U.S. Department of Commerce Volume ttO Number 2 April 2012 Fishery Bulletin U.S. Department of Commerce John E. Bryson Secretary of Commerce National Oceanic and Atmospheric Administration Jane Lubchenco, Ph.D. Administrator of NOAA National! Marine Fisheries Service Samuel D. Rauch III Acting Assistant Administrator for Fisheries Scientific Editor Bruce C. Mundy Associate Editor Julie Scheurer National Marine Fisheries Service Pacific Islands Fisheries Science Center Aiea Heights Research Facility 99-193 Aiea Heights Drive, Suite 417 Aiea, Hawaii 96701-3911 Managing Editor Sharyn Matriotti National Marine Fisheries Service Scientific Publications Office 7600 Sand Point Way NE Seattle, Washington 98115-0070 The Fishery Bulletin (ISSN 0090-0656) is published quarterly by the Scientific Publications Office, National Marine Fish- eries Service, NOAA, 7600 Sand Point Way NE, Seattle, WA 98115-0070. 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: $32.00 domestic and $44.80 foreign. Cost per single issue: $19.00 domestic and $26.60 foreign. See back for order form. Editorial Committee Richard Brodeur John Carlson Kevin Craig Jeff Leis Rich McBride Rick Methot Adam Moles Frank Parrish Dave Somerton Ed Trippel Mary Yoklavich National Marine Fisheries Service, Newport, Oregon National Marine Fisheries Service, Panama City, Florida National Marine Fisheries Service, Beaufort, North Carolina Australian Museum, Sydney, New South Wales, Australia National Marine Fisheries Service, Woods Hole, Massachusetts National Marine Fisheries Sen/ice, Seattle, Washington National Marine Fisheries Service, Auke Bay, Alaska National Marine Fisheries Service, Honolulu, Hawaii National Marine Fisheries Service, Seattle, Washington Department of Fisheries and Oceans, St. Andrews, New Brunswick, Canada National Marine Fisheries Service, Santa Cruz, California Fishery Bulletin web site: www.fisherybulletin.noaa.gov The Fishery Bulletin carries original research reports and technical notes on investigations in fishery science, engineering, and economics. It began as the Bulletin of the United States Fish Commission in 1881; it became the Bulletin of the Bureau of Fisheries in 1904 and the Fishery Bulletin of the Fish and Wildlife Service in 1941. Separates were issued as documents through volume 46; the last document was No. 1103. Beginning with volume 47 in 1931 and continuing through volume 62 in 1963, each separate appeared as a numbered bulletin. A new system began in 1963 with volume 63 in which papers are bound together in a single issue of the bulletin. Beginning with volume 70, number 1, January 1972, the Fishery Bulletin became a periodical, issued quarterly. In this form, it is available by subscription from the Superintendent of Documents, U.S. Government Printing Office, Washington, DC 20402. It is also available free in limited numbers to libraries, research institutions, State and Federal agencies, and in exchange for other scientific publications. U.S. Department of Commerce Seattle, Washington Volume 110 Number 2 April 2012 Fishery Bulletin Contents Articles 143-155 Matkin, Craig O., John W. Durban, Eva L. Saulitis, Russel D. Andrews, Janice M. Straley, Dena R. Matkin, and Graeme M. Ellis Contrasting abundance and residency patterns of two sympatric populations of transient killer whales ( Oranus orca) in the northern Gulf of Alaska 156-175 Rudershausen, Paul J., Jeffrey A. Buckel, Greg E. Bolton, Randy W. Gregory, Tyler W. Averett, and Paul B. Conn A comparison between circle hook and J hook performance in the dolphinfish, yellowfin tuna, and wahoo troll fishery off the coast of North Carolina The National Marine Fisheries Service (NMFS) does not approve, recommend, or endorse any proprie- tary product or proprietary material mentioned in this publication. No reference shall be made to NMFS, or to this publication furnished by NMFS, in any advertising or sales promotion which would indicate or imply that NMFS approves, recom- mends, or endorses any proprietary product or proprietary material mentioned herein, or which has as its purpose an intent to cause directly or indirectly the advertised product to be used or purchased because of this NMFS publication. 176-192 Able, Kenneth W., Thomas M. Grothues, Jason T. Turnure, Donald M. Byrne, and Paul Clerkin Distribution, movements, and habitat use of small striped bass ( Morone soxatilis ) across multiple spatial scales 193-204 Tinus, Craig A. Prey preference of Imgcod ( Ophiodon elongatus), a top marine predator: implications for ecosystem-based fisheries management 205-222 Keller, Aimee A., John R. Wallace, Beth H. Horness, Owen S. Hamel, and Ian J. Stewart Variations in eastern North Pacific demersal fish biomass based on the U.S. west coast groundfish bottom trawl survey (2003-2010) The NMFS Scientific Publications Office is not responsible for the con- tents of the articles or for the stan- dard of English used in them. 223-229 Ohnishi, Shuhei, Takashi Yamakawa, Hiroshi Okamura, and Tatsuro Akamine A note on the von Bertalanffy growth function concerning the allocation of surplus energy to reproduction II Fishery Bulletin 1 10(2) 230-241 Baremore, Ivy E., Dana M. Bethea, and Kate 1. Andrews Gillnet selectivity for juvenile blacktip sharks ( Carcharhinus limbotus) 242-256 Beacham, Terry D., Brenda McIntosh, Cathy MacConnachie, Brian Spilsted, and Bruce A. White Population structure of pink salmon (Oncorhynchus gorbuscha) in British Columbia and Washington, determined with microsatellites 257-270 Caldarone, Elaine M., Sharon A. MacLean, and Beth Sharack Evaluation of bioelectrical impedance analysis and Fulton's condition factor as noniethal techniques for estimating short-term responses in postsmolt Atlantic salmon (Salmo salar) to food availability 271-279 Stachura, Megan M., Chris R. Lunsford, Cara J. Rodgveller, and Jonathan Heifetz Estimation of discard mortality of sablefish (Anoplopoma fimbria) in Alaska longline fisheries 280-281 Guidelines for authors 143 Contrasting abundance and residency patterns of two sympatric populations of transient killer whales ( Orcinus orca ) in the northern Gulf of Alaska Abstract — Two sympatric popula- tions of “transient” (mammal-eating) killer whales were photo-identified over 27 years (1984-2010) in Prince William Sound and Kenai Fjords, coastal waters of the northern Gulf of Alaska (GOA). A total of 88 indi- viduals were identified during 203 encounters with “ATI” transients (22 individuals) and 91 encounters with “GOA” transients (66 individuals). The median number of individuals identified annually was similar for both populations (AT1=7; GOA = 8), but mark-recapture estimates showed the ATI whales to have much higher fidelity to the study area, whereas the GOA whales had a higher exchange of individuals. Apparent survival esti- mates were generally high for both populations, but there was a signifi- cant reduction in the survival of ATI transients after the Exxon Valdez oil spill in 1989, with an abrupt decline in estimated abundance from a high of 22 in 1989 to a low of seven whales at the end of 2010. There was no detectable decline in GOA population abundance or survival over the same period, but abundance ranged from just 6 to 18 whales annually. Resight- ing data from adjacent coastal waters and movement tracks from satellite tags further indicated that the GOA whales are part of a larger popula- tion with a more extensive range, whereas ATI whales are resident to the study area. Manuscript submitted 1 June 2011. Manuscript accepted 18 October 2011. Fish. Bull. 110:143-155 (2012). The views and opinions expressed or implied in this article are those of the author (or authors) and do not necessarily reflect the position of the National Marine Fisheries Service, NOAA. Craig O. Matkin (contact author)1 John W. Durban2 3 Eva L. Saulitis' Russel D. Andrews4 Killer whales (Orcinus orca ) in the eastern North Pacific can be geneti- cally and acoustically separated into three nonassociating lineages: “resident,” “transient,” and “offshore” (Ford and Ellis, 1999; Matkin et ah, 1999; Barrett-Lennard, 2000; Yurk et al., 2002). Of these lineages, Morin et al. (2010) found the transients to be the most genetically divergent and indicated that they should be considered a separate species. Only the transient form has been observed consuming marine mammals in this region and observations indicate that they feed on marine mammals exclusively (Ford et al., 1998; Sauli- tis et al., 2000; Herman et al., 2005; Matkin et al., 2007a, 2007b; Barrett- Lennard et al., 2011). The potential for these whales to affect trajectories of prey populations has led to con- siderable debate over the role of pre- dation by transient killer whales in the decline of coastal pinnipeds and sea otters in western Alaska (e.g., Estes et al., 1998, 2009; Springer et al., 2003, 2008; DeMaster et al., 2006; Wade et al., 2007, 2009). In Janice M. Straley5 Dena R. Matkin1 Graeme M. Ellis6 addition to data on feeding habits, evaluation of their top-down impact requires data on abundance and res- idency patterns of these transient killer whales within specific marine systems, particularly with respect to the abundance and trends of their primary prey. The coastal waters of Prince Wil- liam Sound and the Kenai Fjords in the northern Gulf of Alaska are unique in being regularly used by two sympatric populations of tran- sient killer whales (Matkin et al., 1999). Members of both the Gulf of Alaska-Aleutian Islands-Bering Sea transient stock and the ATI transient stock (Allen and Angliss, 2010) have been photographically identified over the past 27 summer seasons (Matkin et al., 1999, 2008). Individuals from both populations regularly use the same region but have never been recorded swimming together and do not associate (Mat- kin et al., 1999), and they can be separated by behavior (Matkin et al., 1999; Saulitis et al., 2000), by acous- tics (Yurk et al., 2002, 2010; Saulitis Email address for contact author cmatkm@acsalaska.net 1 North Gulf Oceanic Society 3430 Mam St., Suite B1 Homer, AK 99603 2 National Marine Mammal Laboratory Alaska Fisheries Science Center National Marine Fisheries Service, NOAA 7600 Sand Point Way NE Seattle, WA 98115 3 Protected Resources Division Southwest Fisheries Science Center National Marine Fisheries Service, NOAA 3333 N. Torrey Pines Ct. La Jolla, CA 92037 4 School of Fisheries and Ocean Sciences University of Alaska Fairbanks and Alaska SeaLife Center 301 Railway Ave. Seward, AK 99664 5 University of Alaska Southeast Sitka Campus 1332 Seward Ave Sitka, AK 99835 6 Department of Fisheries and Oceans Pacific Biological Station 3190 Hammond Bay Rd. Nanaimo, British Columbia, V9R 5K6 Canada 144 Fishery Bulletin 1 10(2) et al., 2005), and by genetics (Barrett-Lennard et al., 2000). Because of the lack of conclusive studies of genetic divergence across their range, the Gulf of Alaska- Aleutian Islands-Bering Sea transient stock includes all transient killer whales found in Alaskan waters west of southeastern Alaska other than the ATI stock (Allen and Angliss, 2010). However, photographic mark-recapture analyses indicate little apparent overlap between the Gulf of Alaska whales and the western segment of the stock (Matkin et al., 2007a; Durban et al., 2010). In this article we will refer to the non-ATI transients in the study area only as the Gulf of Alaska (GOA) transient population and consider their range to be the Gulf of Alaska and north gulf coast, which stretches from southeast- ern Alaska west through the Kodiak Island region. Although the full range and offshore distribution of the GOA transients is poorly defined, they have been photographed irregularly to the southwest of Prince William Sound-Kenai Fjords study area in Kachemak Bay, lower Cook Inlet, and Kodiak Island waters (Maniscalco et al., 2007; Matkin et al., 1999; C. Matkin, unpubl. data). The ATI transients are considered a separate stock, are classified as depleted under the Marine Mammal Protection Act, and currently number only seven in- dividuals (Allen and Angliss, 2010). The home range of the ATI transient population appears much more restricted than that of the sympatric GOA transients (Matkin et al., 1999; Scheel et al., 2001) or the para- patric west coast transients of southeastern Alaska, British Columbia, and Washington State coastal wa- ters (Ford and Ellis, 1999). ATI individuals have not been identified outside of the coastal waters of Prince William Sound and the Kenai Fjords (Matkin et al., 1999; Saulitis et al., 2005). Because of its limited range, small population size, and the consistent re- sightings of subgroups and individuals, the population dynamics of the ATI population have been monitored directly from annual photographic data (Matkin et al., 2008). However, for the GOA transients, the infrequent resightings of individuals, fluidity in group structure, and larger population size have made it impossible to directly track births and deaths and require a mark- recapture sampling approach to estimate abundance and assess population changes. In this article we fit mark-recapture models to long-term photographic identification data (1984 to 2010) to examine abundance trends, site fidelity, and demography for the ATI and GOA transients in the coastal waters of Prince William Sound and the Ke- nai Fjords. We compare our results with previously described changes in the ATI population (Matkin et al., 2008) and contrast these results with our paral- lel analysis of the GOA transient population. We use photographic resighting data and satellite telemetry data to further differentiate the range of the two populations and provide a context for their differing abundance trends. Materials and methods Photographic mark-recapture Identification photographs of killer whales were obtained from the waters of Prince William Sound, Kenai Fjords National Park, and the adjacent coastal waters of the northern Gulf of Alaska (Fig. 1A). The entire region was not surveyed in any given year; how- ever, survey effort was focused towards Prince William Sound in the earlier years of the study (1980s) and was more evenly balanced across the region in later years. Photographic surveys were conducted between April and September over the 27-year period between 1984 and 2010. In order to increase capture probabilities, survey effort was focused in areas known to be used by killer whales, or in response to sighting reports. Data were collected from a variety of platforms; all were small vessels less than 15 meters in length powered by either gasoline-outboard or diesel-inboard engines. During an encounter, whales were approached at a distance of 15 to 45 m and photographs were taken of the left side of each whale present, showing details of the dorsal fin and saddle patch (Matkin et al., 1999). Photographs were obtained with either 1) a Nikon F-100 SLR camera1 with fixed 300-mm lens and Fuji Neopan 1600 black and white film, or 2) Nikon D70 and D200 digital cameras with 80-200 mm zoom or 300-mm fixed lenses. Individual whales were distin- guished by the shape and pattern of natural markings on their dorsal fins and adjacent saddle patch (Matkin et al., 1999) and were subsequently matched to cata- logs of photographs from previous years. Individual matches were corroborated by using co-occurrence with consistent associates because transient killer whales have been shown to travel in stable (and often life-long) matrilineal groupings (Ford and Ellis 1999; Matkin et al., 1999). Photographs were evaluated for quality, and only photographs resulting in reliable identifications were used. Typically, the entire group was photographed. Membership in the ATI or GOA transient population was determined either by genetic sampling, acoustic analysis, or observation of repeated association with other members of the population. We treated these photographic identifications and re-identifications as “captures” and “recaptures” to which analytical mark-recapture techniques could be applied for estimation of abundance and demographic parameters (Hammond, 1986, 1987, 1990). Individual whales were not seen in every year that they were known to be alive, likely in part because of the movement patterns of whales relative to the geo- graphical boundaries of the study area. This factor highlighted the need to allow for temporary emigration in the capture-recapture modeling. The popular Cor- mack Jolly Seber (CJS) model for estimating survival 1 Mention of trade names or commercial companies is for identification purposes only and does not imply endorsement by the National Marine Fisheries Service, NOAA. Matkm et al Abundance and residency patterns of two sympatric populations of Orcmus orca in the northern Gulf of Alaska 145 1 50°W 1 40C,W 1 30°W B Prince William Sound ' * * fjords 4 • "* *rf r Montague Strait WSg "Wm f r^r^i— 1 — | — I — r 0 37.5 75 150 Kilometers ! 000m contour Figure 1 (A) Location of the coastal study area of Prince William Sound and Kenai Fjords. (B) Locations of encounters with ATI (203, closed circles) and Gulf of Alaska (GOA) (91, open circles) transient killer whales (Orcinus orca) between 1984 and 2010, during which photo-identification data were collected. (Lebreton et al., 1992) does not account for animals that emigrate from the study area and return later. Instead, we followed Whitehead (1990) in de- veloping a mark-recapture model that parameterized emigration and re-im- migration probabilities in addition to survival. Our model was based on an individual-specific factorization (e.g., Schofield et al., 2009), allowing modu- larization into conditional distributions for capture probability, availability of whales for capture (temporary emigra- tion), and death. This formulation al- lowed imputation of partially observed data on availability in the study area (available in the study area when ac- tually identified) and survival status (alive when identified and between years of repeat identification), provid- ing identifiability of parameters and enabling time-varying formulations. Specifically, the model had the param- eters tpf, Kt, A,, and 03, where <\>t l is the probability of survival from time t— 1 to time t; A,_; is the probability of tempo- rary emigration from the study area at time t- 1; Kt is the annual probability of re-immigration back into the study ar- ea; and 03, is the probability of capture at time t for whales alive and avail- able to be captured in the study area. Note that owing to the geographic re- strictions of the surveys and the likely wider ranging patterns of the whales, survival in this case represented ap- parent survival that could comprise either death or permanent emigration (at least for the duration of the study). To fully quantify uncertainty about the unknown parameters, we adopted a Bayesian approach to model fitting and inference, where estimates were pre- sented as full probability distributions (Gelman et al., 1995). The Bayesian approach requires prior distributions to be specified for all model parameters, and we adopted similar hierarchical priors for each set of probability terms ® ) g _ Bernoulli(0.5), where logit(a) = log(a/(l-a). The prior distribution for each parameter was thus determined by two hyper-parameters: p represented the mean value across each set of parameters and the standard deviation term o represented the year-to-year variability over the set, on the logit scale. Uniform(0,l) prior distributions were placed on each of the five mean probabilities pAA.h-.ra anc] a uniform(0,10) prior distri- bution was adopted for ctAA kb auow annual differ- ences from the logit-transformed means to emerge. The probability (evidence) of temporal variability in each parameter vector was assessed through indicator vari- 146 Fishery Bulletin 1 10(2) ables g (e.g., Kuo and Mallick, 1998). Each of these indicators was assigned a Bernoulli prior distribution, such that the prior probability of including any annual effect was 0.5. We used the freely available WinBlJGS software (Lunn et ah, 2000) to implement Markov Chain Monte Carlo (MCMC) sampling to make repeated draws from the “posterior distribution” of each parameter — the pri- or distribution was updated conditionally on the data and structural relationships of the model. We sampled 10,000 values from the posterior distribution of each parameter, after discarding an initial burn-in deter- mined by the method of Brooks and Gelman (1998). The sampled values were then used to estimate sum- mary statistics for the posterior distributions. MCMC approaches can similarly be used to sample from the posterior distribution of quantities that can be derived as functions of parameters. Notably, we used the same MCMC simulation approach to generate predictive ob- servations from the model parameters and compared the fit of our re-immigration model to a standard Cor- mack Jolly Seber model based on the mean squared pre- dicted error (MSPE; Gelfand and Ghosh, 1998; Durban et al., 2010). As with other model selection methods, this predictive approach achieves a compromise between the goodness-of-fit and a penalty for model complexity (Gelfand and Ghosh, 1998). As such, the model with the smallest MSPE was estimated to provide the best fit. Assessing trends We used estimates of the capture probabilities (03,) to derive estimates of the abundance of animals {Nt) using the study area in any given annual survey period (t). These parameters were linked to the observed data by specifying the number of individuals actually observed in the study area (nt) as a binomial sample from the study area abundance (Nt) with the binomial proportion given by the estimated G3r To assess trends across years, we modeled each Nt as Poisson distributed and adopted a model for the unknown Poisson means (m() that governed the form of the variation between years. Specifically, we therefore adopted a flexible change-point model to describe temporal transitions (e.g., Carlin et ah, 1992): log (mt) = /30 + gPfifiit - c) + e,N etN ~ N(0, c A) (2) gP ~ Bernoulli(0.5). The parameter )30 described the general intercept of the model (or level of abundance on the log scale before the change-point), and the function 50 represented a step function, defined as 1 if its argument was zero or posi- tive and zero otherwise. The parameter j\ described the magnitude of a step change (on a log scale), at time c (known as a change-point). We assumed the timing of the change-point was unknown and used the data to assess the evidence for a change-point in each of the 27 years. This problem therefore involved estimating the posterior distribution of the unknown temporal change- point (c) to identify when a change-point may have occurred, and with what probability. The model offers a flexible approach for modelling changes in abundance, because uncertainty about the year of the change-point results in uncertainty over how the trend is apportioned over the time series of between-year transitions. Because the step function 50 was specified on a discrete time period (t - c), we placed a discrete uniform prior for c over T-21 years) (e.g., Carlin et ah, 1992): c ~ U(1,T) (3) with discrete prior probability of 1/T being placed on each of the 27 years. We assumed that the direction and magnitude of the change was unknown, and we there- fore assigned diffuse prior distributions for the hyper- parameters [i0 and /3,, each with mean 0 and standard deviation of 10. We assessed the probability of a trend in abundance by estimating the indicator probability^ of including the trend parameter /3j in the model for the abundance estimates. Rather than perform this trend analysis independent- ly of the mark-recapture model, we combined these two components into a single Bayesian hierarchical model to propagate uncertainty in estimation of capture probabil- ities (03,) into estimates of abundance (Nt) and trend pa- rameters. We did not assume that the Nt fell exactly on the trend line, or had a common variance, but instead we included annual random-effects terms (etN) that al- lowed over-dispersion in contrast to a fixed-effects Pois- son trend model. A normal random effects distribution was adopted for the etN ~ N(0, oN), with overdispersion controlled by the standard deviation ( oN ), which was assigned a uniform (0,10) prior distribution. As with the mark-recapture parameters, we used WinBUGS to sample 10,000 values from the marginal posterior dis- tributions for the annual estimates of abundance, Nr Additionally, interest was focused on making inference about the posterior distributions of the parameters of the trend model, specifically the change-point (c), the rate of change (/3j), and the probability of a trend (gP). Tracking whale movements To examine movements of whales relative to our mark- recapture modeling estimates (extent of temporary emigra- tion away from the study area), we compared photographs used in our analysis with those taken during parallel research efforts in southeastern Alaska, British Columbia, and Washington State (e.g., Ford and Ellis, 1999) to iden- tify annual overlap of individuals. Previous analyses had shown no overlap of ATI or GOA transients with those in the Aleutian Islands (Durban et al., 2010). In addition, we attached satellite transmitter tags to individual GOA and ATI transient whales to provide fine-scale tracks of daily movements. The tag design was a low impact minimally percutaneous external-electronics transmitter (LIMPET) satellite tag (Andrews et al., 2008). In this tag, the main electronics package, an Argos-linked, location-only SPOT Matkin et at: Abundance and residency patterns of two sympatric populations of Orcmus orca in the northern Gulf of Alaska 147 ■ ATI GOA lllllllllllllllll 4^* of") 4^ oT'' 4^ *4? cO' O?- Cp Cp CS^ oP cO' Cv Q\^ Cv cC^ CV C\^ op cjo cp cjo .Op .dp c? J or cp . o? < cp cf J®* ^ ^ ^ ^ ^ ^ Year Figure 2 The number of individuals photographically identified from the ATI and Gulf of Alaska (GOA) transient killer whale ( Orcinus orca) populations in each annual April-September period, 1984-2010. 5 transmitter (Wildlife Computers, Redmond, WA), is housed in an epoxy casing with dimensions of 65x30x22 mm. The tag is held externally on the dorsal fin of the whale by two 4-mm-diameter medical-grade tita- nium darts that were affixed to the bottom of the tag, for a total mass of 49 g. The darts were designed to penetrate 6.5 cm into the connective tissue in the dorsal fin and remain embedded with a series of backward- facing barbs which acted as anchors for the darts. The LIMPET tags were projected onto the whales by using a crossbow with 150-lb draw weight, and the tag was held on the end of an arrow in a special rubber boot. This type of satellite tag transmits ultra-high frequency (UHF) radio signals to Argos receivers onboard weather satellites in sun-synchro- nous polar orbits. To conserve power, transmissions are limited by a sub- mersion sensor to times when the whale is at the surface. Locations were calculated by the Argos sys- tem by the method of least squares ( http://www.argos-svstem.org. accessed October 2007), and we determined the plausibility of each location us- ing the Douglas Argos filter, vers. 7.03 (Douglas2). We retained locations with high location accuracies (LC2 and LC3), as well as consecutive points separated by less than 3 km. All other locations were removed if the rate of movement between consecutive locations exceeded 25 km/h or the angle formed by the previous and sub- sequent locations indicated extreme return-movements. The angle of each triad of points and the distance be- tween the shortest leg of the triad was assessed by the filter and compared with the dimensionless rate coeffi- cient (Ratecoef) that was set to 25. Location data were imported into Google Earth (Google, Mountain View, CA) for basic visual inspection and into ArcMap 9.3.1 (ESRI, Redlands, CA) for further analysis. Distance traveled was calculated for each tagged animal, as well as a calculation of oceanic home range developed by subtracting the land area from the total area in the Minimum Convex Polygon, which was the polygon that described the perimeter of all filtered satellite locations received during the period of attachment. Results During the 27 years of this study we averaged 106 (range=29-249) vessel days per year with at least 59 2 Douglas, D. 2007. The Douglas Argos-Filter. [Available at http://alaska.usgs.gov/science/biologv/spatial/manual.html. accessed 1 October 2007.] days logged in all years except for 1987 (29 days). During these surveys we recorded a total of 203 encounters with members of the ATI transient population and 91 encounters with members of the GOA transient popula- tion (Fig. 1). Over 27 annual (May-September) periods, a total of 88 individual whales were documented. There were three times as many GOA individuals (66) as ATI individuals (22), but the average number of individu- als identified in each summer interval was similar for both populations (GOA: median=8, range 0-18; ATI: median=7, range 4-22; Fig. 2). This finding reflected a higher resighting rate for individual ATI whales; indi- viduals were seen in a median of nine different annual intervals (range 3-25) compared with a median of just two intervals (range 1-16) for GOA whales (Fig. 3). It is notable that 7 of 22 total ATI whales were identified in more than 20 annual intervals, whereas only 1 of 66 GOA whales was identified in more than 10 intervals (Fig. 3). The number of ATI individuals seen each year clearly declined across the study period from around 20 individuals in the 1980s to fewer than 10 individuals in the 2000s, whereas the number of GOA individuals remained at a more consistent but low number with a median 8 individuals identified per year (Fig. 2). However, to formally assess changes in abundance, we adjusted our sighting data for capture probabilities using mark-recapture models. The mark-recapture model with emigration and re- immigration provided a better fit to the photo-identifi- cation data than the standard CJS model, for both ATI and GOA individuals. For GOA whales, there were 49 discrepancies between 1079 observed and predicted data 148 Fishery Bulletin 1 10(2) points under the re-immigration model, compared with 190 of 1079 data points for the CJS model, translating to a mean squared predicted error of 0.04 and 0.18 re- spectively. For ATI whales, there were 25 discrepancies from 569 data points under the re-immigration model, compared with 91 in 569 for the CJS model, correspond- ing to MSPEs of 0.04 and 0.16, respectively. Inference was therefore based on parameter estimates from the flexible emigration-re-immigration model, which in- dicated notable differences in the fidelity of the two populations to the study area (Table 1). GOA transients showed a much higher rate of ex- change of individuals in the study area, with a rela- tively high probability of emigration (posterior median pA=0.55) and low rate of re-immigration (pK'=0.17), compared to a low rate of emigration and high rate of re-immigration for the ATI population (pA=0.08, pK'=0.77), implying high study area fidelity for the ATI whales. Similarly, the average probability of capture was higher for AT1(/jb = 0.98) compared with GOA (jU® = 0.83) individuals, imply- ing that almost all of the ATI in- dividuals in the study area were photographed in each year, likely because of a higher fidelity to the study area and smaller range. Al- though the average apparent sur- vival was high for both populations (GOA ^=0.98; ATI /i*=0.99), there were noticeable annual deviations from the average (Fig. 4). Although there was a substantial dip in the GOA transients’ apparent survival in one year, 1986, there was a con- sistent trend in the ATI population, with survival from 1989 to 1990 showing a marked decrease (poste- rior median = 0.68, 95% probability interval = 0.48 to 0.86) compared with the overall average, with no overlap in 95% probability intervals between this estimate and those for most other years. The trends in abundance of the two populations, based on estimates of abundance and parameters of the 20 Number of years seen Figure 3 Frequency plot of the number of individual whales photographed in different numbers of annual sampling periods, for both the ATI and Gulf of Alaska (GOA) transient killer whale (Orcinus orca) populations. Table 1 Fit of photographic identification data to the mark-recapture model with emigration and re-immigration, for both ATI and Gulf of Alaska (GOA) transient killer whale ( Orcinus orca) populations. l=the probability that an individual in the study area migrates out of it each year; v=the probability that an individual not in the study area population migrates back into it each year; 0=the annual probability of survival, G5=the annual probability of capture (identification) in the study area. Estimates are presented as the 0.025,0.50, and 0.975 probability intervals of the posterior probability distribution (i.e., median surrounded by 95% prob- ability intervals) for the average (p) value across May-September periods, plus the probability of between-year differences in parameters over the 27 time periods, given by the posterior probability p(g= 1) of each respective time-varying indicator variable g. Additionally, the parameter /3j is included to indicate the magnitude and direction of abundance trend (on the log scale). Posterior estimates Population Emigration ft [p(gA=D] Re-immigration gK lp(gK= D) Survival ft [p(ft=l)] Capture ft [p(gw= D] Trend ft [p(ft=l>l GOA 0.21,0.55,0.80 [1.00] 0.02,0.17,0.67 [1.00] 0.94,0.98,0.99 [1.00] 0.55,0.83,0.99 [1.00] -1.2,04,1.3 [0.02] ATI 0.02,0.08,0.23 [1.00] 0.18,0.77,0.97 [0.71] 0.96,0.99,1.00 [1.00] 0.92,0.98,1.00 [0.64] -1.1, -0.8, -0.5 [1.00] Matkin et at: Abundance and residency patterns of two sympatric populations of Orcmus orca in the northern Gulf of Alaska 149 T T T T T 0.8 > 0.6 O .S' I n I t n i ATI 6^ 6^ of?’ cf*® bP>' 6piV cA cv v't <=r V V* V V V V V \’ V V S’ V V V V \ Year 9- "co > £ 3 00 O £ j5 s o cl HHH GOA 0.2 Year Figure 4 Annual estimates of the probability of survival t, for both ATI and Gulf of Alaska (GOA) transient killer whales (Orcinus orca). Solid squares represent the posterior median, and bars represent the 95% probability intervals. trend model, revealed contrast- ing patterns (Figs. 5 and 6). The estimated number of GOA whales using the study area in each an- nual period, Nt, showed relatively little variation from a low pos- terior median of six whales in 1996 and 2005 to a high of 18 in 1990. The ATI whales showed evidence of greater abundance changes, from an estimated high of around 22 in 1989 to a low of seven at the end of the series. As a result, there was strong evi- dence that the abundance of ATI whales declined over the study pe- riod and unequivocal support for inclusion of the trend model for abundance with p(gP= 1) = 1. The entire posterior distribution for the trend parameter, /3lt fell be- low zero, indicating a probability of 1.00 of a downward trend. In contrast, the posterior distribu- tion for the trend parameter was evenly spread above and below zero for GOA whales, with 51% of the posterior density in favor of a negative trend. As a result, there was little support for including a model for trends in abundance with p(gP= 1) = 0.02. Correspond- ingly, the posterior density for the change-point was distributed evenly across all years for GOA whales and reflected no obvious changes in abundance. For ATI transients, in contrast, there was a distinct peak in the poste- rior probability distribution for a change-point, and 97% of the pos- terior density for an abundance change occurred in the five years after 1989. Emigration of GOA transients away from the study area was also supported by photographic resighting data from southeast- ern Alaska and British Columbia (Table 2). For 1995-2007 there were 16 encounters with GOA transient whales in these adjacent regions, including one (5 June 2001) with GOA transients in association with known mem- bers of the west coast transient population (Matkin et al., 2007b). There were no resighting data outside of Prince William Sound-Kenai Fjords for ATI tran- sients despite substantial survey effort in southeastern Alaska (Dahlheim and White, 2010) and in adjacent regions to the west of the study area (Matkin et al., 1999, 2007a; Durban et al., 2010; senior author, un- publ. data). Additionally, the one tagged ATI transient did not travel out of the area. All these observations support the inference from the mark-recapture model of high fidelity to the study area. Satellite-monitored LIMPET tags were attached to GOA transients in Prince William Sound on four occa- sions for a total of 73 days of transmissions (Table 3). One individual (AT73) was tagged on two occasions in different years. Tagged whales traveled a total dis- tance of 7107 km during 73 days for an average move- 150 Fishery Bulletin 1 10(2) 25 ! cA eft. op, ,o. ^ Noi° Nor Ncr scr sncsuedu 1 Department of Biology Center for Marine Science and Technology North Carolina State University 303 College Circle Morehead City, North Carolina 28557 2 North Carolina State University Department of Food, Bioprocessing, and Nutrition Sciences Center for Marine Science and Technology 303 College Circle Morehead City, North Carolina 28557 Rudershausen et al.: A comparison between circle hook and J hook performance in the troll fisheries off North Carolina 157 Stream waters off of North Carolina. Catch composi- tion of dolphinfish, yellowfin tuna, and wahoo in this troll fishery varies widely among vessels, seasons, and locations in the Gulf Stream. There is concern in the charter boat industry that circle hook regulations (developed for and based on bill- fish), if ever mandated outside of U.S. Atlantic billfish tournaments, would negatively impact catch rates of dolphinfish, yellowfin tuna, and wahoo. Chartering an offshore fishing trip in the southeastern United States is an expensive endeavor (~$2000/day; senior author, personal observ. ) and reductions in catch may have eco- nomic influences on charter fishing businesses. Success of the offshore troll fishery relies on clientele having a reasonable chance to catch and keep fish that are highly valued as seafood. In North Carolina, there are few charter captains willing to use or experiment with circle hooks when targeting non-billfish species because there is a widespread perception that trolling circle hooks for non-billfish species results in reduced catch rates, and therefore greater chances for customer dis- satisfaction, compared with J hooks. The charter ocean fishing industry in North Carolina includes roughly 750 vessels and receives $65 million annually in for- hire fees (Dumas et ah, 2009). Economic ramifications of requiring circle hooks outside U.S. Atlantic billfish tournaments have not been quantified. Our purpose in undertaking this study was to deter- mine the effects of using circle hooks on catch levels of non-billfish species in the U.S. southeastern offshore troll fishery in comparison with catch levels with J hooks. Mechanisms that might explain differences or similarities in catch between hook types were also ex- amined. Questions were the following: 1) Did predators strike circle and J hook rigged baits at similar rates?; 2) Once struck, did circle and J hook rigged baits have similar proportions of hook-ups?; and 3) Once hooked up, did circle and J hook rigged baits have similar pro- portions of retained fish (brought to the boat)? Materials and methods Fishing techniques workshop In November 2007 we convened a workshop attended by state and federal biologists, fishery managers, charter boat captains and mates, private boat anglers, and bill- fish tournament directors. The purpose of the workshop was to select hook types, hook styles, rigging techniques, and fishing techniques (see below) that could be used to compare trolled circle and J hooks in Gulf Stream waters off North Carolina during troll fishing days aboard charter vessels. Defining and selecting circle and J hooks was a cen- tral part of the workshop. A circle hook was defined as having the point perpendicular to the hook shank. A J hook was defined as having the point and point shank parallel to the hook shank. We selected circle and J hooks that would be comparable in bend diameter (gap between hook shank and point shank). For both hook types, we selected barbed hooks with zero offset and straight hook eyes (eye parallel to the hook shank). The circular shape, hook point turned perpendicularly toward the shank, and zero offset insured that the circle hooks we selected conformed to the National Ma- rine Fisheries Service definition in the current billfish tournament regulations (Federal Register, 2006). Other hook characteristics (hook size, hook gauge, gap width, and shape) were selected to avoid compromising the action and durability of the trolled dead whole fish (bal- lyhoo [Hemiramphus brasiliensis]). Participants decided that circle and J hooks with a gap width large enough that allowed space between the bait and hook for hook- ing fish but with a relatively low profile (by virtue of the gauge of hook wire) would be most appropriate for testing. Bait rigging and fishing techniques The bait rigging techniques for each non-billfish species presented at the workshop were those used by the local charter industry. Circle and J hooks were embedded in ballyhoo except for directed trips for dolphinfish, when circle hooks were rigged externally (Fig. 1). Other dif- ferences in bait rigging and fishing techniques are described below by species. Flook sizes and styles, leader characteristics, and rigging techniques differed slightly on recreational trips because these fishermen often troll with smaller hooks and different rigging techniques from those used by charter captains. For charter and recreational trips targeting dolphin- fish we used Mustad 9175 7/0 J hooks (Mustad, Gjovik, Norway1) that were rigged inside ballyhoo; the chin weight was affixed to 30 cm of rigging wire. We used Eagle Claw 2004ELG 8/0 circle hooks (Eagle Claw Fishing Tackle Co, Denver, CO) rigged externally to the ballyhoo with a 7-g chin weight and swivel at the top of the head, with 30-cm of rigging wire (no pin). The leader was 1.8 m of 36 kg of monofilament attached to the standing line with a 31-kg Sampo ball-bearing swivel (Sampo Inc., Barneveld, NY). The leader was attached to the hook by using a loose crimp with tag end opposite the point (Fig. 1A). We used lever drag reels affixed to “thirty pound-class” stand-up rods at all locations. Reels were spooled with a 14-kg test Dia- mond® monofilament line (Diamond Fishing Products, Pompano Beach, FL). The drag upon strike of a fish was set just above “free spool” (reel gear not engaged) with the clicker in the “on” position. The drag during the fight of a fish (regardless of species) was roughly 6.4 kg. Baits were dropped back (line allowed to come off the spool with no drag) to missed fish (that struck) immediately after the strike. Recreational rigging tech- niques for dolphinfish were similar to those used on the 1 Mention of trade names or commercial companies is for identification purposes only and does not imply endorsement by the National Marine Fisheries Service, NOAA. 158 Fishery Bulletin 1 10(2) charter vessel except that 1) circle and J hooks were one size smaller. Charter trips targeting yellowfin tuna used Mustad 7692 9/0 J hooks and Eagle Claw 2004ELG 9/0 circle hooks rigged inside ballyhoo, with 7-g chin weights affixed to a pin. Hook and leader were secured to the bait with a rubberband (for wahoo see next paragraph). The leader was 9 m of 36 kg of clear fluorocarbon leader through which a blue and white Seawitch lure (C&H Lures, Jacksonville, FL) with a 14-g lead head was threaded and positioned above the eye of the hook (Fig. IB). The leader was attached to the standing line with a 36-kg SPRO power swivel (SPRO Corp., Kennesaw, GA). Both hook types were attached to the leader with a loose crimp with the tag end opposite the point. We used Penn “50-wide” reels (Penn, Philadelphia, PA) affixed to “fifty pound class” stand-up rods at all loca- tions. Reels were spooled with 27-kg Diamond® line. The drag upon strike was set at roughly 4.5 kg while the drag during fight (regardless of fish species) was set to roughly 6.4 kg. Baits were dropped back to missed strikes and then only until a fish picked up the bait. Recreational rigging techniques for yellowfin tuna were similar except that 1) circle hooks were the same type and style but one size smaller, 2) J hooks were Mustad 3407, 7/0 size, 3) the fluorocarbon leader was 3.7 m long, and 4) “thirty pound class” stand up rods were used. Charter trips targeting wahoo used Mus- tad 7731A 8/0 J hooks and Eagle Claw 2004 ELG 9/0 circle hooks rigged inside ballyhoo with a 7-g chin weight and pin that comprised part of the wire leader. The leader was 3.7 m of #9 (41 kg) piano wire (Fig. 1C) with hay- wire twists for attaching leader to a hook at one end and for forming a loop at other end; the leader was attached to the standing line with a 59-kg ball bearing clip swivel. As with yellowfin tuna, a blue and white Sea- witch lure with a 14-g lead head was thread- ed through the leader and positioned above the eye of the hook. The same rod and reel types used for yellowfin tuna were used for wahoo. Baits were dropped back to missed strikes and then only until a fish picked up the bait. Recreational rigging techniques for wahoo were similar to those used for charter fishing except that 1) circle hooks were one size smaller; 2) J hooks were Mustad 3407, 7/0 size and 3) “thirty pound class” stand up rods were used. Figure 1 Circle and J hook rigging techniques and leader types used in trolling ballyhoo for (A) dolphinfish ( Coryphaena hippurus ) on monofilament leaders, (B) yellowfin tuna ( Thunnus albacares ) on fluorocarbon leaders, and (C) wahoo ( Acanthocybium solandri) on single-strand wire leaders. The circle hook is the bottom hook type in each of the three photographs. Data collection Circle and J hooks were trolled side-by-side for both the charter and recreational groups. Fishing occurred in Gulf Stream and nearby ocean waters off North Carolina. The two charter boats were employed in order to simu- late a typical for-hire fishing operation in this region. Each of the two captains and mates used for charter trips in this study had over 20 years experience in this fishery, as well as experience rigging and trolling circle hooks for billfishes. Fishing aboard a research vessel was conducted to simulate a recreational operation where fishermen have no mate to coordinate the fishing operation (i.e., to coordinate, rig, and check baits; moni- tor and clear lines; check drags; and hook Rudershausen et al.: A comparison between circle hook and J hook performance in the troll fisheries off North Carolina 159 and gaff fish), but instead do these activities them- selves. Before each charter trip, the captain and first author determined which non-billfish species would be targeted and adjusted the tackle class, leader, and rig type accordingly; the first author made this determina- tion for the recreational trips. This determination was based on water temperature, time of year, and fish- ing reports that indicated which species we would be most likely to interact with. We fished monofilament, fluorocarbon, and wire leaders a total of 6, 12, and 18 days on the charter vessels, and 18, 7, and 14 days on the recreational vessel. There were not equal numbers of days fished between the two user groups. At most, one boat trip was taken per day. On each of the two charter vessels we fished pairs of standing rods (held by fixed rod holders) from four positions. These four pairs were flat lines, short outrig- gers (riggers), long riggers, and bridge poles. On windy days, rods were not fished from the bridge because of increased likelihood of tangles in the lines. On days di- rected for wahoo, we used in-line planers on the flat-line rods in order to fish baits deeper in the water column and elicit a greater number of interactions with this species. We randomly selected which side of the boat (port or starboard) would receive the circle and J hook treatment on each day of charter fishing. We fished two pairs of lines simultaneously aboard the research vessel. These pairs were flat lines and poles fished from rod holders on a canopy “t-top.” For each day of fishing on the recreational vessel, we stag- gered hook types so that a hook type on a flat line was on the side of the boat opposite that same hook type on the t-top. The three vessels used to collect data trolled at be- tween six and seven knots (regardless of species tar- geted). Chains of combined artificial lures consisting of four 23-cm long squids ending with a blue and white Hand Lure® (L&S Bait Company, Inc., Largo, FL)- ballyhoo bait combination were deployed as teasers (no hooks) from each vessel during the collection of data. A chain of pink squids was deployed on the starboard side of the boat and a chain of green squids was deployed on the port side of the boat. Baits were medium ballyhoo that were replaced upon washout. We recorded fish total length (mm) when it was pos- sible to associate a fish length with a hook type. This was not always possible because of multiple fish being caught and placed in fish box at nearly the same time. Hooking location was recorded for all captured fish. Data analysis Four response variables were measured: numbers caught, numbers of strikes, proportion hooked up, and proportion retained. Numbers of fish caught reflected the cumulative results for the strike, hook-up, and reten- tion levels. A fish interacting with the gear in a manner such that the line was pulled from the outrigger clip or that engaged the reel clicker when no clip was used was considered a strike (Prince et al., 2002). A fish that had been hooked for 10 seconds after striking was considered a hook-up. A retained fish was one where the leader was touched or the fish put into the boat (“boated”). The proportion of fish that hooked up was relative to the number that struck (Prince et al., 2002); similarly, the proportion of fish that were retained on the hook was relative to the number that hooked up. Strikes and hook-ups for fish not caught or visually identified were included in the analysis. When the ap- pearance of a struck bait (e.g., bite marks), water tem- perature, time of year, fishing location, fish behavior (jumping), and order of fish landed each day indicated a particular species, we attributed these interactions to that species. When these six factors did not com- bine to indicate a particular species, these interactions were considered to be from an unidentified species. We allocated strikes and hook-ups from unidentified fish to each species in the same proportion as that for fish boated for that day of fishing. At each level of interac- tion we found similar best fitting models for data that excluded or included unidentified fish (Rudershausen et al., 2010). Generalized linear models (GLMs) were used to deter- mine the effects of hook type (circle or J), leader type (monofilament, fluorocarbon, or wire), species (dolphin- fish, yellowfin tuna, or wahoo), user group (recreational or charter), wave height, and potentially important interactions on the numbers of fish caught and each of three mechanisms leading to a caught fish. We con- structed a sequence of Poisson GLMs for the numbers caught and numbers of strikes data sets. For hook-up and retention data, we used binomial GLMs to repre- sent the conditional nature of the hook-up and retention processes (e.g., the number of fish that hooked up in a given trip was conditional on the number that struck). In each case, hook type, leader type, species, and user group were treated as categorical variables, whereas wave height was treated as a continuous variable. Spe- cies caught on days where they were not the main tar- get were included in all analyses and are referred to as “nondirected” species. At each level of interaction, plots were constructed to help better visualize the relative ef- fectiveness of circle and J hooks on directed leader types. We collected the same response variable data for other species that have feeding styles similar to those of yellowfin tuna and wahoo to provide additional data to clarify trends in relative hook-type effectiveness. The four model sets described above were also fitted to data sets that included blackfin tuna (Thunnus at- lanticus), skipjack tuna (Euthynnus pelamis), and false albacore ( Euthynnus alletteratus), which were combined with yellowfin tuna data to form a “tuna” group (fam- ily Scombridae, tribe Thunnini), and king mackerel ( Scomberomorus cavalla) and Spanish mackerel ( Scomb - eromorus maculatus ), which were combined with wahoo data to form a “mackerel” group (family Scombridae, tribe Scomberomorini). This additional model fitting kept dolphinfish as a single-species group. We adopted an information-theoretic perspective to compare the parsimony of relatively simple models 160 Fishery Bulletin 1 10(2) that we believed would help determine relative ef- fectiveness of each hook type at catching fish and on mechanisms during the fish-hook interaction (strike, hook-up, and retention). We inspected data plots to determine factors other than hook that contributed to variability in catch rates. Base models for each level of fish interaction were then constructed without hook main effects and hook interactions. For each of these potential base models we calculated a quasi- Akaike’s information criterion (QAIC; Burnham and Anderson, 2002). QAIC was computed instead of AIC because of potential over-dispersion of the data used as the response variable in each model (Burnham and Anderson, 2002). At each level of fish interaction, we selected the base model with the lowest QAIC value. The most parsimonious base models had 1) main ef- fects (excluding hook) plus a leader-species interaction at the catch level; 2) main effects (excluding hook) plus leader-species and species-user interactions at the strike level; 3) main effects (excluding hook) plus a leader-user interaction at the hook-up level; and 4) main effects (excluding hook) plus a species-user in- teraction at the retention level. After the base model was selected, we developed incrementally more com- plex models that then included a hook effect and in- teraction terms between hook and other factors. This sequential model building allowed us to determine if the main factor of interest — hook type — covaried with other factors potentially influencing interactions with fishes. Any models with three-way interactions also included two-way subinteractions. QAIC( values were then used to compare fits among all i models (includ- ing the base model) at each level of fishing interac- tion to help determine the combination of predictors that best explained variation in the data. The AQAIC value for each model was calculated as the differ- ence between any particular model (QAIC,) and the minimum QAIC for the best fitting model in the set (QAICmin). The model with the QAICmin value was, for each model set, considered to be the one representing the data adequately with the fewest parameters; how- ever, we regarded models that differed by <~ 4 AQAIC as all having reasonable support (Burnham and An- derson, 2002). We also computed Akaike weights (wt) for each model to help gauge the relative support for each model in the model set; the value of wt varies between 0 and 1, with a greater value indicating that a particular model better fits the data. See Burnham and Anderson (2002) for equations used to compute QAIC and wt. Highly parameterized models often resulted in sin- gular Hessian matrices, indicating that one or more parameters were nonidentifiable. However, we retained these models in each model set because our primary goal was to obtain parsimonious predictions of how hook type affected catch rates. In an information-the- oretic context, over-parameterized models would simply be penalized for requiring additional parameters to explain the same amount of variation in the data and therefore would be unlikely to be selected with QAIC. The selection of base models and development of more complex models incorporating a hook main effect and hook interactions by using data on taxa (e.g., dolphin- fish, “tunas,” and “mackerels”) followed the process used for the three species. Base models at each level of fish interaction were the same as in the species analyses described above with the exception of the retention level, where a model consisting of main effects (except hook) plus a leader-species interaction best fitted the taxa data. We computed the relative effectiveness of circle and J hooks (effect size) by comparing predicted circle and J hook catch rates of dolphinfish, yellowfin tuna, and wahoo on their respective directed leader types. Effect size was calculated for each catch model with a positive Akaike weight ( wi ) (see Results section). Effect size (ES) for each of these models was com- puted as ES = ^-, (1) where fix and - the predicted mean catch-per-trip values on circle and J hooks, respectively. Effect size theoretically ranges from zero to greater than one. An effect size less than, equal to, or greater than one indicates that circle hooks are less, equally, or more effective than J hooks, respectively. The variance (a2) about each effect size was calculated as where ox2 and o2 are the variances about the mean predicted mean catch-per-trip values of circle and J hooks. The values for user and wave were held constant (at 0.48 and 0.79 m, respectively) when computing effect size for the three species-leader combinations from each aforementioned catch model. The effect size from each model was weighted by the relative w( value. Weighted effect size values from each model were summed to determine an overall effect size for each of the three species caught on its directed leader type. This model- averaging procedure was repeated to compute overall variance about each average effect size; model averag- ing for variance was conducted by multiplying each model’s variance by the squared value of the Akaike weight (w2). Computations of predicted effect sizes and associated variances were repeated with the data on taxa. For each species, we compared median lengths be- tween hook types with a median ranks test (a=0.05). Data were combined across leader types and user groups for each of these size-based analyses. For each species, we compared rates of jaw (mouth) and deep hooking (gut, gills, or eyes) among hook types using a chi-square square test of independence. Rudershausen et al.: A comparison between circle hook and J hook performance in the troll fisheries off North Carolina 161 Table t Number of fish caught on circle and J hooks from 39 recreational and 36 charter trips trolling both hook types with natural and combination baits offshore of North Carolina, 2006-10. Each number (no.) and percent (%) column is specific to user group (rec- reational vs. charter) and hook type (circle vs. J ). Each column of % values adds up to 100%. Species Recreational Charter Circle J Circle J No. % No. % No. % No. % Dolphinfish (Coryphaena hippurus) 35 63.6 71 77.2 45 40.2 73 38.8 Yellowfin tuna ( Thunnus alhacares ) 7 12.7 5 5.4 25 22.3 47 25.0 Wahoo ( Acanthocybium solandri ) 0 0.0 1 1.1 20 17.9 22 11.7 Blackfin tuna (Thunnus atlanticus ) 0 0.0 0 0.0 14 12.5 26 13.8 King mackerel ( Scomberomorus cavalla ) 8 14.5 3 3.3 0 0.0 4 2.1 Barracuda (Sphyraena barracuda) 1 1.8 0 0.0 1 0.9 3 1.6 Spanish mackerel ( Scomberomorus maculatus) 0 0.0 3 3.3 0 0.0 0 0.0 False albacore ( Euthynnus alletteratus) 2 3.6 6 6.5 4 3.6 5 2.7 Greater amberjack (Seriola dumerili ) 0 0.0 1 1.1 0 0.0 1 0.5 Bluefish (Pomatomus saltatrix) 0 0.0 1 1.1 0 0.0 0 0.0 Atlantic sailfish ( Istiophorus platypterus) 1 1.8 0 0.0 2 1.8 4 2.1 White marlin (Tetrapturus albidus) 0 0.0 1 1.1 0 0.0 0 0.0 Blue marlin (Makaira nigricans) 0 0.0 0 0.0 0 0.0 1 0.5 Skipjack tuna (Euthynnus pelamis) 0 0.0 0 0.0 1 0.9 2 1.1 Bullet mackerel (Auxis spp.) 1 1.8 0 0.0 0 0.0 0 0.0 Results Catch composition The three most abundant species captured on recre- ational trips were dolphinfish, yellowfin tuna, and king mackerel, which together constituted 91% of the catch on circle hooks and 86% on J hooks. The three most abundant species captured on charter trips were dolphinfish, yellowfin tuna, and wahoo, which together constituted 80% of the catch on circle hooks and 76% on J hooks. Blackfin tuna were commonly caught on charter trips, constituting 13% of the catch on circle hooks and 14% of the catch on J hooks. Billfishes made up 1% of the catch on recreational trips and 3% of the catch on charter trips (Table 1). Pooling across both user groups, we found that 74% of dolphinfish were caught on monofilament leaders, 96% of yellowfin tuna were caught on fluorocarbon leaders, and 98% of wahoo were caught on wire leaders; that is, the vast majority of individuals from each species were cap- tured on the respective directed leader type. Species identity could not be determined in 14.0% of strike and 2.9% of hook-up interactions over the course of the study. Comparisons of catch and examination of mechanisms influencing catch Hook type influenced catch rate (Fig. 2). For the three- species analysis of catch rate, the base model plus a hook main effect received majority support (Table 2). For directed leaders, J hooks caught more dolphinfish than circle hooks for both recreational and charter groups. Higher catches on J hooks were also observed in the charter group for yellowfin tuna; however, there was no clear hook effect within the recreational group for yellowfin tuna or wahoo or charter group for wahoo. Partial support for models containing hook-user and hook-species interactions confirms these observations (Table 2). The hook-leader interaction also had support and was most obvious in the dolphinfish data where the hook effect was not consistent across leader types (Fig. 2). Model fitting to numbers-caught data with taxa (i.e., dolphinfish, tunas, and mackerels) provided similar results to those for species data (Table 2; Fig. 3); the base model plus a hook main effect received majority support as the best fitting model and models that included hook-user, hook-leader, and hook-species interactions had QAIC values within four units of the best fitting model. Tunas were caught more often on J hooks and fluorocarbon leaders than other hook-leader combinations. Mackerels were caught slightly more often on J hooks than circle hooks, and most often on wire leaders (Fig. 3). The first mechanism contributing to catch was strike. Hook type had little effect on strikes for each of the three species examined (Fig. 4). No single model re- ceived majority support when fitted to strike data for the three species and the base model with a hook factor received only slightly greater support than the base model without the hook parameter (Table 3). Models 162 Fishery Bulletin 1 10(2) Recreational Charter Circle J Circle J Hook type Figure 2 Plots of the average catch per trip (±standard error) on circle hooks (open bars) and J hooks (gray bars). Data for each species are from both directed and nondirected trips for that species. Plots are broken down by user group (recreational [left column, panels A-C] and charter [right column, panels D-F]) and species (dolphinfish [Coryphaena hippurus] [A, D], yellowfin tuna [Thunnus albacares] [B, E], and wahoo [ Acanthocybium solandri] [C, F] ). The legend denoting fill pattern for each leader type applies to all panels. No bar for a particular hook-type + species+user-group+leader-type combination indicates no catch. with hook-user, hook-leader, and hook-species interac- tions each received a relatively small amount of sup- port. Greater numbers of strikes occurred 1) on charter boats (owing to a greater number of rods fished), 2) when using monofilament leaders, and 3) from dolphin- fish than any other species. As with the three species data, there was little difference in the average strikes per trip between circle and J hooks for each taxa (Fig. 5). Similarly, the model that best fitted strike data for the taxa was the base model with hook, but the base model without hook received only slightly less support (Table 3; Fig. 5). Models with hook-user, hook-species, and hook-leader interactions received relatively minor support. The second mechanism contributing to catch was hook-up. J hooks were more effective at hooking fish for many user group-species combinations (Fig. 6). Hook was a main effect in the model that best fit- ted the proportional hook-up data (Table 4). Models that received less support included hook-user, hook- species, hook-leader, and hook-species + hook-user interactions. The base model received no support. There was a reduction of hook-ups for dolphinfish when circle hooks were used on both recreational and Rudershausen et al A comparison between circle hook and J hook performance in the troll fisheries off North Carolina 163 Table 2 Candidate models fitted to catch-per-trip data for three species (dolphinfish | Coryphaena hippurus 1, yellowfin tuna I Thunnus albacares], and wahoo \Acanthocybium solandri], and taxa (dolphinfish, tunas, and mackerels]) when trolling circle and J hooks in Gulf Stream waters off North Carolina. Quasi-Akaike information criterion (QAIC) was used to evaluate model performance, with the lowest value indicating the most parsimonious model. Categorical predictor variables included hook type (hook), leader type (leader), species or taxa, and user group (user). Wave height was used as a continuous predictor variable. K= number of param- eters for each model; ic=Akaike weight. Base models included all predictor variables with exception of hook and any hook interac- tions; see Methods section for a full description of base models. 4QAIC values ~<4 were considered models with reasonable support. Interaction Data type Distribution Model K QAIC AQAIC w Catch: species Count Poisson base + hook 13 356.77 0.00 0.54 base + hook + hook*user 14 358.42 1.65 0.23 base + hook + hook*leader 15 360.49 3.72 0.08 base + hook + hook*species 15 360.90 4.14 0.07 base + hook + hook*user + hook*leader 16 361.98 5.21 0.04 base + hook + hook*species + hook*user 16 362.54 5.77 0.03 base + hook + hook*species + hook*leader 17 364.91 8.15 0.01 base + hook + hook*species + hook*leader 21 373.59 16.83 0.00 + hook*species*leader base 12 385.14 28.37 0.00 Catch: taxa Count Poisson base + hook 13 477.17 0.00 0.55 base + hook + hook*user 14 479.11 1.94 0.21 base + hook + hook*leader 15 480.63 3.46 0.10 base + hook + hook*taxa 15 481.14 3.97 0.07 base + hook + hook*user + hook*leader 16 482.54 5.37 0.04 base + hook + hook*taxa + hook*user 16 483.16 5.99 0.03 base + hook + hook*taxa + hook*leader 17 485.11 7.94 0.01 base + hook + hook*taxa + hook*leader 21 493.60 16.43 0.00 + hook*taxa*leader base 12 501.35 24.18 0.00 charter trips. This trend was most pronounced on charter trips for all leader types (Fig. 6). The excep- tion was a slightly greater hook-up rate for yellowfin tuna on circle hooks than on J hooks when fishing fluorocarbon leaders on recreational trips. For the taxa analysis, trends in model fitting to proportional hook-up data were similar to three species (Table 4; Fig. 7); hook was a main effect in the best fit- ting model and it was a main effect and interaction term in models receiving lesser support. The base model received no support (Table 4). The addition of mackerel data on recreational trips strengthened the trend of greater effectiveness of J hooks in hooking up these taxa on wire, the directed leader type for that group (Fig. 7). The third mechanism contributing to catch was retention. Hook type did not appear to have a pro- nounced effect on proportion of fish retained (Fig. 8). For models fitted to species data, the base model re- ceived majority support (Table 5). A base model with a hook effect was the only other model receiving support, but it was minor. The proportion retained on circle hooks generally equaled (dolphinfish and yellowfin tuna) or slightly exceeded (wahoo) those retained on J hooks on directed leader types (Fig. 8). Proportional retention data for the taxa also showed that retention was high, with little to no difference between hook types (Table 5; Fig. 9). The base model received major- ity support and the base model with hook as a main effect received less support. Two other models that received minor support had hook-species and hook-user interactions (Table 5). Estimates of effect size on catch rates determined from model-averaged predictions showed that J hooks were more effective than circle hooks. This trend held across the species and taxa levels. For the three spe- cies, mean predicted effect size (± standard deviation [ SD] ) for dolphinfish, yellowfin tuna, and wahoo on directed leader types was 0.60 (0.05), 0.60 (0.07), and 0.65 (0.09), respectively (Fig. 10), meaning that circle hooks were roughly 60% as effective as J hooks. For the taxa groups, mean predicted effect size (±SD) for dolphinfish, tunas, and mackerels was 0.62 (0.05), 0.62 (0.06), and 0.67 (0.08), respectively (Fig. 10). There were no significant between-hook differences in the distribution of lengths for dolphinfish (^2 = 0.973, P=0.324), yellowfin tuna (^2=0.003, P=0.958), or wahoo (^2 = 0.068, P=0.795). Thus, hook type was not size selec- tive within a species. The effect of hook type on deep hooking was species dependent. Rates of deep hooking were significantly less for dolphinfish caught on circle hooks than J hooks (Table 6). However, there was no effect of hook type on proportion of deep-hooked wahoo or blackfin tuna. Rates 164 Fishery Bulletin 110(2) Recreational Charter Hook type Figure 3 Plots of the average catch per trip (±standard error) on circle hooks (open bars) and J hooks (gray bars). Data for each group are from both directed and nondirected trips for that species. Plots are broken down by user group (recreational [left column, panels A-C], and charter [right column, panels D-F[), and taxa (dolphinfish Coryphaena hip- purus [A, D[, tunas [B, E], and mackerels [C, Fj). The tuna group included yellowfin tuna ( Thunnus albacares ), blackfin tuna (Thunnus atlanticus), skipjack tuna ( Euthyn - mis pelamis ), and false albacore ( Euthynnus alletteratus). The mackerel group included wahoo (Acanthocybium solandri), king mackerel ( Scomberomorus cavalla ), and Spanish mackerel (Scomberomorus maculatus). The legend denoting fill pattern for each leader type applies to all panels. No bar for a particular hook-type+taxon+user-group + leader- type combination indicates no catch. of deep hooking were 0% for both circle and J hooks that caught yellowfin tuna. Discussion There is increased interest in requiring circle hooks in the recreational bluewater troll fishery in the United States. This is largely due to studies finding that circle hooks maintain catch rates but reduce rates of deep hooking compared with J hooks in billfishes (see Serafy et ah, 2009, for review). In contrast, we found for non- billfishes that observed catch rates were reduced with circle hooks under that for J hooks in the charter group; similar findings were found in the recreational group for dolphinfish. Predictions of relative catch (through effect Rudershausen et al A comparison between circle hook and J hook performance in the troll fisheries off North Carolina 165 Recreational Charter Hook type Figure 4 Plots of the average number of strikes per trip (±standard error) on circle hooks (open bars) and J hooks (gray bars). Data for plots includes strikes from unidentified fish later apportioned to species that could be identified. Data for each species are from both directed and nondirected trips for that species. Plots are broken down by user group (recreational lleft column, panels A-C] and charter I right column, panels D-F]) and species (dolphinfish [ Coryphaena hippurus] |A, D], yellowfin tuna [ Thunnus albacares] [B, E 1 , and wahoo ( Acanthocybium solandri ] [C, F]). The legend denoting fill pattern for each leader type applies to all panels. No bar for a particular hook-type + species+user- group + leader-type combination indicates no catch. size calculations) indicate that fishermen can expect 65% greater catches of the three species or taxa groups on J hooks than on circle hooks. The similar findings between the species and taxa analyses indicate that morphologi- cal features of fish, attack styles, and hook effectiveness are consistent among the species of the tuna group and among the species of the mackerel group. Additionally, the similar results when smaller tunas and mackerels were included in the taxa analysis indicate that the inef- fectiveness of circle hooks compared with J hooks is not size dependent within the range of fish sizes in our study. The similarities between our findings and prior hook comparisons of hooks on longlines depend on the species being considered. In a Brazilian longline fishery, Sales et al. (2010) found a similar trend in dolphinfish catches to that found in our study (lower catches on circle hooks than on J hooks) but significantly more tunas caught on circle hooks than on J hooks. The increased catch 166 Fishery Bulletin 1 10(2) Table 3 Candidate models fitted to strike data for three species (dolphinfish \Coryphaena hippurus], yellowfin tuna [ Thunnus albacares], and wahoo 1 Acanthocybium solandri]), and taxa (dolphinfish, tunas, and mackerels) when trolling circle and J hooks in Gulf Stream waters off North Carolina. Quasi-Akaike information criterion (QAIC) was used to evaluate model performance, with the lowest value indicating the most parsimonious model. Categorical predictor variables included hook type (hook), leader type ( leader), species or taxa, and user group ( user). Wave height was used as a continuous predictor variable. /f=number of parameters for each model; w=Akaike weight. Base models included all predictor variables with exception of hook and any hook interactions; see Methods section for a full description of base models. AQAIC values ~<4 were considered models with reasonable support. Interaction Data type Distribution Model K QAIC hQAIC w Strike: Count Poisson base + hook 15 979.96 0.00 0.36 species base 14 980.17 0.21 0.33 base + hook + hook*user 16 981.89 1.93 0.14 base + hook + hook*leader 17 983.42 3.46 0.06 base + hook + hook*species 17 983.81 3.86 0.05 base + hook + hook*user + hook*leader 18 985.40 5.44 0.02 base + hook + hook*species + hook*user 18 985.75 5.79 0.02 base + hook + hook*species + hook*leader 19 987.85 7.89 0.01 base + hook + hook*species + hook*leader 23 996.39 16.43 0.00 + hook*species*leader B. Strike: Count Poisson base + hook 15 1050.57 0.00 0.40 taxa base 14 1051.09 0.52 0.31 base + hook + hook*user 16 1052.66 2.08 0.14 base + hook + hook*taxa 17 1054.54 3.97 0.05 base + hook + hook*leader 17 1054.56 3.99 0.05 base + hook + hook*taxa + hook*user 18 1056.59 6.01 0.02 base + hook + hook*user + hook*leader 18 1056.64 6.07 0.02 base + hook + hook*taxa + hook*leader 19 1058.92 8.34 0.01 base + hook + hook*taxa + hook*leader 23 1067.40 16.83 0.00 + hook*taxa*leader rate of tunas on circle hooks over that for J hooks has been observed in other longline studies (Falterman and Graves, 2002). It is unclear what the mechanism is that leads to higher tuna catches on longline circle hooks, but lower tuna catches on trolled dead baits rigged with circle hooks; it is likely that tuna ingested the bait and hook more deeply in comparison to the actively trolled bait in our study. Actively trolling hooks (versus passive fishing on a longline) may be the mechanism contribut- ing to these hook-type differences. Most comparative studies of hooks in the dead bait troll fishery have been designed to estimate catch- and-release mortality in billfishes (Prince et al., 2002; Horodysky and Graves, 2005; Graves and Horodysky, 2010). The species that we examined in this study are not generally released; therefore, our focus was on the influence of hook type on catch rates and the potential mechanisms responsible for similarities or differences in catch by hook type, rather than on postrelease mor- tality. This was our focus because many charter boat captains suspect that circle hooks negatively impact catches of dolphinfish, tunas, and mackerels in the North Carolina dead-bait troll fishery. Our results con- firm this suspicion. Model-averaged estimates suggest a strong negative effect of hook type on catch rates for all three species; however, examination of the raw data for individual species suggests that the effect of hook type on wahoo catch may be minor. Future studies with increased sample sizes would help to refine estimates of species by hook-type interactions, providing greater resolution of the importance and magnitude of hook ef- fects for individual species. Thus, this is the first study to find that catch rates in a dead bait troll fishery can be negatively impacted by circle hooks. Horodysky and Graves (2005) and Graves and Horodysky (2010) did not provide comparisons of catch data between circle hooks and J hooks in their hook comparative studies on billfish. Differences in strike, hook-up, and retention rates be- tween hook types all have the potential to contribute to differences in catch rates. There was little evidence for a hook effect on strike rate; therefore, J and circle hook rigged baits were equally attractive to these three fish groups. Other studies that have compared hook types in the dead bait troll fishery have not reported data on strike rate by hook type; we recommend that this in- formation be collected so that the specific mechanisms responsible for potential differences in catch rate can be determined. The greater effectiveness of J hooks at hooking fish once they struck generally held across the three species and dolphinfish and the two taxa groups. Circle hooks Rudershausen et al A comparison between circle hook and J hook performance in the troll fisheries off North Carolina 167 Recreational Charter Hook type Figure 5 Plots of the average number of strikes per trip (±standard error) on circle hooks (open bars) and J hooks (gray bars). Data for plots include strikes from unidentified fish later apportioned to species that could be identified. Data for each group are from both directed and nondirected trips for that species. Plots are broken down by user group (recreational [left column, panels A-C] and charter [right column, panels D-F]) and taxa (dolphinfish [ Coryphaena hippurus] [A, D[, tunas [B, E], and mackerels [C, F] ) . The tuna group included yellowfin tuna ( Thunnus albacares), blackfin tuna (Thunnus atlanticus), skipjack tuna (Euthynnus pelamis), and false albacore (Euthynnus allet- teratus). The mackerel group included wahoo ( Acanthocybium solandri), king mackerel ( Scomberomorus caualla), and Spanish mackerel (Scomberomorus maculatus ). The legend denoting fill pattern for each leader type applies to ail panels. No bar for a particular hook-type + taxon+user-group + leader-type combination indicates no catch. are designed to hook fish if the hook rounds a corner within the jaw area. In theory, this would be most com- mon for fish that turn their mouth opening away from the direction of the fishing line. However, if a fish is not seen during a strike, if is difficult to know when to reel the line tight (i.e., when the fish has turned). Our workshop panel (see Methods section) argued for a drop back for dolphinfish because this species is known to swim with the bait in their mouth in the direction that the line is trolled. The drop back for dolphinfish was done to allow enough time for the dolphinfish to turn. Even with these efforts, hook-up rates of dolphin- fish were lower with circle hooks than J hooks for both user groups. Prince et al. (2002) found that hook-up 168 Fishery Bulletin 1 10(2) Recreational Charter D Dolphinlish ( Coryphaena hippurus) Yellowfin tuna ( Thunnus albacares) F Wahoo {Acanthocybium solandri) iil Circle Hook type Figure 6 Plots of the average proportion of fish that hooked up (±standard error) on circle hooks (open bars) and J hooks (gray bars). Data for plots includes hook-ups from unidentified fish later apportioned to species that could be identified. Data for each species are from both directed and nondirected trips for that species. Plots are broken down by user group (recreational [left column, panels A-C] and charter [right column, panels E-F]) and species (dolphinfish [Coryphaena hippurus ] [A, D], yellowfin tuna [Thunnus albacares ] [B, E], and wahoo [ Acanthocybium solandri] [C, F]). The legend denoting fill pattern for each leader type applies to all panels. No bar for a particular hook-type + species+user- group+leader-type combination indicates no catch. rate on circle hooks was significantly higher than J hooks in a dead bait troll fishery for sailfish. The abil- ity for the angler to visually see the fish with the bait in its mouth may allow for higher hook-ups on circle hooks in that fishery. In contrast, fishing for yellowfin tuna and wahoo involved using a heavy drag because the fish are aggressive and generally hook themselves upon striking (see Graves and Horodysky [2010] for a similar description and approach when targeting blue marlin). Theoretically, the circle hook should work in this heavy-drag situation only if the fish’s mouth is at an angle to the direction of the line when the bait is taken into the mouth. Hook-up rates for yellowfin tuna and wahoo were slightly higher on J hooks on charter trips (for which we had the most data); this finding may be a result of some strikes on circle hooks where the mouth opening faced the direction that the bait was be- ing trolled or because of bait rigging (see below). Graves Rudershausen et at: A comparison between circle hook and J hook performance in the troll fisheries off North Carolina 169 Table 4 Candidate models fitted to hook-up data for three species (dolphinfish [ Coryphaena hippurus], yellowfin tuna [ Thunnus alba- cares I, and wahoo ( Acanthocybium solandri]), and taxa (dolphinfish, tunas, and mackerels) when trolling circle and J hooks in Gulf Stream waters off North Carolina. Quasi-Akaike information criterion (QAIC) was used to evaluate model performance, with the lowest value indicating the most parsimonious model. Categorical predictor variables included hook type (hook), leader type (leader), species or taxa, and user group (user). Wave height was used as a continuous predictor variable. /C=number of parameters for each model; w=Akaike weight. Base models included all predictor variables with exception of hook and any hook interactions; see Methods section for a full description of base models. AQAIC values ~<4 were considered models with reasonable support. Interaction Data type Distribution Model K QAIC AQAIC U) Hook-up: Proportion Binomial base + hook 11 -1159.03 0.00 0.38 species base + hook + hook*user 12 -1158.16 0.88 0.25 base + hook + hook*species 13 -1156.75 2.29 0.12 base + hook + hook*leader 13 -1156.53 2.51 0.11 base + hook + hook*species + hook*user 14 -1155.07 3.96 0.05 base + hook + hook*user + hook*leader 14 -1154.81 4.23 0.05 base + hook + hciok*species + hook*leader 15 -1154.75 4.28 0.04 base + hook + hook*species + hook*leader + species*leader + hook*species*leader 23 -1148.53 10.51 0.00 base 10 -1134.22 24.81 0.00 Hook-up: Proportion Binomial base + hook 11 -1393.91 0.00 0.40 taxa base + hook + hook*taxa 13 -1392.58 1.33 0.21 base + hook + hook*user 12 -1392.17 1.74 0.17 base + hook + hook*leader 13 -1390.90 3.00 0.09 base + hook + hook*taxa + hook*user 14 -1390.48 3.43 0.07 base + hook + hook*user + hook*leader 14 -1388.80 5.11 0.03 base + hook + hook*taxa + hook*leader 15 -1388.72 5.19 0.03 base + hook + hook*taxa + hook*leader + taxa*leader + hook*taxa*leader 23 -1385.27 8.64 0.01 base 10 -1368.17 25.74 0.00 and Horodysky (2010) did not report hook-up percentage data for blue marlin and therefore it is unknown what hook-up rates would be for this aggressive feeder that is hooked upon strike. One rigging tactic when trolling is to rig the circle hook so that it is completely external to the bony or fleshy portions of the bait to maximize the exposed gap width (e.g., the hook is placed on top of the bait’s head; Prince et ah, 2002). This placement is thought to work best for “dropping back” to fish because the fish have enough time to swallow the bait and the hook (dolphinfish and billfish trolling) and turn their body, while the exposed gap width of the circle hook is maximized. We did not employ the external rig- ging technique on days when yellowfin tuna or wahoo were targeted. Hooks were rigged internally for these two species because these species hook themselves upon striking; drop-backs are not typically required by charter or recreational fishers targeting these species. An additional reason for embedding hooks in baits was so that we could fish “combo” baits (lure and natural bait combinations) because colored lures (skirts) elicit more strikes than plain ballyhoo on most days for yellowfin tuna and wahoo. The cooperating mates on charter trips embedded the hook as close to the tail as possible without compromising the swimming ac- tion of the bait. Using larger circle hooks would have increased the gap width between the point and the point shank, potentially making hook-ups more likely, but this change could have compromised the strike rate by making the hook more visible to the fish or causing the bait to wash out faster. There was little to no hook effect at the propor- tional retention level (caught once hooked) for dol- phinfish, yellowfin tuna, and tunas, although there was increased retention of wahoo and mackerels on circle hooks and yellowfin tuna on circle hooks in the recreational fishery. The latter result is consistent with the findings of Prince et al. (2002) when trolling dead baits with circle and J hooks for sailfish. The increased retention on circle hooks relative to J hooks has been used as a selling point for circle hooks, but we did not find this result for the majority of species that we caught. The procedure for assigning interactions with un- identified fish to a particular species is not ideal. For instance, if individuals of one species generate behav- ioral cues or are landed more readily than individuals for another species, species assignments may be biased toward more readily identified fish. In general, this approach decreased our ability to detect species effects on landing probabilities and hookup rates. However, we expected the reduction in statistical power to be relatively small and to affect only inferences about 170 Fishery Bulletin 1 10(2) Recreational Charter Circle D Dolphinfish ( Corvphaena hippurus) E T unas Mackerels Circle J. . m Hook type Figure 7 Plots of the average proportion of fish that hooked up (±standard error) on circle hooks (open bars) and J hooks (gray bars). Data for plots includes hook-ups from unidentified fish later apportioned to species that could be identified. Data for each group are from both directed and nondirected trips for that species. Plots are broken down by user group (recreational 1 left column, panels A-C] and charter [right column, panels E-F]) and taxa (dolphinfish [Coryphaena hippurus] [A, D], tunas [B, E], and mackerels [C, F] ). The tuna group included yellowfin tuna ( Thunnus albacares ), blackfin tuna ( Thunnus atianticus), skipjack tuna (Euthynnus pelamis), and false albacore (Euthynnus aliet- teratus). The mackerel group included wahoo (Acanthocybium solandri), king mackerel (Scomberomorus cavalla), and Spanish mackerel ( Scomberomorus maculatus). The legend denoting fill pattern for each leader type applies to all panels. No bar for a particular hook-type+taxon+user-group+leader-type combination indicates no catch. species-hook interactions; main effects for hook type remained unbiased. If fishermen are interested in releasing dolphinfish, our results provide evidence that released fish are not hooked as deeply and thus have a higher likelihood of survival if circle hooks are used. The drop-back tech- nique that we commonly used for dolphinfish likely led to a higher percentage of dolphinfish becoming deep hooked with J hooks over that for the tuna and mackerel taxa groups. The reduction in gut hooking with circle hooks has been found in most other stud- ies comparing circle and J hooks (Cooke and Suski, 2004). Managers should factor in the high rate of deep hooking for J-hooked dolphinfish as they imple- Rudershausen et al A comparison between circle hook and J hook performance in the troll fisheries off North Carolina 171 Recreational Charter Circle J Circle J Hook type Figure 8 Plots of the average proportion of fish retained (±standard error) on circle hooks (open bars) and J hooks (gray bars). Data for each species are from both directed and non- directed trips for that species. Plots are broken down by user group (recreational lleft column, panels A-C] and charter [right column, panels D-F]) and species (dolphinfish Coryphaena hippurus [A, D], yellowfin tuna \Thunnus albacares ] [B, E], and wahoo \Acanthocybium solandri ] [C, F]). The legend denoting fill pattern for each leader type applies to all panels. No bar for a particular hook-type+species + user-group + leader-type combination indicates no catch. ment new minimum size regulations for this species in the U.S. South Atlantic (SAFMC, 2011). However, managers should also consider that there can be a trade-off when using circle hooks. Although rates of deep hooking are relatively low on circle hooks, handling time and air exposure are increased while dislodging them from captured fish owing to their in- herently deeper bend than J hooks (Cooke and Suski, 2004; senior author, personal observ.). Along with outreach efforts to encourage the use of circle hooks where appropriate, instructions should be available on how to quickly remove the hooks with little injury to the fish. Circle hooks remain vaguely defined. The federal definition of a circle hook (Federal Register, 2006) is somewhat arbitrary. Numerous circle hooks may meet the federal specifications, yet may not simultaneously reduce deep hooking in billfishes and maintain catch rates of non-billfishes. For example, some manufac- turers advertise circle hooks with parallel or nearly parallel point shanks and hook shanks (like a J hook), but which simply have the tip of the point bent 90° 172 Fishery Bulletin 1 10(2) Table 5 Candidate models fitted to retention data for three species (dolphinfish [ Coryphaena hippurus j, yellowfin tuna [Thunnus alba- cares], and wahoo \Acanthocybium solandri]), and taxa (dolphinfish, tunas, and mackerels) when trolling circle and J hooks in Gulf Stream waters off North Carolina. Quasi-Akaike information criterion (QAIC) was used to evaluate model performance, with the lowest value indicating the most parsimonious model. Categorical predictor variables included hook type (hook), leader type ( leader), species or taxa, and user group ( user). Wave height was used as a continuous predictor variable. K= number of parameters for each model; m=Akaike weight. Base models included all predictor variables with exception of hook and any hook interactions; see Methods section for a full description of base models. AQAIC values ~<4 were considered models with reasonable support. Interaction Datatype Distribution Model K QAIC AQAIC w Retention: Proportion Binomial base 10 -876.22 0.00 0.63 species base + hook 11 -874.19 2.02 0.23 base + hook + hook*leader 13 -871.13 5.09 0.05 base + hook + hook*species 13 -870.76 5.46 0.04 base + hook + hook*species + hook*user 14 -869.35 6.87 0.02 base + hook + hook*user + hook*leader 14 -869.16 7.05 0.02 base + hook + hook*species + hook*leader 15 -867.10 9.12 0.01 base + hook + hook*user 12 -857.77 18.45 0.00 base + hook + hook*species + hook*leader 23 -854.71 21.51 0.00 + species + leader hook*species*leader Retention: Proportion Binomial base 12 -1112.66 0.00 0.53 Taxa base + hook 13 -1110.55 2.11 0.19 base + hook + hook*species 15 -1108.74 3.92 0.08 base + hook + hook*user 14 -1108.74 3.93 0.08 base + hook + hook*leader 15 -1108.54 4.12 0.07 base + hook + hook*species + hook*user 16 -1106.71 5.95 0.03 base + hook + hook*user + hook*leader 16 -1106.40 6.26 0.02 base + hook + hook*species + hook*leader 17 -1104.87 7.79 0.01 base + hook + hook*species + hook*leader 23 -1092.10 20.56 0.00 + hook*species*leader Table 6 Percentage of fish caught in two anatomical locations (jaw vs. “deep” [body, gill, gut, eye]) with trolled circle and J hooks. The x2 test statistic and P-value from each test of independence comparing hooking locations between hook types are presented for each species. A x2 test was not conducted for king mackerel because of small sample size. Circle hook J hook Species Jaw Deep Jaw Deep X2 P Dolphinfish ( Coryphaena hippurus) 98.5 1.5 61.3 38.7 31.35 <0.001 Yellowfin tuna ( Thunnus albacares) 100 0 100 0 — — Wahoo ( Acanthocybium solandri) 100 0 91.3 8.7 1.82 0.177 Blackfin tuna (Thunnus atlanticus) 100 0 92.6 7.4 1.13 0.287 King mackerel (Scomberomorus cavalla ) 100 0 66.7 33.3 toward the shank. Having discussed the structure of the hooks with captains, Smith (2006) postulated that a greater turn in the point shank (a point shank that turns back towards the hook shank by >33°) re- duces the chances for deep hooking in billfishes. This outcome has yet to be determined with experimental fishing and would be a useful area of future research. We measured the angle between the point shank and hook shank to be roughly 25 degrees for the circle hooks we used (regardless of the size). Compared with the circle hook styles we tested, other circle hooks with different point shank angles that still satisfy federal requirements may have performed better at catching non-billfish species. The fishing tackle industry and charter boat opera- tors continually adapt gear and techniques to increase Rudershausen et al A comparison between circle hook and J hook performance in the troll fisheries off North Carolina 173 T3 CD C 03 0) CD o3 $ SZ U) o o •c o Q. O Q. < 1.0 0.8 0.6 0.4 0.2 0.0 1.0 0.8 0.6 0.4 0.2 0.0 1.0 0.8 0.6 0.4 0.2 0.0 1.0 0.8 0.6 0.4 0.2 0.0 1.0 0.8 0.6 0.4 0.2 0.0 1.0 0.8 0.6 0.4 0.2 0.0 Circle J Circle J Hook type Recreational Figure 9 Plots of the average proportion of fish retained (±standard error) on circle hooks (open bars) and J hooks (gray bars). Data for each group are from both directed and nondirected trips for that species. Plots are broken down by user group (recreational [left column, panels A-C] and charter [right column, panels D-F[) and taxa (dolphinfish Coryphaena hippurus [A, D], tunas [B, E], and mackerels [C, FI). The tuna group included yellowfin tuna ( Thunnus albacares ), blackfin tuna ( Thunnus atlanticus ), skipjack tuna ( Euthyn - nus pelamis), and false albacore ( Euthynnus alletteratus). The mackerel group included wahoo (Acanthocybium solandri), king mackerel ( Scomberomorus cavalla), and Spanish mackerel ( Scomberomorus maculatus). The legend denoting fill pattern for each leader type applies to all panels. No bar for a particular hook-type+taxon+user-group+leader- type combination indicates no catch. catch efficiency. There are likely untested techniques that allow fishermen to catch non-billfish with circle hooks more efficiently than we found in this study. Cooke and Suski (2004) report that the choice of circle hook size is an important consideration in order to maximize their effectiveness. Hook size seems to be an especially important consideration in a mixed-species and mixed-size fishery such as the one we examined. Hook choice (size and style) was a central topic in the workshop we convened; in targeting each of the main species (dolphinfish, yellowfin tuna, and wahoo), we selected hook sizes and styles recommended by expe- rienced offshore fishermen. It is likely that fishermen would be more inclined to experiment with circle hooks and novel rigging strategies if they knew there would be a pending re- 174 Fishery Bulletin 1 10(2) Species group Figure 10 Mean predicted effect size (±standard deviation) of circle versus J hooks on catch rates by species or taxa group. Dolphinfish ( Coryphaena hip- purus) is listed twice because the predicted effect size changes slightly in comparisons with the “tunas” and “mackerels” taxa groups. An effect size greater than 1 indicates greater effectiveness of circle hooks than J hooks; the opposite is true for an effect size less than 1. An effect size equal to 1 (dashed line) indicates that the hook types are equally effective. The mean and variance of each effect size was calculated by using weighted model averages from each model with positive Akaike weight (w{) at the catch level (see Materials section for details). quirement to use them outside of Atlantic f billfish tournaments. Industry willing- 1.2- ness to refine rigging techniques and fishing strategies in the face of future hook-type regulations could help increase experimentation with circle hooks, and thus catch rates of non-billfish species when trolling for them in this fishery. The workshop we convened generated many novel rigging and fishing tech- niques with circle hooks, only a fraction of which we used for the field experiment of this project. We urge future studies to provide catch rates (numbers standardized to effort), strike, hook-up, and retention data for both hook types so that trade-offs be- tween catch-and-release survival and catch rates can be evaluated. In addition, the terms used when discussing these variables should also be standardized. For example, the catch rate for trolled baits as defined by Serafy et al. (2009) equals a retention rate (caught if hooked), but a fisherman’s interest lies in know- ing how many fish will be caught per trip which is the product of number of strikes, proportion hooked, and propor- tion retained. Without knowledge of the first two variables, the third variable only provides information about a hook’s effectiveness at retaining a fish on the line and not its overall effectiveness. Conclusions We examined three mechanisms that may have been responsible for the hook effect on catch rates. These were strike, hook-up, and retention. There was little to no hook effect at the strike and retention levels. However, the differences in catch rates we observed resulted from a lower hook-up rate on circle hooks compared with J hooks. This trend was generally consistent across analy- ses of data on three species and on three broader taxa. It is unknown whether a requirement to troll exclu- sively circle hooks in the offshore fishery would have an economic impact on either the recreational or charter fisheries in this region. It is likely that circle hooks need to catch fish at rates near, equal to, or higher than J hooks to gain wider acceptance among offshore troll fishermen (Jordan, 1999). We hope that angler experimentation will lead to improvements in circle hook catch rates for non-billfish species caught during trolling operations. Acknowledgments This study was funded by North Carolina Sea Grant Fishery Resource Grant awards 08-FEG-02 and 10-FEG- 06, and North Carolina Sea Grant awards E/GS-6, FEE-1, and the Big Rock Blue Marlin Tournament. We thank D. Kerstetter and two anonymous reviewers for their constructive comments on the manuscript. Literature cited Burnham, K. R, and D .R. Anderson. 2002. Model selection and inference; a practical infor- mation-theoretic approach, 2nd ed., 488 p. Springer- Verlag, New York. Cooke, S. J., and C. D. Suski. 2004. Are circle hooks effective tools for conserving fresh- water and marine recreational catch and release fisher- ies? Aquat. Conserv. Mar. Freshw. Ecosyst. 14:299-326. Dumas, C. F., J. C. Whitehead, C. E. Landry, and J. H. Herstine. 2009. Economic impacts and recreation value of the North Carolina for-hire fishing fleet. North Carolina Sea Grant Fishery Resource Grant (FRG) Report FRG # 07-FEG-05, 169 p. North Carolina Sea Grant Fish- ery Resource Program, Raleigh, NC. [Available from http://www.ncseagrant.org/home/research/core-research/ research-item?view=item& id = 956.1 Falterman, B., and J. E. Graves. 2002. A comparison of the relative mortality and hook- ing efficiency of circle and straight shank (“J”) hooks used in the pelagic longline industry. Am. Fish. Soc. Symp. 30:80-87. Federal Register. 2006. Atlantic highly migratory species: recreational Rudershausen et al A comparison between circle hook and J hook performance in the troll fisheries off North Carolina 175 Atlantic blue and white marlin landings limit: amend- ments to the Fishery Management Plan for Atlantic Billfish. Vol. 71(190):58102. Government Printing Office, Washington, DC. Graves, J. E., and A. Z. Horodysky. 2010. Asymmetric conservation benefits of circle hooks in multispecies billfish recreational fisheries: a synthesis of hook performance and analysis of blue marlin post- release survival. Fish. Bull. 108:433-441. Horodysky, A. Z., and J. E. Graves. 2005. Application of pop-up satellite archival tag tech- nology to estimate postrelease survival of white marlin ( Tetrapturus albidus) caught on circle and straight-shank (“J”) hooks in the western North Atlantic recreational fishery. Fish. Bull. 103:84-96. Jordan, J. 1999. Going full circle. Big Game Fish. J. 12(31:52-62. NCDMF (North Carolina Division of Marine Fisheries). 2010. North Carolina marine recreational finfish har- vest statistics. [Available at http://www,ncfisheries. net/statistics/recstat/i ndex.html. accessed May 2011.] Prince, E. D., M. Ortiz, and A. Venizelos. 2002. A comparison of circle hook and “j” hook performance in recreational catch-and-release fisheries for billfish. In Catch and release in marine recreational fisheries (J. A. Lucy and A. L. Studholme, eds.l, p. 66-79. Am. Fish. Soc. Symp. 30. Am. Fish. Soc., Bethesda, MD. Prince, E. D., D. Snodgrass, E. S. Orbesen, J. P. Hoolihan, J. E. Serafy, and J. E. Schratwieser, 2007. Circle hooks, ‘J’ hooks and drop-back time: a hook performance study of the south Florida recreational live- bait fishery for sailfish, Istiophorus platypterus. Fish. Manage. Ecol. 14:173-182. Rudershausen, P. J., J. A. Buckel, R. Gregory, T. Averett, G. Bolton, and D. Britt. 2010. A comparison between circle hook and J hook performance in the dolphinfish, yellowfin tuna, and wahoo bluewater troll fishery off North Carolina. Final Report. North Carolina Sea Grant Project Number FRG # 08-FEG-02, 43 p. North Carolina Sea Grant Fish- ery Resource Program, Raleigh, NC. [Available from http://www.ncseagrant.org/home/research/core-research/ research-item?view = item & id = 1055.1 SAFMC (South Atlantic Fishery Management Council). 2003 Fishery management plan for the dolphin and wahoo fishery of the Atlantic, 308 p. South Atlantic Fishery Management Council, Charleston, SC. 2011. Summary of the comprehensive annual catch limit (ACL) amendment for the South Atlantic region, 81 p. South Atlantic Fishery Management Council, Charleston, SC. Sales, G., B.B. Giffoni, F.N. Fiedler, V.G. Azevedo, J.E. Kotas, Y. Swimmer, and L. Bugoni. 2010. Circle hook effectiveness for the mitigation of sea turtle bycatch and capture of target species in a Bra- zilian pelagic longline fishery. Aquat. Conserv. Mar. Freshw. Ecosyst. 20(41:428-436. Serafy, J. E., D. W. Kerstetter, and P. A. Rice. 2009. Can circle hook use benefit billfishes? Fish Fish. 10:132-142. Smith, S. 2006. Disconnected government circles: which circle hooks will meet non-offset circle hook requirement for Atlantic billfish tournaments? Billfish Mag. 3:26-27. 176 Distribution, movements, and habitat use of small striped bass ( Morone saxatilis ) across multiple spatial scales Kenneth W. Able' Thomas M. Grothues' Jason T. Turnure1 Donald M. Byrne2* Paul Clerkin' E-mail address for contact author: able@marine.rutaers.edu ' Rutgers University Marine Field Station 800 c/o 132 Great Bay Blvd. Tuckerton, New Jersey 08087 2 New Jersey Department of Environmental Protection Nacote Creek Research Station P.O Box 418 Port Republic, New Jersey 08241 * Posthumous Abstract — Distribution, movements, and habitat use of small (<46 cm, juveniles and individuals of unknown maturity) striped bass (Morone saxa- tilis) were investigated with multiple techniques and at multiple spatial scales (surveys and tag-recapture in the estuary and ocean, and telemetry in the estuary) over multiple years to determine the frequency and dura- tion of use of non-natal estuaries. These unique comparisons suggest, at least in New Jersey, that smaller individuals (<20 cm) may disperse from natal estuaries and arrive in non-natal estuaries early in life and take up residence for several years. During this period of estuarine resi- dence, individuals spend all seasons primarily in the low salinity portions of the estuary. At larger sizes, they then leave these non-natal estuaries to begin coastal migrations with those individuals from nurseries in natal estuaries. These composite observa- tions of frequency and duration of habitat use indicate that non-natal estuaries may provide important habitat for a portion of the striped bass population. Manuscript submitted 23 February 2011. Manuscript accepted 10 November 2011. Fish. Bull. 110:176-192 (2012) The views and opinions expressed or implied in this article are those of the author (or authors) and do not necessarily reflect the position of the National Marine Fisheries Service, NOAA. A full understanding of the distribu- tion, movements, and habitat use of juvenile and other subadult striped bass ( Morone saxatilis ) is central to deciphering the role, impacts, and management of this abundant and important species in estuarine and coastal ocean habitats. This is espe- cially true for the populations between Chesapeake Bay and the Gulf of Maine where adults can be highly migratory and are seasonal partici- pants in coastal migrations (Chapoton and Sykes, 1961; Boreman and Lewis, 1987; Waldman et ah, 1990; Able and Grothues, 2007; Welsh et ah, 2007; Grothues et ah, 2009; Mather et ah, 2010). More recently, it has become apparent that some components of these same populations may be resi- dent in estuaries throughout their life cycle (Secor, 1999; Secor and Pic- coli, 2007; Wingate and Secor, 2007). Despite the accumulating understand- ing of life cycle diversity for adults (see Secor and Kerr, 2009 for M. saxatilis and other species), we have an incom- plete understanding for juveniles and other subadults (Pautzke et ah, 2010). The conventional interpretation based on studies of natal estuaries, includ- ing Chesapeake Bay and other large estuaries (Merriman, 1941; Mansueti, 1961; Waldman et ah, 1990), is that juveniles remain in estuaries for the first few years of life before leaving to join the coastal migration and may stay longer, if they are natal estuarine residents (Secor 1999; Ashley et ah, 2000; Secor and Piccoli, 2007). For those individuals that even- tually become coastal migrants, the available data suggest that the du- ration of residency in estuaries ap- pears to vary with location and be- tween years, potentially relative to year class strength and associated density dependence (e.g., Kohlenstein, 1981), as well as size and maturity for males and females (e.g., Kohlen- stein, 1981; Secor and Piccoli, 2007). For example, an early interpreta- tion was that a mass emigration of small individuals takes place from Chesapeake Bay after ages 2 and 3 (Kohlenstein, 1981). More recently, analysis with otolith microchemistry suggests a gradual shift associated with sexual maturation at ages 5-8 for upper Chesapeake Bay individuals (Secor and Piccoli, 2007). In the Hud- son River, it is estimated that emi- gration from the estuary can occur into adjacent Long Island Sound and the New York Bight at ages 1 and 2 (Secor and Piccoli, 1996) or earlier by age-0 individuals (Dovel 1992, Dun- ning et ah, 2009). Able et al : Distribution, movements, and habitat use of small Morone saxatilis across multiple spatial scales 177 Our understanding of the distribution, movements, and habitat use of small striped bass is largely based on studies that occurred before the recovery in the late 1980s (Nichols and Miller, 1967; Clark, 1968; Kohlenstein, 1981; Boreman and Lewis, 1987; Wool- ey et ah, 1990; Richards and Rago, 1999). Further, most studies have focused on large natal estuaries such as the Hudson River (Secor and Piccoli, 1996), Chesapeake Bay (Mansueti, 1961; Kohlenstein, 1981; Secor, 2007), and to some extent the Delaware River (Waldman and Wirgin, 1994; Able et ah, 2007). There has been little emphasis on non-natal estuaries even though small striped bass are common and even abun- dant components of the fauna (for reviews see Able and Fahay, 1998, 2010). Therefore, we lack a clear understanding of their pattern of habitat use within estuaries, duration of residency, and patterns of tim- ing of emigration (Grothues et al., 2009). These pat- terns are especially confounded because the sources of small individuals in non-natal estuaries are largely unknown. The purpose of this article is to evaluate the distri- bution, movements, and habitat use of small striped bass in and adjacent to non-natal estuaries in New Jersey and adjacent areas. We approach this evaluation using multiple sources including information from sea- sonal catches from trawl, seine, and gill net surveys, tag-recapture studies, and telemetry. Most of these data relate to a period during or after the recovery of the population along the east coast. Further, we evalu- ate these patterns of estuarine and coastal ocean use at three scales: throughout the Middle Atlantic Bight continental shelf (Cape Cod to Cape Hatteras); on the inner continental shelf off New Jersey; and in the Mul- lica River-Great Bay estuary in southern New Jersey. Although the focus is on small individuals, i.e., from young-of-the-year to sexual maturity, the duration of this stage is sometimes difficult to define because the age (and size) at maturity varies between sexes, popu- lations, and even within the same estuary (see Fig. 1 in Specker et ah, 1987). We define the upper size limit for our treatment as 46 cm total length (TL) (approx, age 3-5 years; Merriman, 1941). In addition, there appears to be a natural difference in the size modes of several extensive sampling programs around this size (see below). The rationale for using this cutoff is that it includes the size at first maturity for some populations and that it complements our earlier telemetry studies of larger striped bass in the Mullica River-Great Bay estuary (Able and Grothues, 2007; Ng et ah, 2007; Grothues et al., 2009). Thus, those individuals <46 cm include those likely to be resident in estuaries, such as mature males (e.g., Wingate and Secor, 2007), but also include those that may begin leaving estuaries to par- ticipate in coastal migrations. For the purposes of this article, we make a distinction, where possible, between dispersal (from natal estuaries) by juveniles (< 20 cm) and other individuals of unknown maturity (>20-46 cm) and dispersal by those that make (directed, an- nual) coastal migrations. Materials and methods Study areas We used three geographical areas in this study (Fig. 1): l) continental shelf waters (to depths greater than 450 m) along the east coast of the United States between Cape Hatteras and Cape Cod; 2) a portion of the inner continental shelf (depths of 5.5-27.4 m) off the coast of New Jersey; and 3) the Mullica River-Great Bay estuary (average depth 2 m, some portions to 26 m) which is part of the Jacques Cousteau National Estuarine Research Reserve (JCNERR). Aspects of the geomorphology and hydrology of each of these areas is characterized in further detail elsewhere (Able and Fahay, 1998; 2010). Occurrence and distribution based on surveys Seasonal, coast-wide distributions for small ( <46 cm) striped bass on the continental shelf were determined with data from National Marine Fisheries Service (NMFS) Northeast Fisheries Science Center bottom trawl surveys (Azarovitz, 1981; Grosslein and Azarovitz, 1982) (Fig. 1, Table 1). Samples were collected on the continental shelf at stratified random stations between Cape Hatteras, North Carolina, and the Gulf of Maine during fall (September-October), winter (January-Feb- ruary) and spring (March-April) (Grosslein and Azarov- itz, 1982; Able and Fahay, 2010). The geographical limits of the sampling program, however, varied with season and between years. Similar sampling effort and distribu- tion of samples occurred in the fall (n= 7379 tows) and spring (n =7418 tows) over the period from 1982 through 2003. The winter sampling effort was reduced in terms of number of tows (n- 1552 tows) and geographical extent during the years in which it occurred (1992-2003). It was limited to the southern portion of Georges Bank and south of Cape Cod to just north of Cape Hatteras. In addition, the number of samples in the shallow waters (less than 25 m) off Massachusetts and from New Jersey to North Carolina was reduced in the winter but not in the fall and spring. The distribution of samples over all seasons varied with depth as well, with some less than 20 m (17%), a large portion less than 100 m (81%), fewer between 100 and 250 m (16%) and fewer still in depths >251 m (2%). See Able and Fahay (2010) for additional details. An estimate of the length distribution by age of striped bass was based on data from Mansueti (1941) and Able and Fahay (1998) and back-calculated length at age was based on otoliths of striped bass collected in Delaware Bay by the New Jersey Department of Envi- ronmental Protection (Baum1). Distribution data for small (<46 cm) striped bass off New Jersey were collected seasonally by otter trawl from 1996 to 2003 by randomly selecting sites in each of 15 sampling strata by the New Jersey Department 1 Baum, T. 2006. Personal commun. New Jersey Dep. Environmental Protection, Nacote Creek Research Center, Port Republic, NJ 08241. 178 Fishery Bulletin 1 10(2) Figure 1 Collection sites for striped bass ( Morone saxatilis) within the Mid-Atlantic Bight. Striped bass were collected by the National Marine Fisheries Service’s otter trawl survey (between Cape Cod and Cape Hatteras), New Jersey Department of Environmental Protection’s otter trawl survey (coast of New Jersey), and Rutgers University Marine Field Station’s estuarine-ocean beach-seine and estuarine gillnet surveys. Stationary telemetry hydrophone and water quality data logger locations (in the vicinity of Little Egg Inlet and the Mullica River-Great Bay estuary [inset]) are also shown. See Table 1 for timing of sampling. of Environmental Protection (NJDEP) (Fig. 1, Table 1). See Byrne2 and Sackett et al. (2007) for additional details. These sites were divided into three depth stra- ta and categorized as inshore (5. 5-9.1 m), mid-shore 2 Byrne, D. M. 1989. New Jersey trawl surveys. In Special Report No. 17 of the Atlantic States Marine Fisheries Com- mission (Azarovitz, T. R., J. McGurrin, and R. Seagraves, eds.), p. 46-48. Atl. States, Marine Fish. Comm. .Woods Hole, MA. Able et al.: Distribution, movements, and habitat use of small Morone saxatilis across multiple spatial scales 179 (9.1-18.3 m), and offshore (18.3-27.4 m). Trawl loca- tions were mapped with GIS (ArcGIS3, vers. 9.2, ESRI, Redlands, CA). The entire otter trawl data matrix con- sisted of 2872 records of catch per unit of effort (CPUE; number of individuals per tow), average depth, date, season (spring, April; summer, June-August; fall, Sep- tember-October; winter, January-February), and depth category (inshore, midshore, and offshore). Additional collections from the surf zone adjacent to and within the Mullica River-Great Bay estuary were collected by seine during 1998-99 and 2004-06 (Table 1, Fig. 1). See Taylor et al. (2007) for additional details. In order to determine the estuarine distribution of other small (<46 cm) striped bass in space and time, we sampled with anchored multimesh gill nets (15 mx2.4 m nets with five panels of five box-mesh sizes 2.5, 3.8, 5.1, 6.4, and 7.6 cm) in the Mullica River-Great Bay estuary at several locations (Table 1, Fig. 1). Gill nets were set (for approximately 60 min during the day) at biweekly intervals during the spring, summer, and fall in upper creek, creek mouth, and nearshore bay habitats. Within each area, the position in which each net was set varied such that no two locations were sampled twice. See Able and Fahay (2010) for additional details. Another sampling program was conducted with small otter trawls between 1988-90 and 1996-2009 at a va- riety of stations and habitats located throughout the Mullica River-Great Bay-Inner Continental Shelf cor- ridor (Table 1). These stations were distributed along the salinity gradient from the ocean to tidal freshwater. Other individuals were collected in composite surveys in Delaware Bay with a variety of gear types and from habitats during 1998-2006 (Table 1; Able et al., 2007; Able and Fahay, 2010). Still others came from an exten- sive seine survey in the Hudson River estuary (Table 1). Tag-recapture The tagging procedure outlined in Boreman and Lewis (1987) for their study with American Littoral Society (ALS) data is consistent with the protocol followed in our study. After initial capture, code-specific loop tags were inserted into the dorsal region of each fish and the fish was released. Length, general capture and release location, and date were recorded for each animal on a supplied tagging card and mailed to ALS. The ALS sends raw data to the National Marine Fisheries Service in Woods Hole, Massachusetts, for processing and entry into a long-term database (Shepherd4). We limited the query of records to two subsets of data: 1) striped bass initially captured in New Jersey waters and recaptured at less than 46 cm TL along the eastern United States coast; and 2) striped bass initially captured in nearby 3 Mention of trade names or commercial companies is for identification purposes only and does not imply endorsement by the National Marine Fisheries Service, NOAA. 4 Sheperd, G. 2009. Personal commun. NMFS Northeast Fisheries Science Center, 166 Water Street, Woods Hole, MA 02543-1026. natal estuaries (Hudson and Delaware rivers) and recap- tured in New Jersey waters at less than 46 cm TL (Table 1). The latitude and longitude coordinates associated with each general capture and recapture location were assigned by ALS and NMFS by calculating the spatial average of each location submitted by volunteer taggers. Telemetry We determined dynamic habitat use and movements of small (32.4-42.5 cm fork length [ FL] ) striped bass in the Mullica River-Great Bay estuary using acoustic telemetry. Wireless hydrophones were moored as a series of gates in order to determine occurrence and residency of tagged individuals along the estuarine gradient (Fig. 1). Fishes surgically implanted with individually coded acoustic transmitters (76.8 kHz) were detected when they came within range (approximately 500 m) of moored wireless hydrophones (WHS-1100, Lotek Wireless, Inc., St. Johns, Newfoundland, Canada) suspended at a depth of approximately 3.2 m (see Grothues et al. [2005] for additional details). Permanent environmental-monitor- ing instruments in the Jacques Cousteau National Estu- arine Research Reserve included data loggers recording salinity, temperature, pH, and water depth (Kennish and O’Donnell, 2002) along the estuarine gradient (Fig. 1). In addition, mobile tracking methods were used to determine fine-scale patterns of habitat use. In order to spatially and temporally standardize tracking, 113-120 fixed locations were georeferenced with a global po- sitioning system (GPS) unit in universal transverse mercator (UTM) coordinates by using a GIS software package (ArcGIS, vers. 9.2, ESRI) and visited with a directional mobile hydrophone on a weekly basis (LHP_1; Lotek Wireless). Listening range with the mo- bile hydrophone was typically about 500 m, determined by signal range tests. At each of the above locations, the hydrophone was lowered 1.0 m into the water and pointed at the four principal ordinates for 5 seconds in each direction. When a fish was detected, its position was triangulated by moving until a reading of 115 dB or above was detected at a gain of 15 or less (ap- proximately 2 m from the hydrophone). Measurements of water temperature and salinity were collected (YSI Model 85; Yellow Springs Instruments, Inc., Yellow Springs, Ohio), along with date, time, tag number, and depth at each confirmed fish detection. Tracking was not conducted when the listening range was less than 500 m (which corresponded to wind velocities greater than 30 km/h) or on days when there was heavy rainfall or thunderstorms. See Ng et al. (2007) and Sackett et al. (2008) for further details on mobile tracking proto- col. To determine patterns of seasonal habitat use in relation to physical habitat variables, the distances of individually tagged striped bass from emergent (marsh) and submerged (channel) embankment edges were cal- culated by using a GIS software package. The loca- tions of submerged edges were derived from estuarine bathymetry data by calculating high slope areas (i.e., channel edges; >2.5°) and submerged or emergent edge 180 Fishery Bulletin 1 10(2) Summary of data Table 1 sources for juvenile striped bass ( Morone saxatilis) examined in the current study. See Fig. 1 for sampling locations. General habitat Geographic location Sampling gear Sampling frequency/duration Ocean Atlantic coast Otter trawl Fall, winter, spring/1982-2003 Atlantic coast Tag-recapture Fall, winter, spring, summer/1962, 1967, 1973, 1977-2009 New Jersey coast Otter trawl Fall, winter, spring, summer/1988-2003 Central New Jersey coast Seine Biweekly/June - November 1998, May-October 1999-2000, July 2004, May-October 2005, August-October 2006 Estuary Mullica River-Great Bay Otter trawl Monthly/July and September 1988-1990, 1996-2009 Mullica River-Great Bay Multi-mesh gill net Monthly/August-October 2001; Semi-monthly/May-October 2002 Mullica River- Great Bay Seine Biweekly/June - November 1998, May-October 1999-2000, July 2004, May-October 2005, August-October 2006 Mullica River-Great Bay Acoustic telemetry Mobile (Weekly/2006-2008) Passive (Continuous/2006-2008) Delaware Bay Otter trawl/weirs Monthly/Aprii - November 1996-2000; May-November 2001-2005 Hudson River Seine July-November 1990-2009 distances were calculated as the straight-line distance (m) to the nearest edge. Results Occurrence and distribution based on surveys Small ( < 4 6 cm TL) striped bass were represented in multiple sampling gears from multiple locations throughout the study area (Table 1, Fig. 2). However, individuals <20 cm (presumed age 0-1 years) were seldom collected in the coastal ocean, including the NMFS trawl survey between Cape Hatteras and Cape Cod (n = 2 individuals), the NJDEP trawl survey (n = 61 individuals), and the Rutgers University Marine Field Station (RUMFS) beach seine survey along the inner continental shelf off New Jersey (n = 21 individuals) despite the large number of samples. These smaller individuals were also not abundant in estuarine seine, gill net, or otter trawl collections within the Mullica River-Great Bay estuary based on over 3100 samples (Table 1, Fig. 2). Of these, individuals <20 cm were collected only within the estuary during otter trawl (n = 21, 3.4-19.5 cm) and gillnet (n = 1, 16.4 cm) sam- pling. Alternatively, large numbers of small individuals <20 cm have been collected from the Delaware River and Hudson River estuaries, both known spawning areas (Fig. 2, G and H). Larger juveniles (21-46 cm, presumed age 2-5 years) were better represented in surveys in most locations including the Mullica River- Great Bay estuary (n = 55; Fig. 2). The seasonal patterns of distribution were similar regardless of the spatial scale examined. Individuals 20 to 46 cm, according to the NMFS surveys on the continental shelf, were seldom collected in the fall and winter (a period of restricted sampling in shallow waters) surveys. During the spring (February-March) they were more abundant and largely restricted to the inner portion of the shelf according to compos- ite collections during 1982-2003 (Fig. 3). Most were restricted to an area from north of the Chesapeake Bay mouth to Long Island including the coast of New Jersey. A similar shallow-water distribution, in space and time, of individuals <46 cm is evident from depth strati- fied sampling off the coast of New Jersey during all seasons from 1988 through 2003 (Figs. 4 and 5). Both smaller (<20 cm), although less common, and larger (21-46 cm) individuals were most abundant in the spring but also occurred during the winter months and were either rare or absent in the summer and relatively rare during the fall. Over all these seasons, both of these size groups were most abundant in the nearshore depth strata (5.5-9. 1 m) with a trend to decreasing abundance with depth with the least number of collec- tions in the offshore strata (18.3-27.4 m). During the winter and spring the larger individuals (21-46 cm) were found all along the coast from the mouth of Dela- ware Bay to the tip of Sandy Hook (Fig. 5). Able et al Distribution, movements, and habitat use of small Morone saxatilis across multiple spatial scales 181 Sampling events Or tOWS (72) Water depths sampled (m) No. of individuals (<46 cm) Data source >16,000 5-481 438 National Marine Fisheries Service; Grosslein and Azarovitz (1982); Able and Brown (2005) >300,000 (captures); No data 1529 (recaptures) American Littoral Society; current study >19,000 (recaptures) 2872 3-80 2930 New Jersey Department of Environmental Protection; Byrne (1989); current study 526 <2 9 Able et al. (2003); current study 2328 0.35-26.0 27 Able and Fahay (2010) 599 0-8 28 Able and Fahay (2010) 243 <2 9 Able et al. (2003); current study Mobile (80) 1-25 14 Current study Passive (>50,000) >15,000 1-24 5343 Nemerson and Able (2003); Able et al. (2007) — <2 108,445 (1-39 cm) New York Department of Environmental Conservation Movements determined with tag-recapture methods Tagged and recaptured individuals revealed that they could move from natal estuaries to the vicinity of non- natal sources along the New Jersey coast and that those individuals that were found along the New Jersey coast could move to other areas. Few individuals tagged in the nearest natal estuaries (Hudson River and vicinity, n= 25, and Delaware River, n- 1) were recaptured along the New Jersey coast (n= 26 total, Fig. 6A). Small striped bass captured in neighboring natal estuaries ranged in size from 30-46 cm before being recaptured in New Jersey (33-46 cm). Days at liberty for fish captured in nearby natal estuaries ranged from 13-892 (mean 276 days). Individuals tagged in or along the New Jersey coast (n = 152 total) were recaptured throughout the northeastern United States from northern Chesapeake Bay (n = 4; 3%), Delaware Bay (n = 19; 13%), and Long Island and Connecticut (n= 27; 18%), with some found as far north as Cape Cod and Maine (re=21; 14%). The majority of recaptures, however, occurred along the New Jersey coast (n = 81; 53%; Fig. 6B). The time between capture and recapture was similar for this subset of fish (1-868 days; mean 244 days). For those fish originally captured in New Jersey and recaptured elsewhere along the coast, sizes were generally smaller than in the other subset of fish analyzed and lengths ranged from 25 to 46 cm during capture and from 28 to 46 cm during recap- ture. A relatively small number of fish were recaptured at sizes less than 40 cm TL (n = 18; 11.8%), with all but one of these individuals recaptured less than 100 km from their original release location in New Jersey waters (Fig. 6B). Estuarine habitat use determined with acoustic telemetry From 2006 through 2008, 14 small striped bass (32.4- 42.5 cm FL) were tagged with acoustic transmitters within the Mullica River-Great Bay estuary in south- ern New Jersey (Tables 1 and 2, Fig. 7). Most were consistently detected (11 of 14 individuals, 72 = 114 detec- tions) based on mobile telemetry. An examination of the seasonal distribution revealed consistent use of the mesohaline portions of the river all the way up to, and occasionally above, the freshwater-saltwater interface, whereas fewer were found in polyhaline waters near Little Egg Inlet (see Fig. 1). In the summer, fall, and spring some individuals were detected downstream near Little Egg Inlet, or in Great Bay, but during the winter all juveniles were detected upstream in the river (Fig. 7). During December 2006, four fish (42-48 cm FL) were tagged in the ocean off Long Beach Island (Fig. 7C). Of these, one (code 104) moved into the estuary by way of Main Marsh Thorofare (see Fig. 1) on December 24 and remained there for approximately 125 days. The use of upriver habitats (such as Lower Bank) was evident by the temperature (Fig. 8A) and salinities (Fig. 8B) at which tagged juvenile striped bass were frequently detected. Juveniles inhabited the warmer water temperatures found upstream in the summer 182 Fishery Bulletin 1 10(2) 20000 1S000 10000 5000 0 ALS tag-recaptlire I 1 Recapture (77=14,564) l*#g^l Capture (n=338, 666) .11 J J3 RUMFS telemetry study (/?= 134) Current study Other tagged Striped bass 40 cm TL) that may be joining the annual coastal migration. These larger individuals were also frequently recaptured in presumably non-natal habitats in the Gulf of Maine and along the coasts of Connecticut and Rhode Island after initial release in New Jersey waters. This same pattern has been reported for age 2 and 3 fish moving into non- natal estuaries, such as the Connecticut River (Kynard and Warner, 1987) and in Massachusetts where 40-50 cm TL individuals (most age 2-5) apparently feed during the summer, make coastal migrations during the fall through spring, but return in subsequent years (Mather et al., 2009, 2010; Pautzke et al., 2010). Certainly, other estuarine-dependent fish species leave estuaries when they reach a size threshold (Rountree and Able, 1992; Potter et ah, 1997). This pattern for striped bass may vary with sex, i.e., females are likely to leave at earlier ages or smaller sizes, whereas males tend to remain, at least in natal estuaries, for longer periods of time (Secor and Piccoli, 1996). In general, overall patterns of use of a non-natal estu- ary and scheduling of departure appear similar between natal and non-natal estuaries and we also suspect that there are no major changes in these patterns before and after the recovery of the striped bass population in recent years (Boreman and Lewis, 1987). However, we hasten to point out that there was little emphasis on non-natal estuaries as secondary nursery habitat before the recovery. Conclusion As we have demonstrated, non-natal estuaries are potentially important habitat for small (20-46 cm) striped bass. This finding may further complicate our understanding of life cycle diversity (see Secor and Kerr, 2009) for this species because the prior focus has been on natal estuaries. Further, as these individu- als from non-natal estuaries join the annual coastal 190 Fishery Bulletin 1 10(2) migration, grow, and mature, one wonders where they are likely to spawn. One possibility is that they will attempt to spawn in the non-natal estuaries where they have previously spent several months to years. This could account for the seeming unsuccessful attempts in the Mullica River-Great Bay estuary (Able and Grothues, 2007; Grothues et ah, 2009). One could also argue that these individuals may be responsible for colonizing new spawning sites, as has previously been suggested (Grothues et ah, 2009). Alternatively, they may join other maturing individuals as they migrate back to their natal rivers and streams that provided primary nurseries. Otolith microchemistry might be the appropriate means to distinguish the ultimate source of individuals that use non-natal estuaries and the site of their subsequent spawning. Acknowledgments Numerous individuals from the National Marine Fisher- ies Service, New Jersey Department of Environmental Protection-Bureau of Marine Fisheries (NJDEP-BMF), and the Rutgers University Marine Field Station (RUMFS) assisted in data collection, surveys, and telem- etry. The angling skills of D. Messerschmidt and L. Webb assisted our tagging efforts. K. Hattala (New York Dept. Environmental Conservation) provided length data from the Hudson River young-of-the-year survey. B. Muffley and R. Allen (NJDEP-BMF) provided comments on earlier drafts. Funding sources for this study were pro- vided by the Rutgers University Institute of Marine and Coastal Sciences (IMCS), RUMFS, the Bluefish/Striped Bass Dynamics Research Program of Rutgers University, and the NOAA National Marine Fisheries Service. J. Plangere Jr. generously supported the telemetry study. This article is contribution number 2012-4 of the Rutgers University Institute of Marine and Coastal Sciences. Literature cited Able, K. W„ J. H. Balletto, S. M. Hagan, P. R. Jivoff, and K. Strait. 2007. Linkages between salt marshes and other nekton habitats in Delaware Bay, USA. Rev. Fish. Sci. 15:1-61. Able, K.W. and R. Brown. 2005. Distribution and abundance of young-of-the- year estuarine fishes: seasonal occurrence on the Middle Atlantic Bight continental shelf. IMCS Tech- nical Report 2005-14, 62 p. Rutgers Univ., New Brunswick, NJ. Able, K. W„ and M. P. Fahay. 1998. The first year in the life of estuarine fishes in the Middle Atlantic Bight, 342 p. Rutgers Univ. Press, New Brunswick, NJ. 2010. Ecology of estuarine fishes: temperate waters of the western North Atlantic, 566 p. Johns Hopkins Univ. Press, Baltimore, MD. Able, K. W„ and T. M. Grothues. 2007. Diversity of estuarine movements of striped bass ( Morone saxatilis): a synoptic examination of an estuarine system in southern New Jersey. Fish. Bull. 105:426-435. Able, K.W., P. Rowe, M. Burlas, and D. Byrne. 2003 Use of ocean and estuarine habitats by young-of- year bluefish ( Pomatomus saltatrix) in the New York Bight. Fish. Bull 101:201-214. Ashley, J. T. F., D. H. Secor, E. Zlokovitz, S. Q. Wales, and J. E. Baker. 2000. Linking habitat use of Hudson River striped bass to accumulation of polychlorinated biphenyl conge- ners. Environ. Sci. Technol. 34:1023-1029. Azarovitz, T. R. 1981. A brief historical review of the Woods Hole Labo- ratory trawlsurvey time series. Bottom trawl sur- veys. Can. Spec. Publ. Fish. Aquat. Sci. 58:1-273. Bass, A. J. 1978. Problems in the studies of sharks in the southwest Indian Ocean. In Sensory biology of sharks, skates and rays (E. S. Hodgson, and R. F. Mathewson, eds.), p. 545-594. Office of Naval Research, Dept. Navy, Arlington, VA. Beck, M. W., K. L. Heck Jr., K.W. Able, D. Childers, D. Eggleston, B. M. Gillanders, B. Halpern, C. Hays, K. Hoshino, T. Minello, R. Orth, P. Sheridan, and M.Weinstein. 2001. The identification, conservation, and management of estuarine and marine nurseries for fish and inver- tebrates. Bioscience 51:633-641. Boreman, J. and R. R. Lewis. 1987. Atlantic coastal migration of striped bass. Am. Fish. Soc. Symp. 1:331-339. Byrne, D. M. 1989. New Jersey trawl surveys. In Special report no. 17 of the Atlantic States Marine Fisheries Commission (T. R. Azarovitz, J. McGurrin, and R. Seagraves ,eds.), p. 46-48. Atlantic States Mar. Fish. Comm., Woods Hole, MA. Chapoton, R. B., and J. E. Sykes. 1961. Atlantic coast migration of large striped bass as evidenced by fisheries and tagging. Trans. Am. Fish. Soc. 90:13-20. Clark, J. 1968. Seasonal movements of striped bass contingents of Long Island Sound and the New York Bight. Trans. Am. Fish. Soc 97:320-343. Collette, B. B., and G. Klein-MacPhee (eds.) 2002. Bigelow and Schroeder’s fishes of the Gulf of Maine, 3rd ed., 748 p. Smithsonian Inst. Press, Washington, D.C. Dahlgren, C. P., G. T. Kellison, A. J. Adams, B. M. Gillanders, M. S. Kendall, C. A. Layman, J. A. Ley, I. Nagelkerken, J. E. Serafy. 2006. Marine nurseries and effective juvenile habi- tats: concepts and applications. Mar. Ecol. Prog. Ser. 312:291-295 Deegan , L. A. 1993. Nutrient and energy transport between estuaries and coastal marine ecosystems by fish migration. Can. J. Fish Aquat. Sci. 50:74-79. 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. Dovel, W. L. 1992. Movements of immature striped bass in the Hudson Able et al Distribution, movements, and habitat use of small Morone saxatilis across multiple spatial scales 191 Estuary. In Estuarine research in the 198Q’s (C. L. Smith, ed.), p. 276-300. State Univ. New York Press, Albany, NY. Dunning, D. J., Q. E. Ross, K. A. Mckown, and J. B. Socrates. 2009. Effect of striped bass larvae transported from the Hudson River on juvenile abundance in western Long Island Sound. Mar. Coastal Fish.: Dynamics, Manage. Ecosystem Sci. 1:343-353. Fodrie, F. J., L. A. Levin, and A. J. Lucas. 2009. Use of population fitness to evaluate the nursery function of juvenile habitats. Mar. Ecol. Prog. Ser. 385:39-49. Gillanders, B. M., K. W. Able, J. A. Brown, D. E. Eggleston, and P. F. Sheridan. 2003. Evidence of connectivity between juvenile and adult habitats for mobile marine fauna: an important compo- nent of nurseries. Mar. Ecol. Prog. Ser. 247:281-295. Grosslein, M. D., and T. R. Azarovitz (eds.) 1982. Fish distribution. Mar. Ecosyst. Anal. N.Y. Bight Atlas Monogr. 15, 182 p. Grothues, T. M., K. W. Able, J. Carter, and T. Arienti. 2009. Migration patterns of striped bass through non- natal estuaries of the U. S. Atlantic coast. Am. Fish. Soc. Symp. 69:135-150. Grothues, T. M., K. W. Able, J. McDonnell, and M. M. Sisak. 2005. An estuarine observatory for real-time telemetry of migrant macrofauna: design, performance and con- straints. Limnol. Oceanogr. Methods 3:275-289. Hurst, T. P., and D. O. Conover. 1998. Winter mortality of young-of-the-year Hudson River striped bass ( Morone saxatilis ): size-dependent patterns and effects on recruitment. Can. J. Fish. Aquat. Sci. 55:1122-1130. Hurst, T. P., E. T. Schultz, and D. O. Conover. 2000. Seasonal energy dynamics of young of-the-year Hudson River striped bass (Morone saxatilis ): size- dependent patterns and effects on recruitment. Can. J. Fish. Aquat. Sci. 55:1122-1130. Kennish, M. J., and S. O’Donnell. 2002. Water quality monitoring in the Jacques Cousteau National Estuarine Research Reserve System. Bull. NJ Acad. Sci. 47(21:1— 14. Kohlenstein, L. C. 1981. On the proportion of the Chesapeake Bay stock of striped bass that migrates into the coastal fish- ery. Trans. Am. Fish. Soc. 110:168-179. Kynard, B. and J. P. Warner. 1987. Spring and summer movements of subadult striped bass, Morone saxatilis, in the Connecticut River. Fish. Bull. 85:143-147. Little, M. J. 1995. A report on the historic spawning grounds of the striped bass, Morone saxatilis. Maine Naturalist 3:107-113. 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. Mather, M. E., J. T. Finn, K. H. Ferry, L. A. Deegan, and G. A. Nelson. 2009. Use of non-natal estuaries by migratory striped bass (Morone saxatilis) in summer. Fish. Bull. 107:329-338. Mather, M. E., J. T. Finn, S. M. Pautzke, D. Fox, T. Savoy, H. M. Brundange III, L. A. Deegan, and R. M. Muth. 2010. Diversity in destinations, routes, and timing of small adult and sub-adult striped bass Morone saxatilis on their southward autumn migration. J. Fish Biol. 77: 2326-2337. Merriman, D. 1941. Studies of the striped bass ( Roccus saxatilis) of the Atlantic coast. U.S. Fish Wildlife Serv. Fish. Bull. 50(351:1-77. Merson, R. R., and H. L. Pratt Jr. 2007. Sandbar shark nurseries in New Jersey and New York: evidence of northern pupping grounds along the United States East Coast. In Shark nursery grounds of the Gulf of Mexico and the East Coast waters of the United States (C. T. McCandless, N. E. Kohler, and H. L. Pratt Jr., eds.), p. 35-43. Am. Fish. Soc., Bethesda, MD. Nemerson, D. M., and K. W. Able. 2003. Spatial and temporal patterns in the distribution and feeding habits of Morone saxatilis in marsh creeks of Delaware Bay, USA. Fish. Manage. Ecol. 10:337-348. Ng, C. L., K. W. Able, and T. M. Grothues. 2007. Habitat use, site fidelity, and movement of adult striped bass in a southern New Jersey estuary based on mobile acoustic telemetry. Trans. Am. Fish. Soc. 136:1344-1355. Nichols, P. R. and R. V. Miller. 1967. Seasonal movements of striped bass, Roccus saxati- lis (Walbaum), tagged and released in the Potomac River, Maryland, 1959-1961. Chesapeake Sci. 8(21:102-124. Pautzke, S. M., M. E. Mather, J. T. Finn, L. A. Deegan, and R. M. Muth. 2010. Seasonal use of a New England estuary by forag- ing contingents of migratory striped bass. Trans. Am. Fish. Soc. 139:257-269. Potter, I. C., D. Tiivel, F. J. Valesini, and G. A. Hyndes. 1997. Comparisons between the ichthyofaunas of a tem- perate lagoonal-like estuary and the embayment into which that estuary discharges. Int. J. Salt Lake Res. 5:337-358. Richards, R. A. and P. J. Rago. 1999. A case history of effective fishery management: Chesapeake Bay striped bass. N. Am. J. Fish. Manage. 19:356-375. Rountree, R. A., and K. W. Able. 1992. Fauna of polyhaline subtidal marsh creeks in southern New Jersey: Composition, abundance and biomass. Estuaries 15:171-185. Sackett, D. K., K. W. Able, and T. M. Grothues. 2007. Dynamics of summer flounder, Paralichthys den- tatus, seasonal migrations based on ultrasonic telem- etry. Estuar. Coast. Shelf Sci. 74:119-130. Sackett, D. K., K. W. Able, and T. M. Grothues. 2008 Habitat dynamics of summer flounder, Paralich- thys dentatus, within a shallow USA estuary, based on multiple approaches using acoustic telemetry. Mar. Ecol. Prog. Ser. 364:199-212. Secor, D. H. 1999. Specifying divergent migrations in the concept of stock: the contingent hypothesis. Fish. Res. 43:13-34. 2007. The year-class phenomenon and the storage effect in marine fishes. J. Sea Res. 57:91-103. Secor, D. H., and L. A. Kerr. 2009. Lexicon of life cycle diversity in diadromous and other fishes. Am. Fish. Soc. Symp. 69:1-20. Secor, D. H., and P. M. Piccoli. 1996. Age- and sex-dependent migrations of striped bass in the Hudson River as determined by chemical micro- analysis of otoliths. Estuaries 19:778-793. 2007. Oceanic migration rates of upper Chesapeake 192 Fishery Bulletin 1 10(2) Bay striped bass ( Morone saxatilis ) determined by otolith microchemical analysis. Fish. Bull. 105: 62-73. Setzler-Hamilton, E. M., and L. Hall Jr. 1991. Ch. 13: Striped bass Morone saxatilis. In Habitat requirements for Chesapeake Bay living resources (S. L. Funderburk, J. A. Mihursky, S. J. Jordan, and D. Riley, eds.), p. 13.1-13.31. Chesapeake Res. Consor- tium, Annapolis, MD. Sheaves, M. 2005. Nature and consequences of biological connectiv- ity in mangrove systems. Mar. Ecol. Prog. Ser. 302: 293-305. Sheaves, M., R. Baker, and R. Johnston. 2006. Marine nurseries and effective juvenile habitats: an alternative view. Mar. Ecol. Prog. Ser. 318:303-306. Specker, J. L., D. L. Berlinsky, and H. D. Bibb. 1987 Oocyte development in striped bass: factors influ- encing estimates of age at maturity. Am. Fish. Soc. Symp. 1:162-174. Taylor, D. L., R. Nichols, and K. W. Able. 2007. Habitat selection and quality for multiple cohorts of young-of-the-year bluefish Pomatomus saltatrix: comparisons between estuarine and ocean beaches in southern New Jersey. Estuar. Coast. Shelf Sci. 73:667- 679. Waldman, J. R., D. J. Dunning, Q. E. Ross, and M. T. Mattson. 1990. Range dynamics of Hudson River striped bass along the Atlantic coast. Trans. Am. Fish. Soc. 119:910-919. Waldman , J. R., and I. I. Wirgin. 1994. Origin of the present Delaware River striped bass population as shown by analysis of mitochondrial DNA. Trans. Am. Fish. Soc. 123:15-21. Weisberg, S. B., P. Himchak, T. Baum, H. T. Wilson, and R. Allen. 1996. Temporal trends in abundance of fish in the tidal Delaware River. Estuaries 19: 723-729. Welsh, S. A., D. R. Smith, R. W. Laney, and R. C. Tipton. 2007. Tag-based estimates of annual fishing mortality of a mixed Atlantic coast stock of striped bass. Trans. Am. Fish. Soc. 136:34-42. Wingate, R. L., and D. H. Secor. 2007. Intercept telemetry of the Hudson River striped bass resident contingent: migration and homing pat- terns. Trans. Am. Fish. Soc. 136:95-104. Wooley, C. M., N. C. Parker, B. M. Florence, and R. W. Miller. 1990. Striped bass restoration along the Atlantic coast: a multistate and federal cooperative hatchery and tagging program. Am. Fish. Soc. Symp. 7:775-781. 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. 193 Prey preference of lingcod ( Ophiodon elongatus), a top marine predator: implications for ecosystem-based fisheries management Craig A. Tinus Email address for contact author: craiq.tinus@oreaonstate.edu Oregon State University Department of Fisheries and Wildlife 104 Nash Hall Corvallis, Oregon 97331-3803 Present address: Oregon Department of Fish and Wildlife Corvallis Research Lab 28655 Hwy 34 Corvallis, Oregon 97330 Abstract — Many highly exploited eco- systems are managed on the basis of single-species demographic informa- tion. This management approach can exacerbate tensions among stakehold- ers with competing interests who in turn rely on data with notoriously high variance. In this case study, an application of diet and dive survey data was used to describe the prey preference of lingcod ( Ophiodon elon- gatus) in a predictive framework on nearshore reefs off Oregon. The ling- cod is a large, fast-growing generalist predator of invertebrates and fishes. In response to concerns that lingcod may significantly reduce diminished populations of rockfishes ( Sebastes spp. ), the diets of 375 lingcod on near- shore reefs along the Oregon Coast were compared with estimates of relative prey availability from dive surveys. In contrast to the transient pelagic fishes that comprised 46% of lingcod diet by number, rockfishes comprised at most 4.7% of prey items. Rockfishes were the most abundant potential prey observed in dive sur- veys, yet they were the least preferred. Ecosystem-based fisheries manage- ment (EBFM) requires information about primary trophic relationships, as well as relative abundance and dis- tribution data for multiple species. This study shows that, at a minimum, predation relative to prey availability must be considered before predator effects can be understood in a man- agement context. Manuscript submitted 2 February 2011. Manuscript accepted 28 November 2011. Fish. Bull. 110:193-204 (2012). The views and opinions expressed or implied in this article are those of the author (or authors) and do not necessarily reflect the position of the National Marine Fisheries Service, NOAA. Ecosystem-based fisheries manage- ment (EBFM) has generated consider- able interest over the last decade as a way of better managing depressed fisheries stocks (Pikitch et ah, 2004; Gaichas et ah, 2010). This interest has been in large part a reaction to the perceived failure of traditional single-species fisheries management to prevent the collapse of exploited and ancillary populations in many systems worldwide (Dulvy et ah, 2003; Hutchings and Baum, 2005; Myers and Worm, 2005). One implication of EBFM is the capacity to forecast changes in managed populations in reaction to fluctuations in linked predator and prey populations. This requires understanding what species of interest consume in a given tempo- ral and spatial context. An additional consideration is that many exploited fishes are generalist consumers and shifts in densities and distributions may produce complex top-down effects (Bruno and O’Connor, 2005). These are among numerous challenges in gathering the information required to describe even a subset of primary tro- phic relationships in a dynamic system. Prey preference is the differential consumption of some prey types over others given equal availability. It is considered a fixed behavioral char- acteristic and, as a way of forecast- ing predation intensity on managed stocks, has received little attention. However, preference models may be both useful and efficient as an exten- sion of food web models to aid man- agement of exploited stocks (Gaichas et al., 2010). In this study I used an analysis of dive survey data with consumption data in a prey prefer- ence model to better understand the likely effects of a rapidly increas- ing predator population on managed prey. Consumption rates for gener- alist consumers like lingcod ( Ophi- odon elongatus Girard) may be either positively or negatively correlated among different prey types, or they may be uncorrelated, and these ef- fects can be important in actively managed systems (Dill et ah, 2003). If spatially and temporally transient prey species predominate in the diet of a resident predator, they may con- stitute subsidies to the local preda- tor population (Anderson and Polis, 1998). When subsidies occur there may be a concomitant suppression of local prey species through apparent competition among prey types with a common predator (Holt, 1977; Chan- eton and Bonsall, 2000). Especially in marine systems where trophic webs may be poorly defined (Thompson et ah, 2007), initial consideration of predator-prey relationships requires dietary analysis (Heithaus et ah, 2008). Diets of targeted fishery spe- cies provide necessary information for understanding food web structure, which is an important requirement for ecosystem-based fisheries science 194 Fishery Bulletin 1 10(2) and management (Francis et al., 2007). However, in addition to these basic trophic relationships, it is neces- sary to understand the context in which prey are being selected. The effects of predation on both predator and prey populations change as prey densities vary. Although EBFM requires even more information than traditional single-species management approaches, managers, scientists, and stakeholders make use of less certain information both in less accessible systems and in those that are accessible but where temporal and spatial scales far exceed the capacity to collect local demographic data. For these reasons identifying specific management triggers based on comprehensive and col- lectable information has been proposed (Samhouri et ah, 2010) and the case made that uncertain data and imperfect advice must be embraced, as long as they are appropriate data (Ludwig et ah, 1993; Johannes, 1998; Frid et ah, 2008). Challenges to the use of EBFM in- clude “species conflicts,” where management and stake- holder interest in one target species may interfere with other species and often involve the assumed effects of large generalist predator(s) on recovering high value prey species, sometimes in and out of marine protected areas. Examples of generalist predators involved in management conflicts are groupers ( Epinephelus spp. [Ault et ah, 2006; Coleman and Koenig, 2010]), red snapper ( Lutjanus campechanus [Wells et ah, 2008; Cowan et ah, 2010]), cod ( Gadus morhua [Link and Garrison, 2002]), and striped bass ( Morone saxatilis [Paolisso, 2002; Walter et ah, 2003]). Marine reserves are becoming more widely consid- ered as a management tool for protecting a portion of breeding populations as interest in EBFM increases. However, in addition to providing a refuge from fishing mortality, marine reserves can enhance local popula- tions of large, resident, top-level predators (Martell et ah, 2000; McClanahan and Arthur, 2001). Among possible effects of a local increase in predator biomass is a decrease in a particular prey type (Graham et ah, 2003). For example, this kind of interaction has been proposed for lingcod predation on rockfishes (Sebastes spp.) within marine reserves (Beaudreau and Essing- ton, 2007; 2009) and both are major targets of com- mercial fisheries. The following case study exemplifies necessary con- siderations for EBFM. Lingcod are targeted by both recreational and commercial fishermen along the west coast of North America. The 2000 stock assessment of lingcod from British Columbia to northern California estimated biomass at 11% of precommercial exploitation levels (Jagielo, et ah1) and management substantially reduced fishing mortality to allow recovery of this stock. By 2006, lingcod stocks were declared fully recovered by the Pacific Fisheries Management Council. Lingcod are 1 Jagielo, T. H., F. R. Wallace, and Y. W. Cheng. 2003. Assess- ment of lingcod (Ophiodon elongatus). Amendment 16-2: Rebuilding plans for darkblotched rockfish, Pacific ocean perch, canary rockfish, and lingcod. Environmental impact statement and regulatory analysis, 129 p. Pacific Fishery Management Council, Portland, OR. large (up to 152 cm total length [TL] and 59 kg) and fast growing. They are relatively site-attached, demer- sal, generalist predators, found on shallow northeastern Pacific rocky reefs. They roam across both rocky habitat and soft-bottom over distances of at least hundreds of meters, yet they demonstrate a high degree of site fidel- ity for time scales of at least weeks to months (Jagielo, 1990; Smith et ah, 1990; Mathews, 1992; Yamanaka and Richards, 1993; Jagielo, 1999; Starr et ah, 2004). Although lingcod population dynamics have been studied from a fisheries perspective, very little is un- derstood about how this predator affects the structure of fish populations and assemblages on rocky reefs. A previous study of diet and habitat associations of demersal fishes on nearshore reefs along the Oregon Coast revealed that 282 adult lingcod had consumed 27 identifiable species of fish and invertebrates. Of those 134 prey items, no adult rockfishes were found and the contribution to total biomass by all rockfish prey was less than one percent (Steiner, 1979). However, no prior lingcod studies have described diet in relation to prey abundance. In order to assess differential selection, and thus characterize which prey types will most likely be selected, there must be an estimate of prey availability relative to consumption (Manley et al., 2002). The goal of this study was to describe the diet of adult lingcod off the coast of Oregon, to characterize relative patterns of consumption of transient and resident prey species by lingcod, and describe whether or not preference, defined as the differential consumption of one prey type over others in relation to availability, was evident. Specifi- cally, by using lingcod diet and prey abundance esti- mates off the coast of Oregon, I addressed the following questions: 1) Do lingcod prefer particular prey species, and 2) do lingcod preferentially target rockfishes? The answers to these questions were yes and no, respec- tively. This information can be used to more effectively manage a reserve system where both predator and prey populations are the focus of conservation efforts. Materials and methods Study area The nearshore zone off Oregon is generally exposed, has relatively high wave energy, and is influenced by long- shore currents. I sampled lingcod from two nearshore subtidal sites along the coast of Oregon: one south of Newport, referred to as site 1 (44°31'N lat.; 124°08/W long.), and another south of Coos Bay, referred to as site 2 (43°16'N; 124°25'W) (Fig. 1). Both sites comprised high relief rocky reef, rocky flats, cobble, and sand at depths of 20 to 50 m. The reefs varied from small pin- nacles encompassing <10 m2 to large boulder fields and bedrock flats that may exceed one km2 in area. The area of exposed rock changes on temporal scales of months to decades, however, sand transport is greatest during the stormy winter months and relatively stable during the summer (Kulm et al., 1968). Tinus Prey preference of Ophiodon elongatus, a top marine predator 195 Prey availability Prey availability was compared with observed prey consumption to evaluate prey preference. Lingcod are highly generalized visual preda- tors and visual surveys provide an estimate of relative prey density within a visual field. I evaluated prey availability with dive surveys 44°30'0"N in the areas where lingcod were collected for gut analyses (Starr et al., 2010). Dive surveys were conducted from a relatively small boat equipped with standard electronics. Ocean conditions had to be sufficiently benign for both safe boat handling and diver deploy- ment and recovery. Weather conditions were 44°0'0"N a limiting factor for dive surveys. In general, combined seas (wave and swell height) of less than two to three meters and wind velocities of less than 20 knots are necessary. Additionally, fog and strong currents at times prohibited safe dive and boat operation near shallow reefs. A single dive survey consisted of a single 43°30'0"n 100x4 m visual-count transect (Bohnsack, 1996) during daylight between 1000 and 1500 hours. I conducted surveys at site 1 in Janu- ary and June 2004, and in June 2005 (three surveys total), and at site 2 in January and October 2004, and in June (three surveys) and September 2005 (six surveys total) (Table 1). The exact locations of transects were deter- 43 00 N mined haphazardly from the surface by drop- ping a weighted line in an area as close as possible to where fishing for lingcod occurred and where depths were sufficiently shallow so that single dive surveys could be completed within one scuba dive (<35 m). Visibility was variable but was always sufficient to identify fish within two meters of the transect line, and fishes and invertebrates were approachable. I surveyed three basic habitat types within each transect: high- relief rocky reef, boulder mixed with cobble, and broken shell mixed with sand. I quantified the relative abundance of potential prey within the foraging range of lingcod, estimating age groups of rockfishes (year 1, 1-2, 3+) from estimated total lengths. During dives I estimated fish lengths by comparing them against objects of similar shape and color of known lengths at various distances. I observed only adult lingcod on rocky reef habitat. Relative prey availability between sites 1 and 2 were compared by one-way analysis of similarity (ANOSIM; Clark, 1993). The ordination, associated tests, and species accumulation curves were produced with PRIMER analytical software (vers. 6.1.6, PRIMER-E Ltd., Plymouth, U.K.2) by using an included ANOSIM method (Clark and Gorley, 2006). Additionally, a rank concordance test of prey category abundance was 2 Mention of trade names or commercial companies is for identification purposes only and does not imply endorsement by the National Marine Fisheries Service, NOAA. 124°30'0"W 124WW 123°30'0"W 123 WW Figure 1 Map of study region (inset) and study sites (1 and 2) within the Oregon nearshore zone where stomach samples of lingcod (Ophiodon elongatus) were collected and dive surveys of potential prey were conducted. conducted for sites 1 and 2 (Sokal and Rohlf, 1995). For comparisons of two groups, t-tests were used unless a Shapiro-Wilk test for normality, equal variance, or both, failed, in which case a Mann-Whitney rank sum test was used. The Michaelis-Menten equation (MME) was used to generate species accumulation curves to evaluate how quickly the number of new species became asymptotic (curve stability) with additional sampling effort (Willott, 2001; Williams et al., 2007). The beta value for the MME represents the number of samples required to detect 50% of the total number of species, or groups. Dietary composition Multiple anglers using lines with a single hook and attrac- tor on a chartered recreational fishing vessel in July (19 and 17 fish in two sampling trips), August (12 fish), and September (12 fish) of 2003 (one trip each) collected a total of 60 lingcod at site 1. The lingcod collected at Site 196 Fishery Bulletin 1 10(2) 2 were by a commercial fisherman in the months of May (21 fish), June (48 fish), and October (59 fish) of 2004, and May (49 fish), June (45 fish), August (46 fish), and October (40 fish) of 2005 (Table 1). The commercial gear used was a “dingle-bar” — an iron bar trolled just off the seafloor with a set of three rubber jigs with large hooks and an attractor. An additional set of three jigs with hooks was trolled mid-water (about 10-20 m off-bottom). When multiple lingcod hit the jigs, they generally did so simultaneously on both the bottom and mid-water sets. In the first year, lingcod stomachs were labeled, placed in cloth bags, and preserved in ethanol. In subsequent years, stomachs were labeled, wrapped in cheesecloth, packed in ice, and examined within 24 hours. The number and identity of items in each stomach were identified to the lowest possible taxon. When the identification of a prey fish was not possible from external characteristics, I attempted to identify the prey by otoliths or skeletal elements (or both). A second, blind reading of a subsample of otoliths and skeletal elements was done by a recognized expert who confirmed prior determinations. When possible, beaks were used to estimate size and infer species of octopus by comparison with other samples that were identified to species from external characteristics. Observed consumption provides a description of a local prey base if sample sizes are large enough to cap- Table 1 Dates and locations of dive surveys and stomach collec- tions for diet samples of lingcod (Ophiodon elongatus ) in the nearshore zone off the Oregon Coast. Site 1 is located south of Newport, OR, and site 2, south of Coos Bay, OR. An X indicates that data were collected. Data used for prey preference analysis are within-season. Dive survey Diet samples Site Date X 1 16/07/03 X 1 23/07/03 X 1 25/08/03 X 1 24/09/03 X 1 09/01/04 X 2 22/01/04 X 2 24/05/04 X 1 09/06/04 X 2 24/06/04 X 2 05/10/04 X 2 22/10/04 X 2 13/05/05 X 2 03/06/05 X 2 08/06/05 X 1 22/06/05 X 2 26/06/05 X 2 27/06/05 X 2 17/08/05 X 2 28/09/05 X 2 20/10/05 ture the diversity within a population and incorporate representative temporal and spatial scales. Although lingcod are highly generalized, the incidence of new prey types in gut samples was asymptotic with in- creasing sample size. The MME was used to generate a species-accumulation curve and test for sufficiency of sampling effort. A rank concordance test of prey cat- egory abundance was conducted for sites 1 and 2 (Sokal and Rohlf, 1995). For comparisons of two groups, /-tests were used unless a Shapiro-Wilk test for normality or equal variance test (or both) failed, in which case a Mann-Whitney rank sum test was used. Additionally, one-way analysis of variance (ANOVA) was used to test for differences among sampling trips for differences in stomach fullness and for differences in consumption of transient and resident prey types. Stomach content data were analyzed by frequency of occurrence, %Fo - (n-100)/Ns; and percentage of prey, %N = (n'-100)/Afp; where n=the number of stomachs containing a particular prey type, Ns=the total number of lingcod stomachs examined, n'=the total number of individuals of a particular prey type, and Wp=the total number of prey items (Hyslop, 1980). Prey-preference model A preference model describes the relative selection of resources in relation to the availability of those resources. If a particular prey type is selected more or less frequently than would be predicted by relative avail- ability, that prey type is said to be either preferred or avoided relative to other prey types. The general formu- lation of the preference model (Johnson, 1980) is as fol- lows. Let r - be the rank of some measure of consumption of prey component (i) by an individual predator (J) and s(/ be the rank of an observed measure of the availability of prey component (i) to individual predator (j). The indi- vidual differences in these ranks, ttJ = rtJ - sy , are then averaged across animals to indicate the relative prefer- ence of all prey types across all predators, as given in Equation 1 below. The advantage of this nonparametric approach is that information about prey preference can be gleaned from imperfect field data. The use data and availability data are ranked for each animal and even if a particular prey type is not observed, those data can be used in the analysis. If a known prey type was not observed, the availability of that particular prey type would be considered low by comparison with other prey types in the analysis. I = Z(rv-sy)/J- (1) Results Prey availability Observed fish abundance was overwhelmingly dominated (over 90%) by demersal rockfishes ( Sehastes spp.) at both Tinus: Prey preference of Ophiodon elongatus, a top marine predator 197 sites. Aggregations of black rockfish ( S . melanops ) typically exceeded 100 individual adults and were the most common rockfish species. There was no evidence of a group effect between sites 1 and 2 (one-way ANOSIM, global /? = — 0. 115; significance of sample stat. = 64.2%). An MME for species accu- mulation indicated sampling effort was sufficient to achieve a stable asymptotic curve (Smax=17.94; /3= 0.72). In the pooled data, black rockfish were 41.1% of a total of 2640 fish recorded in nine dive surveys. When Sebastes species were aggregated into a single prey category (demersal rockfishes) there was no difference in mean abundance between sites (two-sided (-test, P=0.84, df=7) or of lingcod abundance between sites (Mann-Whitney (7=26.5, P=1.0). There was a mean of 177.7 (stan- dard error [SE ] = 14 . 8 ) demersal rockfish and 4.3 (SE = 0.33) lingcod observed at Site 1 and a mean of 148.8 (SE = 37.8) rockfish and 3.8 (SE = 0.98) lingcod at site 2. A rank concordance test of prey common to both sites was significant with respect to abundance of potential prey species between sites (Kendall’s rank concordance test, PcO.Ol, n = 16, s = 1.91). Striped surfperch (Embiotoca late- ralis) and yellowtail rockfish (S. flavidus ) were recorded only at site 1, whereas canary rockfish (S. pinniger) were observed only at site 2. Besides those species, the sites did not differ with respect to either the presence of potential prey species or relative abundance by genus, with the exception of significantly more sculpins (family Cottidae) at site 1 than site 2 (Mann-Whitney (7=18.0, P=0.3). Geographic ranges of all species in this study are known to overlap both sites. The smallest lingcod sampled or observed on a reef was 42 cm TL. Lingcod may be retained by the fishery at 61 cm TL and larger. Although undersized lingcod were sampled by special permit at both sites, the com- mercially caught samples were biased toward larger lingcod and most of the lingcod sampled were within a relatively narrow size range, likely because the local lingcod population was rebuilding and was dominated demographically by only a few cohorts. Lingcod juve- niles settle onto a variety of habitats but were not ob- served on reefs. This is not surprising because lingcod are periodically cannibalistic, as shown in this and other studies. Dietary composition Of the 60 lingcod stomachs sampled at site 1, 12 were empty and 48 contained prey that were aggregated into 10 categories. At site 2, of the 315 lingcod stomachs sampled, 177 were empty and 138 contained the same 10 prey categories plus Pacific sandlance ( Ammodytes hexapterus) as a major prey item, as well as market squid ( Loligo opalescens ), Pacific lamprey (Lampetra tridentata), and northern anchovy (Engraulis mordax) as minor items (Table 2). Because both the number of samples and sampling effort was much greater at site 2, it was expected that more prey types were found in c 50 a Site 1 empty d Site 1 non-empty ■ Site 2 empty ■ Site 2 non-empty 30 - t L t ■■ 45 50 55 60 65 70 75 80 85 90 95 >95 Lingcod total length (cm) Figure 2 Lingcod (Ophiodon elongatus) total length (cm) and frequency of occurrence for empty and nonempty stomachs sampled at sites 1 and 2 off Oregon, 2003-05. Gear types were: multiple fishermen with single hooks and lines at site 1 (gray), and a commercial fisherman with multiple hooks and lines for site 2 (black). Lingcod under the legal limit for total length (<61 cm) were retained by permit at both locations, but were propor- tionately more abundant among sampled lingcod at site 1, and proportionately more lingcod stomachs were empty at site 2. lingcod from there (see Bock, 1987). There were pro- portionately fewer empty stomachs among captured lingcod, and lingcod were smaller on average at site 1 than at site 2 (Fig. 2). Among prey categories common to both sites, there were significantly more Pacific her- ring (Clupea pallasii) consumed by lingcod sampled at site 1 than at site 2 (two-sided (-test, P=0.01, df=373). A rank concordance test was significant, indicating that prey consumption by category did not differ between sites (Kendall’s rank concordance test, P<0 .01, n = 15, s=2.03) and therefore the data were pooled for the pref- erence analysis. After sites were pooled, there were 21 identified spe- cies aggregated into 14 ecologically similar prey catego- ries. Among the 342 prey items found in 375 stomachs (50.4% of lingcod stomachs were empty) major prey items were Pacific herring, Pacific sandlance, unidenti- fied fishes, two-spotted octopus ( Octopus bimaculatus), and pandalid shrimps ( Pandalus spp.). All other prey groups, including rockfishes, each comprised less than five percent of the total gut contents (Fig. 3). A MME for species accumulation indicated that sampling effort was sufficient (Smax=14.91; /3= 17. 2 1 ). Of the prey items that were measurable to total length, 14 were confirmed to be rockfishes. The largest of those was 28 cm (the on- ly potential adult), and none was estimated to be more than three years old based on published length-at-age curves (Love et al., 2002). Of the identified young-of- year rockfishes, five were of the “black-spot” group and one was a stripetail (S. saxicola). Nearly all rockfishes identified to species were S. melanops and less than two years old as inferred by length (Love et al., 2002). 198 Fishery Bulletin 110(2) Table 2 Prey found in stomachs of 375 lingcod ( Ophiodon elongatus) collected off Oregon, where n is the number of stomachs containing a particular prey type and ri is the total number of individuals of a particular prey type; %F0 is the frequency of occurrence, and %N is the percentage of prey items. The preference rank for each of 14 aggregated prey categories is also provided, where l=most preferred prey and 14 = least preferred prey. Prey species n n %F0 %N Preference rank Transient and pelagic fishes Lampetra tridentata i i 0.27 0.31 8 Engraulis mordax 2 2 0.53 0.62 6 Clupea pallasii 37 109 9.87 33.64 7 Merluccius productus 7 8 1.87 2.47 4 Ammodytes hexapterus 15 49 4 15.12 2 Skates and flatfishes (soft bottom) 3 Raja spp. 1 1 0.27 0.31 Hippoglossus stenolepis 1 1 0.27 0.31 Citharichthys sordidus 3 3 0.8 0.93 Parophrys vetulus 1 1 0.27 0.31 Platichthys stellatus 2 2 0.53 0.62 unidentified flatfishes 5 5 1.33 1.54 Reef-dwelling fishes Rockfishes Sebastes melanops 6 6 1.6 1.85 13,14 Sebastes saxicola 1 1 0.27 0.31 unidentified rockfishes Greenlings 7 9 1.87 2.78 12 Hexagrammos decagrammus 3 3 0.8 0.93 11 Ophiodon elongatus 2 2 0.53 0.62 Unidentified fishes 28 33 7.47 10.19 Sculpins Hemilepidotus hemilepidotus 3 3 0.8 0.93 10 Scorpaenichthys marmoratus 2 2 0.53 0.62 Unidentified sculpins 9 9 2.4 2.78 Invertebrates Octopus Octopus bimaculatus 26 30 6.93 9.26 9 Octopus dofleini 5 5 1.33 1.54 Loligo opalescens 3 3 0.8 0.93 5 Pandalus spp. 23 27 6.13 8.33 1 Cancer magister 2 6 0.53 1.85 The dominant prey type was Pacific herring (%F=9.87, %A = 33.64), a transient and pelagic species. Other prey types were clustered and far less dominant in the di- et (Fig. 4). Among sampling periods, empty stomachs ranged from 8-81% (mean 56% empty, n = 10 sampling periods, SE = 5.7). Among sampling months, May-Oc- tober, the presence of consumed prey among lingcod was unpredictable, regardless of the sampling month (ANOVA, Fx 9=1.77, P=0.22) and consumption of resi- dent prey appeared to be independent of consumption of transient prey (ANOVA, F1 9=2.46, jP=Q.15). There were 41 unidentified prey items, 33 of which were confirmed not to be rockfishes. Lingcod eat parts of animals they cannot swallow whole by tearing prey apart (e.g., Pacific giant octopus; personal observ.) and are thus not considered gape-limited with respect to prey preference. Larger lingcod consumed larger prey (Fig. 5) but not distinctly different prey types (adjusted coefficient of determination r2=0.29, one-way ANOVA F, 71 = 30.3, P < 0.01, n= 12 measurable prey items). Typi- cally, a single prey item (but as many as 17) was found in a stomach containing prey, and among those stom- achs containing more than one prey item, as many as four different species were found. Numerically, 52% of prey were transient and pe- lagic, 4% were associated with soft-bottom seafloors, 44% were demersal reef-dwelling species, and of the latter, half were invertebrates. The importance of macroinvertebrates among local prey species is dif- ferent from what was found in previous studies. Sand Tinus Prey preference of Ophiodon elongatus, a top marine predator 199 Figure 3 Proportional relative availability (dark bars) of potential prey as determined from nine dive surveys off Oregon for sites 1 and 2 combined, and propor- tional consumption (light bars) of prey by lingcod (Ophiodon elongatus) for sites 1 and 2 combined. The lack of overlap between realized consumption and relative availability indicates that lingcod were not preferentially consuming the most apparently abundant prey types, which were rock- fishes ( Sebastes spp.). consistently occurred in lingcod stomachs containing both octopus and shrimps, but never with flatfishes of any species. This pattern suggests that these lingcod did not forage for flatfishes directly over the seafloor, but were eating them in the water column. Because lingcod were captured on mid-water lures, they are apparently capable of foraging in the pelagic as well as the benthic zones. Prey preference Analysis of identified prey in the pooled data showed that prey selection was not proportional to availability (Johnson’s preference, F13 132 = 943, P<<0.001). Rock- fishes were significantly “avoided” among prey categories (Waller-Duncan [1969] multiple comparisons, P =0.01, n = 145). In order of preference, adult rockfishes were ranked last followed by subadult rockfishes (Fig. 6). Preference ranking also indicated that transient and pelagic prey (Pacific herring and Pacific sandlance) were among the most preferred prey. The January surveys could not be temporally matched with consumption data and therefore were excluded from this analysis, as were empty stomachs. However, because of the inherent tem- poral and spatial patchiness of transient prey, as well as the difficulty in comparing very different types of prey, it was not possible to differentiate prey preference ranks among Pacific herring, Pacific sandlance, shrimps, and octopus. Other categories fell between these extremes (Fig. 6). Discussion These data indicate that lingcod off the coast of Oregon 1) are highly generalized predators of both fish and invertebrates in multiple habitats; 2) select prey dis- proportionately to prey abundance; and 3) do not dif- ferentially target rockfish as prey. Rockfishes may not be preferred because, unlike any other identified prey items, they have robust, venomous spines (Smith and Wheeler, 2006). In this case, experimental manipulation of predator and prey densities at meaningful temporal and spatial scales is not possible. For this reason it is necessary to use consumption and relative density estimates in a static model to find evidence of an effect. If consumption is very low relative to prey abundance, as is the case with predation on rockfishes, then any 200 Fishery Bulletin 1 10(2) direct effects on the population dynamics of either are unlikely to be strong. There is incomplete spatial and temporal overlap between prey availability and consumption data sets and the variance may be greater than it otherwise would be because the dive surveys are disjunct. Still, the MME beta value and asymptotic curve stability of the combined surveys suggest that the heterogene- ity of available prey can be detected with this level of effort and that the data are representative at this spatial and temporal scale. The prey availability data are not intended to reflect regional abundance. When a prey type such as Pacific herring is ranked low with respect to availability relative to rockfishes, it suggests rockfishes have more constant (less patchy) temporal and spatial overlap with lingcod. In this way the po- tential for encounter is much higher between lingcod and rockfishes. Large, highly generalized predators eat many differ- ent prey types and often do so infrequently, and there- fore sample sizes must be relatively large to adequately capture the heterogeneity of the consumption data (e.g., Kingsford, 1992). With 375 samples, the dietary data reported here describe the relative abundance of prey categories in the diet of lingcod over a limited geo- graphical area during half the year. However, Steiner (1979) collected summer and winter stomach samples and did not show an increase in lingcod consumption of rockfishes in winter and the number of samples col- lected appears to have captured the heterogeneity in the consumption data. The primary sources of error in these data include potential misidentification of prey and undefined rates of egesting stomach contents. Ad- ditionally, the digestion rates for free-living lingcod are unknown. Although they do reflect the relative temporal distributions of different prey types, the data from dive surveys were biased by both the spatial and temporal patchiness of transient prey, and by asymmetric sam- pling accuracy among habitats for prey types that were difficult to observe. However, rockfishes are highly ob- servable and there was clearly a strong negative prefer- ence (or avoidance) for rockfishes than for all other prey types. Hydro-acoustic tracking studies of black rockfish have shown they move less than a few hundred meters over periods of months (Parker et al., 2008). The gape-limitation hypothesis predicts that prey- size selection is consistent with optimal diet theory at the lower bound and the physical constraint of mouth size at the upper bound (Schmitt and Holbrook, 1984) and can be useful for predicting foraging behavior in fish (e.g., Persson et al., 1996). Larger lingcod tend to consume larger prey, but the gape-limitation hypothesis, or size-spectrum hypothesis (Scott and Murdoch, 1983), is not particularly useful for predicting prey selection in these animals because all sizes of lingcod eat small prey and lingcod consume parts of larger prey. Gape- limitation does not effectively predict which prey species or functional groups adult lingcod of different sizes will prefer to consume, nor do these data show a distinct shift to larger prey with increasing lingcod size. In relatively long-lived generalist predators such as lingcod, dietary sampling at temporal scales over two years may be required for meaningful patterns in consumption to emerge. The variance in consumption by local predators of transient prey is high and may be independent of regional prey abundance. If consumption of resident prey is relatively even over time, the resident prey types may provide a maintenance resource and more ephemeral prey may provide sporadic opportuni- ties for enhanced growth and reproduction. Addi- tionally, indirect effects can be important to the distribution of predators. Besides direct consump- tion, risk effects (modification of prey distribution or behavior because of a perceived predation risk) may have an important influence on community structure (Creel and Christianson, 2007; Madin et al., 2010). There is concern that lingcod predation may re- duce the efficacy of marine reserves in the recov- ery of some overfished populations of rockfishes. In a recent study that addressed this issue in Puget Sound, Washington, Beaudreau and Essing- ton (2007) found that in 560 lingcod (<30-108 cm TL) sampled inside and outside marine reserves, 6.8% of the total number of prey items were rock- fishes. All individual rockfish identified to species were Puget Sound rockfish (S. emphaeus ) and 0.4% of all prey were confirmed to be other spe- cies of Sebastes. The Puget Sound rockfish is a %Fn Figure 4 Percentage of lingcod ( Ophiodon elongatus) prey (%N) as a function of frequency of occurrence in the diet (%F0). Prey categories are as follows: TP=transient-pelagic fishes (pre- dominantly Pacific herring [Clupea pallasii]); SF=skates ( Raja spp.) and flatfishes; S = sculpins (family Cottidae); R=rockfishes ( Sebastes spp.); G=greenling (family HexagrammidaeXinclud- ing cannibalism by lingcod); 1= invertebrates; 0 = other (includ- ing uncategorized, unidentified fishes). TP, for example, was both a relatively large percentage of the overall diet, and also commonly occurred as a prey type among lingcod sampled. Tinus Prey preference of Ophiodon elongatus, a top marine predator 201 very small species that matures in 1-2 years. It is a schooling species and is often found in high densities. It is not fished either recreationally or commercially and thus is not the focus of recovery efforts. The largest measurable rockfish in Beau- dreau and Essington’s (2007) study was 16.6 cm. Combined with the results from Steiner (1979) and this study, lingcod of any size rarely prey on larger-body roekfishes. Beaudreau and Essington (2007) state that model results suggest intensive lingcod fishing is likely to disproportionately al- leviate predation pressure on larger roekfishes. However, combined empirical evidence from this study and the two studies cited immediately above does not support this assertion. Of all prey items found in this study, only one was a potentially reproductive rockfish and it apparently had been ingested within 24 hours of capture. This ratio simplifies to less than one adult rockfish consumed per adult lingcod per year, whereas the dive surveys revealed an aver- age of 40 adult roekfishes living in the vicinity of each lingcod. If these ratios are representa- tive, they suggest that lingcod predation is not a primary source of mortality for nearshore adult roekfishes off the coast of Oregon. Nor do lingcod appear to be a primary source of mortality of juvenile or young-of-the-year roekfishes because they were only slightly more likely than adult roekfishes to be eaten by lingcod. Hobson et al. (2001) found predation by black rockfish, blue rockfish, and kelp greenling was the primary source of mortality for postsettlement juvenile roekfishes in northern California. 50 40 • 40 50 60 70 80 90 100 Lingcod length (cm) Figure 5 Lingcod ( Ophiodon elongatus) total length (cm) versus prey length for each of 73 measurable prey items. The largest prey item was a Pacific giant octopus (Octopus dofleini) estimated at 70 cm in length that two individual lingcod, 72 cm and 93 cm respectively, appeared to have each eaten half. This prey item is represented by two data points, each with an assigned prey length value of 35 cm. The slope of the regression line (adj. R2 = 0.29) is influenced by the three largest prey items. Dashed lines represent the 95% confidence interval. Conclusions The results of this study show that lingcod are highly generalized predators that consume a broad variety of prey in terms of taxa, body form, and habitat. Lingcod are mobile, opportunistic, ambush predators that do not appear to be individually specialized. On the basis of the number of empty stomachs, they frequently go at least several days without eating, indicating there may be large differences between local prey abundance and prey availability (see Menge, 1972; Kelly, 1996). Better information is required on foraging range in relation to differences in habitat and prey availability to better understand lingcod foraging behavior as it relates to prey density. Nevertheless, this study strongly indicates that lingcod do not pose a threat to rockfish populations. EBFM requires more information than single-species management approaches. In data poor systems, and particularly those that are difficult to access, higher echelon data describing interactions among both tar- geted and nontargeted species will be very difficult to develop. However, this study shows that untested as- sumptions about trophic relationships may lead to coun- terproductive management decisions, particularly with respect to large predatory species (Baum and Worm, 2009). Marine reserves can be an effective manage- ment tool for the conservation and recovery of exploited and other species, and particularly so where species of particular interest have relatively site-attached adult populations. In these cases trophic relationships, es- pecially among resident and transient species, are a critical uncertainty and these relationships can only be fully understood through both consumption and relative prey availability measures. In this case, a preference index provides much more information about the likely result of fluctuations in predator and prey populations than would be the case with diet data alone. Acknowledgments This study was funded in part by a Cooperative Institute for Marine Resource Studies (CIMRS) grant. I thank R. Cross, D. Edge, M. Hixon, and J. Miller for their helpful suggestions in the preparation of this manu- script; S. Reimers for help with sample identification; B. Gallagher, P. Heikkila, S. Heppell, K. Schultz, and S. Theberge for help with sample collection; E. Gilbert 202 Fishery Bulletin 1 10(2) 6 1 5 - O ▲ ▲ o O Other resident reef dwelling A Transient □ Subadult rockfishes ■ Adult rockfishes O o o □ -1 1 2 3 4 5 6 7 9 10 11 12 13 14 Prey category rank Figure 6 Mean relative prey “preference” (Johnson’s [19801 - t) of 186 lingcod ( Ophiodon elongatus ) captured at two sites off Oregon containing prey from 14 prey categories, listed from most preferred (rank 1) to least preferred (rank 14). Values with the same underline are not significantly different (P>0.05, Waller-Duncan multiple comparison). Rank numbers identify prey categories as listed in Table 1. Both juvenile and adult rockfishes ( Sebastes spp.) were significantly avoided as prey types in comparison with all other prey groups. for help with Figure 1; three anonymous reviewers; and Captain J. W. Cheesman (1962-2011). Literature cited Anderson, W. B., and G. A. Polis. 1998. Marine subsidies of island communities in the Gulf of California: evidence from stable carbon and nitrogen isotopes. Oikos 81:75-80. Ault, J. S., S. G. Smith, J. A. Bohnsack, J. Luo, D. E. Harper, and D. B. McClellan. 2006. Building sustainable fisheries in Florida’s coral reef ecosystem: positive signs in the Dry Tortugas. Bull. Mar. Sci. 73:633-654. Baum, J. K., and B. Worm. 2009. Cascading top-down effects of changing oceanic predator abundances. J. Anim. Ecol. 78:699-714. Beaudreau, A. H., and T. E. Essington. 2007. Spatial, temporal, and ontogenetic patterns of predation on rockfishes by lingcod. Trans. Am. Fish. Soc. 136:1438-1452. 2009. Development of a new field-based approach for estimating consumption rates of fishes and comparison with a bioenergetics model for lingcod ( Ophiodon elon- gatus). Can. J. Fish. Aquat. Sci. 66:565-578. Bock, C. E. 1987. Distribution-abundance relationships of some Ari- zona landbirds: A matter of scale? Ecology 68:124-129. Bohnsack, J. A. 1996. Two visually based methods for monitoring coral reef fishes. In Coral reef symposium on practical, reli- able, low cost monitoring methods for assessing the biota and habitat conditions of coral reefs, January 26-27, 1995 (M. P. Crosby, G. R. Gibson, and K. W. Potts, eds.), p. 31-36. Office of Ocean and Coastal Resource Management, NOAA, Silver Spring, MD. Bruno, J. F., and M. I. O’Connor. 2005. Cascading effects of predatory diversity and omnivory in a marine food web. Ecol. Lett. 8:1048-1056. Chaneton, E. J., and M. B. Bonsall. 2000. Enemy-mediated apparent competition: empirical patterns and the evidence. Oikos 88:380-394. Clarke, K. R. 1993. Non-parametric multivariate analyses of changes in community structure. Aust. J. Ecol. 18:117-143. Clarke, K. R., and R. N. Gorley. 2006. Primer, vers. 6: User manual/tutorial. PRIMER- E, Plymouth, U.K. Coleman, F. C., and C. C. Koenig. 2010. The effects of fishing, climate change, and other anthropogenic disturbances on red grouper and other reef fishes in the Gulf of Mexico. Integ. Comp. Biol. 50:201-212. Cowan Jr., J. H., C. B. Grimes, W. F. Patterson III, C. J. Walters, A. C. Jones, W. J. Lindberg, D. J. Shelby, W. E. Pine III, J. E. Powers, M. D. Campbell, K. C. Lindeman, S. L. Diamond, R. Hilborn, H. T. Gibson, and K. A. Rose. 2010. Red snapper management in the Gulf of Mexico: sci- ence- or faith-based? Rev. Fish. Biol. Fish. 21:187-204. Creel, S., and D. Christianson. 2007. Relationships between direct predation and risk effects. Trends Ecol. Evol. 23:194-201. Tinus Prey preference of Ophiodon e/ongatus, a top marine predator 203 Dill, L. M., M. R. Heithaus, and C. J. Walters. 2003. Behaviorally mediated indirect interactions in marine communities and their conservation implica- tions. Ecology 84:1151-1 157. Dulvy, N. K., Y. Sadovy, and J. D. Reynolds. 2003. Extinction vulnerability in marine popula- tions. Fish Fish. 4:25-64. Francis, R. C., M. A. Hixon, M. E. Clarke, S. A. Murawski, and S. Ralston. 2007. Ten commandments for ecosystem-based fisheries scientists. Fisheries 32:217-233. Frid, A., G. G. Baker, and L. M. Dill. 2008. Do shark declines create fear-release systems? Oikos 117:191-201. Gaichas, S. K., K. Y. Aydin, and R. C. Francis. 2010. Using food web model results to inform stock assessment estimates of mortality and production for ecosystem-based fisheries management. Can. J. Fish. Aquat. Sci. 67:1490-1506. Graham, N. A. J., R. D. Evans, and G. R. Russ. 2003. The effects of marine reserve protection on the trophic relationships of reef fishes on the Great Barrier Reef. Environ. Cons. 30:200-208. Heithaus, M. R., A. Frid, A. J. Wirsing, and B. Worm. 2008. Predicting ecological consequences of marine top predator declines. Trends Ecol. Evol. 23:202-210. Hobson, E. S., J. R. Chess, and D. Howard. 2001. Interannual variation in predation on first-year Sebastes spp. by three northern California preda- tors. Fish. Bull. 99:292-302. Holt, R. D. 1977. Predation, apparent competition, and the structure of prey communities. Theor. Popul. Biol. 12:197-229. Hutchings, J. A., and J. K. Baum. 2005. Measuring marine fish biodiversity: temporal changes in abundance, life history and demogra- phy. Phil. Trans. R. Soc. B. 360:315-338. Hyslop, E. J. 1980. Stomach content analysis - a review of methods and their applications. J. Fish Biol. 17:411-429. Jagielo, T. H. 1990. Movement of tagged lingcod Ophiodon elongatus at Neah Bay, Washington. Fish. Bull. 88:815-820. 1999. Movement, mortality, and size selectivity of sport and trawl caught lingcod off Washington. Trans. Am. Fish. Soc. 128:31-48. Johannes, R. E. 1998. The case for data-less marine resource manage- ment: examples from tropical nearshore finfisher- ies. Trends Ecol. Evol. 13:243-246. Johnson, D. H. 1980. The comparison of usage and availability mea- surements for evaluating resource preference. Ecology 61:65-71. Kelly, J. F. 1996. Effects of substrate on prey use by belted king- fishers ( Ceryle alcyon ): a test of the prey abundance- availability assumption. Can. J. Zool. 74:693-697. Kingsford, M. J. 1992. Spatial and temporal variation in predation on reef fishes by coral trout ( Plectropomus leopardus , Serani- dae). Coral Reefs 11:193-198. Kulm, L. D., K. F. Scheidegger, J. V. Byrne, and J. J. Spigai. 1968. A preliminary investigation of the heavy mineral sites of the coastal rivers and beaches of Oregon and Northern California. The Ore Bin 30:165-180. Link, J. S., and L. P. Garrison. 2002. Trophic ecology of Atlantic cod Gadus morhua on the northeast US continental shelf. Mar. Ecol. Prog. Ser. 227:109-123. Love, M. S., M. Yoklavich, and L. Thorsteinson. 2002. The rockfishes of the northeast Pacific, 414 p. Univ. California Press, Berkeley, CA. Ludwig, D., R. Hilborn, and C. Walters. 1993. Uncertainty, resource exploitation, and conserva- tion: lessons form history. Science 260:17-18. Madin, E. M. P, S. D. Gains, and R. R. Warner. 2010. Field evidence for pervasive indirect effects of fishing on prey foraging behavior. Ecology 91:3563- 3571. Manly, B. F. T., L. L. McDonald, D. L. Thomas, T L. McDonald, and W. P. Erickson. 2002. Resource selection by animals: statistical design and analysis for field studies, 2nd ed., 223 p. Kluwer Academic Pubis., Dordrecht, The Netherlands. Martell, S. J. D„ C. J. Walters, and S. S. Wallace. 2000. The use of marine protected areas for conserva- tion of lingcod ( Ophiodon elongatus). Bull. Mar. Sci. 66:729-743. Mathews, K. R. 1992. A telemetric study of the home ranges and homing routes of lingcod Ophiodon elongatus on shallow rocky reefs off Vancouver Island, British Columbia. Fish. Bull. 90:784-790. McClanahan, T. R., and R. Arthur. 2001. The effect of marine reserves and habitat on popu- lations of east African coral reef fishes. Ecol. Appl. 11:559-569. Menge, B. A. 1972. Foraging strategy of a starfish in relation to actual prey availability and environmental predictability. Ecol. Mon. 42:25-50. Myers, R. A., and B. Worm. 2005. Extinction, survival, or recovery of large preda- tory fishes. Phil. Trans. R. Soc. Lond. B. 360:13- 20. Paolisso, M. 2002. Blue crabs and controversy on the Chesapeake Bay: a cultural model for understanding waterman’s reasoning about blue crab management. Human Orga- nization 61:226-239. Parker, S. J., J. M. Olson, P. S. Rankin, and J. S. Malvitch. 2008. Patterns in vertical movements of black rockfish Sebastes melanops. Aquat. Biol. 2:57-65. Persson, L., J. Andersson, E. Wahlstrom, and P. Eklov. 1996. Size-specific interactions in lake systems: preda- tor gape limitation and prey growth rate and mortal- ity. Ecology 77:900-91 1. Pikitch, E., C. Santora, E. Babcock, A. Bakun, R. Bonfil, D. Conover, P. Dayton, P. Doukakis, D. Fluharty, B. Heneman, E. Houde, J. Link, P. Livingston, M. Mangel, M. McAllister, J. Pope, and K. Sainsbury. 2004. Ecosystem-based fishery management. Science 305:346-347. Samhouri, J. F., P. S. Levin, and C. H. Ainsworth. 2010. Identifying thresholds for ecosystem-based man- agement. Plos One 5:e8907. Schmitt, R. J., and S. J. Holbrook. 1984. Gape-limitation, foraging tactics and prey-size selectivity of two microcarnivorous species of fish. Oeco- logia 63:6-12. 204 Fishery Bulletin 1 10(2) Scott, M. A., and W. W. Murdoch. 1983. Selective predation by the backswimmer, Noto- necta. Limnol. Oceanogr. 28:352-366. Smith, B. D., G. A. McFarlane, and A. J. Cass. 1990. Movements and mortality of tagged male and female lingcod in the Strait of Georgia, British Colum- bia. Trans. Am. Fish. Soc. 119:813-824. Smith, W. L., and W. C. Wheeler. 2006. Venom evolution widespread in fishes: A phy- logenetic road map for the bioprospecting of piscine venoms. J. Heredity 97:206-217. Sokal, R. R., and F. J. Rohlf. 1995. Biometry: the principles and practice of statistics in biological research, p. 593. Freeman, New York. Starr, R. M., M. Carr, D. Malone, A. Greenley, and S. McMillan. 2010. Complimentary sampling methods to inform eco- system-based management of nearshore fisheries. Mar. Coastal Fisheries 2:159-179. Starr, R. M., V. O’Connell, and S. Ralston. 2004. Movements of lingcod ( Ophiodon elongatus ) in southeast Alaska: potential for increased conservation and yield from marine reserves. Can. J. Fish. Aquat. Sci. 61:1083-1091. Steiner, R. G. 1979. Food habits and species composition of neritic reef fishes off Depot Bay, Oregon. M.S. thesis, 59 p. Oregon State Univ., Corvallis, OR. Thompson, R. M., M. Hemberg, B. M. Starzomski, and J. B. Shurin. 2007. Trophic levels and trophic tangles: the prevalence of omnivory in real food webs. Ecology 88:612-617. Waller, R. A., and D. B. Duncan. 1969. A Bayes rule for the symmetric multiple compari- sons problem. J. Am. Stat. Assoc. 64:1484-1503. Walter III, J. F., A. S. Overton, K. H. Ferry, and M. E. Mather. 2003. Atlantic coast feeding habits of striped bass: a synthesis supporting a coast-wide understanding of trophic biology. Fish. Manag. Ecol. 10:349-360. Wells, D. R. J., J. H. Cowan Jr., and B. Fry. 2008. Feeding ecology of red snapper Lutjanus campecha- nus in the northern Gulf of Mexico. Mar. Ecol. Prog. Serv. 361:213-225. Williams, V. L., E. T. F. Witkowski, and K. Balkwill. 2007. The use of incidence-based species richness esti- mators, species accumulation curves and similarity measures to appraise ethnobotanical inventories from South Africa. Biodivers. Conserv. 16:2495. Willott, S. J. 2001. Species accumulation curves and the measure of sampling effort. J. Appl. Ecol. 38:484-486. Yamanaka, K. L., L. J. Richards. 1993. Movements of transplanted lingcod (Ophiodon elongatus ) determined by ultrasonic telemetry. Fish. Bull. 91:582-587. 205 Variations in eastern North Pacific demersal fish biomass based on the U.S. west coast groundfish bottom trawl survey (2003-2010) Email address for contact author: aimee.kellertS) noaa.gov National Oceanic and Atmospheric Administration National Marine Fisheries Service Northwest Fisheries Science Center Fishery Resource Analysis and Monitoring Division 2725 Montlake Blvd E. Seattle, Washington 98112 Abstract-— In response to declining biomass of Northeast Pacific ground- fish in the late 1990s and to improve the scientific basis for management of the fishery, the Northwest Fish- eries Science Center standardized and enhanced their annual bottom trawl survey in 2003. The survey was expanded to include the entire area along the U.S. west coast at depths of 55-1280 m. Coast-wide biomass and species richness significantly decreased during the first eight years (2003-10) of this fishery-independent survey. We observed an overall ten- dency toward declining biomass for 62 dominant taxa combined (fishery target and nontarget species) and four of seven subgroups (including cartilaginous fish, flatfishes, shelf rockfishes, and other shelf species), despite increasing or variable biomass trends in individual species. These decreases occurred during a period of reduced catch for groundfish along the shelf and upper slope regions relative to historical rates. We used informa- tion from multiple stock assessments to aggregate species into three groups: 1) with strong recruitment, 2) without strong recruitment in 1999, and 3) with unknown recruitment level. For each group, we evaluated whether declining biomass was pri- marily related to depletion (using year as a proxy) or environmental factors (i.e., variation in the Pacific Decadal Oscillation). According to Akaike’s information criterion, changes in aggregate biomass for species with strong recruitment were more closely related to year, whereas those with no strong recruitment were more closely related to climate. The significant decline in biomass for species with- out strong recruitment confirms that factors other than depletion of the exceptional 1999 year class may be responsible for the observed decrease in biomass along the U.S. west coast. Manuscript submitted 3 June 2011. Manuscript accepted 29 November 2011. Fish. Bull. 110:205-222 (2012). The views and opinions expressed or implied in this article are those of the author (or authors) and do not necessarily reflect the position of the National Marine Fisheries Service, NOAA. Aimee A. Keller (contact author) John R. Wallace Beth H. Horness Owen S. Hamel Ian J. Stewart Over the past 20 years, a number of changes have occurred in the North- east Pacific groundfish fishery with low abundance observed for multiple species (Field and Fox, 2006; Levins et ah, 2006). Historically catch and annual exploitation rates increased from the 1950s through the 1980s and then declined rapidly. Between 1999 and 2002, nine important fish stocks in the eastern North Pacific off the U.S. west coast were declared overfished, at which time the Pacific Fishery Management Council (PFMC) introduced a series of regulatory mea- sures to reduce fishing pressure. Man- agement actions included reducing total allowable catch and fleet size, and closure of large areas of the upper continental shelf to fishing (PFMC, 2008a). In response to manage- ment concerns the NOAA Northwest Fisheries Science Center (NWFSC) also expanded and standardized the annual west coast groundfish bottom trawl survey to provide enhanced scientific information for managers. Since 2003, the Northwest Fisheries Science Center (NWFSC) has con- ducted a comprehensive fishery-inde- pendent bottom trawl survey covering the entire coast from the U.S. -Canada to the U.S.-Mexico borders, at depths of 55 to 1280 m (Keller et ah, 2008). This groundfish survey follows strict sampling protocols with stan- dardization of vessels, fishing gear, deployment methods, catch sampling practices, and geographic extent from 2003 onward (Stauffer, 2004). Here we summarize variations in biomass indices for species collected during the 2003-10 fisheries-inde- pendent west coast groundfish bot- tom trawl survey. We evaluate if an observed decline in biomass of demer- sal fish (target and nontarget species) from 2003 through 2010 can be at- tributed primarily to recruitment (i.e., depletion after strong recruitment events for multiple species in the late 1990s) or to climate variability (i.e., poor environmental conditions). The 2003-10 survey time series covers a period within the Califor- nia Current system characterized by 1) reduced catch and exploitation of groundfish species relative to histori- cal rates (Worm et ah, 2009; Hilborn et ah, in press); 2) the population ef- fects of a very strong 1998-99 year class observed for many west coast groundfish species (e.g.. Pacific hake [Merluccius productus ], English sole [Parophrys vetulus], http://www. pcouncil.org/. accessed June 2011); and 3) a phase shift in the Pacific Decadal Oscillation (PDO), an El Ni- no-like pattern of Pacific climate vari- ability linked to productivity (Mantua et ah, 1997). The PDO is detected as warm or cool surface waters in the western Pacific Ocean, north of 20°N, that shift phases on a scale of about 206 Fishery Bulletin 110(2) 10 to 30 years. During a “warm” or “positive” phase, part of the eastern Pacific Ocean warms and productiv- ity of waters off the U.S. west coast declines; during a “cool” or “negative” phase, the opposite pattern occurs (Schwing et ah, 2009). The aim of this study was to evaluate the importance of depletion after strong recruitment versus environ- mental effects on declining biomass observed during groundfish surveys in the western U.S. shelf system. We used data from 24 stock assessments conducted since 2005 (http://www.pcouncil.org/. accessed Septem- ber 2011). With information contained in the assess- ments we separated 62 dominant species into three groups: those with strong recruitment during the late 1990s-early 2000s, those without a strong recruitment during this period, and those with unknown year-class strength. For each group and the overall biomass indi- ces for all groups we evaluated regression models be- tween demersal fish biomass (2003 through 2010) along the U.S. west coast versus year (as a proxy for gradual depletion after recruitment of exceptional year classes to the fishery) and the PDO index, an ecosystem-level indicator of climate variability. For each comparison, the most appropriate model for describing the relation- ship with biomass was determined. A similar analysis was undertaken for species richness. We additionally present information on frequency of occurrence (number of positive hauls) and depth distribution by species. Materials and methods Survey design and methods The NWFSC conducted annual bottom trawl surveys of groundfish resources off the U.S. West Coast using stan- dardized procedures from 2003 through 2010 (Keller et ah, 2008). Surveys occurred May through October from the area off Cape Flattery, Washington (lat. 48°10'N), to the U.S. -Mexico border (lat. 32°30'N) at depths of 55-1280 m (Fig. 1). The entire geographic extent of the survey was covered twice each year by two west coast commercial fishing vessels (20 to 28 m in length) per pass. Each year sampling extended from late May through late July for the first period and mid-August through late October for the second. A stratified random sampling design was used, in which the surveyed region was subdivided into -13,000 cells of equal area (1.5 nmi longitude by 2.0 nmi latitude) (Fig. 1). An average of 700 primary cells was randomly selected each year, strati- fied by geographic location and depth. The geographic allocation was based on a simple north-south division at 34°30'N lat. (Point Conception, California) with 80% of the effort in the northern portion of the survey and 20% in the southern range. The survey area was further stratified into depth zones as follows: north of Point Conception, 40% of the cells were in the shallow depth zone (55-183 m), 30% at mid-depths (184-549 m), and 30% in the deep stratum (550-1280 m); and south of Point Conception, 25% were in the shallow depth zone, 45% at mid-depth, and 30% in the deep stratum. Four chartered west coast fishing vessels were assigned an equal portion of stations to sample per year except in 2004 when only three vessels were used. Vessels were equipped with customized Aberdeen- style nets with a small mesh (3.8 cm stretched measure) liner in the codend, a 25.9-m headrope, and a 31.7-m foot rope. All fishing operations were conducted in strict compliance with national and regional protocols de- tailed in Stauffer (2004). Simrad Integrated Trawl In- strumentation (ITI , Kongsberg Simrad Mesotech Ltd., Port Coquitlam, B.C., Canada1) was used to monitor and record net performance and position for each haul. A differential global positioning system (DGPS) naviga- tion unit (Northstar 500, Northstar Technologies, Ac- ton, MA) was used to monitor towing speed during each haul. Standard survey haul positions were estimated from DGPS data — generally the mid-point between the net touchdown and net liftoff positions. Average net speed over ground and distance fished were calculated from the position data for the trawl and actual bottom time (Keller et ah, 2008). Samples were collected by trawling within the ran- domly selected cells (Fig. 1) for a target fishing time of 15 minutes at a target speed of 1.13 m sec 1 (2.2 knots). All fish and invertebrates were sorted to species (or the lowest possible taxon), and then weighed by using an electronic, motion-compensated scale (Marel, Reykjavik, Iceland). Abundance was not analyzed in this study be- cause not all individuals were counted. Total abundance is estimated from biomass and the two cannot be con- sidered independent without analysis of the variability of mean weights. That analysis is beyond the scope of the present study. Near bottom temperature (°C) and depth (m) were measured during each trawl with an SBE 39 temperature and pressure recorder (Sea-Bird Electronics, Inc., Bellevue, WA) attached to the head rope. Mean tow depths were computed as the average of all depth recordings from the center 80% of the trawl duration (net touch down to lift off). Only tows judged to be acceptable (based on postcollection analysis of bottom contact, net performance, and other metrics; Stauffer, 2004) were included in the data analyses. Analyses of catch To limit this analysis to the most reliably sampled species, we initially examined catch for 310 individual fish species summed over the 2003-10 period. When graphed by species in order of descending catch, no obvious break was apparent and therefore we included all demersal species with an overall catch greater than 450 kg. This break point included the 62 most abundant species in the survey and incorporated 45 of the demersal groundfish species present in the Pacific Fishery Management Council Pacific Coast groundfish 1 Mention of trade names or commercial companies is for identification purposes only and does not imply endorsement by the National Marine Fisheries Service, NOAA. Keller et al : Variations in eastern North Pacific demersal fish biomass, 2003-10 207 Newport Cape Flattery f>art Franciscc Morro Bay 0 25 50 Pt Conception Los Angeles Newport Bead- NWFSC West Coast Groundfish Survey (2003-2010) | 1-5 samples 7“W Figure 1 Geographical extent of the Northwest Fisheries Science Center’s West Coast Groundfish Bottom Trawl Survey and the location of stations (shaded) trawled one to five times from 2003 through 2010. (A) Stations off Washington and Oregon. (B) Stations off northern California. (C) Southern California stations. The Cowcod ( Sebastes levis ) Conservation Area in Southern California was excluded from the experimental design because it is closed to fishing for groundfish species. fishery management plan (PFMC, 2008b), as well as two important benthic invertebrates and 15 non-fished species. The break point also represented a greater than 10% difference in catch relative to the next most abundant species. Species-specific catch per unit of effort (CPUE, kg ha-1) was calculated for each tow on the basis of area swept. Area swept was computed from the mean net width for each tow multiplied by the distance fished. Mean CPUE was calculated for each stratum (depth and geographic region, including those with zero catch), by species. Species-specific biomass indices (6, in metric tons [t] ) were computed by multi- plying the mean CPUE by the appropriate stratum area and then summing the strata biomasses. The estimate of the variance of the biomass was the sum of the vari- ances of the strata involved: Var(b) = X ^(VariCPUE, ) x A?, with n equal to the number of strata, and A equal to the area of each stratum (km2). Standard errors (SE) for the annual species-specific biomass indices were calculated using standard statistical techniques (Cochran, 1977). To examine variation in biomass over time, species groups were initially designated on the basis of either taxonomy or depth (e.g., shelf and slope rockfishes). Subsequently we examined relationships between bio- mass and both year and the PDO index using groups designated by the presence, absence, or unknown oc- currence of a strong recruitment during the mid to late 1990s and early 2000s. In all cases biomass was summed over the appropriate group and variance was calculated as previously described. For those species included in the 2003-10 biomass analyses, we also cal- culated mean depth (m) by averaging station values weighted by catch for each species in each year (Hsieh et al., 2008). 208 Fishery Bulletin 1 10(2) Recruitment versus PDO indices Stock assessments for 24 of the 62 species in the analyses have been published by the PFMC since 2005 (available at http://www.pcouncil.org/groundfish/stock-assess- ments/safe-documents/2011-safe-document/, accessed September 2011) and provide information on the number of recruits by year. We examined stock assessments for each species to determine whether strong recruitment events occurred during the period from the mid to late 1990s through 2002. Annual recruitment strength is generally modeled in the assessments as random devia- tions about a stock- recruitment (S-R) relationship. These deviations and the central tendency of the S-R curve are informed by all other sources of available information (i.e., observed lengths, weights, age, and trend informa- tion from fishery-dependent and independent sources) and will reflect predation intensity, climate, and other influences (Methot, 2011). For this analysis, we defined strong recruitment as 1.7-5 times greater than the average recruitment during the 10 to 14 years before the most recent assessment. We subsequently subdivided the 62 species included in our study into three groups: those with strong re- cruitment during the late 1990s, those without strong recruitment during this period, and those with un- known recruitment levels. We summed the biomass indices for all species within each group and the overall biomass indices for all three groups and regressed these summed values versus year. We reasoned that declining biomass indices would be more tightly tied to time for those species with elevated recruitment as the resulting exceptionally strong cohorts declined due to natural and fishing-induced mortality in the early 2000s. Biomass indices for the aggregated subgroups and overall were also compared with the PDO, a widely used index of climate variability for the California Cur- rent system. The PDO is an index based on patterns of variation in sea surface temperature of the North Pacific from 1900 to the present (Mantua et al., 1997; Schwing et ah, 2009). Although derived from sea sur- face temperature data, the PDO index is well correlated with other environment factors, including sea level pres- sure, winter air temperature, wind shear, and precipi- tation, as well as other Pacific climate indices (ENSO [El Nino-Southern Oscillation] and MEI [multivariate ENSO index]). For comparison with the annual survey data, monthly PDO values were averaged annually (November to October) to include the survey period each year (Mantua2). Species richness Coast-wide estimates of species richness were calcu- lated as area-weighted mean number of fish species taken per trawl sample. Estimates were stratified by survey year (2003-10), depth (55-183 m, 184-549 m. 2 Mantua, N. 2010. Personal commun. Dep. Atmospheric Sciences, Univ. Washington, Seattle, WA 98195. and 550-1280 m), and geographic region (one degree latitudinal increments from 32° to 49°N) for all fish spe- cies. Estimates were built upon the number of distinct fish species reported for each trawl sample. Mean species counts were determined for each stratum and weighted by the proportion of stratum area within the total area. Annual species richness estimates were computed as the sum of these area-weighted species counts within the area of interest (per depth range or coast-wide) and survey year. Species richness variance within each area was similarly estimated as the sum of stratum vari- ances weighted by their associated squared proportion of stratum area within the total area. Standard errors of the mean were computed as the square root of the ratio of the variance estimate to the stratum count for each area (i.e., within a specific depth stratum or coast-wide). We compared species richness over time by regressing against year and also evaluated the relationship between species richness and the annual Pacific Decadal Oscilla- tion (PDO) index. In both cases, regression analyses by depth strata and overall depth were undertaken Statistical analyses For the biomass data we examined individual species, and present for comparison several aggregate groups formed by summing species coast-wide biomass indices (metric tons, t). For each species, regression analy- sis was used to initially investigate the relationship between annual biomass indices and year. To account for the large number of tests conducted, a sequential Bonferroni correction with a significance level of 0.05 was applied to the data (Peres-Neto, 1999). Grouping data for later analyses (initially by depth or taxonomic group to examine trends over time for aggregated data and subsequently by the presence, absence, or unknown occurrence of exceptionally large year classes after recruitment) resulted in fewer tests and no Bonferroni correction was applied. Results for biomass and species richness were statistically compared with year and the PDO index by linear and multiple regression (GLM) by using SAS for Windows (SAS Institute, Inc., Cary, North Carolina). To stabilize the variance, the natural logarithm of the response variable was used in the regression models; however, even after the transforma- tion, annual variance estimates were highly variable for some species. Regressions weighted by the variance estimate of the annual values were therefore used to examine interactions between annual biomass indices and species richness versus year, PDO values, or both (Draper and Smith, 1981). The Akaike information criterion (AIC) was used to choose between competing models (i.e., recruitment, environmental variability or both) when comparing biomass values, summed by groups, versus year and the PDO index (Sakamoto et al., 1986). For each group, the best model was selected on the basis of the smallest AIC value ( AlCmin ). A similar comparison was done between species richness versus year, the PDO index, and both year and the PDO index. To determine whether a model Keller et al. : Variations in eastern North Pacific demersal fish biomass, 2003-10 209 other than the best model was plausible, the difference in AIC values for each model was calculated as A=A 1C r A ICm m. Models with At < 2 are considered equivalent to the best model ( AICmin ) and candidate models with At >10 are highly unlikely to be plausible alternatives for the best model. Candidate models with Al between 3 and 7 have less support than the best model (Burnham and Anderson, 2002). Results Biomass Between 2003 and 2010, 5271 trawls were successfully conducted as part of the groundfish survey with an annual average of 659 trawls yr-1 (range: 505 to 722 trawls; Fig. 1). Although an average of 265 individual fish taxa were identified each year (range: 252- 310), the 60 demersal fishes and two benthic invertebrates included in this analysis comprised greater than 99% of the total catch. Annual biomass indices (t) for the 62 individual species revealed variable trends over time (2003-10) (Figs. 2 and 3). Six species exhibited signifi- cant (P<0.001) increases in biomass indices over time (Fig. 2A), 20 species displayed significant (P<0.05) or near significant (P<0.10) negative trends (Fig. 3, Califor- nia skate [Raja inornata] and pygmy rockfish [Sebastes wilsoni] not shown), and 36 species exhibited nonsig- nificant trends over time. Representative examples for species with no significant trends are shown for the most abundant species within each group (Fig. 2B). Regardless of trends, both target and nontarget species occurred in each group. Mean, minimum, and maximum depth (m), and total numbers of positive hauls over the eight year study are shown for the 62 individual species included in the analyses (Table 1). Catch for these 62 species was initially partitioned into seven groups based on tax- onomy and depth (in order of decreasing biomass): flat- fish (30%), other shallow to mid-depth species (20%), shelf rockfish (15%), sharks, skates, and ratfish (13%), other deep water species (9%), thornyheads (8%), and slope rockfish (5%), to examine trends over time. The weighted mean depths for shallow to mid-depth spe- cies was <500 m, and the weighted average for deep water species was >650 m. Shelf rockfish occurred at average depths ranging from 101 to 209 m, whereas slope rockfish were somewhat deeper (226-456 m). In general, rockfish were encountered in fewer hauls than other subgroups. Despite variations in biomass indices at the species level, four of the seven groups initially examined here (with the exception of slope rockfish, thornyheads, and other deep water species) and overall biomass indi- ces decreased significantly (P<0.05) over time (Fig. 4). Overall aggregate coastwide biomass indices for all 62 species decreased approximately 60% from 2,308,207 t in 2003 to 1,384,391 t in 2010. However the lowest biomass (1,373,473 t) was recorded in 2008 (the year of the lowest PDO and also possibly a good recruitment year) followed by slight increases in 2009 and 2010. Recruitment versus PDO indices Our initial analyses based on biomass indices for spe- cies grouped taxonomically or by depth indicated that deepwater species, such as thornyheads and slope rock- fish, did not significantly decrease over time (Fig. 4). However because some species within these groups either decreased significantly or displayed decreasing trends (Figs. 2 and 3), we included all 62 species when we separated demersal catch into categories based on recruitment. Our examination of 24 stock assessment models revealed that 13 species had large recruitment events occurring primarily in 1999 (arrowtooth flounder [Atheresthes stomias], English sole, Pacific hake, sable- fish [ Anoplopoma fimbria], bocaccio [Sebastes paueispi- nis ], chilipepper rockfish [Sebastes goodie ], splitnose rockfish [Sebastes diploproa], but occasionally some- what earlier (petrale sole [Eopsetta jordani ], longspine thornyhead [Sebastolobus altivelis]), or later (Dover sole [Microstomas pacificus], greenstriped rockfish [Sebastes elongates ], darkblotched rockfish [Sebastes crameri ], Pacific ocean perch [Sebastes alutus]). Eleven additional species assessed since 2005 did not display significantly larger individual recruitment levels during the period examined (spiny dogfish [Squalus acanthias], longnose skate [Raja rhina], lingcod [Ophiodon elongatus], black- gill rockfish [Sebastes melanostomus], canary rockfish [Sebastes pinniger], greenspotted rockfish [Sebastes chlorostictus], shortbelly rockfish [Sebastes jordani[, widow rockfish [Sebastes entomelas] yelloweye rockfish [Sebastes ruberrimus], yellowtail rockfish [Sebastes flavi- dus], shortspine thornyheads [Sebastolobus alascanus]). Stock assessments have not yet been conducted on the remaining 38 species included in our analyses. During the 1999-2010 period, when depletion of a strong year class by fisheries was expected to result in declining biomass trends, we observed variable trends in the PDO index (Fig. 5). Changes in the PDO index from 1999 to 2010 indicate that average climate in the California Current system gradually shifted from cool (1999-2001) to warm (2003-06) and back to cool ( 2007— 10) conditions (Fig. 5). We summed biomass for each of the three groups previously described and overall (all species combined) and regressed aggregate biomass (natural log transformed) versus year and annual PDO indices (Fig. 6). We noted significant (P<0.001) or near- significant (P=0.06) inverse relationships for all groups and overall versus year (Fig. 6). For species with strong recruitment events in 1999, the trend in biomass over time was increasingly downward, whereas for those with no or unknown recruitment there was a tendency for biomass to increase in recent years (Fig. 6). All groups also demonstrated significant relationships with the annual PDO indices (Fig. 6); however, the greatest Fishery Bulletin 1 10(2) A Increasing trends 250.000 200.000 150.000 100.000 50,000 0 longspine thornyhead ( Sebastolobus altivelis ) ft ft ft ft ft B Nonsignificant trends Dover sole ( Microstomus pacificus) Pacific grenadier 12 ooo ( Embassichthys bathybius) yellowtail rockfish (Sebastes flavidus) 40.000 20.000 0 3000 snakehead eelpout ( Lycenchelys crotalinus) 20,000 15.000 10.000 5000 Pacific cod (Gadus macrocephalus) ft i a i . SB -01 , L 2003 2004 2005 2006 2007 2008 2009 2010 Figure 2 (A) Increasing and (B) nonsignificant trends in annual, coast-wide demersal biomass indices (in metric tons, t) for 12 taxa caught in the Northwest Fisheries Science Center’s West Coast Groundfish Bottom Trawl Survey, 2003-10. Taxa are ranked from highest (top) to lowest (bottom) biomass for each category, and standard errors are shown. All six species exhibiting significant (P<0.05) or near-significant (P<0.10) increases in biomass indices over time are shown; however representative examples of species with nonsignificant trends (36 spe- cies) are shown only for the most abundant taxa for each subgroup described in Table 1. Keller et al. : Variations in eastern North Pacific demersal fish biomass, 2003-10 211 350.000 280.000 210,000 140.000 70.000 0 200.000 150.000 100.000 50.000 0 75.000 50.000 25.000 0 75.000 50.000 25.000 Pacific hake ( Merluccius productus ) ft & c tj ft a & x LI L C=B □ 15,000 10,000 4000 3000 2000 1000 Ai sablefish (Anoplopoma fimbria) A a, £ lingcod ( Ophiodon elongatus) Hi, spotted ratfish (Hydrolag us collief) n n o LI Li U U ft A bocaccio ( Sebastes paucispinis) fi a black eelpout ( Lycodes diapterus) ft ft i ft 4 i ft 350.000 280.000 210,000 140,000 70.000 0 180.000 135,000 90.000 45.000 0 75.000 50.000 25.000 50.000 40.000 30.000 20.000 10,000 0 15.000 10.000 '5000 0 3000 2000 spiny dogfish ( Squalus acanthias) □ 0 fi 0 => ft sharpchin rockfish (Sebastes zacentrus) rex sole (Glyptocephalus zachirus) ft ft darkblotched rockfish ( Sebastes cramen) fi a i fi a white croaker (Genyonemus lineatus) 0 ■ »-* 2003 2004 2005 2006 2007 2008 2009 20 1 0 redbanded rockfish (Sebastes babcocki ) 250.000 200.000 150.000 100.000 50.000 0 125.000 100.000 75.000 50.000 25.000 0 60.000 40.000 20.000 40.000 30.000 20.000 10,000 0 6000 4000 chilipepper rockfish ( Sebastes goodei) shortbelly rockfish ( Sebastes jordani) ft . i £ ft English sole (Parophrys vetutus ) o u u u giant grenadier (Albatrossia pectoralis ) A A A 4 flathead sole ( Hippoglossoides elassodon) ft 1 a s a 2000 1500 1000 500 curlfin sole (Pteuronichthys decurrens) 2003 2004 2005 2006 2007 2008 2009 2010 2003 2004 2005 2006 2007 2008 2009 201 0 Figure 3 Decreasing trends in demersal biomass indices in metric tons (t) (±standard error) for 18 of 20 taxa caught in the Northwest Fisheries Science Center’s West Coast Groundfish Bottom Trawl Survey, 2003-10. With the exception of California skate and pygmy rockfish, all species exhibiting significant (P<0.05) or near- significant (P<0.1Q) decreases in biomass indices over time are shown. amount of variation in biomass was explained by the PDO indices for those species with no strong recruit- ment during the late 1990s. We used AIC to determine which model (i.e., based on year, PDO indices or com- bined) provided the best fit to the data for each group (Table 2). For species with strong recruitment, the re- gression of biomass versus year had the minimum AIC value; for species without strong recruitment the model incorporating PDO indices provided the best fit. For both other groups (unknown recruitment and overall), 212 Fishery Bulletin 1 10(2) Table 1 Common name, scientific name, and group designations according to taxonomy or depth for 62 dominant demersal species col- lected during the 2003 to 2010 Northwest Fisheries Science Center West Coast Groundfish Bottom Trawl Surveys. Mean, mini- mum, and maximum depth, and the total number of positive hauls per species are shown for the 2003 to 2010 period. Mean values are weighted by catch per unit of effort to accurately reflect average depth. Depth (m) Hauls Common name Scientific name mean min. max. (No. of hauls) Sharks, skates, ratfish big skate California skate filetail cat shark longnose skate sandpaper skate spiny dogfish spotted ratfish Flatfish arrowtooth flounder curlfin sole Dover sole English sole flathead sole Pacific sanddab petrale sole rex sole slender sole Shallow to mid-depth water (<500 m) bigfin eelpout black eelpout Dungeness crab lingcod Pacific cod Pacific hake plainfin midshipman sablefish white croaker Rockfish — shelf bocaccio canary rockfish chilipepper greenspotted rockfish greenstriped rockfish halfbanded rockfish pygmy rockfish redstripe rockfish rosethorn rockfish sharpchin rockfish shortbelly rockfish squarespot rockfish stripetail rockfish swordspine rockfish widow rockfish yelloweye rockfish yellowtail rockfish Raja binoculata 104 Raja inornata 107 Parmaturus xaniurus 489 Raja rhina 265 Bathyraja kincaidii 284 Squalus acanthias 171 Hydrolagus colliei 185 Atheresthes stomias 208 Pleuronichthys decurrens 94 Microstomus pacificus 382 Parophrys vetulus 130 Hippoglossoides elassodon 146 Citharichthys sordidus 107 Eopsetta jordani 136 Glyptocephalus zachirus 235 Lyopsetta exilis 209 Lycodes cortezianus 343 Lycodes diapterus 470 Cancer magister 139 Ophiodon elongatus 137 Gadus macrocephalus 133 Merluccius productus 281 Porichthys notatus 109 Anoplopoma fimbria 495 Genyonemus hneatus 85 Sebastes paucispinis 159 Sebastes pinniger 139 Sebastes goodei 167 Sebastes chlorostictus 146 Sebastes elongatus 156 Sebastes semicinctus 115 Sebastes wilsoni 138 Sebastes proriger 158 Sebastes helvomaculatus 207 Sebastes zacentrus 209 Sebastes jordani 170 Sebastes hopkinsi 101 Sebastes saxicola 175 Sebastes ensifer 156 Sebastes entomelas 195 Sebastes ruberrimus 149 Sebastes flavidus 141 56 332 668 56 792 517 113 1224 472 57 1227 2934 59 1173 1631 57 1143 1737 56 1241 2521 58 992 1726 56 440 473 56 1246 4205 56 480 1957 62 346 332 56 491 1554 56 541 2059 56 937 3147 57 830 2564 57 1095 1418 82 1143 896 56 835 1633 56 417 1553 56 285 273 56 1213 2775 56 464 656 57 1268 3320 56 181 302 56 333 263 57 264 340 56 464 656 62 348 303 64 474 1299 57 440 378 64 268 95 66 271 120 65 447 420 76 455 307 71 406 405 59 203 80 59 436 1076 100 236 30 64 399 182 66 250 107 60 343 315 continued Keller et al : Variations in eastern North Pacific demersal fish biomass, 2003-10 213 Table 1 (continued) Depth (m) Hauls (No. of hauls) Common name Scientific name mean min. max. Rockfish — slope aurora rockfish Sebastes aurora 456 129 814 689 bank rockfish Sebastes rufus 267 101 499 91 blackgill rockfish Sebastes melanostomus 406 133 647 264 darkblotched rockfish Sebastes crameri 226 84 538 920 Pacific ocean perch Sebastes alutus 279 87 715 344 redbanded rockfish Sebastes babcocki 269 84 550 398 rougheye rockfish Sebastes aleutianus 291 124 798 253 splitnose rockfish Sebastes diploproa 287 68 559 1107 Thornyhead rockfish longspine thornyhead Sebastolobus altiuelis 767 104 1271 1844 shortspine thornyhead Sebastolobus alascanus 611 67 1268 2590 Deep (>650 m) brown cat shark Apristurus brunneus 672 82 1241 1717 California slickhead Alepocephalus tenebrosus 909 477 1268 1097 deepsea sole Embassichthys bathybius 874 276 1271 1116 giant grenadier Albatrossia pectoralis 943 443 1271 838 grooved Tanner crab Chionoecetes tanneri 782 64 1271 1431 Pacific flatnose Antimora microlepis 931 262 1271 890 Pacific grenadier Coryphaenoides acrolepis 915 313 1271 994 roughtail skate Bathyraja trachura 941 107 1271 622 snakehead eelpout Lycenchelys crotalinus 893 63 1264 783 twoline eelpout Bothrocara brunneum 861 343 1271 898 Table 2 Akaike information criterion (AIC), P, coefficient of determination (r2), and zf, for different models fitting annual biomass indices versus year, Pacific Decadal Oscillation (PDO) values, and both variables for three subgroups of demersal species determined by the occurrence of strong recruitment during the late 1990s (with recruitment: « = 13, without recruitment: ra=ll, and unknown: n = 38) and overall (n = 62). The best model for each group is determined by the minimum AIC value (!) within each category, with lower AIC values indicating a better fit. Additionally, when Al values are <7 (2) there is some support for the alternative model. Similar results are shown for species richness subdivided by depth strata (shallow 55-183 m; mid-depth 184-549 m; deep 550-1280 m and overall 55-1280 m). With year With PDO With year and PDO Models: AIC P r2 AIC P r2 AIC P r2 4 Biomass with recruitment -8.91 0.007 0.87 -4.8 0.02 0.62 -1.7 0.008 0.93 7.2 without recruitment 2.8 0.002 0.81 -0.6J 0.001 0.84 5.6 0.01 0.91 6.22 unknown recruitment 0.8 0.07 0.45 -5.1 0.01 0.70 -8.71 0.0002 0.99 9.5 all -6.3 0.001 0.85 -7.5 0.002 0.81 -1 1.97 0.0001 0.99 5.62 Richness shallow 23.1 0.15 0.34 15. 77 0.009 0.71 18.4 0.10 0.75 7.4 mid-depth 17.5 0.06 0.49 12. 6; 0.01 0.68 16.1 0.12 0.73 4.92 deep 18.7 0.28 0.19 16.0 0.20 0.20 12. 81 0.05 0.84 5.92 all 15.9 0.08 0.45 7.61 0.004 0.80 13.4 0.08 0.79 8.3 214 Fishery Bulletin 1 10(2) 13.2 12 8 12.4 12.0 11.6 11.2 sharks, skates, ratfish L t y=-0.131x+275 ^=0.79 * 13.5 13.3 13.1 12.9 flatfish y=-0.027x+66 ^=0 58 13.5 13.0 12.5 12.0 115 % 11.0 to >/> 03 E 13.5 o CD 13 0 12.5 - shelf rockfish i 4 y=-0 18x+373 r*=0.89 others shallow-mid y=-0 138X+290 ^=0.89 12.0 * J 12.4 119 114 10.9 10.4 12.5 12.4 12.3 12.2 12.1 slope rockfish ttti. i ^=0 003 thornyheads JtM 1 I y=0.027x-41 1 I ^=0 59 12.3 12.1 119 others deep I * * it tu ^=0.0004 1 1 7 4 2002 2004 2006 2008 2010 14 8 14 6 14 4 14.2 -I total groundfish y=-0 091x+198 ^=0.93 14.0 2002 2004 2006 2008 2010 Figure 4 Expanded coast-wide biomass indices (expressed as the natural log in metric tons, [t] ) and standard errors as a function of time (year) for 62 dominant demersal taxa collected during the 2003-10 groundfish survey, subdivided into seven subgroups (Table 1), and summed overall. Regression equations for subgroups with significant decreasing trends (P<0.05) over time and coefficients of determination (r2) are displayed. the minimum AIC value occurred for models incorporat- ing both year and PDO indices combined (Table 2). The A- for the four biomass models indicate that none of the models are equivalent to the best model; however, the observed A- <7 for two of the four groups suggest some support for alternative models (Table 2). Species richness Species richness indices incorporated all fish collected during the 2003-10 surveys, including rare deep-water species and those not normally associated with the bottom. Regressions between species richness and year indicate near significant negative trends for mid-depth (P=0.06) and overall (P=0.08), but insignificant rela- tionships for shallow and deep depth strata (Fig. 7). With the exception of the deep depth stratum (P=0.20), significant positive relationships (P<0.01) were observed between species richness and the PDO. The number of fish species present within two depth strata (shallow and mid-depth) and overall increased during the warm PDO phase. Additionally, species richness indices declined within increasing depth strata (Fig. 7). Based on minimum AIC values, the models which provided the best fit to changes in species richness during the survey period were either models incor- porating only PDO indices (shallow, mid-depth, and overall) or models incorporating both year and PDO values (deep), but not models based solely on year (Table 2). Keller et al : Variations in eastern North Pacific demersal fish biomass, 2003-10 215 Discussion The Pacific Coast groundfish fishery is exceed- ingly complex to manage because a wide range of species (90+), including a number of overfished and rebuilding stocks, are caught with the same trawl gear. Beginning in 2000, the PFMC ini- tiated a series of measures designed to reduce catch along the west coast, including fleet reduc- tions, closed areas, and catch restrictions. These measures were initiated in direct response to nine stocks being declared overfished from 1999 to 2002. Landed catch from 1980 through 2010 shows that groundfish harvests off Washington, Oregon, and California were significantly lower (rockfishes, flatfishes, sablefish) or relatively con- stant (Pacific hake, thornyheads) in recent years relative to historical rates (Fig. 8). This period of decreased catch corresponds to the implementa- tion of the PFMC’s management plan. And yet, despite much reduced fishing effort (and landings), overall survey indices of groundfish biomass off the western United States declined from 2003 through 2010. The PFMC’s management directives also coincided with the period following expansion of the NWFSC’s West Coast Goundfish Bottom Trawl Survey to annu- ally include both the shelf region, as well as the upper continental slope waters along the entire coast (U.S.- Canada to U.S.-Mexico). We used data from this fishery- independent survey combined with information from 24 stock assessments since 2005 to attempt to unravel the causes for the decline in biomass indices despite strict adherence to fishery management plans. Our results indicated that from 2003 through 2010 individual groundfish stocks along the U.S. west coast responded in varying ways to the newly imposed man- agement measures, with many of the overfished spe- cies of concern exhibiting increases in spawning stock biomass (PFMC, 2008a). However, despite improve- ments in individual stocks there has been a gradual decline in overall groundfish biomass, measured as the sum of the biomass indices for 60 abundant groundfish and two benthic invertebrate species, as well as in major groundfish groups (e.g., sharks, flatfishes, and rockfishes). Twenty of the 62 species described here significantly decreased in observed biomass from 2003 through 2010, whereas 6 species significantly increased. Of the remaining 36 species, 20 exhibited decreas- ing trends and 16 exhibited increasing trends. These changes indicate that the decline in biomass is not attributable to just a few species. Similar declines in Northwest Atlantic fish stocks have previously been attributed to a variety of factors including overfish- ing and environmental effects (Haedrich and Barnes, 1997). Hilborn et al. (in press) noted that the dramatic decline in catch can be interpreted in two ways: as an indication of collapsing stocks caused by overfishing, or as a demonstration that management has effectively reduced catch to prevent overfishing of sensitive species. However, the continued decline in overall observed bio- 03 13 C C 03 O Q CL 1 5 1 0 0 5 0 0 -0 5 -1 0 -1 5 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 Figure 5 Annual variation in the Pacific Decadal Oscillation index (PDO) from January 1999 through December 2010 as it fluctuated between cool negative and warm positive phases. Annual values are the mean of the 12 months beginning in November of the year preceding the start of each annual survey and ending in October with the completion of the annual survey. mass along the west coast suggests that reduced catch in itself may not be sufficient to prevent biomass from further decreasing when environmental conditions are poor. Hsieh et al. (2008) further concluded that fishing pressure reduces the resilience of exploited populations facing negative climatic effects. By chance our study occurred during a period after the emergence of a particularly strong 1999 year class for many groundfish species inhabiting the California Current system (Haltuch and Hicks, 2009) and a pe- riod of changing environmental conditions as indicated by variability in the annual PDO index. The observed declining survey trends are consistent with natural and fishing-induced mortality estimated for the 1999 cohort in many stock assessments, especially flatfishes, sablefish, and Pacific hake (Schirripa, 2007; Haltuch and Hicks, 2009; Stewart et al., 2011). However, the prevalence of this large year class among many west coast groundfish and its gradual depletion over the sur- vey period (2003-10) may not be entirely responsible for the dramatic decline in overall biomass estimated with data from our west coast fishery-independent survey. The decline in biomass may additionally be tied to environmental conditions. The annual PDO index, a measure of climate variability, declined from a high value at the start of our standardized survey (2003) to low and negative values near the end of the series examined (2007-10). Numerous studies correlate shifts in the abundance and distribution of marine fish to oscillations in ocean conditions (Francis et al., 1998; Beamish et ah, 1999; Hare et ah, 1999). Within the northern California Current system, changes in salmon production (Mantua et ah, 1997), landed sardine catch (Smith and Moser, 2003; Norton et al., 2008), and Pa- cific hake distribution (Benson et al., 2002) are linked to decadal-scale fluctuations in climate. Hollowed et al. (2001) further reported that production of commercial fish stocks (32 pelagic fish and groundfish species) in 216 Fishery Bulletin 1 10(2) 14.1 13.9 13.7 13 5 13.3 with recruitment t y=-0.059x+1 31 r>=0.87 14 1 13.9 13.7 13.5 13.3 with recruitment # y=0.159x+13.7 ^=0.62 i 13.4 without recruitment 13.0 12.6 i 12 2 y=-0 124X+262 ^ I " r*=0.81 T 11.8 Hi 13.4 13.0 12.6 12.2 # * 11.8 without recruitment 13.6 unknown recruitment 13.6 13.4 13.4 13.2 13.2 13.0 * 13.0 12.8 y=-0.046x+106 * 12.8 12 6 ^=0.45 12.6 14 8 total 14.8 14.6 i 14.6 14.4 Vj. ’ V*. 14,4 14.2 T y=-0.066x+147 • 14.2 14.0 ^=0.85 " 14.0 2002 2004 2006 2008 2010 If a .1 ' • y=0 41x+12.5 ^=0.84 " y=0.186x+13.0 r==0.70 total ■I Year • #- " « y=0.209x+14.3 I -- " r*=0.81 1.5 -1.0 -0.5 0.0 0.5 1.0 1.5 PDO Figure 6 Expanded coast-wide biomass indices (natural log of biomass, [ t] ) and standard errors as a function of year and annual Pacific Decadal Oscillation (PDO) values for 62 dominant demersal taxa collected during the 2003-10 groundfish survey subdivided into three subgroups (with, without, and unknown recruitment) and summed overall (total). Regression equations for subgroups with signifi- cant (P<0.05) or near significant (P<0.10) trends are displayed in the figure. Significance levels (P) are listed in Table 2. the North Pacific responded to climatic shifts. Major changes in coastal ocean productivity of the eastern Pacific are correlated with phase changes in the PDO and cold eras are associated with enhanced productiv- ity. Warm periods are correlated with low productivity in the California current system (Brodeur and Ware, 1992; Hare and Francis, 1995) and the observed de- crease in coastwide biomass could reflect the decline in biomass during the most recent warm period (2003-06) as productivity levels drop. Results indicate that not only for individual species but also for species grouped taxonomically or by depth the change in biomass over time (2003-10) was vari- able. For the slope rockfish group (aurora rockfish [Se- bastes aurora], bank rockfish [Sebastes rufus], blackgill rockfish, darkblotched rockfish, Pacific ocean perch, redbanded rockfish, [Sebastes babcocki], rougheye rock- fish [Sebastes aleutianus], and splitnose rockfish) and the deepwater species group (mean depth >650 m) composed of brown cat shark ( Apristurus brunneus), Keller et al Variations in eastern North Pacific demersal fish biomass, 2003-10 217 shallow 55-183 m shallow 55-183 m in a> O CD CL CO 18 i i ♦ * 18 16 * # * * 16 ^=0.34 14 14 18 mid-depth 184-549 m 18 * 16 * 16 T" 14 y=-0.200x+417.2 14 d=0 49 12 12 15 deep 15 550-1280 m 14 i , } i 14 13 i * i T I 13 12 1 r2=0 19 t * 12 11 11 18 all depths 18 55-1280 m 16 16 14 i "r~ 14 y=-0 160x+336 3 12 r*=0.45 12 2002 2004 2006 2008 2010 * i y=1 086x+17 0 r*=0 71 mid-depth 184-549 m ^"1" * y=0.757x+15.5 r'=0 68 deep 550-1280 m 1 i f all depths 55-1280 m r*=0 20 y=0 691x+15 0 r>=0.80 -1 -0 5 0 0 5 1 1.5 Year PDO Figure 7 Coast-wide estimates of species richness (± standard error) by depth (shallow 55-183 m; mid 184-549 m; deep 550-1280 m; all depths combined 55-1280 m) versus year and annual Pacific Decadal Oscillation (PDO) values, 2003-10. Regression equations for relationships with significant (P<0.05) trends are displayed. Coefficients of determination (r2) are shown here for the regression equations while significance levels ( P ) are shown in Table 2. California slickhead ( Alepocephalus tenebrosus), deep- sea sole (Embassichthys bathybius), giant grenadier ( Albatrossia pectoralis ), Pacific flatnose ( Antimora mi- crolepis), Pacific grenadier ( Coryphaenoides acrolepis ), roughtail skate (Bathyraja trachura), snakehead eel- pout ( Lycenchelys crotalinus), twoline eelpout (Bothro- cara brunneum), and grooved Tanner crab ( Chionoece - tes tanneri), no significant relationship with time was detected over the survey period. Thornyheads were the sole group to exhibit an overall positive correlation with time (with higher biomass occurring in recent years). Shallow to mid-water species (mean depth <500 m) consisting of groups composed of cartilaginous fish, flatfish, shelf roekfish, or a mixed subgroup of fish, and overall biomass all significantly declined over time. These declines occurred despite the classification of the west coast fishery as having the lowest overall exploita- tion rates of ten ecosystems examined by Worm et al. (2009) and multiple management measures introduced to reduce catch. 218 Fishery Bulletin 1 10(2) 80,000 60,000 40,000 20,000 Rockfish ■o— Flatfish ■*— Sablefish o Thornyheads 400,000 300,000 200,000 100,000 B -° Pacific hake - Total groundfish 1980 1985 1990 1995 Year 2000 2005 2010 Figure 8 Landed catch (in metric tons, t) for (A) rockfish ( Sebastes spp.), flatfish, sablefish ( Anoplopoma fimbria ), and thornyheads (Sebastolobus spp.), and (B) total groundfish and Pacific hake (Merluccius productus), off Washington, Oregon, and California from 1980 through 2010. Data were obtained from the Pacific Fisheries Information Network: http:// pacfin.psmfc.org/pacfin pub/data.php (accessed September 2011). Note the different scales for overall catch and Pacific hake (0-400,000 t) and for the remainder of the species (0-80,000 t). To evaluate the decline in overall groundfish biomass indices off the western United States, despite much re- duced fishing effort, we investigated two potential and perhaps overlapping factors: depletion after strong re- cruitment and environmental effects. Decreases in total biomass indices occur within an ecosystem when catch is greater than net population growth. To evaluate the contribution of these factors to the observed decline in biomass, we separated the species examined here into subgroups based on the presence or absence of strong recruitment during the late 1990s. We used information contained in 24 stock assessments to assign species to subgroups with strong recruitment (primarily in 1999), without strong recruitment, and with unknown recruitment. We developed regressions models for each subgroup and overall versus year, annual PDO indices, and including both variables. We assumed that year was a good proxy for depletion of exceptionally strong cohorts as they recruited to the groundfish fishery in subsequent years (Haltuch and Hicks, 2009). For mul- tiple species with highly successful recruitment events in 1999, there has been a gradual decline in biomass since 2003. Regression models for total biomass of these species were best fitted by time, whereas regression models for those species without strong recruitment were best fitted to variation in climate as measured by annual PDO indices. Our analysis also indicated that for models comparing summed biomass for groups with Keller et at. Variations in eastern North Pacific demersal fish biomass, 2003-10 219 unknown recruitment and overall, the best fit incorpo- rated both year and PDO indices combined. We think it is important to note that for all subgroups, particularly the group without species with known strong recruit- ment events, there is still a significant decrease in biomass indices over time. This decrease indicates that the depletion of an exceptionally strong 1999 year class for multiple species is not the only factor contributing to decreasing biomass of west coast demersal fish spe- cies. Despite the overall decline in biomass observed in recent years, very few of the species examined here are considered overfished (NMFS, 2009) and therefore a loss of yield cannot be inferred. However, the continued decline in overall observed biomass from 2003 to 2010, despite enactment of multiple management measures and reduced catch, emphasizes the need for the state of the ecosystem to be considered when setting catch limits. During periods of low ocean productivity, a pre- cautionary approach is advised. During the NWFSC bottom trawl survey, random, rare, and very large tows of schooling rockfish are oc- casionally taken. If a tow occurs near rocky but still trawlable habitat when schooling rockfishes are pres- ent, then very large catches of rockfish are possible (Stewart, 2007). Occurrence of these large catches may prevent detection of underlying biomass trends when only a small number of years are available for analysis, as with the 2003-10 survey. Although this phenomenon appears clear for canary and widow rockfish, for spe- cies like sharpchin and redbanded rockfish, it is harder to clearly separate large rare tows from an increase or decrease in biomass, and certainly random large catches and biomass trends may be occurring at the same time (Fig. 3). The higher productivity associated with the cool PDO phase in the California Current system may have re- sulted in another strong recruitment event in 2008 (Ralston3) and if cool conditions continue, the associated higher productivity could promote enhanced growth and survival of groundfish. Our data indicate that the return to cool conditions in 2007-10 was followed by a slight increase in overall biomass in 2010 — an increase that suggests a time lag between cool conditions and increased demersal biomass within the California Cur- rent. Although many rockfish species are long lived and exhibit highly variable recruitment, previous studies have additionally indicated that both rockfish recruit- ment and juvenile growth respond to broad indicators of productivity, and that juvenile abundance is corre- lated with large-scale oceanographic events such as El Niho-Southern Oscillation and the PDO (Ainley et ah, 1993; Laidig et ah, 2007). For now we find our results interesting and recognize that the value of the ground- fish survey increases with each annual increment and over time will provide additional information to unravel these relationships. 3 Ralston, S. 2011. Personal commun. National Marine Fisheries Service, Southwest Fisheries Science Center, Fish- eries Ecology Division. Santa Cruz, CA 95060. The best models to describe the variation in species richness (restricted to fish only) included the PDO indi- ces (shallow, mid-depth, and overall) or a combination of the PDO indices and year (deep depths) but were not based on year alone. Species richness was positively correlated with the annual PDO index indicating that more species were present within the survey area and, in particular, at shallow and mid-depth strata during the warm phase of the PDO. Tolimieri and Levin (2006) and Tolimieri (2007) examined patterns of diversity in groundfish assemblages in relation to depth (200-1200 m) and latitude (33-37° N) along the U.S. Pacific Coast. They found, as we did, that species richness declined with depth (Fig. 7), but they did not examine changes over time or in relation to climate indices. However, their observation that patterns of diversity were cor- related with temperature may partially support our ob- servation of elevated species richness during the warm phase of the PDO. Tolimieri (2007) points out that lati- tude and depth are factors well known to correlate with diversity and assemblage structure. Both species rich- ness and biomass decreased along the U.S. west coast for demersal groundfish as the PDO index shifted from a warm to a cool phase. Given that the swept area per haul remains constant, fewer species per haul are ex- pected if the underlying population densities decrease; however, as demonstrated above, the densities for the 62 most abundant species exhibited variable trends from 2003 through 2010. Mueter and Litzow (2008) provided convincing evi- dence of climate-linked changes in the distribution of demersal fishes in the Bering Sea (1982-2006), cou- pled with reorganization in community composition by latitude. They observed increases in both biomass and species richness in an area characterized by warming temperatures. Community-wide patterns indicated that taxa shifted northward and also were captured with increasing frequency at shallower stations from 1982 to 2006. Interestingly, they noted that mean species rich- ness significantly increased within the portion of the survey area termed the “cold-pool” as it warmed over time, in a similar relationship to that observed here. Species richness along the U.S. west coast was elevated during the warmer phase of the PDO and lower during the cool phase. Changes in species richness are most likely caused by movement of species in response to environmental conditions (Trenkel and Cotter, 2009). Understanding the mechanisms underlying the ob- served relationships between biomass indices, species diversity, depletion, and climate remains a challenge. However, Brodeur et al. (2008) recently related varia- tions in the abundance of dominant ichthyoplankton in the northern California Current to oceanic and climatic indices, thus providing a link between climate and recruitment success. Both larval concentrations and diversity varied on a semidecadal basis in conjunction with fluctuations in the PDO. Zheng and Kruse (2006) found some evidence that recruitment variation in east- ern Bering Sea crabs may be related to climate forcing, although interaction with groundfish predators likely 220 Fishery Bulletin 1 10(2) contributed to recruitment success as well. Brodeur and Ware (1992) and Sugimoto and Tadokoro (1997) both demonstrated changes in zooplankton biomass and distribution coupled with changes in ocean conditions, and Lenarz et al. (1995) reported reduced primary pro- ductivity and zooplankton biomass, coupled with poor rockfish recruitment off the west coast during El Nino events. Recruitment success of salmon within the north- ern California Current system has previously been tied to fluctuations in the PDO index (Mantua et ah, 1997). Hollowed et al. (2001) examined the timing of the PDO and ENSO and correlated changes in both with recruit- ment success in groups such as flatfish and gadoids, but further study is needed to confirm their findings. Like others, they caution that additional factors could also affect biomass, species richness, and distributional changes such as density-dependent habitat selection, timing of migrations, changes in local currents, catch- ability, and shifts between nursery and feeding grounds (Swain et al., 1994; Delworth et al., 1997; Attrill and Power, 2002; Magill and Sayer, 2002). The interactions among climate variability, currents, and the seasonal strength of upwelling and downwelling is particularly interesting given our prior research where we directly linked changes in demersal biomass and species diver- sity to depressed oxygen levels along the Oregon shelf (Keller et al., 2010). At the species level, changes appear driven by cli- mate-induced variation in primary and secondary productivity and recruitment (Beaugrand et al., 2003; Steingrund and Gaard, 2005), although the nature of the relationship has not been deciphered. We recognize that the relatively short time series examined may in- crease the likelihood that the results are spurious. The observed tight correlations between total and grouped groundfish biomass indices and the PDO are expected if the underlying relationship results from reduced popu- lation growth due to poor environmental conditions or if environmental conditions, such as phase shifts in regional climate (Mantua et al., 1997; Hollowed et ah, 2001; Castonguay et al., 2008) are also coupled with pe- riodic strong recruitment events, such as the emergence of the 1999 year class. Natural mortality for species with relatively high natural mortality rates could play an additional role in declining biomass indices. However, for rockfish with low natural mortality rates, it is likely that growth would be more important than mortality during most of the time period. The relationship between the PDO and biomass indices may also be due to changes in catch- ability or selectivity, rather than to actual population changes. A decrease could be due to higher selection of younger fish (i.e., to peak selection around age 4 or 5 and a decline afterwards). Although understanding the mechanisms supporting the relationships observed here remains problematic, our results demonstrate the importance of incorporating environmental conditions in management decisions. Despite enactment of highly effective management measures and the occurrence of periodic strong recruitment, biomass indices declined as oceanographic conditions changed throughout much of the survey period. Acknowledgments The authors are indebted to the captains and the crew of the chartered fishing vessels Excalibur, Ms. Julie , Noah’s Ark, Raven , B. J. Thomas, and Sea Eagle for providing at-sea support. We especially thank the West Coast Groundfish Bottom Trawl Survey team (V. Simon, M. Bradburn, K. Bosley, D. Kamikawa, V. Tuttle, J. Buchanan, and J. Harms) for their skill and dedication in collecting high quality data for the groundfish survey and C. Whitmire for preparing GIS charts. 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. CalCOFI Rep. 34:133-139. Attrill, M. J., and M. Power. 2002. Climatic influence on a marine fish assem- blage. Nature 417:275-278. Beamish, R. J., D. Noakes, G. A. McFarlane, L. Klyashtorin, V. V. Ivonov, and V. Kurashov. 1999. The regime concept and natural trends in the production of Pacific salmon. Can. J. Fish. Aquat. Sci. 56:506-515. Beaugrand, G., K. Brander, J. A. Lindley, S. Souissi, and P. C. Reid. 2003. Plankton effect on cod recruitment in the North Sea. Nature 426:661-664. Benson, A. J., G. A. McFarlane, S. F. Allen, and D. E. Dower. 2002. Changes in Pacific hake (Merluccius produc- tus) migration patterns and juvenile growth related to the 1989 regime shift. Can. J. Fish. Aquat. Sci. 59:1969-1979. Brodeur, R. D., W. T. Peterson, T. D. Auth, H. L. Soulen, M. M. Parnel, and A. A. Emerson. 2008. Abundance and diversity of coastal fish larvae as indicators of recent changes in ocean and climate conditions in the Oregon upwelling zone. Mar. Ecol. Prog. Ser. 366:187-202. Brodeur, R. D., and D. M. Ware. 1992. Long-term variability in zooplankton biomass in the subarctic Pacific Ocean. Fish. Oceanogr. 1:32-37. Burnham, K. P., and D. R. Anderson. 2002. Model selection and multimodel inference: a practical information-theoretic approach, 2nd ed., 488 p. Springer, New York. Castonguay, M., S. Plourde, D. Robert, J. A. Runge, and L. Fortier. 2008. Copepod production drives recruitment in a marine fish. Can. J. Fish. Aquat. Sci. 65:1528-1531. Cochran, W. G. 1977. Sampling techniques, 2nd ed., 428 p. John Wiley & Sons, Inc., New York. Delworth, T. L., S. Manabe, and R. J. Stouffer. 1997. Multidecadal climate variability in the Greenland Sea and surrounding regions: a coupled model simula- tion. Geophys. Res. Lett. 24:257-260. Keller et al.: Variations in eastern North Pacific demersal fish biomass, 2003-10 221 Draper, N. R., and H. Smith. 1981. Applied regression analysis, 2nd ed., 709 p. John Wiley & Sons, Inc., New York. Field, J. C., and R. C. Fox. 2006. Considering ecosystem-based Fisheries management in the California Current. Mar. Policy 30:552-569. Francis, R. C., S. R. Hare, A. B. Hollowed, and W. S. Wooster. 1998. Effects of interdecadal climate variability on the oceanic ecosystems of the NE Pacific. Fish. Oceanogr. 7:1-21. Haedrich, R. L., and S. M. Barnes. 1997. Changes over time of the size structure in an exploited shelf fish community. Fish. Res. 31:229-239. Haltuch, M. A., and A. Hicks. 2009. Status of the U.S. petrale sole resource in 2008. In Pacific Coast groundfish fishery through 2009, stock assessment and fishery evaluation, 309 p. (Available from Pacific Fishery Management Council, 7700 NE Ambassador Place, Suite 101, Portland, OR, 97220. J Hare, S. R., and R. C. Francis. 1995. Climate change and salmon production in the Northeast Pacific Ocean. Can. Spec. Publ. Fish. Aquat. Sci. 121:357-372. Hare, S. R., N. J. Mantua, and R. C. Francis. 1999. Inverse production regimes: Alaskan and West Coast salmon. Fisheries 24:6-14. Hilborn, R., I. J. Stewart, T. A. Branch, and O. P. Jensen. In press. Defining trade-offs between conservation of species diversity, profitability and food security in the California Current bottom trawl fishery. Conserv. Biol. Hollowed, A. B., S. R. Hare, and W. S. Wooster. 2001. Pacific Basin climate variability patterns of North- east Pacific marine fish production. Prog. Oceanogr. 49:257-282. Hsieh, C-h., C. S. Ress, R. P. Hewitt, and G. Sugihara. 2008. Spatial analysis shows that fishing enhances the climatic sensitivity of marine fishes. Can. J. Fish. Aquat. Sci. 65:947-961. Laidig, T. E., J. R. Chess, and D. F. Howard. 2007. Relationship between abundance of juvenile rockfishes ( Sebastes spp.) and environmental vari- ables documented off northern California and poten- tial mechanisms for the covariation. Fish. Bull. 105:39-48. Keller, A. A., B. H. Horness, E. L. Fruh, V. H. Simon, V. J. Tuttle, K. L. Bosley, J. C. Buchanan, D. J. Kamikawa, and J. R. Wallace. 2008. The 2005 U.S. West Coast bottom trawl survey of groundfish resources off Washington, Oregon, and Cali- fornia: Estimates of distribution, abundance, and length composition. NOAA Tech. Memo. NMFS-NWFSC-93, 136 p. Keller, A. A., V. Simon, F. Chan, W. W. Wakefield, M. E. Clarke, J. A. Barth, D. Kamikawa, and E. L. Fruh. 2010. Demersal fish and invertebrate biomass in rela- tion to an offshore hypoxic zone along the U.S. West Coast. Fish. Oceanogr. 19:76-87. Lenarz, W. H., D. VenTresca, W. M. Graham, F. B. Schwing, and F. Chavez. 1995. Exploration of El Nino events and associated bio- logical population dynamics off central California. Cal- COFI Rep. 36:106-119. Levins, P. S., E. E. Holmes, K. R. Piner, and C. Harvey. 2006. Shifts in a Pacific Ocean fish assemblage: the potential influence of exploitation. Conserv. Biol. 20:1181-1190. Mantua, N. J., S. R. Hare, Y. Zhang, J. M. Wallace, and R. C. Francis. 1997. A Pacific interdecadal climate oscillation with impacts on salmon production. Bull. Am. Meteorol. Soc. 78:1069-1079. Magill, S. H., and M. D. J. Sayer. 2002. Seasonal and interannual variation in fish assem- blages of northern temperate rocky subtidal habitats. J. Fish Biol. 61:1198-1216. Methot, R. D. 2011. User manual for stock synthesis, 165 p. Model ver- sion 3.21d, May 12, 2011. NOAA-NWFSC, Seattle, WA. Mueter, F. J., and M. A. Litzow. 2008. Sea ice retreat alters the biogeography of the Bering Sea continental shelf. Ecol. Appl. 18:309-320. NMFS (National Marine Fisheries Service). 2009. Our living oceans. Report on the status of U.S. living marine resources, 6th ed. NOAA Tech. Memo. NMFS-F/SPO-80, 369 p. Norton, J. G., S. F. Herrick, and J. E. Mason. 2008. Fisheries abundance cycles in ecosystem and eco- nomic management of California fish and invertebrate resources. In The future of fisheries sciences in North America (R. J. Beamish, and B. J. Rothschild, eds.), p. 227-244. Fish Fish. Series, vol. 31. Springer, New York. Peres-Neto, P. R. 1999. How many statistical tests are too many? The problem of conducting multiple ecological inferences revisited. Mar. Ecol. Prog. Ser. 176:303-306. PFMC (Pacific Fishery Management Council). 2008a. Pacific coast groundfish fishery stock assessment and fishery evaluation, vol. 1, 221 p. (Available from Pacific Fishery Management Council, 7700 NE Ambas- sador Place, Suite 101, Portland, OR, 97220.] 2008b. Pacific coast groundfish fishery management plan for the California, Oregon, and Washington ground- fish fishery as amended through amendment 19, 155 p. (Available from Pacific Fishery Management Council, 7700 NE Ambassador Place, Suite 101, Portland, OR, 97220.) Sakamoto, Y., M. Ishiguro, and G. Kitagawa. 1986. Akaike information criterion statistics, 290 p. D. Reidel Publ. Co., Tokyo. Schirripa, M. J. 2007. Status of the sablefish resource off the Continental U.S. Pacific coast in 2007. In Pacific Coast groundfish fishery stock assessment and fishery evaluation, 104 p. (Available from Pacific Fishery Management Coun- cil, 7700 NE Ambassador Place, Suite 101, Portland, OR, 97220.) Schwing, F. B., W. T. Peterson, N. Cyr, and K. E. Osgood. 2009. Future research requirements for understanding the effects of climate variability on fisheries and their management. In The future of fisheries sciences in North America (R. J. Beamish, and B. J. Rothschild, eds.), p. 621-636. Fish Fish. Series, vol. 31. Springer, New York Smith, P. E., and H. G. Moser. 2003. Long-term trends and variability in the larvae of pacific sardine and associated fish species of the California Current region. Deep-Sea Res. 50:2519- 2536. Stauffer, G. 2004. NOAA protocols for groundfish bottom trawl sur- 222 Fishery Bulletin 1 10(2) veys of the nation’s fishery resources. NOAA Tech. Memo. NMFS-F/SPO-65, 204 p. Steingrund, P., and E. Gaard. 2005. Relationship between phytoplankton production and cod production on the Faroe Shelf. ICES J. Mar. Sci. 62:163-172. Stewart, I. J. 2007. Status of the U.S. canary rockfish resource in 2007. In Pacific Coast groundfish fishery stock assess- ment and fishery evaluation, 362 p. (Available from Pacific Fishery Management Council, 7700 NE Ambas- sador Place, Suite 101, Portland, OR, 97220.] Stewart, I. J., R. E. Forrest, C. Grandin, O. S. Hamel, A. C. Hicks, S. J. D. Martell, and I. G. Taylor. 2011. Status of the Pacific hake (whiting) stock in U.S. and Canadian waters in 2011. In Pacific Coast ground- fish fishery stock assessment and fishery evaluation, 206 p. (Available from Pacific Fishery Management Council, 7700 NE Ambassador Place, Suite 101, Port- land, OR, 97220.1 Sugimoto, T., and K. Tadokoro. 1997. Interannual-interdecadal variations in zooplankton biomass, chlorophyll concentration, and physical envi- ronment in the subarctic Pacific and Bering Sea. Fish. Oceanogr. 6:74-93. Swain, D. P, G. A. Nielsen, A. F. Sinclair, and G. A. Chouinard. 1994. Changes in catchability of Atlantic cod ( Gadus morhua ) to an otter-trawl fishery and research survey in the southern Gulf of St. Lawrence. ICES J. Mar. Sci. 51:493-504. Tolimieri, N. 2007. Patterns of species richness, species density, and evenness in groundfish assemblages on the continental slope of the U.S. Pacific coast. Environ. Biol. Fish. 78:241-256. Tolimieri, N., and P. S. Levin. 2006. Assemblage structure of eastern Pacific ground- fishes on the U.S. continental slope in relation to physi- cal and environmental variables. Trans. Am. Fish. Soc. 135:317-332. Trenkel, V. M., and J. Cotter. 2009. Choosing survey time series for populations as part of an ecosystem approach to fishery manage- ment. Aquat. Living Resour. 22:121-126. Worm, B., R. Hilborn, J. K. Baum, T. A. Branch, J. S. Collie, C. Costello, M. J. Fogarty, E. A. Fulton, J. A. Hutchings, S. Jennings, O. P. Jensen, H. K. Lotze, P. M. Mace, T. R. McCla- nahan, C. Minto, S. R. Palumbi, D. Ricard, A. A. Rosenberg, R. Watson, and D. Zeller. 2009. Rebuilding global fisheries. Science 325:578-585. Zheng, J., and G. H. Kruse. 2006. Recruitment variation of eastern Bering Sea crabs: climate-forcing or top-down effects? Prog. Oceanogr. 68:184-204. 223 A note on the von Bertalanffy growth function concerning the allocation of surplus energy to reproduction Shuhei Ohnishi (contact author)1 Takashi Yamakawa2 Hiroshi Okamura3 Tatsuro Akamine4 Email address for contact author: ohnishi @scc.u-tokai.ac. ip 1 School of Marine Science and Technology Tokai University Shimizu, Shizuoka 424 8610, Japan 2 Graduate School of Agricultural and Life Sciences The University of Tokyo Bunkyo, Tokyo 1 13-8657, Japan 3 National Research Institute of Far Seas Fisheries Fisheries Research Agency Yokohama, Kanagawa 236-8648, Japan 4 National Research Institute of Fisheries Science Fisheries Research Agency Yokohama, Kanagawa 236-8648, Japan Abstract— We propose an extended form of the von Bertalanffy growth function (VBGF), where the allocation of surplus energy to reproduction is considered. Any function can be used in our model to describe the ratio of energy allocation for reproduction to that for somatic growth. As an exam- ple, two models for energy allocation were derived: a step-function and a logistic function. The extended model can jointly describe growth in adult and juvenile stages. The change in growth rate between the two stages can be either gradual or steep; the latter gives a biphasic VBGF. The results of curve fitting indicated that a consideration of reproduc- tive energy is meaningful for model extension. By controlling parameter values, our comprehensive model gives various growth curve shapes ranging from indeterminate to determinate growth. An increase in the number of parameters is unavoidable in practical applications of this new model. Addi- tional information on reproduction will improve the reliability of model estimates. Manuscript submitted 17 February 2011. Manuscript accepted 5 December 2011. Fish. Bull. 110:223-229(2012). The views and opinions expressed or implied in this article are those of the author (or authors) and do not necessarily reflect the position of the National Marine Fisheries Service, NOAA. The von Bertalanffy growth function (VBGF) has been used to analyze somatic growth data in a wide range of studies. It is now commonly put into practice to partially reparameterize the VBGF parameters to avoid their covariation and to ensure statistical accuracy (Quinn and Deriso, 1999). Although variations in the growth rate influenced by extrinsic envi- ronmental fluctuations have been examined in many studies, we pro- pose that intrinsic physiological dy- namics are also of great importance. The interaction between growth rate and sexual maturation has often been debated in life history studies (Roff, 1984; Beverton, 1992; Jensen, 1996). Mathematical treatments for representing the switch in energy allocation between growth and re- production have been introduced to discuss optimal life-history strategies (Roff, 1983; Kozlowski, 1992, 1996; Kozlowski and Teriokhin, 1999). When the energy budget of fishes has been quantified, dynamics of the energy distribution between growth and reproduction have often been considered in simulated scenarios (Jorgensen and Fiksen, 2006; Pec- querie et ah, 2009). The effects of reproductive energy should also be important for practi- cal curve fitting studies. A biphasic growth model derived by connect- ing two independent VBGFs at an arbitrary age is often employed for curve fitting (Soriano et ah, 1992; Porch et ah, 2002; Araya and Cu- billos, 2006; Quince et ah, 2008a, 2008b; Alos et ah, 2010; Tribuzio et ah, 2010). Although a biphasic VBGF is one approach used to account for inflections in growth and is similar to the higher-parameter model (Sch- nute and Richards, 1990), results of model selection based on the Akaike information criterion (AIC; Akaike, 1973) often indicate that the bipha- sic VBGF is a more suitable model than the original monophasic VBGF (Porch et ah, 2002; Araya and Cubil- los, 2006; Tribuzio et ah, 2010). The better fit implies that the delay in growth due to a reallocation of en- ergy may be detected as a change in the growth trajectory. 224 Fishery Bulletin 1 10(2) Day and Taylor (1997) and Czarnot^ski and Kozlowski (1998) identified the lack of an explicit formula for the reproductive process in the VBGF. Although the bipha- sic VBGF is an empirical approach, a deductive model that can incorporate both growth and reproduction should be developed to help to understand the process of energy allocation and to improve curve fit. In this study, we begin with an extension of the VBGF with respect to a continuous change in energy allocation. We also present an application of curve fitting and model selec- tion. An overview of changes in growth-curve shapes is subsequently shown. Finally, we briefly discuss the features of our model. Methods We start with the general form of VBGF given by = hw2/> - kw, (1) dt where w, t, h, and k are body weight, age, and coefficients of anabolism and catabolism, respectively. The right hand side of Equation 1 is the total production rate of surplus energy. If we consider the reallocation of surplus energy for reproduction. Equation 1 can be expanded as — + c- = hw2/3-kw. (2) dt dt Two newly introduced terms, f and c, denote the cumula- tive energy investment for reproduction until age t and the conversion factor of reproductive energy to body weight, respectively. Note that f is not equivalent to the weight of the gamete (i.e., eggs or spermatozoa). Equa- tion 2 is equivalent to the exoskeleton growth model (Ohnishi and Akamine, 2006) in that energy allocation to activities or appendages unrelated to somatic growth are explicitly described. Suppose w=fH3, where l is body length and /3 is a con- stant proportionality coefficient. Dividing dw/dt=3pl2dl/ dt by each side of Equation 2 and substituting w=f3l3 yields the following equation: ctt dt dw df 'l + c — — dt dt ) K(L-l), (3) where K=k/3 and loa(=hk~1j3~1/3) is the asymptotic length. Let p be the ratio (0trn), where 0tm ) Equation 7 represents the time delay to attain a cer- tain body size in t>tm due to change in energy alloca- tion. Consequently, the growth curve becomes biphasic, combining two independent VBGFs. The alternative model assumes that pit) changes continuously throughout an individual’s lifetime. In particular, an S-type curve that has an inflection point around t=tm is suitable for describing a change in pit) due to sexual maturation. Let pit) be p(D = u/(l+exp(- ait-tm))) as a general logistic curve such that the ana- lytical solution for Tit) is given by T(t) = (l-o)(t-t„)- (g) where u and a are the upper limit of the allocation rate in reproductive energy and the rapidity of maturation, respectively. The logistic function converges to a step- function when a—> By inspection, Equation 7 is a special case of Equation 8. The solution for Equation 5 is complicated when pspil) (or p^piw)) such that |0£i,<" ,9) The explicit solution for / is a biphasic VBGF when pit) is a step-function that has discontinuous switching at Ohmshi et al.: The von Bertalanffy growth function concerning the allocation of surplus energy to reproduction 225 Table 1 Parameter estimates for two types of von Bertalanffy growth functions (VBGFs). Both types of VBGF have three common parameters: asymptotic length (/„), growth coefficient (K), and initial condition of age (t0). Additional parameters, namely age at maturity (tm), the upper limit of the allocation rate in reproductive energy (u), and rapidity of maturation (a), were used in the extended model. Values within parentheses show the square root of the variance of the estimates derived from the matrix inverse of the Hessian matrix. The 4AIC shows the relative difference of the Akaike information criterion (AIC) value compared with the minimum AIC. Type of VBGF L K ^0 V a AIC 4AIC Conventional model 260.72 0.34 -0.45 — — — 26881.3 20.8 (1.51) (0.01) (0.04) Extended model 463.01 0.15 -0.77 3.41 0.79 1.01 26860.5 0 (49.08) (0.01) (0.14) (0.35) (0.05) (0.21) the boundary of mature “size.” In most cases, however, Results it is not easy to obtain an explicit solution, as shown in Equation 6, owing to the complexity of the integrand in Equation 9. Allocated reproductive energy can be derived as fol- lows by rearranging Equation 4 with the condition p^pit) as (i-PMEf-PWf. Substituting dwldt=3fH2dlldt and Equation 5 into this equation, one obtains f = 3-frW(L-D, f = ^~ J'p(s)/2 {L-l)ds. (10) (11) Equation 10 represents the instantaneous reproduc- tive energy at age t. Equation 11 shows the cumulative investment of reproductive energy until age t. Thus, changes in two quantities (body size in Eq. 6 and energy investment in Eq. 10) are treated in an extended VBGF. We fitted the growth curve in Equations 6 and 8 to individual measurements in length-at-age as Ll(i = 1,..., AD, where N is the total number of samples. Param- eters were estimated by minimizing the residual sum of squares ofS=X, iL-lf2. The numerical optimization for S was accomplished by using the quasi-Newton method (BFGS algorithm) in “optim( )” with R statistical soft- ware (R Development Core Team, 2011). The comparison between this model and the original monophasic VBGF was based on the AIC value as follows: AIC=ATogS+20, where 8 is the number of free parameters. We used measurement data on willowy flounder ( Tanakius kitaharai) males collected by bottom-trawl surveys in the coastal area of Fukushima Prefecture, Japan, from 2004 to 2006. The sample size was n=2169. Age ranged from 1.38 to 14.30 years and length ranged from 113 to 298 mm (standard length). Otoliths were used to determine yearly ages and dates of birth were assigned as January 1st. Results for curve fitting and model selection are sum- marized in Table 1 and Figure 1. As shown in Table 1, the AIC difference (AAIC=20.8) between the two types of VBGF suggests that the trajectory given by the extended model more appropriately described lifetime growth. This result implies that a consideration of reproductive energy can be meaningful for model extensions. The variance of two common parameters (i.e., and t0) in the extended model was larger than that in the conventional VBGF (Table 1). Twelve types of energy allocation schedules, pit), and the corresponding somatic growth (in length /) based on different combinations of parameter values in Equa- tion 8 (u = Q.4, 0.6, 0.8, 1.0 and a- 1, 3, 100) are shown in Figure 2. The behavior of df/dt and f describing the energy investment in reproduction is shown in Figure 3. When v = 0, the growth curve is identical to the origi- nal VBGF (Fig. 2). Although somatic growth curves generated by lower v values (i.e., v = 0.4, Fig. 2, A-C) do not differ substantially from the original VBGF, there are distinctive differences for shapes with higher v values (i.e., u = 0.8, 1.0, Fig. 2, G-L). In these cases, the somatic growth of the adult and juvenile stages can be clearly distinguished. Gradual but steady growth after maturation is typical with indeterminate growth (Fig. 2, A-I). We can see a continuous phase shift of inde- terminate growth in Figure 2. When u = 1.0, the growth rate after maturation converges to zero because most surplus energy is devoted to reproduction, generating more determinate growth (Fig. 2, J-L). The variation in a leads to a difference in the degree of continuity of growth rate during the sexual matura- tion transition period (Fig. 2, A, D, G, and J vs. Fig. 2, C, F, I, and L). The curves given by sufficiently high a (a = 100) represent biphasic VBGF resulting from an abrupt change in growth rate around age tm (Fig. 2, C, F, I, and L). In Figure 3, an apex can be found on the convex shape of df/dt, and the height and degree of curvature changes according to the values of v and a. An increase in the value of v raises the reproductive investment df! 226 Fishery Bulletin 1 10(2) Figure 1 Two types of growth curve estimated with the conventional von Bertalanffy growth function (A) and with the extended von Bertalanffy growth function (B) for individual measurement data of the willowy flounder ( Tanakius kitaharai ) male (n = 2169). dt and shifts the maximum df/dt to older ages. When u = 1.0 (Fig. 3, J-L), df/dt converges to a constant value after maturation as a result of determinate growth and constant surplus energy, defined by Equation 2. Lower a values show a slower initial rise in df/dt around tm (Fig. 3, A, D, G, and J), whereas higher a values yield a steeper initial rise in df/dt around tm (Fig. 3, C, F, I, and L). Discussion A notable feature of our mode! is that energy allocation can be quantified by the arbitrary functional form p(-). The introduction of p(-) provides a unified platform to treat the trade-off between somatic growth and repro- duction. The extended model can jointly describe adult and juvenile growth. The change in growth rate between the two stages can be either gradual or steep, with the latter case showing a biphasic VBGF. By controlling the value of p(-), our comprehensive model yields various shapes of growth curves that range from indeterminate to determinate growth. Therefore, our model can be used for life history studies, as well as practical curve fitting studies. When allocation dynamics are not fully described by a simple model, such as seen in Equation 8, additional parameters beyond a, u, and tm, or a par- ticularly designed form ofp(-) would be useful for further model development. The extended VBGF in Equation 5 can theoretically incorporate an unlimited number of parameters. How- ever, an increase in the number of free parameters in p(-) will be disadvantageous for model estimation because the functional form of p(-) does not directly appear in the age-length relationship. Increases in the variance of estimates imply instability due to curve fitting (Table 1). Therefore, it is necessary to consider methods of overcoming the trade-off between an in- creased number of parameters and estimation stability. Data sets other than those for length-at-age data will be useful for estimating the parameters in p(-) because the dynamics of p(-) are readily apparent in the behav- ior of df/dt (Fig. 3) rather than in length (Fig. 2). We expect that the robustness of this estimation will be improved by means of a combined likelihood-function (Martin and Cook, 1990; Eveson et ah, 2004) described by two heterogeneous relationships: length-at-age and reproductive energy-at-age. Our model development has similarities to that of Lester et al. (2004). Both studies explicitly give a growth function that can quantify a delay in somatic growth due to reproductive energy allocation. Lester et al. (2004) initially assumed a linear function of pre- mature growth in length and derived the conventional VBGF by introducing an intensive energy investment at postmature ages. Additionally, Lester et al. (2004) assumed that the ratio of gonad to body weight at postmature ages is constant. This assumption causes the linear function to yield a delay in growth after maturation equivalent to that yielded with the VBGF. Alternatively, our model derivation started from the VBGF. Additional hypotheses for model formulation other than ours and those of Lester et al. (2004) are possible. Hence, the adequacy of these assumptions for model derivation must be evaluated with a wide range of practical applications. Ohnishi et al.: The von Bertalanffy growth function concerning the allocation of surplus energy to reproduction 227 A ( 0 4 . 1 ) B ( 0.4 , 3 ) C ( 0 4 . 100 ) Age (years) Figure 2 Variation in somatic growth due to differences in energy allocation (A-L). Thick, thin, and broken lines correspond to growth in length lit), growth in length lit) with u = 0 (i.e., the original von Bertalanffy growth function), and the energy allocation function pit), respectively. The numerical values in parentheses correspond to the combination of parameters r(=0.4, 0.6, 0.8, 1.0) and a( = l, 3, 100) employed in pit). The parameters used for all cases include the following: fl=1.0, c = 1.0, A = 0.15, 7^ = 1.0, t0= 0, and tm = 5. Acknowledgments The authors would like to thank to Y. Narimatsu of Tohoku National Fisheries Research Institute, Fisher- ies Research Agency, who provided us with the willowy flounder data for analysis. The authors also express our gratitude to the anonymous reviewers who provided valuable, insightful suggestions for the improvement of the manuscript. 228 Fishery Bulletin 1 10(2) o 0 N in tt> 3 CD l3 0 5 10 15 20 25 300 5 10 15 20 25 300 5 10 15 20 25 30 Age (years) Figure 3 Variation in variables related to reproduction that result from differences in energy allocation (A-L). Thick, thin, and broken lines correspond to the instantaneous reproductive energy df/dt, cumulative energy investment fit), and the energy alloca- tion function pit), respectively. The numerical values in parentheses correspond to the combination of parameters i>(=0.4, 0.6, 0.8, 1.0) and a( = l, 3, 100) employed in pit). The parameters used for all cases include the following: /3= 1.0, c = 1.0, /f=0.15, Z„=1.0, t0=0 , and tm = 5. Literature cited Akaike, H. 1973. Information theory and an extension of the maximum likelihood principle. In 2nd international symposium on information theory (B. N. Petrov, and F. Csaki, eds.), p. 267-281. Akademiai Kiado, Budapest. Aids, J., M. Palmer, A. Alonso-Fernandez, and B. Morales-Nin. 2010. Individual variability and sex-related differ- ences in the growth of Diplodus annularis (Linnaeus, 1758). Fish. Res. 101:60-69. Araya, M., and L. A. Cubillos. 2006. Evidence of two-phase growth in el as mo - branchs. Environ. Biol. Fish. 77: 293-300. Beverton, R. J. H. 1992. Patterns of reproductive strategy parameters in some marine teleost fishes. J. Fish Biol. 41:137- 160. Czarnol?ski, M., and J. Kozlowski. 1998. Do Bertalanffy’s growth curves result from optimal resource allocation ? Ecol. Lett. 1:5-7. Day, T., and P. D. Taylor. 1997. Von Bertalanffy’s growth equation should not Ohnishi et al The von Bertalanffy growth function concerning the allocation of surplus energy to reproduction 229 be used to model age and size at maturity. Am. Nat. 149:381-393. Eveson, J. P., G. M. Laslett, and T. Polacheck. 2004. An integrated model for growth incorporating tag-recapture, length-frequency, and direct aging data. Can. J. Fish. Aquat. Sci. 61:292-306. Jensen, A. L. 1996. Beverton and Holt life history invariants result from optimal trade-off of reproduction and survival. Can. J. Fish. Aquat. Sci. 53:820-822. Jorgensen, C., and 0. Fiksen. 2006. State-dependent energy allocation in cod ( Gadus morhua). Can. J. Fish. Aquat. Sci. 63:186-199. Kozlowski, J. 1992. Optimal allocation of resources to growth and reproduction: Implications for age and size at matu- rity. Trends Ecol. Evol. 7:15-19. 1996. Optimal allocation of resources explains inter- specific life-history patterns in animals with inde- terminate growth. Proc. R. Soc. Lond. B 263:559 -566. Kozlowski, J., and A. T. Teriokhin. 1999. Allocation of energy between growth and reproduc- tion: The Pontryagin Maximum Principle solution for the case of age- and season-dependent mortality. Evol. Ecol. Res. 1:423-441. Lester, N. P., B. J. Shuter, and P. A. Abrams. 2004. Interpreting the von Bertalanffy model of somatic growth in fishes: the cost of reproduction. Proc. R. Soc. Lond. B 271:1625-1631. Lincoln, R. J., G. A. Boxshall, and P. F. Clark. 1998. A dictionary of ecology, evolution and systematics, 361 p. Cambridge Univ. Press, New York. Martin, I., and R. M. Cook. 1990. Combined analysis of length and age-at-length data. J. Cons. Int. Explor. Mer 46:178-186. Ohnishi, S., and T. Akamine. 2006. Extension of von Bertalanffy growth model incor- porating growth patterns of soft and hard tissues in bivalve molluscs. Fish. Sci. 72:787-795. Pecquerie, L., P. Petitgas, and S. A. L. M. Kooijman. 2009. Modeling fish growth and reproduction in the con- text of the Dynamic Energy Budget theory to predict environmental impact on anchovy spawning duration. J. Sea Res. 62:93-105. Porch, C. E., C. A. Wilson, and D. L. Nieland. 2002. A new growth model for red drum ( Sciaenops ocella- tus) that accommodates seasonal and ontogenic changes in growth rates. Fish. Bull. 100:149-152. Quince, C., P. A. Abrams, B. J. Shuter, and N. P. Lester. 2008a. Biphasic growth in fish 1: theoretical founda- tions. J. Theor. Biol. 254:197-206. Quince, C., B. J. Shuter, P. A. Abrams, and N. P. Lester. 2008b. Biphasic growth in fish II: empirical assess- ment. J. Theor. Biol. 254:207-214. Quinn, T. J., II, and R. B. Deriso. 1999. Quantitative fish dynamics, 542 p. Oxford Univ. Press, New York. R Development Core Team. 2011. R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. [Available from: URL http://www.R- proiect.org] Roff, D. A. 1983. An allocation model of growth and reproduction in fish. Can. J. Fish. Aquat. Sci. 40:1395-1404. 1984. The evolution of life history parameters in teleosts. Can. J. Fish. Aquat. Sci. 41:989-1000. Schnute, J. T., and L. J. Richards. 1990. A unified approach to the analysis of fish growth, maturity, and survivorship data. Can. J. Fish. Aquat. Sci. 47:24-40. Soriano, M., J. Moreau, J. M. Hoenig, and D. Pauly. 1992. New functions for the analysis of two-phase growth of juvenile and adult fishes, with application to Nile perch. Trans. Am. Fish. Soc. 121:486-493. Tribuzio, C. A., G. H. Kruse, and J. T. Fujioka. 2010. Age and growth of spiny dogfish ( Squalus acanthias) in the Gulf of Alaska: analysis of alternative growth models. Fish. Bull. 108:119-135. 230 Abstract — Gillnet mesh selectivity parameters were estimated for juve- nile blacktip sharks ( Carcharhinus limbatus) by using length data from an experimental fishery-independent gillnet survey in the northeastern Gulf of Mexico. Length data for 1720 blacktip sharks were collected over 17 years (1994-2010) with seven mesh sizes ranging from 7.6 to 20.3 cm. Four selectivity models, a normal model assuming fixed spread, a normal model assuming that spread is proportional to mesh size, a log- normal model, and a gamma model were fitted to the data by using the SELECT (share each length’s catch total) method. Each model was run twice under separate assumptions of 1) equal fishing intensity; and 2) fishing intensity proportional to mesh size. The normal, fixed-spread selec- tivity curve where fishing intensity is assumed to be proportional to mesh size provided the best fit to the data according to model deviance estimates and was chosen as the best model. Results indicate that juvenile blacktip sharks are susceptible as bycatch in some commercial gillnet fisheries. Manuscript submitted 23 March 2011. Manuscript accepted 23 December 2011. Fish. Bull. 110:230-241 (2012). The views and opinions expressed or implied in this article are those of the author (or authors) and do not necessarily reflect the position of the National Marine Fisheries Service, NOAA. Gillnet selectivity for juvenile blacktip sharks ( Carcharhinus limbatus ) Ivy E. Baremore (contact author)1 Dana M. Bethea1 Kate I. Andrews2 Email address for contact author: Ivy. Baremore (Sjnoaa.qov 1 Panama City Laboratory Southeast Fisheries Science Center National Marine Fisheries Service National Oceanic and Atmospheric Administration 3500 Delwood Beach Rd Panama City, Florida 32408 2 Beaufort Laboratory Southeast Fisheries Science Center National Marine Fisheries Service National Oceanic and Atmospheric Administration 101 Pivers Island Road Beaufort, North Carolina 28516-9722 In the late 1980s, a gillnet fishery for sharks developed in the Atlantic Ocean off the coasts of Florida and Georgia (Trent et ah, 1997). Fishing area varied with seasons, and shark drift gillnet vessels operated in near- shore waters between 4.8 and 14.4 km offshore, ranging from West Palm Beach, Florida (~26°46'N), to Alta- maha Sound, Georgia (~31°45'N). A variety of methods were used to deploy gillnets, including drifting the net on the surface (Trent et ah, 1997), strik- ing around a school of sharks (Carlson and Baremore1), and anchoring the net to the bottom (Carlson and Bethea, 2007). Fishermen targeted a variety of coastal species of sharks, from black- tip sharks (Carcharhinus limbatus) to Atlantic sharpnose sharks (Rhi- zoprionodon terraenouae ) depending on market conditions and fishery clo- sures. Over the last 10 years, the size and scope of the commercial shark gillnet fishery has decreased primar- 1 Carlson, J. K., and I. E. Baremore. 2003. The directed shark gillnet fish- ery: catch and bycatch 2003, NOAA Sus- tainable Fisheries Division Contribution PCB-03/07, 8 p. Panama City Labora- tory, National Marine Fisheries Service, Panama City, Florida. (Available from http://www.sefsc.noaa.gov/labs/panama/ documents/observer documents/gillnet/ SDG2003.pdf. accessed December 2011.] ily owing to regulations that restrict gear, fishing areas, and trip limits for sharks. In 2008, Amendment 2 to the Consolidated Atlantic Highly Migra- tory Species Fishery Management Plan (NMFS, 2008) limited landings of large coastal sharks to 33 sharks per trip. The high cost of fuel and low market value for shark meat, in conjunction with these regulations, caused most commercial fishermen in the U.S. south Atlantic Ocean to abandon the gillnet fishery for sharks. Although shark-targeted gillnet trips are currently rare in the U.S. Atlantic Ocean, blacktip sharks are still caught as bycatch in other gill- net fisheries that target species such as Spanish mackerel ( Scomberomorus maculatus ) and king mackerel (S. ca- valla) (Passerotti et ah, 2010; Thorpe and Frierson, 2009). These coastal teleost gillnet fisheries are expansive, and had more than 65 active fishing vessels in 2010. 2 The fishing locations of these vessels span the U.S. east coast throughout the range of the blacktip shark. The blacktip shark is a cosmopoli- tan species, ranging from Massachu- 2 Southeast Fisheries Science Center Coastal Fisheries Logbook, available at http://www.sefsc.noaa.gov/fisheries/ reporting.htm. accessed March 2011. Baremore et at: Gillnet selectivity for |uvenile Carcharhinus limbatus 231 setts throughout the Gulf of Mexico in U.S. coastal wa- ters (McEachran and Fechhelm, 1998). Juvenile blacktip sharks use nursery areas such as bays and nearshore habitats during spring and summer months (Castro, 1993; Heupel and Hueter, 2002). Because of their range and life history characteristics, juvenile blacktip sharks are likely to encounter commercial gillnets. Gillnet selection curves are a useful way to represent the retention probabilities of different mesh sizes for a given species of fish. Retention probability by gillnets is usually considered to be dome-shaped and can be described by the equation where r-(Z-) is the retention probability that a fish of length l in size class i is caught by mesh size j, and n and a represent the mean and spread of the curve (Millar and Fryer, 1999). However, the selection curve may be skewed because of snagging, rolling, or entangling of animals, and can result in a gamma or lognormal curve (Millar and Fryer, 1999). Generally, selectivity can be measured in two ways: directly and indirectly (Millar and Fryer, 1999). Direct experiments are performed on a population for which the size distribution is known, and size selection is calculated by comparison of the population with the catch distributions. Indirect, or comparative, experi- ments are more common and usually involve simulta- neously fishing gillnets of differing mesh sizes with equal effort. Commercial fishing gear selectivity curves are in- corporated into modern stock assessment models, and changes in the parameters have the potential to impact the assessed status of the stock (Maunder, 2002). Size selectivity is used in the estimation of the length-frequency of a stock, estimation of fishing-in- duced mortality, and in age-based assessment models (Millar and Fryer, 1999). Although important for the stock assessment models, fishery-independent selec- tivity models are rare for many large shark species (McAuley et al., 2007). Selectivity for bycatch species is also becoming an important issue in stock assess- ment, but direct estimates are likewise rare for most fisheries. The goal of this study is to determine the relationship between gillnet mesh size and selectivity for juvenile blacktip sharks using fishery-independent data. Materials and methods Sampling Data necessary for indirect calculation of gillnet mesh selectivities were obtained from the Gulf of Mexico Shark Pupping and Nursery (GULFSPAN) survey, which is a fishery-independent gillnet survey of coastal shark populations in the northeastern Gulf of Mexico (Carlson and Brusher, 1999). Catch data for C. lim- batus were collated over 17 years (1994-2010) from five bay systems in northwest Florida: St. Andrew Bay, Crooked Island Sound, St. Joseph Bay, the gulf side of St. Vincent Island, and Apalachicola Bay (Fig. 1). Six gillnet panels of differing stretched mesh sizes were strung together in increasing mesh size, anchored, and fished concurrently as a single gillnet. Each panel was 30.1 m long and 3.4 m deep (Table 1). From 1994 through 2005, stretched mesh sizes ranged from 8.9 cm to 14.0 cm, increasing by 1.3-cm (0.5-in) intervals, with an additional panel of 20.3 cm. In 2006, the 20.3-cm panel was removed and a 7.6-cm panel was added ad hoc. The largest mesh panel was removed because of its historically low catch of juvenile small coastal shark species, and the 7.6-cm panel was added to increase catch of small neonatal sharks. Unless otherwise indi- cated, all mesh sizes reported in the present study are stretched mesh sizes. Sampling occurred each year from late March through October. Net set locations within bay sys- tems were randomly chosen over a variety of habitat and depth combinations. The majority of sets were short (<1 hr) as a means of reducing mortality, es- pecially when water temperatures were above 25°C. However, some nets were soaked for longer periods of time, depending on the research priorities at the time. Captured sharks were removed from the net, their sex was determined, and they were measured for fork length (FL) on a rigid measuring board in a straight line from the tip of the nose to the fork in the tail. Sharks in poor condition were sacrificed for research projects and those in good condition were tagged and released. Maturity state was determined by clasper calcification for males, internal examina- tion for sacrificed female sharks, and released females were considered to be mature when greater than 115 cm FL (Carlson et al., 2006). Sexes were combined for data analyses. Data analysis Catch data were pooled by mesh size into 5-cm-FL size bins, and the midpoint of each size class (/,) was used to calculate a selectivity curve for each mesh size (Millar and Holst, 1997). Four gillnet selectivity models were fitted to the /( for each mesh size (m) (Millar and Holst, 1997), by using the SELECT (share each length’s catch total) method (Millar and Holst, 1997; Millar and Fryer, 1999; Millar, 2003, 2010). The selection curves were fitted to the data by using the “gillnetfunctions” package in R statistical software (Millar, 2003, 2010; R Develop- ment Core Team, 2009). The SELECT method applies the method of maximum likelihood, which estimates selectivity parameters from a general log-linear model. The expected catch of sharks of length class i in gillnet j is described by vu = PArj’ (1) 232 Fishery Bulletin 1 10(2) 85°30,0’,w 84°20'0"W Figure 1 Location of the Gulf of Mexico Shark Pupping and Nursery (GULFSPAN) survey in northwest Florida. Sampling sites were located in St. Andrew Bay, Crooked Island Sound, St. Joseph Bay, the gulf side of St. Vincent Island, and Apalachicola Bay. Sampling occurred from 1994 through 2010. Table 1 Gillnet specifications used in the Gulf of Mexico Shark Pupping and Nursery (GULFSPAN) survey 1994-2010. For all net con- figurations, the hanging ratio (length to height ratio of the meshes) was 0.5, leadline weight was 4.5 kg, 2.3 kg of buoyancy was used, and each panel length was 30.1 m. Stretch mesh size (in/cm) Twine size no. No. of meshes deep Thickness of twine (mm) Breaking strength (kg) Years fished 3. 0/7.6 208 45 0.52 11.8 2006-2010 3. 5/8. 9 208 40 0.52 11.8 1994-2010 4.0/10.2 208 35 0.52 11.8 1994-2010 4.5/11.4 208 35 0.52 11.8 1994-2010 5.0/12.7 277 30 0.62 18.2 1994-2010 5.5/14.0 277 25 0.62 18.2 1994-2010 8.0/20.3 24 20 1.00 115.9 1994-2005 where pf = the relative fishing intensity of gillnet j: A- = the abundance of sharks in length class i; and r = the selection curve for each gillnet j. Relative fishing intensity represents fishing effort and fishing intensity combined and is the conditional prob- ability that a fish contacted gillnet pane! j, with the assumption that it made single contact with the entire combined gillnet panel (Millar, 1992). The normal, gamma, and lognormal models observe geometric simi- larity (mean p • and spread a, proportional to mesh size), whereas the normal model with fixed spread is not geo- metrically similar (mean p ■ and spread cr equal across mesh sizes). When p; is assumed to be equal among mesh sizes, the form of the log-linear model is as follows: 1 o g ( v J = fac to r ( /, ) + /3, fx ( m J , j ) + /J2 ■ f2 ( m J , j ) , ( 2 ) Baremore et at. Gillnet selectivity for |uvenile Carcharhmus limbatus 233 Table 2 Selectivity curves for normal, gamma, and lognormal models used to estimate gillnet selectivity for blacktip sharks (Carcharhi- nus limbatus ): m/ is the mesh size for panel j ( j= 1-7 panels) and /( is the midpoint of length class i (i= 1-22 length classes). Rela- tive fishing intensity is modeled separately. Equations in the right hand column are the last two terms in the log-linear model Model Selection curve A /i("vO+A Normal: fixed spread exp ( / \2 (, l,-k m J 2cx 2 Normal: proportional spread exp {l, ~ai mj)~ 2 a2 2 Gamma: proportional spread /, (a-l)-k-m a- 1 - [«-!]■ /, m j Lognormal: proportional spread -exp ^+logl H y log(/,)-p, - log 2d"1 ^ | log (Z,.)- log — 1 --log' m, J 2 Table 3 Equations used to estimate the modal length of blacktip sharks (Carcharhinus limbatus) caught with each gillnet mesh size {mf for all four gillnet selectivity models. Model Mode Normal (fixed and proportional spread) Mode (m])=k-mJ Gamma Mode (m )=(a- 1) • k-m} Lognormal f m \ Mode (m,)=exp(ii-a2) • — J ^ r l ml ) where factor (l J indicates that length class is fitted as a factor of the model, and fx(mt, j) and f2(mr j ) are the selectivity functions of rrij and j (right hand column of Table 2). When p is assumed to be proportional to mesh size (logp-=logm-j, the form of the log-linear model is as follows: l°g ( vu ) = log (my) + factor (/,) + Pif,(mJ,j) + l32f2(mJj). The parameters /3; and /3, are related to the form of the selectivity curve and are defined in Table 2. The follow- ing assumptions were made for all models: 1) catches were independent; and 2) gillnet panels were fished with equal effort. The mode, or maximum selected size for each panel, was calculated according to equations listed in Table 3. All models were fitted to the data twice, once under the assumption of equal fishing intensity and again under the assumption of fishing intensity proportional to mesh size. Overdispersion, or lack of fit, was tested by calculating the dispersion parameter, which is the model deviance divided by the degrees of freedom. When the dispersion parameter is >1, data are considered to be overdispersed. Results A total of 1720 blacktip sharks were measured from 1994 through 2010 (Table 4). Blacktip sharks were collected during 14 of the 17 years of the survey. Average net soak time was 2.67 hr (range: 0.17-23.83 hr) over 1573 sets. Some outliers were excluded when sampling protocol was considered to be out of the ordinary procedure. The majority (97%) of blacktip sharks caught in all panels were immature and less than 110 cm FL (mode = 65 cm FL, Fig. 2). There was a general increase in the mean size of blacktip sharks with increasing mesh size. For the panels that were fished concurrently for all years (8.9-14.0 cm mesh), the total sample sizes of measured sharks were similar (Table 4). 234 Fishery Bulletin 1 10(2) Table 4 Length distribution for all blacktip sharks ( Carcharhinus limbatus ) caught in each gillnet mesh panel in the Gulf of Mexico Shark Pupping and Nursery (GULFSPAN) Survey in northwest Florida, 1994-2010. Fork Mesh sizes (cm) lengin (cm) 7.6 8.9 10.2 11.4 12.7 14.0 20. 42.5 4 4 4 0 1 0 0 47.5 2 10 14 18 6 4 0 52.5 13 25 29 57 21 22 2 57.5 13 20 32 53 30 43 3 62.5 12 24 21 41 63 41 0 67.5 7 30 31 29 47 31 4 72.5 6 26 30 34 20 32 6 77.5 13 36 34 21 30 20 5 82.5 8 15 20 22 17 28 7 87.5 4 29 24 34 25 16 9 92.5 7 15 17 20 14 15 9 97.5 0 14 12 15 13 14 18 102.5 2 4 16 15 9 8 17 107.5 0 10 5 1 5 4 12 112.5 0 1 0 2 4 0 6 117.5 0 2 0 4 3 0 8 122.5 0 2 1 1 0 0 4 127.5 0 0 0 1 0 0 2 132.5 0 0 0 0 0 0 0 137.5 0 0 0 0 0 0 2 142.5 0 0 0 1 0 0 1 147.5 1 0 1 0 0 0 0 Totals 92 267 291 369 308 278 115 The normal, fixed-spread models had the lowest model deviance overall, with the model incorporating fishing intensity proportional to mesh size having the low- est total model deviance (Fig. 3, Table 5). The ratio of model deviance to degrees of freedom was higher than 1 (2.9), indicating overdispersion of the data. This result indicates that blacktip sharks may not have behaved independently (e.g., with schooling behavior), violating the first assumption of independent catches. Residual plots showed a similar degree of bias for all models (Fig. 3), with none demonstrating markedly different fits to the data. The biggest difference among models was for the largest mesh (20.3 cm) for which the normal (proportional spread), lognormal, and gamma curves under-represented some of the smaller length classes. The highest number of positive residuals was seen for the smaller length classes (50-70 cm FL) in mesh sizes 11.4 cm and 12.7 cm and, to a lesser degree, the 14.0 cm panel for all models (Fig. 3). The plots indicated that more of the smaller individuals were caught in these panels than predicted by the models. The largest and smallest mesh sizes (20.3 and 7.6 cm) caught fewer of the smallest sharks than predicted by the models. The residuals did not indicate systematic bias in any of the models aside from the lack of fit to the smallest size classes (Fig. 3). Predicted selectivity curves for the normal, fixed-spread model assuming proportional fish- ing intensity plotted with observed length-frequencies for each mesh size (Fig. 4) showed that the model fitted the observed data well. Discussion In previous gillnet selectivity studies on sharks, a gamma-shaped distribution has been assumed (Carlson and Cortes, 2003; Kirkwood and Walker, 1986; McLoughlin and Stevens, 1994; Simpfendorfer and Unsworth, 1998), based on the specialized SELECT method described by Kirkwood and Walker (1986). However, a more recent study on the gillnet selectivity for sandbar sharks C. plumbeus (McAuley et al., 2007) found that all four models estimated by the Millar and Holst (1997) method provided better fits than the Kirkwood and Walker (1986) gamma model. Our study on blacktip sharks indicated that the normal, fixed spread models provided the best fit. A more limited study in North Caro- lina (Thorpe and Frierson, 2009) found that the normal model with spread proportional to mesh size generally provided the best fit for blacknose (C. acronotus), bonnethead ( Sphyrna tiburo), and blacktip sharks. Although the method of Kirkwood and Walker (1986) was not employed in this study, the gamma curve estimated by the Millar and Holst (1997) SELECT method provided a poorer fit than the normal and lognormal models. Therefore, it Baremore et at: Gillnet selectivity for juvenile Carcharhinus hmbatus 235 Table 5 Gillnet selectivity parameter estimates for each model for blacktip sharks ( Carcharhinus limbatus ) in northwest Florida, 1994- 2010. All four models were run twice: first assuming fishing intensity to be equal across mesh sizes and again assuming that fishing intensity was proportional to mesh size. Model deviance is the likelihood ratio goodness of fit, with 130 degrees of freedom for each model. Equal fishing intensity Proportional fishing intensity Model Parameters Model deviance Parameters Model deviance Normal (fixed spread) (k, ct) = (5.98, 30.98) 411.79 (k, cr) = ( 6.94 , 34.91) 371.36 Normal (prop, spread) (a^ a2)=(6.8Q, 10.11) 536.33 (aj, a2) = (8.11, 8.20) 553.73 Lognormal (Pj, ol = (4.00, 0.41) 440.33 (jUj, ct) = <4.17, 0.41) 440.33 Gamma (a, k) = ( 6.39, 1.17) 469.68 (a, k ) = ( 7 . 3 9 , 1.17) 469.68 was not necessary to test a separate method to estimate a gamma selectivity curve. Residual plots from all selectivity models showed some degree of bias for the smaller (50-70 cm FL) size classes in the 11. 4-, 12. 7-, and 14.0-cm mesh sizes. This finding indicated that all models underestimated the numbers of small blacktip sharks caught in these mesh sizes, and these underestimates could be an artifact of the sampling design of the GULFSPAN juvenile shark survey (Carlson and Brusher, 1999). In this survey gillnet panels were arranged in increasing order by mesh size, and the order of panels was not randomized. Randomization of gillnet panels is common in selectiv- ity experiments because it is thought to reduce the potential preference of fish for any one area of the net. However, because fixed stations were not used, and the nets were fished at a variety of depths, habitats, and seasons, sampling design was probably not a factor in the model’s lack of fit to the data. The overdispersion of the data could be a result of the pooling of the data into 5-cm bins, or could indicate schooling behavior by some size classes of blacktip sharks. Shark species are known to segregate by size and sex; therefore the cap- ture of a cluster of similar-size blacktip sharks is likely. Overdispersion does not necessarily affect parameter estimation (Millar and Fryer, 1999), although an initial model assumption may have been violated. Although the assumption of equal catches may have been violated, the second assumption of equal fishing effort among gillnet panels was most likely met. The shallow bays and estuaries sampled, along with the length of the net (-600 m), decreased the probability of different panels fishing in different habitats and depth zones. Commercial gear can be several kilometers in length, and sagging can cause the middle part of the gear to fish in different depth strata than those at the ends. Blacktip sharks were therefore equally likely to encounter each panel of the GULFSPAN survey gillnet. On occasion, adult blacktip sharks (>130 cm FL) have been captured in the survey areas on longlines (Bethea and Carlson3). However, larger sharks are less likely to be caught in gillnets with mesh sizes smaller than 20 cm, and those few large sharks captured in the smaller mesh sizes were generally entangled by roll- ing in the gear — a phenomenon that was also noted for finetooth sharks (C. isodon) (Carlson and Cortes, 2003). All gillnet panels, except the 20.3 cm panel, were monofilament, and large sharks were able to break the monofilament and escape the gear. Such cases where larger sharks were entangled in smaller mesh sizes or where they broke free of the net could also have affected the lack of fit because the assumption of geometric similarity would not stand. The occurrence of larger sharks in small mesh sizes may have been reflected by the high model deviances for the models (normal proportional spread, lognormal, and gamma) where geometric similarity of the data was assumed. However, other than the lack of fit to the smallest size classes, the models described the data very well, with residu- als showing mostly equal error distribution and little systematic bias. Because of the change in the gear from 2005 through 2006, several attempts were made to account for a year effect within the SELECT method. Because of low sample sizes within years, especially for the 7.6- and 20.3-cm panels, it was not possible to incorporate year as a factor. For instance, a total of 92 and 115 blacktip sharks were captured by the 7.6- and 20.3-cm panels, respectively. Although these sample sizes were adequate for the overall model, when broken down by year the sample sizes were in the single digits for most size classes. The data were also separated into two time periods (1994-2005 and 2006-10), and the SE- LECT method was used to estimate selectivity models for each time period. The first time period produced reasonable results; however, no realistic solution was found for the second time period. This could also be due to sample sizes in the second time period. Although 3Dana M. Bethea and John. K. Carlson. 2011. Unpubl. data. Panama City Laboratory, Southeast Fisheries Sci- ence Center, National Marine Fisheries Service, National Oceanic and Atmospheric Administration, 3500 Delwood Beach Rd., Panama City, Florida 32408. 236 Fishery Bulletin 1 10(2) Normal (fixed spread) retention curve Deviance residuals Normal (proportional spread) retention curve Log-normal retention curve Deviance residuals Deviance residuals Figure 3 Gillnet selectivity curves and residuals estimated for blacktip sharks (Carcharhinus limbatus ) in the Gulf of Mexico Shark Pupping and Nursery (GULFSPAN) survey in northwest Florida calculated from the normal (fixed spread), normal (proportional spread), lognormal, and gamma distributions. The plots on the left are the estimated gillnet selectivity curves with relative retention probability on they axis. Increasing height of the curves indicates increas- ing mesh sizes. The plots on the right show the residuals of the models and mesh size on the y axis increases from bottom to top. Filled circles represent positive residuals and open circles represent negative residuals. The area of the circle is proportional to the square of the residual. there may have been a year effect that we were unable to account for, this is unlikely because of the nature of the survey and the species studied. Generally, year- to-year variability in recruitment is lower in sharks than in teleosts because of the production of large, well-developed young and low natural mortality (Smith et al., 1998; Walker, 1998). The GULFSPAN survey primarily targets juvenile sharks in nursery areas, and the majority of the blacktip sharks captured were juveniles. Therefore it is probable that interannual size variability was low in the survey area for black- tip sharks. Although this is an important factor that could be applied to other selectivity studies with more robust sample sizes, current stock assessment models Baremore et al Gillnet selectivity for juvenile Carcharhmus hmbatus 237 FL (cm) FL (cm) Figure 4 Selectivity curves for the normal proportional spread model where fishing intensity is assumed to be proportional to mesh size. Curves were plotted with observed data on frequency of fork lengths (FL) for blacktip sharks ( Carcharhmus limbatus) caught in the Gulf of Mexico Shark Pupping and Nursery (GULFSPAN) survey. Selectivity curves and length frequencies are plotted separately by mesh size. Relative retention is the probability that a fish of a given length class that comes into contact with that mesh size is captured. 238 Fishery Bulletin 1 10(2) for sharks do not include year-specific selectivity func- tions (SEDAR, 2006). Gillnet selectivity is more highly influenced by mor- phological features such as girth and the presence or absence of hard structures than it is by the length of a fish (Reis and Pawson, 1999; Carol and Garcia Berthou, 2007). Nevertheless, straight-line measured length can often be used as a proxy for girth in selectivity studies because of the close direct relationship between the two parameters (Reis and Pawson, 1999), with exceptions for cases of unusual morphological features (e.g., in hammerheads [Thorpe and Frierson, 2009]) or behav- ioral response to entanglement (e.g., finetooth sharks [Carlson and Cortes, 2003]). Many sharks in the family Carcharhinidae share similar body shape and struc- ture (Compagno and Niem, 1998), with girth and the rigidity of fins acting as a limiting factor for capture by gillnets (Carlson and Cortes, 2003; McAuley et ah, 2007). Girth-to-length relationships have been found to be similar among related species of sharks (McLoughlin and Stevens, 1994). It is therefore possible that selectiv- ity curves could be family-specific rather than species- specific for sharks. The most recent data have shown that the normal selectivity curves may provide the best fit for sharks in the family Carcharhinidae (McAuley et ah, 2007; Thorpe and Frierson, 2009), indicating that the results for blacktip sharks could be useful for other carcharhinids of similar size. Selectivity param- eters estimated for the blacktip shark could be used as a proxy for other species in the same family when species-specific selectivity estimates are unavailable. This theory could be tested by applying this method to other similar-size shark species for which a gillnet selectivity curve has been estimated, and should be pursued further as more data become available. Thorpe and Frierson (2009) found length modes of 97 and 88 cm FL for blacktip sharks caught in mesh sizes 7.6 and 10.2 cm, respectively, whereas we estimated modes of 46 and 62 cm FL for the same mesh sizes (Table 6). However, Thorpe and Frierson (2009) failed to fit a selectivity curve to the individual mesh sizes because of the wide spread of the sparse length data. Their study was based on a small number of samples (n = 76) and the modes for only two mesh sizes were estimated. The low sample size reported by Thorpe and Frierson (2009) was likely due to the relatively short duration of sampling, which was conducted over a period of eight months. Additionally, Thorpe and Frierson (2009) conducted their survey more than 1 km from shore, where the likelihood of small juveniles coming in contact with the gear was low. Total effort was not reported; however, catch rates were low in all gillnet panels (<0.15 blacktip sharks caught per hour of fishing). It is also possible that the size classes sampled in both studies were not reflective of the true size structure of the population because localized con- centrations of sharks in each area that were available to the gear probably differed. The true availability of blacktip sharks to gillnets in different regions cannot be known; therefore applying selectivity functions should Baremore et al.: Gillnet selectivity for |uvenile Carcharhmus limbatus 239 be done with a proper context and with supporting length-frequency data when possible. Based on data from fisheries observers, the average mesh size used from 2005 through 2010 in the commer- cial anchored gillnet fishery in the U.S. Atlantic Ocean was 11.1 cm, with a range of 8.5-16.0 cm (Passerotti4). The modal length calculated by using the equation in Table 3 for normal models indicates that blacktip sharks approximately 77 cm FL should be most vulner- able to the average mesh size in the commercial gillnet fishery. When calculated by using the full range of mesh sizes, the predicted modes range from 59 to 111 cm FL. Average lengths of blacktip sharks measured by ob- servers captured by commercial anchored gillnets from 2005-2010 ranged from 79 to 107 cm FL (Baremore et al., 2007; Passerotti and Carlson, 2009, 2010; Passerotti et al., 2010, 2011). The observed lengths are consis- tent with the selectivity model estimated for blacktip sharks. Blacktip sharks are born at approximately 40 cm FL and mature between 120 and 130 cm FL (5-7 yr) in the U.S. Atlantic Ocean (Carlson et al., 2006), suggesting that the blacktip sharks most vulnerable to commercial gear are juveniles. Juvenile blacktip sharks use inshore nursery areas during spring and summer months, but migrate into deeper waters in the fall and winter (Castro, 1993; Heupel et al., 2007). Commercial gillnet fishermen operating in states with gillnet bans are required to fish at least 4.8 km from shore (federal waters) in the U.S. Atlantic Ocean; therefore the small- est juvenile blacktip sharks may not be as vulnerable to bycatch in these areas, especially during summer months. However, in states such as North Carolina, which allow commercial gillnet fishing in state waters, the potential for gear interaction with juvenile blacktip sharks year-round is higher. Observer data show that blacktip sharks <120 cm FL are captured in commercial gillnet fisheries and therefore juvenile blacktip sharks are likely affected by both offshore and inshore gillnet fisheries. Thorpe and Frierson (2009) reported a mortality rate of 90.5% for blacktip sharks captured in experimental gillnets. Although soak time was not reported, the gill- nets and sampling protocol in their study were designed to mimic those commonly used by commercial gillnet fishermen in North Carolina; therefore it is probable that juvenile blacktip sharks interacting with commer- cial gillnets may also experience high bycatch mortal- ity. Demographic evidence suggests that population growth rates are more sensitive to survival of juvenile life stages of sharks than adults (Cortes, 2002). There- fore, modeling of the gear selectivity of gillnet fisheries, and particularly modeling bycatch from fisheries that have the potential to impact juveniles, is especially important. 4 Passerotti, M. 2011. Personal commun. Panama City Laboratory, Southeast Fisheries Science Center, National Marine Fisheries Service, National Oceanic and Atmos- pheric Administration, 3500 Delwood Beach Rd., Panama City, Florida 32408. Blacktip sharks are a commercially exploited species in U.S. waters, and the stock status in the Atlantic Ocean and Gulf of Mexico is assessed by the National Marine Fisheries Service on a regular basis (NMFS, 2002; SEDAR, 2006). Bycatch estimates for blacktip sharks are available from observer data (Passerotti et al., 2010), and fishing intensity of the Spanish and king mackerel gillnet fisheries has been previously es- timated (SEDAR 2008; 2009). These fishery-dependent data, along with selectivity curves provided by this study, can be used by assessment scientists to estimate the selectivity of blacktip sharks caught as bycatch by commercial gillnet fisheries in the U.S. Atlantic Ocean. Bycatch data are equally as important as primary catch data for stock assessment models (NMFS5; SEDAR, 2006), though often more difficult to attain because bycatch is generally discarded at sea. This study pro- vides valuable information for assessment scientists and managers tasked with estimating the size structure of blacktip sharks caught by commercial gillnet fisheries. Conclusions Juvenile blacktip sharks are caught as bycatch in com- mercial gillnet fisheries in the U.S. Atlantic Ocean, although the impact on the population has not been assessed. The results from this study showed that gill- net selectivity for juvenile blacktip sharks caught in the fishery-independent survey was best described by a normal selectivity curve with fixed spread and with fish- ing intensity proportional to mesh size. Because many commercial gillnet fisheries use mesh sizes similar to those used to produce these results, it may be possible to estimate the length frequencies of juvenile blacktip sharks influenced by these coastal fisheries. Selectiv- ity estimates may also be applicable to other sharks of similar size for which species-specific information is unavailable. Future studies should focus on fishery- dependent gillnet selectivity estimates to determine if selectivity changes with gear, location, or target species. Acknowledgments We thank everyone at the NOAA Fisheries Panama City Laboratory who aided in field work and gear maintenance. We would especially like to thank the numerous volunteers and unpaid interns who gave countless hours collecting data over the years. J. Carl- son reviewed a version of the manuscript and gave useful comments. E. Cortes imparted expertise on the subject of gillnet selectivity and M. Passerotti provided 5 NMFS (National Marine Fisheries Service). 2002. Stock assessment of large coastal sharks in the U.S. Atlantic Ocean and Gulf of Mexico: final meeting report of the 2002 shark evaluation workshop. Contribution report 02-03-177, 64 p. Sustainable Fisheries Div., National Marine Fisheries Service, NOAA, Silver Spring, MD. 240 Fishery Bulletin 1 10(2) information on commercial gillnet fisheries. The com- ments of three anonymous reviewers also added to the overall quality of the article. Sharks were collected under Florida Fish & Wildlife Conservation Commis- sion special activity licenses 02R-075, 03SR-075A, 04SR-075 and 08SR-075. Literature cited Baremore, I. E., J. K. Carlson, L. D. Hollensead, and D. M. Bethea. 2007. Catch and bycatch in the US southeast gillnet fisheries, 2007. NOAA Tech. Memo. NMFS-SEFSC-565, 23 p. Carlson, J. K., and D. M. Bethea. 2007. Catch and bycatch in the shark gillnet fishery 2005-2006. NOAA Tech. Memo. NMFS-SEFSC-552, 29 p. Carlson, J. K., and J. H. Brusher. 1999. An index of abundance for coastal species of juve- nile sharks from the northeast Gulf of Mexico. Mar. Fish. Rev. 61:37-45. Carlson, J. K., and E. Cortes. 2003. Gillnet selectivity of small coastal sharks off the southeastern United States. Fish. Res. 60:405-414. Carlson, J. K., J. A. Sulikowski, and I. E. Baremore. 2006. Do differences in life history exist for blacktip sharks, Carcharhmus limbatus, from the United States South Atlantic Bight and eastern Gulf of Mexico? Envi- ron. Biol. Fish. 77:279-292. Carol, J., and E. Garcia-Berthou. 2007. Gillnet selectivity and its relationship with body shape for eight freshwater fish species. J. Appl. Ich- thyol. 23:654-660. Castro, J. I. 1993. The shark nursery of Bulls Bay, South Carolina, with a review of the shark nurseries of the southeast- ern coast of the United States. Environ. Biol. Fish. 38:37-48. Compagno, L., and V. Niem. 1998. Carcharhinidae. Requiem sharks, FAO identifica- tion guide for fishery purposes. FAO, Rome. Cortes, E. 2002. Incorporating uncertainty into demographic mod- eling: application to shark populations and their con- servation. Conserv. Biol. 16:1048-1062. Heupel, M. R., J. K. Carlson, and C. A. Simpfendorfer. 2007. Shark nursery areas: concepts, definition, char- acterization and assumptions. Mar. Ecol. Prog. Ser. 337:287-297. Heupel, M. R., and R. E. Hueter. 2002. Importance of prey density in relation to the move- ment patterns of juvenile blacktip sharks ( Carcharhmus limbatus) within a coastal nursery area. Mar. Freshw. Res. 53:543-550. Kirkwood, G. P., and T. I. Walker. 1986. Gillnet mesh selectivities for gummy shark, Mus- telus antarcticus Gunther, taken in south-eastern Aus- tralian waters. Aust. J. Mar. Freshw. Res. 37:689-697. Maunder, M. N. 2002. The relationship between fishing methods, fisheries management and the estimation of maximum sustain- able yield. Fish Fish. 3:251-260. McAuley, R. B., C. A. Simpfendorfer., and I. W. Wright. 2007. Gillnet mesh selectivity of the sandbar shark ( Car - charhinus plumbeus ): implications for fisheries manage- ment. ICES J. Mar. Sci. 64:1702-1709. McEachran, J. D., and J. D. Fechhelm. 1998. Fishes of the Gulf of Mexico, 1112 p. Univ. Texas Press, Austin, TX. McLoughlin, K., and J. Stevens. 1994. Gill-net mesh selectivities for two species of com- mercial carcharhinid shark taken in northern Austra- lia. Mar. Freshw. Res. 45:521-534. Millar, R. B. 1992. Estimating the size-selectivity of fishing gear by conditioning on the total catch. J. Am. Stat. Assoc. 87:962-968. 2003. R CODE for fitting SELECT models to gillnet data, [http://www.stat.auckland.ac.nz/-millar/select ware/R/gillnets/. accessed February 2011.] 2010. Next generation R functions for trawl and net (or hook) selectivity. I http://www.stat.auckland.ac.nz/ -millar/selectware/RNext/1 Millar, R. B., and R. J. Fryer. 1999. Estimating the size-selection curves of towed gears, traps, nets and hooks. Rev. Fish Biol. Fish. 9:89-116. Millar, R. B., and R. Holst. 1997. Estimation of gillnet and hook selectivity using log-linear models. ICES J. Mar. Sci. 54:471-477. NMFS (National Marine Fisheries Service) 2008. Final amendment 2 to the Consolidated Atlantic Highly Migratory Species Fishery Management Plan, 80 p. NOAA Office of Sustainable Fisheries, Highly Migratory Species Management Division. Public Docu- ment, Silver Spring, MD. Passerotti, M. S., and J. K. Carlson. 2009. Catch and bycatch in the U.S. southeast gillnet fisher- ies, 2008. NOAA Tech. Memo. NMFS-SEFSC-583, 22 p. 2010. Errata for: Catch and bycatch in the shark gill- net fishery: 2005-2006. NOAA Tech. Memo. NMFS- SEFSC-552, 26 p. Passerotti, M. S., J. K. Carlson, and S. J. B. Gulak. 2010. Catch and bycatch in U.S. southeast gillnet fisheries, 2009. NOAA Tech. Memo. NMFS-SEFSC-600, 24 p. 2011. Catch and bycatch in the U.S. southeast gillnet fisher- ies, 2010. NOAA Tech. Memo. NMFS-SEFSC-612, 19 p. R Development Core Team. 2009. R: A language and environment for statistical computing, R Foundation for Statistical Computing, Vienna, Austria. [ISBN 3-900051-07-0; http://www.R- proieet.orgl Reis, E., and M. Pawson. 1999. Fish morphology and estimating selectivity by gillnets. Fish. Res. 39:263-273. SEDAR (Southeast Data, Assessment, and Review). 2006. SEDAR11, Stock assessment report: Large coastal shark complex, blacktip and sandbar shark, 31 p. NOAA Office of Sustainable Fisheries, Highly Migratory Spe- cies Management Division, Silver Spring, MD. [Avail- able from: http://www.sefsc.noaa.gov/sedar/. accessed December 2011.] 2008. SEDAR17, Complete stock assessment report 1: South Atlantic Spanish mackerel, 95 p. South Atlantic Fishery Management Council, North Charleston, SC. 2009. SEDAR16, Stock assessment report: South Atlantic and Gulf of Mexico king mackerel, 484 p. South Atlantic Baremore et at Gillnet selectivity for |uvemle Carcharhinus limbatus 241 Fishery Management Council, North Charleston, SC. [Available http://www.sefsc.noaa.gov/sedar/. accessed December 2011.1 Simpfendorfer, C. A., and P. Unsworth. 1998. Gill-net mesh selectivity of dusky sharks ( Car- charhinus obscurus ) and whiskery sharks ( Furgaleus macki) from south-western Australia. Mar. Freshw. Res. 49:713-718. Smith, S. E., D. W. Au, and C. Show. 1998. Intrinsic rebound potentials of 26 species of Pacific sharks. Mar. Freshw. Res. 49:663-678. Thorpe, T , and D. Frierson. 2009. Bycatch mitigation assessment for sharks caught in coastal anchored gillnets. Fish. Res. 98:102-112. Trent, L., D. E. Parshley, and J. K. Carlson. 1997. Catch and bycatch in the shark drift gillnet fishery off Georgia and east Florida. Mar. Fish. Rev. 59:19-28. Walker, T. I. 1998. Can shark resources be harvested sustainably? A question revisited with a review of shark fisheries. Mar. Freshw. Res. 49:553-572. Population structure of pink salmon (Oncorhynchus gorbuscha ) in British Columbia and Washington, determined with microsatellites E-mail address for contact author: Terry. Beacham@dfo-mpo.Qcc 1 Fisheries and Oceans Canada Pacific Biological Station 3190 Hammond Bay Road Nanaimo, B C., Canada V9T 6N7 2 Fisheries and Oceans Canada 417-2nd Avenue West Prince Rupert, B C Canada V8J 1G8 3 Pacific Salmon Commission 600-1155 Robson Street Vancouver, B C., Canada V6E 1B5 Abstract — Population structure of pink salmon ( Oncorhynchus gorbus- cha) from British Columbia and Wash- ington was examined with a survey of microsatellite variation to describe the distribution of genetic variation. Variation at 16 microsatellite loci was surveyed for approximately 46,500 pink salmon sampled from 146 loca- tions in the odd-year broodline and from 116 locations in the even-year broodline. An index of genetic differ- entiation, Fst, over all populations and loci in the odd-year broodline was 0.005, with individual locus values ranging from 0.002 to 0.025. Popula- tion differentiation was less in the even-year broodline, with a ^st value of 0.002 over all loci, and with individ- ual locus values ranging from 0.001 to 0.005. Greater genetic diversity was observed in the odd-year brood- line. Differentiation in pink salmon allele frequencies between broodlines was approximately 5.5 times greater than regional differentiation within broodlines. A regional structuring of populations was the general pat- tern observed, and a greater regional structure in the odd-year broodline than in the even-year broodline. The geographic distribution of microsatel- lite variation in populations of pink salmon likely reflects a distribution of broodlines from separate refuges after the last glaciation period. Manuscript submitted 3 October 2011. Manuscript accepted 5 January 2012. Fish. Bull. 110:242-256(2012) The views and opinions expressed or implied in this article are those of the author (or authors) and do not necessarily reflect the position of the National Marine Fisheries Service, NOAA. Terry D. Beacham (contact author)1 Brenda McIntosh' Cathy MacConnachie1 Brian Spilsted2 Bruce A. White3 Pink salmon ( Oncorhynchus gor- buscha) spawn in more than 700 rivers in British Columbia (Aro and Shepard, 1967), and the distribution of spawning populations varies in odd- numbered years and even-numbered years. Spawning occurs primarily only in odd years in the Fraser River drainage, mainly in even years on the Queen Charlotte Islands, and in both years in central and northern coastal areas of British Columbia (Neave, 1952). Because virtually all pink salmon mature at two years of age (Bilton and Ricker, 1965), two dis- tinct broodlines in the species have developed (odd-year and even-year), with virtually no gene flow between the broodlines. This reproductive iso- lation of the broodlines has resulted in reported differences in body size (Godfrey, 1959), morphological char- acteristics (Ricker, 1972; Beacham, 1985), timing of spawning (Aro and Shepard, 1967; Dyagilev and Markev- ich, 1979), and genetic differentiation (Aspinwall, 1974; Beacham et al., 1988). Evaluation of the genetic pop- ulation structure both between and within broodlines in British Columbia and Washington form the basis of the current study. Estimation of genetic population structure has been a key area of research in salmon assessment and management. Identification of a ge- netically distinct group of populations in the distribution of a species is a key step in conserving and maintain- ing genetic diversity. Genetically dis- tinct populations or regional groups of populations (stocks) were determined through surveys of genetic variation to evaluate the population structure of a species (Shaklee and Bentzen, 1998). Identification of genetically distinct groups in the distribution of pink salmon in British Columbia and Puget Sound may lead to the conser- vation of genetic diversity by fisheries and resource management. An evalu- ation of genetic variation is effective in describing the population struc- ture of salmonids, is a key part in the elucidation of management units or conservation units for a species and can be applied to manage fisheries exploiting specific stocks of salmon. Determination of genetic population structure is an important part in de- veloping a genetically based method for estimation of stock composition. Allozymes were initially the key set of genetic markers used in evaluat- Beacham et al : Population structure of Oncorhynchus gorbuscha in British Columbia and Washington, determined with microsatellites 243 ing population structure in pink salmon in British Co- lumbia (Beacham et ah, 1988; Shakiee et ah, 1991) and elsewhere in the Pacific Rim distribution of the species (Aspinwall, 1974; Varnavskaya and Beacham, 1992; Shakiee and Varnavskaya, 1994; Noil et ah, 2001; Hawkins et ah, 2002). Analysis of allozyme variation provided the general pattern of marked differentiation between the broodlines, but identification of fine-scale regional structure was limited, although estimation of stock composition of samples from mixed-stock fisheries was conducted (Beacham et ah, 1985; Shakiee et ah, 1991). Development of DNA markers has led to new avenues of research for using genetic variation in de- fining population structure. Mitochondrial DNA varia- tion again showed strong differentiation between the broodlines, but differentiation among regional groups of pink salmon was limited (Brykov et ah, 1996, 1999; Churikov and Gharrett, 2002). However, Golovanov et ah (2009) noted that differentiation among even-year populations was higher than among odd-year popula- tions in the northern Sea of Okhotsk region. Microsatellites are reported to be useful for evaluat- ing fine-scale population structure in salmonids (Banks et ah, 2000) and have been used to evaluate large- scale and regional variation in chum salmon (O. keta) (Beacham et ah, 2009). Initial applications of micro- satellite variation to evaluate individual identification and population structure were reported by Olsen et ah (1998, 2000a). However, surveys of population variation were quite limited in these studies, and no comprehen- sive evaluation of variation at microsatellites has been conducted for pink salmon. A survey of microsatellite variation over a broader geographic range of pink salm- on distribution would likely be valuable for evaluating population structure. In the current study, we outline the microsatellite- based population structure of pink salmon in British Columbia and Washington as an initial step in evalu- ating whether higher resolution in estimation of stock composition may be possible when compared with esti- mates previously derived with aliozymes. This objec- tive was accomplished by analyzing variation at 16 microsatellite loci to evaluate relationships in popula- tion structure of pink salmon, as well as by analyzing regional differences in allelic variation. The distribu- tion of genetic diversity among broodlines, regions, and populations was estimated in the study, as well as the stability of population structure. Materials and methods Collection of DNA samples and laboratory analysis Tissue samples were collected from mature pink salmon. Samples were preserved in 95% ethanol, and sent to the Molecular Genetics Laboratory at the Pacific Biologi- cal Station of Fisheries and Oceans Canada. DNA was extracted from the tissue samples by using a variety of methods, including a chelex resin protocol outlined Table 1 Microsatellite loci surveyed in pink salmon ( Oncorhyn- chus gorbuscha) and their associated annealing and extension temperatures and times (seconds), as well as the number of cycles used in polymerase chain reaction amplifications. Locus Annealing Extension Cycles OkilO 53°C/30s 70°C/30s 39 OkilOl 53°C/45s 68°C/30s 40 OnelOl 50°C/30s 70°C/30s 39 Onel02 50°C/30s 70°C/30s 39 Onel04 50°C/30s 70°C/30s 36 Onel09 55°C/30s 70°C/30s 34 Onelll 55°C/30s 70°C/30s 34 Onell4 50°C/30s 70°C/45s 38 Ots213 52°C/45s 72°C/60s 38 Ots7e 51°C/30s 72°C/30s 35 OtsG253b 60°C/45s 72°C/45s 35 OtsG311 50°C/45s 68°C/45s 34 OtsG68 50°C/30s 70°C/30s 36 Ssa407 60°C/30s 70°C/30s 39 Ssa408 60°C/45s 70°C/45s 40 Ssa419 50°C/30s 70°C/30s 40 by Small et al. (1998), a Qiagen 96-well DNeasy®1 pro- cedure (Qiagen, Mississauga, Ontario), or a Promega Wizard SV96 Genomic DNA Purification system (Pro- mega, Madison, WI). Once extracted DNA was avail- able, surveys of variation at 16 microsatellite loci were conducted: OkilO (Smith et a!., 1998), OkilOl (Beacham et ah, 2011), OtsG68, OtsG253b, OtsG311 (Williamson et ah, 2002), Ots213 (Greig et ah, 2003), Ots7e (Wright et ah, 2008), OnelOl, Onel02, Onel04, Onel09 , Onelll , Onell4 (Olsen et ah, 2000b), Ssa407 , Ssa408, Ssa419 (Cairney et ah, 2000) In general, polymerase chain reaction (PCR) DNA amplifications were conducted by using a DNA En- gine Cycler Tetrad2 (BioRad, Hercules, CA) in 6-pL volumes consisting of 0.15 units of Taq polymerase, 1-pL of extracted DNA, lxPCR buffer (Qiagen), 60 pM each nucleotide, 0.40 pM of each primer, and deionized water. The thermal cycling profile involved one cycle of Taq activation for 15 minutes at 95°C, followed by a denaturation cycle of 30 seconds at 94°C, with anneal- ing and extension conditions for each locus as outlined in Table 1. PCR fragments were initially size fraction- ated in denaturing polyacrylamide gels with an ABI 377 automated DNA sequencer, and genotypes were scored by Genotyper, vers. 2.5 software (Applied Biosystems, Foster City, CA) by using an internal lane sizing stan- dard. Later in the study, microsatellites were size frac- tionated in an ABI 3730 capillary DNA sequencer, and 1 Mention of trade names or commercial companies is for identification purposes only and does not imply endorsement by the National Marine Fisheries Service, NOAA. 244 Fishery Bulletin 1 10(2) Table 2 Pink salmon (Oncorhynchus gorbuscha) spawning regions, number of populations sampled, years when sampling occurred, aver- age number of fish sampled per population per year over all populations within a region, and average total number of fish sampled (AO per population within the region for 146 odd-year and 116 even-year populations in 15 geographic regions (Fig. 2). Listing of populations in each region as well as allele frequencies for all population samples surveyed in this study are available at the Molecular Genetics Laboratory website http://www.pac.dfo-mpo.gc.ca/science/facilities-installations/pbs-sbp/mgl-lgm/ data-donnees/index-eng.htm. Values shown are averages with the ranges of samples sizes in parentheses. Average annual Average population Region Populations Years population sample size size (A0 Washington (odd) 12 1995-2009 110 (44_498) 257 (98-755) Fraser River-upper (odd) 9 1987-2009 95 (69-102) 201 (100-463) Fraser River-lower (odd) 6 1987-2009 87 (50-100) 245 (98-463) East Coast Vancouver Island (odd) 10 1987-2009 108 (85-219) 173 (85-397) South Coast British Columbia (odd) 13 1987-2009 105 (25-200) 169 (38-390) Central Coast British Columbia (odd) 59 2003-2009 111 (2-227) 164 (32-394) Skeena River (odd) 10 2003-2007 163 (67-229) 228 (123-393) North Coast British Columbia (odd) 25 2003-2009 151 (9-233) 211 (87-381) Queen Charlotte Islands (odd) 2 2005 210 (200-219) 210 (200-219) East Coast Vancouver Island (even) 2 2006-2008 83 (50-113) 124 (85-163) South Coast British Columbia (even) 11 2002-2010 119 (27-237) 162 (47-452) Central Coast British Columbia (even) 50 2002-2010 96 (5-202) 133 (18-312) Skeena River (even) 7 2002-2006 162 (63-228) 208 (100-381) North Coast British Columbia (even) 25 2002-2010 113 (3-215) 190 (24-425) Queen Charlotte Islands (even) 21 2002-2006 135 (32-200) 176 (32-364) genotypes were scored by GeneMapper software, vers. 3.0 (Applied Biosystems) by using an internal lane siz- ing standard. Allele identification between the two se- quencers was standardized by analyzing approximately 600 individuals on both platforms and converting the sizing in the gel-based data set to match that obtained from the capillary-based set. Data analysis All annual samples available for a location were com- bined to estimate population allele frequencies, as is recommended by Waples (1990). Each population in each broodline at each locus was tested for departure from Hardy-Weinberg equilibrium of genotypic frequencies by using the software Genetic Data Analysis (GDA; Univ. of Connecticut, Storrs, CT). Critical significance levels for simultaneous tests were evaluated by using Bon- ferroni adjustment for each broodline separately (odd- year broodline: 0.05/146 = 0.00034, even-year broodline 0.05/116 = 0.00043; Rice, 1989). Weir and Cockerham’s (1984) Fst estimates for each locus over all populations were calculated with FSTAT, vers. 2. 9. 3. 2 (Goudet, 1995). The significance (P< 0.05) of the multilocus FST value over all samples was determined by jackknifing over loci. Populations were combined into 14 regional groups in order to develop a practical method to display mean pairwise FST values between regions, as well as the mean number of alleles observed per locus in each region. These 14 regional groups were constructed from the 15 regional groups outlined in Table 2 by combin- ing the upper and lower Fraser River regions into a single region, with other regions remaining as outlined in Table 2. Broodlines were separated in odd-year and even-year spawning lines. Geographic areas were out- lined in Figure 1. The 14 regional groups correspond to the geographic areas and broodlines as follows: 1) Washington odd-year; 2) Fraser River (upper+lower) odd-year; 3) east coast Vancouver Island odd-year; 4) south coast British Columbia odd-year; 5) central coast British Columbia odd-year; 6) Skeena River odd-year; 7) north coast British Columbia odd-year; 8) Queen Charlotte Islands odd-year; 9) east cCoast Vancouver Island even-year; 10) south coast British Columbia even- year; 11) central coast British Columbia even-year; 12) Skeena River even-year; 13) north coast British Colum- bia even-year; 14) Queen Charlotte Islands even-year. Individual populations remained discrete within these larger regional groups for determination of pairwise Fst values. Genotypic disequilibrium and potential genetic link- age among loci were tested with GDA with 1500 itera- tions per test. The number of pairs of loci exhibiting potential linkage was summed for the 146 populations sampled in the odd-year broodline and 116 popula- tions sampled in the even-year broodline (Table 2). Statistical significance was evaluated by using a Bon- ferroni adjustment as outlined previously.FSTAT was used to measure the “allelic richness” (allelic diversity standardized to a sample size of 240 fish per region) for the 14 regional groups of populations. Computa- tion of the number of alleles observed per locus was carried out with GDA. Cavalli-Sforza and Edwards (CSE) chord distance (1967) was used to estimate Beacham et al. : Population structure of Oncorhynchus gorbuscha in British Columbia and Washington, determined with microsatellites 245 Figure 1 Map of British Columbia and northern Washington coasts indicating the general geographic regions where pink salmon (Oncorhynchus gorbuscha) from 146 odd-year and 116 even-year populations were surveyed, with the regions listed in Table 2. The regions depicted in the figure are the fol- lowing: 1 Washington; 2 lower Fraser River; 3 upper Fraser River; 4 east coast Vancouver Island (ECVI); 5 southern British Columbia; 6 central coast; 7 Skeena River; 8 north coast; and 9 Queen Charlotte Islands. In 6 of these regions (ECVI, southern British Columbia mainland, central coast, Skeena River, north coast, Queen Charlotte Islands), pink salmon spawn in both even and odd years. Because 6 regions have both broodlines present, the map encompasses 12 regional groups of populations when even and odd populations are separated. The other 3 regional groups (upper and lower Fraser River, and Washington) have pink salmon only in odd years. This summary accounts for the 15 regional groups outlined in Table 2. genetic distances among all populations. An unrooted neighbor-joining tree based upon CSE was generated with NJPLOT (Perriere and Gouy, 1996). Bootstrap support for the major nodes in the tree was evalu- ated with the CONSENSE program from PHYLIP based upon 500 replicate trees (Felsenstein, 1993). The distribution of genetic variation in pink salmon was evaluated with a gene diversity analysis with the analysis structured between broodlines, among regions within broodlines, and among populations within regions. The analysis was conducted with GDA, which will support a maximum of three nested levels of variation in addition to the error mean square. All populations outlined in Table 2 were included in the analysis. Allele frequencies for all location samples surveyed in this study are available at the Molecular Genetics Laboratory website http://www.pac.dfo-mpo. gc.ca/science/facilities-installations/pbs-sbp/mgl-lgm/ data-donnees/index-eng.htm. Results Variation within populations Variation was displayed in the number of observed alleles at the 16 microsatellite loci surveyed in the study. The fewest number of alleles was observed at Ots7e (12 alleles odd-year broodline, 13 alleles even- year broodline), and the greatest number of alleles was observed at OkilO (85 alleles odd-year broodline, 83 alleles even-year broodline) (Table 3). Heterozy- gosities were generally above 90%, with notable excep- tions observed at Ots7e (both broodlines) and Onelll (odd-year only). Genotypic frequencies at all 16 loci surveyed typically conformed to those expected under Hardy-Weinberg equilibrium (HWE) for populations in both broodlines. Greater overall population differen- tiation was observed in the odd-year broodline ( Fsr = 0.005) than in the even-year broodline ( FST =0.002). 246 Fishery Bulletin 1 10(2) Table 3 Number of alleles per locus, an index of genetic differentiation FST (standard deviation in parentheses), expected heterozygosity (He), observed heterozygosity (H0), and percent significant Hardy-Weinberg equilibrium (HWE) tests for 16 microsatellite loci among 146 odd-year and 116 even-year pink salmon (Oncorhynchus gorbuscha) populations. Locus Number of alleles 1 ST He H0 HWE Odd-year broodline OkilO 85 0.004 (0.000) 0.94 0.94 0.0 OkilOl 83 0.002 (0.000) 0.97 0.96 0.7 OnelOl 52 0.003 (0.000) 0.96 0.95 0.7 One 102 32 0.006 (0.001) 0.92 0.91 0.0 One 104 40 0.007 (0.001) 0.95 0.94 0.0 Onel09 32 0.005 (0.001) 0.90 0.90 0.0 Onelll 36 0.012 (0.001) 0.72 0.71 0.0 Onell4 49 0.003 (0.000) 0.96 0.95 2.1 Ots213 69 0.003 (0.001) 0.97 0.95 2.8 Ots7e 12 0.025 (0.003) 0.45 0.45 0.0 OtsG253b 46 0.005 (0.001) 0.95 0.94 1.4 OtsGSll 34 0.009 (0.001) 0.89 0.88 1.4 OtsG68 34 0.004 (0.001) 0.93 0.93 0.0 Ssa407 73 0.006 (0.001) 0.94 0.94 0.0 Ssa408 69 0.002 (0.000) 0.97 0.95 6.2 Ssa419 66 0.003 (0.000) 0.96 0.93 7.5 Total 0.005 (0.001) Even-year broodline OkilO 83 0.001 (0.000) 0.94 0.93 1.8 OkilOl 73 0.002 (0.000) 0.97 0.96 0.9 OnelOl 65 0.001 (0.000) 0.96 0.95 1.8 One 102 35 0.002 (0.000) 0.93 0.93 0.0 Onel04 36 0.004 (0.000) 0.92 0.92 0.0 Onel09 29 0.003 (0.000) 0.91 0.90 0.0 Onelll 38 0.004 (0.001) 0.92 0.91 2.6 Onell4 42 0.002 (0.000) 0.96 0.95 0.0 Ots213 71 0.002 (0.000) 0.97 0.97 5.1 Ots7e 13 0.005 (0.001) 0.58 0.58 1.8 OtsG253b 48 0.002 (0.000) 0.95 0.94 3.5 OtsG311 34 0.003 (0.000) 0.93 0.91 0.0 OtsG68 32 0.003 (0.000) 0.93 0.93 0.0 Ssa407 77 0.002 (0.000) 0.96 0.93 2.6 Ssa408 69 0.002 (0.000) 0.97 0.95 1.8 Ssa419 66 0.002 (0.000) 0.97 0.95 2.6 Total 0.002 (0.000) Individual locus FST values ranged from 0.002 to 0.012 for the odd-year broodiine, and between 0.001 and 0.005 for the even-year broodline. For the odd-year broodline, individual pairs of loci displayed poten- tial linkage of between 0% and 28% ( Ssa408 and Ssa419) of the populations surveyed. In the even-year broodline, individual pairs of loci displayed potential linkage of between 0% and 13% ( Ssa407 and Ssa419 ) of the populations surveyed. Potential linkage among loci was higher in the Ssa series of loci than that observed in the other microsatellites, but different patterns were observed within the broodlines. Poten- tial linkage among the Ssa loci was not judged to be at a level that required removal from subsequent analyses. The number of alleles observed displayed variation among the regional groups of pink salmon surveyed. With the number of alleles observed standardized to a sample size of 240 individuals per region for both broodlines, the odd-year broodline populations with the fewest number of observed alleles originated from Washington (486 alleles), whereas populations with the greatest number of alleles originated from the north- ern coastal region of British Columbia (551 alleles) (Table 4). In the even-year broodline, populations with the fewest numbers of alleles originated from the east Beacham et al Population structure of Oncorhynchus gorbuscho in British Columbia and Washington, determined with microsatellites 247 Table 4 Mean number of alleles observed per locus at 16 microsatellite loci for pink salmon (Oncorhynchus gorbuscha) from 14 geographic broodline areas standardized to a sample size of 240 fish per geographic area. Regional groups (see Table 2), are as follows: 1) Washington odd-year, 2) Fraser River (upper-i- lower) odd-year, 3) east coast Vancouver Island odd-year, 4) south coast British Columbia odd-year, 5) central coast British Columbia odd-year, 6) Skeena River odd-year, 7) north coast British Columbia odd- year, 8) Queen Charlotte Islands odd-year, 9) east coast Vancouver Island even-year, 10) south coast British Columbia even-year, 11) central coast British Columbia even-year, 12) Skeena River even-year, 13) north coast British Columbia even-year, 14) Queen Charlotte Islands even-year. The two Fraser River regions in Table 2 were combined for the analysis. Locus 1 2 3 4 5 6 7 8 9 10 11 12 13 14 Odd Even OkilO 30.77 33.14 41.59 40.84 47.21 40.53 48.42 44.81 26.93 31.14 38.21 36.03 43.31 35.67 40.91 35.21 OkilOl 50.95 53.09 54.73 55.30 55.83 51.38 54.65 55.79 49.88 48.04 51.90 52.11 55.65 48.74 53.96 51.05 OnelOl 35.30 37.33 35.93 37.70 36.39 36.18 36.36 35.78 35.78 36.66 39.15 36.55 39.41 39.42 36.37 37.83 Onel02 22.73 22.25 23.22 23.45 24.44 20.79 24.40 23.99 23.94 25.73 25.53 26.08 26.16 25.68 23.16 25.52 Onel04 30.29 31.81 30.85 31.78 30.64 28.23 31.10 30.49 29.89 27.95 30.04 24.63 30.40 27.40 30.65 28.39 One 109 17.31 18.00 17.51 19.13 19.62 20.18 20.23 18.90 16.00 18.00 18.60 18.54 19.33 16.98 18.86 17.91 Onelll 19.79 21.45 21.52 21.81 21.01 19.73 21.24 20.54 21.97 24.05 25.27 22.29 23.51 22.40 20.88 23.25 Onell4 37.24 39.89 38.61 38.84 37.30 34.99 36.19 35.26 30.00 32.23 34.74 35.46 35.78 34.58 37.29 33.80 Ots213 43.25 42.92 45.32 49.90 51.23 50.38 51.77 50.54 43.91 50.37 50.34 48.99 51.27 46.15 48.16 48.51 Ots7e 5.79 5.90 7.00 6.43 7.68 7.12 7.80 6.48 4.00 5.84 5.82 6.17 7.27 4.92 6.77 5.67 OtsG311 16.65 15.91 17.97 18.32 20.50 20.44 20.24 21.25 20.00 21.41 21.06 18.75 21.49 19.69 18.91 20.40 OtsG68 24.95 25.11 25.12 24.44 26.38 27.07 26.01 23.80 20.98 21.94 21.99 20.99 25.23 21.14 25.36 22.04 OtsG253b 33.96 33.88 34.08 35.64 33.92 31.77 34.03 32.57 32.83 34.17 35.74 34.44 36.26 33.12 33.73 34.43 Ssa407 27.87 27.35 31.41 31.74 33.37 31.83 32.28 25.33 47.90 48.86 52.42 51.36 51.01 49.25 30.15 50.13 Ssa408 52.54 53.04 54.75 51.62 52.89 49.94 54.09 52.31 49.90 48.73 54.88 52.09 55.73 51.32 52.65 52.11 Ssa419 36.33 40.19 49.00 46.13 53.07 47.58 52.36 50.93 48.00 46.94 47.97 44.46 50.95 44.42 46.95 47.12 Total 485.69 501.24 528.60 533.04 551.46 518.14 551.17 528.78 501.90 522.06 553.65 528.92 572.73 520.90 524.76 533.36 coast of Vancouver Island (502 alleles), and those popu- lations with the greatest number of alleles originated from the northern coastal region of British Columbia (573 alleles). The greatest difference in number of al- leles between the broodlines was observed at the locus Ssa407, with an average of 30 alleles observed at the locus in the odd-year populations, and 50 alleles ob- served in the even-year populations (P<0.01). Within the odd-year broodline, lower numbers of observed al- leles of Washington populations, and to some extent Fraser River populations, compared with other regional groups of populations, were concentrated in the OkilO and Ssa419 loci (Table 4). Distribution of genetic variance Gene diversity analysis of the 16 microsatellites sur- veyed was used to evaluate the distribution of genetic variation between two broodlines, among regions (15 regions, Table 2) within broodlines, and among popula- tions within regions (146 odd-year, 116 even-year popu- lations). The amount of variation within populations ranged from 85.1% ( Onelll ) to 99.8% ( OkilOl , OnelOl ), and averaged 98.1% across all loci (Table 5). Variation between the two broodlines accounted for 1.5% of total observed variation and was the largest source of varia- tion after within-population variation. Variation among regions within broodlines was the next largest source of variation and accounted for 0.3% of total observed variation. Variation among populations within regions accounted for 0.2% of total observed variation. Differen- tiation between the broodlines was approximately three times greater than any combined regional or population source of variation. Significant broodline differentia- tion in allele frequencies was observed in 13 of the 16 loci surveyed, with the greatest difference observed at Onelll , which was nine times larger than that observed at any other locus (Table 5). For the geographic range of populations surveyed in the study, broodline differences contributed more to differentiation of allele frequencies than any regional or population source of differentiation. Population structure Regional genetic differentiation was observed among pink salmon populations sampled in the different geo- graphic regions. As expected, the largest differences in genetic differentiation were observed between regional groups of populations when compared with populations in the alternate broodline (regional FST values ranging from 0.014 to 0.036) (Table 6). In the odd-year brood- line, the largest average population differentiation was observed in comparisons of populations originat- ing from Washington compared with populations from British Columbia (regional FST values ranging from 0.011 to 0.016) (Table 6). Within British Columbia, pink salmon in the odd-year broodline originating from the Fraser River were the most genetically distinct 248 Fishery Bulletin 1 10(2) Table 5 Hierarchical gene-diversity analysis of regional FST values for 262 populations of pink salmon (Oncorhynchus gorbuscha) within 15 regions (9 regions in odd-year broodline, 6 regions in even-year broodline) for 16 microsatellite loci, with the regions outlined in Table 2. **P<0.01 *P<0.05 Locus Within populations Among populations within regions Among regions within broodlines Between brood lines OkilO 0.9967 0.0014" 0.0017” 0.0002 OkilOl 0.9976 0.0013" 0.0009" 0.0002 OnelOl 0.9976 0.0012" 0.0009" 0.0003 One 102 0.9942 0.0021" 0.0023" 0.0014* Onel04 0.9782 0.0021" 0.0045" 0.0152” Onel09 0.9902 0.0021" 0.0025" 0.0052” Onelll 0.8507 0.0022" 0.0057" 0.1414“ Onell4 0.9944 0.0013" 0.0013" 0.0030” Ots213 0.9952 0.0019" 0.0012" 0.0017" Ots7e 0.9207 0.0056" 0.0111" 0.0626” OtsG253b 0.9955 0.0019” 0.0023" 0.0003 OtsG311 0.9772 0.0023" 0.0047" 0.0158" OtsG68 0.9896 0.0019” 0.0020" 0.0065** Ssa407 0.9930 0.0018" 0.0034” 0.0018* Ssa408 0.9942 0.0013" 0.0009" 0.0036’* Ssa419 0.9949 0.0016" 0.0012” 0.0023" Total 0.9806 0.0019" 0.0027" 0.0148" Table 6 Mean pairwise FST values averaged over 16 microsatellite loci from 14 regional groups of pink salmon (Oncorhynchus gorbuscha) outlined in Table 2 that were sampled at 262 locations (146 odd-year, 116 even-year) in British Columbia and Washington. Com- parisons were conducted between individual populations in each region. Values in bold on the diagonal represent comparisons among populations within each region. ^st values are listed above the diagonal, and standard deviations are shown below the diagonal. RC is region code, and codes are as follows: 1) Washington odd-year, 2) Fraser River odd-year, 3) east coast Vancouver Island odd-year, 4) southern coast British Columbia odd-year, 5) central coast British Columbia odd-year, 6) Skeena River odd- year, 7) north coast British Columbia odd-year, 8) Queen Charlotte Islands odd-year, 9) east coast Vancouver Island even-year, 10) southern Coast British Columbia even-year, 11) central coast British Columbia even-year, 12) Skeena River even-year, 13) north coast British Columbia even-year, 14) Queen Charlotte Islands even-year. RC 1 2 3 4 5 6 7 8 9 10 11 12 13 14 1 0.011 0.011 0.013 0.011 0.013 0.016 0.014 0.015 0.036 0.034 0.032 0.032 0.031 0.034 2 0.010 0.001 0.009 0.006 0.007 0.010 0.008 0.010 0.029 0.027 0.026 0.025 0.025 0.028 3 0.009 0.003 0.006 0.006 0.006 0.008 0.006 0.008 0.025 0.023 0.021 0.021 0.020 0.023 4 0.009 0.002 0.003 0.004 0.003 0.006 0.004 0.006 0.025 0.023 0.021 0.020 0.020 0.024 5 0.009 0.002 0.003 0.002 0.001 0.003 0.001 0.003 0.019 0.017 0.015 0.015 0.014 0.018 6 0.009 0.003 0.003 0.003 0.002 0.001 0.004 0.007 0.019 0.017 0.015 0.015 0.014 0.018 7 0.008 0.003 0.003 0.003 0.002 0.002 0.001 0.003 0.019 0.017 0.015 0.015 0.014 0.018 8 0.009 0.001 0.003 0.003 0.001 0.003 0.002 0.003 0.022 0.020 0.018 0.019 0.018 0.021 9 0.010 0.003 0.006 0.004 0.003 0.003 0.003 0.003 0.008 0.005 0.004 0.006 0.005 0.007 10 0.009 0.003 0.006 0.004 0.003 0.003 0.003 0.003 0.002 0.003 0.002 0.004 0.002 0.005 11 0.009 0.003 0.006 0.004 0.002 0.002 0.002 0.002 0.002 0.003 0.001 0.003 0.001 0.004 12 0.008 0.002 0.005 0.004 0.002 0.002 0.002 0.002 0.002 0.003 0.002 0.002 0.002 0.006 13 0.009 0.002 0.006 0.004 0.002 0.002 0.002 0.002 0.002 0.002 0.001 0.002 0.001 0.004 14 0.009 0.002 0.006 0.004 0.002 0.002 0.002 0.002 0.003 0.002 0.001 0.002 0.001 0.002 Beacham et al.: Population structure of Oncorhynchus gorbuscha in British Columbia and Washington, determined with microsatellites 249 group of populations (regional FST values ranging from 0.006 to 0.010). The least differentiation was observed between populations from the northern and central coastal regions of British Columbia (F ST= 0.001). In the even-year broodline, the largest regional genetic differentiation was observed between populations from the east coast of Vancouver Island and those from the Queen Charlotte Islands (FST= 0.007), whereas the least differentiation between regional groups of populations was observed between the northern and central coastal regions of British Columbia (Fsr=0.001) (Table 6). Two distinct lineages of pink salmon were observed in the cluster analysis, and they were clearly based on whether pink salmon spawned in odd-numbered or even-numbered years. All odd-year populations clus- tered together with 100% bootstrap support, as did all even-year populations (Fig. 2). Within the odd-year broodline, a Washington group of populations was well supported (Washington populations clustered together in 83% of dendrograms evaluated). Within Washing- ton, further geographic subdivision was observed, with populations from the Strait of Juan de Fuca (Gray Wolf River, Dungeness River) clustering with the Hood Canal hatchery population in 100% of dendrograms evaluated. The Hood Canal hatchery population was the most genetically distinct population included in the survey (Fig. 2). The three remaining populations from Hood Canal (Dosewallips River, Hamma Hamma River, and Duckabush River), in addition to the Gray Wolf River, Dungeness River, and Hood Canal hatchery populations, were well separated from other popula- tions in Washington, clustering together in 100% of dendrograms evaluated. Pink salmon populations from the Fraser River in southern British Columbia were a well-defined geographic cluster — all 15 populations clustered together in 98% of dendrograms evaluated. Furthermore, populations in the upper portion of the drainage were separated from those populations in the lower portion of the drainage, with upper populations clustering together in 98% of dendrograms evaluated. Populations from the central portion of the east coast of Vancouver Island (Quinsam River, Puntledge River, Oyster River, Big Qualicum River, and Nanaimo River) clustered together in 100% of dendrograms evaluated, as did 98% of populations from the northern portion of the east coast of Vancouver Island (Keogh River, Quatse River, Cluxewe River). Populations from the northern portion of the South Coast of British Colum- bia (Kakweiken River, Lull Creek, Ahta Creek, Heydon River, Glendale River) constituted a well-defined group (96% of dendrograms evaluated). Those in the southern portion of the south coast (Cheakamus River, Ashlu River, Mamquam River, Squamish River, Indian River) were not well supported, but displayed some affinity to each other in the cluster analysis (Table 6). All 10 odd-year populations sampled from the Skeena River formed a distinct regional group (50% bootstrap sup- port). Populations sampled from the central coast and north coast regions of British Columbia did not cluster into distinct geographic units. Although some sepa- ration was observed, genetic differentiation between populations in the two regions was limited, and this was reflected in the lack of consistency in population clustering in the dendrograms evaluated. Cluster analysis of the populations sampled in the even-year broodline revealed a general lack of consisten- cy in geographically based clustering of the populations. The only exception was observed for the Skeena River drainage, where all seven populations sampled clus- tered together in 97% of dendrograms evaluated. There was some evidence for a weak association for 19 of 21 populations from the Queen Charlotte Islands, but the cluster was not well supported (22% bootstrap support). As with the odd-year broodline, populations sampled from the central coast and north coast regions of Brit- ish Columbia did not cluster into distinct geographic units (Fig. 2), which again reflected the overall lack of genetic differentiation (Fsr=0.001) between populations in the two regions. Discussion In the current study of microsatellite variation in pink salmon, approximately 46,500 individuals were surveyed from 146 odd-year and 116 even-year populations, 16 microsatellites were analyzed encompassing 812 alleles, and 12-85 alleles were identified per locus. Sample size ranged from 18 to 755 individuals per population, with at least 100 individuals sampled in 127 of the 146 odd-year populations, and 92 of the 116 even-year populations. Only six odd-year and seven even-year populations had fewer than 40 individuals surveyed. With a range in the number of individuals sampled per population, sampling errors may have influenced the estimated allele frequen- cies within populations, particularly for populations with fewer than 40 individuals sampled. If sampling errors are large in estimation of allele frequencies, there is a potential for these errors to obscure genetic relationships among related populations. Kalinowski (2005) reported that loci with larger numbers of alleles produced esti- mates of genetic distance with lower coefficients of varia- tion than loci with fewer numbers of alleles, without requiring larger sample sizes from each population. Given the results from the cluster analysis, variation in the number of individuals sampled per population likely did not result in misidentification of genetic relationships among populations. Inferences from the genetic relationships of popula- tions surveyed in our study were dependent upon ac- curate determination of population allele frequencies. Microsatellite alleles differ in size, but alleles of the same size at a locus in geographically separate popula- tions may not have the same origin as a result of size homoplasy. Convergent mutations in different lineages may produce alleles of the same size, with the result that there may be greater differentiation among lin- eages than revealed by analysis of size variation alone. However, with over 800 alleles observed across all loci in the study, the large amount of variation present 250 Fishery Bulletin 1 10(2) Kwakusdis even Kainet even Crane Bay even Skowquiltz even — Salmon Bay even 1 lartley Bav even Kilasoo even Borrowman even — Evelyn even — Barnard even Kitkiata even — West Arm even — Neckas even Amoup even — Turn even — Clatse even Nekite even - Martin even 0.002 Central Keogh even — Embiey even -Wakeman even — Big Qualicum even — Phillips even - Clearwater even rfi Kakweiken even — Ahta even Kvvalate even Glendale even Klinaklini even -Clear even Ahnuhati even South / ECV1 — Ambaek even — Atnarko even Kcmano even Clyak even - Chuckwalla even Dallery even —Green even Khutze even ■ Kxngeal even 100 Even — Bel owe even - Wilauks even IJliancc even Hankin even Head even - Stewart even -Elcho even - Kiltuish even "Crag even Bullock Channel even Miiton even - Quartcha even Spi Her even Skedans even Centra! / North -Johnston even - Security even QC1 I Central Lignite even Datlaman even Mamin even Yakoun even Gregory even — Kano Inlet even Celestial even T asu even - Browns Cabin even Big Goose even Windy Bay even — Copper even Pallant even Tarundl even - 1 lonna even Decna even — Echo 1 Iarbour even QCI Salmon even - La I Iou even -Stumaun even — Sandy Bay even — Shaw even -Gilttoyees even Tlell even - Bish even — Duthie even Central / North QCI I Central -McNichol even Ensheshese even — Mouse even — Little Tsamspanaknok even — Big Tsamspanaknok even — Dogfish even Crow Lagoon even North Figure 2 Neighbor-joining dendrogram of Cavalli-Sforza and Edwards (1967) chord distance for 146 odd-year and 116 even-year populations of pink salmon (Oncorhynchus gorbuscha ) surveyed at 16 microsatellite loci. Bootstrap values at major tree nodes indicative of a regional group of populations display the percentage of 500 trees where populations beyond the node clustered together. QCI = Queen Charlotte Islands and ECVI = east coast Vancouver Island. Beacham et at: Population structure of Oncorhynchus gorbuscha in British Columbia and Washington, determined with microsatellites 251 Quaal even Kwinamass even Killope even Lizard even Kincolith even Gill even Lachmaeh even Toon even Kateen even — Mussel even Kleanza even Kalum even Kitwanga even — Babine even — Bulklcy even Kispiox even Lakelse even Khutzeymateen even Larch even Chambers even Cedar even Kitimat even Silver even Oona even Kumealon even Pa-aat even Central / North Skccna North Alpha even W 1 1 auks odd i oo Odd Illiance odd — Gilttoyees odd — Foch odd Cedar odd Skowquiltz odd — Kitimat odd — Kiltuish odd — Bish odd Weewante odd — Union odd Toon odd — Tsimitack odd — Copper odd — Tlellodd Central / North Central Central / North QCI - Stewart odd Alpha odd — Sandy Bay odd — Crow Lagoon odd Kxngeal odd -—Head odd — Han kin odd - Kumealon odd lit - Spiller odd - Oona odd -Silver odd LalTou odd — Stumaun odd — Lizard odd Dogfish odd Khutzeymateen odd -Chambers odd — Mouse odd it — Kwinamass odd — Quaal odd j — Eagle Bay odd II — Ensheshese odd — L.Tsamspanaknok odd Big Tsamspanak Crag odd — Larch odd — Lachmaeh odd — Kateen odd — Evelyn o Kitkiata odd Amoup odd - Martin odd North Central / North Central North - Tyler odd Nais odd - Kainet odd East Arm odd - Hartley Bay odd - West Arm odd -Gill odd - Barnard odd — Checnis Lake odd - Nckite odd Cooper Inlet odd — Kwakusdis odd Central Figure 2 (continued) 252 Fishery Bulletin 1 10(2) Central Central / South Central EC VI South Mamma Mamma odd Grev Wo if odd Dungencss odd Lower Fraser Upper Fraser Washington I iood Canal odd Figure 2 (continued) Beacham et al: Population structure of Oncorhynchus gorbuscha in British Columbia and Washington, determined with microsatellites 253 at these loci largely compensates for size homoplasy (Estoup et al., 2002), and therefore the pattern of popu- lation structure identified was unlikely to be obscured or distorted by size homoplasy. The distribution of genetic variance indicated that differences in allele frequencies between the brood- lines were about three times greater than differences for all regional and population sources of variation, but that regional differences in allele frequencies were only about 1.4 times larger than differences among populations within regions. Previous analyses of an- nual variation in allele frequencies within populations have indicated that this source of variation can be as large as differences among populations within regions (Golovanov et al., 2009). Although not specifically il- lustrated in the current study because of limitations of the software to facilitate a four-level nested analysis of variance, similar results (not shown) were observed in the current study. The larger importance of annual variation within populations relative to population dif- ferentiation compared with other Pacific salmon species likely reflected the reduced population differentiation observed in pink salmon relative to that observed in other salmon species (Beacham et al., 2009, 2011). The current study indicates that the largest deter- minant of population structure of pink salmon in Brit- ish Columbia and Washington was year of spawning (odd or even), with a distinct separation of the two broodlines. With the odd-year broodline, regional dif- ferentiation was stronger in the southern portion of the distribution of populations, with the greatest population differentiation within a region observed in Washing- ton. The Hood Canal hatchery population (also known as Hoodsport hatchery 47°23'37"N, 123°08'54"W) was the most distinct, even though this population was derived from adults returning to the Dungeness River and Dosewallips River in 1953 (Hard et al., 1996). The distinctiveness of this population was reflected in allelic frequency differentiation. For example, the frequency of the Onel023U allele was 0.54 in the Hood Canal hatchery population, 0.34 in the Gray Wolf River popu- lation, and <0.20 in all other Washington populations. Additionally, the frequency of Onel09133 was 0.33 in the Hood Canal hatchery population, but <0.20 in all other Washington populations. All populations from drainages entering Hood Canal (Hood Canal hatchery, Dosewal- lips River, Duckabush River, Hamma Hamma River) or the Strait of Juan de Fuca (Dungeness River, Gray Wolf River) were distinct from those on the eastern side of Puget Sound (Snohomish River, Stillaguamish River, Skagit River, Green River, Puyallup River). Genetic separation of Strait of Juan de Fuca populations and Hood Canal populations from those in eastern Puget Sound was initially described by Shaklee et al. (1991) in an analysis of allozyme variation. However, as described by Shaklee et al. (1991), the Nooksack River population, located in the northeastern section of Puget Sound and nearest to the border with Canada, clustered with Hood Canal and Strait of Juan de Fuca populations, rather than with geographically closer populations on the east side of Puget Sound. Similar results were observed in the current study. Shaklee et al. (1991) suggested that the genetic similarity of the Nooksack River population to that of Hood Canal populations was a consequence of a 1977 transfer of fertilized eggs from the Hood Canal hatchery to a tributary of the Nooksack River and a reduction of the native population due to habitat degra- dation. As Shaklee et al. (1991) outlined, this enhance- ment effort may have caused a genetic change in the characteristics of this population that has persisted over time (Hard et al., 1996). Fraser River populations were separate from those in southern British Columbia (east coast of Vancouver Island, south coast mainland) and Washington, confirm- ing the results from the previous analysis of allozyme variation reported by Beacham et al. (1988) and Shak- lee et al. (1991). In the Fraser River drainage, some separation was observed between populations spawning upstream from the Fraser River canyon (southern limit approximately 175 km upstream from the mouth) from those spawning downstream of the canyon. Genetic separation between upriver and downriver populations had also been had been outlined previously by Beacham et al. (1988) and Shaklee et al. (1991). Similar genetic separation between upper drainage and lower drain- age populations has been observed in coho salmon (O. kisutch) (Beacham et al., 2011) and reflects geographic separation between the two groups of populations. In northern British Columbia, odd-year broodline populations in the Skeena River drainage were separate from those farther south in the central coastal region of British Columbia and from those farther north on the northern coastal region of British Columbia in a simi- lar pattern to that outlined by Beacham et al. (1988). Similar differentiation was also observed in the even- year broodline, with Skeena River drainage populations distinct from other populations in northern British Columbia. Some differentiation was observed in the current study between even-year broodline pink salmon populations from the Queen Charlotte Islands and other regions in northern British Columbia (central coast, Skeena River, north coast) ( FST =0.004-0.006); differen- tiation of populations from the Queen Charlotte Islands had also been observed by Beacham et al. in 1988. Studies of population structure in Pacific salmon are a useful initial step in developing and applying genetic variation to the problem of estimating stock composition in mixed-stock salmon fisheries. The key to successful application of genetic variation to estimation of stock composition centers around whether or not there is a regional basis to population structure. This is a key consideration because a regionally based population structure is generally required for genetic stock identi- fication estimation, with the assumption that the por- tion of the mixed-stock sample derived from unsampled populations is allocated to sampled populations from the same region. With this assumption, the cost and com- plexity of developing a baseline for stock composition analysis is reduced, and refinements in estimated stock compositions are possible as the baseline is enhanced 254 Fishery Bulletin 1 10(2) in stages. For the odd-year broodline, applications in southern British Columbia would appear to be possible if fishery management objectives are to separate pink salmon of Washington, Fraser River, and southern Brit- ish Columbia origin. Finer subdivision of stock composi- tion estimation, particularly in the Washington region, may be possible, as separation of Hood Canal and Strait of Juan de Fuca populations from those in Puget Sound may be practical. Studies of population structure in pink salmon have revealed some consistent patterns. The great- est differentiation observed in population structure has been consistently reported to occur between the two broodlines, whether in Asia or North America (Beacham et al., 1988; Kartavtsev, 1991; Varnavs- kaya and Beacham, 1992; Zhivotovsky et ah, 1994; Salinenkova et ah, 2006; Golovanov et ah, 2009). In Asia, studies have indicated that genetic differen- tiation among populations is greater in the even-year broodline than in the odd-year broodline (Hawkins et ah, 2002; Golovanov et ah, 2009). In North America, the reverse situation occurs, with population differ- entiation among populations greater in the odd-year broodline than in the even-year broodline (Beacham et ah, 1988; Gharrett et ah, 1988; current study). These findings support the concept of two main refugia oc- cupied by pink salmon during the most recent Pleisto- cene Era glaciation some 10,000 years ago (Aspinwall, 1974). The even-year broodline may have survived the glaciation in a northern refugium (Aspinwall, 1974). Once the glaciation ended, the even-year broodline dispersed from the northern refugium, colonizing southern regions more recently than northern ones. Conversely, the odd-year broodline may have occupied a southern refugium during the Pleistocene Era gla- ciation (McPhail and Lindsey, 1970), and dispersed northward, with northern populations derived more recently than southern ones. As populations closer to the refugium have had greater time to accumulate genetic mutations and thus display greater population differentiation, the current pattern of broodline and population differentiation is consistent with dispersal from a northern refugium for the even-year broodline (greater population genetic differentiation in even-year broodline) and dispersal from a southern origin for the odd-year broodline (greater population differentiation in odd-year broodline). Additionally, embryonic sur- vival of the even-year broodline has been reported to be higher than that of the odd-year broodline in a cold (4°C) incubation environment, with higher alevin and fry growth of the even-year broodline also observed in the cold incubation environment (Beacham and Murray, 1988). Greater suitability of the even-year broodline to a colder environment is also illustrated by the spawning distributions of the broodlines in North America, with the even-year broodline in very low abundance from the southern portion of the range (Fraser River, Washington) and the odd-year broodline in low abundance in western Alaska. Alternatively, Krkosek al. (2011) suggested that the distribution of even-year and odd-year populations result from densi- ty-dependent mortality caused by interactions between the broodlines. However, it seems difficult to account for genetic population structure observed in pink salm- on as a result of broodline interactions. Conclusion The level of differentiation observed among the pink salmon populations within broodlines surveyed in the current study was considerably less than in other species of Pacific salmon. Sockeye salmon (O. nerka) typically display high levels of genetic differentiation ( FST =0.097, 14 loci, average 30 alleles per locus, 299 populations) (Beacham et al., 2006), with the other species displaying levels of genetic differentiation ranging between sockeye salmon and pink salmon. The low level of differentia- tion observed in pink salmon may be a result of a more recent colonization history (Hawkins et al., 2002), but may also be a result of straying among local populations within regions. As pink salmon juveniles spend little time in fresh water after fry emergence, imprinting on natal streams may not be as strong as in other species, and as a result may stray more upon returning spawn- ing migrations (Quinn, 1993). Chum salmon ( O . keta ) juveniles spend similar amounts of time in fresh water as pink salmon, and population differentiation in the species is higher only than pink salmon ( FST =0.033, 14 loci, average 57 alleles per locus, 380 populations) (Beacham et al., 2009). The low level of genetic differ- entiation observed in pink salmon population structure likely reflects higher levels of straying among popula- tions during spawning than those observed for other Pacific salmon species. Acknowledgments A very substantial effort was undertaken to obtain the pink salmon samples used in this study. We thank vari- ous staff of Fisheries and Oceans Canada (DFO), the Pacific Salmon Commission (PSC), and the Washington Department of Fish and Wildlife (WDFW) for sample collection, as well as First Nations staff. We acknowl- edge those within the Kitasoo Fisheries Program who sampled pink salmon, as well as the Gitxsan Watershed Authority, the Kitselas and Kitsumkalum field staff, the Skeena Fisheries Commission, the Nisga’a First Nation, the Haida Fisheries Program, and the crew of the Canadian Coast Guard Vessel Arrow Post. L. Fitzpatrick drafted the map. C. Wallace assisted in the analysis. Funding for the study was provided by DFO and the PSC. Literature cited Aro, K. V., and M. P. Shepard. 1967. Salmon of the North Pacific Ocean, part V. Spawn- Beacham et al.: Population structure of Oncorhynchus gorbuscha in British Columbia and Washington, determined with microsatellites 255 ing populations of the North Pacific Ocean. Pacific salmon in Canada. Int. N. Pac. Fish. Comm. Bull. 23:225-337. Aspinwall, N. 1974. Genetic analysis of North American populations of the pink salmon, Oncorhynchus gorbuscha', possible evidence for the neutral mutation-random drift hypoth- esis. Evolution 28:295-305. Banks, M. A., V. K. Rashbrook, M. J. Calavetta, C. A. Dean, and D. Hedgecock. 2000. Analysis of microsatellite DNA resolves genetic structure and diversity of Chinook salmon ( Oncorhyn- chus tshawytscha) in California’s Central Valley. Can. J. Fish. Aquat. Sci. 57:915-927. Beacham, T. D. 1985. Meristic and morphometric variation in pink salmon (Oncorhynchus gorbuscha) in southern British Columbia and Puget Sound. Can. J. Zool. 63:366-372. Beacham, T. D., J. R. Candy, K. D. Le, and M. Wetklo. 2009. Population structure of chum salmon ( Oncorhynchus keta) across the Pacific Rim determined from microsat- ellite analysis. Fish. Bull. 107:244-260. Beacham, T. D., B. McIntosh, C. MacConnachie, K. M. Miller, R. E. Withler, and N. V. Varnavskaya. 2006. Pacific Rim population structure of sockeye salmon as determined from microsatellite analysis. Trans. Am. Fish. Soc. 135:174-187. Beacham, T. D., and C. B. Murray. 1988. Variation in developmental biology of pink salmon (Oncorhynchus gorbuscha) in British Columbia. Can. J. Zool. 66:2634-2648. Beacham, T. D., M. Wetklo, L. Deng, and C. MacConnachie. 2011. Coho salmon population structure in North America determined from microsatellites. Trans. Am. Fish. Soc. 140:253-270. Beacham, T. D., R. E. Withler, and A. P. Gould. 1985. Biochemical genetic stock identification of pink salmon (Oncorhynchus gorbuscha) in southern British Columbia and Puget Sound. Can. J. Fish. Aquat. Sci. 42:1474-1483. Beacham, T. D., R. E. Withler, C. B. Murray, and L. W. Barner. 1988. Variation in body size, morphology, egg size, and biochemical genetics of pink salmon in British Columbia. Trans. Am. Fish. Soc. 117:109-126. Bilton, H. T., and W. E. Ricker. 1965. Supplementary checks on the scales of pink salmon (Oncorhynchus gorbuscha) and chum salmon (Oncorhynchus keta). J. Fish. Res. Board Canada 22:1477-1489. Brykov, V. A., N. Polyakova, L. A. Skurikhina, and A. D. Kukhlevsky. 1996. Geographical and temporal mitochondrial DNA variability in populations of pink salmon. J. Fish Biol. 48:899-909. Brykov, V. A., N. Polyakova, L. A. Skurikhina, A. D. Kukhlevsky, O. N. Kirillova, D. Churikov, A. I. Pudovkin, and A. J. Gharrett. 1999. Analysis of mtDNA indicates weak temporal genetic heterogeneity in pink salmon spawning runs in two rivers on Sakhalin Island. J. Fish Biol. 55:617- 635. Cairney, M., J. B. Taggart, and B. Hoyheim. 2000. Characterization of microsatellite and minisatel- lite loci in Atlantic salmon (Salmo salar L.) and cross- species amplification in other salmonids. Mol. Ecol. 9:2175-2178. Cavalli-Sforza, L. L., and A. W. F. Edwards. 1967. Phylogenetic analysis: models and estimation pro- cedures. Am. J. Hum. Genet. 19:233-257. Churikov, D., and A. J. Gharrett. 2002. Comparative phylogeography of the two pink salmon broodlines: an analysis based on a mitochondrial DNA genealogy. Mol. Ecol. 11:1077-1101. Dyagilev, S. Ye., and N. B. Markevich. 1979. Different times of maturation of pink salmon, Oncorhynchus gorbuscha, in even and uneven years as the main factor responsible for different acclimatization results in the northwestern U.S.S.R. J. Ic-hthol. 19(2): 30—44 . Estoup, A., P. Jarne, and J. M. Cornuet. 2002. Homoplasy and mutation model at microsatel- lite loci and their consequences for population genetics analysis. Mol. Ecol. 11:1591-1604. Felsenstein J. 1993. PHYLIP: phylogeny inference package. Univ. Washington, Seattle, WA. Gharrett, A. J., C. Smoot, A. J. McGregor, and P. B. Holmes. 1988. Genetic relationships of even-year northwestern Alaskan pink salmon. Trans. Am. Fish. Soc. 117:536- 545. Godfrey, H. 1959. Variations in annual average weight of British Columbia pink salmon, 1944-1958. J. Fish. Res. Board Can. 16:329-337. Golovanov, I. S., S. L. Marchenko, and S. P. Pustovoit. 2009. Genetic monitoring of northern sea of Okhotsk populations of pink salmon ( Oncorhynchus gorbuscha). Cytol. Genet. 43(61:18-27. Goudet, J. 1995. FSTAT A program for IBM PC compatibles to calculate Weir and Cockerham’s (1984) estimators of F-statistics (version 1.2). J. Heredity 86:485-486. Greig, C., J. P. Jacobson, and M. A. Banks. 2003. New tetranucleotide microsatellites for fine-scale discrimination among endangered Chinook salmon (Oncorhynchus tshawytscha). Mol. Ecol. Notes 3:376- 379. Hard, J. J., R. G. Kope, W. S. Grant, F. W. Waknitz, L. T. Parker, and R. S. Waples. 1996. Status review of pink salmon from Washington, Oregon, and California. NOAA Tech. Memo. NMFS- NWFSC-25, 141 p. Hawkins, S. L., N. V. Varnavskaya, E. A. Matzak, V. V. Efremov, C. M. Guthrie III, R. L. Wilmot, H. Mayama, F. Yamazaki, and A. J. Gharrett. 2002. Population structure of odd-broodline Asian pink salmon and its contrast to the even-broodline structure. J. Fish Biol. 60:370-388. Kalinowski, S. T. 2005. Do polymorphic loci require large sample sizes to estimate genetic distances? Heredity 94:33-36. Kartavtsev, Y. F. 1991. Temporal and spatial variability in allele frequen- cies in populations of pink salmon, Oncorhynchus gor- buscha. J. Ichthyol. 31:87-98. Krkosek, M. R. Hilborn, R. M. Peterman, and T. P. Quinn. 2011. Cycles, stochasticity, and density dependence in pink salmon population dynamics. Proc. Royal Soc. Series B 278:2060-2068. McPhail, J. D., and C. C. Lindsey. 1970. Freshwater fishes of northwestern Canada and Alaska. Bull. Fish. Res. Board Can. 173, 381 p. 256 Fishery Bulletin 1 10(2) Neave, F. 1952. “Even-year” and “odd-year” pink salmon popula- tions. Trans. R. Soc. Can. 46(series 3, section 5 ) : 55 — 70. Noll, C., N. V. Varnavskaya, E. A. Matzak, S. L. Hawkins, V. V. Midanaya, O. N. Katugin, C. Russell, N. M. Kinas, C. M. Guthrie III, H. Mayama, F. Yamazaki, B. P. Finney, and A. J. Gharrett. 2001. Analysis of contemporary genetic structure of even- broodyear populations of Asian and western Alaskan pink salmon, Oncorhynchus gorbuscha. Fish. Bull. 99:123-138. Olsen, J. B., P. Bentzen, M. A. Banks, J. B. Shaklee, and S. Young. 2000a. Microsatellites reveal population identity of individual pink salmon to allow supportive breeding of a population at risk of extinction. Trans. Am. Fish. Soc. 129:232-242. Olsen, J. B., L. W. Seeb., P. Bentzen, and J. E. Seeb. 1998. Genetic interpretation of broad-scale microsatellite polymorphism in odd-year pink salmon. Trans. Am. Fish. Soc. 127:535-550. Olsen, J. B., S. L. Wilson, E. J. Kretschmer, K. C. Jones, and J. E. Seeb. 2000b. Characterization of 14 tetranucleotide micro- satellite loci derived from sockeye salmon. Mol. Ecol. 9:2185-2187. Perriere, G., and M. Gouy. 1996. WWW-query: An on-line retrieval system for biological sequence banks. Biochi mie 78:364- 369. Quinn, T. P. 1993. A review of homing and straying of wild and hatchery-produced salmon. Fish. Res. 18:29-44. Rice, W. R. 1989. Analyzing tables of statistical tests. Evolution 43:223-225. Ricker, W. E. 1972. Heredity and environmental factors affecting cer- tain salmonid populations. In The stock concept in Pacific salmon (R. C. Simon and P. A. Larkin, eds.), p. 27-160. H. R. MacMillan Lectures in Fisheries, Univ. British Columbia, Inst. Fisheries, Vancouver, Canada. Salmenkova, E. A., N. V. Gordeeva, , V. T. Omel’chenko, Yu. P. Altukhov, K. I. Afanas’ev, G. A. Rubtsova, and Yu. V. Vasil’eva. 2006. Genetic differentiation of pink salmon Oncorhyn- chus gorbuscha in the Asian part of the range. Rus. J. Genet. 42:1148-1163. Shaklee, J. B. and P. Bentzen. 1998. Genetic identification of stocks of marine fish and shellfish. Bull. Mar. Sci. 62:589-621. Shaklee, J. B., D. C. Klaybor, S. Young, and T. F. Cross. 1991. Genetic stock structure of odd-year pink salmon Oncorhynchus gorbuscha (Walbaum) from Washington and British Columbia and potential mixed-stock fisheries applications. J. Fish Biol. 39 (suppl. A):21-34. Shaklee, J. B., and N. V. Varnavskaya. 1994. Electrophoretic characterization of odd-year pink salmon (Oncorhynchus gorbuscha) populations from the Pacific coast of Russia, and a comparison with selected North American populations. Can. J. Fish. Aquat. Sci. 51 (suppi. 1 ) : 158— 17 1 . Small, M. P., T. D. Beacham, R. E. Withler, and R. J. Nelson. 1998. Discriminating coho salmon ( Oncorhynchus kisutch ) populations within the Fraser River, British Columbia using microsatellite DNA markers. Mol. Ecol. 7:141-155. Smith, C. T., B. F. Koop, and R. J. Nelson. 1998. Isolation and characterization of coho salmon ( Oncorhynchus kisutch) microsatellites and their use in other salmonids. Mol. Ecol. 7:1613-1621. Varnavskaya, N. V., and T. D. Beacham. 1992. Biochemical genetic variation in odd-year pink salmon (Oncorhynchus gorbuscha) from Kamchatka. Can. J. Zool. 70:2115-2120. Waples, R. S. 1990. Temporal changes of allele frequency in Pacific salmon populations: implications for mixed-stock fishery analysis. Can. J. Fish. Aquat. Sci. 47:968-976. Weir, B.S., and C. C. Cockerham. 1984. Estimating F-statistics for the analysis of popula- tion structure. Evolution 38:1358-1370. Williamson, K. S., J. F. Cordes, and B. P. May. 2002. Characterization of microsatellite loci in Chinook salmon (Oncorhynchus tshawytscha) and cross-species amplification in other salmonids. Mol. Ecol. Notes 2:17-19. Wright, J. J., K. P. Lubieniecki, J. W. Park, S. H. S. Ng, R. H. Devlin, J. Leong, B. F. Koop and W. S. Davidson. 2008. Sixteen Type 1 polymorphic microsatellite mark- ers from Chinook salmon (Oncorhynchus tshawytscha) expressed sequence tags. Animal Genet. 39:84-85. Zhivotovsky, L. A., A.- J. McGregor, K. Glubokovsky, and M. W. Feldman. 1994. Gene differentiation in Pacific salmon ( Oncorhyn- chus sp.hfads and models with reference to pink salmon (O. gorbuscha). Can. J. Fish. Aquat. Sci. 51(suppl. l):223-232. 257 Abstract — We evaluated measures of bioelectrical impedance analysis (BIA) and Fulton’s condition factor (K) as potential nonlethal indices for detecting short-term changes in nutritional condition of postsmolt Atlantic salmon ( Salmo salar). Fish reared in the laboratory for 27 days were fed, fasted, or fasted and then refed. Growth rates and proximate body composition (protein, fat, water) were measured in each fish to evalu- ate nutritional status and condition. Growth rates of fish responded rap- idly to the absence or reintroduction of food, whereas body composition (% wet weight) remained relatively stable owing to isometric growth in fed fish and little loss of body con- stituents in fasted fish, resulting in nonsignificant differences in body composition among feeding treat- ments. The utility of BIA and Ful- ton’s K as condition indices requires differences in body composition. In our study, BIA measures were not sig- nificantly different among the three feeding treatments, and only on the final day of sampling was K of fasted vs. fed fish significantly different. BIA measures were correlated with body composition content; however, wet weight was a better predictor of body composition on both a content and concentration (% wet weight) basis. Because fish were growing isometrically, neither BIA nor K was well correlated with growth rate. For immature fish, where growth rate, rather than energy reserves, is a more important indicator of fish condition, a nonlethal index that reflects short- term changes in growth rate or the potential for growth would be more suitable as a condition index than either BIA measures or Fulton’s K. Manuscript submitted 16 February 2011. Manuscript accepted 10 January 2012. Fish. Bull. 110:257-270 (2012). The views and opinions expressed or implied in this article are those of the author (or authors) and do not necessarily reflect the position of the National Marine Fisheries Service, NOAA. Evaluation of bioelectrical impedance analysis and Fulton's condition factor as nonlethal techniques for estimating short-term responses in postsmolt Atlantic salmon ( Salmo salar ) to food availability Elaine M. Caldarone (contact author)1 Sharon A. MacLean1 Beth Sharack2 Email address for contact author elaine.caldarone@ noaa.gov ' Narragansett Laboratory Northeast Fisheries Science Center National Marine Fisheries Service National Oceanic and Atmospheric Administration 28 Tarzweil Drive Narragansett, Rhode Island 02882 2 J. j. Howard Laboratory Northeast Fisheries Science Center National Marine Fisheries Service National Oceanic and Atmospheric Administration 74 McGruder Road Highlands, New Jersey 07732 Populations of Atlantic salmon ( Salmo salar) are broadly distributed across the North Atlantic Ocean, repro- ducing in coastal rivers of Iceland, Europe, northwestern Russia, and northeastern North America. Histori- cally, native Atlantic salmon ranged throughout New England waters, but by the late 1880s, they were extir- pated from many of the rivers. Cur- rently, native populations exist only in central and southeast Maine (North- east Fisheries Science Center, http:// www.nefsc.noaa.gov/sos, accessed July 2011). A significant decline in these populations during the 1990s resulted in the listing of the Gulf of Maine Distinct Population Segment (DPS) as endangered under the United States Endangered Species Act (Federal Reg- ister, 2009). Hatchery-based restora- tion of salmon to this area began in 1970 and continues today. Although all life-stages have been released to the rivers, recent emphasis has focused on releasing fry and smolts produced in the hatchery from field-caught adults from the DPS region. Once smolts reach the sea, they must adapt to an environment and food source radi- cally different from their freshwater habitats. To evaluate the success of these restoration efforts, managers need tools to assess whether hatchery- reared fish are thriving in the natural environment and to assess condition of the postsmolt Atlantic salmon popula- tion as a whole. Growth rate and fat content are key measures of condition in fish. During a fish’s early life-history stage, rapid growth rates increase the probability of survival and recruitment, primar- ily through decreased vulnerability to predation and starvation (see re- view by Fonseca and Cabral, 2007). Fat, the primary energy repository in marine fish, increases when food intake exceeds metabolic needs and decreases during food-limited times when it provides energy for main- tenance, growth, and reproduction (Shulman and Love, 1999; Jobling, 2001). Growth rates of fish caught in the wild are often estimated by us- ing nucleic acid analysis (the ratio of RNA to DNA) or otolith microstruc- ture analysis (Chambers and Miller, 258 Fishery Bulletin 110(2) 1995). The former requires removal of a plug of muscle tissue (MacLean et al., 2008), and the latter requires removal of an otolith, a lethal procedure. Direct mea- surement of fat content is also lethal, requiring chemi- cal analysis of a sacrificed fish. Identifying a minimally invasive, nonlethal index that could estimate growth rate or body composition in postsmolt salmon would al- low restoration managers to evaluate the condition of field-captured fish. We chose to evaluate two nonlethal techniques which had the potential to reflect fish condition: bioelectrical impedance analysis (BIA) and Fulton’s condition factor (hereafter, also called Fulton’s K). BIA is a technique which has been applied to humans and other mammals as a means to estimate nutritional status and body composition (Baumgartner et al., 1988; Marchello and Slanger, 1994; Schwenk et al., 2000; Barbosa-Silva et al., 2003). Recently, BIA has been used to estimate body composition content in fish (Cox and Hartman, 2005; Pothoven et al., 2008; Hanson et al., 2010). For BIA, a small, portable, battery-operated instrument is used to generate a mild alternating current between two sets of electrodes that have been placed on the subject. The resulting voltage drop is recorded as resistance ( R ) and capacitive reactance ( Xc ) in series. When an alternating current passes around a cell, R can be af- fected by extracellular water (good conductor) and fat (poor conductor). When a constant signal frequency is applied, a geometrical system can be modeled as a cylinder (conductor volume = pL2/R, where L is length) (Lukaski et al., 1985). With this conductor volume ap- proach, predictive equations have been constructed to estimate water content and fat-free mass in humans (Lukaski et al., 1985; Chumlea et al., 2002), and water-, fat-, protein-, and ash-content in fish (Cox and Hart- man, 2005; Pothoven et al., 2008; Hanson et al., 2010). Impedance values have also been used to calculate a condition index (phase angle, arctangent Xc/R converted to degrees) in both humans and fish (Schwenk et al., 2000; Barbosa-Silva et al., 2003; Cox and Heintz, 2009). In an organism, Xc is a measure of the phase shift that results from an electrical charge being momentarily stored in the double phospholipid layer of a cell mem- brane. When cells die, Xc drops to zero; phase angles thus range from zero (zeroXc, all cells dead) to 90° (ze- ro R). In humans, lower phase angle values have been associated with conditions such as reduced survival in HIV-infected patients, and malnutrition (Schwenk et al., 2000; Barbosa-Silva et al., 2003). In fish, signifi- cant decreases in phase angles have been observed in juvenile rainbow trout ( Oncorhynchus mykiss) and brook trout ( Salvelinus fontinalis) after three weeks of fasting, and in juvenile Chinook salmon (O. tshawytscha) after eight weeks of fasting (Cox and Heintz, 2009). In those studies, fish were repeatedly measured over a period of weeks with no mortalities associated with the BIA procedure — an attribute that made it attractive for our application. Fulton’s K (weight/length3) (Ricker, 1975) is a widely used fish condition index based on morphometries (e.g., Anderson and Gutreuter, 1983; Stevenson and Woods Jr., 2006), measurements that can be obtained easily in the field. This index is based on the assumption that within a cohort, individuals with higher K values (more rotund fish) contain more energy reserves (fat and pro- tein), and thus are in better condition than those with lower K values. The response time of a condition index can be af- fected by factors such as water temperature, life-stage, season, and species (Busacker et al., 1990). Cox and Heintz (2009) measured phase angles in food-deprived rainbow trout and brook trout on a weekly basis, and in food-deprived Chinook salmon intermittently for 13 weeks. Because our field recaptures of hatchery-reared postsmolts occur two to three weeks after release, we measured response of BIA measures and K to varying food availability every 3-4 days throughout a 3-week time period. Thus the objectives of our study were 1) to assess and validate the relations between two nonlethal condition indices (BIA measures and Fulton’s condi- tion factor) and two measures of nutritional condition (growth rate and body composition); and 2) to determine the short-term response time (days to a few weeks) of these measures to varying food availability. Materials and methods Smolts used in this study were progeny of field-caught Atlantic salmon from the Penobscot River, Maine, which had been spawned at Craig Brook National Fish Hatch- ery, East Orland, Maine, and reared at the Green Lake National Fish Hatchery, Ellsworth, Maine, for 13-15 months. Randomly selected smolts (52-113 g, 16-21 cm) were anesthetized in buffered tricaine methane sulfonate (MS-222, 150 mg/L) and implanted intra- muscularly with a passive integrated transponder tag (PIT tag, Biomark, Boise, ID1) to permit identification of individuals. The smolts were then returned to the hatchery tank to allow time for full recovery, resumption of feeding, and removal of any tagging-related mortali- ties. Twenty-five days later the fish were transported to the University of Rhode Island’s Blount Aquarium facility in Narragansett, Rhode Island, where they were randomly placed into two aerated, flow-through tanks (360-L capacity) initially filled with freshwater trucked from the hatchery. Over a period of five to six hours, freshwater was gradually replaced with sand-filtered seawater (10°C, 30 ppt). During the subsequent three weeks, while the fish were recovering from the transfer and acclimating to seawater, the water temperature was gradually raised to 12°C. During this period fish were fed to satiation twice per day with a commercial feed (Corey Optimum Hatchery Feed for Salmonids, Corey Nutrition Co., Fredericton, NB, Canada) supplemented with freeze-dried krill ( Euphausia pacifica, Aquatic 1 Mention of trade names or commercial companies is for identification purposes only and does not imply endorsement by the National Marine Fisheries Service, NOAA. Caldarone et al.: Nonlethal techniques for estimating responses of postsmolt Salmo solar to food availability 259 Table 1 Sampling schedule for Atlantic salmon ( Salmo salar ) postsmolts reared at 12°C under three feeding regimens (fed; fasted; fasted then refed) in order to obtain a range of nutritional condition and growth rates. Nonlethal condition indices (Fulton’s condition factor \K\ and bioelectrical impedance analysis [BIA] measures) and two measures of nutritional condition (wet-weight based growth rate and proximate body composition) were determined for each fish. The refed group was fasted for 11 days and then fed. Numbers listed are number of fish sampled. Sampling day and feeding regimen Day 0 Day 3 Day 7 Day 11 Base-line Fed Fasted Refed Fed Fasted Refed Fed Fasted Refed Fed Fasted Refed Weight, length 5 24 24 22 4 4 0 4 4 0 4 4 22 (Fulton’s K) BIA measures 5 24 24 0 4 4 0 4 4 0 4 4 0 Proximate body 5 0 0 0 4 4 0 4 4 0 4 4 0 composition Sampling day and feeding regimen Day 15 Day 19 Day 23 Day 27 Fed Fasted Refed Fed Fasted Refed Fed Fasted Refed Fed Fasted Refed Weight, length (Fulton’s K) 4 4 5 4 4 5 4 4 5 0 0 7 BIA measures 4 4 5 4 4 5 4 4 5 0 0 7 Proximate body composition 4 4 5 4 4 5 4 4 5 0 0 7 Eco-Systems, Inc., Apopka, FL). Twenty-five days after the initial seawater transfer, when the now postsmolts appeared to be acclimated and feeding well, the experi- ment commenced (day 0). Throughout the experiment, water temperature in each tank was recorded hourly with a HOBO® data log- ger (Onset Computer Corp., Bourne, MA), and ammonia levels and salinity were tested weekly. Water tempera- tures averaged 12.0°C, standard deviation (SD) = 0.2; salinity averaged 31 ppt, SD = 1; and the photoperiod was 15 hours of light to 9 hours of dark. Two-thirds of each tank surface was covered with black plastic to provide a low-light refuge, and the remaining third was exposed to overhead fluorescent lighting covered with red plastic. All experiments were conducted in accordance with guidelines established by the Institu- tional Animal Care and Use Committee (IACUC) at the University of Rhode Island. Sampling protocols Day 0 sampling On day 0, five fish were randomly selected and sacrificed to provide baseline body compo- sition data. To obtain a range of nutritional condition levels, all remaining postsmolts were subdivided into three different feeding treatments (tanks); fed, fasted, and fasted then refed. The fed treatment (re = 24) was continually fed, the fasted treatment (re =24) received no food, and the fasted, then refed treatment (re =22) received no food for 11 days followed by feeding for 16 days. Individuals were anesthetized with buffered MS-222 (150 mg/L) in chilled (12°C) seawater, blotted dry, measured for initial weight (wet weight, WW, near- est 0.1 g) and fork length (FL, nearest 0.1 cm), and the PIT tag number and any gross external abnormalities were noted. BIA measurements were taken on all fish assigned to the fed and fasted treatments. Fish assigned to the fasted, then re fed treatment had their fork length and wet weight recorded at the start of their fasting (day 0), and again on day 11 when they were fed for the remainder of the experiment. Total handling time per fish for wet weight, fork length, and BIA measurements was no more than 30 seconds. During daylight hours, fish in the fed treatment and refed group were provided freeze-dried krill ad libitum from a belt feeder. On days fish were sampled, no food was provided until after sampling was completed (-1200). Days 3-27 sampling In order to determine the response time of BIA measures and K to the three feeding treat- ments, and to construct predictive equations for body composition, fish were sampled and sacrificed every three to four days over a 23-27 day period. On day 3 and every 4 days thereafter, 4 fish each from the fed and fasted treatments were sacrificed. On day 15 and every 4 days thereafter, 5 fish from the refed group were sacrificed and sampled, except on the final day when 7 fish were sacrificed (Table 1). 260 Fishery Bulletin 1 10(2) All fish were killed by an overdose of buffered MS- 222 (300 mg/L) in chilled (12°C) seawater. The fish were immediately blotted dry, measurements of wet weight, fork length, and BIA (in that order) were taken, and PIT tag numbers and external abnormalities were noted. Internal temperatures (muscle and stomach) were determined by inserting an instant-read digital thermometer into the dorsal musculature of the fish, and down the esophagus into the stomach. Total sam- pling time for each fish was ~1.5 min. Each fish was then dissected, its liver removed, its gut evacuated, and sex and maturity status was determined. Livers and carcasses were wrapped separately in aluminum foil before being vacuum-sealed in plastic bags and stored frozen at -80°C until subsequent analysis. Body composition analysis Liver and carcass wet weights were determined to the nearest 0.1 mg before being freeze-dried to a constant weight and reweighed. Each dried sample was ground in a Foss Tecator® Cyclotec 1093 sample mill (FOSS, Hilerqd, Denmark) and stored at -20°C in glass scintillation vials under nitrogen gas until further analysis. Total water ( TWa ) was calculated by subtracting total dry weight (liver dry weight plus carcass dry weight, DW) from total wet weight. Freeze-dried carcasses were analyzed for proximate body composition (protein, fat, ash) by an indepen- dent laboratory (A&L Great Lakes Laboratory, Fort Wayne, IN) by using Association of Official Analytical Chemists international certified methods and were reported to us on a percent DW basis. Nitrogen was determined by using a LECO nitrogen combustion ana- lyzer (LECO Corp., St. Joseph, MI; Dumas method), protein was calculated by multiplying nitrogen values by 6.25 (Jones, 1931), fat was obtained with a 4-hr ether reflux extraction, and ash was determined after combustion at 600°C for 2-4 hr. Body composition (g) (total amount of each proximate body constituent) were calculated from percent dry weight concentrations by dividing the independent laboratory values by 100 and multiplying by the total dry weight. To conform to the format most often reported in the literature, percent dry-weight-based concentrations were converted to a percent wet-weight-based concentration by dividing body composition (g) by wet weight and multiplying by 100 (body composition [%WW]). Liver lipids, which are often mobilized first during fasting (Love, 1970; Black and Love, 1986), were mea- sured separately in order to detect changes more easily. Liver lipid content was determined in-house by using a modification of Folch et al. (1957). Entire freeze-dried livers were first extracted by ultrasonic homogenization with 2:1 methylene chloride:methanol solvent (20 mL/g tissue), then back extracted with aqueous 0.1 M KC1 and centrifuged to remove water, methanol, and water- soluble and water-insoluble tissue components by phase separation. The remaining nonaqueous fractions were evaporated to remove methylene chloride, and the non- volatile lipid residue was weighed on a Mettler® AE240 balance (nearest mg, Mettler-Toledo, Inc., Columbus, OH). Individual livers were not analyzed for protein or ash because of their small size; therefore body composi- tion values were obtained for carcass water, liver water, total water (liver water plus carcass water), carcass fat, liver fat, total fat (liver fat plus carcass fat), carcass protein, and carcass ash. Growth-rate calculations Individual instantaneous wet-weight based growth rates were calculated with the following formula (Ricker, 1979): growth rate (per d) = (In WWl2-\nWWtl)l(t2- tj ) , (1) where WW = the wet weight of an individual at time t (day). Growth rates for fish in the fed and fasted treat- ments were calculated from day 0 (t;) until the day they were sacrificed ( t2 ). For the fasted portion of the fasted, then refed treatment, growth rates were cal- culated from day 0 (t7) until day 11 (<2); for the refed portion, growth rates were calculated from the first day of refeeding (day 11, t:) until the day they were sacrificed (t2). Growth rates calculated over intervals of less than five days were excluded from our analyses because minimal weight changes over those short time intervals, combined with the inherent variability of measuring wet weight, resulted in inaccurate growth rate estimates. BIA measurement protocol and BIA measures BIA measurements were determined with a Quantum- X® (RJL Systems, Point Heron, MI) four-electrode single frequency (800 pA, 50 KHz) analyzer. Needle electrode probes were constructed in-house according to Cox and Hartman (2005). For each probe, two 12-mmx28-gauge electroencephalographic (EEG) needles (Grass Telefac- tor, West Warwick, RI) were mounted in balsam wood 1 cm apart and with 0.5 cm of the needle exposed. The fish were placed on their right sides on a nonconduc- tive board. The detector electrode of the anterior probe was inserted midway between the posterior edge of the operculum and the leading edge of the dorsal fin, and midway between the base of the dorsal fin and the lateral line. The signal electrode of the posterior probe was inserted at the leading edge of the adipose fin, and midway between the base of the adipose fin and the lateral line. Serial R, serial Xc, and the distance between the inside (detector) electrodes, were recorded for each fish. All BIA measures were initially calculated by using both their series and parallel forms. Results of statis- tical analyses indicated that the parallel forms were more highly correlated to the independent variables. For this reason, as well as the instrument manufac- turer’s recommendation that the parallel forms most Caldarone et al Nonlethal techniques for estimating responses of postsmolt Sa/mo salar to food availability 261 closely approximate the real electrical values of biologi- cal tissue (RJL Systems, http://www.rjlsystems.com/ docs/bia_info/principles/, accessed April 2008), only the parallel forms of the BIA measures are discussed and reported. Serial R and Xc values were transformed to their parallel equivalents with the following formulas: Rpar(Q) = R + Xc2/ R, (2) XcpJQ) = Xc + R2/ Xc . (3) Because R and Xc are dependent upon the distance the current must travel ( D , distance between the electrodes in cm), the BIA instrument manufacturer advises that when these variables are used in prediction equations, the effect of this distance must be accounted for (RJL Systems, http://www.rilsvstems.com/docs/bia info/prin- ciples/. accessed April 2008). We therefore also calcu- lated standardized Rpar and Xcpar values by dividing Kpar ™*Xcpar by D(RpJD,XcpJD). Conductor volumes were calculated by using the fol- lowing formulas: R„ar conductor volume = D2/R„„r , (4) pdi pdi Xc conductor volume = Z)2/Xc .. . (5) pdr r Capacitance (a measure of the electrical storage ca- pacity) and impedance (a measure of the opposition to the flow of electrical current) were calculated using the following formulas: capacitance (pF) = 1 x 1012/(2tt •50000»Xcpar), (6) impedance (Q) = sqrt((Rpar2) + ( Xcpar 2)) , (7) where 50,000 is the frequency applied by the BIA instru- ment in Hertz. In order to conform to values previously reported in the literature, phase angles were calculated with Xc and R in their series form: phase angle (°) = arctan (Xc //? ) • 180/zr, (8) where Xc and R are the vertical and horizontal axis, respectively. The arctangent of the ratio will yield the angle of the impedance vector in radians, which is then converted to degrees by multiplying by 180/7T. A series- based phase angle is equal to 90° minus the parallel- based phase angle. Impedance measurements are negatively related to temperature (van Marken Lichtenbelt, 2001). Water temperature, room temperature, and internal fish tem- peratures (mean muscle temperature=13.4°C, SD = 0.9; mean stomach temperature = 12.6°C, SD = 1.0) were con- stant in our study and therefore not included as fac- tors in our analyses. Unless otherwise noted, the term “BIA measures” refers collectively to i?par, Xcpar, Rp.JD , Xcpar/D, phase angle, /?par and XcpaT conductor volumes , capacitance, and impedance. Fulton's K Fulton’s condition factor (K) was calculated with the following formula (Ricker, 1975): K = 100 • WW/FL3 , (9) where WW (in g) and FL (in cm) are values from the day the fish was sacrificed. Data analysis Within treatments A Dunnett two-tailed /-test with final FL as a covariate was used to detect changes in body composition (% WW), BIA measures, and K, within the fed and fasted treatments and the refed group. Baseline values (day 0) were specified as the control for all variables except BIA measures, where day-3 fed values were the specified controls. Between treatments A two-way multivariate analysis of covariance (MANCOVA) for unbalanced design was used to compare body composition (%WW) (total fat concentration, TF%; total water concentration, TWa%\ carcass protein concentration, CP%), BIA measures, growth rate, and K between the three feeding treatments (fed; fasted; fasted, then refed) and sampling times, with final FL as the covariate. When interactions were significant, feeding treat- ment was nested in day and follow-up comparisons were examined by using Tukey’s HSD multiple range tests. Because we found no significant differences in percent liver fat between any of the treatments or days, and liver fats comprised <1% of total fats (range: 0.39-0.99%), only total fat values were used in all analyses. Prediction models Prediction models for body compo- sition expressed as both content (total fat, TF\ TWci; carcass protein, CP), and concentration ( CP% , TF%, TWa%), and growth rate were developed. We used an information-theoretic approach for small sample sizes (Akaike’s information criterion, AICc) to select the “best-fit” models (Wagenmakers and Farrell, 2004). Because we had no prior knowledge of the variables or combination of variables that would be the best predictors of the dependent variables, all nine BIA measures plus the interaction of Rpar, and Xc with D were tested, along with WW, FL, and K. Testing 14 independent variables generated a large number of models for each dependent variable, with many models significant at the PcO.0001 level. For brevity, only the top three most parsimonious models (as indicated by the smallest AICc values) are reported and discussed. Correlations Pearson product-moment correlations were used to investigate the relations between body compo- sition (both g and %WW), and WW, BIA measures, K, and growth rate. All statistical analyses were carried out with SAS software vers. 9.1 (SAS Inst., Inc., Cary, 262 Fishery Bulletin 1 10(2) Table 2 Proximate body composition (% whole-body wet weight), capacitance, and % change in wet weight of Atlantic salmon ( Salmo sala r) postsmolts reared at 12°C under 3 feeding regimens (fed; fasted; fasted then refed). The refed group was fasted for 11 days and then fed. TF%= total fat concentration; TWa%= total water concentration; CP%=carcass protein concentration. Capacitance (pF) is a measure of the electrical storage capacity of cells. Percent change in wet weight was calculated by subtracting the initial wet weight of a fish on day 0 (fed and fasted treatments) or day 11 (fasted then refed treatment, day they were fed) from its wet weight on the day it was sacrificed, expressed as a percent of its initial wet weight. Asterisks (*) in the body composition columns indicate a significant difference from baseline values, and in the capacitance column they indicate a significant difference from day 3 fed values (Dunnett two-tailed t-test, P<0.05). Values are means with standard deviation (SD) in parentheses. Mean initial wet weight of all fish was 76.0 g (SD = 12.6, [no. of fish sampled=74|). NA=not available. n=number of fish sampled. Feeding regimen Sampling day n TF% TWa% CP% Capacitance % change in wet weight Baseline 0 5 7.1 (1.1) 73.2 (1.2) 17.6(0.5) NA 0 Fed 3 4 7.4 (1.8) 73.9(1.5) 17.0 (0.4) 2275 (322) NA 7 4 6.3 (1.4) 74.1 (1.3) 17.4 (0.7) 2057 (66) 1.2 (1.9) 11 4 5.9 (0.9) 74.9(1.3) 16.9(0.7) 2012 (333) 9.0 (2.3) 15 4 5.5 (0.8) 74.9 (1.0) 17.4 (0.4) 2144 (150) 15.5 (6.5) 19 4 6.2 (1.1) 73.8(1.3) 17.8(0.3) 2024 (113) 25.9 (5.0) 23 4 6.7 (2.0) 73.9(2.6) 17.6(0.7) 2024 (126) 20.7 (7.2) Fasted 3 4 7.9 (1.2) 72.6 (1.1) 17.0 (1.0) 2112 (268) NA 7 4 6.4 (1.3) 74.8 (1.6) 16.6 (0.4) 2042 (156) -2.7 (0.8) 11 4 6.0 (1.0) 74.5(1.2) 16.9(0.2) 1861 (180)* -4.2 (0.9) 15 4 5.2 (0.6)* 75.9 (1.1)* 17.0 (1.0) 1919 (152)* -4.6 (1.4) 19 3 5.3 (1.6) 75.8 (1.6)* 17.3 (0.2) 1988 (175) -7.3 (2.5) 23 4 4.9 (1.1)* 76.5(1.5)* 16.2 (1.1) 1744 (297)* -7.3 (2.5) Fasted, then refed 15 5 6.5 (2.1) 74.8 (2.8) 17.3 (0.9) 2005 (228) NA 19 5 5.7 (0.8) 75.7 (1.2) 16.8 (0.7) 1876 (151) 4.0 (4.5) 23 5 5.8 (0.8) 74.7 (0.9) 17.4 (0.7) 1978 (136) 17.5 (2.7) 27 7 5.8 (0.8) 75.1 (0.7) 17.2 (0.5) 1953 (74) 15.5 (5.9) NC). Unless otherwise stated, the level of significance was set at P<0.05. Results General observations At the start of the experiment we observed frayed or eroded dorsal fins in -78% of the fish, and fraying of the pelvic fin in -12% of the fish. During the experi- ment there was no change in the frequency of these abnormalities and there were no mortalities; however, one severely emaciated, moribund fish was excluded from the data set. None of the fish had well developed gonads and therefore all were considered to be imma- ture. Because sex did not emerge as a significant factor in any of the statistical analyses, it was not included in the data set. Within two weeks after their transfer to seawater, the postsmolts were actively feeding and appeared to be acclimated to the seawater and their surroundings. On day 0, wet weight ranged from 43 to 132 g, and fork length from 18 to 23 cm; size distributions of fish were not significantly different between feeding treatments (fed mean=76 g, SD = 12; fasted mean = 75 g, SD = 13; fasted then refed mean = 80 g, SD = 4). On day 0 we encountered instability problems with our handmade probes, which were resolved by the end of the day. Because we were not completely confident in our BIA measurements that day, we have not included these initial BIA values in any of our analyses. Cur- rently, BIA needle-electrodes are not manufactured and must be made in-house. Standardized manufactured probes would be necessary if BIA is to be routinely used. Within-treatment effects Over the course of the experiment, changes in fish weight and body composition (%WW) were observed within a treatment. On the final day of the experiment, fish in the fed treatment had increased in weight by 14-28%, whereas fasted fish had lost 5-10% of their weight (Table 2). Fish in the fasted, then refed treat- ment lost an average of 4% in weight during their 11 days without food, and gained 10-21% after 16 days of refeeding. Within each of the 3 feeding treatments, CP% (Table 2) and CA% (not shown) remained fairly constant. Changes in TF% (decreasing) and TWa% (increasing) occurred in fasted fish only, and differences from base- line values became significant after 15 d of fasting (Table 2). Within a treatment and sampling day, TF% was the most variable of the body components, averaging an Caldarone et al : Nonlethai techniques for estimating responses of postsmolt Salmo solar to food availability 263 18% coefficient of variation (CV) compared to 3.7% for CP% and 1.9% for TWa%. In continually fasted fish only, changes in two BIA measures from day-3 fed values were observed: capacitance significantly decreased beginning on day 11 (Table 2) and impedance increased on day 23 only (not shown). Between- treatment effects Results of the MANCOVA indicated that of the vari- ables tested (body composition [%-WW], BIA measures, growth rate, K), only growth rate (PcO.OOOl, df=13) and K (P=0.0009, df= 16) revealed significant differences due to feeding treatment. Results of Tukey’s HSD multiple range tests indicated that beginning on day 11 and con- tinuing until the end of the experiment, growth rates (negative) of the continually fasted fish were statisti- cally significantly slower than the fed treatment (Fig. 1A). Eight days after refeeding (day 19), growth rates of the refed group were significantly faster than those of the continually fasted treatment. On day 19 the fed treatment also had faster growth rates than those of the refed group, whereas the relation was reversed on day 23. On day 23, K values were significantly smaller in the continually fasted fish than those of the fed fish (Fig. IB). There was a trend of higher mean phase angle values in the fed treatment than those in the fasted treatment (Fig. 1C), but the values between the feeding treatments were not statistically different owing to high variability within a day’s sample. Prediction models Body composition (g), /? and Xcpar measured in the postsmolts encompassed a range of values (Table 3). The best predictor models for body composition (g) all contained fff and FL as independent variables, with some models also including BIA measures or K (Table 4). In these models, K can be viewed as an interaction term between fW and FL (i.e., a size-related variable). The models for TWa and CP had high predictive capabili- ties (coefficient of determination (r2)>0. 98), whereas the model for TF was less so (r2 range: 0.74-0.76). Adding any of the BIA measures to models containing only size or size-related independent variables (size-based-only models) increased the explanatory capabilities by <1.5%. The best predictor models for body composition (%WW) also contained size-related variables (WW or FL, and often K) (Table 4). The models for CP% ad- ditionally contained two or three BIA measures. Add- ing the BIA measures to a size-based-only CP% model increased the explanatory capabilities by <3.4%. The best predictor models for growth rate included size-related variables plus two BIA measures (Table 4). Adding the two BIA measures to size-based-only models increased the predictive capability of the growth equa- tions by 18-22%. Models with AAICc values <2 are considered to be equally probable to the “best fit” model (Burnham and Anderson, 2002). Based on the AAICc and r2 values, (A) Instantaneous wet-weight-based growth rate (per d); (B) Fulton’s condition factor (K= 100»final wet weight/ fork length3); and (C) phase angle (arctangent reactance/ resistance converted to degrees) of laboratory-reared Atlantic salmon (Salmo salar) postsmolts measured over 27 days. Values are mean (±standard deviation [ SD I ) for each sampling day. Fish were either fed ad libitum , fasted, or fasted until day 11 and then fed for the remainder of the experiment (refed). Within a sampling day, food treatments sharing a common superscript or without superscripts do not differ significantly (Tukey’s HSD multiple range tests). For fed and fasted fish n = 4 for each sampling day. For refed fish n = 22 for day 11, n = 5 for days 15, 19, and 23, and n = 7 for day 27. 264 Fishery Bulletin 1 10(2) little difference was evident between the top three mod- els for each dependent variable measured. Statistical significance of all models was high (P<0.0001); however, the predictive capability of the body composition content (g) models (74-99%) was much higher than the con- centration (%WW) and growth rate models (all <50%), with the TF% models having the lowest explanatory capability (<33%). Correlations Body composition (g) values were most highly correlated with WW (Table 5), although CP and TWa were also well correlated with Xcpar conductor volume (Table 5). CP and TWa were highly positively correlated with each other, and TF and TWa were less so; thus the majority of the changes in WW would be from the relation of water to protein, not water to fat. The ratio of water to protein averaged 4.4 (SD = 0.2), which is very similar to the value reported by Breck (2008) from other studies. Variables most highly correlated with TF% (posi- tively) and TWa% (negatively) were WW and K (Table 5), whereas CP% and growth rate were most correlated with WW. TF%, TWa%, and CP% had no or low corre- lation with growth rate, indicating that the fish were growing isometrically and not storing fat (Table 5). Overall, correlations of body composition (%WW) were much lower than correlations of body composition (g) with WW, BIA measures, and K. Discussion Treatment effects Feeding treatment (fed; fasted; fasted then refed) clearly impacted weight-based growth rates of post- smolt salmon. Growth rates of fed fish (mean=0. 0078/d; range: -0.0018 to 0.0139/d) were significantly faster than those of fasted fish (mean = -0. 0036/d; range: -0.0020 to -0.0053/d) and fell within ranges reported for other laboratory studies using similarly aged Atlantic salmon postsmolts (Handeland et ah, 2000; Jobling et al., 2002; Bendiksen et al., 2003; Sissener et al., 2009). Growth rates of fish responded quickly to the absence or reintroduction of food; decreasing after seven days of fasting, and increasing eight days after refeeding. Wilkinson et al. (2006) reported a similar response time for growth rates of Atlantic salmon smolts after 15 days of fasting (their first sampling date) and seven days of refeeding. Somatic growth rate in immature fish is an important index of condition because faster growers are considered to have a higher probability of survival (Lundvall et al., 1999; Craig et al., 2006; Fonseca and Cabral, 2007). Rapid growth rate, which results in larger individuals, is also thought to be critical for over-winter survival (Beamish and Mahnken, 2001). The widespread importance of rapid growth rates during early life his- tory stages is demonstrated in a meta-analysis of 40 fish studies (Perez and Munch, 2010) where 77% of estimated Table 3 Mean, standard deviation (SD) in parentheses, and range of body composition (g) and bioelectrical impedance anal- ysis measures (BIA; parallel resistance [i?par] and reac- tance [Xcpar] ) of postsmolt salmon (Salmo salar) reared at 12°C and three feeding regimens in order to obtain a range of nutritional condition. Fish weight ranged from 43 to 132g and length from 18 to 23 cm (no. of fish sam- pled=74). Variable Mean Range Total water content (TWa) (g) 60.0 (11) CO CO -92.4 Carcass protein content (CP) (g) 13.7 (3.1) 6.4- -24.3 Total fat content (TF) (g) 5.1 (1.8) 1.8- -12.7 Carcass ash content (CA) (g) 1.9 (0.4) 1.0- -3.0 Rpar(tl) 408 (24) 340- -480 Xcpar<£>) 1613 (178) 1204- -2437 selection differentials favored a larger fish size during this stage. During our study there was little impact of feeding treatment on the body composition (%WW) of postsmolt salmon. The reason for this relative constancy among feeding treatments was two-fold: 1) isometric growth in fed fish resulted in stable body composition (%WW) throughout a range of growth rates; and 2) there was little loss of proximate body constituents in fasting fish. When excess energy in young fish is directed primar- ily toward isometric growth, differences in growth rate cannot be discerned from body composition (%WW). For this reason, indices based on body composition (%WW) may not be the best metrics for assigning nutritional status and condition in immature fish. Generally, fish are well adapted to fluctuations in the food supply — a scenario they encounter often in the wild (Jobling, 2001). Atlantic salmon and other fish can respond to instances of food deprivation by reducing oxygen consumption, resulting in a decreased rate of fat and protein catabolism (Beamish, 1964; Metcalfe, 1998; O’Connor et al., 2000). Our results indicate that fasted postsmolts used only a small por- tion of fat for their metabolic needs during the 3-week experiment. The combination of isometric growth in the fed fish, and short-term food deprivation period in the fasting fish, resulted in nonsignificant differ- ences in body composition (%WW) among the feeding treatments. The utility of BIA as a condition index is based on differences in proportions of body constituents trans- lating into measurable differences in impedance when an electric current is applied. Because body composi- tion (%WW) of the postsmolts did not differ among the feeding treatments, it was not surprising that none of the nine BIA measures differed among the feeding treatments. Within the fasted treatment, capacitance values did decrease with increasing time fasted, pos- sibly reflecting the decline in fat concentration in this Caldarone et al Nonlethal techniques for estimating responses of postsmolt Sa/mo sa/ar to food availability 265 Table 4 Coefficients and Akaike’s second-order information criterion for small sample sizes (AICc) for the top 3 most parsimonious regres- sion models for proximate body composition (expressed as g and % wet weight) and growth rate of postsmolt Atlantic salmon ( Salmo salar) reared at 12°C under 3 feeding regimens in order to obtain a range of nutritional condition and growth rates. tyW=wet weight (g); FL = fork length (cm); i?par=resistance in parallel (Q); Acpar=reactance in parallel (12); Z) = distance between bioelectric impedance detector electrodes (cm); Rpar conductor volume =D2 /Rpar\ F=Fulton’s K (100 •WW/FL3)', capacitance (pF) is a measure of the electrical storage capacity of cells; impedance (Q) is a measure of the opposition to the flow of electrical current; RpaJD=Rpar standardized for D\ phase angle (°)=arctangent of Xc/R converted to degrees; AAICc= difference in AICc values with respect to the most parsimonious model. For all models PcO.OQOl. r2 = coefficient of determination. Dependent variable n Model r 2 AICc AAICc Total fat content (TF)( g) 60 0.733 + 0.172HVW) - 0.696(FL) + 0.090 (RpJD) 0.755 -6.72 0 8.326 + 0.150( VFIV) - 0.755(FL) 6.573 + 0.173GVW) - 0.648(FL) - 0.742 -5.71 1.0 54.90 )Rpar conductor volume) 0.750 -5.49 1.2 Carcass protein 65 content (CP)(g) 13.794 + 0.260(WWO - 0.825 (FL) - 4.453(F) 28.269 + Q.285(WW) — 1.131(FL) - 0.002 (capacitance) - 0.985 -119.18 0 0.002 (impedance) - 6.62(F4 0.986 -117.97 1.2 2.261+0.2151 WW) - 0.234 (FL) - 0.0005(capacficmce) 0.985 -117.16 2.0 Total water 67 -12.187 + 0.616( WW) + 1.109(FL) 0.993 -3.20 0 content (TWa Mg) -30.19 + 0.54KWW) + 2.002(FL) + 6.210(F) 0.993 -2.93 0.3 -7.791 + 603(WW) + 1.087(FL) - 0.055 (RpJD) 0.993 -2.35 0.8 Total fat concentration 60 15.272 + 0.099( WW) - 0.842 (FL) 0.326 9.81 0 ( TF% ) -1.568 + 0.28 (WW) + 5.70(F) 0.324 10.02 0.2 Carcass protein 65 8.086 + 0.330IFL) + 7.910(F) 32.185 + 0.044( WW) - 0.218 (FL) - 0.318 10.55 0.7 concentration (CP%) 0.004 (capacitance) - 0.004 [impedance) 27.318 + 0.026(WW) - 0.003 (capacitance) - 0.503 -83.17 0 0.004 (impedance) 28.090 + 0.036( WW) + 0.041 (Rpl,r/D) - 0.483 -82.96 0.2 0.004 (capacitance) - 0.005 {impedance) 0.498 -82.52 0.6 Total water 67 86.901 - 0.044( WW) - 9.180(F) 0.499 20.00 0 concentration ( TWa% ) 97.183 - 0.522(FL) - 12.553(F) 0.496 20.35 0.3 Instantaneous wet -weight 56 based growth rate 60.435 - 0.154IWW) - 1.310(FL) -0.6866 - 0.0016( WW ) + 0.0219(FL) + 0 . 3882 ( R par conductor volume) + 0.485 21.86 1.9 (per day) 0.0025 (RpJD) + 0.1931(F) -0.1087 + 0.0033(FL)-0.00003(capacfiance) 0.481 -582.29 0 + 0.00374 (phase angle) + 0.0560(F) -0.2156 + 0.0034IFL) + 0.00003(impecfance) + 0.453 -581.85 0.4 0.0031 (phase angle) + 0.0578(F) 0.444 -580.98 1.3 group. However, when the whole data set was exam- ined, capacitance and fat concentration (range: 4-10%) were only somewhat correlated (coefficient of correlation [r] = 0.41, P<0.005). Isometric growth is assumed for Fulton’s K, and dif- ferences in the weight-length relation are interpreted as an indication of stored energy. Because the posts- molts were growing isometrically with little energy storage, Fulton’s K was unable to distinguish between fast and slow growers within the fed treatment, and K values of fed fish were significantly higher than those of fasted fish only on the final day of sampling (day 23). Generally, Fulton’s K tends to have a long tempo- ral response (weeks to months) (Busacker et al., 1990). The relations among a fish’s wet weight, water weight, protein weight, and fat weight may explain this lag time. The wet weight of a fish is highly related to wa- ter weight (as was observed by Sutton et al. [2000] in Atlantic salmon parr), and water weight is much more strongly associated with protein weight than fat weight (20-40x more) (Breck, 2008). Therefore during the early stages of fasting, when fat stores are used first (Shulman and Love 1999, Jobling 2001), changes in a fish’s wet weight may be fairly subtle, but once a fish be- gins to use protein for energy, water loss (and thus wet weight loss) would accelerate. In our study, mean fat concentration in fasted fish decreased slightly with time while mean protein concentration remained constant. Within the fasted treatment there was a decreasing trend in mean K values, but owing to high variability 266 Fishery Bulletin 110(2) Table 5 Pearson product-moment correlations (r) between (A) proximate body composition (g) or (B) wet-weight-based proximate body composition or instantaneous wet-weight-based growth rate of Atlantic salmon ( Salma salar) post-smolts, and wet weight, bioelec- trical impedance analysis (BIA) measures and Fulton’s K. i?par=resistance in parallel; Xcpar= reactance in parallel; D = distance between BIA detector electrodes; Rpar and Xcpar conductor volume =D2 /Rpgr and D2 lXcpar, respectively; phase angle (“^arctan- gent of Xc/R converted to degrees; capacitance is a measure of the electrical storage capacity of cells; impedance is a measure of the opposition to the flow of electrical current; Fulton’s A=(100»wet weight/fork length3). Boldface type highlights highest correlation between body composition or growth rate, and wet weight, BIA measures, and Fulton’s K. * P<0.050, ** P<0.005, *** P0.95) and total water and protein content in their fish. The relation of total fat to conductor volume was equally high in the brook trout (r2=0.96; Cox and Caldarone et al.: Nonlethal techniques for estimating responses of postsmolt Sa/mo salar to food availability 267 Hartman, 2005) but was not significant in the steelhead (r2 = 0.02; Hanson et al., 2010). Pothoven et al. (2008) observed significant relations between conductor volume and total fat or dry mass in yellow perch ( Perea flaves- cens), walleye ( Sander vitreus), and lake whitefish ( Core - gonus clupeaformis), with r2 values ranging from 0.62 to 0.93. In our study, we also found conductor volume to be highly correlated with carcass protein content and total water (r=0.93 and 0.95, respectively) but more weakly correlated with total fat (r=0.74). Pothoven et al. (2008) suggested that significant correlation between published conductor volumes and body composition (g) is most likely due to biases imposed by the distance between the electrodes. The consistent placement of BIA electrodes on each fish results in the numerator of the conductor volume equation becoming a proxy for the size of the fish, and fish size is highly correlated with body contents (i.e., the larger the animal, the greater its total fat, water, protein, and ash content). In both our study (protein, fat, water) and Pothoven et al. (2008; yellow perch, fat and dry weight; walleye, fat), wet weight had a stron- ger relation to body composition (g) than did conductor volume. Neither Cox and Hartman (2005) nor Hanson et al. (2010) reported correlation results between wet weight and body composition (g). It would be valuable to know whether wet weight could estimate total fat, protein, and water content equally weil or better than BIA conductor volume in their studies as well. Prediction modeis In our study, the most parsimonious models for predict- ing body composition (g) all contained wet weight and fork length, and frequently a weight-length interaction term (Fulton’s K). Adding any combination of the nine BIA measures to size-based-only models increased the explanatory capabilities by less than 2%. Bosworth and Wolters (2001) estimated carcass fat content in channel catfish ( Ictalurus punctatus) using wet weight, R, and Xc as the predictor variables, resulting in an r2 = 0.75. They determined that adding the BIA measures to a model containing wet weight only increased the pre- dictive capability by 71%; however, R and Xc in their model had not been corrected for the distance between the electrodes. Because these impedance measurements are highly dependent upon the distance the current must travel between the electrodes, they must be stan- dardized to distance when used in prediction equations (RJL Systems, http://www.rjlsystems.com/docs/bia_info/ principles/, accessed April 2008) or they simply become proxies for length. Therefore the Bosworth and Wolters (2001) carcass fat content model essentially contains size-related only variables — and size is highly correlated with content values. When body constituents are expressed as concentra- tions (e.g., percent wet weight), the values are less size- dependent. Pothoven et al. (2008) examined the capabil- ity of 4 models (3 of which contained BIA measures) to predict percent total fat in 3 fish species. Interestingly, all of the variables used in their models were size re- lated (body mass, total length, conductor volume, Rpar and Xc|)ar uncorrected for distance between electrodes). In their study, fat concentration within a species encom- passed a range of values (yellow perch: 2. 7-8. 7%; wall- eye: 6.0-18.2%; lake whitefish: 2.4-14.7%), but overall predictive capability of the most parsimonious model for each species was low (r2 range: 0.17-0.53). In our study, all of the body composition (%WW) mod- els also had low predictive capabilities, with less than 50% of the variability explained by any of the top 3 models. Our most parsimonious models for TF% and TWa% contained only size variables (Table 4), whereas the CP% and growth rate models contained both size and either 2 or 3 BIA measures (Table 4). In the CP% models there was little added value of BIA measures to models containing only size variables (explanatory capabilities increased by <3.4%), and in the growth-rate models, adding two BIA measures did increase the r2 of size-based-only models by -20%. However, the overall predictive capability of the growth-rate models was still less than 50%. In a recent study by Hartman et al. (2011), highly predictive models for estimating percent dry weight in coastal bluefish (Pomatomus saltatrix) were constructed by using BIA measures. The most highly predictive models for their fish (15°C, r2=0.86; 27°C, r2 = 0.91) con- tained either phase angle or capacitance, plus RID and Xc/D. Ideally, standardizing R and Xc by the distance between the electrodes (D) would eliminate the effect of size on the impedance values; however, in our study we determined that even after standardization, RpaJD and Xcpar/D were still highly correlated with size (wet weight, r>0.91, Table 5B). Correlations between size (wet weight) and candidate BIA predictor variables were not reported in the coastal bluefish study, and size (wet weight or length) was not tested as a predictor variable in any of the models. In our study, size continually emerged as a significant variable in all of the body composition models. To our knowledge, no experiment has been conducted which controls the effect of size well enough to determine the actual contribution of BIA measures to the estimation of body composition independent of size. Until this is better understood, it will remain unclear to what extent BIA measures can improve upon size-only estimates of body composition. Phase angle The use of BIA-estimated phase angle as a measure of fish condition has been proposed by Cox and Heintz (2009). They concluded that in the 5 species they stud- ied, fish with phase angles >15° were in better condition than those with phase angles <15°. Caution should be used before universally applying this 15°cut-off value. In their own study, Cox and Heintz (2009) observed 12° phase angles in field-caught Pacific herring ( Clupea pallasii) with mass-specific energy contents of 7.15 kJ/g, and 15° phase angles in Pacific herring caught four months later with energy contents of 5.02 kJ/g. They also 268 Fishery Bulletin 1 10(2) reported phase angles >15° in large juvenile rainbow trout which had experienced a 9% loss of body weight after four weeks of food deprivation (Cox and Heintz, 2009). In our study, we found a low correlation (r= 0.37) between growth rate and phase angle. On the basis of our results and those of Cox and Heintz (2009), we feel it is unlikely that a single phase angle value can cor- rectly distinguish good from poor condition fish in all applications, and that the cut-off value will be influenced by factors such as life-stage, age, reproductive status, species, and temperature. In the Cox and Heintz (2009) study, the time frame necessary to clearly differentiate fasted from fed labo- ratory fish by phase angle varied by species and size. Significant differences between feeding treatments were observed in small juvenile rainbow trout after 14 days of fasting, in large juvenile rainbow trout after 21 days, in juvenile brook trout after 28 days, and in juvenile Chinook salmon after >77 days. This large range of response times may reflect differences in fat reserves in the test animals. If fat reserves do affect response time, then more than just a decline in fat content would be necessary to elicit a decrease in phase angle. A de- crease in protein content (with its concomitant water loss) may also be necessary for a significant decline in phase angle. If this is true, then the response time of K to food deprivation may be similar to (or possibly better than) that of phase angle. On the final sampling day of our study, K values, not phase angle, were sig- nificantly different between the fasted and fed fish. It would be interesting to know what the response time of K values were in the Cox and Heintz (2009) food- deprivation study. Conclusions The use of BIA as a proximate body composition esti- mator or fish condition index is relatively new. As with any condition index, its validity must be established for the specific application in which it will be used. Field personnel are seeking a nonlethal index that can reflect the condition of Atlantic salmon postsmolts 2 to 3 weeks after the fish are released from the hatchery, when the majority of postsmolts are captured on targeted trawl surveys. We designed an experiment to evaluate the utility of BIA and Fulton’s condition factor as indices of condition during that time-frame. Results of our study indicated that 1) growth rates of postsmolts responded rapidly to the withholding and re-introduction of food; 2) fed postsmolts grew isometrically; and 3) 3 weeks of withholding food is not sufficient time to elicit signifi- cant declines in proximate body constituents in fasting postsmolts. This combination of isometric growth in fed fish and short-term starvation period in fasting fish resulted in nonsignificant differences in body composi- tion (%WW) among the feeding treatments (fed; fasted; fasted, then refed). The utility of BIA and Fulton’s K as condition indices depends upon detecting differences in proportions of body constituents. During our study, BIA measures were not significantly different among the 3 feeding treatments, and only on the final day of sam- pling was K in fasted fish significantly less than in fed fish. Our study has demonstrated that neither BIA nor Fulton’s K would be an appropriate choice of index 1) to reflect short-term changes (weeks rather than months) in postsmolt condition, or 2) to monitor fish condition during a life-stage where excess energy is primarily directed toward isometric growth rather than energy storage. We propose that a methodology that measures growth rate (directly or indirectly) would be a more suitable condition index for isometric growth life-stages. We will be reporting on the utility of two potential growth-rate indices in Atlantic salmon postsmolts in a later publication. Results from our study also supported conclusions by Pothoven et al. (2008) that simple measures of size were better predictors of body composition than BIA measures. Additionally, we observed that Fulton’s K responded more quickly to food deprivation than BIA measures, and that a single cut-off value for phase angle, as a distinction between good and poor condition fish, should be used with caution. BIA is an emerging technique in fishery biology, and as such, its application will require more research to identify its appropriate use. Acknowledgments The authors would like to thank M. K. Cox for help constructing the BIA electrodes and demonstrating the BIA technique, E. Baker, and K. Fredrick for assistance in aquarium setup and temperature control, M. Prezioso and J. St. Onge-Burns for help rearing the salmon, the University of Rhode Island Graduate School of Ocean- ography for the use of Blount Aquarium, and anony- mous reviewers whose suggestions greatly improved this article. Literature cited Anderson, R. G., and S. J. Gutreuter. 1983. Length, weight, and associated structural indi- ces. In Fisheries techniques (L. A. Nielsen and D. L. Johnson, eds.), p. 283-300. Am. Fish. Soc., Bethesda, MD. Barbosa-Silva, M. C. G., A. J. D. Barros, C. L. A. Post, D. L. Wait- zberg., and S. B. Heymsfield. 2003. Can bioelectrical impedance analysis identify mal- nutrition in preoperative nutrition assessment? Nutri- tion 19:422-426. Baumgartner, R. N., W. C. Chumlea, and A. F. Roche. 1988. Bioelectric impedance phase angle and body com- position. Am. J. Clin. Nutr. 48:16-23. Beamish, F. W. H. 1964. Influence of starvation on standard and routine oxygen consumption. Trans. Am. Fish. Soc. 93:103-107. Beamish, R. J., and D. Mahnken. 2001. A critical size and period hypothesis to explain natural regulation of salmon abundance and the link- Caldarone et at: Nonlethal techniques for estimating responses of postsmolt Salmo solar to food availability 269 age to climate and climate change. Prog. Oceanogr. 49:423-437. Bendiksen, E. A., A. M. Arnesen, and M. Jobling. 2003. Effects of dietary fatty acid profile and fat con- tent on smelting and seawater performance in Atlantic salmon ( Salmo salar L.). Aquaculture 225:149-163. Black, D., and R. M. Love. 1986. The sequential mobilisation and restoration of energy reserves in tissues of Atlantic cod during starva- tion and refeeding. J. Comp. Physiol. B. 156:469-479. Bosworth, B. G., and W. R. Wolters. 2001 Evaluation of bioelectric impedance to predict car- cass yield, carcass composition, and fillet composition in farm-raised catfish. J. World Aquacult. Soc. 32:72-78. Breck, J. E. 2008. Enhancing bioenergetics models to account for dynamic changes in fish body composition and energy density. Trans. Am. Fish. Soc. 137:340-356. Burnham, K. P., and D. R. Anderson. 2002. Model selection and multimodel inference: a practical information-theoretic approach, 2nd ed., 488 p. Springer Science + Business Media, LLC, New York. Busacker, G. P., I. R. Adelman, and E. M. Goolish. 1990. Growth. In Methods for fish biology (C. B. Schreck and P. B. Moyle, eds.), p. 363-387. Am. Fish. Soc., Bethesda, MD. Chambers, R. C., and T. J. Miller. 1995. Evaluating fish growth by means of otolith incre- ment analysis: special properties of individual-level lon- gitudinal data. In Recent developments in fish otolith research (D. Secor, J. Dean, and S. Campana, eds.), p. 155-174. Univ. South Carolina Press, Columbia, SC. Chumlea, W. C., S. S. Guo, R. J. Kuczmarksi, K. M. Glegal, C. L. Johnson, S. B Heymsfield., H. C. Lukaski, K. Friedl, and V. S. Hubbard. 2002. Body composition estimated from NHANES III bio- electrical impedance data. Int. J. Obesity 26:1596-1609. Cox, M. K., and K. J. Hartman. 2005. Nonlethal estimation of proximate composition in fish. Can. J. Fish. Aquat. Sci. 62:269-275. Cox, M. K., and R. Heintz. 2009. Electrical phase angle as a new method to measure fish condition. Fish. Bull. 107:477-487. Craig, J. K., B. J. Burke, L. B. Crowder, and J. A. Rice. 2006. Prey growth and size-dependent predation in juve- nile estuarine fishes: experimental and model analy- ses. Ecology 87:2366-2377. Federal Register. 2009. Endangered and threatened species: determination of endangered status for the Gulf of Maine Distinct Popu- lation Segment of Atlantic salmon; notice. Vol. 74, no. 117, June 19, p. 69459-69483. GPO, Washington, D.C. Folch, J., M. Lees, and G. H. Sloane-Stanley. 1957. A simple method for the isolation and purifica- tion of total lipids from animal tissues. J. Biol. Chem. 226:497-509. Fonseca, V. F., and H. N. Cabral. 2007. Are fish early growth and condition patterns related to life-history strategies? Rev. Fish Biol. Fish. 17:545-564. Handeland, S. O., A. Berge, B. Th. Bjornsson, 0. Lie, and S. O. Stefansson. 2000. Seawater adaptation by out-of-season Atlantic salmon ( Salmo salar L.) smolts at different tempera- tures. Aquaculture 181:377-396. Hanson, K. C., K. G. Ostrand, A. L. Gannam, and S. L. Ostrand. 2010. Comparison and validation of nonlethal techniques for estimating condition in juvenile salmon. Trans. Am. Fish. Soc. 139:1733-1741. Hartman, K. J., B. A. Phelan, and J. E. Rosendale. 2011. Temperature effects on bioelectrical impedance analysis (BIA) used to estimate dry weight as a condition proxy in coastal bluefish. Mar. Coast. Fish. 3:307- 316. Jobling, M. 2001. Nutrient partitioning and the influence of feed composition on body composition. In Food intake in fish (D. Houlihan, T. Boujard, and M. Jobling, eds.), p. 354-375. Blackwell Scientific, Oxford, U K. Jobling, M., B. Andreassen, A. V. Larsen, and R. L. Olsen. 2002. Fat dynamics of Atlantic salmon Salmo salar L. smolt during early seawater growth. Aquat. Res. 33:739-745. Jones, D. B. 1931. Factors for converting percentages of nitrogen in foods and feeds into percentages of proteins. U.S. Dep. Agr. Circ. 183(1). Love, R. M. 1970. The chemical biology of fishes, 547 p. Academic Press, Inc., New York, NY. Lukaski, H. D., P. E. Johnson, W. W. Bolonchuk, and F. I. Lykken. 1985. Assessment of fat-free mass using bioelectrical impedance measurements of the human body. Am. J. Clin. Nutr. 41:810-817. Lundvall, D, R. Svanback, L. Persson, and P. Bystrom. 1999. Size-dependent predation in piscivores: interac- tions between predator foraging and prey avoidance abilities. Can. J. Fish. Aquat. Sci. 56:1285-1292. MacLean, S. M., E. M. Caldarone, and J. M. St. Onge-Burns. 2008. Estimating recent growth rates of Atlantic salmon smolts using RNA-DNA ratios from nonlethally sampled tissues. Trans. Am. Fish. Soc. 137:1279-1284. Marchello, M. J., and W. D. Slanger. 1994. Bioelectrical impedance can predict skeletal muscle and fat-free skeletal muscle of beef cows and their car- casses. J. Anim. Sci. 72:3118-3123. Metcalfe, N. B. 1998. The interaction between behavior and physiol- ogy in determining life history patterns in Atlantic salmon ( Salmo salar). Can. J. Fish. Aquat. Sci. 55 (suppl. 1 ) : 93—103. O’Connor, K. I., A. C. Taylor, and N. B. Metcalfe. 2000. The stability of standard metabolic rate during a period of food deprivation in juvenile Atlantic salmon. J. Fish. Biol. 57:41-51. Perez, K. O., and S. B. Munch. 2010. Extreme selection on size in the early lives of fish. Evolution 64:2450-2457. Pothoven, S. A., S. A. Ludsin, T. O. Hook, D. L. Fanslow, D. M. Mason, P. D. Collingsworth, and J. J. Van Tassell. 2008. Reliability of bioelectrical impedance analysis for estimating whole-fish energy density and percent lipids. Trans. Am. Fish. Soc. 137:1519-1529. Ricker, W. E. 1975. Computation and interpretation of biological sta- tistics of fish populations. Bull. Fish. Res. Board Can. 191:1-382. 1979. Growth rates and models. In Fish physiology, vol. 8 (W. S. Hoar, D. J. Randall, and J. R. Brett, eds), p. 677-743. Academic Press, New York, NY. 270 Fishery Bulletin 110(2) Schwenk, A., A. Beisenherz, K. Romer, G. Kremer, B. Salzberger, and M. Elia. 2000. Phase angle from bioelectrical impedance analysis remains an independent predictive marker in HIV-infected patients in the era of highly active anti- retroviral treatment. Am. J. Clin. Nutr. 72:496-501. Shulman, G. E. and R. M. Love. 1999. The biochemical ecology of marine fishes. Adv. Mar. Biol. 36, 351 p. Sissener, N. H., M. Sanden, A. M. Bakke, A, Krogdahl, and G.-I. Hemre. 2009. A long term trial with Atlantic salmon ( Salmo salar L.) fed genetically modified soy; focusing gen- eral health and performance before, during and after the parr-smolt transformation. Aquaculture 294:108- 117. Stevenson, R. D., and W. A. Woods, Jr. 2006. Condition indices for conservation: new uses for evolving tools. Integr. Comp. Biol. 46:1169-1190. Sutton, S. G. , R. P. Bult, and R. L. Haedrich. 2000. Relationships among fat, weight, body weight, water weight, and condition factors in wild Atlantic salmon parr. Trans. Am. Fish. Soc. 129:527-538. van Marken Lichtenbelt, W. D. 2001. The use of bioelectrical impedance analysis (BIA) for estimation of body composition. In Body composi- tion analysis of animals: a handbook of non-destructive methods (J. R. Speakman, ed.), p. 161-187. Cambridge Univ. Press, Cambridge, U.K. Wagenmakers, E. J., and S. Farrell. 2004. AIC model selection using Akaike weights. Psychon. Bull. Rev. 11:192-196. Wilkinson, R. J., M. Porter, H. Woolcott, R. Longland, and J. F. Carragher. 2006. Effects of aquaculture related stressors and nutritional restriction on circulating growth factors (GH, IGF-I and IGF-II) in Atlantic salmon and rainbow trout. Comp. Biochem. Phys. A 145:214-22. 271 Estimation of discard mortality of sablefish ( Anopiopoma fimbria) in Alaska longline fisheries Email address for contact author: chns.lunsford tat noaa.gov 1 Auke Bay Laboratories Ted Stevens Marine Research Institute Alaska Fisheries Science Center National Marine Fisheries Service National Oceanic and Atmospheric Administration 17109 Pt Lena Loop Rd. Juneau, Alaska 99801 2 Present address: School of Aquatic and Fishery Sciences University of Washington Box 355020 Seattle, Washington 98195 Abstract — -Sablefish ( Anopiopoma fimbria) are often caught inciden- tally in longline fisheries and dis- carded, but the extent of mortality after release is unknown, which creates uncertainty for estimates of total mortality. We analyzed data from 10,427 fish that were tagged in research surveys and recovered in surveys and commercial fisheries up to 19 years later and found a decrease in recapture rates for fish originally captured at shallower depths (210— 319 m) during the study, sustaining severe hooking injuries, and sustain- ing amphipod predation injuries. The overall estimated discard mortality rate was 11.71%. This estimate is based on an assumed survival rate of 96.5% for fish with minor hooking injuries and the observed recapture rates for sablefish at each level of severity of hook injury. This esti- mate may be lower than what actu- ally occurs in commercial fisheries because fish are likely not handled as carefully as those in our study. Com- paring our results with data on the relative occurrence of the severity of hooking injuries in longline fisheries may lead to more accurate account- ing of total mortality attributable to fishing and to improved management of this species. Manuscript submitted 4 October 2011. Manuscript accepted 22 February 2012. Fish. Bull. 110:271-279 (2012). The views and opinions expressed or implied in this article are those of the author (or authors) and do not necessarily reflect the position of the National Marine Fisheries Service, NOAA. Megan M. Stachura' 2 Chris R. Lunsford (contact author)1 Cara J. Rodgveller1 Jonathan Heifetz' For stock assessment, accurate ac- counting of discard mortality is impor- tant for estimating total mortality attributable to fishing. Studies of sablefish ( Anopiopoma fimbria) and other fish species show that catch- related injuries can cause delayed mortality after a fish is discarded. For example, sablefish laboratory experiments have shown that the level of physical injury, reflex impair- ment, and behavior impairment may be useful proxies for delayed mortal- ity (Davis, 2005; Davis and Ottmar, 2006). Pacific halibut ( Hippoglossus stenolepis) with more severe hook injuries had increased mortality and reduced growth compared to those with less severe injuries (Kaimmer, 1994; Kaimmer and Trumble, 1998) and were visually impaired after exposure to simulated sunlight (Brill et ah, 2008). For Atlantic cod ( Gadus morhua L), injuries to the eyes, gills, and belly were more lethal than injuries to other anatomical parts (Palsson et ah, 2003). After release, Atlantic cod had inhibited activity for 4 days, during which there was potentially increased susceptibility to predation and delayed mortality (Neat et ah, 2009). A first step in estimating discard mortality is to estimate the propor- tion of fish that die after being dis- carded. Estimates of sablefish discard mortality rates and the derivation methods for determining these esti- mates vary regionally and by man- agement agency. In the southeast Alaska sablefish stock assessment conducted by the Alaska Depart- ment of Fish and Game for state wa- ters, a 25% discard mortality rate in the Pacific halibut longline fishery is assumed for sablefish (Dressel1). For both trawl and longline federal groundfish fisheries in Alaska, 100% mortality is assumed for all sablefish that are discarded (Hanselman et al., 2010). In the federal Pacific Coast sablefish stock assessment a much lower discard mortality rate of 10% is assumed for longline gear (Schir- ripa, 2008). Sablefish support one of the most valuable fisheries in Alaska (Hiatt et ah, 2010). The fixed gear fishery in 1 Dressel, S. C. 2009. 2006 northern southeast inside sablefish stock assess- ment and 2007 forecast and quota. Fish- ery Data Series 09-50, 78 p. Alaska Dep. Fish Game, Anchorage, AK. 272 Fishery Bulletin 1 10(2) federal waters off Alaska is managed by a catch shares program, where annual individual fishing quota (IFQ) shares are allocated to fishermen, for fish that can be caught anytime during the eight and a half month sea- son. For fishermen with IFQs, full retention of all sable- fish caught is required. However, sablefish are often legally discarded in other commercial longline fisheries, primarily in those targeting Pacific halibut and Pacific cod (Gadus macrocephalus). In the sablefish fishery, the practice of releasing small sablefish and retaining only the larger fish because of the greater value per pound of larger fish (a technique known as “highgrading” [Davis, 2002] ) is illegal. However, because there is an incentive to retain larger fish and not all fishing trips are monitored, highgrading may occur. Factors affecting discard mortality likely vary by spe- cies, gear type, depth, and other environmental factors. Injury location on fish has proven to be an indicator of short and long-term discard mortality (e.g., Bar- tholomew and Bohnsack, 2005). In Alaska, sablefish inhabit a wide range of depths and are caught primar- ily on longline gear, which can cause external injuries to different areas of the body. Fish tethered to longline gear for extended periods are subject to predation by parasitic amphipod crustaceans. Also fish size may affect mortality of discarded fish. The objective of our study is to determine if the location and severity of the hook injury, line and roller gear injury, water depth, fish size, and the level of amphipod predation affect the discard mortality rate in Alaskan longline fisher- ies. To answer these questions, the recapture rates of fish tagged and released in the marine environment were related to each factor. In addition, an absolute discard mortality rate was computed on the basis of the observed severity of hooking injuries. Materials and methods Tagging and data collection In 1989 and 1990, research surveys were conducted by the National Marine Fisheries Service (NMFS), Alaska Fisheries Science Center (AFSC) in Southeast Alaska. In 1989, sablefish were tagged during August and September in Chatham Strait; in 1990, sablefish were tagged during April and May in Clarence Strait (Fig. 1). Longline gear was fished on the bottom at depths from 210 to 419 m with a minimum 3-hour soak time. Gear configuration consisted of size 13/0 circle hooks baited with squid attached to 38-cm gangions that were secured to beckets tied in a 9.5-mm (3/8 in) groundline at 2-m intervals. This gear configuration is similar to that used in the commercial sablefish fishery in Alaska. However, the Pacific halibut fishery typically uses larger hooks (16/0). All sablefish, except those with extremely severe injuries, were tagged with plastic T-bar style anchor tags, and injuries were classified by the following 4 variables: location of hook injury, sever- ity of hook injury, severity of injuries due to amphipod predation, and the presence of injury sustained on fins or body from line and roller gears. Within each variable, a categorical condition code describing the injury was recorded (Table 1). The date of capture, capture location, and depth of capture were also documented. Fish were promptly released after they were measured (fork length, nearest mm) and tagged. To determine recapture rates of fish within each cat- egory, tagged fish were recovered in commercial fisher- ies and tags were returned to the AFSC for a reward (Maloney2). Tags were also recovered during subsequent research studies. Data for fish recaptured from the time of tagging to June 2009 were used in our analysis (up to 19 years at liberty). Analysis A logistic regression model was constructed to determine which factors were related to significant differences in recapture rates. The relationship between the binary, dependent variable, Tj-, which represents whether a fish was recaptured or not, and seven independent explana- tory variables was estimated with the following full model, Logit(Yj) = a + bYr, + cL, + dD. + eHLi + ftiS. + gAt + hG(, (1) where a Yn L, Di HL. HSt A, G( the intercept, and b to h are estimated model coefficients; year of tagging (1989, 1990); fish length at capture; capture depth group (210-269, 270-319, 320-419 m); location of the hook injury (cheek, upper jaw, lower jaw, nose, throat, eye, gill); severity of the hook injury (minor, moderate, severe); severity of amphipod predation injury (no injury, <10% scale loss, >10% scale loss); and type of injury sustained on fins or body from line and roller gears (no injury, fin damage, lacerations) for fish i (Table 1). Year can also be considered to be the effect of location because in each year fish were tagged at different loca- tions. All independent variables were treated as categori- cal except for length, which was continuous. Interaction terms were not included in the model because of the small sample sizes available across multiple categori- cal variables, which resulted in an inability to estimate these interaction parameters. Forward-stepwise model selection was performed to simplify the model to factors that significantly improved 2 Maloney, N. E. 2002. Report to industry on the Alaska sablefish tag program, 1972-2001. AFSC Processed Rep. 2002-01, 44 p. Alike Bay Laboratory, NMFS, NOAA, 11305 Glacier Highway, Juneau, AK 99801. Stachura et al. : Estimation of discard mortality of Anoplopoma fimbria in Alaska longline fisheries 273 135°W 130°W Figure I Map of the areas in Southeast Alaska where sablefish (Anoplopoma fimbria) were tagged during research surveys in 1989 (•) and 1990 ( + ). model fit. The model with the minimum Akaike infor- mation criteria (AIC) value was chosen. A Wald chi- squared test was used to calculate the overall signifi- cance of categorical variables with multiple coefficients. All statistical analysis was implemented in R software, vers. 2.11.1 (R Development Core Team, 2010) including use of the aod package, vers. 1.2 (Lsenoff and Lancelot, 2010). Recapture rates for categories within each variable were calculated by dividing the number of recaptured fish by the number of tagged fish for each category. Absolute survival rates were calculated for each level of hook severity on the basis of observed recapture rates and the survival rate of a Pacific halibut with minor hooking injuries (Kaimmer, 1994; Kaimmer and Trum- ble, 1998; Trumble et. al., 2000). Previous studies have determined that the expected survival of a properly handled Pacific halibut is in the 95-98% range; a re- leased fish with minor injuries has an estimated 96.5% survival rate (Trumble et al., 2000). We used the Pacific halibut estimate of survival rate as a proxy for that of sablefish for the following reasons: these species do not experience barotrauma as a result of rapid decompres- sion; they co-occur in the same water temperatures, areas, and depths; they are caught with nearly identical gear types; and they are commonly fished by the same fishing vessels and crew. Like Pacific halibut, sablefish are hardy and, when handled appropriately, have high survival rates after capture and discard. Long-term tag- ging programs for both species provide evidence of their hardiness (Kaimmer, 2000; Maloney2). The hardiness of sablefish is also supported by previous research in a laboratory setting where there was 100% survival after 60 days (Davis et al., 2001). Ours is the first dedicated study to estimate sablefish discard mortality. Previous estimates of Pacific halibut survival rates are the best available data to use as a proxy for sablefish. The average survival rate of fish with different severi- ties of hook injury, i.e., the absolute survival rate, was estimated on the basis of recapture rates and relative frequency of all 3 levels of hook injury (minor, moderate, severe) by using the methods in Kaimmer and Trumble (1998). The overall absolute survival rate (S) of cap- tured fish was calculated with the following formula: S = ( R0 + R1+R2 ) {Tc + ^+T.t + NT j Ro To x 0.965, (2) where T0, Tv and T2 and R0, R ,, and R2 are the number of fish tagged ( T ) and recovered (R) with minor (0), mod- erate (1) and severe hook injuries (2). Fish that were not 274 Fishery Bulletin 1 10(2) Table 1 Description and assigned injury code for injury types and severities for sablefish ( Anoplopoma fimbria ) caught on longline gear, for an estimation of discard mortality. Factor Description Hook injury location 0 Hooked in cheek or parts of operculum 1 Hooked in upper jaw: maxilla or premaxilla 2 Hooked in lower jaw: dentary (mandible) 3 Hooked in nose or snout 4 Hooked in throat 5 Hooked in eye 6 Hooked around gill or gill arches Hook injury severity 0 Minor: small puncture, flesh not torn, no abrasion 1 Moderate: flesh torn; some abrasion; bones intact, eye orbit not punctured 2 Severe: bones torn at insertion, severed or shattered, gills hooked but no broken gill arches, hooked through palatine into nose capsule, cheek bones shattered, hooked in throat and bleeding but not torn NT No tag: gill arches torn or bleeding, hook swallowed with substantial tears in throat; maxillary and premaxillary or dentary torn off; nose or snout smashed Amphipod predation injury 0 No injury 1 Moderate scale loss: 10% or less 2 Heavy scale loss: greater than 10% Line and roller gear injury sustained on fins or body 0 No injury 1 Fin damage: caudal, pectoral, pelvic, dorsal, or anal fin 2 Lacerations: line markings across body tagged because of the extreme severity of their injuries were assumed to have 0% survival and are represented in the equation as NT (having no tag and they were included in the total number of fish caught when cal- culating the recovery rate for each injury group). The survival rate of fish in each category was calculated with the following formula: Sx = — x 0.965, (3) •TCq T 1 0 where all variables are the same as in Equation 2 and x represents the severity of the hook injury (0, 1, or 2). Results A large number of sablefish were captured (10,940) and tagged (10,508): 8838 fish were tagged during the 1989 survey and 1670 during the 1990 survey. A substantial number of fish were recaptured (1207 fish, 11.49% recapture rate of tagged fish) between 9 days and 19.2 years (mean = 3.4 yr, standard devia- tion = 4.5 yr) after tagging. Because some data were lacking for 81 fish, analyses were run with data from 10,427 fish. An additional 432 fish were captured but not tagged because of the extreme injuries from capture or amphipod predation (NT in Eq. 2, see Materials and methods section). Logistic regression model The reduced model was chosen on the basis of the small- est AIC value. Several parameters were found to sig- nificantly affect recapture rates: year (which also can be considered to be a location effect), depth, severity of hook injury, and amphipod predation (Table 2, Fig. 2). Fish tagged in 1989 had a lower recapture rate (11.26%) than those tagged in 1990 (12.66%) (Table 3). Fish from the greatest depths (320-419 m) had a greater rate of recapture (14.33%) than fish captured at shallower depths (210-269 m, 10.61%; 270-319 m, 10.43%; Table 3). Severity of hook injury also exhibited a significant effect on the recapture of tagged fish (Table 2). Fish with severe injuries had a lower recapture rate (8.49%) than those with minor (12.05%) or moderate (11.81%) injuries (Table 3). The confidence intervals surround- ing the parameters for severity of injury were relatively narrow, with the 95% confidence interval of the odds Stachura et al Estimation of discard mortality of Anoplopoma fimbria in Alaska longline fisheries 275 Year 1990 Depth: 270-319 m Depth: 320^119 m Severity of hook injury: moderate Severity of hook injury : severe Amphipod predation: s10% scale loss Amphipod predation: >10% scale loss Higher mortality Hig <■ Lower mortality 0.5 1.0 1.5 Effect on recapture rate 2.0 Figure 2 Comparison of the effects of the variables in the final model on the recapture rate of tagged sablefish ( Anoplopoma fimbria). The effect on recapture (circle) is the exponent of the estimated parameter for the variable in the logistic regression and is the odds ratio: the odds of recapture of a fish in a category compared to the odds of recapture in the initial category of the categorical variable (year: 1989; depth: 210-269 m; severity of hook injury: minor; amphipod predation: no predation). Horizontal lines are 95% confidence intervals for the estimate. ratio for the effect of severe hooking injuries <1, indicat- ing a significant negative effect on recapture (Fig. 2). Although only a small portion of the fish sampled suf- fered from amphipod predation, it significantly affected recapture and was included in the final mode! (Table 2). Fish with no observed amphipod predation had a higher rate of recapture (11.86%) than fish with <10% scale loss (8.44%) and fish with >10% scale loss (7.84%) owing to amphipod predation (Table 3). The 95% confidence inter- val of the odds ratio for the effect of less than or equal to 10% scale loss was less than 1, indicating a significant negative effect on recapture (Fig. 2). However, there was a high amount of variability around the estimated parameter for the effect of >10% scale loss because of a low number of samples (Fig. 2). The majority of fish (51.06%) that were too severely injured to be tagged had suffered from amphipod predation, and only 11.09% of fish that were healthy enough to be tagged had suffered amphipod predation. In our study location of hook injury, fish length, and type of gear injury did not significantly affect recap- ture rates. Hook injuries were not in critical locations that would likely cause mortality alone. Most injuries were located on the cheek and upper and lower jaws (95.53%). There were a small number of fish observed that had hook injuries to other areas of the body (nose, throat, eye, gill; 4.47%) (Table 3). A wide range of fish lengths were included in our study, but length did not Table 2 Significant effects included in the reduced logistic regres- sion model, where the response is whether a sablefish (Anoplopoma fimbria) was successfully recaptured after tagging. The overall variable significance was calculated by using a Wald chi-squared test. Variable t df P(>X2) Intercept 571.0 1 <0.001 Year 11.6 1 <0.001 Depth (m) 34.8 2 <0.001 Severity of hook injury 12.4 2 0.002 Amphipod predation 7.7 2 0.021 have a significant effect on recapture rate. Injuries caused by the line and roller gear also did not have a significant effect on recapture rate. This result may be the consequence of low statistical power because few fish (5.58%) sustained injuries caused by gear other than hooks (Table 3). Survival rates The absolute survival rate of fish in each category of severity of hook injury was calculated with Equation 3 276 Fishery Bulletin 1 10(2) Table 3 Number of sablefish (Anoplopoma fimbria) tagged and recaptured (number and %) by each variable. Length variable in the analysis but is categorized here for summary purposes. The estimated absolute survival was levels of the severity of hook injury on the basis of an assumed 96.5% survival of fish with minor injuries. was a continuous estimated for the Variable Tagged Recaptured Estimated % Recaptured absolute survival % Year 1989 8768 987 11.26 1990 1659 210 12.66 Length (cm) <60 1280 152 11.88 60-69 5250 585 11.14 70-79 3048 360 11.81 >80 849 100 11.78 Depth (m) 210-269 3354 356 10.61 270-319 4428 462 10.43 320-419 2645 379 14.33 Hook location Cheek 3290 396 12.04 Upper jaw 1759 199 11.31 Lower jaw 4912 559 11.38 Nose 123 14 11.38 Throat 212 17 8.02 Eye 120 11 9.17 Gill 11 1 9.09 Severity of hook injury Minor 2963 357 12.05 96.50 Moderate 6204 733 11.81 94.63 Severe 1260 107 8.49 68.01 Extreme Total 432 10,859 1197 11.02 0.0 88.29 Amphipod predation No predation 9271 1100 11.86 <10% scale loss 1054 89 8.44 >10% scale loss 102 8 7.84 Gear injury No injury 9845 1133 11.51 Fin damage 539 57 10.58 Lacerations 43 7 16.28 (Table 3). The overall absolute survival of released sablefish was estimated, with Equation 2, to be 88.29%, or an overall mortality rate of 11.71% (Table 3). The absolute survival of fish with severe injuries (68.01%) was much lower. Discussion Our results indicate that the severity of hook injury is related to recapture rates for tagged sablefish. Most inju- ries were to the cheek and jaw and not to critical areas, such as the gills and brain. The severe injuries that we saw likely resulted in delayed mortality following the tagging event which would explain lower recapture rates. The severity of an injury is likely influenced by the technique for hook removal. Previous studies with Pacific halibut (Kaimmer, 1994; Kaimmer and Trumble, 1998) found that the removal of the hook affects the severity of the hook injury and, as with sablefish, sur- vival decreased with an increase in the severity of hook injury. Severity of hook injury is a logical parameter for estimation of discard mortality because it significantly affects recapture rate. Stachura et al Estimation of discard mortality of Anoplopoma fimbria in Alaska longline fisheries 277 In our study, the location of hooking injury did not significantly affect recapture rates. However, a large portion of injuries occurred on the cheek and upper and lower jaws — locations that are typically affected by circle hooks. We likely did not have enough samples of fish with injuries in other locations to detect the effects of those injuries. Unlike our results, results from stud- ies of catch-and-release of recreationaliy caught species have indicated that hooking location was the most sig- nificant factor in estimating mortality (reviewed in Bar- tholomew and Bohnsack, 2005). Deep-hooking injuries in critical locations such as the esophagus, stomach, gills, eyes, and brain significantly increase mortality in many species (e.g., Muoneke and Childress, 1994; Palsson et al., 2003; Aalbers et al., 2004; Aids et al., 2009). The circle hooks that are used in Alaska longline fisheries usually hook fish in the mouth and injuries in critical locations are not common (reviewed in Trumble et al., 2000). Capture with other hook types or fishing gears, such as trawl gear, would likely produce injuries on other locations of the body. Depth of capture significantly affected the recapture rate of sablefish, which is common for other fish species. We found a positive relationship between depth of cap- ture and assumed survival (i.e., fish caught at shallow depths were less likely to be recaptured). The sablefish fishery extends to at least 800 m in many areas and so the effect of depth on recapture rates may be even more pronounced at depths greater than 419 m, the maximum sampling depth in our study. The opposite has been observed in physoclistous species due to baro- trauma, because of organ damage caused by gas expan- sion in the body cavity during capture (e.g., Gitschlag and Renaud, 1994; Wilson and Burns, 1996; Collins et al., 1999; St. John and Syers, 2005). Sablefish lack a swim bladder, thus no correlation between mortality and depth of capture was expected. Deeper-dwelling fish can also have increased injuries with greater cap- ture depths, indicating that injuries are inflicted while fish struggle during hauling (Atlantic cod; Palsson et al., 2003). There are some potential explanations for why fish caught at shallow depths had lower recapture rates. First, sablefish caught at deeper depths (320-419 m) could be less vigorous because of the longer retrieval time and the increased time spent fighting the line during retrieval and therefore they are less likely to become injured during the landing process when out of the water and onboard the fishing vessel. Differential predation in the depth categories may also affect the mortality of released sablefish, if they return to their previous depths after release. Two major predators of sablefish have greater concentration at shallower depths, Pacific halibut (27-274 m; IPHC, 1998) and Pacific sleeper sharks ( Somniosus pacificus) (Yano et al., 2007). Second, fishing effort likely differs by depth and therefore may affect depth-related recapture rates of tagged fish. Data on fishing effort by depth were not available from the Pacific halibut fishery or the southeast Alaska state sablefish fishery and there- fore a full examination of this supposition was not possible. In our study, amphipod predation was related to the recapture rate of sablefish and was prevalent for fish that were too severely injured to tag. Similarly, Pacific halibut that were tethered to longlines for extended periods suffered from amphipod predation and had a low survival rate (Trumble et al., 2000). Fishery-specific amphipod predation rates would need to be investigated to accurately assess this effect on the discard mortality of sablefish. The year of capture significantly affected the re- capture rate of sablefish. A greater recapture rate was found for fish tagged in Clarence Strait in 1990 and several factors likely contributed to this differ- ence. First, a greater proportion of fish tagged in 1990 (18.57%) were recaptured within 60 days of tagging compared to those tagged in 1989 (7.42%). This is likely explained by the occurrence of an Alaska De- partment of Fish and Game sablefish survey and the southern southeast Alaska directed sablefish fishery both occurring within 60 days of the initial tagging effort. Tagging conducted in Chatham Strait in 1989 occurred after both the state survey and fishery period and therefore the grounds were not fished for nearly a year after the tagging effort. A minimum time at liber- ty was not used in our study because the year or loca- tion of tagging was secondary to our primary objective of determining the factors related to discard mortality and estimating absolute discard mortality based on the severity of injuries to sablefish. Second, longline fishing is permitted in the Chatham Strait fishery, and in Clarence Strait both longline and pot gear are allowed. Animals can exhibit varying levels of “trap addiction” (attraction to fishing gear) or “trap shyness” (an aversion to the gear) depending on the gear type (Seber, 1982). Previous tagging analyses have shown that sablefish may be trap shy towards longline gear within the first year after capture, likely because of the stress incurred during the initial capture (Carlile, et al.3). Because many of our fish were caught soon after capture in the fishery, some of the difference in recapture rate that we saw may be explained by the differential recapture catch rates between pot and longline gear types. Finally, amphipod predation was significantly higher in 1989 (12.45%) than in 1990 (3.86%) indicating that Chatham Strait may have a higher incidence of amphipods, which we found to be related to a decreased recapture rate. We calculated an absolute mortality rate for each level of severity of hook injury. The overall mortality rate of 11.71% is substantially lower than the 25% mortality rate assumed for sablefish discarded in the Pacific halibut fishery in state waters (i.e., Chatham 3 Carlile, D., B. Richardson, M. Cartwright, and V. M. O’Connell. 2002. Southeast Alaska sablefish stock as- sessment activities 1998-2001. Regional Information Report IJ02-02, 86 p. Alaska Dep. Fish and Game, Douglas, AK. 278 Fishery Bulletin 1 10(2) and Clarence Straits; Dressel1), and the assumed 100% mortality of sablefish caught in other target fisheries in federal waters in Alaska (Hanselman et ah, 2010). Applying the 11.71% mortality rate to the average catch of sablefish discarded in federally managed hook- and-line fisheries (491 t, 2004-09 average; Hanselman et al., 2010), yields an annual discard mortality of 57.5 tons. There are two reasons why our estimate of absolute discard mortality may be lower than what occurs in the commercial fishery. First, in our study fish were handled carefully and released, whereas in commer- cial fisheries we would expect a greater proportion of moderate and severe injuries that would result in a higher discard mortality. Second, commercial fishery discards come from multiple fisheries that use numer- ous gear types, most notably different hook types and sizes. Larger hooks have been shown to result in higher discard mortality (Trumble et al., 2000). Because the halibut fishery in Alaska uses larger hooks than we used in our study, a higher discard mortality rate for sablefish would be expected in the halibut fishery. Care- ful hook removal during release of fish could potentially minimize discard mortality rates observed in commer- cial fisheries. Conclusion In this study we examined some of the factors that affect the discard mortality rate of sablefish in Alaskan long- line fisheries. We found a decrease in recapture rates for fish originally captured at shallower depths (210-319 m) in our study, sustaining severe hooking injuries, and sustaining amphipod predation injuries. Based on the severity of hook injury, we estimated an overall discard mortality rate of 11.71%. Obtaining data on the relative occurrence of the severity of hook injuries that occur in these fisheries is a logical next step. Such data would allow us to extrapolate our findings more reliably and may lead to a more accurate accounting of total mortal- ity attributable to fishing and to improved management of this species. Acknowledgments This analysis was performed while author M. Stachura was a student intern from the University of Miami and funded by the National Oceanic and Atmospheric Admin- istration (NOAA) Ernest F. Hollings Undergraduate Scholarship Program administered by Oak Ridge Associ- ated Universities through a Cooperative Grant sponsored by NOAA. We thank N. Maloney for managing the tag data for the entirety of this project. We also thank E. Varosi and J. Fujioka for help in designing hook injury codes, and the crew of the RV Townsend Cromwell. This manuscript has benefited from review by D. DiResta, G. Thomas, J. Richardson, J. Murphy, K. Echave, P. Rigby, and three anonymous reviewers. Literature cited Aalbers, S. A, G. M. Stutzer, and M. A. Drawbridge. 2004. The effects of catch-and-release angling on the growth and survival of white seabass captured on offset circle and J-type hooks. N. Am. J. Fish. Manage. 24:793-800. Alos, J., M. Palmer, and A. M. Grau. 2009. Mortality of Diplodus annularis and Lithognathus mormyrus released by recreational anglers: implications for recreational fisheries management. Fish. Manage. Ecol. 16:298-305. Bartholomew, A., and J. A. Bohnsack. 2005. A review of catch-and-release angling mortality with implications for no-take reserves. Rev. Fish. Biol. Fish. 15:129-154. Brill, R., C. Magel, M. Davis, R. Hannah, and P. Rankin. 2008. Effects of rapid decompression and exposure to bright light on visual function in black rockfish ( Sebastes melanops) and Pacific halibut (Hippoglossus stenole- pis). Fish. Bull. 106:427-437. Collins, M. R., J. C. McGovern, G. R. Sedberry, H. S. Meister, and R. Pardieck. 1999. Swim bladder deflation in black sea bass and ver- milion snapper: potential for increasing postrelease survival. N. Am. J. Fish. Manage. 19:828-832. Davis, M. W. 2002. Key principles for understanding fish bycatch dis- card mortality. Can. J. Fish. Aquat. Sci. 59:1834-1843. 2005. Behaviour impairment in captured and released sablefish: ecological consequences and possible substi- tute measures for delayed discard mortality. J. Fish Biol. 66:254-265. Davis, M. W., B. L. Olla, and C. B. Schreck. 2001. Stress induced by hooking, next towing, elevated sea water temperature and air in sablefish: lack of con- cordance between mortality and physiological measures of stress. J. Fish. Biol. 58:1-15. Davis, M. W., and M L. Ottmar. 2006. Wounding and reflex impairment may be predic- tors for mortality in discarded or escaped fish. Fish. Res. 82:1-6. Gitschlag, G. R., and M. L. Renaud. 1994. Field experiments on survival rates or caged and released red snapper. N. Am. J. Fish. Manage. 14:131-136. Hanselman, D. H., C. R. Lunsford, and C. J. Rodgveller. 2010. Assessment of the sablefish stock in Alaska. In Stock assessment and fishery evaluation report for the groundfish resources of the Gulf of Alaska, chapter 3, p. 329-468. [Available from North Pacific Fishery Man- agement Council, 605 W. 4th Ave, Anchorage, AK 99510.] Hiatt, T., M. Dalton, R. Felthoven, B. Fissel, B. Garber-Yonts, A. Haynie, S. Kasperski, D. Lew, C. Package, J. Sepez, and C. Seung. 2010. Economic status of the groundfish fisheries off Alaska, 2009. In Stock assessment and fishery evalu- ation report for the groundfish fisheries of the Gulf of Alaska and Bering Sea/Aleutian Islands area, 254 p. (Available from North Pacific Fishery Management Council, 605 W. 4th Ave., Anchorage, AK 99510.] IPHC (International Pacific Halibut Commission). 1998. The Pacific halibut: biology, fishery, and manage- ment. Int. Pac. Halibut Comm. Tech. Rep. 40, 64 p. IPHC, Seattle, WA. Stachura et al.: Estimation of discard mortality of Anoplopoma fimbria in Alaska longline fisheries 279 Kaimmer, S. M. 1994. Halibut injury and mortality associated with manual and automated removal from setline hooks. Fish. Res. 20:165-179. 2000. Pacific halibut tag release programs and tag release and recovery data, 1925 through 1998. Int. Pac. Halibut Comm. Tech. Rep. 41, 32 p. IPHC, Seattle, WA. Kaimmer, S. M., and R. J. Trumble. 1998. Injury, condition, and mortality of Pacific halibut bycatch following careful release by Pacific cod and sablefish longline fisheries. Fish. Res. 38:131-144. Lsenoff, M., and R. Lancelot. 2010. aod: Analysis of overdispersed data. (Available from http://CRAN.R-project.org/package = aod. accessed October 2010.] Muoneke, M. I., and W. M. Childress. 1994. Hooking mortality: a review for recreational fish- eries. Rev. Fish Sci. 2:123-156. Neat, F. C., M. Breen, R. M. Cook, I. M. Gibb, and P. J. Wright. 2009. Electronic tags reveal behavior of captured and discarded fish. J. Fish Biol. 74:715-721. Palsson, 6. K., H. A. Einarrson, and H. Bjornsson. 2003. Survival experiments of undersized cod in a hand- line fishery in Iceland. Fish. Res. 61:73-86. R Development Core Team. 2010. R: A language and environment for statistical computing. R Foundation for Statistical Computing. (Available from: http://www.R-proiect.org. accessed August 2010.1 Seber, G. A. F. 1982. Estimation of animal abundance and related parameters, 2n(i ed., 653 p. Macmillan Publ. Co., Inc., New York. St. John, J., and C. J. Syers. 2005. Mortality of the demersal West Australian dhufish, Glaucosoma hebraicum (Richardson 1845) following catch and release: the influence of capture depth, venting and hook type. Fish. Res. 76:106-116. Schirripa, M. J. 2008. Status of the sablefish resource off the continental U.S. Pacific coast in 2007. In Pacific Coast groundfisn fishery stock assessment and fishery evaluation, on what page or pages? [Available from Pacific Fishery Management Council, 7700 NE Ambassador PI., Port- land, OR 97220.] Trumble, R. J., S. M. Kaimmer, and G. H. Williams. 2000. Estimation of discard mortality rates for Pacific halibut in groundfish longline fisheries. N. Am. J. Fish. Manage. 20:931-939. Wilson, R. R., and K. M. Burns. 1996. Potential survival of released grouper caught deeper than 40 m based on shipboard and in-situ observation, and tag-recapture data. Bull. Mar. Sci. 58:234-247. Yano, K., J. D. Stevens, and L. J. V. Compagno. 2007. Distribution, reproduction and feeding of the Green- land shark Somniosus ( Somniosus ) microcephalus, with notes on two other sleeper sharks, Somniosus ( Somnio- sus) pacificus and Somniosus ( Somniosus ) antarcticus. J. Fish. Biol. 70:374-390. 280 Fishery Bulletin 1 10(2) Fishery Bulletin Guidelines for authors Manuscript preparation Contributions published in Fishery Bulletin describe original research in marine fishery science, fishery engineering and economics, as well as the areas of marine environmental and ecological sciences (including modeling). Preference will be given to manuscripts that examine processes and underlying patterns. Descriptive reports, surveys, and observational papers may occa- sionally be published but should appeal to an audience outside the locale in which the study was conducted. Although all contributions are subject to peer review, responsibility for the contents of papers rests upon the authors and not on the editor or publisher. Submission of an article implies that the article is original and is not being considered for publication elsewhere. Articles may range from relatively short contributions (10-15 typed, double-spaced pages, tables and figures not included) to extensive contributions (20-30 typed pages). Manuscripts must be written in English; authors whose native language is not English are strongly advised to have their manuscripts checked by English-speaking colleagues before submission. Title page should include authors’ full names and mailing addresses and the senior author’s telephone, fax number, and e-mail address. Abstract should be limited to 250 words (one-half typed page), state the main scope of the research, and emphasize the author’s conclusions and relevant findings. Do not review the methods of the study or list the contents of the paper. Because abstracts are circulated by abstracting agen- cies, it is important that they represent the research clearly and concisely. General text must be typed in 12-point Times New Roman font throughout. A brief introduction should convey the broad significance of the paper; the remain- der of the paper should be divided into the following sections: Materials and methods, Results, Discus- sion, Conclusions, and Acknowledgments. Headings within each section must be short, reflect a logical sequence, and follow the rules of subdivision (i.e., there can be no subdivision without at least two items). The entire text should be intelligible to interdisciplinary readers; therefore, all acronyms, abbreviations, and technical terms should be written out in full the first time they are mentioned. For general style, follow the Style Manual (2008. [Available at http://www,gpoaccess. gov/stvlemanual/index.htmll published by the U.S. Gov- ernment Printing Office and Scientific Style and Format: the CSE Manual for Authors, Editors, and Publishers (2006, 7th ed.) published by the Council of Science Edi- tors. For scientific nomenclature, use the current edition of the American Fisheries Society’s Common and Sci- entific Name of Fishes from the United States, Canada, and Mexico and its companion volumes (Decapod Crus- taceans, Mollusks, Cnidaria and Ctenophora, and World Fishes Important to North Americans). For species not found in the above mentioned AFS publications and for more recent changes in nomenclature, use the Integrated Taxonomic Information System (available at http://itis. gov/), or, secondarily, the California Academy of Sciences’ Catalog of Fishes (available at http://researcharchive. calacademv.org/research/ichthvologv/catalog/fishcatmain. asp) for species names not included in ITIS. Dates should be written as follows: 11 November 2000. Measurements should be expressed in metric units, e.g., 58 metric tons (t); if other units of measurement are used, please make this fact explicit to the reader. Use numerals, not words, to express whole and decimal numbers in the general text, tables, and figure captions (except at the beginning of a sentence). For example: We considered 3 hypotheses. We collected 7 samples in this location. Refrain from using the shorthand slash (/), an ambiguous symbol, in the general text. Equations and mathematical symbols should be set from a standard mathematical program (MathType) or tool (Equation Editor in MS Word). LaTex is accept- able for more advanced computations. For mathematical symbols in the general text (a, yf, n, ±, etc.), use the symbols provided by the MS Word program and italicize all variables. Do not use the photo mode when creating these symbols in the general text. Literature cited section comprises published works and those accepted for publication in peer-reviewed journals (in press). Follow the name and year system for citation format in the “Literature cited” section (that is to say, citations should be listed alphabetically by the authors’ last names, and then by year if there is more than one citation with the same authorship). If there is a sequence of citations in the text, list chronologically: (Smith, 1932; Green, 1947; Smith and Jones, 1985). Abbreviations of serials should conform to abbreviations given in Cambridge Scientific Abstracts (http://www. csa.com/ids70/serials source list.php?db = aquclust- set-c). Authors are responsible for the accuracy and completeness of all citations. Literature citation format: Author (last name, followed by first-name initials). Year. Title of article. Abbreviated title of the journal in which it was published. Always include number of pages. Cite all software and special equipment or chemical solu- tions used in the study within parentheses in the text (e.g., SAS, vers. 6.03, SAS Inst., Inc., Cary, NC). Footnotes are used for all documents that have not been formally peer reviewed and for observations and communications. These types of references should be cited sparingly in manuscripts submitted to the journal. All reference documents, administrative reports, inter- nal reports, progress reports, project reports, contract reports, personal observations, personal communica- tions, unpublished data, manuscripts in review, and council meeting notes are footnoted in 9 pt font and placed at the bottom of the page on which they are first cited. Footnote format is the same as that for formal literature citations. A link to the online source (e.g., 281 [Available at http://www/ , accessed July, 2007.]), or the mailing address of the agency or department holding the document, should be provided so that readers may obtain a copy of the document. Tables are often overused in scientific papers; it is seldom necessary or even desirable to present all the data associated with a study. Tables should not be excessive in size and must be cited in numerical order in the text. Headings should be short but ample enough to allow the table to be intelligible on its own. All unusual symbols must be explained in the table legend. Other incidental comments may be footnoted with italic numeral footnote markers. Use asterisks only to indicate significance in statistical data. Do not type table legends on a separate page; place them above the table data. Do not submit tables in photo mode. Figures must be cited in numerical order in the text. Graphics should aid in the comprehension of the text, but they should be limited to presenting patterns rather than raw data. Figures should not exceed one figure for every four pages of text. Figures must be labeled with number of the figure. Avoid placing labels verti- cally (except fory axis). Figure legends should explain all symbols and abbreviations seen in the figure and should be double-spaced on a separate page at the end of the manuscript. Color is allowed in figures to show morphological differences among species (for species identification), to show stain reactions, and to show gradations in temperature contours within maps. Color is discouraged in graphs, and for the few instances where color may be allowed, the use of color will be determined by the Managing Editor. • Probability is notated with a capital, italic P. • Zeros should precede all decimal points for values less than one. • Sample size, n, should be italicized. Do not confuse sample size (n) with number of samples, number of transects, number of hauls, etc. • 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. • Use a comma in numbers of five digits or more (e.g. 13,000 but 3000). • Maps require a North arrow and degrees latitude- longitude (e.g., 170°E). Failure to follow these guidelines and failure to correspond with editors in a timely manner will delay publication of a manuscript. Copyright law does not apply to Fishery Bulletin, which falls within the public domain. However, if an author reproduces any part of an article from Fishery Bulletin in his or her work, reference to source is con- sidered correct form (e.g., Source: Fish. Bull. 97:105). Submission Submit manuscript online at http://mc.manuscriptcentral. com/fishervbulletin. Commerce Department authors should submit papers under a completed NOAA Form 25-700. For further details on electronic submission, please contact the Associate Editor, Julie Scheurer, at julie.scheurer@noaa.gov When requested, the text and tables should be submit- ted in Word format. Figures should be sent as PDF files, Windows metafiles, tiff files, or EPS files. Send a copy of figures in the original software if conversion to any of these formats yields a degraded version. Questions? If you have questions regarding these guidelines, please contact the Managing Editor, Sharyn Matriotti, at sharyn.matriotti@noaa.gov Questions regarding manuscripts under review should be addressed to Julie Scheurer, Associate Editor. Fishery Bulletin Subscription form Superintendent of Documents Publications Order Form *5178 1 I YES, please send me the following publications: Subscriptions to Fishery Bulletin for $32.00 per year ($44.80 foreign) The total cost of my order is $ . Prices include regular domestic postage and handling and are subject to change. (Company or Personal Name) (Please type or print) (Additional address/attention line) (Street address) (City, State, ZIP Code) (Daytime phone including area code) (Purchase Order No.) Charge your order. IT’S EASY! Please Choose Method of Payment: ] Check Payable to the Superintendent of Documents ] GPO Deposit Account -□ ] VISA or MasterCard Account To fax your orders (202) 512-2104 (Credit card expiration date) (Authorizing Signature) Mail To: Superintendent of Documents P.O. Box 371954, Pittsburgh, PA 15250-7954 Thank you for your order! Also available online at http://bookstore.gpo.gov/actions/GetPublication.do?stocknumber=703-023-00000-2