528 



Fishery Bulletin 102(3) 



Overlap of size of pollock consumed by Steller sea lions 

 with size of pollock caught by the fisheries 



The Canadian commercial pollock trawl fishery in Dixon 

 Entrance between 1993 and 1999 landed mostly (93%) 

 adult fish (mean FL = 52.2 ±5.9 cm, «=2103, modal range: 

 48-54 cm). The majority (79%) of scats containing pol- 

 lock from the Forrester Island rookery in June and July 

 also contained remains of adult pollock (mean corrected 

 FL=51.4 ±10 cm, a? = 192, modal range: 46-52 cm, n s =81 

 scats). Percentage overlap based on a comparison of 

 size-frequency distributions totaled 75.1% for those fish 

 eaten around Forrester Island and 52.1% for all fish 

 eaten. However, the estimated overlap would have been 

 assumed incorrectly to be half these values if DCFs had 

 not been applied to the selected digested otoliths and 

 bones (i.e., 36.7% overlap at Forrester and 24.1% for all 

 areas combined). Clearly overlap levels would have been 

 further underestimated if structures in poor condition 

 had been included in our analyses. 



Discussion 



Only 57% of the scats (303 of 531) that contained suit- 

 able pollock remains had structures that were in good 



enough condition to be measured reliably. Numbers of 

 elements in good or fair condition (rc = 909) averaged 

 three per scat, and a very small fraction of these con- 

 sisted of otoliths (<4%). The most numerous structures 

 were DENT, QUAD, and ANGU (Table 1). This finding 

 is inconsistent with feeding trials with captive Steller 

 sea lions where otoliths were found to be the most com- 

 monly occurring structure (Cottrell and Trites, 2002; 

 Tollit et al., 2003). 



Different structures yielded somewhat different mean 

 sizes of pollock, although 95% confidence intervals gen- 

 erally overlapped, ranging between 37 and 52 cm for 

 bones (Table 2). Such discrepancy is not surprising 

 given that different bones originate from different scats 

 and possibly different fish (even within a single scat). 

 Our comparison of estimates with all structures versus 

 MNI selections indicates that the potential effect of 

 double counting (and measuring) fish within a single 

 scat is likely negligible with large sample sizes (Fig. 2). 

 Although the use of all-structure data to estimate fish 

 length results in a greatly increased sample size, there 

 remains an underlying assumption that all structures 

 are affected equally by digestion. Tollit et al. (2004, 

 this issue) found no significant difference in the degree 

 of erosion across the three size ranges (28.5-45.0 cm 

 FL) for each structure within each condition category. 



