We present a study of the process whereby juveniles of a species disperse out of their natal landscape and recolonize rejuvenating landscapes nearby. This involves a significant number of long-distance dispersal events, which must occur, but which are difficult to measure in practise. Much previous research has used logical arguments coupled with mathematical models to derive plausible dispersal kernels for a variety of dispersal hypotheses. We turn this process on its head,

developing an agent-based model incorporating well-studied dispersal behaviour, and then observe the dispersal kernel that results under various landscape configurations. We find that while the decision rules of the dispersing agents does not change, the dispersal kernel changes significantly under different landscape configurations and agent densities. In particular, as the recolonization invasion proceeds, the dispersal kernel varies across the landscape and across time.