Ecosystem-based fishery management (EBFM) requires minimizing the risk of irreversible change to a natural assemblage of species. Because pelagic fisheries modify the biotic community rather than the physical environment and often target migratory species, the pelagic community can be expected to respond quickly after fishing pressure is reduced. We monitored the pelagic zooplanktivore community over 4 decades in the central Bering Sea before and after the closures of large-scale pelagic fisheries. We found a quick and steep change in the pelagic fish community structure after fishing pressure was reduced, which reflected increased abundance of salmon species; however, walleye pollock stocks, one of the most valuable fisheries resources in this area, have not recovered. A lower population growth rate might contribute to the slow recovery of pollock stocks. Long-term ecosystem monitoring of fishing grounds and surrounding ecosystems will be required for successful implementation of EBFM of pelagic fish stocks.

Salmon food habits studies are important to BASIS research because they help identify salmon densitydependent effects on growth and survival and can be used for food web models (NPAFC 2001). Since the 1960s these studies have been conducted in the western (e.g., Ito 1964; Andrievskaya 1966; Machidori 1968; Karpenko 1982a; Karpenko and Maksimenkov 1988; Chuchukalo et al.1995; Klovach et al. 1996; Koval and Karpenko 1998; Bugaev and Shaporev 2002; Karpenko 2003; Smorodin et al. 2004), eastern (e.g., Nishiyama 1974; Straty 1974; Carlson et al. 1998; Murphy et al. 2003; Davis et al. 2004; Farley et al. 2004), and central (e.g., Kanno and Hamai 1972; Azuma 1992; Davis et al. 2000; Myers et al. 2004) Bering Sea. In this abstract, we review information on Bering Sea salmon food habits useful to BASIS researchers investigating salmon prey, diet overlap, ration, and bioenergetics. Salmon distribution during their ocean migration has been linked to prey availability (Nishiyama 1974; Straty 1974; Carlson 1976; Karpenko 1979, 1983; Karpenko and Piskunova 1984; Sobolevskiy et al. 1994; Farley et al. 2004; Klovach and Gruzevich 2004). Review of 40 Bering Sea salmon food habits studies showed the major prey common to all species of salmon include euphausiids (Thysanoessa), hyperiid amphipods (Themisto), pteropods (Limacina), and juvenile squids (Gonatopsis, Gonatus, and Berryteuthis). Important fish prey include Stenobrachius leucopsarus and juvenile greenlings (Pleurogrammus) in basin habitats, and juvenile walleye pollock (Theragra chalcogramma), capelin (Mallotus villosus), and sand lance (Ammodytes) in shelf habitats. Future studies should focus on distribution and abundance of these major prey because this will improve our assessment of diet overlap and feeding competition. In response to the biennial cycle of pink salmon abundance in the Bering Sea, density-dependent shifts in prey composition have been observed in the food habits of pink, chum, and sockeye salmon (Ito 1964; Andrievskaya 1966; Tadokoro et al. 1996; Karpenko et al. 1998; Davis 2003). During even-numbered years (low abundance of maturing pink salmon), total stomach content weight and proportions of euphausiids, copepods, fish, and squid increase in sockeye and pink salmon, and the proportion of euphausiids and other crustaceans increase in chum salmon stomach contents (Fig.1). Salmon diet overlaps (percent similarity index) were calculated for sockeye, chum, pink, and chinook salmon collected in the basin during the summer cruises (1991–2003) of the Wakatake maru (Fig. 2). Results showed a very high (> 75%) diet overlap between sockeye and pink salmon in oddand evennumbered years. In odd-numbered years, however, diet overlaps between chum and sockeye, and chum and pink salmon were reduced. Comparing summer to fall overlaps using data collected onboard the Northwest Explorer in 2002 (Davis et al. 2004), indicated seasonal reduction in diet similarity between sockeye and chum salmon, and sockeye and chinook salmon (Fig. 2). In fall, overlap between sockeye and chum salmon in the Aleutian Islands was very high (> 75%), while overlap between chum and chinook salmon was moderate (28–30%) in the basin and the eastern shelf.

In recent decades, the marine production of Asian and North American Pacifi c salmon and steelhead populations has undergone signifi cant variability linked to climate change. Improved forecasts of the abundance and distribution of salmon are needed that will benefi t stock management in all salmon producing countries around the North Pacifi c Rim. The North Pacifi c Anadromous Fish Commission (NPAFC) Science Plan is a long-term comprehensive strategy for international cooperative research. The primary goal of the 2011–2015 Science Plan was to explain and forecast annual variations in Pacifi c salmon production. The plan was developed with an overarching research theme “Forecast of Pacifi c Salmon Production in the Ocean Ecosystems under Changing Climate” and fi ve research topics. This paper describes progress made on each research topic and the overarching theme, much of which was assessed at an international symposium in Kobe, Japan, on May 17–19, 2015. In summary, the reliability of stock identifi cation methods including genetic and otolith mark analyses has improved, enabling better monitoring of stock-specifi c ocean distribution and abundance. Salmon marine survival depends on early marine coastal environments but also on conditions later in life, including winter. Models incorporating fi sh mortality and various environmental factors improve our ability to forecast returns of specifi c salmon stocks. However, limitations on our ability to accurately explain and forecast annual variations in Pacifi c salmon production remain, in part because of uncertainty in the factors responsible for salmon mortality and from the eff ects of climate warming on the marine distribution and abundance of salmon. It is more important than ever to promote cooperative and innovative international research to identify and better understand the ecological mechanisms regulating the distribution and abundance of salmon populations for sustainable salmon and steelhead management.

How Pacifi c salmon and steelhead (Oncorhynchus spp.) respond to climate-driven changes in their oceanic environment is highly uncertain, in part due to limited information on winter distribution in international waters (high seas) of the North Pacifi c Ocean and Bering Sea. We review what is known and summarize what should be known to properly address the question: Where do Pacifi c salmon go in the high seas during winter and why, and how might this be aff ected by climate change? Historical high-seas research (1950s–1970s, all seasons) discovered that there are species and stock-specifi c distributions in the high seas; winter survey results provided some clues as to important winter locations and dominant oceanographic features of winter habitat. In succeeding decades (1980–2015), new fi sheries-oceanographic survey methods, stock-identifi cation techniques, remote-sensing technologies, and analytical approaches have enabled us to expand our knowledge of the winter distribution and ecology of salmon, although empirical data are still very limited. In general, we learned that the “why” of ocean distribution of salmon is complex and variable, depending on spatio-temporal scale and synergies among heredity, environment, population dynamics, and phenotypic plasticity. The development of quantitative multispecies, multistage models of salmon ocean distribution linked to oceanographic features would help to identify key factors infl uencing winter distribution and improve understanding of potential climate change eff ects.

To assess effects of intra- and inter-specific interactions on chum salmon in the central Bering Sea, chum salmon lipid content was analyzed as a proxy for body condition. We measured the lipid contents of 466 immature individuals collected during summer from 2002 to 2007. Individual variation in log-transformed lipid content was tested using multiple regression analysis with biological and environmental variables. A regression model that included chum salmon fork length and pink salmon CPUE (number of fish caught per 1500 m of gillnet) was the most effective in describing variation in lipid content. Path analysis showed that the negative effect of pink salmon CPUE was stronger than the effect of chum salmon CPUE on chum salmon lipid content. Stomach content analysis of 283 chum salmon indicated non-crustacean zooplankton (appendicularian, chaetognath, cnidarian, ctenophore, polychaete, and pteropod) was higher under conditions of high pink salmon CPUE. Increased consumption of non-crustacean zooplankton containing a low lipid level could lower the lipid content of chum salmon. Thus, chum salmon lipid content could be affected directly by their shift in prey items and indirectly by interspecific competition with pink salmon.

Information on prey availability, diets, and trophic levels of fish predators and their prey provides a link between physical and biological changes in the ecosystem and subsequent productivity (growth and survival) of fish populations. In this study two long-term data sets on summer diets of steelhead (Oncorhynchus mykiss) in international waters of the central North Pacific Ocean (CNP; 1991–2009) and Gulf of Alaska (GOA; 1993–2002) were evaluated to identify potential drivers of steelhead productivity in the North Pacific. Stable isotopes of steelhead muscle tissue were assessed to corroborate the results of stomach content analysis. We found the composition of steelhead diets varied by ocean age group, region, and year. In both the GOA and CNP, gonatid squid (Berryteuthis anonychus) were the most influential component of steelhead diets, leading to higher prey energy densities and stomach fullness. Stomach contents during an exceptionally warm year in the GOA and CNP (1997) were characterized by high diversity of prey with low energy density, few squid, and a large amount of potentially toxic debris (e.g., plastic). Indicators of good diets (high proportions of squid and high prey energy density) were negatively correlated with abundance of wild populations of eastern Kamchatka pink salmon (O. gorbuscha) in the CNP. In conclusion, interannual variations in climate, abundance of squid, and density-dependent interactions with highly-abundant stocks of pink salmon were identified as potential key drivers of steelhead productivity in these ecosystems. Additional research in genetic stock identification is needed to link these potential drivers of productivity to individual populations.

SUMMARY The activity of the pituitary–gonadal axis (PG axis) in pre-migratory and homing chum salmon was examined because endocrine mechanisms underlying the onset of spawning migration remain unknown. Pre-migratory fish were caught in the central Bering Sea in June, July and September 2001, 2002 and 2003, and in the Gulf of Alaska in February 2006. They were classified into immature and maturing adults on the basis of gonadal development. The maturing adults commenced spawning migration to coastal areas by the end of summer, because almost all fish in the Bering Sea were immature in September. In the pituitaries of maturing adults, the copy numbers of FSHβ mRNA and the FSH content were 2.5- to 100-fold those of the immature fish. Similarly, the amounts of LHβ mRNA and LH content in the maturing adults were 100- to 1000-fold those of immature fish. The plasma levels of testosterone,11-ketotestosterone and estradiol were higher than 10 nmol l–1 in maturing adults, but lower than 1.0 nmol l–1 in immature fish. The increase in the activity of the PG-axis components had already initiated in the maturing adults while they were still in the Gulf of Alaska in winter. In the homing adults, the pituitary contents and the plasma levels of gonadotropins and plasma sex steroid hormones peaked during upstream migration from the coast to the natal hatchery. The present results thus indicate that the seasonal increase in the activity of the PG axis is an important endocrine event that is inseparable from initiation of spawning migration of chum salmon.

Declining runs of Chinook salmon in western Alaska have focused interest on the ocean condition and food habits of Chinook salmon in the Bering Sea, including potential mortality from bycatch in the pollock fishery. Examination of Chinook salmon stomach contents collected in the eastern Bering Sea by the U.S. North Pacific Groundfish Observer Program (NOAA Fisheries) revealed isolated pieces of skin, bones, and fins (offal) belonging to large-bodied fish which were physically identified as either walleye pollock ( Theragra chalcogramma) or Pacific cod (Gadus macrocephalus). To confirm the species identification of the offal, we matched DNA sequences of these offal samples to known sequences of walleye pollock and Pacific cod. Novel mitochondrial DNA (mtDNA) primers were designed to amplify a 174-base pair (bp)-long section of the cytochrome c oxidase subunit I (COI) gene, which was sequenced and compared with sequences downloaded from the GenBank database. Typically, much longer sections (~700 bp) of DNA are used for species identification but due to the state of digestion of the samples, long sequences of DNA were no longer present. The specific design of our primers, however, allowed us to make positive identification and differentiation of walleye pollock and Pacific cod. Of the 15 offal samples, nine yielded usable sequences, all of which were positively identified as walleye pollock. Our results clearly demonstrate the utility of a short COI sequence for species identification of Chinook salmon stomach contents that might otherwise be unidentifiable due to either the state of digestion, or because the salmon consumed isolated body parts (offal) rather than whole fish. These results suggest that walleye pollock offal supplements the diet of Chinook salmon during winter.

The BASIS food habits studies of sockeye, chum, pink, and Chinook salmon conducted in 2002-2006 were summarized. These studies identified important (≥ 10% of prey composition by weight) prey taxa of salmon. Salmon diet composition differed between the western region, where diets contained more zooplankton, and the eastern region, where diets contained more ichthyoplankton and nekton. Salmon feeding conditions, growth, and survival in the eastern region were more favorable in relatively warm years, as compared to cool years. However, warmer conditions may not be favorable for all salmon species, such as chum salmon. These studies significantly increased the available information on salmon food habits during the fall in the western, central, and eastern regions. Salmon diet composition shifted from zooplankton to fish and squid, or to larger sizes of fish prey, with increasing salmon body size, age, or maturity. Continued monitoring of salmon food habits will contribute to understanding how future climate changes will affect salmon populations in the Bering Sea.

This is the first study of winter diets of Chinook salmon in the eastern Bering Sea. We analyzed Chinook salmon stomach samples collected by U.S. observers on board commercial groundfish trawlers from January to March and July to August, 2007. The proportion of empty stomachs was higher in winter (45%) than summer (8%), suggesting longer time periods between meals in winter. Diversity of squid species in Chinook salmon diets was higher in winter than summer, when more fish, particularly juvenile walleye pollock, were consumed. All age groups of Chinook salmon collected in winter consumed fish offal, likely generated by fishery catch-processing activities, however, fish offal was not observed in summer samples. In winter, the ratio of euphausiids and fish offal weight to Chinook salmon body weight was significantly higher in samples collected at shallow depths (< 200 m), and the ratio of squid was significantly higher in salmon collected at deeper depths (201-600 m). The ratio of euphausiids to fish body weight was significantly higher in immature than maturing Chinook salmon.

We estimated bias-corrected mean fork lengths of gillnet-caught chum salmon using a size selectivity estimate of the gillnet to test how the bias correction affects the estimated temporal pattern of chum salmon body size, during 1971-1994 and 1994-2007. Results showed bias-corrected mean fork lengths were smaller than uncorrected means. Therefore, when examining ontogenetic changes in fish size (e.g. the growth trajectory) using data collected by research gillnets, the uncorrected mean fork length can overestimate the true value. Comparison of temporal trends in bias-corrected mean fish lengths to uncorrected means showed similar results because both illustrated a decrease in chum salmon fork length in 1971-1994, and a stable fish size after 1994. Uncorrected mean values of chum salmon fork length for fish caught using research gillnets can be used as a proxy for fish size to examine temporal trends. We conclude that interpreting temporal trends using either uncorrected or bias-corrected data will support the same general conclusions regarding long-term changes in chum salmon body size.

Some research gillnets with size combinations based on a geometric series have been used for research surveys underpinning the stock assessment of fresh-water and marine fish. We assessed a bias in size composition of chum salmon caught using a research gillnet consisting of ten different mesh sizes based on a geometric series of factor 1.14. In all, 11 fishing operations were conducted for gear intercalibration between the research gillnet and a midwater trawl in the central Bering Sea. The best-fit selectivity model to pooled catch data included different fishing intensities among gillnet meshes. The pooled catch efficiency and the maximum catch efficiency of the gillnet increased with fish size. Estimated size composition of chum salmon was more similar to trawl catches than to research gillnet catches. Bias in size composition of research gillnet catches may be caused by the difference in encounter probability among mesh sizes, variability in fish swimming speed based on fish size, mesh visibility influencing fish behaviour, and diel vertical migration of chum salmon. When conducting multimesh gillnet surveys for stock assessment, researchers should correct a bias in size composition by performing gear intercalibrations.

To improve understanding of the mechanisms affecting growth and survival, we evaluated the summer diets and feeding patterns (prey composition, energy density, and stomach fullness) of hatchery and wild juvenile pink salmon Oncorhynchus gorbuscha in Prince William Sound (PWS) and the northern coastal Gulf of Alaska (CGOA). Our study (1999-2004) included 2 years of low (similar to 3%), mid (similar to 5%), and high (similar to 8-9%) survival of PWS hatchery pink salmon. Because variations in diet should affect growth and ultimately survival, we expected that the variations in diet, growth, and survival would be correlated. During August in the CGOA, pteropod-dominated diets and higher gut fullness corresponded to high survival (5-9%), and copepod-dominated diets and lower gut fullness corresponded to low survival (3%). Within years, no significant differences were found in diet composition or gut fullness between hatchery and wild fish or among the four PWS hatchery stocks. Diets varied by water mass (habitat) as juveniles moved from PWS to more saline habitats in the CGOA. In July, when juveniles were most abundant in PWS, their diets were dominated by pteropods and hyperiid amphipods. The diets of fish that moved to inner-shelf (i.e., the least-saline) habitat in the CGOA in July were dominated by larvaceans in low-survival years and pteropods in high-survival years. Diet quality was higher in CGOA habitats than in PWS in July. In August, fish moved to the more productive, more saline water masses in the CGOA, where large copepods and pteropods were dominant prey and diet quality was better than in PWS. Our results indicate that spatial variation in the diets of juvenile pink salmon in July and the timing of migration to the CGOA play a critical role in marine growth and survival.

The Model for Assessing Links Between Ecosystems (MALBEC) is a policy gaming tool with potential to explore the impacts of climate change, harvest policies, hatchery policies, and freshwater habitat capacity changes on salmon at the North Pacific scale. This article provides background information on the MALBEC project, methods, input data, and preliminary results pertaining to (1) hatchery versus wild salmon production in the North Pacific Ocean, (2) rearing, movement, and interactions among Pacific salmon populations in marine environments, (3) marine carrying capacities, density-dependent growth, and survival in Pacific salmon stocks, and (4) climate impacts on productivity in salmon habitat domains across the North Pacific. The basic modeling strategy underlying MALBEC follows the full life cycle of salmon and allows for density-dependence at multiple life stages, and it includes spatially explicit ecosystem considerations for both freshwater and marine habitat. The model is supported by a data base including annual run sizes, catches, spawning escapements, and hatchery releases for 146 regional stock groups of hatchery and wild pink, chum, and sockeye salmon around the North Pacific for the period 1952–2006. For this historical period, various hypotheses about density-dependent interactions in the marine environment are evaluated based on the goodness-of-fit between simulated and observed annual run sizes. Based on the information we used to inform our ocean migration table, interactions among stocks that originate from geographically distant regions are greatest in the Bering Sea in summer–fall and in the eastern subArctic in winter–spring. While the model does not reproduce the observed data for some specific stock groups, it does predict the same overall production pattern that was observed by reconstructing run sizes with catch and escapement data alone. Our preliminary results indicate that simulations that include density-dependent interactions in the ocean yield better fits to the observed run-size data than those simulations without density-dependent interactions in the ocean. This suggests that for any level of ocean productivity, the ocean will only support a certain biomass of fish but that this biomass could consist of different combinations of stocks, stock numbers and individual fish sizes. MALBEC simulations illustrate this point by showing that under scenarios of Pacific-wide reduced hatchery production, the total number of wild Alaskan chum salmon increases, and that such increases are large where density-dependent effects on survival are large and small where they are not. Under scenarios with reduced freshwater carrying capacities for wild stocks, the impacts of density-dependent interactions also lead to relative increases in ocean survival and growth rates for stocks using ocean habitats where density-dependence is large. All correspondence should be addressed to N. Mantua. e-mail: nmantua@u.washington.edu The Salmon MALBEC Project: A North Pacific-scale Study to Support Salmon Conservation Planning Nathan J. Mantua1, Nathan G. Taylor1, Gregory T. Ruggerone2, Katherine W. Myers1, David Preikshot3, Xanthippe Augerot4, Nancy D. Davis1, Brigitte Dorner5, Ray Hilborn1, Randall M. Peterman5, Peter Rand6, Daniel Schindler1, Jack Stanford7, Robert V. Walker1, and Carl J. Walters3 1School of Aquatic and Fishery Sciences, University of Washington, Box 355020, Seattle, WA 98195-5020, USA 2Natural Resources Consultants, Inc., 4039 21st Avenue West, Suite 404207, Seattle, WA 98199, USA 3Fisheries Centre, 2204 Main Mall, University of British Columbia, Vancouver, BC V6T 1Z4, Canada 4Pangaea Consulting, LLC, 1615 SE Bethel Street, Corvallis, OR 97333-1251, USA 5School of Resource and Environmental Management, Simon Fraser University, Burnaby, BC V5A 1S6, Canada 6The Wild Salmon Center, 721 NW 9th Ave, Suite 300, Portland, OR 97209, USA 7Flathead Lake Biological Station, University of Montana, 32125 Bio Station Lane, Polson, MT 59860-6815, USA