Wilson; Effects of year and region on abundance and size of Theragra chalcogramma 



829 



Table 2 



Catch and sample statistics, by region and year, for age-0 pollock collected in the western Gulf of Alaska during 1985-88. Mean 

 logn,, mean fork length (FL), mean sample depth, and mean sea-surface temperature were based on samples where age-0 pollock 

 were caught. As stated in the text, fish density and length were adjusted to the median sui-vey date (3 Sept) to account for differ- 

 ences in collection date. (Freq. occur.=frequency of occurrence). 



Year 



class 



1985 



1986 



1987 



1988 



Region 



Total sample 

 number 



Kodiak 



Shumagin 



Kodiak 



Shumagin 



Kodiak 



Shumagin 



Unimak 



Kodiak 



Shumagin 



Unimak 



62 



41 

 52 

 40 

 37 

 29 

 44 

 20 

 23 

 36 



Freq. occur. 



Mean logj^ 

 (fish/ni^) 



Mean 



(fish/m3) 



Median 

 (fish/nv*) 



Mean FL 

 (cm) 



Mean depth 



(m) 



Mean temp. 



CO 



77 

 90 

 83 

 88 

 65 

 83 

 98 

 95 

 83 

 92 



-2.19 

 -2.19 

 -2.84 

 -2.74 

 -3.06 

 -2.32 

 -1.86 

 -1.76 

 -2.59 

 -2.12 



0.0303 

 0.0496 

 0.0060 

 0.0147 

 0.0046 

 0.0180 

 0.0302 

 0.0544 

 0.0242 

 0.0263 



0.0038 

 0.0073 

 0.0008 

 0.0015 

 0.0001 

 0.0053 

 0.0264 

 0.0225 

 0.0024 

 0.0063 



7.5 

 6.9 

 7.0 

 6.1 

 8.2 

 7.5 

 7.6 

 8.4 

 7.4 

 7.7 



101 

 88 

 136 

 116 

 75 

 70 

 40 

 95 

 64 

 35 



10.5 



10.9 



9.3 



10.8 



11.2 



11.6 



11.2 



10.8 



11.7 



11.0 



(Table 2). The data from 1985 clearly show how the Kodiak 

 and Shumagin regions contrast in terms offish size (Fig. 4). 

 Note that the difference in mean size coincides roughly with 

 regional delineation. On closer inspection, it was apparent 

 that large fish tended to concentrate near the northeast- 

 and east-side of Kodiak Island in 1985 and 1988. Thus, 

 size variability appears to also exist on a finer geographic 

 scale. In Figure 4, most mean lengths from the Shumagin 

 region in 1986 are above their respective line because one 

 sample of relatively small fish accounted for almost 30% of 

 the total density During 1987 and 1988. means from the 

 Unimak region were not different from those in the Shuma- 

 gin region (Bonferroni. P=0.054) but they were different 

 from the Kodiak group (Bonferroni. P=0.003). This differ- 

 ence was indicated by a second ANCOVA, and a second pos^ 

 hoc multiple comparison test, wherein day (df=l and 156. 

 F=196, P<0.001) and region (df=2 and 157. F=22, P<0.001) 

 were the only significant effects. 



To address a possible association between regional mean 

 size and mean sample depth (Table 2). the ANCOVA of 

 1985-88 data was rerun with sample depth as a third 

 effect. Sample depth was included as a categorical effect 

 with two levels: 50-100 and 100-150 m. Only 174 mean 

 length estimates were used because the number of esti- 

 mates at other depths was low or zero in some regions and 

 years. Results were similar to the first run and there was 

 no significant depth effect. Depth was neither significant 

 by itself (df=l and 157,F=0.118, P=0.731) nor did it inter- 

 act with other effects (P>0.05). The difference in fish size 

 between the Kodiak and Shumagin regions was therefore 

 not likely due to differences in sampling depths. 



Recruitment 



Future development of a direct relationship between age-0 

 abundance and recruitment seems most likely for the 



Kodiak region (Fig. 5). In this region, linear relationships 

 were evident for the average, log-transformed, nonzero 

 density and for the overall mean density. There was no 

 convincing evidence of a direct linear relationship between 

 mean length and recruitment. 



Discussion 



The geographic distribution of age-0 fish integrates the 

 effects of dispersal of young from spawning areas and of 

 spatial differences in mortality; therefore, it helps reveal 



