Phylogenetic informativeness analyses to clarify past diversification processes in Cucurbitaceae

Abstract

Phylogenomic studies have so far mostly relied on genome skimming or target sequence capture, which suffer from representation bias and can fail to resolve relationships even with hundreds of loci. Here, we explored the potential of phylogenetic informativeness and tree confidence analyses to interpret phylogenomic datasets. We studied Cucurbitaceae because their small genome size allows cost-efficient genome skimming, and many relationships in the family remain controversial, preventing inferences on the evolution of characters such as sexual system or floral morphology. Genome skimming and PCR allowed us to retrieve the plastome, 57 single copy nuclear genes, and the nuclear ribosomal ITS from 29 species representing all but one tribe of Cucurbitaceae. Node support analyses revealed few inter-locus conflicts but a pervasive lack of phylogenetic signal among plastid loci, suggesting a fast divergence of Cucurbitaceae tribes. Data filtering based on phylogenetic informativeness and risk of homoplasy clarified tribe-level relationships, which support two independent evolutions of fringed petals in the family. Our study illustrates how formal analysis of phylogenomic data can increase our understanding of past diversification processes. Our data and results will facilitate the design of well-sampled phylogenomic studies in Cucurbitaceae and related families.