Can parallel ecological speciation be detected with phylogenetic analyses?

Abstract

Distinguishing parallel divergence from vicariance scenarios is crucial to establish the determinism of natural selection in the formation of new species. It has been proposed that phylogenetic methods can be used to disentangle a single origin in allopatry and multiple origins in sympatry for ecological speciation. However, a key issue is to what extent introgression in a patchy environment may complicate the distinction between both origins through the analysis of variation at neutral markers. While divergence at environmentally-based selected loci retains the initial correlation with ecological variables, such association may be lost at neutral loci unlinked to any selected locus. Thus, neutral divergence might reflect in the long-term the molecular fingerprint of isolation by distance regardless of the model of speciation considered, and a question arises as to whether phylogenetic analyses of neutral markers are able or not to retrieve the signals acquired in the ancestral populations. Here, we use computer simulations to show that the detection of the original signal using a phylogenetic method strongly depends on the migration rates among populations. Recombination accelerates the loss of the initial phylogenetic signal, but this effect is rather small compared with the effect of migration, and only substantial when recombination is very large. For model species with reduced gene flow between distant populations and between populations adapted to different habitats, the phylogenetic approach is able to distinguish a single origin in allopatry from multiple origins in sympatry.