OUP user menu

Seasonal and Spatial Differences in Diet in the Western Stock of Steller Sea Lions (Eumetopias Jubatus)

E. H. Sinclair, T. K. Zeppelin
DOI: http://dx.doi.org/10.1644/1545-1542(2002)083<0973:SASDID>2.0.CO;2 973-990 First published online: 27 November 2002


We identified prey remains from 3,762 scats (feces) of Steller sea lions (Eumetopias jubatus). Scats were collected from 1990–1998 on island sites across most of the range of the United States western stock of the species. Walleye pollock (Theragra chalcogramma) and Atka mackerel (Pleurogrammus monopterygius) were the 2 most common species of prey, followed by salmonids (Oncorhynchus) and Pacific cod (Gadus macrocephalus). An additional 16 species of fish and unidentified cephalopods were considered primary in the diet, either because they occurred in >5% of scats collected across the range in winter and summer or because they consistently occurred among the top 3 prey items in particular islands or island groups. Capelin (Mallotus villosus) occurred at very low frequencies despite their predominance in the diet of Steller sea lions before the 1980s. Regions of diet similarity suggest area-specific foraging strategies, with strong seasonal patterns in consumption of most species of prey. Patterns in prey consumption and characteristics of prey indicate that Steller sea lions target prey that are densely schooled in spawning or migratory aggregations at the continental shelf or along oceanographic boundary zones. We suggest that regional diet patterns among the western stock reflect regional foraging strategies of females learned at islands near the natal rookery site.

Key words
  • diet
  • Eumetopias jubatus
  • food
  • foraging
  • otariid
  • prey
  • sea lions
  • Steller sea lions

Steller sea lions (Eumetopias jubatus) have declined precipitously throughout the western portion of their range across the North Pacific Rim (Kenyon and Rice 1961; Loughlin et al. 1992). In 1997, the newly defined western stock of Steller sea lions (Bickham et al. 1996; Loughlin 1997) was listed as endangered under the U.S. Endangered Species Act. It is not known whether the decline in Steller sea lions is due to decreased reproduction or increased mortality. As with other North Pacific apex predators, the Steller sea lion decline is thought to be associated with diet or foraging restrictions due to natural and anthropogenic influences (National Marine Fisheries Service 2001; National Research Council 1996). Commercial fisheries have been implicated as playing a role in jeopardizing the recovery of Steller sea lions through resource alteration and competition (National Marine Fisheries Service 2001). These recent events underscore the importance of understanding Steller sea lion diet and foraging patterns.

Steller sea lions breed, bear, and nurse their young on remote island sites called rookeries, which are occupied during summer months by territorial adult males, adult females, and pups of the year (Kenyon and Rice 1961). Yearlings and older juveniles are more transient than breeding adults during summer and are present on rookeries to a much lesser extent. After breakup of rookeries in August, adult males disperse from the islands, whereas adult females and their pups remain or move to other islands referred to as haul-outs. After September, haul-outs are occupied by all ages and both sexes. Some haul-outs may have served as rookeries during the breeding season. Our objectives were to evaluate spatial and seasonal trends in the diet of Steller sea lions based on analysis of scat (fecal) material collected throughout the 1990s on rookeries and haul-outs across the range of the western stock.

Materials and Methods

Field and laboratory

Steller sea lion scats were collected from 1990–1998 across most of the United States range of the western stock. We collected 2,340 samples from 31 sites (rookeries) in summer (May–September) and 1,843 samples from 31 sites (haul-outs) in winter (December–April; Fig. 1; Appendix I). Some sites were visited during both winter and summer, but typically, sites were visited no more than once per season (Appendix I).

Fig. 1.

Scat collection sites on island rookeries and haul-outs of the western stock of Steller sea lions, 1990–1998. Site names are given in Appendix I

Only scats considered as whole samples from an individual animal were collected. Scats were initially stored dry in plastic bags and then frozen before processing in the laboratory. Each bagged sample was thawed separately in soapy water before being either hand washed through nested sieves of 4.8-, 1.4-, 0.7- and 0.5-mm mesh or passed from an elutriator (Bigg and Olesiuk 1990) into nested sieves of 0.7- and 0.5-mm mesh. All prey remains (e.g., lenses, scales, bones including otoliths, and cephalopod rostra or beaks) were recovered from the sieves and stored for analysis. Beaks and lenses were stored in 50% isopropyl alcohol. Scales and bones were stored dry in glass vials. This diet analysis is based on beaks, bones, scales, and otoliths identified to the lowest possible taxon by using a reference collection.

We assume that collections made in summer primarily represent the diet of adult females because adult males usually fast during the breeding season. But juveniles of both sexes haul out on rookeries during the summer and are undoubtedly represented in the data. Winter collection sites are all considered haul-outs. Scat collections during winter presumably represent a greater cross-section of ages and sexes than do collections from summer.


Relative importance of each prey species (or prey family) was based on the frequency with which it occurred in scats, and each scat was treated as an independent sample. Percentage frequency of occurrence (FO) of individual prey species was calculated by dividing number of scats in which a prey item occurred by total number of scats that contained identifiable prey remains. All samples were pooled for FO calculations at sites with multiple collections within a year and in multiple years. This technique allows interpretation of trends in species composition of prey and provides an index of the proportion of the predator population consuming a particular prey item. FO provides an estimate of the number of sea lions that consumed each species of prey, but it does not provide an estimate of the number of each species of prey consumed. The number of prey consumed is not estimated in this study because the structure currently used to count individual prey (otoliths) is often absent in Steller sea lion scats. But a wide range of marine mammal diet studies (Antonelis et al. 1997; Kajimura 1984; Perrin et al. 1973; Sinclair 1992, 1994; Sinclair et al. 1994, 1996; Walker 1996) indicate that with large sample sizes, prey rank is equally represented by either FO or prey number.

To determine site-specific differences in Steller sea lion diet, FO values were compared on a site-by-site basis for summer and winter, for all collections and years combined. Collection sites were then grouped to describe diet patterns in an area and to make seasonal comparisons (because not all sites were visited in both seasons). Sites were grouped based on similarities between the FO of prey species, using principal components analysis (PCA) and an agglomerative hierarchical cluster analysis (Ludwig and Reynolds 1988).

Principal components analysis was first used to reduce data into representative species of prey that accounted for most of the variance in the data set. PCA was calculated on a correlation matrix using species of prey as variables and 31 sites as observations. To minimize zeros in the analysis, only prey that occurred in >5% of the scats across all sites and seasons were included as variables in the PCA. Cluster analysis was then conducted on the PCA factors using squared Euclidian distance (Ludwig and Reynolds 1988) as a measure of similarity between sites and Ward's (1963) method to compare cluster distances. Spatial breaks in the clusters were determined visually from output dendograms of the data from summer using the software package SPSS 1999 (Statistical Packages for the Social Studies Incorporated, 233 South Wacker Drive, Chicago, Illinois) and used to group sites into regions. Data from winter were laid over these same spatial breaks for comparison with diets in summer.

Chi-square analysis was conducted using the software package Splus 2000 (Insightful Corporation, 1700 Westlake Avenue North, Suite 500, Seattle, Washington) to test a hypothesis of no difference in the proportion of scats containing a particular prey item (FO) within each of the regions (island groupings as defined by PCA and cluster analysis) between winter and summer. It was assumed that all species of prey represented within a scat were consumed independent of each other, and analyses were limited to prey items occurring in >5% of scats across all regions and seasons. A 2 × 2 contingency table was used to compare the proportion of scats containing a particular prey item in summer and winter. Chi-square statistics were not calculated for species of prey with any cell count (observed value) ≤5 (Ramsey and Schafer 1996). Significance was determined at the 0.01 level for the contingency analysis.

Diet diversity was calculated for each region using Shannon's index of diversity, H, where pi is the proportion of the ith family in the sample (Ludwig and Reynolds 1988): Embedded Image Prey identified to family rather than species were used for the diet diversity analysis to include all data on an equivalent taxonomic level. Exceptions to this were flatfish and cephalopods, which we were unable to categorize into family groups.


Prey identification and island comparisons

A total of 3,762 scats contained identifiable prey items in summer (n = 2,102) and winter (n = 1,660). FO values combined across years, seasons, and sites depict walleye pollock (Theragra chalcogramma; FO = 46.4%) and Atka mackerel (Pleurogrammus monopterygius; FO = 39.6%) as the 2 most common species of prey, followed by Pacific salmon (Oncorhynchus; FO = 20.4%) and Pacific cod (Gadus macrocephalus; FO = 16.1%; Table 1). Based on comparison with prey reference collections, remains recovered from these primary prey were predominantly from late-stage juvenile and adult-sized fish. In addition to the dominant species, those occurring at frequencies ≥5% that were included in PCA and cluster analysis were arrowtooth flounder (Atheresthes stomias; FO = 7.4%), Pacific herring (Clupea pallasi; FO = 6.9%), Pacific sand lance (Ammodytes hexapterus; FO = 6.3%), Irish lord (Hemilepidotus; FO = 8.3%), and cephalopods (squid and octopus; FO = 8.8%; Table 1).

View this table:
Table 1.
View this table:
Table 1.
View this table:
Table 1.
View this table:
Table 1.

In general, collections from island rookeries in close proximity to each other had similar prey matrices (same species of prey occurring at similar frequencies). But interisland comparisons of diet suggest that some species of prey had a low FO when averaged throughout the year and across sites but high FO values within a particular season on particular islands (Appendix I). Larger sample sizes are needed to draw conclusions regarding the consistency of interisland patterns in prey consumption.

Principal components analysis and cluster analysis

Most of the variance (67.7%) in the diet during summer was explained by 2 principal components. The 1st component accounted for 43.6% of the variance. The 2nd component accounted for 23.1% of the variance. The remaining components accounted for 9.5% of the variance or less and were not considered useful in determining regional divisions based on prey. Principal components indicate that 3 related groups of prey occur based on similar loading scores (Fig. 2). Likewise, 3 spatial breaks were evident from the cluster analysis, with the exception of 1 outlier, a Bering Sea rookery (Sea Lion Rock near Amak Island). Cluster 1 had negative values for the 1st principal component and generally positive values for the 2nd principal component, and these sites were characterized by relatively high occurrences of Atka mackerel and cephalopods. Cluster 2 had positive values for the 1st principal component and negative values for the 2nd principal component. These sites were characterized by relatively high values of pollock, salmon, and arrowtooth flounder. Cluster 3 had positive values for both principal components. Although pollock, salmon, and Atka mackerel occurred within cluster 3, the relatively high occurrence of sand lance, herring, Pacific cod, and Irish lord differentiated cluster 3 from the other clusters.

Fig. 2.

Plot of principal component 1 against principal component 2 representing species of prey that account for most of the variance in diet of Steller sea lions across the study area. Prey species are designated by the following abbreviations: arr = arrowtooth flounder; sal = salmon; pol = walleye pollock; her = Pacific herring; sln = Pacific sand lance; ild = Irish lord; pcd = Pacific cod; ceph = cephalopod; atka = Atka mackerel

The diet in winter had 3 principal components that accounted for 68% of the variance. Four distinct clusters were very similar to spatial patterns in the summer diet (Fig. 3). In both summer and winter, sites in each cluster were generally separated spatially (Fig. 3). One site in cluster 2 (Ogchul Island) fell out far west of the other sites in summer, but sample sizes were small. There was also some spatial overlap between clusters 2 and 3 in winter and in summer. For the purposes of this analysis, 4 regions were defined based on the spatial distribution of the cluster assignments for each site. In summer, diets were clearly distinctive in regions 1, 3, and 4. In region 2, diet overlap occurred between clusters 2 and 3 (Fig. 3). Regional divisions are representative of both summer and winter prey consumption, except that in winter there is more extensive overlap of cluster 3 into region 2.

Fig. 3.

Regional variation in patterns in the diet of Steller sea lions as depicted by principal components analysis and hierarchical cluster analysis. Lines depict the boundaries between each of the 4 regions (REG)

Regional and seasonal comparisons

Patterns in prey consumption by region (island group) illustrate changing foraging patterns from east to west (Table 1; Fig. 3). In general, there was a shift in regional diet dominance from pollock to Atka mackerel and cephalopods moving west from region 3 to region 4. The general profile of regions 1, 2, and 3 was similar in that they were all most strongly defined by the relatively high FO of walleye pollock in the diet during both winter (region 1, 56%; region 2, 86%; region 3, 59%) and summer (region 1, 64%; region 2, 80%; region 3, 54%), and high FO values for salmon in summer (region 1, 41%; region 2, 45%; region 3, 35%). Region 1 differs from all others in that the diet there was characterized by arrowtooth flounder in both winter (21%) and summer (35%). The FO values of Irish lord, Pacific sand lance, and Pacific herring were also generally highest in regions 1 through 3. Region 2 had the highest FO levels of Pacific cod consumption in both winter (36%) and summer (11%) compared with all other regions. Region 4 is generally distinguished by low frequencies of all prey except cephalopods and Atka mackerel.

Several trends were apparent when examining prey FO seasonally within regions (Fig. 4). Pollock FO was not significantly different between seasons in regions 1, 2, and 3 and not statistically comparable due to low values in region 4. Cephalopod FO was not significantly different between seasons in any region. Otherwise, the seasonal shifts in prey among regions were significant (Table 2). Pacific cod occurrence was higher in winter in every region. Salmon was lower during winter in regions 1–3 and higher in region 4. The FO of arrowtooth flounder was lower in winter in region 1, even though the species was most abundant in scats from this area and well represented the year round. Atka mackerel was lower in the diet in winter in region 4, where it was the most common prey the year round. Herring was higher in the diet in winter in region 1 and lower in regions 2 and 3. Pacific sand lance FO was higher during winter in region 1 and lower in region 2. Irish lord FO was higher in winter than in summer in regions 1, 3, and 4. Although sandfish and snailfish occur rarely during summer and are not included in chi-square analysis, they had relatively high occurrences during winter in most regions.

View this table:
Table 2.
Fig. 4.

Seasonal variation in the diet of the western stock of Steller sea lions by region. Prey species abbreviated as in Fig. 2

Except in region 2, diversity values were higher in winter than in summer. During summer, diet diversity was highest in region 2 (H = 2.1) and region 3 (H = 2.0) and lowest in region 4 (H = 1.5), where the diet was made up almost entirely of Atka mackerel and cephalopods. In winter, diversity was highest in region 1 (H = 2.1) and region 3 (H = 2.1) and lowest in region 2 (H = 1.8).


Seasonal and regional patterns in the diet of Steller sea lions are strongly defined. The species of prey consumed by Steller sea lions in summer, when the sampled population is predominantly adult females, varies from that in winter, when the sampled population is presumably a greater mix of ages and possibly sexes. But the boundaries that define regions of diet shift only slightly between winter and summer. The static regional definition in diet together with significant seasonal variation illustrates that prey are targeted when they are nearshore and densely schooled in spawning or migratory aggregations.

The seasonal offshore–onshore or substrate to epi-neritic movements of many species of fish are age related and precise, allowing a predator some degree of predictability in locating forage. Steller sea lions appear to tailor their foraging behavior to these patterns in their prey. The temporal and spatial distribution of nearshore spawning aggregations of Pacific herring (Grosse 1988) and Pacific cod (Shimada and Kimura 1994) and the migratory movements of sand lance (Blackburn and Anderson 1997; Eschmeyer et al. 1983) and Pacific salmon (Pearcy 1992) parallel their highest seasonal and regional frequencies as prey among Steller sea lions. The FO of Pacific salmon best illustrates this relationship between predator and prey. Salmon were consumed by Steller sea lions at late juvenile to adult stages and mostly during summer in all regions except region 4. Most Pacific salmon live pelagically as juveniles for 1–3 years, making spring and summer alongshore migrations to breeding streams in high densities as adults (Pearcy 1992). Region 4 is within the projected fall and winter migratory corridor for Bristol Bay salmon stocks that have spent their 1st summer in the Bering Sea and are migrating south to the open Pacific for their pelagic phase (Pearcy 1992). Region 4 is the single area where salmon consumption was significantly higher in winter than in summer. By comparison, walleye pollock and Atka mackerel are nearshore the year round in the areas where they are consumed as year-round staples by Steller sea lions (North Pacific Fisheries Management Council 1998).

Regional divisions in the diet during summer are closely aligned with those defined by patterns in declines of populations of females and their pups (York et al. 1996; Fig. 5), suggesting that diet and decline of Steller sea lions are linked. York et al. (1996) modeled metapopulation breaks of the western stock of Steller sea lions based on trends in population counts of adult females and pups in summer. The results of their work determined that rookeries in close proximity to each other demonstrated similar trends in patterns of population change, just as our results show that scat collections from islands in close proximity to each other demonstrate similar patterns in diet. York et al. (1996) drew boundaries around the rookery clusters (metapopulations) defined by these population trends that are nearly identical to those clustered by diet (Fig. 5). The close-knit regional patterns of diet and decline warrant research into a finer definition of stock separation for Steller sea lions because separate stocks might be differently influenced by human or environmental perturbation.

Fig. 5.

Population trends of Steller sea lions (York et al. 1996) within regions (REG) of diet defined in this study. Colors represent sites clustered together based on patterns of decline between 1976 and 1994 and are still descriptive of current patterns in population (A. York, in litt)

The suggestion by York et al. (1996) that dispersal by breeding Steller sea lion females away from natal rookeries is low in summer is supported by the regional feeding patterns indicated here. The similarity in region boundaries between summer and winter is a further indication of strong site fidelity in year-round feeding. We suggest that this is a reflection of site fidelity in female Steller sea lions similar to that demonstrated by female northern fur seals who return to their natal rookeries to breed in summer (Baker et al. 1995; Kenyon and Wilke 1953). In addition, the information on diets presented here may represent mostly those of female and young-of-the-year Steller sea lions. Prey consumed pelagically (such as deep sea smelt, Leuroglossus) will be detected in scat if sea lions return to haul-outs with regularity. But collections at sea (Barabash-Nikiforov 1938; Calkins 1998; Tikhomirov 1964; P. J. Gearin, in litt.) show that adult and juvenile male Steller sea lions may not return to haul-out areas regularly from pelagic foraging bouts. If the strong parallel patterns of diet and population decline indicate that widespread dispersal from the region of natal rookeries is low for females, then the region of female birth may dictate foraging behavior and future reproductive success of Steller sea lions. Further, prey species that form a minor part of the diet range-wide may play a very important role in the foraging success of regional populations of Steller sea lions and their young.

Reduced prey diversity has been proposed as a potential factor in widespread declines of pinnipeds and other apex predators in the Bering Sea and North Pacific ecosystem since the broad-scale climatic change (regime shift) of 1976 (Merrick et al. 1997; National Research Council 1996; Sinclair 1988; Sinclair et al. 1994). The availability of a diverse prey base is probably important in terms of nutritional balance and predictability of forage. But regions of high prey diversity may simply reflect an environment of generally high productivity or habitat diversity.

The Unimak Pass area as well as Sea Lion Rock (Amak Island; Fig. 1) encompassed the regions of highest prey diversity. Unimak Pass is generally regarded as an oceanographically dynamic area that supports large numbers of apex predators and possibly nursery stocks of critical prey (Sinclair and Stabeno, in press). In the midst of precipitous population declines of the western stock of Steller sea lions (Loughlin et al. 1992), Amak Island was among 6 rookeries that demonstrated persistently stable or increasing populations (Amak, Akun, Akutan, Chernabura, Clubbing, Ugamak—York et al. 1996). All these sites fall within regions 2 and 3, the areas of highest diversity and greatest overlap in prey matrices (Fig. 3). The temporal model for extinction of the western stock (York et al. 1996) predicted that in the face of extinction of all other rookery sites, these 6 would remain viable. Examination of current population data confirmed that the overall patterns of decline and stability cited by York et al. (1996) in this region persist (A. E. York, in litt.). Implications of the importance of diversity in diet should be addressed further, with special attention given to the dynamics of physical and bottom-up processes that influence the nearshore habitat of rookery regions and, ultimately, the population stability of Steller sea lions.

Our results are very similar to those of historical studies in terms of indicated foraging patterns on demersal (flatfish), semidemersal (greenlings), and epipelagic (herring) prey groups, but the species of prey consumed differ markedly from studies conducted in the same area before the mid-1970s. Pollock were absent from the diet of Steller sea lions in studies conducted between 1958 and 1969 within the range of the western stock (Fiscus and Baines 1966; Mathisen et et al. 1962; Thorsteinson and Lensink 1962; Tikhomirov 1964). Imler and Sarber (1947) reported pollock in 2 stomachs collected near Kodiak Island in 1945–1946. Otherwise, the high occurrence of pollock in this study is most comparable to studies of the diet conducted since 1975 (Calkins 1998; Frost and Lowry 1986; Merrick et al. 1997; Pitcher 1981). This study also highlights the importance of Pacific cod in the diet of Steller sea lions during the winter months. Relatively few studies have focused on diet during winter, so it is difficult to assess whether this is a recent trend. Pacific cod was a top prey item in Calkins (1998) Bering Sea winter collections and in stomachs collected in the Gulf of Alaska from 1973 to 1975 (Pitcher 1981).

Overall, the most common prey items in studies conducted before the mid-1970s included capelin (Mallotus villosus), sand lance, cephalopods, herring, greenlings (Hexagrammidae), rockfishes, and smelts. Capelin, which were important in the diet of Steller sea lions through the 1970s (Fiscus and Baines 1966; Pitcher 1981), do not have an occurrence >5% in this study. Salmon were present in the diet in earlier studies but not at the high frequencies found across the western range during the summer of our study. The occurrence of flatfish, especially arrowtooth flounder, in the diet of Steller sea lions in the Gulf of Alaska is substantially higher in this study than in any previous studies. Cephalopods were among the top prey items found in Steller sea lion stomachs in many early studies (Mathisen et al. 1962; Pitcher 1981; Thorsteinson and Lensink 1962), sometimes ranking as the most frequently occurring prey item (Fiscus and Baines 1966). Cephalopod occurrence in this study was primarily limited to the central and western Aleutian Islands of region 4 and was highest during the summer months, but it never reached the high frequencies of the 1960s.

Historical studies of diet based on stomach contents are comparable to scat studies conducted since the mid-1980s in determining changes in most species of prey consumed by Steller sea lions. But each technique has its own set of biases (Pierce et al. 1991). Hard parts from very large prey may be regurgitated or retained in the stomach, never passing through to scat (Perrin et al. 1973). Small otoliths tend to flush through the digestive system more quickly than larger ones, remaining intact and more abundant in scats than in stomachs (Sinclair et al. 1996). Therefore, the adult- and late-stage juvenile fish identified in this study may represent the minimum sizes consumed, and the largest species of prey, such as the giant octopus (Octopus dofleini), may never be detected in scats. The giant octopus is highly represented in stomach collections of Steller sea lions (Pitcher 1981; A. M. Springer, in litt.; P. J. Gearin, in litt.) but was not found in our study. Despite the influence of potential biases between current and historical studies, the composition of the diet of Steller sea lions and other apex predators in the North Pacific (Alaska Sea Grant 1997; Decker et al. 1995; Hunt and Byrd 1999; Piatt and Anderson 1996; Sinclair 1988; Sinclair et al. 1994) has changed in ways that reflect a different foraging environment from that of the 1950s and 1960s.

The results of this study as well as previous observations (Fiscus and Baines 1966; Pitcher 1981) suggest that Steller sea lions are predators that capitalize on precise timing of high-density, seasonal aggregations of their prey. Steller sea lions may depend on the presence of large, dense prey patches associated with bathymetric or temperature gradients (Nishiyama et al. 1986; Sinclair et al. 1994) to detect or capture prey successfully and may encounter reduced foraging success if prey school structure is disrupted or marginalized and local prey resources are depleted (National Marine Fisheries Service 2001). Together with indications of high regional fidelity in feeding patterns (at least for females) presented here, fish school structures compromised by climate or anthropogenic factors could be a driving mechanism in either a decrease in survival or reduced intraregion reproductive success of Steller sea lions.

View this table:
Appendix I
View this table:
Appendix I.
View this table:
Appendix I.


Fish remains were identified by Pacific ID, Vancouver, BC, Canada, through the use of prey reference collections of Pacific ID and the National Marine Mammal Laboratory. Portions of the 1990–1993 data set were reported in Merrick et al. (1997). A. York was instrumental in providing statistical guidance. Reviews by S. Melin, S. Moore, K. Pitcher, P. Wade, and an anonymous reviewer were instrumental in improving the quality of this paper.

Literature Cited

View Abstract