Browne et al.: Improving pinniped diet analyses 



429 



Smelt Shiner surf Pacific herring Pacific Sebastes spp Pacific lomcod Salmonid spp Salmonid spp 



species perch staghorn (juvenile) (adull) 



scuipin 



Prey taxa 



Figure 1 



For eight prey taxa. ratio of estimated mean number of individuals estimated from the number of otoliths 

 recovered from scat multiplied by a species-specific correction factor for recovery rate (Harvey, 1989) to MNI 

 estimated from all structures recovered from harbor seal scats collected from Desdemona Sands during 

 spring, summer, and fall 199.5 through 1997. Values for smelt species are based on correction factors for euch- 

 alon. Values for salmonid species are based on correction factors for steelhead salmon. 



opalescens and Octopus spp.) mass was estimated from 

 regressions of beak measurements on mantle length and 

 mass (Wolf. 1982). 



Discussion 



Diet of harbor seals in the Columbia River 



Identification of prey remains indicated that the diet of 

 harbor seals in the Columbia River was temporally vari- 

 able and seals appeared to exploit prey when species 

 were abundant. Many of the dominant prey by number 

 and frequency were small fish such as herring, smelts, 

 northern anchovy, juvenile flatfish, and sculpins (Tables 3 

 and 4). Pacific herring. Pacific staghorn scuipin iLeptocot- 

 tus armatus), and smelts were three of the top six prey 

 taxa by number and frequency for all three seasons (Table 

 4), although estimated masses varied greatly between 

 season, indicating that seals preyed on different size 

 classes (Table 5). Interestingly, scats without remains 

 were most common during late spring and summer (Table 

 2). Olesiuk et al. ( 1990) reported similar results from Brit- 

 ish Columbia and suggested harbor seals were feeding on 

 soft-bodied prey and roe. Occurrence of these scats in our 

 study coincided with pupping on the lower Columbia River 



^ Huber, H. R. 1997. Unpubl. data. National Marine Mammal 

 Laboratory, 7600 Sand Point Way, NE, Bldg. 4, Seattle, WA 

 98115. 



(Huber*'), but an alternative explanation is that these may 

 have been from nursing pups. 



Most common harbor seal prey were variable by season 

 or year (or both). Seasonal effects in the diet indicated pe- 

 riods when prey were reproducing, when young of the year 

 were available, or perhaps an absence of highly abundant 

 prey when seals relied more heavily on consistently avail- 

 able species. Annual differences in frequency of occur- 

 rence may have been largely the result of differences in 

 sample timing or prey cohort strength (Moyle and Cech, 

 1982; Dark and Wilkins, 1994). Significant year and inter- 

 active effects between season and year probably reflected 

 differences in prey year class (Table 3). For example, FO of 

 anchovy (known to have high variability in recruitment) 

 collected during fall of 1996 was 63%, whereas during fall 

 of 1997 it was only 2%, although samples were collected 

 on similar dates (Table 1). Harbor seals are generalist 

 feeders and differences in frequency and number of prey 

 probably reflect the temporal availability of prey rather 

 than predator selection. This hypothesis is supported by 

 AIC over-dispersion constants (6) greater than 1.0 for all 

 prey taxa (Table 3). A binomial model assumes constant 

 probabilities of a prey taxon occurring in scats collected on 

 any sampling date within a season or year. Ephemerally 

 abundant prey will have highly variable probabilities of 

 occurrence in harbor seal scats collected in each season. It 

 is likely that the overdispersion constant underestimates 

 deviations for taxonomic gi'oups, including more than one 

 species such as salmonids and smelts, because temporal 

 abundance of the different species in the group may be 

 offset. 



