Pliocene forest dynamics as a primary driver of African bird speciation

Aim Montane tropics are areas of high endemism, and mechanisms driving this endemism have been receiving increasing attention at a global scale. A general trend is that climatic factors do not explain the species richness of species with small to medium-sized geographic ranges, suggesting that geological and evolutionary processes must be considered. On the African continent, several hypotheses including both refugial and geographic uplift models have been advanced to explain avian speciation and diversity in the lowland forest and montane regions of central and eastern Africa; montane regions in particular are recognized as hotspots of vertebrate endemism. Here, we examine the possible role of these models in driving speciation in a clade of African forest robins. Location Africa. Methods We constructed the first robustly supported molecular phylogenetic hypothesis of forest robins. On this phylogeny, we reconstructed habitat-based distributions and geographic distributions relative to the Albertine Rift. We also estimated the timing of lineage divergences via a molecular clock. Results Robust estimates of phylogenetic relationships and clock-based divergences reject Miocene tectonic uplift and Pleistocene forest refugia as primary drivers of speciation in forest robins. Instead, our data suggest that most forest robin speciation took place in the Late Pliocene, from 3.2 to 2.2 Ma. Distributional patterns are complex, with the Albertine Rift region serving as a general east-west break across the group. Montane distributions are inferred to have evolved four times. Main conclusions Phylogenetic divergence dates coincide with a single period of lowland forest retraction in the late Pliocene, suggesting that most montane speciation resulted from the rapid isolation of populations in montane areas, rather than montane areas themselves being drivers of speciation. This conclusion provides additional evidence that Pliocene climate change was a major driver of speciation in broadly distributed African animal lineages. We further show that lowland forest robins are no older than their montane relatives, suggesting that lowland areas are not museums which house 'ancient' taxa; rather, for forest robins, montane areas should be viewed as living museums of a late Pliocene diversification event. A forest refugial pattern is operating in Africa, but it is not constrained to the Pleistocene.