The colonisation of Madagascar by land-bound vertebrates.

Despite discussions extending back almost 160 years, the means by which Madagascar's iconic land vertebrates arrived on the island remains the focus of active debate. Three options have been considered: vicariance, range expansion across land bridges, and dispersal over water. The first assumes that a group (clade/lineage) occupied the island when it was connected with the other Gondwana landmasses in the Mesozoic. Causeways to Africa do not exist today, but have been proposed by some researchers for various times in the Cenozoic. Over-water dispersal could be from rafting on floating vegetation (flotsam) or by swimming/drifting. A recent appraisal of the geological data supported the idea of vicariance, but found nothing to justify the notion of past causeways. Here we review the biological evidence for the mechanisms that explain the origins of 28 of Madagascar's land vertebrate clades [two other lineages (the geckos Geckolepis and Paragehyra) could not be included in the analysis due to phylogenetic uncertainties]. The podocnemid turtles and typhlopoid snakes are conspicuous for they appear to have arisen through a deep-time vicariance event. The two options for the remaining 26 (16 reptile, five land-bound-mammal, and five amphibian), which arrived between the latest Cretaceous and the present, are dispersal across land bridges or over water. As these would produce very different temporal influx patterns, we assembled and analysed published arrival times for each of the groups. For all, a 'colonisation interval' was generated that was bracketed by its 'stem-old' and 'crown-young' tree-node ages; in two instances, the ranges were refined using palaeontological data. The synthesis of these intervals for all clades, which we term a colonisation profile, has a distinctive shape that can be compared, statistically, to various models, including those that assume the arrivals were focused in time. The analysis leads us to reject the various land bridge models (which would show temporal concentrations) and instead supports the idea of dispersal over water (temporally random). Therefore, the biological evidence is now in agreement with the geological evidence, as well as the filtered taxonomic composition of the fauna, in supporting over-water dispersal as the mechanism that explains all but two of Madagascar's land-vertebrate groups.