Meeting Abstract
Polygenic selection (selection on multiple loci contributing to the same phenotype) is expected to result in covariance of allelic states among phenotype-associated variants; in other words, it should give rise to linkage disequilibrium (LD) among sites in the absence of physical linkage between them. Here, I used this idea to identify sites under selection against introgression between ecomorphs of the great star coral (Montastrea cavernosa) from the Florida Keys. This species comes in four genetically distinct but hybridizing ecomorphs: two of them are specialized for deep reef habitat (>15 m) but are sympatric with respect to each other, the third is offshore-specialized, and the fourth one is inshore-specialized. To identify co-varying groups of sites, I adopted the Weighted Gene Coexpression Networks Analysis (WGCNA) to analyze the matrix of allelic correlations (pairwise r-square estimated with the EM algorithm using ngsLD software). WGCNA identified five groups of covarying sites (“modules”), four of them capturing derived alleles resisting introgression in each of the four ecomorphs, and the fifth one capturing sites that are selected for the ancestral state in the deep habitat. The new approach (“LD networks”) inherits the main advantages of WGCNA: (1) it is model-free, meaning that it does not require the researcher to specify which samples correspond to which deme or environmental condition; (2) post hoc, it allows simultaneous evaluation of large number of factors for possible role in shaping polygenic selection signatures.
