Microsatellites reveal population identity of individual pink salmon to allow supportive breeding of a population at risk of extinction

Efforts to restore depressed populations of Pacific salmon Oncorhynchus spp . are often hampered by the inability to assign population identity to indi viduals in an admixture. This knowledge is of particular concern in support ive breeding programs, in which misidentification of individuals to populat ion may result in progeny of mixed heritage, which, in turn, results in the erosion of the genetic population structure and of the existing genetic di versity and local adaptations of the target population. We evaluated two cl asses of genetic markers, allozymes and microsatellites, for estimating pop ulation identity of pink salmon Oncorhynchus gorbuscha in a supportive bree ding program on the Dungeness River in Washington State. Fall-run pink salm on of the Dungeness River are the target of restoration, but they presumabl y overlap, in terms of timing, with an earlier summer run. Both marker type s revealed similarly low estimates of relative genetic differentiation (<(t heta)over cap> = 0.02), which suggests that there is little variation in al lele frequency among populations. However, microsatellites provided a more accurate estimate of population identity. When applying a log-likelihood ra tio criterion of greater than 1.3, 74.8% of individuals were correctly assi gned to population using microsatellites (versus 3.1% of individuals using allozymes). The difference in assignment accuracy was best predicted by the statistic <(delta)over cap>, which estimates cumulative allele frequency d ifferences among populations. Our results suggest that genetic markers with many alleles are preferred when populations exhibit little genetic differe ntiation (as is the case in pink salmon), because <(delta)over cap> is more likely to be large, presumably as a result of genetic drift at each allele . The use of microsatellites to select fall-run pink salmon for supportive breeding confirmed the run-timing overlap and prevented unintentional cross es between the two populations.