82 



Fishery Bulletin 90(1), 1992 



where Nm is the average number of migrants exchang- 

 ing genes per generation. Equation (1) was solved for 

 Nm by setting Fst equal to the relative gene diversity 

 appropriate for the hierarchical level of interest. This 

 formulation provided an estimate of the number of 

 migrant fish exchanging genes among samples per 

 generation under the assumptions of selective neutral- 

 ity of alleles and Wright's (1943) island model of migra- 

 tion. Slatkin and Barton (1989) discussed the sensitivity 

 of equation (1) relative to various methods of estimating 

 Fgx in the presence of selection and alternative popu- 

 lation structures, and found it to be fairly robust. 



Results 



A total of 96 isozyme loci were examined. Thirty-one 

 loci were monomorphic, 47 were categorized as poly- 

 morphic (Appendix A), whereas variability of an un- 

 known and undefined nature was detected at 18 loci. 

 Details of genetic polymorphisms not described else- 

 where are outlined in Appendix B. The enzyme systems 

 involving the 18 loci for which evidence of probable 

 polymorphisms was detected (not listed in Table 2) and 

 warrant further study included: two adenylate kinase 

 loci, creatine kinase, four fructose biphosphate aldolase 

 loci, four glyceraldehyde-3-phosphate dehydrogenase 

 loci, two beta-galactosidase loci, alpha-glucoside, super- 

 oxide dismutase, two peptidase loci, and a highly anodal 

 acromatic band. Because of difficulties defining a gene- 

 tic model of inheritance, poor band resolution, or in- 

 complete data, these 18 loci were not included in the 

 analyses. 



Tests of conformance to Hardy Weinberg genotypic 

 proportions revealed 37 out of 462 cases (8%) of dis- 

 equilibria. For wild samples of chinook salmon, 13 of 

 252 tests (5%) revealed disequilibrium, whereas in 

 hatchery samples, 24 of 210 tests (11%) showed non- 

 conformance to Hardy- Weinberg expectations. How- 

 ever, in the Klamath Basin, a higher percentage of 

 disequilibrium was found (13 of 97 cases or 13%) in 

 hatchery and wild samples. The proportion of disequi- 

 librium observed in Klamath and non-Klamath samples 

 was found to be significantly different (P<0.05) when 

 tested for equality by the generalized likelihood-ratio 

 test for binomial data (Larsen and Marx 1981) . The 

 proportion of disequilibrium observed in hatchery 

 (including pond rearing programs) and wild chinook 

 salmon populations also was significantly different 

 (P<0.05). The nature of the observed disequilibrium 

 appeared to be random. That is, we did not observe con- 

 sistent excesses or deficiencies of heterozygotes, nor 

 did we observe specific loci that consistently deviated 

 from Hardy- Weinberg expectations. 



Estimates of average heterozygosity ranged from a 

 low value of 0.028 in Shasta River (#16) to a high of 

 0.076 in the Morgan Creek (#2) and Elk River (#5) 

 hatcheries. The Middle Oregon samples (#1-6) tended 

 to have high estimates of average heterozygosity, 

 whereas values for the Klamath-Trinity samples 

 (#12-21) tended to be lower (Table 1). 



Although genetic identity indices between all pairs 

 of samples were greater than 0.982 (data not shown), 

 the geographic distribution of alleles suggested popula- 

 tion subdivision within the study area. For example, 

 we found the Aat-2(85), Aat-3(90), Aat-Ml30), and 

 Iddh-l(O) alleles predominantly in Oregon and north- 

 coastal California (collections 1-11). The mAh-Jt(112), 

 Gpi-H(*), and Pgdh<90) alleles were present mainly in 

 the Sacramento/San Joaquin system (collections 33- 

 37), whereas Mdhp-1(92) and Gpi-2(60) were less abun- 

 dant in the Sacramento Basin compared with more 

 northern areas. Mdhp-2(78) was a characteristic of the 

 Klamath-Trinity system and a few coastal samples. 



Cluster analysis of genetic identities revealed a 

 strong geographic component to the grouping of 

 chinook salmon samples. Five distinct clusters that 

 reflected geographic areas were evident (Fig. 2): (1) 

 Smith River-Southern Oregon rivers, (2) Klamath- 

 Trinity Rivers, (3) Eel River system-California coastal 

 rivers, (4) Middle Oregon rivers, and (5) Sacramento- 

 San Joaquin system. The Smith River (#11) and the 

 Rowdy Creek Hatchery (#10) samples were the most 

 northern samples collected from California. Therefore, 

 it is reasonable that they would be genetically similar 

 to the southern Oregon samples. The sample from the 

 Fall Creek Hatchery (#1) was the only sample from 

 northern Oregon and therefore, appears as an indepen- 

 dent cluster. Three samples. Rock Creek Hatchery (#6, 

 middle Oregon), Blue Creek (#12, Klamath-Trinity 

 Basin), and Omagar Creek (#13, Klamath-Trinity 

 Basin), did not cluster in accordance with their geo- 

 graphic location. 



Total gene diversity was 0.0620 (Hx) and average 

 sample diversity was 0.0554 (Hg). Therefore, approx- 

 imately 89.4% of the total genetic diversity was due 

 to intrasample variability and 10.6% was due to inter- 

 sample variation (Table 3). Further examination of the 

 intersample diversity showed that genetic differences 

 among samples within the five geographic groups iden- 

 tified from the dendrogram (see Table 1) accounted for 

 about 3.2% of the total variation and 7.4% of the total 

 diversity was due to differences between the major 

 geographic areas. Gene diversity analysis for each 

 geographic area treated separately revealed that 

 although the Klamath-Trinity system possessed the 

 lowest total gene diversity for a given area (Hd), rela- 

 tive gene diversity (Ggo) for this drainage was high 



