Using Nuclear Genomic Data to Address Intractable Relationships and Gene Tree Discordance in an Ancient Group of Gymnosperms ( <i>Ephedra</i> , Gnetales)

ABSTRACT The evolutionary history of Ephedra L. has over time proved to be a very difficult phylogenetic problem, something which probably is best reflected by the incongruent phylogenetic relationships demonstrated in and among previous studies. A low number of phylogenetically informative sites and different rooting strategies have been suggested as explanations, but the role of gene tree discordance and its underlying causes and influence on phylogenetic reconstruction in Ephedra have largely remained unexplored. In the present study, we address the phylogeny of Ephedra using information from a large set of nuclear low‐copy genes obtained with a specifically designed bait set. We explore the potential effect of putative paralogy, intra‐locus recombination, and chimeric sequences on phylogenetic estimation in Ephedra , while striving to minimize the risk for gene tree error. Our results indicate that recombination and putatively paralogous sequences have limited influence on phylogenetic results, although handling of putative paralogs and recombinants did improve overall species tree resolution and statistics. A robust Ephedra species phylogeny was obtained by analyzing gene trees, in which orthologous clades were inferred from data composed of orthologous and putatively paralogous sequences. The results support the division of Ephedra into three geographically defined clades, with the American clade as sister to the Mediterranean clade and the Asian clade. Relationships among species within these three clades were relatively consistent among our analytical approaches, but our results highlighted moderate to strong signals of gene tree discordance at many branches including the deepest split. Incomplete lineage sorting is a possible explanation for this discord for many branches, but hybridization/introgression is also commonly indicated.