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



The population unit of Georgia Strait (unit 11) com- 

 prises populations forming clusters 3 and 6 in the 

 dendrogram, plus the San Juan River population. 

 These six populations aggregate adjacently in the 

 plottings of PCS and PC4. Populations of Unit II 

 typically have relatively high allele frequencies of 

 Aat-3 (90), Pkg-2 (90), and Tapep-1 (130), although 

 exceptions occur at each locus. Carl and Healey 

 (1984) reported similar high frequencies for allelic 

 variations of Aat-3 and Tapep in a study of chinook 

 salmon populations of the Nanaimo River which 

 flows from Vancouver Island into Georgia Strait. 



Populations in the Puget Sound unit (Unit III), 

 bounded to the north by the population from the 

 South Fork of the Nooksack River, aggregate fair- 

 ly clearly in both the dendrogram (clusters 8D and 

 8E) and the plots of PC3 and PC4. The cohesiveness 

 among the fall-run populations vary likely reflects 

 both genetic isolation and present (or very recent) 

 gene flow through transfers among hatcheries. Like 

 unit II, populations of unit III also have high allele 

 frequencies for Tapep-1 130; in fact, it has the high- 

 est mean frequencies for this allele among the nine 

 population units that were formulated. However, the 

 mean frequencies of the common (i.e., 100) alleles 

 for Aat-3 and Pgk-2 are much higher in unit III than 

 in unit II. No influence of reported transfers of lower 

 Columbia River fish to Puget Sound hatcheries (e.g., 

 Ricker 1972) is apparent from the graphic projec- 

 tions or the allelic data. 



An extended grouping of coastal populations (unit 

 IV) ranges from northern California (see Utter et 

 al. 1980) to Robertson Creek on the west coast of 

 Vancouver Island. Populations of the Columbia, 

 Klamath, and Sacramento Rivers are excluded from 

 unit IV. This unit is distinguished by high frequen- 

 cies of the Gpi-S (60) allele and (in most instances) 

 some Pgm variation. Most populations appear either 

 in clusters 4 or 8B of the dendrogram and aggregate 

 distinctly in the plottings of PCI and PC2. Two 

 populations are retained in Unit IV for geographic 

 consistency which do not congregate with other 

 populations of this unit; the spring run returning to 

 the Soleduck River on the Washington coast, and 

 the Lobster Creek population returning to the upper 

 Rogue River on the Oregon coast. The outlying of 

 the Soleduck spring-run population appears to be 

 related to its heterogeneous origins. Records in- 

 dicate that this run originated from crosses of fish 

 from the Cowlitz River (lower Columbia River) and 

 Umpqua River (Oregon coast) with some contribu- 

 tion from the spring run of the Dungeness River, 

 a drainage entering the Strait of Juan de Fuca (C. 

 Johnson^). An explanation for the outlying of the 



Lobster Creek population is less apparent and re- 

 quires further investigation. 



Two individual and four paired hatchery popula- 

 tions sampled from the lower Columbia River form 

 a geographically and genetically discrete unit (unit 

 V). This group represents the most divergent pair 

 of clusters (1 and 2) on the dendrogram and general- 

 ly aggregates distinctly in the plotting of PCI and 

 PC2. Populations of unit V are particularly distin- 

 guished by high allele frequencies of (yr (85) and Mpi 

 (109). Unit V is bounded upstream by the U.S. Fish 

 and Wildlife Service Spring Creek Hatchery popu- 

 lation (Spring Creek Hatchery is located on the pool 

 impounded by Bonneville Dam). The pairing of four 

 of the six populations is consistent with high levels 

 of gene flow resulting from an extensive history of 

 transplantation among the populations of the lower 

 Columbia River (Simon 1972; Howell et al. 1985). 

 This group's distinctness from other groups is also 

 consistent with a minimal impact of transplantations 

 of these populations beyond the lower Columbia 

 River on indigenous populations in other areas (e.g., 

 Cowlitz spring-run fish to the Snake River, C. 

 Burley^; Kalama fall-run fish to Puget Sound, men- 

 tioned above). 



The upper Columbia River unit (unit VI)— more 

 than any of the other groupings— is composed of 

 genetically diverse elements placed together more 

 on the basis of geographic convenience rather than 

 genetic unity. Unit VI is somewhat loosely bounded 

 downstream by populations of the Klickitat and 

 Deschutes Rivers; both rivers enter the Columbia 

 River near The Dalles Dam. Unit VI's component 

 populations include individuals of mixed ancestral 

 origins, along with others of presumably pure line- 

 age. Two populations known to have mixed ances- 

 tral origin are those of the U.S. Fish and Wildlife 

 Service Carson and Leavenworth Hatcheries. The 

 Carson Hatchery population (located on the Wind 

 River which drains into the Bonneville pool) was 

 derived from interceptions of spring-run fish 

 destined for areas of the upper Columbia and Snake 

 Rivers. The Leavenworth population (combined with 

 Carson in the analyses) has been largely maintained 

 by continued infusions from fish of the Carson 

 Hatchery (Howell et al. fn. 5). The Ice Harbor pop- 

 ulation—another group of mixed ancestral origins- 

 is composed of fall-run fish destined for different 

 areas within the Snake River that were intercepted 

 at Ice Harbor Dam near the mouth of the Snake 



'^C. Johnson, Washington Department of Fisheries, General Ad- 

 ministration Bldg., Olympia, WA 98504, pers. commun. May 1985. 



«C. Burley, U.S. Fish and Wildhfe Service, 9317 Highway 99, 

 Vancouver, WA 98665, pers. commun. May 1985. 



248 



