Abstract

Farley, E. V., Starovoytov, A., Naydenko, S., Heintz, R., Trudel, M., Guthrie, C., Eisner, L., Guyon, J. R. 2011. Implications of a warming eastern Bering Sea for Bristol Bay sockeye salmon. – ICES Journal of Marine Science, 68: 1138–1146.

Overwinter survival of Pacific salmon (Oncorhynchus sp.) is believed to be a function of size and energetic status they gain during their first summer at sea. We test this notion for Bristol Bay sockeye salmon (O. nerka), utilizing data from large-scale fisheries and oceanographic surveys conducted during mid-August to September 2002–2008 and from February to March 2009. The new data presented in this paper demonstrate size-selective mortality for Bristol Bay sockeye salmon between autumn and their first winter at sea. Differences in the seasonal energetic signatures for lipid and protein suggest that these fish are not starving, but instead the larger fish caught during winter apparently are utilizing energy stores to minimize predation. Energetic status of juvenile sockeye salmon was also strongly related to marine survival indices and years with lower energetic status apparently are a function of density-dependent processes associated with high abundance of juvenile sockeye salmon. Based on new information regarding eastern Bering Sea ecosystem productivity under a climate-warming scenario, we hypothesize that sustained increases in spring and summer sea temperatures may negatively affect energetic status of juvenile sockeye salmon, potentially resulting in increased overwinter mortality.

Introduction

The Intergovernmental Panel on Climate Change Regional Climate Projections for Polar regions suggests that (i) the Arctic is very likely to warm during this century more than the global mean; (ii) warming is projected to be largest in winter and smallest in summer; and (iii) sea ice is very likely to decrease in its extent and thickness (Christensen et al., 2007). Probably the biggest effect on the eastern Bering Sea ecosystems will be the extent, duration during winter, and thickness of sea ice. These characteristics are believed to determine benthic vs. pelagic productivity on the eastern Bering Sea shelf (Hunt and Stabeno, 2002), as well as the boundary between Arctic and Subarctic demersal fish communities (Wyllie-Echeverria and Wooster, 1998; Grebmeier et al., 2006).

The western Alaska and Arctic regions are home to five species of salmonid (Oncorhynchus spp.) fish. These salmon are anadromous, spawning and residing in freshwater rivers and lakes for 1–3 years before migrating to the ocean. As juveniles, they spend their first summer within the relatively shallow waters of the eastern Bering Sea and Chukchi Sea feeding and growing, before moving offshore into the Bering Sea basin and North Pacific Ocean. These salmon provide economic benefits in the form of subsistence, commercial, and recreational fisheries and contribute to the cultural enrichment of the regions where they occur. Their ecological role is complex, because they facilitate energy transfer directly and indirectly at multiple trophic levels in many ecosystems. Their ability to occupy habitats in fresh, salt, and brackish water has resulted in a remarkable diversity of life histories, but climate change threatens to alter their distribution, abundance, and growth rate during critical life-history stages.

The Bristol Bay region in southwestern Alaska supports one of the largest populations of sockeye salmon (O. nerka) in North America (Burgner, 1991). Bristol Bay sockeye salmon spawn during late summer and autumn within the Bristol Bay watershed. Sockeye salmon fry typically spend 1–2 years rearing in the freshwater lakes in Bristol Bay. Outmigration to marine waters generally happens during May and early June, and they spend their first summer at sea foraging and growing within the southeastern Bering Sea region, before moving to offshore regions of the North Pacific Ocean, where they spend 2–3 more years before returning to spawn.

Understanding how Bristol Bay sockeye salmon respond to loss of sea ice and warming air and sea temperatures along the eastern Bering Sea shelf is critical for appropriate management of these important species. Pacific salmon in general experience relatively high mortality rates during their first few months at sea (Parker, 1968; Hartt, 1980), and it is believed that the high mortality rates may be partly related to size (Pearcy, 1992). Size-dependent marine mortality of juvenile salmon may be concentrated during two specific early marine life-history stages. The first stage may happen just after juvenile salmon enter the marine environment, where smaller, slower-growing individuals are believed to experience greater size-selective predation (Parker, 1968; Willette et al., 1999). The second stage is thought to happen following the first summer at sea, when smaller individuals may not have sufficient energy reserves to survive late autumn and winter (Beamish and Mahnken, 2001). The energy reserves (lipids and proteins) are likely a function of the type, abundance, and quality of prey consumed during the summer growth season. Because of low prey productivity, salmon often face food shortages during winter and must rely on the lipids accumulated during summer to fuel their metabolic functions. Hence, overwinter survival of salmon is expected to be influenced by the quantity of lipids stored before and utilized during winter.

The eastern Bering Sea shelf is an important nursery ground for juvenile Bristol Bay sockeye salmon (Farley et al., 2009), and the mechanism regulating size and condition of juvenile salmon is believed to be bottom-up control of the trophic structure (Farley et al., 2007b). A leading hypothesis for ocean productivity on the eastern Bering Sea shelf suggests that the southern extent and duration of sea ice in spring affects whether the benthic or pelagic communities benefit from spring and summer production (Hunt et al., 2002). Warmer winters with less sea ice are believed to favour pelagic productivity, potentially benefitting salmon growth and early marine survival. Changes in size, survival, distribution, diet, and growth-rate potential for western Alaska salmon in response to changing spring and summer sea surface temperatures have been noted (Farley et al., 2005, 2007b; Farley and Moss, 2009; Farley and Trudel, 2009). Although there is evidence that reduced size of juvenile pink (O. gorbuscha) and coho (O. kisutch) salmon results in higher overwinter mortality (Moss et al., 2005; Beamish et al., 2004), direct evidence that the first winter at sea is the critical period for Pacific salmon that spend more than 1 year in the ocean has not been fully documented (i.e. sockeye salmon; Farley et al., 2007a).

In this study, we develop further the critical size and period hypothesis for Bristol Bay sockeye salmon. The critical size portion of this hypothesis assumes that larger juvenile fish have more energy reserves before winter; they therefore have higher overwinter survival. Therefore, we tested this part of the hypothesis by first relating fork length to energy density (kJ g−1 wet weight) for juvenile sockeye salmon collected during 2003–2008 along the eastern Bering Sea shelf. Next, we computed a marine survival index (MSI) for juvenile sockeye salmon for 2002–2007 and related these indices to annual averages of juvenile sockeye salmon energy density. Sockeye salmon collected during autumn surveys are known to come from Bristol Bay (Farley et al., 2005), and genetic analyses described in this paper for the sockeye salmon collected during winter 2009 indicated that more than 55% of the ocean age-1 sockeye salmon were from Bristol Bay. To examine the critical period for Bristol Bay sockeye salmon, we compared size distributions and seasonal energy budgets for juvenile sockeye salmon collected during autumn 2008 with ocean age-1 sockeye salmon during winter 2009 to examine size-selective mortality and to determine the amount of energy lost from storage (lipid) and structural reserves (protein) of Bristol Bay sockeye salmon during their first winter at sea. Lastly, production of large lipid-rich crustacean zooplankton was reduced during years with anomalously warm spring and summer sea temperatures (2002–2005) on the eastern Bering Sea shelf; this is believed to have contributed to reduced energy reserves in juvenile pink salmon (Andrews et al., 2009) and lower recruitment for age-0 walleye pollock (Theragra chalcogramma; Hunt et al., 2011). Therefore, we compared mean annual energy density of juvenile sockeye salmon collected during anomalously warm (2003–2005) and cold (2006–2008; Hunt et al., 2011) spring and summer sea surface temperatures and discuss the implications of a warming Bering Sea on the early marine health of Bristol Bay sockeye salmon.

Methods

Surveys

Fisheries and oceanographic data were collected during the US Bering Aleutian Salmon International Surveys (BASIS) conducted in the southeastern Bering Sea from mid-August to September 2002–2008 (Figure 1) by scientists from the Alaska Fisheries Science Center and during a February–March 2009 winter survey conducted in the North Pacific Ocean (Figure 1) by scientists from the TINRO Center. In the southeastern Bering Sea, juvenile sockeye salmon were collected following the methods described in Farley et al. (2005). Briefly, fish were collected using a midwater rope trawl that was 198 m long, had hexagonal mesh in wings and body, and had a 1.2-cm mesh liner in the codend. The rope trawl was towed at 6.5–9.3 km h−1, at or near the surface, and had a typical spread of 55 m horizontally and 25 m vertically. Trawl stations were sampled during daylight (07:30–21:00, Alaska Daylight Savings Time), and all tows lasted 30 min and covered 2.8–4.6 km.

Figure 1.

Map providing examples of stations sampled by scientists with the Alaska Fisheries Science Center, BASIS project during mid-August–September 2002–2008 (black dots) and stations sampled by scientists with the TINRO Center during February–March 2009 (black triangles).

Figure 1.

Map providing examples of stations sampled by scientists with the Alaska Fisheries Science Center, BASIS project during mid-August–September 2002–2008 (black dots) and stations sampled by scientists with the TINRO Center during February–March 2009 (black triangles).

During the winter survey (see Starovoytov et al., 2009, for details), fish were collected using the standard midwater rope trawl that was 30 m long, with quadrangular mesh in the wings and body and a 1-cm small mesh section in the codend. The rope trawl was towed at 7.2–10.3 km h−1, at or near the surface, and had a typical spread of 49.0 m horizontally and 32.2 m vertically. Trawl stations were sampled during 24 h operations, and all tows lasted 1 h.

At each trawl station, juvenile sockeye salmon were selected at random for further analyses. Standard biological attributes, including fork length (nearest 1.0 mm) and body weight (nearest 1.0 g) of juvenile sockeye salmon were measured on board the vessels. Length frequency analyses revealed two distinct distributions for immature sockeye salmon collected during the winter 2009 survey (Figure 2). We interpret the first distribution (FL >200 and <320 mm) to be ocean age-1 sockeye salmon and the distribution of larger fish to be ocean age-2+.

Figure 2.

Fork length frequency for sockeye salmon caught during the February–March 2009 survey.

Figure 2.

Fork length frequency for sockeye salmon caught during the February–March 2009 survey.

Marine survival index

An MSI for juvenile sockeye salmon was estimated by dividing the relative abundance of juvenile sockeye salmon collected during autumn (RAt) by the number of adult sockeye salmon returning to Bristol Bay, 2 and 3 years later (Nt+2, Nt+3): 

formula
Juvenile sockeye salmon relative abundance was estimated by: 
formula
where TA is the total area surveyed each year (153 437 km2), ASt the average area sampled by our trawl gear (km2) during year t, Avecpuet the average catch per unit effort for the survey within year t, and q the catchability of our trawl, estimated to be 0.3 (Farley et al., 2007c). In addition, the estimated total area surveyed is less than that reported in Farley et al. (2007c). This change was made to standardize the area surveyed during 2002–2007 to a region that was consistently sampled during all years.

Age-structured adult sockeye salmon return data for Bristol Bay were provided by the Alaska Department of Fish and Game (T. Baker, Alaska Department of Fish and Game, pers. comm.). Age-structured return data were not available for ocean age-3 sockeye salmon returning during 2010, hence the ocean age-3 returns were estimated using the Alaska Department of Fish and Game Bristol Bay sockeye salmon forecast for 2010.

The confidence intervals for relative abundance and MSI were estimated using the bootstrap method (Efron, 1982; Gunderson, 1993). The surveys were performed using a systematic grid design; the bootstrap process was used to conduct 1000 simulated surveys, each simulation using a different random selection (with replacement) of sampling stations within a given year. This provided 1000 new estimates of relative abundance within a given year and the adult sockeye salmon returns in juvenile year t were divided by these new relative abundance estimates to provide 1000 new estimates for MSI for each year. The lower and the upper confidence intervals for the new mean relative abundance and MSI estimates were calculated using 

formula
 
formula
where forumla and forumla are the mean and variance of the 1000 bootstrap simulations for relative abundance and MSI in year t, respectively.

Genetic analyses

Distribution and migration pathways for juvenile Bristol Bay sockeye salmon were described in Farley et al. (2005). Genetic analyses were performed on juvenile sockeye salmon collected during 2002–2007; although not published, these analyses established that juvenile caught during the surveys were Bristol Bay stocks (J. Seeb, University of Washington, pers. comm.). Therefore, we focused the genetic analyses on ocean age-1 sockeye salmon collected during the winter 2009 survey, following methods described in Guthrie et al. (2010). Briefly, genomic DNA was extracted using a DNeasy 96 Tissue Kit by QIAGEN (reference to trade names does not imply endorsement by the National Marine Fisheries Service). A panel of 45 single-nucleotide polymorphism (SNP) markers was assayed using TaqMan reactions (Habicht et al., 2007). This panel included three mitochondrial and 42 nuclear SNPs represented in the Alaska Department of Fish and Game coast-wide SNP baseline (Habicht et al., 2010). After amplification, the Taqman genotyping reactions were assayed on an ABI PRISM 7900HT Sequence Detection System and scored using Sequence Detection Software 2.2 (Applied Biosciences, Inc.). Individual genotypes were imported into our genetic database, which was developed with Progeny software (Progeny, Inc.).

A mixture analysis using a Bayesian estimation method was implemented using BAYES software (Pella and Masuda, 2001). For this analysis, eight Monte Carlo chains starting at disparate values of stock proportions were configured such that 95% of the stocks came from one of the eight previously designated regions (Habicht et al., 2010, for baseline regional designations), with weights distributed equally among the stocks of that region. The remaining 5% was distributed equally among the remaining stocks from all other regions. A flat prior of 0.003289 was used (1/number baseline populations). The analyses were completed for a chain length of 25 000, with the first 12 500 deleted during the burn-in phase when determining overall stock compositions. Convergence of the chains to posterior distributions of stock proportions was determined with Gelman and Rubin shrink statistics, which were all 1.01 or less, conveying strong convergence to a single posterior distribution (Pella and Masuda, 2001).

Energetic analyses

Energy content of salmon was estimated from subsamples of sockeye salmon collected during the autumn 2003–2008 (no data available for 2002) and winter 2009 surveys. Juvenile sockeye salmon retained for energetic analyses during autumn surveys were frozen whole. Whole sockeye salmon were not available from samples collected during the winter 2009 survey. Instead, a chunk of muscle tissue taken from a section posterior to the gill plate just above the lateral line was used to determine the energy content of these fish (Trudel et al., 2005). The energetic budgets for lipid, protein, and energy (kJ) for sockeye salmon samples collected during autumn 2008 and winter 2009 were determined from muscle chunks following methods described by Vollenweider (2005). Briefly, lipid was extracted from ∼1 g of wet sample homogenate using a modification of Folch's method outlined by Christie (1982). Lipid, moisture, and ash content were measured gravimetrically. Protein content was estimated from the total nitrogen content, as determined by the Dumas method (Association of Official Analytical Chemists, 2002).

The energy densities (ED; kJ g−1 wet weight) for juvenile sockeye salmon collected during 2003–2007 were determined using bomb calorimetry. Before bomb calorimetry analysis, we obtained whole-body wet weight (g) and removed otoliths and stomach contents. Fish were dried in a VWR 1324 convection oven at 60–65°C until a constant weight (within 0.005 g) was obtained. Dried fish were stored in a desiccator until further processing. Individual sockeye salmon were homogenized using a pulverizer for 30 s, then transferred to a mortar and pestle and pulverized further until the sample consisted of a uniform powder. Pressed pellets weighing ∼0.15 g for each sockeye salmon sample were created and stored in a desiccator until further processing. These pellets were then combusted in a Parr 1425 Semimicro Calorimeter to determine whole-body energy content. The values generated by the calorimeter were converted from Cal g−1 dry weight to J g−1 dry weight. Energy density represents the energy per gramme of juvenile sockeye salmon and was calculated by dividing the whole-body energy content (J) by the wet weight of the fish, then dividing by 1000.

Lipid content of juvenile sockeye salmon was available only for 2008. We note that there is a good relationship between ED and lipid (ED = 4.86 + 0.132 × Lip; r2 = 0.72) for juvenile sockeye salmon taken from our autumn 2008 survey. Similar relationships between ED and lipid were found for juvenile coho salmon (ED = 3.76 + 0.049 × Lip; r2 = 0.86) and juvenile Chinook salmon (ED = 3.60 + 0.047 × Lip; r2 = 0.87) in the eastern North Pacific Ocean during several seasons spanning spring, summer, autumn, and winter (Trudel et al., 2005). Therefore, ED was used as a proxy for lipid in some of the statistical analyses.

Statistical analyses

We used analyses of covariance (ANCOVAs) to test whether the straight-line relationship between juvenile sockeye salmon ED (response) and fork length (mm; covariate) was the same for each year (2003–2008; factor) and to test whether the mean juvenile sockeye salmon ED levels for each year differ significantly, after taking into account the possible confounding effects of differing fork length distributions among years. We note that the assumption of equal slopes for the ANCOVA may be violated by including data from 2004 (Figure 3). In this analysis, it may be that the slope of the relationship between ED and weight during 2004 is not parallel with the slopes of this relationship among the other years. However, the lines are not far from being parallel. Therefore, these data are left in the ANCOVAs. Tukey's multiple comparison test was used to test for pairwise differences in ED among years.

Figure 3.

Scatterplot displaying the relationship between energy density (ED; kJ g−1 wet weight) and fork length (mm) for juvenile sockeye salmon collected during autumn 2003–2008 in the eastern Bering Sea. Linear lines for each year were fit to the data to illustrate differences in the relationship between ED and fork length for juvenile salmon among years.

Figure 3.

Scatterplot displaying the relationship between energy density (ED; kJ g−1 wet weight) and fork length (mm) for juvenile sockeye salmon collected during autumn 2003–2008 in the eastern Bering Sea. Linear lines for each year were fit to the data to illustrate differences in the relationship between ED and fork length for juvenile salmon among years.

The extent of size-selective mortality in juvenile sockeye salmon during their first winter at sea was assessed by comparing the mean and variance between length of juvenile and ocean age-1 sockeye salmon sampled during autumn 2008 and winter 2009, respectively. Mean size and variance are expected to increase and decrease, respectively, over winter, because of size-selective mortality against smaller fish. We also compared the size frequency distribution between juvenile and ocean age-1 sockeye salmon caught during autumn 2008 and winter 2009, using quantile–quantile plots, where the slopes of the quantile–quantile plots are expected to be smaller than 1.0 when there is size selective mortality on smaller fish (Post and Evans, 1989).

Lipid depletion in juvenile sockeye salmon during winter was tested by comparing the slope of the relationship between lipid and size for fish that were collected during autumn 2008 and winter 2009. Because smaller fish were expected to utilize a larger fraction of their lipid reserve over the winter period, the slope and intercept of the relationship between lipid and size were expected to be higher and lower, respectively, for fish collected during winter. Slopes of the relationship between lipid (g) and fork length (L; mm) were compared among seasons as follows: 

formula
where γ, β, and δ are the regression coefficients, D a dummy variable that takes the value of zero for fish caught during autumn and 1 for those caught during winter, and ɛ a randomly distributed error. Positive values of δ indicate that lipid is depleted faster in smaller fish, whereas negative values indicate that lipid is depleted faster in larger fish. The difference in slopes between the relationships for protein (g) and fork length was also tested by replacing lipid values in the model above with values for protein.

Results

Tests for critical size

The scatterplot of ED and fork length for juvenile sockeye salmon indicated differences in this relationship among years (Figure 3). The ANCOVA results reveal that fork length is strongly related to ED (p < 0.001). The juvenile sockeye salmon forumla estimates during autumn 2002–2008 varied from a low of 65 million during 2004 to a high of 359 million during 2007 (Table 1). Adult sockeye salmon that survived 2–3 more years in the ocean and returned to Bristol Bay to spawn varied from a low of 33 million for juvenile year 2003 to a high of 59.2 million for juvenile year 2002. The forumla ranged from a low of 14.8% during 2007 to a high of 61.5% during 2004. The highest uncertainty in the forumla was evident during even years and 2007. In addition, the forumla and the forumla values had an odd/even year pattern, where forumla values were evident during even years and forumla values occurred during odd years (Table 1).

Table 1.

Bootstrap estimates of relative abundance (forumla, millions) and MSI forumla with upper and lower confidence bounds (95% confidence intervals, LCI and UCI) for juvenile sockeye salmon collected during autumn 2002–2007 in the eastern Bering Sea and subsequent number of adult sockeye salmon returns to Bristol Bay (millions) 2 and 3 years later. The dash (–) indicates that the LCI was below 0%.

Year forumla
 
Adult returns forumla
 
LCI Est. UCI LCI (%) Est. (%) UCI (%) 
2002 64.2 136.9 209.6 59.2 21.1 46.4 71.7 
2003 98.4 181.6 264.7 33.0 8.9 19.3 29.6 
2004 36.3 65.8 95.4 38.4 31.6 61.5 91.4 
2005 160.8 338.3 515.8 49.2 7.5 15.5 23.6 
2006 27.2 83.4 139.5 38.6 9.6 52.7 95.9 
2007 46.3 359.4 672.6 41.0 – 14.8 36.2 
Year forumla
 
Adult returns forumla
 
LCI Est. UCI LCI (%) Est. (%) UCI (%) 
2002 64.2 136.9 209.6 59.2 21.1 46.4 71.7 
2003 98.4 181.6 264.7 33.0 8.9 19.3 29.6 
2004 36.3 65.8 95.4 38.4 31.6 61.5 91.4 
2005 160.8 338.3 515.8 49.2 7.5 15.5 23.6 
2006 27.2 83.4 139.5 38.6 9.6 52.7 95.9 
2007 46.3 359.4 672.6 41.0 – 14.8 36.2 

Test for critical period

The genetic analysis suggested that more than 55% of ocean age-1 sockeye salmon caught during winter 2009 were from the Bristol Bay region: 34% from eastern Bristol Bay stocks, 21% from western Bristol Bay stocks, and 12% from Alaska Peninsula stocks, where a portion of these stocks migrate through Bristol Bay (Table 2). The remaining 34% were from western and eastern Gulf of Alaska stocks, of which a small amount (6%) was from western Kamchatka, with no apparent contribution from Norton Sound or eastern Kamchatka.

Table 2.

Mean, standard deviation (s.d.), median, and 97.5% confidence intervals for genetic regional group identification of ocean age-1 sockeye salmon collected during winter 2009 in the North Pacific Ocean.

Regional group Mean s.d. 2.5% Median 97.5% 
Norton Sound 0.00 0.00 0.00 0.00 0.00 
Western Bristol Bay 0.21 0.03 0.15 0.21 0.27 
Eastern Bristol Bay 0.34 0.03 0.27 0.34 0.41 
Alaska Peninsula 0.12 0.03 0.07 0.12 0.18 
Eastern Kamchatka 0.00 0.00 0.00 0.00 0.01 
Western Kamchatka 0.06 0.02 0.02 0.06 0.10 
Western GOA 0.18 0.03 0.13 0.18 0.24 
Eastern GOA 0.10 0.02 0.03 0.10 0.14 
Regional group Mean s.d. 2.5% Median 97.5% 
Norton Sound 0.00 0.00 0.00 0.00 0.00 
Western Bristol Bay 0.21 0.03 0.15 0.21 0.27 
Eastern Bristol Bay 0.34 0.03 0.27 0.34 0.41 
Alaska Peninsula 0.12 0.03 0.07 0.12 0.18 
Eastern Kamchatka 0.00 0.00 0.00 0.00 0.01 
Western Kamchatka 0.06 0.02 0.02 0.06 0.10 
Western GOA 0.18 0.03 0.13 0.18 0.24 
Eastern GOA 0.10 0.02 0.03 0.10 0.14 

The length frequency distributions for juvenile and ocean age-1 sockeye salmon indicate the possibility of size-selective mortality on smaller sockeye salmon within the population between autumn 2008 and winter 2009 (Figure 4). On average, the length of ocean age-1 sockeye salmon increased by 53 mm over winter (196.75–249.8 mm), with a decrease in the variance from 1116.1 for juvenile sockeye salmon to 247 for ocean age-1 sockeye salmon. In addition, the slope of the quantile–quantile plot (not illustrated) was <1. These results suggest that size-selective mortality on smaller juvenile Bristol Bay sockeye salmon likely took place between autumn 2008 and winter 2009.

Figure 4.

Fork length frequency for juvenile sockeye salmon caught during autumn 2008 (filled bar) in the eastern Bering Sea and ocean age-1 sockeye salmon caught during winter 2009 (open bar) in the North Pacific Ocean.

Figure 4.

Fork length frequency for juvenile sockeye salmon caught during autumn 2008 (filled bar) in the eastern Bering Sea and ocean age-1 sockeye salmon caught during winter 2009 (open bar) in the North Pacific Ocean.

Lipid in juvenile sockeye salmon collected during autumn 2008 ranged from 1.8 to 13.4 g and increased with size, whereas lipid for ocean age-1 sockeye salmon collected during winter 2009 ranged from 1.4 to 4.4 g, only slightly increasing with size (Figure 5a). The slope of the relationship between lipid and size was steeper for juvenile sockeye salmon caught during autumn 2008 than for ocean age-1 sockeye salmon caught during winter 2009. In fact, δ was negative (LIP = −18.6 + 0.11 × L + 18.7 × D – 0.10 × D × L + ɛ; r2 = 0.80; p < 0.001 for all coefficients), indicating that lipid depletion was higher in larger than in smaller ocean age-1 sockeye salmon collected during winter 2009. However, the slopes of the regression lines for the relationship between the amount of protein (structure) and size for juvenile and ocean age-1 sockeye salmon do not appear to differ (Figure 5b). When protein was related to length in the regression analysis that included dummy variable for season, δ was positive and moderately significant (δ = 0.09; p = 0.06). In this case, the positive δ suggests that larger sockeye salmon are gaining protein faster than smaller sockeye salmon do during winter.

Figure 5.

The relationship between fork length (mm) and (a) lipid (g) and (b) protein (g) for juvenile (squares) and ocean age-1 (diamonds) sockeye salmon collected during autumn 2008 in the eastern Bering Sea and winter 2009 in the North Pacific Ocean, respectively.

Figure 5.

The relationship between fork length (mm) and (a) lipid (g) and (b) protein (g) for juvenile (squares) and ocean age-1 (diamonds) sockeye salmon collected during autumn 2008 in the eastern Bering Sea and winter 2009 in the North Pacific Ocean, respectively.

Test for differences in energetic status among years

The ANCOVA indicated a significant difference in ED among years (p < 0.001). The unadjusted and the adjusted mean EDs are displayed in Table 3. The multiple comparison test indicated that adjusted mean ED during 2003 was significantly lower than all other years (p < 0.05). Adjusted mean ED during 2004 was significantly higher than 2005 (p < 0.001) and 2007 (p < 0.01). In addition, adjusted mean ED for 2005 was significantly lower than 2006 (p = 0.05).

Table 3.

Energy density (ED; kJ g−1 wet weight) mean and standard error (s.e.), ED adjusted mean (Adj.; from ANCOVA) and standard error (s.e.), and sample size (n) for juvenile sockeye salmon collected during autumn 2003–2008 in the eastern Bering Sea.

Year n Mean s.e. Adj. mean Adj. s.e. 
2003 139 5.22 0.06 4.98 0.05 
2004 118 5.31 0.05 5.49 0.04 
2005 89 5.04 0.05 5.18 0.05 
2006 35 5.39 0.10 5.43 0.07 
2007 64 5.04 0.09 5.23 0.05 
2008 23 5.56 0.10 5.33 0.09 
Year n Mean s.e. Adj. mean Adj. s.e. 
2003 139 5.22 0.06 4.98 0.05 
2004 118 5.31 0.05 5.49 0.04 
2005 89 5.04 0.05 5.18 0.05 
2006 35 5.39 0.10 5.43 0.07 
2007 64 5.04 0.09 5.23 0.05 
2008 23 5.56 0.10 5.33 0.09 

A positive relationship (r2 = 0.69) was found between forumla and adjusted mean ED (Figure 6), indicating that juvenile sockeye salmon that contained higher energy density before winter had higher forumla. It is noteworthy that the odd years had the lowest ED and forumla values, whereas the highest ED and forumla values were evident during even years. Hence, significantly higher and lower ED values are mixed within years with anomalously warm and cold spring and summer sea temperatures.

Figure 6.

The relationship between MSI and energy density (ED; kJ g−1 wet weight) for juvenile salmon caught during autumn of 2003–2007 in the eastern Bering Sea.

Figure 6.

The relationship between MSI and energy density (ED; kJ g−1 wet weight) for juvenile salmon caught during autumn of 2003–2007 in the eastern Bering Sea.

Discussion

Our results provide evidence that marine mortality of juvenile salmon after their first summer at sea can be large and that the first winter at sea may be the critical period for survival. This is consistent with other observations of size-selective mortality during winter for juvenile salmon (Beamish et al., 2004; Moss et al., 2005). The comparison of sockeye salmon size between autumn 2008 and winter 2009 revealed that size-selective mortality on smaller individuals is occurring for Bristol Bay sockeye salmon in between the two periods. However, the seasonal energy budgets demonstrate that larger ocean age-1 sockeye salmon were losing more lipid than smaller conspecifics, but that larger ocean age-1 fish were gaining protein.

Can the seasonal energy budgets provide insights into potential mechanisms for size-selective mortality of Bristol Bay sockeye salmon during winter? Food resources are thought to decline rapidly during winter (Foy and Paul, 1999); therefore, it is believed that fish that do not have sufficient energy reserves either die or become more vulnerable to predators (Beamish and Mahnken, 2001). An animal moves through three phases to get from extended natural fasting to the point of starvation (Castellini and Rea, 1992). During the first phase, animals burn glucose and begin to mobilize stored lipids as the body starts to switch to fat oxidation and reduce protein catabolism. The second phase involves increased oxidation of lipids and the partial sparing of protein. Animals in this phase are still considered fit enough to survive once prey items are again available. Phase 3, or terminal starvation, sets in when 30–50% of the body protein has been used. Cardiac muscle tissue is one of the first sources of protein utilized and other organs become compromised too. This phase is often associated with death of the animal, because of extreme reduction in fitness.

The observed changes in seasonal energy budgets indicate that Bristol Bay sockeye salmon may be in phase 2 during winter. We found that larger ocean age-1 sockeye salmon are depleting lipids. The nearly constant ED for all sizes of age-1 sockeye salmon suggests that larger ocean age-1 sockeye salmon contain enough lipids to potentially fast longer, whereas smaller sockeye salmon must feed to maintain enough energy (prolong phase 2) to keep from moving into phase 3. Fasting may be a strategy to avoid predation, whereas feeding during winter may increase predation risk. This observation could explain the size-selective mortality of smaller sockeye salmon between autumn 2008 and winter 2009. However, sockeye salmon of all size ranges were actively feeding during late winter in the North Pacific Ocean (Starovoytov et al., 2009), suggesting that this strategy may not explain changing seasonal energy budgets and size-selective mortality for these fish.

Alternatively, the energy depletion of larger sockeye salmon may be because of energetic costs associated with swimming. Between autumn 2008 and winter 2009, juvenile Bristol Bay sockeye salmon migrated from the eastern Bering Sea shelf to the North Pacific, south of the Aleutian Islands. The energetic cost associated with swimming and foraging may represent a significant portion of the energy budget of fish; it has been established to be higher in larger fish (Boisclair and Rasmussen, 1996; Rowan and Rasmussen, 1996; Trudel and Rasmussen, 2001; Pazzia et al., 2002). Therefore, the differences in lipid signatures between juvenile (autumn) and ocean age-1 (winter) sockeye salmon could also be because of higher weight-specific metabolic rates in larger fish that are actively swimming and foraging. In this case, size-selective mortality would be reduced for larger fish that can swim faster away from predators.

Lastly, the changes in size and energetic status between juvenile and ocean age-1 sockeye salmon may be biased, because of incomplete sampling of the ocean age-1 Bristol Bay sockeye salmon population during winter. We found that more than 55% of the ocean age-1 sockeye salmon sampled during winter were from Bristol Bay. In addition, the winter migration models for ocean age-1+ western Alaska sockeye salmon indicate that these fish would be distributed south of the Aleutian Islands (Burgner, 1991; Myers et al., 2007) within the survey area sampled by the RV “TINRO” during February to March. These data and the fact that size-selective mortality for other Pacific salmon happens during the first winter at sea (Beamish et al., 2004; Moss et al., 2005) suggest that the chance of bias in our analysis is minimal.

Energy density of juvenile sockeye salmon is likely a function of prey quality and quantity and sea temperature. In the eastern Bering Sea, spring and summer sea surface temperatures were anomalously warm during 2002–2005 and anomalously cold during 2006–2008 (Hunt et al., 2011). Previous studies utilizing the same juvenile sockeye salmon used in our study have established that during years with warm sea temperatures, juvenile sockeye salmon were feeding primarily on age-0 walleye pollock, whereas during years with cold sea temperatures, they were primarily feeding on euphausiids, copepods, and larval and juvenile sand lance (Ammodytes hexapterus; Farley et al., 2009). Juvenile sockeye salmon growth-rate potential models that included these prey items suggested that their growth-rate potential was higher during years with warm spring and summer sea temperatures on the eastern Being Sea shelf (Farley and Trudel, 2009). These and other studies (Farley et al., 2007b, c) have revealed that juvenile Bristol Bay sockeye salmon are fitter and have higher marine stage survival during years with warm sea temperatures on the eastern Bering Sea shelf.

We found a strong relationship between juvenile sockeye salmon ED and MSI, suggesting that juvenile sockeye salmon marine survival is linked to their energetic status before winter. Presumably, years with higher ED would prevail during periods of warm spring and summer sea temperatures on the eastern Bering Sea shelf. However, we found that juvenile sockeye salmon ED varied significantly among years, with the highest ED during years with warm and cold spring and summer sea surface temperatures. We note that reduced ED and forumla were prevalent during years with highest forumla for juvenile sockeye salmon, indicating the possibility of density-dependent processes affecting energetic status of juvenile sockeye salmon during summer and autumn.

We also noted that the forumla, forumla and ED values for juvenile sockeye salmon displayed an odd/even year pattern, similar to that found for immature (ocean age-2+) Bristol Bay sockeye salmon stocks, whose density-dependent growth and survival was linked to abundance of Asian pink salmon (Ruggerone et al., 2003). Although juvenile pink salmon are found in the eastern Bering Sea during odd years, their relative abundance is low and their early marine distribution does not tend to overlap that of juvenile sockeye salmon (Farley et al., 2009). Hence, the odd/even year pattern in these values is apparently not linked to juvenile pink salmon abundance.

Concluding remarks

The new data presented in this paper demonstrate size-selective mortality of Bristol Bay sockeye salmon between autumn and their first winter at sea. Differences in the seasonal energetic signatures for lipid and protein suggest that these fish are not starving, but instead the larger fish apparently are utilizing energy stores to minimize predation. In addition, the energetic status of juvenile sockeye salmon was also strongly related to MSIs, and years with lower energetic status apparently are a function of density-dependent processes, because of high juvenile sockeye salmon abundance. Even so, adult returns to Bristol Bay have been relatively high during the past 5 years, suggesting that although mortality was high during some years, there were enough juvenile sockeye salmon to sustain healthy runs.

It is generally agreed that the Bering Sea will continue to warm up (Christensen et al., 2007); consequently, the expectation would be for continued healthy returns of sockeye salmon to Bristol Bay watersheds (Farley et al., 2007b, c). Many juvenile salmon and, in particular, juvenile sockeye salmon relied heavily on age-0 walleye pollock for prey during years with anomalously warm sea temperatures (Farley et al., 2009). However, there is new evidence that extended periods of warming may reduce the availability of lipid-rich crustacean zooplankton, negatively affecting walleye pollock recruitment (Hunt et al., 2011). This suggests that continued high sea temperatures could reduce the availability of age-0 pollock, causing juvenile sockeye salmon to seek other, potentially lipid-poor, prey items. In addition, a previous analysis suggested that if summer sea temperatures increased by 5°C, the largest decrease in growth-rate potential for juvenile Bristol Bay sockeye salmon would occur during years where observed sea temperatures (2000–2006) were already anomalously warm (Farley and Trudel, 2009). Hence, under a climate-warming scenario, we hypothesize that sustained increases in sea temperatures higher than those observed during 2002–2005 may affect the energetic status and growth-rate potential for juvenile Bristol Bay sockeye salmon, potentially resulting in increased overwinter mortality.

Acknowledgments

The research reported here was supported by NOAA, Alaska Fisheries Science Center, the Pacific Scientific Research Fisheries Center, TINRO Center, the Bering Sea Fisherman's Association, and the North Pacific Research Board's BEST/BSIERP programmes. The findings and conclusions in the paper are those of the author(s) and do not necessarily represent the views of the National Marine Fisheries Service.

References

Andrews
A. G.
Farley
E. V.
Moss
J. H.
Murphy
J. M.
Husoe
E. F.
Farley
E.
Azumaya
T.
Beamish
R.
Koval
M.
Myers
K.
Seong
K. B.
Urawa
S.
Energy density and length of juvenile pink salmon (Oncorhynchus gorbuscha) in the eastern Bering Sea from 2004 to 2007: a period of relatively warm and cool sea surface temperatures
Climate Change, Production Trends, and Carrying Capacity of Pacific Salmon in the Bering Sea and Adjacent Waters
 , 
2009
Vancouver, BC
North Pacific Anadromous Fish Commission, Bulletin 5
(pg. 
183
-
189
)
Association of Official Analytical Chemists
Official methods of analysis
 , 
2002
Washington, DC
Association of Official Analytical Chemists
Beamish
R. J.
Mahnken
C.
A critical size and period hypothesis to explain natural regulation of salmon abundance and the linkage to climate and climate change
Progress in Oceanography
 , 
2001
, vol. 
49
 (pg. 
423
-
437
)
Beamish
R. J.
Mahnken
C.
Neville
C. M.
Evidence that reduced early marine growth is associated with lower marine survival of coho salmon
Transactions of the American Fisheries Society
 , 
2004
, vol. 
133
 (pg. 
26
-
33
)
Boisclair
D.
Rasmussen
J. B.
Empirical analysis of environmental variables on perch growth, consumption and activity rates
Annales Zoologici Fennici
 , 
1996
, vol. 
33
 (pg. 
507
-
515
)
Burgner
R. L.
Groot
C.
Margolis
L.
Life history of sockeye salmon (Oncorhynchus nerka)
Pacific Salmon Life Histories
 , 
1991
Vancouver, BC
University of British Columbia Press
(pg. 
3
-
117
)
Castellini
M. A.
Rea
L. D.
The biochemistry of natural fasting at its limits
Experientia
 , 
1992
, vol. 
48
 (pg. 
575
-
582
)
Christensen
J. H.
Hewitson
B.
Busuioc
A.
Chen
A.
Gao
X.
Held
I.
Jones
R.
, et al.  . 
Solomon
S.
Qin
D.
Manning
M.
Chen
Z.
Marquis
M.
Averyt
K. B.
Tignor
M.
, et al.  . 
Regional climate projections
Climate Change 2007: the Physical Science Basis. Contribution of Working Group I to the Fourth Assessment Report of the Intergovernmental Panel on Climate Change
 , 
2007
Cambridge, UK and New York, NY, USA
Cambridge University Press
Christie
W. W.
Lipid Analysis: Isolation, Separation, Identification and Structural Analysis of Lipids
 , 
1982
2nd edn
New York, NY
Pergamon Press
Efron
B.
The jackknife, the bootstrap, and other resampling plans
 , 
1982
Philadelphia, PA
SIAM, CBMS-NSF Regional Conference Series in Applied Mathematics, 38
Farley
E. V.
Moss
J. H.
Farley
E.
Azumaya
T.
Beamish
R.
Koval
M.
Myers
K.
Seong
K. B.
Urawa
S.
Growth rate potential of juvenile chum salmon on the eastern Bering Sea shelf: an assessment of salmon carrying capacity
Climate Change, Production Trends, and Carrying Capacity of Pacific Salmon in the Bering Sea and Adjacent Waters
 , 
2009
Vancouver, BC
North Pacific Anadromous Fish Commission, Bulletin 5
(pg. 
265
-
277
)
Farley
E. V.
Moss
J. H.
Beamish
R. J.
Beamish
R.
Radchenko
V.
A review of the critical size, critical period hypothesis for juvenile Pacific salmon
Status of Pacific Salmon and Their Role in North Pacific Marine Ecosystems
 , 
2007
North Pacific Anadromous Fish Commission, Bulletin 4
(pg. 
311
-
317
)
Farley
E. V.
Murphy
J. M.
Adkison
M.
Eisner
L.
Juvenile sockeye salmon distribution, size, condition, and diet during years with warm and cool spring sea temperatures along the eastern Bering Sea shelf
Journal of Fish Biology
 , 
2007
, vol. 
71
 (pg. 
1145
-
1158
)
Farley
E. V.
Murphy
J. M.
Adkison
M. D.
Eisner
L. B.
Helle
J. H.
Moss
J. H.
Nielsen
J.
Early marine growth in relation to marine-stage survival rates for Alaska sockeye salmon (Oncorhynchus nerka)
Fishery Bulletin US
 , 
2007
, vol. 
105
 (pg. 
121
-
130
)
Farley
E. V.
Murphy
J. M.
Moss
J.
Feldmann
A.
Eisner
L.
Krueger
C. C.
Zimmerman
C. E.
Marine ecology of western Alaska juvenile salmon
Pacific Salmon: Ecology and Management of Western Alaska's Populations
 , 
2009
American Fisheries Society Symposium, 70
(pg. 
307
-
329
)
Farley
E. V.
Murphy
J. M.
Wing
B. W.
Moss
J. H.
Middleton
A.
Distribution, migration pathways, and size of western Alaska juvenile salmon along the eastern Bering Sea shelf
Alaska Fishery Research Bulletin
 , 
2005
, vol. 
11
 (pg. 
15
-
26
)
Farley
E. V.
Trudel
M.
Growth rate potential of juvenile sockeye salmon in warmer and cooler years on the eastern Bering Sea shelf
 , 
2009
Journal of Marine Biology, 2009
pg. 
10 pp
 
Foy
R. J.
Paul
A. J.
Winter feeding and changes in somatic energy content of age-0 Pacific herring in Prince William Sound, Alaska
Transactions of the American Fisheries Society
 , 
1999
, vol. 
128
 (pg. 
1193
-
1200
)
Grebmeier
J. M.
Overland
J. E.
Moore
S. E.
Farley
E. V.
Carmack
E. C.
Cooper
L. W.
Frey
K. E.
, et al.  . 
A major ecosystem shift in the northern Bering Sea
Science
 , 
2006
, vol. 
311
 (pg. 
1461
-
1464
)
Gunderson
D. R.
Surveys of Fisheries Resources
 , 
1993
New York, NY
Wiley & Sons, Inc.
pg. 
248 pp
 
Guthrie
C. M.
Masuda
M.
Hguyen
H.
Cheng
W.
Wilmot
R. L.
Guyon
J. R.
Northern boundary area sockeye salmon genetic stock identification for year 2006 and 2007 District 101 gillnet and District 104 purse seine fisheries
 , 
2010
Juneau, AK
Report to the Pacific Salmon Commission. NOAA, NMFS, Auke Bay Laboratories
pg. 
22 pp
 
Habicht
C.
Seeb
L. W.
Myers
K. W.
Farley
E. V.
Seeb
J. E.
Summer–fall distribution of stocks of immature sockeye salmon in the Bering Sea as revealed by single-nucleotide polymorphisms
Transactions of the American Fisheries Society
 , 
2010
, vol. 
139
 (pg. 
1171
-
1191
)
Habicht
C.
Templin
W. D.
Willette
T. M. F.
Raborn
S. W.
Seeb
L. W.
Post-season stock composition analysis of upper Cook Inlet sockeye salmon harvest, 2005–2007
 , 
2007
Alaska Department of Fish and Game Manuscript, Number 07-07
Hartt
A. C.
McNeil
W. J.
Himsworth
D. C.
Juvenile salmonids in the oceanic ecosystem: the critical first summer
Salmonid Ecosystems of the North Pacific
 , 
1980
Corvallis, OR
Oregon State University Press
(pg. 
25
-
57
)
Hunt
G. L.
Coyle
K. O.
Eisner
L. B.
Farley
E. V.
Heintz
R. A.
Mueter
F.
Napp
J. M.
, et al.  . 
Climate impacts on eastern Bering Sea food webs: a synthesis of new data and an assessment of the Oscillating Control Hypothesis
ICES Journal of Marine Science
 , 
2011
, vol. 
68
 (pg. 
1230
-
1243
)
Hunt
G. L.
Stabeno
P. J.
Climate change and the control of energy flow in the southeastern Bering Sea
Progress in Oceanography
 , 
2002
, vol. 
55
 (pg. 
5
-
22
)
Hunt
G. L.
Stabeno
P.
Walters
G.
Sinclair
E.
Brodeur
R. D.
Napp
J. M.
Bond
N. A.
Climate change and control of the southeastern Bering Sea pelagic ecosystem
Deep Sea Research II
 , 
2002
, vol. 
49
 (pg. 
5821
-
5853
)
Moss
J. H.
Beauchamp
D. A.
Cross
A. D.
Myers
K.
Farley
E. V.
Murphy
J. M.
Helle
J. H.
Higher marine survival associated with faster growth for pink salmon (Oncorhynchus gorbuscha)
Transactions of the American Fisheries Society
 , 
2005
, vol. 
134
 (pg. 
1313
-
1322
)
Myers
K. W.
Klovach
N. V.
Gritsenko
O. F.
Urawa
S.
Royer
T. C.
Beamish
R.
Radchenko
V.
Stock-specific distributions of Asian and North American salmon in the open ocean, interannual changes, and oceanographic conditions
Status of Pacific Salmon and their Role in North Pacific Marine Ecosystems
 , 
2007
North Pacific Anadromous Fish Commission, Bulletin 4
(pg. 
159
-
177
)
Parker
R. R.
Marine mortality schedules of pink salmon of the Bella Coola River, central British Columbia
Journal of the Fisheries Research Board of Canada
 , 
1968
, vol. 
25
 (pg. 
757
-
794
)
Pazzia
I.
Trudel
M.
Ridgway
M.
Rasmussen
J. B.
Influence of food web structure on the growth and bioenergetics of lake trout (Salvelinus namaycush)
Canadian Journal of Fisheries and Aquatic Sciences
 , 
2002
, vol. 
59
 (pg. 
1593
-
1605
)
Pearcy
W. G.
Ocean Ecology of the North Pacific Salmonids
 , 
1992
Seattle, WA
University of Washington Press
pg. 
179 pp
 
Pella
J.
Masuda
M.
Bayesian methods for analysis of stock mixtures from genetic characters
Fishery Bulletin US
 , 
2001
, vol. 
99
 (pg. 
151
-
167
)
Post
J. R.
Evans
D. O.
Size-dependent overwinter mortality of young-of-the year yellow perch (Perca flavescens): laboratory, in situ enclosure, and field experiments
Canadian Journal of Fisheries and Aquatic Sciences
 , 
1989
, vol. 
46
 (pg. 
1958
-
1968
)
Rowan
D. J.
Rasmussen
J. B.
Measuring the bioenergetic cost of fish activity using a globally dispersed radiotracer (137Cs)
Canadian Journal of Fisheries and Aquatic Sciences
 , 
1996
, vol. 
53
 (pg. 
734
-
745
)
Ruggerone
G. T.
Zimmermann
M.
Myers
K. W.
Nielsen
J. L.
Rogers
D. E.
Competition between Asian pink salmon (Oncorhynchus gorbuscha) and Alaska sockeye salmon (O. nerka) in the North Pacific Ocean
Fisheries Oceanography
 , 
2003
, vol. 
12
 (pg. 
209
-
219
)
Starovoytov
A. N.
Naydenko
S. V.
Kurenkova
E. V.
Ocheretyany
M. A.
Vanin
N. S.
Composition and structure of epipelagic nekton communities in the central and western parts of Subarctic frontal zone in winter and spring 2009 (result of 2009 research cruise of R/V TINRO)
 , 
2009
North Pacific Anadromous Fish Commission, 1188
pg. 
29 pp
 
Trudel
M.
Rasmussen
J. B.
Predicting mercury concentration in fish using a mass balance model
Ecological Applications
 , 
2001
, vol. 
11
 (pg. 
517
-
529
)
Trudel
M.
Tucker
S.
Morris
J. F. T.
Higgs
D. A.
Welch
D. W.
Indicators of energetic status in juvenile coho salmon and Chinook salmon
North American Journal of Fisheries Management
 , 
2005
, vol. 
25
 (pg. 
374
-
390
)
Vollenweider
J. J.
Variability in Steller sea lion (Eumetopias jubatus) prey quality in southeastern Alaska
 , 
2005
Fairbanks, AK
MSc thesis, Juneau Center, School of Fisheries and Ocean Sciences, University of Alaska
Willette
T. M.
Cooney
R. T.
Hyer
K.
Predator foraging mode shifts affecting mortality of juvenile fish during the subarctic spring bloom
Canadian Journal of Fisheries and Aquatic Sciences
 , 
1999
, vol. 
56
 (pg. 
64
-
376
)
Wyllie-Echeverria
T.
Wooster
W. S.
Year-to-year variations in Bering Sea ice cover and some consequences for fish distributions
Fisheries Oceanography
 , 
1998
, vol. 
7
 (pg. 
159
-
170
)