Tollit et al.: Sizes of walleye pollock consumed by Eumetopias /ubatus 



529 



They also found that the relative shape, structure, and 

 proportion of the morphological features used to es- 

 timate erosion were consistent for both smaller and 

 larger fish. We therefore assumed that the DCFs in that 

 study could be used reliably for the fish in our study 

 outside of the experimental size range in which they 

 were considered. 



Applying DCFs increased mean fork length estimates 

 by 23% (from 34.4 to 42.4 cm) on average and resulted 

 in adult fish contributing 44% to the sea lion diet by 

 number and 74% by mass. The contribution of juvenile 

 fish was insignificant. Applying valid correction factors 

 clearly provides better insights into prey-size selection 

 and consequently niche overlap. It should also lead to 

 more precise estimates of mass of prey consumed and 

 the number of prey within a scat (Ringrose, 1993; Tollit 

 et al., 1997; Laake et al., 2002). 



Over 61 species of prey were identified in the diet of 

 Steller sea lions in Southeast Alaska from 1993 to 1999 

 (Trites et al. 3 ). The most common prey were walleye pol- 

 lock, Pacific herring (Clupea pallasi). Pacific sand lance 

 (Ammodytes hexapterus), salmon iOncorhynchus spp.), 

 arrowtooth flounder (Remhardtius stomias), rockfish (Se- 

 bastes spp.), skates {Raja spp.), and cephalopods. During 

 summer, gadids (most of which were pollock) made up 

 27% of the diet, and increased to 49-62% of the diet at 

 other times of the year (Trites et al. 3 ), confirming that 

 pollock are a significant component of the diet. 



Steller sea lions consumed a wide size range of pollock 

 in Southeast Alaska; the bulk of fish fell between 20 and 

 60 cm and peaked between 44 and 52 cm (Fig. 2). The 

 contribution of juvenile fish (<20 cm) was insignificant. 

 The only historical data to compare with these results 

 are those from the stomach samples of eight Steller sea 

 lions collected from Southeast Alaska in 1986 (Calkins 

 and Goodwin 1 ). Pollock lengths backcalculated from all 

 otoliths found in the stomachs were generally shorter 

 (mean FL = 25.5 ±10.4 cm, range; 4. 8-55. 7cm, n = 80) 

 than our estimates from multiple structures found in 

 scats collected during the 1990s (mean FL = 42.4 ±11.6 

 cm, range: 10.0-78.1 cm, n = 909). It should be noted 

 that we derived our estimates after removing heavily 

 eroded structures and applying DCFs, whereas Calkins 

 and Goodwin 1 did not account for partial digestion. 

 However our estimates of pollock length would have 

 been similar to those of Calkins and Goodwin 1 if we 

 had used only otoliths and had not corrected for diges- 

 tion (Table 2). Although Frost and Lowry (1980) found 

 no significant difference between the size of otoliths 

 obtained from stomachs and intestines of ribbon seals, 

 underestimates of fish size determined from otoliths 

 from stomach samples will depend on the time since 

 ingestion (i.e., on the extent of digestion). 



One possible explanation for the virtual absence of ju- 

 venile pollock in the scats we examined is that the rela- 

 tively smaller structures of smaller fish were more likely 

 to be completely digested, and were therefore underrep- 

 resented in the scats (Tollit et al., 1997; Bowen, 2000). 

 However, juvenile pollock otoliths and bones were found 

 in large numbers in a number of scats collected from 



the western stock (Zeppelin et al., 2004, this issue). 

 Clearly, the potential for underestimating smaller fish 

 depends heavily on the balance between relative re- 

 covery rates and the number of different size fish con- 

 sumed in a meal. For example, if an animal needs to 

 eat 5 kg a day, then it would have to consume 195 15.5- 

 cm pollock, but less than ten 41-cm pollock. Given that 

 large pollock bones are at least three times more likely 

 than small bones to pass through the digestive tract 

 (Tollit et al., 2003; D. J. Tollit, unpubl. data), the sheer 

 numbers of small pollock in this example would lead 

 to a conclusion that smaller fish were more important 

 numerically, when in fact they were equally important. 

 Conversely, the relative proportion of large fish is likely 

 to be overestimated if ten large and ten small pollock 

 are consumed together. The generally low number of 

 structures per scat provides little information to assess 

 this balance. Hence we must assume that our results 

 are representative and unbiased. 



Steller sea lions in Southeast Alaska did not seem to 

 eat fish over 65 cm. Whether or not sea lions do not tar- 

 get large fish, or whether large fish are harder to catch 

 and handle, or are encountered at a lower rate is not 

 known. However, large fish could be under-represented 

 in scats if large fish cannot be swallowed whole, and 

 head skeletal parts are lost while the fish is torn apart 

 on the surface (Olesiuk et al., 1990; Wazenbock, 1995) 

 or if bone regurgitation is size specific. 



Regional, geographical, and temporal variation in sizes of 

 pollock consumed 



Stomach samples collected in 1975-78 and 1985-86 in 

 the Gulf of Alaska contained substantial numbers of 

 juvenile pollock, as well as larger fish (mode: 39-43 cm). 

 In 1985, the distribution of sizes consumed by sea lions 

 around Kodiak Island appeared to mimic that of the pol- 

 lock population (Merrick and Calkins. 1996). However, 

 juvenile sea lions ate significantly smaller and relatively 

 more juvenile pollock than adult sea lions. Stomachs 

 from the Gulf of Alaska contained an average of 49 pol- 

 lock (1975-78) and 72 pollock (1985) compared with 1.6 

 pollock per scat in Southeast Alaska. In the Bering Sea, 

 90 stomachs were examined between 1975 and 1981 by 

 using only non-eroded otoliths, and these also contained 

 mainly (76%) juvenile pollock (mean FL=29.3 cm), but 

 also some adult fish (Frost and Lowry, 1986). 



Between 1998 and 2000, Steller sea lions across the 

 range of the western population in Alaska consumed 

 pollock averaging 39.3 ±14.3 cm (range: 3.7-70.8 cm, 

 Zeppelin et al., 2004, this issue). This finding suggests 

 that sea lions may have been less reliant on juvenile 

 pollock than they were during the 1970s and 1980s. 

 Apparent differences may reflect differences in pollock 

 year-class strength, and thus differences in the domi- 

 nant size classes that were available to be consumed. 

 However, Zeppelin et al. (2004, this issue) reported 

 that the size distribution of walleye pollock consumed 

 by Steller sea lions between 1998 and 2000 did not ap- 

 pear to fluctuate with year-class strength, unlike the 



