Plant & Animal Genomes XV Conference Plant & Animal Genomes XV ConferenceJanuary 13-17, 2007
Town & Country Convention Center
San Diego, CA
Hong Gao , Scott Williamson , Carlos D. Bustamante
When studying partially selfing species, it is of considerable interest to distinguish the effects of selfing as a mode of reproduction from population substructure maintained by limited dispersal of gametes. Here, we extend the popular Bayesian clustering approach STRUCTURE for simultaneous inference of selfing rates and population classification using multilocus genetic markers. We present a family of related Markov chain Monte Carlo algorithms for sampling from the posterior distribution of parameters in various models of variation in self-fertilization rates among individuals and populations. In particular, we consider a population specific model for selfing rates (i.e., each population has a distinct selfing rate) and an individual specific model where selfing rates vary randomly among individuals in the sample. We gauge the performance of our method, InStruct, using extensive Monte Carlo simulations and demonstrate that our MCMC approach has good accuracy both in classification of individuals and estimation of self-fertilization rates. We also apply InStruct to estimate outcrossing rates and population substructure in a sample of Oryza rufipogon (n = 16), the wild relative of domesticated rice, for which sequence data across 106 unlinked proteincoding gene fragments is available. We find strong evidence for population structure in the sample with K = 2 populations providing a much better fit to the data than K = 1 population. Assignment of individuals recapitulates geographic sampling (and the results of STRUCTURE) with one cluster composed almost exclusively of Chinese orign O. rufipogon and another composed of samples from Nepal, India, and Laos.