Number of alleles as a predictor of the relative assignment accuracy of short tandem repeat (STR) and single-nucleotide-polymorphism (SNP) baselines for chum salmon

Short tandem repeat (STR) markers, which exhibit many alleles per locus, are commonly used to assign fish to their populations of origin. Single-nucleotide polymorphisms (SNPs), which have many technical advantages over STRs, typically exhibit only two alleles per locus. Simulation studies have indicated that the number of independent alleles is a good predictor of the accuracy of genetic markers for fishery applications. Extant STR baselines for salmon Oncorhynchus spp. contain hundreds, of alleles, and it has been estimated that hundreds of SNP markers will need to be developed before the SNP baselines will be comparable to the STR baselines. We compared 15 STRs exhibiting 349 independent alleles with 61 SNPs exhibiting 66 independent alleles for accuracy in assigning individuals to closely related populations of chum salmon O. keta. The SNP baseline yielded slightly higher mean accuracies for proportional assignment and comparable accuracies for individual assignment. Based on the number of independent alleles in each baseline, the SNP baseline performed considerably better than the STR baseline. We suggest that this discrepancy is due to the fact that the simulation studies do not capture the impacts of the different strategies commonly employed for discovering and selecting STR and SNP markers.