438 



Fishery Bullelin 100(3) 



Results 



Scats were collected on 31 occasions during March-Octo- 

 ber 1995-97 providing 1385 scats with identifiable prey 

 remains (Browne et al., 2002). All seasons were sampled 

 in each year except for the fall of 1995. Remains were 

 identified from 5832 different prey which were assigned to 

 (0=28 prey groups. We excluded the very minor unidenti- 

 fied component. Salmonids of unknown species and size 

 represented by bone (154) were partitioned based on the 

 observed seasonal proportions (Eqs. 4 and 5, Table 1). 



During 16 of the flights in 1997, the proportion of the 26 

 radio-tagged seals hauled out was measured to construct 

 a single average correction factor (\lf) of 1.64 (percent 

 coefficient of variation (CV)=6.7%) to estimate seasonal 

 abundance from the haulout counts (Table 2). Using the 

 assumed age and sex structure and biomass requirements, 

 we estimated that the seals in the Columbia River would 

 consume 704 metric tons (t) of biomass during the 7.5 

 month period (Table 2). 



From the 1385 scats, we reconstructed 1.15 t of biomass 

 which was only 0.16% of the required biomass consump- 

 tion. Because the number of scats and seals varied be- 

 tween seasons, the percentages varied from 0.077% in 

 spring, 0.173% in summer, and 0.272% in fall. Using BR, 



we constructed diet composition estimates (Fig. 1) and 

 seasonal consumption estimates (Fig. 2). Many of the 

 estimates of salmonid consumption had exceedingly poor 

 precision with the coefficient of variation exceeding 0.5 

 (Table 3). Using adult chinook salmon (O. tshawytscha) as 

 an example, an examination of the variance components 

 demonstrated that estimation of biomass (Table 4) was 

 the predominant source of variance. Variance associated 

 with biomass estimation includes variation in predicted 

 weights and estimation of number of prey (n,) for salmo- 

 nids. The latter accounted for one- to two-thirds of the 

 total variance for salmonids depending on the species and 

 size group. The use of genetics to obtain species identifica- 

 tion of the unknown salmonids identified by bone would 

 substantially improve the precision for salmonids. 



Using BR to estimate diet composition, adult chinook 

 salmon was the only salmonid that was consistently in the 

 five most important prey items for each season based on 

 percent of biomass (Fig. 1). Their importance was derived 

 from their average mass which was the largest of all the 

 prey. Adult chinook salmon and the other salmonids ap- 

 peared much less important if the ranking was based on 

 the number consumed (Fig. 2). Smaller prey items such 

 as herring iClupeid spp.), sculpin iLeptocottus armatiis), 

 lamprey (Petromyzontid spp.), smelt iOsmeridae) and 



