Aim

For 80 years, popular opinion has held that most of Madagascar's terrestrial vertebrates arrived from Africa by transoceanic dispersal (i.e. rafting or swimming). We reviewed this proposition, focussing on three ad hoc hypotheses proposed to render this unlikely scenario more feasible: (a) Could hibernation have helped mammals to reach Madagascar? (b) Could the aquatic abilities of hippopotamuses have enabled them to swim the Mozambique Channel? (c) How valid is the Ali‐Huber model predicting that eastward Palaeogene surface currents allowed rafts to reach Madagascar in 3–4 weeks? Finally, we explored the alternative hypothesis of geodispersal via short‐lived land bridges between Africa and Madagascar.

Location

East Africa, Madagascar, Mozambique Channel.

Taxa

Fish, amphibians, reptiles, birds, mammals.

Methods

We established colonization timeframes using molecular divergence dates estimated for Malagasy vertebrate lineages. We reviewed the likelihood of the “torpid waif” and “swimming hippopotamus” hypotheses, and re‐investigated Ali and Huber's model of Eocene jet‐like currents by tracking particle trajectories in currents simulated using the Institut Pierre‐Simon Laplace Earth System Model. Finally, we summarized recent geological findings from the Mozambique Channel, and used them to compile palaeosedimentological maps using PLACA4D.

Results

Madagascar's vertebrate fauna has complex origins. Hibernation is probably an adaptation to Madagascar's hypervariable climate, rather than a facilitator of mammal dispersal. Hippopotamus physiology precludes the ability to cross an oceanic channel deeper than 4 m and hundreds of km wide. The Ali‐Huber model of Palaeogene currents considerably underestimated the time required to cross the Mozambique Channel under simulated palaeogeographic conditions. New geological data indicate the existence of three short‐lived land bridges between Africa and Madagascar at 66–60 Ma, 36–30 Ma and 12–05 Ma.

Main conclusion

The three Cenozoic land bridges afford a more grounded hypothesis for the dispersal of Madagascar's extant biota than transoceanic rafting or swimming, although vicariance, island hopping and limited rafting also played a role.