Winship and Trites: Prey consumption of Eumetop/as /ubatus 



157 



method may overestimate the contribution of small prey 

 and underestimate the contribution of large prey to diet 

 biomass. However, as previously mentioned, such a bias 

 would be reduced if small prey are consumed in greater 

 numbers than large prey in a given meal. A potentially 

 better technique than split-sample frequency of occur- 

 rence is volumetric or biomass reconstruction analysis 

 (i.e. the estimation of the actual size of each prey in a scat 

 from the size of otoliths or other hard parts), but otoliths 

 are usually not available from Steller sea lion scat (Mer- 

 rick et al., 1997) and digestion correction factors (Tollit et 

 al., 1997) and regressions of hard-part size on body size 

 are currently not available for prey of Steller sea lions. 



Olesiuk et al. (1990) and Olesiuk (1993) estimated the 

 error associated with a key assumption of the split-sample 

 frequency of occurrence technique (all prey categories in a 

 scat are consumed in equal quantities) by calculating the 

 minimum and maximum split-sample frequencies of oc- 

 currence of prey. For example, the minimum split-sample 

 frequency of occurrence of a prey category was calculated 

 by assuming that when the prey category was found in 

 a scat with another prey category, it represented a negli- 

 gible proportion of the biomass of the meal represented by 

 that scat. We considered their estimates of the minimum 

 and maximum split-sample frequencies of occurrence of 

 prey to approximate the total potential errors in the diet 

 compositions we used. Thus, the assumed errors in diet 

 were based only on the potential error in estimating diet 

 from scats and not on the potential error due to sampling 

 limitations. 



Our assumed errors in diet composition were relatively 

 large. For example, if a prey category was assumed to 

 comprise a median of 509r of the diet, the proportion of 

 the diet represented by that prey category in any one run 

 of the model ranged from 27.5'7r to 72.5% (before the diet 

 was standardized to 100%). Nevertheless there is still the 

 possibility of sampling biases in the diet compositions that 

 we assumed for Steller sea lions in Alaska. Future studies 

 of the diet of Steller sea lions will allow us to obtain better 

 estimates of the regional, seasonal, and intrapopulation 

 variability in diet. Also, studies of captive Steller sea lions 

 will assist in determining the biases and variability asso- 

 ciated with the estimation of diet from scats (Cottrell and 

 Trites, 2002). 



Effect of diet on food requirements 



We found that changes in the energy density of the diet 

 can have large effects on the amount of food that Steller 

 sea lions need to consume. In southeast Alaska, seasonal 

 changes in the energy density of the diet resulted in large 

 seasonal changes in daily food requirements, even when 

 daily energy requirements were relatively constant. 

 Immature and mature animals (excluding lactating 

 females) required approximately 45-60% more food per 

 day in early spring than in late summer (Fig. 3). Regional 

 differences in the energy density of the diet resulted in 

 smaller, but still substantial differences in food require- 

 ments among Steller sea lions in different regions of 

 Alaska (up to a 24% difference based on summer diets). 



The effect of diet on food requirements, can be further 

 illustrated by considering two diets: one of entirely gadids 

 (walleye pollock, Pacific cod) and one of entirely small 

 schooling fish (herring, sandlance). Based on caloric dif- 

 ferences between prey types and differences in digestive 

 efficiency and the heat increment of feeding, a 10-year-old 

 male would require an average of 30 (±7.7) kg of small 

 schooling fish per day (5% of body mass), but would re- 

 quire 41 (±9.7) kg of gadids (6% of body mass): a 37% 

 increase in prey biomass requirements. A 10-year-old 

 female's (pregnant, no pup) average daily food require- 

 ment would increase by a similar percentage with a diet 

 shift from small schooling fish to gadids ( 15 (±3.71 kg to 21 

 |±4.7| kg or 6% to 8% of body mass). 



A large animal may be able to compensate for changes 

 in prey biomass requirements, but immature or recently 

 weaned animals may be more susceptible to changes in 

 prey biomass requirements because they need to consume 

 more food per unit body mass than adult animals (Win- 

 ship et al., 2002). A 1 -year-old male would require an 

 average of 16 (±4.2) kg of small schooling fish per day 

 (12% of body mass), but would require an average of 22 

 ±5.4 kg of gadids (16% of body mass). Similarly, a 1-year- 

 old female would need 14 (±3.2) kg of small schooling fish 

 (13% of body mass) or 18 (±4.1) kg of gadids (17% of body 

 mass). The difference in energy density between the gadid 

 and forage fish categories was greater in the summer and 

 autumn than in the winter and spring; thus the difference 

 between the daily food requirement of a sea lion consum- 

 ing only gadids and the food requirement of a sea lion 

 consuming only forage fish was greatest in the summer 

 and autumn. Although animals prey on more than one 

 species category in nature, which would buffer the effects 

 of changes in diet composition, differences in the energy 

 density and digestibility of prey can have large effects on 

 prey biomass requirements, especially for young animals. 



Merrick et al. (1997) found a significant relationship be- 

 tween the diversities of Steller sea lion diets and the rates 

 of change in the numbers of adult and juvenile Steller 

 sea lions counted on rookeries between 1990 and 1994 

 in different regions of Alaska (Gulf of Alaska through the 

 western Aleutian Islands). Sea lions in regions with high 

 rates of population decline had low dietary diversity. Plot- 

 ting the rates of population decline against the amount 

 of prey required in each region (using summer diets), we 

 found a significant (a=0.05) relationship (Spearman rank 

 correlation coefficient i\=-0.929, P=0.02; Fig. 6). indicat- 

 ing that sea lions in areas with high rates of decline had 

 higher per capita food requirements. This finding suggests 

 that the energy density of the diet may have had a role in 

 the population decline in some regions of Alaska during 

 the early 1990s. 



The correlation we report between food requirements 

 and population change (Fig. 6) is based on summer diets 

 and limited data on the energy density of sea lion prey 

 categories such as hexagrammids. When seasonal diet in- 

 formation for southeast Alaska was considered, our model 

 predicted a substantially greater per capita food require- 

 ment in that region (Fig. 4). Seasonal data are required 

 from all regions of Alaska to describe diet composition 



