FISHERY BULLETIN: VOL, 87, NO, 2, 1989 



broad geographic range and high heterozygosity of 

 the coastal populations support the long-term exist- 

 ence of unit IV in which cohesiveness among popu- 

 lations appears to have been maintained through 

 some degree of gene flow (Soule 1976; Campton and 

 Utter 1987). Ecological as well as geographic bar- 

 riers to extensive gene flow from the coastal area 

 apparently existed in the Columbia, Klamath, and 

 Sacramento drainages. However, the presence of 

 the Grpi-2(60) allele— typical of coastal populations— 

 in some populations of units V, VI, VIII, and 

 IX suggests some degree of introgression from 

 coastal populations. Natural obstructions of the 

 mid-Columbia River such as Cascade Falls and CelUo 

 Falls (presently obscured by Bonneville and The 

 Dalles Dams, respectively) may have restricted 

 migration between populations of the lower Colum- 

 bia River and those of the upper Columbia and 

 Snake Rivers. 



The relationship of the Snake River populations 

 of unit VII to other groups within and beyond the 

 Columbia River is unclear. Its most distinguishing 

 feature is its very low average heterozygosity (H 

 = 0.04), an attribute shared with the Klamath River 

 populations (H = 0.029) (unit VIII) with which it 

 also aggregates in the dendrogram and the principal 

 component projections. In spite of this similarity, 

 we favor an explanation that both Snake River and 

 Klamath River populations had independent origins. 

 The high frequencies of common (i.e., 100) alleles 

 over the present sampling of loci are interpreted as 

 reflecting loss of variation through genetic drift ac- 

 centuated by periodic bottlenecks and restricted 

 gene flow (see also Winans in press). This explana- 

 tion is consistent with the inland locations of both 

 drainages. In addition, both drainages continued to 

 flow within their present courses during the Pleis- 

 tocene. Thus, similarity is presently interpreted as 

 an artifact based on minimal allelic variation de- 

 tected over most of the loci sampled. However, drift 

 coupled with isolation should lead to divergent fre- 

 quencies of some alternate alleles with an adequate 

 sampling of variable loci. If such differences are not 

 observed as additional genetic marks continue to be 

 detected in chinook salmon, then a zoogeographical 

 explanation based on gene flow or recent ancestry 

 must be pursued for Snake River and Klamath River 

 populations. 



Following glacial regression, the newly habitable 

 regions appear to have been repopulated from 

 diverse sources. Origins of the northern portions of 

 the coastal unit can be readily explained by immi- 

 grations from more southern coastal streams. How- 

 ever, populations of units I, II, and III apparently 



arose from other sources based on their virtual 

 absence of Gpi-2 variation. Seeding of the Fraser 

 River from sources including the upper Columbia 

 River and Snake River units, and of Georgia Strait 

 and Puget Sound drainages from the lower Colum- 

 bia River or Alaska, are possibilities that seem more 

 likely. The Aat-3 (85) allele is recorded in most Alas- 

 kan populations studied by Gharrett et al. (1987) at 

 frequencies up to 0.32. The highest frequencies of 

 this allele occur in populations from Vancouver 

 Island suggesting immigration from northern 

 refugia. 



Comparisons with Sympatric Salmonid 

 Species 



It is of interest to compare the present data set 

 with similar information from other anadromous 

 salmonid species within the same geographic range. 

 These species presently share habitats and have 

 been subjected to the same geological processes 

 throughout their periods of common habitation. 

 Thus, some common patterns of genetic population 

 structuring may be anticipated. However, differ- 

 ences among species in life histories and long-term 

 distributions may likewise result in unique popula- 

 tion structures. Similar data sets have been collected 

 from four species within this range: rainbow trout, 

 Salmo gairdneri; coastal cutthroat trout, S. clarki; 

 chum salmon, 0. keta; and sockeye salmon, 0. nerka. 



Investigations of rainbow trout include both anad- 

 romous (i.e., steelhead) and nonmigratory popula- 

 tions (Huzyk and Tsuyuki 1974; Allendorf 1975; 

 Allendorf and Utter 1979; Allendorf et al. 1980; 

 Busack et al. 1980; Chilcote et al. 1980; Parkinson 

 1984; Wishard et al. 1984). A geographic basis for 

 population structure is also apparent in this species 

 and allelic similarities persist among indigenous 

 populations of a particular region regardless of 

 migratory tendencies, times of migration, or local 

 environments. Apparent population units for 

 chinook salmon and rainbow trout differ, however. 

 A single major population unit of rainbow trout com- 

 prising the upper Fraser River, the upper Colum- 

 bia River, and the Snake River contrasts with at 

 least three distinct groupings for chinook salmon. 

 A clear distinction between coastal streams of 

 Washington and Oregon from those of the lower 

 Columbia River, Puget Sound, and Georgia Strait 

 is also not apparent in rainbow trout as it is in 

 chinook salmon. 



Distribution of sockeye salmon over the geograph- 

 ic range of this study is less continuous than that 

 of chinook salmon because of the more stringent 



250 



