Coalescent-based phylogenetic inference from genes with unequivocal historical signal suggests a polytomy at the root of the placental mammal tree of life

Abstract

Finding the correct place of the root of the Placentalia tree of life is an unsolved problem in mammalian systematics. Three major competing hypotheses were proposed, alternating the position of the supraordinal taxa Afrotheria, Xenarthra, or Atlantogenata as sister lineages to the remaining placentals. While all three phylogenetic associations were inferred by studies conducted so far, recent assessments applying heterogeneous models and coalescent-based phylogenetic methods found robust support for the Atlantogenata hypothesis. Current developments in theoretical phylogenetics, such as the demonstration that concatenating genes may bias the likelihood function, and that summary coalescent-based phylogenetic methods are sensitive to errors in gene tree estimation, calls for a reevaluation of the early placental split problem. We investigated the phylogenetic relationships between placental superorders by restricting the analysis to subsets of genes with unequivocal phylogenetic signal. In contrast to previous works, we show that the hypothesis of a polytomy at the placental root could not be rejected under the multispecies coalescent model. This result endorses conclusions from analyses of retroposon insertion patterns. We provide an analytical framework to access empirical polytomies employing summary coalescent methods and topological tests, helping the investigation of poorly resolved nodes in the tree of life.