Consequences of modelling interpatch migration as a function of patch geometry when predicting metapopulation extinction risk

We have constructed a simulation model with demographic stochasticity for a generalised two-patch system. With the model, it was possible to vary the effect of patch geometry on interpatch migration from no effect, as in traditional models, to a gradually more pronounced effect. In systems with a constant amount of available habitat but with different allocation of habitat among patches we generally found lower metapopulation extinction risks if interpatch migration was adjusted to current sizes of local patches than if interpatch migration was assumed to be independent of patch sizes. In systems with equal patches, but with a different total amount of available habitat we found that simulations based on the assumption of patch-independent interpatch migration rates generate far too optimistic predictions of how well a metapopulation will persist when decreasing the total amount of habitat compared with simulations based on an adjustment of interpatch migration to current patch sizes.