Gudmundson et al : Diet of Callorhmus ursinus 



449 



in "good" or "fair" condition. The occurrence of walleye 

 pollock age classes determined from otolith measure- 

 ments varied by sample type (Fig. 3). 



A multidimensional contingency analysis showed 

 that sample type was independent of island (P= 0.087, 

 X~ = &.2G, df=3); therefore data from St. George and St. 

 Paul Islands were pooled to test for differences in age 

 class of prey between sample types. Spews contained 

 a larger percentage of adult pollock (88.0%, P<0.001, 

 X^=2&Q.21, df=l) and a smaller percentage of juvenile 

 pollock (9.1%, P<0.001, x- = 180.91. df=l) than scats. 

 There was not a significant difference in the occur- 

 rence of mixed-age pollock between scats (6.4%) and 

 spews (4.0%) (P=0.490, x-=QAl, df=l) and only 71 of 

 1125 (6.3%^) samples contained both age classes. A total 

 of 2856 and 5030 Gb-Bm lower beaks were recovered 

 and measured from 225 scat samples and 111 spew 

 samples, respectively, collected at St. George Island 

 (Fig 4.). A total of 719 Gb-Bm lower beaks were re- 

 covered and measured from 110 St. Paul Island scat 

 samples and 47 Gb-Bm lower beaks were recovered 



and measured from five St. Paul Island spew samples. 

 The mean DML of Gb-Bm in St. George Island scat 

 and spew samples was 69 mm (SE = 0.10) and 82 mm 

 (SE = 0.14), respectively. The mean DML of Gb-Bm in 

 St. Paul Island scat samples was 67 mm (SE = 0.21) and 

 in spew samples was 101 mm (SE = 6.35). Gb-Bm data 

 were log transformed to meet assumptions of normal- 

 ity for island and sample-type size comparisons. The 

 DML of Gb-Bm estimated from beaks recovered from 

 spew samples was significantly larger than the DML 

 estimated from beaks recovered from scat samples col- 

 lected from both islands (two-sample ^test, P<0.001 for 

 St. George Island and St. Paul Islands). The DML of 

 Gb-Bm from St. George Island scats was significantly 

 larger than that form St. Paul Island scats (two-sample 

 /-test, P<0.001). In spew samples, estimated Gb-Bm 

 DML was significantly larger on St. Paul Island than 

 on St. George Island (two-sample f-test, P<0.001), How- 

 ever, the number of spews collected from the two is- 

 lands were highly unequal (7i=204, St. George Island; 

 « = 63, St. Paul Island). 



