OTTER ET AL.: GENETIC POPULATION STRUCTURE OF CHINOOK SALMON 



River. This population is included in unit VI because 

 of its geographic proximity and genetic similarity 

 to populations of unit VI contrasted with its dis- 

 tinctness from spring- and summer-run populations 

 of the upper Snake River. 



Populations of purer lineage within unit VI aggre- 

 gate within cluster 8 of the dendrogram. The spring- 

 run population returning to the Lewis River lies 

 geographically within unit V, entering the Colum- 

 bia River below Bonneville Dam. This population is 

 included in unit VI because it is genetically distinct 

 from other downstream populations and more 

 typical of certain spring- and fall-run fish within Unit 

 VI (i.e., Klickitat, Deschutes, and Winthrop popu- 

 lations) with which it closely aggregates on the den- 

 drogram (cluster 8C) and the plots of PCI and PC2. 



The similarity of the populations from Wells Dam 

 and Priest Rapids Dam in unit VI is presumably a 

 reflection of the two groups being different temporal 

 segments of the same major run. All fish migrating 

 past Priest Rapids Dam prior to 13 August are per- 

 mitted to pass upstream and sequentially constitute 

 the spring- and summer-runs of the upper Colum- 

 bia River. The latter segment of this migration ar- 

 riving at Wells Dam is captured and spawned for 

 hatchery production. Most arrivals at Priest Rapids 

 Dam later than 14 August are intercepted and 

 spawned there (Chris Carlson'). This process inevi- 

 tably results in considerable gene flow between 

 these two artificially maintained populations. 



The Snake River unit (unit VII) contains the two 

 combined populations of McCall Hatchery-Johnson 

 Creek and Rapid River Hatchery-Valley Creek- 

 Sawtooth-Red River, all managed by the Idaho 

 Department of Fish and Game; all populations are 

 from the Salmon River drainage of central Idaho. 

 This unit is distinguished by very low average 

 heterozygosities (see Winans in press) and by high 

 frequencies of the Pgk-2 (90) allele. 



The Klamath River populations (unit VIII) are 

 geographically isolated from, but genetically similar 

 to those of the Snake River. However, populations 

 of unit VIII lack variation of Idh-3,4 contrasted with 

 a mean frequency of 0.925 for the Idk-S.J, (100) allele 

 in unit VII. Klamath River populations, like those 

 of unit VII, are characterized by very low aver- 

 age heterozygosities. This characteristic contrasts 

 sharply with most adjacent coastal populations for 

 which the highest heterozygosities among all popu- 

 lations are observed. Allele frequency data from the 

 Shasta and Scott river populations, two wild pop- 



'Chris Carlson, Grant County Public Utility District, P.O. Box 

 878, Ephrata, WA 98823, pers. commun. March 1986. 



ulations of the Klamath River are statistically iden- 

 tical with frequencies in the Iron Gate Hatchery 

 sample; these data were recently collected which 

 precluded their use in most of the analyses of this 

 study. Thus the low heterozygosity of Klamath River 

 populations cannot be attributed to effects of hatch- 

 ery management (see Allendorf and Ryman 1987). 



The three samples from the Sacramento River 

 drainage form a distinct geographic and genetic unit 

 (unit IX). These samples cluster together in the den- 

 drogram (cluster 7) and in PCI and PC2. As men- 

 tioned above, these populations are distinguished by 

 high frequencies of the Gpi-l(H) allele. 



An analysis of gene diversity within and between 

 the nine proposed population units (Table 5, column 

 3), provides further support for the reality of these 

 genetic subdivisions. It is appropriate that almost 

 two-thirds of the total gene diversity due to popula- 

 tion structuring (7.9/12.3 = 64.2%) occurred be- 

 tween the population units. Furthermore, the 

 diversity between populations within the units was 

 smaller than the diversity between populations 

 within areas (Table 5, column 1) calculated prior to 

 the synthesis of the units. 



Relationships and Origins of 

 Population Units 



The common genetic and geographic attributes of 

 populations within units have been stressed, but 

 relationships between units also require considera- 

 tion. The geographic areas of the Fraser River, 

 Georgia Strait, and Puget Sound (units I, II, and III) 

 were completely glaciated during the late Pleisto- 

 cene, and therefore must have been entirely re- 

 populated within roughly the last 15,000 years 

 (McPhail and Lindsey 1986). Those areas of the 

 Columbia River sampled in this study were outside 

 of the ice sheet, although the upper third of the 

 drainage was glaciated. However, downstream pop- 

 ulations (units V and VI) were doubtlessly affected 

 by massive runoffs and temporary impoundments 

 resulting from sudden releases of glacial Lake Mis- 

 soula initially occurring some 18,000 years ago 

 (Bunker 1982); most of the Snake River drainage 

 (unit VII), entering the mid-Columbia River from 

 the south, was presumably unaffected by these 

 events above its lower reaches. The coastal region 

 (Unit IV) from the Chehalis River (Washington) 

 southward, and the entire Sacramento-San Joaquin 

 River drainage (unit HI), were likewise free of 

 glaciation during the late Pleistocene. 



Much of the presently observed genetic diversity 

 almost certainly existed during the Pleistocene. The 



249 



