Statistical hybrid detection and the inference of ancestral distribution areas in Tolpis (Asteraceae)

Many historical biogeographic studies do not account for the effect of hybrid taxa on phylogenetic tree inference, despite recent advances in the statistical identification of such taxa. This investigation aims to illustrate the impact that hybrid taxa can have on ancestral area reconstructions of the plant genus Tolpis, which displays an evolutionary history possibly indicative of a back-colonization of the continental Mediterranean. We evaluate and apply two statistical hybrid detection methods, JML and STEM-hy, which assist in identifying reticulate patterns of allele coalescence. We also evaluate and apply a software tool, P2C2M, to test the fit of genetic loci to the multispecies coalescent model (MSCM). The application of these tools to a previously published data set of three nuclear DNA markers indicates the presence of several potential hybrid taxa in Tolpis. One nuclear marker is found to display a reduced level of reticulate history, a good fit to the MSCM and minimal signal of gene flow across archipelagoes. Ancestral distribution areas are reconstructed on gene and species trees of Tolpis before and after the exclusion of putative hybrid taxa using stochastic character mapping, parameterized likelihood reconstructions, and reconstructions under continuous-time Markov chain models. The results of these reconstructions indicate that taxa of hybrid origin may have a considerable impact on ancestral area reconstruction and that it is important to account for such taxa prior to biogeographic analysis. We conclude that Tolpis has likely had a time-consistent distribution in island habitats and originated on the Canary Islands.