The concepts of metapopulation persistence and source-sink dynamics are central and often untested tenets of marine reserve networks. Effective application of these concepts to marine reserves is limited by data on demographic rates within reserves and larval connectivity among reserves. Using a network of reserves established to restore eastern oyster (Crassostrea virginica) in Pamlico Sound, North Carolina, USA as our model system, we integrated empirically based demographic rates and regional hydrodynamic-based connectivity estimates within a metapopulation model to (1) evaluate the potential for the reserve network to function as a self-persistent oyster metapopulation, (2) determine the relative importance of demographics vs. connectivity on reserve source-sink dynamics, (3) assess the efficacy of stock enhancement in promoting metapopulation persistence, and (4) evaluate whether application of a Few Large or Several Small reserves-a modification of the Single Large or Several Small concept-promoted greater metapopulation retention of larvae within the reserve network (i.e., local larval retention within reserves + larval connectivity among reserves). The reserve network failed to function as a self-persistent metapopulation, despite exceptional demographic rates within reserves. When considering only larval supply from reserves, local retention and connectivity were insufficient to provide annual replacement (>=similar to 0.25 recruits per adult). Accordingly, reserves contributed to the metapopulation primarily via production, not dispersal, of larvae and reserve source-sink dynamics were influenced more by demographics within reserves, particularly adult growth and survival, than larval connectivity among reserves. Addressing recruitment limitation via stock enhancement was not effective at augmenting projected metapopulation declines. Several small reserves initially promoted greater metapopulation retention of larvae than a few large reserves, however, as the number or area of reserves increased, metapopulation retention was equivalent among designs, suggesting that Few Large and Several Small reserves may be the best network design. Marine reserves can be an effective restoration tool for improving demographic rates within reserve boundaries, but designation of multiple reserves does not guarantee a functional reserve network metapopulation. Both demographics and larval connectivity are important considerations to successful application of metapopulation concepts to the design of reserve networks.
