Abstract
Hybridisation is an important biological force that shapes phylogenies. However, hybridisation is often neglected in phylogenetics because it violates the assumption of ever-diverging, tree-like evolution that is required by most evolutionary models. Here we present a new coalescent-based approach for generating species phylogenies that circumvents the tree assumption. The method estimates speciation times for pairs of species (or other OTUs) from a large set of gene coalescent times, by use of a Bayesian hierarchical model under the multispecies coalescent. This approach is suitable for large, phylogenomic data because gene trees are calibrated individually, rather than jointly in a fully parametrised analysis, and because each speciation event is estimated independently from the distribution of gene coalescent times. The output is a pairwise distance matrix for species (or other OTUs), based on which the species phylogenetic network can be estimated and its deviation from the species tree assumption tested.
We applied this approach on the IWGSC genome assemblies of eight species/genome lineages of wheat (Triticum/Aegilops) using a genome-wide sample of 275 genes. We dated the crown node of wheat, i.e. the divergence of the A- and B-genome lineages, to ~7 Ma ago. Two discrete, successive reticulations were detected and interpreted as homoploid hybrid speciations: one involving the A- and B- lineages ~6 Ma ago and giving rise to the D-genome lineage, and a second involving the B- and D-lineages ~2 Ma ago giving rise to the Ae. sharonensis lineage. Furthermore, our results shed light on the timing of the polyploidisations in the hexaploid bread wheat (T. aestivum) and support the view that this species arose with agriculture ~10 ka ago.
Thomas Marcussen
Department of Plant Sciences, Norwegian University of Life Sciences, 1432 Ås, Norway