UTTER ET AL : GENETIC POPULATION STRUCTURE OF CHINOOK SALMON 



Table 5— Summary of distribution of relative gene diversity of chinook salmon in geographical and temporal 

 fiierarcfiies based on 65 individual or pooled populations and 25 polymorphic loci. Areas, regions, and seasons 

 are given for each population in Table 1 . Absolute values of gene diversity include mean average hetero- 

 zygosity (Hs) - 0.1018 and total diversity (Ht) - 0.1161. 



hierarchy resulted from differences between popula- 

 tions within seasons (11.6%), with only 0.7% of the 

 total diversity being due to differences between 

 seasons. 



Interpretation of Observed Variation 



We interpreted the overall data set primarily as 

 a reflection of patterns and levels of gene exchange 

 among populations. This interpretation does not ex- 

 clude the possibility of some selective forces influ- 

 encing the frequencies of some alleles and genotypes 

 in some environments (e.g., Powers et al. 1983; 

 Mork et al. 1984). However, empirical data from 

 diverse animal species justify an assumption of 

 predominant neutrality (Ihssen et al. 1981; Chakra- 

 borty et al. 1980; Eanes 1987). This assumption is 

 strengthened when many polymorphic loci are ex- 

 amined and is particularly pertinent in anadromous 

 salmonids where restricted population sizes accentu- 

 ate the influence of genetic drift (Utter et al. 1980). 

 The data presented here indicate that chinook 

 salmon consist of a genetically complex network of 

 populations throughout the geographic range of this 

 study. This information yields some clear conclu- 

 sions and suggests a number of additional possibil- 

 ities that must await confirmation or rejection from 

 additional studies. 



One conclusion is that the time of return (i.e., 

 season) is not a major factor in establishing relation- 

 ships of stocks among areas. Both the geographic 

 clustering in Figures 2 and 3 and the small between- 

 seasons component of the temporal gene diversity 

 analysis point away from the concept of a recent 

 common ancestry of fish returning at the same time 

 in different areas. This finding comes as no surprise 

 based on published data of other anadromous 

 salmonids (e.g., rainbow trout, Allendorf and Utter 

 1979). However, it is still commonly accepted that 

 the chinook salmon is separated into temporally 



distinct "races" (e.g., McClane 1978). Although a 

 strong genetic component for the time of return has 

 been clearly demonstrated in anadromous salmonids 

 (e.g., Helle 1981), and this is not debated here, it 

 appears that genetic divergence into temporally 

 distinct units tends to occur within an area from a 

 common ancestral stock of chinook salmon. 



In contrast to the lack of evidence for genetic 

 structuring of time of return, a geographic basis for 

 genetic structuring is apparent. The relatively large 

 area component of gene diversity (over half of the 

 between-population diversity in column 1 of Table 

 5) coupled with the predominantly geographic 

 clusterings warrant an attempt to define different 

 geographically discrete population units. Most imits 

 (Figure 2 and Appendix) incorporate one or more 

 of the areas or drainage systems listed in Table 1. 

 Inevitably, overlap occurs between these formulated 

 population units and the a priori groupings of areas 

 or drainages. 



The Fraser River grouping (unit I) is necessarily 

 limited to the upstream areas because no down- 

 stream populations were sampled. The single sam- 

 ple from the Babine River, tributary to the Skeena 

 River and adjacent to drainages of the upper Fraser 

 River, is also tentatively included in Unit I. The 

 Babine population aggregates with those of the 

 Fraser River in the dendrogram (cluster 5A) and the 

 plots of PCI and PC2. Most populations of Unit I 

 (including the Babine) are distinguished by the pres- 

 ence of the Gr-110 allele at a mean frequency of 0.05. 

 This allele was not included in the Appendix or in 

 most analyses because of the incomplete data sets 

 from some coastal populations. The Gr-(llO) allele 

 was not observed in other populations that aggre- 

 gate in the dendrogram and PC plots with those of 

 unit I; these populations include the San Juan River 

 (southern Vancouver Island), the spring- and sum- 

 mer-run fish of the Snake River, and the Klamath 

 River. 



247 



