Bayesian analysis of population mixture and admixture

Eric C. Anderson* and Jonathan K. Pritchard

A common problem in fields such as fisheries management and evolutionary ecology involves the genetic analysis of samples of individuals taken from mixed populations. Typically, one wants to infer which subpopulations are represented in a sample, and in what proportions. Previous analyses have assumed either that every individual is of mixed ancestry (population admixture) or that each individual is purely descended from a specific subpopulation (population mixture). Here, we present a more flexible model in which each individual may be of pure or mixed ancestry. We use MCMC to compute the posterior probability of an individual's type of ancestry and the posterior distribution of associated genetic parameters. We demonstrate the method on data from wild populations which have been interbreeding with domestic strains.