84 



Fishery Bulletin 90(1). 1992 



We do not know if the genetic structure of the Blue 

 and Omagar Creek samples is characteristic of the 

 lower Klamath-Trinity drainage. The Omagar Creek 

 sample consisted of progeny of broodstock captured by 

 instream gill nets at the mouth of Blue Creek and in 

 the main section of the Klamath River; the Blue Creek 

 sample was collected in the main stem of Blue Creek 

 and was presumed to represent progeny of natural 

 spawning. If accurate, our data suggest greater gene 

 exchange between the lower Klamath and coastal 

 populations of northern California-southern Oregon 

 than between the lower and upper Klamath basin. Ap- 

 parently northern California coastal populations of 

 Chinook salmon are genetically similar to southern 

 Oregon populations because the two samples from the 

 Smith River (samples 10 and 11) also clustered with 

 the Oregon populations. This genetic similarity may 

 have resulted from historical gene exchange in the form 

 of transplants into the Klamath basin (Snyder 1931). 

 Chinook salmon in the lower Klamath River are 

 thought to be similar to Oregon populations in other 

 characters, such as timing of spawning migration, 

 fecundity, and size (Snyder 1931; Craig Tuss, U.S. Fish 

 Wildl. Serv., Sacramento, CA 95616, pers. commun., 

 Sept. 1990). 



The relatively high incidence of Hardy-Weinberg 

 disequilibria in hatchery and pond rearing programs 

 may be the result of the limited number of broodstock 

 used in production or non-random sampling of a hatch- 

 ery's production, i.e., only sampling juveniles from a 

 few raceways. For example, the Coleman National Fish 

 Hatchery spawns approximately 10,000 fall-run Chi- 

 nook salmon. It is likely that our sample of 100 juveniles 

 may not be an adequate representation of the hatchery 

 output. The two samples with the highest number of 

 deviations from Hardy-Weinberg expectations were 

 both from pond rearing projects, Omagar and Camp 

 Creeks. These pond rearing projects can serve a useful 

 function by augmenting or establishing runs of chinook 

 salmon in specific streams. However, care must be 

 taken to maximize the effective population size of the 

 broodstock and to prevent changes in the genetic 

 variation. 



The large number of significant departures from 

 Hardy-Weinberg expectations for the Klamath samples 

 compared with other samples was due primarily to the 

 samples from Camp Creek and Omagar Creek. These 

 two samples accounted for nine of the 13 significant 

 tests within the Klamath system. Deleting data for 

 these two Creeks from the comparison resulted in 6% 

 (4 of 72) significant deviations for Klamath system 

 samples versus 7% (24 of 349) for non-Klamath 

 samples. 



Our results indicate a geographic basis for genetic 

 differentiation and subpopulation structure in chinook 



salmon populations from California and Oregon. Geo- 

 graphic affinities among chinook salmon populations 

 have now been demonstrated along most of the western 

 coastline of North America (Gharrett et al. 1987, Utter 

 et al. 1989, Bartley and Gall 1990). Bartley and Gall 

 (1990) identified three major clusters of chinook salmon 

 populations in California that corresponded to the three 

 major river drainages: the Sacramento-San Joaquin, 

 the Eel, and the Klamath-Trinity. Utter et al. (1989) 

 identified nine population imits of chinook salmon over 

 a large area from British Columbia to California. They 

 found coastal populations from Oregon and Washing- 

 ton to be genetically similar to each other. Our data 

 indicate that some coastal populations in California are 

 differentiated from those in Oregon, but that northern 

 California coastal populations of chinook salmon are 

 similar to southern Oregon populations. 



The level of intrasample gene diversity found in the 

 present study, 89.4%, is similar to the values of 82.3 

 and 87.7% reported by Bartley and Gall (1990) and 

 Utter et al. (1989), respectively. Overall estimates of 

 gene flow of 1.16 (Bartley and Gall 1990) and 2.11 (this 

 study) migrants per generation also are similar. The 

 slightly lower level of population subdivision and there- 

 fore, higher level of gene flow found in the present 

 study probably reflect a bias caused by the samples 

 analyzed. Bartley and Gall (1990) analyzed a greater 

 number of inland California populations than the pres- 

 ent study. Most of their samples were from the three 

 major drainages within California: the Klamath- 

 Trinity, the Sacramento-San Joaquin, and the Eel. 

 They suggested that straying and gene flow were 

 higher among coastal streams than among separate 

 drainages. Therefore, by including the large number 

 of coastal samples in the present study, slightly higher 

 overall estimates of gene flow and less apparent 

 subdivision were expected. Separate gene diversity 

 analyses of the groups from Oregon and northern 

 California revealed that approximately 6% of the total 

 diversity of the two Oregon groups was due to inter- 

 population differences compared with 12% for the 

 three California groups. These results further support 

 the expectation of lower levels of population subdivi- 

 sion when analyses involve many coastal samples. 



The estimates of gene flow and population subdivi- 

 sion from hierarchical gene-diversity analyses varied 

 among geographic areas. The Klamath-Trinity system 

 would be expected to display lower levels of gene ex- 

 change if the lower and upper sections of the Klamath 

 are separate subpopulations. However, deletion of the 

 Blue Creek and Omagar Creek samples from the anal- 

 ysis changed the gene diversity estimates by less than 

 2%. The high level of estimated gene flow within the 

 Sacramento-San Joaquin system most likely reflects 

 the fact that four of the five samples were from 



