Dispersal usually is considered an intrinsic species trait, invariant through time and independent of environmental factors. This assumption underlies many aspects of population biology and its application in assessments of global change. For example, dispersal is critical in determining the ecological impact of habitat loss, fragmentation, and climate change. My research shows that realized dispersal distances may depend on landscape configuration as well as intrinsic capabilities. A critical landscape measure is the connectivity between discrete habitat patches. I used spatial modeling to demonstrate that urban growth in some landscapes can increase habitat connectivity, despite the loss of a high proportion of habitat. Landscapes with a high degree of change in structural attributes such as connectivity, habitat area, and inter-patch distances, can exert high selective pressures on organisms. I used results from my simulations and evolutionary theory to structure a review of museum specimens of four butterfly species collected over the past 100 years in the San Francisco Bay area of California. I show that dispersal morphologies may have changed rapidly over this period. These changes are better explained by patterns of urban growth than trends in climatic variables such as temperature and precipitation. These findings combine to suggest that dispersal should be considered a dynamic trait that interacts with environmental change. Some species may be more vulnerable to extinction if climate change interacts with land use change to exacerbate the rate of change in landscape connectivity. Explicit consideration of this dynamic interaction will improve our understanding of the movement and persistence of species in human-dominated landscapes under climate variability and climate change.