Climate change presents a potentially severe threat to biodiversity. Species will be required to disperse rapidly through fragmented landscapes in order to keep pace with the changing climate. An important challenge for conservation is therefore to manage landscapes so as to assist species in tracking the environmental conditions to which they are adapted. Here we develop a stochastic spatially explicit model to simulate plant dispersal across artificial fragmented landscapes. Based on certain assumptions as to the dispersal mechanism, we assess the impact that varying potential for rare long-distance dispersal (LDD) has on the ability to move over landscapes with differing spatial arrangements of suitable habitat (clumped versus fragmented). Simulations demonstrate how the relative importance of landscape structure in determining migration ability may decrease as the potential for LDD increases. Thus, if LDD is the principal mechanism by which rapid large-scale migrations are achieved, strategically planned networks of protected habitat may have a limited impact on rates of large-scale plant migrations. We relate our results to conventional principles for conservation planning and the geometric design of reserves, and demonstrate how reversal of these principles may maximise the potential for conservation under future climates. In particular, we caution against the justification of large-scale corridors on grounds of climate change since migration along corridors by standard dispersal mechanisms is unlikely to keep pace with projected change for many species. An improved understanding of the dispersal mechanisms by which species achieve rapid migrations, and the way that these processes are affected by patterns of landscape fragmentation, will be important to inform future conservation strategies.