Beacham: The use of DMA variation for stock identification of Oncorhynchus keta 



619 



As with the Ssa-A33 locus, genotypic frequencies 

 at the Ssa -A34 locus were in Hardy- Weinberg equi- 

 librium in all of the 42 stocks sampled. Mean het- 

 erozygosity of the Japanese stocks (0.77) was higher 

 than that for any other regional group of stocks (Table 

 3 ) and significantly higher than that for North Ameri- 

 can (G=15.0, df=l,P<0.05) or Russian stocks (G=6.50, 

 df=l, P<0.05). Heterozygosity of Japanese chum 

 salmon was consistently higher than that for other 

 regional groups of chum salmon when both the Ssa- 

 A33 and Ssa-A34 loci were considered. 



Annual variation in allele frequencies 



Samples from two Vancouver Island (Big Qualicum, 

 Nitinat) stocks and one Fraser River (Chilliwack) 

 stock were obtained in both 1991 and 1992. No sig- 

 nificant annual variation was observed in allele fre- 

 quencies at either the Ssa -A33 or Ssa -A34 loci for 

 any stock (P>0.05)(6 tests conducted). 



Stock structure 



Neighbor-joining analysis of the distances between 

 stocks derived from allele frequencies at the Ssa -A33 

 and Ssa -A34 loci and from band counts observed with 

 the Ssa 1 probe indicated that Japanese chum salmon 

 stocks were distinct from all other groups of stocks 

 (Fig. 3). Regional stock structure was also examined 

 with principal-components analysis, and all nine 

 Japanese stocks sampled were clearly separated from 

 all other stocks (Fig. 4). Russian and Yukon River 

 stocks were more similar to each other than to more 

 distant stocks in Japan and British Columbia (Figs. 

 3 and 4). There was weak regional structuring among 

 stocks of British Columbia chum salmon, and there 



was some evidence that southeast Alaska and north- 

 ern British Columbia stocks were more similar to 

 each other than to stocks in southern British Colum- 

 bia (Fig. 3) and some evidence that there was sepa- 

 ration between ECVI and Fraser River stocks 

 (Fig. 4). 



Estimation of stock composition 



The potential use of minisatellite DNA variation for 

 estimation of stock composition of stocks from broad 

 geographic areas was evaluated for two main group- 

 ings of chum salmon, namely mixtures of Japanese 

 and Russian chum salmon and mixtures of Russian 

 and Yukon River chum salmon. On the basis of stock 

 structure, relatively accurate and precise estimates 

 of stock composition should be derived from samples 

 containing Japanese and Russian chum salmon 

 (Table 4). Similarly, in applications where estima- 

 tion of Russian and Yukon River chum salmon com- 

 positions is of interest, relatively accurate and pre- 

 cise estimates of stock composition of the two target 

 groups should be derived (Table 4). The Andrefsky 

 River stock was in a group with Russian stocks in 

 the neighbor-joining analysis (Fig. 3) but seemed 

 separable from Russian stocks in the principal-com- 

 ponents analysis (Fig. 4). The simulations suggested 

 that even if the Andrefsky River comprised all the 

 Yukon River component, separation of Russian and 

 Yukon River stock components should still be pos- 

 sible (Table 4). 



The application of minisatellite DNA variation to 

 local stock composition issues was evaluated for three 

 situations; i.e. mixtures of Japanese, Yukon River, 

 and southern British Columbia chum salmon stocks 

 were evaluated separately. The simulations sug- 

 gested that both discrimination among all of the 

 Japanese stocks sampled and relatively accurate 

 estimates of stock composition observed in both 

 single-stock and multistock simulated mixtures were 

 possible (Table 5). Because less than 20 fish per stock 

 were sampled for five of the nine stocks surveyed 

 (Table 1), increased baseline sampling, however, 

 would be required to confirm these results before 

 application to actual fishery samples. The simula- 

 tions of Yukon River chum salmon mixtures suggested 

 that all five stocks were distinct from each other and 

 that accurate estimates of stock composition may be 

 possible with minisatellite DNA variation (Table 6). The 

 fall-run Sheenjek River stock was distinguishable from 

 fall-run stocks spawning in the Yukon Territory. The 

 simulations of southern British Columbia chum salmon 

 mixtures suggested that minisatellite DNA variation 

 may be applicable for estimation of stock composition 

 in fishery samples. Estimates of stock composition for 



