REISENBICHLER and PHELPS: GENETIC VARIATION IN CHINOOK AND COHO SALMON 



The modification to Chakraborty's (1980) anal- 

 ysis consisted of giving equal weight to subgroups 

 within a cell, rather than weighting them accord- 

 ing to the number of samples within each sub- 

 group. Our sampling design did not include all 

 possible or desirable subgroups; the design was a 

 compromise that allowed us to evaluate the dif- 

 ferent levels of subdivision and still remain 

 within our budget. We felt that equal weighting 

 was necessary because the number of subgroups 

 within a cell usually did not reflect the "true" 

 number of subgroups that may have existed for 

 that cell. Donald Campton (University of Florida, 

 Gainesville) provided a computer program, coded 

 in Fortran 77, that included the required modifi- 

 cation to Chakraborty's equations. 



Cluster Analysis 



The unweighted pair group method of cluster 

 analysis (UPGM analysis; Sneath and Sokal 

 1973) and (nonmetric) multidimensional scaling 

 (Gordon 1981; Kruskal and Wish 1977) were used 

 to illustrate genetic similarities among samples. 

 These two cluster analyses were applied to values 

 of Nei's (1972) genetic distance calculated for 

 each pair of samples. Data from the separate 

 broods were pooled with equal weight for these 

 analyses. 



RESULTS 



Chinook Salmon 



Although fish from two locations showed signif- 

 icant deviation from Hardy-Weinberg propor- 

 tions (P < 0.05/n, where n, was the number of loci 

 tested for location i) — juveniles of the 1982 brood 

 from the Bogachiel River were deficient in het- 

 erozygotes at the Pgk-2 locus and juveniles of the 

 1982 brood from the Hoh River had an excess of 

 heterozygotes at the Gpi-2 locus — these devia- 

 tions are probably spurious, given the large num- 

 ber (20) of samples tested. 



Interbrood variation in allele frequencies was 

 significant (P < 0.01) for wild fish and for hatch- 

 ery fish (Table 4). Six loci, or pairs of loci, showed 

 sufficient variation and were scored for enough 

 fish (« > 25) to be used in the ANOVA (Fig. 3, 

 App. Table 1). Variation between drainages was 

 not significant, although summer-run fish may 

 differ between drainages iP = 0.07, Table 4). 

 Hatchery fish were different from wild fish (con- 

 trasts 3 to 6 in Table 5). 



The UPGM cluster analysis showed that the 

 hatchery populations were distinct from wild ju- 

 veniles and from all but one (Quinault River) 

 sample of adults (Fig. 4). Of the hatchery popula- 

 tions, fall-run fish from Soleduck Hatchery were 



Table 4. — Chinook salmon — likelihood ratio analysis of interbrood variation at 10 codominant loci. Significant levels were 



evaluated for totals only. G = likelihood ratio statistic. 



•P < 0.05 n 

 "P < 01 n 

 tP<0.01. 



where n = 3 for interbrood variation within drainages, n = 5 for variation within streams, and n 

 eries These are corrections for multiple comparisons (Cooper 1968). 



4 for variation within hatch- 



687 



