On the use of allele frequency data within a Bayesian framework: evaluation of the relative probabilities of alternative stock structure hypotheses for North Pacific minke whales

Abstract

Genotype frequency information for one or more loci is used within a Bayesian modelling framework to assign relative probabilities to alternative stock-structure hypotheses using the Bayes factor approach. This framework has advantages over maximum-likelihood estimation as it provides the information needed to select amongst hypotheses. For primarily illustrative purposes, the approach is applied to the data for the Adh-1 and Gpi loci for sub-areas 6, 7, 8, 9 and 11 for North Pacific minke whales. The results confirm those of previous studies that there are (at least) two stocks to the east and west of Japan. In contrast, the results support the hypothesis of a single stock in sub-areas 7, 8 and 9 unless a priori the allele frequencies for stocks that are adjacent spatially are likely to be similar. This last result needs to be interpreted with caution as the mutation rate of allozymes is slow and so this caveat might apply in this case.