Evolution of Migration Rate in a Spatially Realistic Metapopulation Model.

We use an individual-based, spatially realistic metapopulation model to study the evolution of migration rate. We first explore the evolutionary consequences of habitat change in hypothetical patch networks on a regular lattice. If the primary consequence of habitat change is an increase in local extinction risk due to decreased local population sizes, migration rate increases. A non-monotonic response, with migration rate decreasing at high extinction rate, was obtained only by assuming very frequent catastrophic extinctions. If the quality of the matrix habitat deteriorates, leading to increased mortality during migration, the evolutionary response is more complex. As long as habitat patch occupancy does not decrease markedly with increased migration mortality, reduced migration rate evolves. However, once mortality becomes so high that empty patches remain uncolonized for long time, evolution tends to increase migration rate, which may lead to an 'evolutionary' rescue in a fragmented landscape. We examined in detail the role of kin competition in explaining our results. Kin competition has a quantitative effect on the evolution of migration rate in our model but the patterns described above are primarily caused by spatio-temporal variation in fitness and mortality during migration. We apply the model to real habitat patch networks occupied by two checkerspot butterfly (Melitaea) species, for which sufficient data are available to rigorously estimate most of the model parameters. The migration rate predicted by the model is not significantly different from the empirically observed one. Regional variation in patch areas and connectivities leads to regional variation in the optimal migration rate, predictions which can be quantitatively tested with appropriate empirical data.