GRANT ET AL.: POPULATION GENETICS OF YELLOWFIN SOLE 



64°N- 



62°N 



60°N 



58°N 



56°N 



Bering 

 Seo 



I70°W 



I60°W 



50°W 



I40°W 



Figure 1. 



-Map of Bering Sea and Gulf of Alaska showing locations (see Table 1 for numbered locations) of samples of yellowfin sole used in 



this study. 



with N(3-aminopropyl)-morpholine (Clayton and 

 Tretiak 1972); (III) gel 1:4 dilution of tray solution, 

 tray, TRIS 0.18 M, boric acid 0.1 M, EDTA 0.004 M 

 (pH 8.7) (Markert and Faulhaber 1965). 



The system of locus and allelic nomenclature sug- 

 gested by Allendorf and Utter (1979) was used. 

 Locus homologies with other fish (Whitt et al. 1975; 

 Fisher and Whitt 1978) were designated with letters 

 where they could be deduced from tissue dis- 

 tributions (Table 2). The enzymes examined in this 

 study, their abbreviations, and Enzyme Commission 

 (E.C.) numbers are listed in Table 2. 



Statistical Procedures 



Departures from Castle-Hardy- Weinberg propor- 

 tions in each of the samples were detected using the 

 log likelihood-ratio test for goodness of fit (Sokal and 

 Rohlf 1969) with the degrees of freedom equal to the 

 number of phenotypes minus the number of alleles 

 for a codominant locus. 



Stock structure was analyzed using a nested 

 contingency-table analysis where the total hetero- 

 geneity among allelic frequencies at each locus was 



partitioned into orthogonal, regional comparisons in 

 a manner analogous to ANOVA. The log likelihood- 

 ratio test criterion, G, was used to test each com- 

 parison with the degrees of freedom equal to the 

 number of alleles minus one, times the number of 

 areas or samples minus one. Only loci having variant- 

 allele frequencies of 0.05 or greater were used in this 

 analysis to avoid low expected frequencies. Rare 

 alleles at these loci were pooled into the next least- 

 frequent allelic class until the pooled class reached a 

 frequency of at least 0.05. The significance level of 

 each comparison was modified to account for the 

 increase in type I error, when multiple tests of the 

 same comparison are made (Cooper 1968). Com- 

 parisons at each locus were considered significant if 

 G exceeded the value in a chi- square table associated 

 with a probability of 0.05/4 = 0.012, where n was the 

 number of polymorphic loci. In this way the overall 

 probability of rejecting H by chance was 1 — (1 — 

 0.05/4) 4 -0.05. Only data of samples taken in 1979, 

 1980, and 1981 were used in all of the statisical 

 analyses, because the six earlier samples were not all 

 taken from spawning areas and because not all of the 

 loci were examined in these samples. However, allelic 



669 



