512 



Fishery Bulletin 102(3) 



Minimum number of individuals for each species in 

 each scat was estimated by counting species-specific 

 sided elements and choosing the greatest number of left 

 or right elements. If more than one structure had the 

 same number, the structure with the highest r 2 value 

 in its regression on fork length (Table 1) was selected 

 as a representative length estimate for that fish. If an 

 equal number of left and right bones were present, right 

 bones were selected. 



Temporal variation in size of walleye pollock and 

 Atka mackerel consumed by Steller sea lions 



Temporal differences were assessed by grouping fish 

 into stage-class categories. Stage-class categories were 

 defined for pollock as follows: juvenile or 1-year-old 

 fish (<20 cm FL), adolescent (20.1-34 cm FL), subadult 

 (34.1-45 cm FL), and adult (>45.1 cm FL; Dorn et al., 

 2001; Smith, 1981; Walline, 1983). Walleye pollock sub- 

 adults are likely 3-4 years old, of which -50% have 

 matured and recruited into the fishery, whereas adults 

 are sexually mature fish, likely 5 years or older. Stage- 

 class categories for Atka mackeral were defined as fol- 

 lows: juvenile up to 2-year-old fish (<30 cm), adolescent 

 or 3-year-old fish (30.1-35.2 cm), subadult or 4-year-old 

 fish (35.3-45 cm), and adults (>45.1 cm; Lowe et al., 

 2001; McDermott and Lowe, 1997). Atka mackerel ado- 

 lescents are -50% sexually mature and adult-size fish 

 are fully mature. 



A chi-squared contingency test was used to test for 

 differences in the proportion offish stage-classes occur- 



ring in scats among rookeries and haul-out sites, years, 

 and seasons by using corrected fork-length estimates 

 from all cranial structures (S-PLUS 2000, Mathsoft, 

 Inc., Cambridge, MA). To avoid pseudoreplication, we 

 used presence or absence of cranial elements of a stage 

 class in a scat particilarily because multiple elements 

 from the same stage-class within a sample may not be 

 independent (Hunt et al., 1996). By using presence- 

 absence data we also avoided the problems associated 

 with the variability in passage and recovery rates of 

 different size structures (Tollit et al., 1997). Because 

 sample sizes were small, juvenile and adolescent wall- 

 eye pollock stage classes and recruiting adult and adult 

 Atka mackerel stage classes were combined for seasonal 

 comparisons among years. Fisher's exact test was used 

 for comparisons when samples sizes for any stage class 

 were less than 5 (S-PLUS 2000, Mathsoft, Inc., Cam- 

 bridge, MA). 



We obtained size composition data from commercial 

 bottom trawls of walleye pollock and Atka mackerel 

 from the NMFS North Pacific Groundfish Observer Pro- 

 gram. Data were divided into winter (January-April I 

 and fall (August-November) seasons and compared 

 with our seasonal scat data (February-March and 

 June-September). The percentage of overlap in sizes 

 of fish caught by the commercial groundfish fishery 

 with sizes of fish consumed by Steller sea lions was 

 calculated by comparing size-frequency distributions. 

 Two-cm size bins were used for the overlap calculation 

 and Steller sea lion prey-size data were rounded to the 

 nearest integer to be consistent with the fishery data. 



