LAUSR

Aim: The expansion of open habitats during the mid-Miocene has been hypothesized as a driver of allopatric speciation for many African taxa. This habitat-dependent mode of diversification has been implicated in the shift from C3 (e.g. forest/woodland) to C4 dominated systems (i.e. open savanna, grasslands) in a number of African squamates. We examined this hypothesis using a genus of African viperid snakes (Bitis) with both open habitat and forest-dwelling representatives. Location: Africa. Methods: A comprehensive multilocus dataset was used to generate a calibrated species tree using a multispecies coalescent model. Individual gene trees and patterns of nuclear allele sharing were used to assess species monophyly and isolation. To test the habitat-dependent evolution hypothesis, we generated an ancestral character state reconstruction for open and closed habitats using the dated phylogeny. This was related to the timing of open habitat expansion and forest/woodland contraction in Africa. Results: The genus Bitis originated in the Oligocene, with species level diversification in the late Miocene/Pliocene. Four well-supported clades correspond to the recognized subgenera Bitis, Keniabitis, Macrocerastes and Calechidna. Several previously unrecognized lineages potentially represent cryptic species. Main conclusions Habitat-dependent evolution does not appear to have been a main driver for generic level viperine diversification: the ancestral state for Bitis was open habitat and at least one clade moved into forest in the Miocene, long after forest had contracted and fragmented. Forest-dependent species diversified only in the late Miocene, presumably as forest became further reduced in extent, fitting an allopatric model of speciation. Although our results do not favour a general pattern of habitat-dependent diversification in Bitis, cladogenesis within the subgenus Calechidna for “arenicolous” species (Bitis caudalis complex) and “rupicolous” species (B. Atropos-cornuta complex), corresponds to the aridification of southwest Africa. This suggests there are subtleties not captured in the broad open habitat category, which are relevant for understanding the role of habitat-dependent evolution.