A covariant model of molecular and diversification patterns through time and the history of large clades

Abstract

AO_SCPLOWBSTRACTC_SCPLOWRate shifts in speciation and extinction have been recognised as important contributors to the creation of evolutionary patterns. In particular, the distribution of modern clade sizes is difficult to reconcile with models that do not include them. Although recent advances have allowed rate shifts to be integrated into evolutionary models, these have largely been for the purpose of inferring historical rate shifts across phylogenetic trees. In addition, these models have typically assumed an independence between patterns of diversification and rates of molecular and morphological evolution, despite there being mounting evidence of a connection between them. Here, we develop a new model with two principal goals: first, to explore the general patterns of diversification implied by constantly changing rates, and secondly to integrate diversification, molecular and morphological evolution into a single coherent framework. We thus develop and analyse a covariant birth-death process in which rates of all evolutionary processes (i.e. speciation, extinction and molecular and morphological change) covary continuously, both for each species and through time. We use this model to show that modern diversity is likely to be dominated by a small number of extremely large clades at any historical epoch; that these large clades are expected to be characterised by explosive early radiations accompanied by elevated rates of molecular evolution; and that extant organisms are likely to have evolved from species with unusually fast evolutionary rates. In addition, we show that under such a model, the amount of molecular change along a particular lineage is essentially independent of its height, which further weakens the molecular clock hypothesis. Finally, our model predicts the existence of "living fossil" sister groups to large clades that are both species poor and have exhibited slow rates of morphological and molecular change. Although our model is highly stochastic, it includes no special evolutionary moments or epochs. Our results thus demonstrate that the observed historical patterns of evolution can be modelled without invoking special evolutionary mechanisms or innovations that are unique to specific times or taxa, even when they are highly non-uniform: instead they could emerge from a process that is fundamentally homogeneous throughout time.