POPULATION-STRUCTURE AND STOCK IDENTIFICATION OF BRITISH-COLUMBIA COHO SALMON, ONCORHYNCHUS-KISUTCH, BASED ON MICROSATELLITE DNA VARIATION

We used genetic variation at three microsatellite DNA loci to describe population structure in 34 coho salmon populations from British Colum bia and to perform stock composition analysis on simulated mixed-stock fishery samples. Each microsatellite locus was highly polymorphic, wi th 31 alleles at Ots101, 20 alleles at Ots3, and 38 alleles at Ots103. Average observed heterozygosities were 86.3%, 73.3% and 74.9% respect ively. Analysis of genetic distances revealed three relatively homogen eous, geographically based groups of coho salmon populations in the fo llowing regions: the upper Skeena and Nass River watersheds, the lower Fraser River drainage, and the upper Fraser River-Thompson River wate rsheds. Coastal populations from the mainland of British Columbia, Van couver Island, and the Queen Charlotte Islands formed a more heterogen eous regional stock grouping. Significantly different allele frequenci es were observed among populations within regions, and allele frequenc ies were generally temporally stable in multiyear samples. Phylogeneti c lineages within British Columbia coho salmon likely reflect geograph ic patterns of recolonization from at least three separate glacial ref ugia after the last ice age. Local spawning populations within regions may form metapopulations, but current levels of gene flow among subpo pulations are apparently insufficient to prevent differentiation at ne utral genetic loci. Maximum-likelihood estimates of stock composition were accurate and precise, indicating great potential for management o f coho salmon at the level of metapopulations or ''evolutionarily sign ificant units'' in domestic and international mixed-stock fisheries. I ndividual fish were identified to stock by using a discriminant analys is with a high degree of accuracy in a few regions, but more generally with approximately 50% success.