644 



Fishery Bulletin 101 (3) 



cies over both loci and treated as a single tetrasomic locus. 

 Following the recommendations of Waples (1990), allelic 

 frequencies of samples taken in different years from the 

 same location were combined. In general, little temporal 

 allele-frequency variation was detected in coho salmon 

 populations sampled over years (Van Doornik et a!., 2002; 

 present study). Levels and patterns of genetic variation 

 within and between populations were estimated with 56 

 polymorphic loci (Table 2). Average expected heterozygos- 

 ity per locus (isoloci excluded) for each population was 

 calculated by using an unbiased estimator (Nei, 1978). 

 The proportion of P,, (,r, loci was computed for each popula- 

 tion, in which a locus was considered to be polymorphic 

 if the frequency of the most common allele was sO.95. 

 Chord distances (Cavalli-Sforza and Edwards, 1967) were 

 computed between all pairs of populations with BIOSYS 

 (Swofibrd and Selander, 1981), and relationships among 



populations were depicted with multidimensional scaling 

 (MDS, NTSYS-PC, Exeter Software, NY). Allele-frequency 

 variation among baseline populations was partitioned 

 (Chakraborty et al., 19821 into two geographic levels: 1) 

 populations within regions; and 2) among regions (Table 1). 

 These regions were delimited by geography and by genetic 

 groupings in the MDS analyses. 



We used the maximum likelihood procedures of Pella and 

 Milner (1987) and the Statistical Package for Analyzing 

 Mixtures (SPAM; Debevec et al., 2000) to estimate stock 

 contributions to simulated and actual mixtures of coho 

 salmon. Estimates were made by using 56 polymorphic loci 

 (Table 2) and 89 baseline populations, except for analysis 

 of marked (hatchery) fish where only hatchery populations 

 were used (Table 1). Allocations to individual baseline 

 populations were then summed to estimate contributions of 

 regional stock groups (Pella and Milner, 1987). Average mix- 



