Aim To investigate the validity of Simpson's model of sweepstakes dispersal, particularly as it applies to the colonization of Madagascar by African mammals. We chose lemurs as a classic case.
Location The East African coast, the Mozambique Channel and Madagascar.
Methods First, we investigated the assumptions underlying Simpson's statistical model as it relates to dispersal events. Second, we modelled the fate of a natural raft carrying one or several migrating mammals under a range of environmental conditions: in the absence of winds or currents, in the presence of winds and currents, and with and without a sail. Finally, we investigated the possibility of an animal being transported across the Mozambique Channel by an extreme climatic event like a tornado or a cyclone.
Results Our investigations show that Simpson's assumptions are consistently violated when applied to scenarios of over-water dispersal by mammals. We suggest that a simple binomial probability model is an inappropriate basis for extrapolating the likelihood of dispersal events. One possible alternative is to use a geometric probability model. Our estimates of current and wind trajectories show that the most likely fate for a raft emerging from an estuary on the east coast of Africa is to follow the Mozambique current and become beached back on the African coast. Given prevailing winds and currents, transport from Madagascar to Africa is very much more likely than the reverse process. Freak transport by means of a hurricane or tornado is even less likely than rafting for mammals.
Main conclusions Our models suggest that the scenario of sweepstakes dispersal that currently enjoys wide support is not valid at either the theoretical or the applied level when applied to the hypothetical invasion of Madagascar by African mammals. Alternative explanations should be sought.