); 32, USNM 128659 Panama (520); 33, CAS 216871 Tres Marias Island (538); 34, SIO 58-83 Peru (588); 35, SIO 75-383 California, San Diego (643); 36, SIO 65-176A Baja California (670). Caranx hippos: 37, USNM 57294 (176) Florida; 38, ANSP 97864 (176) New Jersey; 39, ANSP 97864 (189) New Jersey; 40, ANSP 151093 (198) Florida; 41, USNM 121801 (205); Venezuela 42, ANSP 151093 Florida (207); 43, USNM 121801 (213) Venezuela; 44, CAS 122133 (223) Brazil; 45, CAS 122133 (228) Brazil; 46, ANSP 158504 (233) Honduras; 47, CAS 122133 (235) Brazil; 48, ANSP 151589 (254) Puerto Rico; 49, ANSP 158504 (251) Honduras; 50, US- NM 12681 (254) Key West; 51, USNM 12681 (269) Key West; 52, AMNH 3889 (274) Brazil; 53, ANSP 148238 (294) French Guiana; 54, ANSP 102112 (298) Brazil; 55, AMNH 3889 (300) Brazil; 56, CAS 153016 (316) Brazil; 57, CAS 153016 (319) Brazil; 58, USNM 114618 (326) Guatemala; 59, USNM 114618 (332) Guatemala; 60, USNM 128658 (334) Panama; 61, USNM 128658 (350) Panama; 62, USNM 114618 (360) Guatemala; 63, USNM 128657 (364) Panama; 64, ANSP 93821 (366) Florida; 65, USNM 29986 (372) Rhode Island; 66, USNM 19821 (386) Cuba; 67, USNM 132964 (400) Cuba; 68, USNM 114594 (420) Guatemala; 69, USNM 332457 (421) Florida; 70, USNM 362541 (430) Florida; 71, USNM 79981 Panama (440); 72, USNM 362541 (460) Florida; 73, USNM 362541 (470) Florida; 74, USNM 114816 (485) Guatemala; 75, CAS 153082 (534) Brazil;76, CAS 153082 (555) Brazil; 77, CAS 153080 (605) Brazil; 78, CAS 153080 (617) Brazil; 79, USNM 22855 Gulf of Mexico (622); 80, ANSP 159674 (655) Mexico; 81, ANSP 102762 (657) Bahamas; 82, USNM 32080 (725) Jamaica; 83, CAS 216873 (1015) Florida; 84, MNHN 1978-230 (274) "tropical West Africa"; 85, MNHN 1978-216 (331) Western Sahara; 86, ZMUC 25 (565) Senegal, Dakar; 87, USNM 377462 (1070 estimated, 130 cm TL measured), Guinea-Bissau. Caranx fischeri: 88, CAS 38375 (159) Nigeria; 89, US- NM 27566 (163) Sierra Leone; 90, CAS-SU 15885 (171) Cameroon; 91, ANSP 158493 (173) Gulf of Guin- ea, Bioko; 92, ANSP 158493 (182) Bioko; 93, ANSP 158493 (197) Bioko; 94, USNM 279566 (203) Sierra Leon; 95, ANSP 158497 (237) Sierra Leone; 96, ANSP 158498 (239) Sierra Leone; 97, BMNH 1939.7.12.12 (271) Gold Coast; 98, MNHN 1978-260 (313) Sen- egal; 99, MNHN 1978-235 (317) "tropical West Af- rica"; 100, ANSP 140256 (328) Cameroon; 101, BMNH 1899.11.27.87 (348) Congo; 102, BMNH 1927.12.7.49 (358) Ascension Island; 103, MRAC 87428 (367) Con- go; 104, MRAC 36 (403) Congo; 105, ZMUC P.46362 (450) Nigeria; 106, IRSNB 829 (530) Senegal. Appendix 2 Catalog numbers, localities, and sizes (mm FL) of specimens used for the outline drawings of the eighth precaudal vertebra in Figure. 13. Number in bold cor- respond to the numbers for the vertebrae illustrated in Figure 13. Caranx hippos: 1, USNM 19821 (357), Cuba; 2, USNM 114618 (360), Guatemala; 3, USNM (400) Cuba; 4, USNM 11490 (420) Guatemala; 5, USNM 3324557 (421) Florida, Swannee River; 6, USNM 79981 (440), Panama, Colon; 7, USNM 362541 (470), Florida, Caloosahatchee River; 8, USNM 115618 (485), Gua- temala; 9, CAS-SU (534) Brazil; 10, CAS-SU 53082 (555) Brazil; 11, USNM 114618 (557) Guatemala; 12, CAS-SU 53080 (605) Brazil; 13, USNM 22855 (622) Florida, Pensacola; 14, ANSP 159674 (655) Mexico, Carmen; 15, ANSP 102762 (657) Bahamas, Andros Island; 16, ANSP 159674 (698) Mexico, Carmen; 17, ANSP 159674 (717) Mexico, Carmen; 18, ZMUC (565) Senegal, Dakar Caranx caninus: 19, USNM 127918 (359) Peru, Lo- bos de Tierra; 20, SIO-62-61 (371) Mexico, Isabel Island; 21, USNM 10099 (381), Mexico, Petarabo Bay; 22, USNM 100998 (399) Mexico; 23, SIO 62- 61 (419) Mexico, Isabel Id.; 24, SIO 65-182 (431) Mexico, Baja; 25, USNM 206999 (490) Colombia, Baja Utria; 26, USNM 128659 (520) Panama, Mi- raflores Lock; 27, CAS 216871 (538) Tres Marias Is.; 28, SIO 58-83 (588) Peru; 29, SIO 75-383 (643), San Diego Bay; 30, SIO 65-176A (670), Baja California. Appendix 3 Catalog numbers, localities, and sizes (mm FL) of speci- mens used for the outline drawings of the first anal-fin pterygiophore in Figure 15. Numbers in bold font cor- respond to the numbers for the pterygiophores seen in Figure 15. Caranx caninus: 1, SIO-62-61 (419) Mexico, Isabel Is- land; 2, CAS 216872 (421) Baja California; 3, SIO- 62-61 (438) Mexico, Isabel Island; 4, USNM 206999 (490) Colombia; 5, CAS 21871(538) Tres Marias Is- land; 6, SIO 58-83 (588) Peru; 7, SIO 75-383 (643) California, San Diego; 8, SIO 65-176A (670) Baja California. Caranx hippos: 9, USNM 114594 (420) Guatemala; 10, USNM 79981 (440) Panama, Colon; 11, USNM 362541 (460) Florida, Coloosahatchee River; 12, USNM 362541 (470), Florida, Coloosahatchee River; 13, USNM 114618 (485) Guatemala; 14, CAS-SU 53082 (534) Brazil; 15, CAS-SU 53082 (534) Brazil; 16, USNM 114618 (557), Guatemala; 17, CAS-SU 53082 (555) Brazil; 18, CAS-SU 53080 (617) Brazil; 19, ANSP 159674 (655) Mexico; 20, ANSP 102762 (657) Bahamas; 21, ANSP 159674 (698) Mexico. 234 Abstract — The diet of Steller sea lions (Eumetopias jubatus) was determined from 1494 scats (feces) collected at breeding (rookeries) and nonbreeding (haulout) sites in Southeast Alaska from 1993 to 1999. The most common prey of 61 species identified were wall- eye pollock (Theragra chalcogramma) , Pacific herring (Clupea pallasii). Pacific sand lance {Ammodytes hexa- pterus). Pacific salmon (Salmonidae), arrowtooth flounder {Atheresthes sto- mias). rockfish [Sebastes spp.), skates (Rajidae), and cephalopods (squid and octopus). Steller sea lion diets at the three Southeast Alaska rook- eries differed significantly from one another. The sea lions consumed the most diverse range of prey catego- ries during summer, and the least diverse during fall. Diet was more diverse in Southeast Alaska during the 1990s than in any other region of Alaska (Gulf of Alaska and Aleutian Islands). Dietary differences between increasing and declining populations of Steller sea lions in Alaska correlate with rates of population change, and add credence to the view that diet may have played a role in the decline of sea lions in the Gulf of Alaska and Aleutian Islands. Diets of Steller sea lions iEumetopias jubatus) in Southeast Alaska^ 1993-1999 Andrew W. Trites' Donald G. Calkins^ Arliss J. Winship^ Email address for A. W. Trites: tritese'zoology.ubc.ca ' Marine Mammal Research Unit, Fisheries Centre Room 247, AERL - Aquatic Ecosystems Research Laboratory 2202 Main Mall, University of British Columbia Vancouver, BC, Canada V6T 1Z4 2 Alaska Department of Fish and Game 333 Raspberry Road Anchorage, Alaska 99518-1599 Manuscript submitted 10 January 2006 to the Scientific Edfitor's Office. Manuscript accepted: 6 October 2006 by the Scientific Editor. Fish. Bull. 105:234-248 (2007).234 Steller sea lion iEumetopias jubatus) populations in the Aleutian Islands and Gulf of Alaska began declining in the mid-1970s and were listed as endangered under the U.S. Endan- gered Species Act in 1997 (NMFSi; Trites and Larkin, 1996; Loughlin, 1998). The cause of the population decline is uncertain but may be linked to a decrease in the quantity, quality, or availability of prey, in turn caused either by commercial fisheries or by a natural change in the ecosystem (Alaska Sea Grant, 1993; DeMaster and Atkinson, 2002; Trites et al., 2007). Stomach contents and scat analysis indicate that the diets of the declining population may have changed from primarily small, fatty, schooling fishes (such as capelin (Mal- lotus villosus) and sand lance {Ammo- dytes hexapterus)) in the 1950s to one increasingly dominated by walleye pol- lock {Theragra chalcogramma), Atka mackerel {Pleurogrammus monopter- ygius), and flatfish (Pleuronectidae) in the 1970s, 1980s, and 1990s (Mathi- sen et al., 1962; Thorsteinson and Lensink, 1962; Pitcher, 1981; Calkins and Goodwin^; Merrick et al., 1997; Sinclair and Zeppelin, 2002).- Merrick et al. (1997) found a posi- tive relationship between the rate of population change and the diversity of summer Steller sea lion diets in the declining population during the early 1990s. Regions that had the highest rates of decline had the lowest diversi- ties of diet. The greater the diet diver- sity, the slower the rate of population decline. Additional diet data (through to 2001) supported the conclusion that diet diversity had some influence on population success (Sinclair and Zeppelin, 2002; Sinclair et al., 2005). Merrick et al. (1997) hypothesized that animals with less diverse diets may have experienced difficulty ob- taining enough prey. Others have hy- pothesized that consumption of larger proportions of lower energy-dense prey may have exacerbated the effect of diet diversity by increasing the food requirements of sea lions (Alverson, 1992; Rosen and Trites, 2000; Trites and Donnelly, 2003). Sea lions with less diverse, low energy-dense diets may also have been more sensitive to changes in overall prey abundance, and could have theoretically incurred higher rates of predation from killer 1 NMFS (National Marine Fisheries Ser- vice). 1992. Recovery plan for the Steller sea lion {Eumetopias jubatus), 92 p. Prepared by the Steller Sea Lion Recovery Team for the National Marine Fisheries Service, 1315 East-West High- way, Silver Spring, MD 20910-3282. 2 Calkins, D. G., and E. Goodwin. 1988. Investigation of the declining sea lion population in the Gulf of Alaska, 76 p. Unpublished report. Alaska Department of Fish and Game, 333 Raspberry Road, Anchorage, AK 99518-1599. Trites et al.: Diets of Eumelopios jubatus in Southeast Alaska 235 whales if they had to forage for longer periods of time. Population trends in Southeast Alaska have been opposite to those observed in the Gulf of Alaska (Trites and Lar- kin, 1996; Calkins et al., 1999; Pitcher et al., 2007). The robustness of the Southeast population compared to the other regions of Alaska may reflect a difference in diet. One explanation for this finding is that Steller sea lions in Southeast Alaska eat a wider range of prey and therefore have a more diverse diet. Another is that low energy-density prey (such as pollock) do not comprise a significant portion of the sea lion diet in Southeast Alaska. Our goal was to determine the diets of Steller sea lions in Southeast Alaska. We sought to test two hypotheses: 1) diet in Southeast Alaska is the most diverse of all regions inhabited by Steller sea lions; and 2) pollock is not an important prey species in Southeast Alaska. We also wanted to document prey associa- tions and seasonal changes in diet. Materials and methods There are three major breeding areas (rookeries) and over 45 major non- breeding areas (haulouts) in Southeast Alaska. We collected 1494 scats from 12 haulouts and all three rookeries from 1993 through 1999 (Fig. 1). Some areas, such as the Forrester rookery, were sampled every year, and others were sampled less frequently (Table 1). We grouped our analyses into rookeries and haulouts, and then into subgroups by sample size, location, and frequency of sampling. Haulouts consisted of 12 nonbreeding sites in the inside protected waters of Southeast Alaska ( Fig. 1). Rookeries consisted of the three breeding areas in Southeast Alaska (Forrester Island, Hazy Island, and White Sisters Islands). Scats were generally collected opportunistically, when rookeries and haulouts were disturbed in order to count pups or for other research purposes. Each scat was placed in a zip-lock plastic bag and frozen in a 5-gal- lon plastic bucket before it was shipped to the Food and Energy Consumption Laboratory at the Vancouver Aquarium Marine Science Centre for cleaning. Only scats that were big enough or solid enough to likely contain prey remains were collected, and only one scat was collected from any group of scats if there was any doubt about whether the scat came from more than one Steller sea lion. Each thawed scat was transferred to a plastic jar and soaked in water for 1-6 days. Periodic Figure 1 Major rookeries (White Sisters, Hazy Island, and Forrester Island) and haul- outs (all other sites) of Steller sea lions (Euinetopias jubatus) in Southeast Alaska during 1993-99. Labeled sites indicate where scats were collected. shaking of the jars ensured that the scats broke down and formed a uniform slurry at the bottom of the jar. Volume was recorded from graduated markings on each jar. An elutriator removed most of the water-soluble elements (Bigg and Olesiuk, 1990) before the remaining sample was washed through a fine mesh screen. Prey species were identified at Pacific IDentifications Inc. (Victoria, BC) from cleaned and dried hard parts; the types of hard parts that were present and the spe- cies from which they came were also noted. Prey hard parts recovered from scats were compared with hard parts from a reference collection of identified skeletal and nonskeletal hard parts. Otoliths and all other hard parts were identified to the lowest possible taxon. Hard parts that were digested beyond recognition or were not diagnostic for prey taxa were not included in our analysis (e.g., ribs). Some recovered structures, such as otoliths or squid beaks, could be used to estimate 236 Fishery Bulletin 105(2) the type and number of prey consumed, but other hard parts, such as scales, teeth, branchials, and gill rak- ers, could only be used to quantify the type of prey consumed. Scats that were empty or contained prey that could not be identified with certainty were not analyzed. These represented few of the scats collected (56 of 1494, 4%). Unrecognizable hard parts could have been from Table 1 Total number of Steller sea lion [Eu metopiasjubatus) scats collected in Southeast Alaska during 1993- 99 by yea ", location, and month Note that North Rocks, Cape Horn Rocks, and Sea Lion Rocks are part of the Forrester Island rookery complex. Year Location Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Total 1993 Benjamin Is. 28 28 1993 Brothers 7 27 34 1993 Cape Horn Rocks 5 5 1993 Hazy 27 27 1993 North Rocks 8 8 1993 Pt. League 38 38 1993 Sail Is. 80 80 1993 Sea Lion Rocks 9 8 17 Total 49 8 7 173 237 1994 Cape Horn Rocks 29 29 1994 Hazy 54 54 1994 North Rocks 32 32 1994 Sea Lion Rocks 73 73 1994 White Sisters 49 49 Total 134 103 237 1995 Benjamin Is. 16 16 1995 Brothers 25 14 39 1995 Cape Horn Rocks 30 30 1995 Circle Pt. 12 22 34 1995 Horn Cliff 10 10 1995 North Rocks 68 68 1995 Pt. League 13 14 27 1995 Sail Is. 26 6 32 1995 Sea Lion Rocks 30 30 1995 Sukoi Is. 7 7 1995 Sunset Is. 4 4 1995 Sunset Pt. 22 22 1995 Turnabout Is. 10 2 12 Total 128 119 84 331 1996 Benjamin Is. 4 10 14 1996 Brothers 15 20 18 53 1996 Cape Horn Rocks 11 13 24 1996 Dorothy 11 11 1996 Horn Cliff 12 12 1996 Liesnoi Is. 22 22 1996 North Rocks 21 21 1996 Pt. League 11 14 25 1996 Sail Is. 11 11 1996 Sea Lion Rocks 8 8 1996 Sunset Pt. 20 13 45 78 1996 Turnabout Is. 34 34 Total 4 90 44 77 19 34 45 313 continued Trites et a\. Diets oi Eumetopios /ubotus in Southeast Alaska 237 species not in the reference skeleton collection at the time of identification or could have been too far digested to be identifiable. We grouped the identified species of prey into eight categories for statistical analysis. These included ga- dids, forage fish, salmon (Salmonidae), flatfish, rockfish iSebastes spp.), cephalopods, hexagrammids, and other prey (Fig. 2). Scats that contained more than one spe- cies from a particular group were scored as containing only a single occurrence of that group. For example, a scat containing both Pacific herring (Clupea pallasii) and sand lance was scored as having a single occur- rence of forage fish. Hexagrammids do not inhabit the waters of Southeast Alaska in significant numbers but were included as a prey category so that diets could be compared across regions of the North Pacific where hexagrammids are consumed in greater numbers (Mer- rick et al., 1997; Sinclair and Zeppelin, 2002). The diversity of the diet was calculated for the eight prey groups by using the Shannon-Wiener species di- versity index (Rickleffs and Miller, 2000), which yields a value between 1 and 8, where a value of 1 indicates that only one of the eight groups was consumed, and a value of 8 indicates that all eight were equally con- sumed. Merrick et al. (1997) used this index to de- termine the dietary diversity of Steller sea lions that consumed seven prey groups in the Gulf of Alaska and Aleutian Islands. We therefore pooled rockfish with other prey to create the same seven categories used by Merrick et al. (1997) to compare the diversity of diet across all regions of Alaska. We compared our estimate of dietary diversity to those presented by Merrick et al. (1997) for diet data collected between 1990 and 1995. However, we recalculated the diet diversities presented in their paper (from their split-sample frequency of oc- currence data) because of a calculation error in their published values. Seasonal diets were calculated for rookeries in sum- mer (Forrester Island. Jun-Aug, 1993-99) and haulouts in fall (Sep-Nov, 1993 and 1995-96), winter (Dec-Feb, 1996-1997), and spring (Mar-May 1996). Average sum- mer diet (Jun-Aug) was calculated from the three rook- eries — weighted by the average number of pups counted at each site during 1993-1997 (pup counts serving as an index of population size; Trites and Larkin, 1996; Pitcher et al., 2007). The summer data were weighted to indicate what the average Steller sea lion ate in Southeast Alaska, rather than to describe what the average rookery diet was. Fall, winter, and spring diets were given equal weight and averaged to describe the nonsummer diet (haulouts, Sep-May) because animals are more evenly distributed during the nonbreeding season and haulout counts were not available for each of the seasons. The relative importance of prey in the diet was quan- tified as "simple" and "split-sample" frequency of occur- rences. The simple frequency of occurrence indicates what proportion of scats contains any particular prey type. They do not sum to 100%. For example, 80% of the scats examined may contain gadids, and 50% may contain forage fish — meaning that some scats contained both prey types, and others contained only gadids or only forage fish. The second method we used, the split- sample frequency of occurrence (Olesiuk et al., 1990; Olesiuk, 1993), yields the proportion of the overall diet made up of any single prey type. These proportions do sum to 100%. With the split-sample method, it is as- sumed that the scat contained remains from all prey consumed in the previous meal and that the prey were consumed in equal volumes. Table 1 (continued) Year Location Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Total 1997 1997 Cape Horn Rocks North Rocks 25 25 25 25 1997 Sea Lion Rocks 25 25 1997 Sunset Pt. 32 32 1997 Turnabout Is. 27 27 Total 59 75 134 1998 1998 1998 1998 Cape Horn Rocks Hazy North Rocks Sea Lion Rocks Total 27 70 21 21 139 27 70 21 21 139 1999 1999 1999 Hazy North Rocks Sea Lion Rocks Total 6 4 10 60 33 93 60 39 4 103 Grand Total 4 59 90 44 77 415 377 126 218 84 1494 238 Fishery Bulletin 105(2) Rookeries Haulouts Pollock - D n Pacrichake - □ D Gadids Pacific cod - D ■□ Salmon - • • Salmon Sandlance - > E> Pacific herring - > > Smett sp - Eulachon - t> t> > Forage Fish Rainbow smell - t> C> Capelin - > > Rockfish sp - - Rockfish Arrowtooth Itounder - O o Flatfish sp - O Halibul - Rock sole - o o Starry flounder - o o Flatfish Sanddab sp o o Hybrid sole - o Petrale sole o Kamchatka flounder • o ■o Squid/Octopus * Squid sp > > ► Cephalopods Octopus sp " -> > Greenling sp - Atka mackerel - o o o o Hexagrammids Skate - Dogfish - Lamprey sp « Pacific lamprey - Sandfish - -* Polychaete unident * Dogth lamptish Insh lord sp • « Bird - Sablefish - Woffeel - Gunnel - Nonfish - Cassins auklel - Searcher - Rhinoceros auklel Pacific saury Auklel - Wolffish - Sculpin sp Other Poachei sp * « Threespine stickleback « Slaghom sculpin - Gymnocanthus sp - Cabezon - Buffalo-type sculpin - Snailfish sp - Gunnel/Prickleback •♦ Great-type sculpin " « ■•■♦ Great sculpin - Calshark sp " Tidepooi sculpin - Snake prickleback - « Smoolhtongue - « - -♦ Sm lumpsucker - - - Prickleback sp - - fylyclophid - Eelpoul - Buffalo sculpin - — 1 1 1 1 1 1 1 1 1 1 1 i 1 r 5 15 25 50 75 1000 5 15 25 50 75 100 Occurrence of prey (%) Figure 2 Frequency of occurrence of individual prey species in Steller sea lion [Eumetopias jubatus) scats from three Southeast Alaskan rookeries (Forrester, Hazy, and White Sisters; n=752) in summer (Jun-Aug) and haulouts (n = 686) during the rest of the year (Sep-May, 1993-99). Data were pooled across months, years, and sites. Plotted data were transformed (square-root transformed) to improve the visual resolution at lower frequencies of occurrence. The eight symbols identify the eight groups of species used to calculate diet diversity. Statistical analyses were performed on the sim- ple frequency of occurrences to determine whether diets varied by sites and time (across years). We used a contingency table analysis for the total num- ber of scats containing particular prey categories (Pearson x". o<0.05). The cephalopod and hexagram- mid prey categories were not considered in these anal- yses because of their low frequencies of occurrence. Differences in the number of categories of prey con- sumed on each foraging trip (i.e., the number of prey Trites et al Diets of Eumetopias jubatus in Southeast Alaska 239 groups per scat) were compared by using analysis of variance. Associations between prey groups recovered from individual scats were identified by calculating partial correlation coefficients for each pair of prey groups by using presence and absence data with each scat as a replicate (Zar, 19961. This analysis was per- formed for all scats collected at the three rookeries during the summer and for all scats collected at the haulouts during autumn-spring. Partial correlations were considered significant at P=0.05. Prey associa- tions were illustrated by using the hcliist function of S-Plus 2000 (Mathsoft Inc., Seattle, WA) and using the "average" clustering method and the distance between two