Maintenance of Species Boundaries Despite Ongoing Gene Flow in Ragworts

Abstract

The role of hybridization between diversifying species has been the focus of a huge amount of recent evolutionary research. While gene flow can prevent speciation or initiate species collapse, it can also generate new hybrid species. Similarly, while adaptive divergence can be wiped out by gene flow, new adaptive variation can be introduced via introgression. The relative frequency of these outcomes, and indeed the frequency of hybridization and introgression in general are largely unknown. One group of closely-related species with several documented cases of hybridization is the Mediterranean ragwort (genus: Senecio) species-complex. Examples of both polyploid and homoploid hybrid speciation are known in the clade, although their evolutionary relationships and the general frequency of introgressive hybridization among them remain unknown. Using a whole genome gene-space dataset comprising eight Senecio species we fully resolve the phylogeny of these species for the first time despite phylogenetic incongruence across the genome. Using a D-statistic approach, we demonstrate previously unknown cases of introgressive hybridization between multiple pairs of taxa across the species tree. This is an important step in establishing these species as a study system for diversification with gene flow, and suggests that introgressive hybridization may be a widespread and important process in plant evolution.

Keywords: clade diversification; hybridization; introgression; phylogenetic incongruence; speciation with gene flow.

Fig. 1.—

Phylogenetic reconstruction and gene tree-species tree incongruence. (A) Species phylogeny estimated using RAxML (Stamatakis 2014) and ASTRAL (Mirarab et al. 2014), both of which produced the same topology. Branch lengths are the mean of those produced using RAxML on both reference-based datasets. For each node, bootstrap support from the method of Mirarab et al. (2014) for the S. flavus and S. madagascariensis reference-based assemblies are shown before and after the bar, respectively. (B) A DensiTree plot of ML gene trees for all contigs. For each gene tree, nodes with <75% bootstrap support are collapsed and gene trees which subsequently contain more than two polytomies are excluded. For each unique topology among gene trees, branch lengths are averaged among all gene trees showing that topology. Results using trees from both reference-based assemblies combined are also shown.

Fig. 2.—

The scenario explaining the results of introgression analyses in this article as well as those in previous study (Osborne et al. 2013) which requires the fewest number of episodes of introgression. Green arrows represent introgression events and letters refer to table 2. Branch lengths are arbitrary. See table 2 for justification.