Bayesian Estimation of Genomic Clines

We developed a Bayesian genomic cline model to study the genetic architecture of adaptive divergence and reproductive isolation between hybridizing lineages. This model quantifies locus-specific patterns of introgression with two cline parameters that describe the probability of locus-specific ancestry as a function of genome-wide admixture. 'Outlier' loci with extreme patterns of introgression relative to most of the genome can be identified. These loci are potentially associated with adaptive divergence or reproductive isolation. We simulated genetic data for admixed populations that included neutral introgression, as well as loci that were subject to directional, epistatic or underdominant selection, and analysed these data using the Bayesian genomic cline model. Under many demographic conditions, underdominance or directional selection had detectable and predictable effects on cline parameters, and 'outlier' loci were greatly enriched for genetic regions affected by selection. We also analysed previously published genetic data from two transects through a hybrid zone between Mus domesticus and M. musculus. We found considerable variation in rates of introgression across the genome and particularly low rates of introgression for two X-linked markers. There were similarities and differences in patterns of introgression between the two transects, which likely reflects a combination of stochastic variability because of genetic drift and geographic variation in the genetic architecture of reproductive isolation. By providing a robust framework to quantify and compare patterns of introgression among genetic regions and populations, the Bayesian genomic cline model will advance our understanding of the genetics of reproductive isolation and the speciation process.