Sources of Gene Tree Discordance and Their Implications for Systematics and Evolution of a Megadiverse Australian Plant Radiation (Subtribe Hakeinae, Proteaceae)

Abstract

Resolving phylogenetic relationships in the presence of conflicting signal across genes is one of the major challenges of the phylogenomic era. Conflicting signal can emerge from biological processes, such as incomplete lineage sorting or introgression, or have technical origins, such as from misaligned sequences. Decisions made in the process of estimating species trees may therefore result in alternative tree topologies and large variation in branch support values with important taxonomic consequences. To explore how these methodological choices affect the estimation of relationships, we compare alternative strategies for alignment cleaning, loci filtering, and phylogenetic estimation for 551 taxa in the Proteaceae subtribe Hakeinae. We found that node support values across gene trees were generally low and gene discordance was high in the Hakeinae, and that the degree of discordance varied across the phylogeny, as well as geographically across Australia. Higher stringency of alignment cleaning tended to decrease node support, while removing undesirable loci tended to increase gene concordance. Cleaning, filtering, and phylogenetic estimation method (short-cut coalescent vs. concatenation) have significant effects on tree topologies with distinct clusters of similar topologies detected in tree space. In some cases, this has important systematic consequences for two of Australia’s largest and best-known plant genera, with concatenated approaches resolving Hakea and Grevillea as reciprocally monophyletic, but coalescent approaches showing that Hakea is nested within Grevillea. Our results suggest that widespread gene discordance may be the result of rapid radiation and incomplete lineage sorting, demonstrating the importance of assessing the drivers of discordance to understand phylogenetic relationships.