Phylogeny of the Rosidae: A dense taxon sampling analysis

Abstract

Rosidae, a clade of approximately 90 000 species of angiosperms, exhibits remarkable morphological diversity and extraordinary heterogeneity in habitats and life forms. Resolving phylogenetic relationships within Rosidae has been difficult, in large part due to nested radiations and the enormous size of the clade. Current estimates of phylogeny contain areas of poor resolution and/or support, and there have been few attempts to synthesize the available data into a comprehensive view of Rosidae phylogeny. We aim to improve understanding of the phylogeny of Rosidae with a dense sampling scheme using both newly generated sequences and data from GenBank of the chloroplast rbcL, atpB, and matK genes and the mitochondrial matR gene. We combined sequences from 9300 species, representing 2775 genera, 138 families, and 17 orders into a supermatrix. Although 59.26% of the cells in the supermatrix have no data, our results generally agree with previous estimates of Rosidae phylogeny and provide greater resolution and support in several areas of the topology. Several noteworthy phylogenetic relationships are recovered, including some novel relationships. Two families (Euphorbiaceae and Salvadoraceae) and 467 genera are recovered as non‐monophyletic in our sampling, suggesting the need for future systematic studies of these groups. Our study shows the value of a botanically informed bioinformatics approach and dense taxonomic sampling for resolving rosid relationships. The resulting tree provides a starting point for large‐scale analyses of the evolutionary patterns within Rosidae.