Coastal habitats have recently received much attention from policymakers, but marine reserve theory still needs to integrate across scales, from local dynamics of communities to biogeographic patterns of species distribution, recognizing coastal ecosystems as complex adaptive systems in which local processes and anthropogenic disturbances can result in large-scale biological changes. We present a theoretical framework that provides a new perspective on the science underlying the design of marine reserve networks. Coastal marine systems may be usefully considered as metacommunities in which propagules are exchanged among components, and in which the persistence of one species depends on that of others. Our results suggest that the large-scale distribution of marine species can be dynamic and can result from local ecological processes. We discuss the potential implications of these findings for marine reserve design and the need for long-term monitoring programs to validate predictions from metacommunity models. Only through an integrated and dynamic global perspective can scientists and managers achieve the underlying goals of marine conservation.