A phylogenetic study of evolutionary transitions in sexual systems in Australasian wurmbea (Colchicaceae)

Using phylogenies to make sound inferences about character evolution depends on a variety of factors, including tree uncertainty, taxon sampling, and the degree of evolutionary lability in the character of interest. We explore the effect of these and other sources of ambiguity for maximum likelihood (ML)-based inferences of sexual-system evolution in Wurmbea, a small genus of geophytic monocots from the Southern Hemisphere. We reconstructed Wurmbea phylogeny using four noncontiguous regions (ca. 5.5 kb) of the plastid genome across a broad sampling of taxa, and we confirm that the genus is divided into two well-supported clades, each defined by its geography (Africa vs. Australasia) and variation in sexual system (i.e., uniformly monomorphic vs. sexually variable, respectively). We demonstrate that the predominantly Australian clade includes the sexually monomorphic species Iphigenia novae-zelandiae. We observe treewide uncertainty in the state of all ancestral nodes, and therefore all state transitions, when all taxa in Wurmbea are considered. We demonstrate that this is primarily a consequence of interspersion of terminals with gender dimorphism vs. monomorphism throughout the Australasian clade, rather than tree uncertainty or the presence of very short internal branches. We accounted for tree uncertainty by randomly sampling alternative resolutions of branches that are poorly supported by ML bootstrap analysis, effectively interpreting these as soft polytomies. Under the assumption that well-supported aspects of our gene tree accurately depict organismal phylogeny, there is a marked evolutionary lability in the sexual systems of Australasian Wurmbea. A more problematic issue is that our results contradict the monophyly of two sexually polymorphic Australian species, Wurmbea dioica and Wurmbea biglandulosa. If this reflects paraphyly at the species level, lateral gene transfer, or failed coalescence, then the interpretations of character transitions will need to be adjusted. Our analysis provides an example of the impediments to linking macroevolutionary pattern with microevolutionary processes for evolutionarily labile traits in recently evolved plant groups that possess a high degree of variation in sexual characters.