148 



Fishery Bulletin 101(1) 



170°E 



65 N 



Mean = 120.7 

 SD = 23.3 



Units: thousands oft 



Figure 1 



Estimated annual food biomass requirements (thousands of tl for Steller sea lions in 1998 in 

 seven study areas of Alaska assuming that the summer diets wore consumed all year long. SDs 

 were obtained by using Monte Carlo simulations ( 1000 runs). Pic charts represent the proportions 

 of diet biomass that each prey species category represents (defined in text). Diameters of the pie 

 charts are proportional to their respective mean food requirement estimates. The map and study 

 areas were adapted from Merrick et al. ( 1997) and Seasc and Loughlin ( 1999). Numbers in paren- 

 theses in the central Aleutians (areas 2-4) are subarea numbers. 



heat increment of feeding or the efficiency of using metab- 

 olizablo energy). Metabolizable energy requirements of 

 individuals were assumed to be the same in all regions 

 of Alaska. Gross energy requirements of individuals 

 were expected to vary among regions of Alaska because 

 digestive efficiency and the heat increment of feeding 

 are dependent upon the energy density of the diet, which 

 varied among regions. Next, population size and composi- 

 tion were calculated by using pup count data from 1998, 

 and a life table for Steller sea lions in Alaska. Population 

 size varied by region of Alaska, but we assumed that 

 population composition (i.e. sex and age structure) was 

 the same for all regions. Finally, food requirements were 



calculated by assuming a given diet composition (percent 

 contribution of each prey category to diet biomass) and by 

 using information on the energy density of prey. Diet com- 

 position varied by region of Alaska, but we assumed the 

 energy density of individual prey categories did not. 



The model incorporated a Monte Carlo random sam- 

 pling routine which allowed us to estimate the error in the 

 model predictions based on the assumed uncertainty in 

 each parameter value. Three types of parameter sampling 

 distributions were used; uniform (defined by upper and 

 lower limits; e.g. 0.1-0.3), triangular (defined by a median, 

 an upper limit and a lower limit; e.g. 0.2, 0.1-0.3), and nor- 

 mal (defined by a mean and SD; e.g. 0.2 ±0.0.5). 



