Analysis of Genetic Diversity in Geographically Structured Populations: A Bayesian Perspective

Abstract

Bayesian approaches have been widely applied to partitioning diversity within and among levels in many different multi-level modeling contexts. In spite of the structural similarities between these Bayesian models and hierarchical approaches to partitioning diversity in population genetics, population geneticists have not explored the use of hierarchical Bayesian models to provide estimates of Wright's F-statistics. In this paper I describe and illustrate the application of a simple multilocus, two-allele model sufficient for partitioning diversity within and among populations. Extenions of the model incorporate both fixed-effect and random-effect models of population sampling at multiple hierarchical levels with multiple alleles per locus. The Bayesian approach developed here is closely related to previously developed methods for likelihood analysis of the same problem. I illustrate the utility of the Bayesian approach with a reanalysis of previously published allozyme data from Argania spinosa.