Metapopulation Dynamics of a Marine Reserve Network: Interacting Effects of Demography and Connectivity

Population dynamics are governed by four demographic rates: births, deaths, immigration, and emigration. Populations of many species exist as spatially separated subpopulations that are connected by migration (e.g., larval dispersal), forming a metapopulation. Understanding how these four demographic rates and their interactions drive metapopulation dynamics is the focus of this dissertation, with application to assessing the efficacy and design of a network of no-take marine reserves. Marine reserve networks, multiple reserves connected by larval dispersal, have proliferated globally in response to declining fisheries and loss of biodiversity. The metapopulation concept is a central tenet of effective reserve networks; however, limited data on spatiotemporally explicit demographic rates and connectivity of target species often precludes application of this concept to assess and design reserve networks.;Using a network of marine reserves designed to restore subpopulations of eastern oyster (Crassostrea virginica) in Pamlico Sound, North Carolina, USA as a model system, I first quantified spatiotemporal variation in oyster demographic rates---recruitment, growth, and survival---within reserves. From 2006 to 2008, average oyster recruitment and total density increased fifteen- and five-fold, respectively, supporting the ability of reserves to rapidly increase density of protected species. The unprecedented high oyster densities in certain reserves (up to 6,500/m2) modified demographic rates such that further density increases may be regulated by density-dependent survival. Oyster demographic rates varied significantly among reserves. Certain reserves were the strong "recruiters", others the fast "growers", and yet others the high "survivors". This demographic mosaic, which may serve as a metapopulation bet hedging strategy buffering biotic and abiotic variability, highlights the need for spatially explicit demographic data to support varying management objectives.;Patterns of larval dispersal and connectivity, and their drivers were quantified using a biophysical model. The location (i.e., natal reserve) and timing of spawning relative to physical processes, particularly frequency of wind reversals, were the dominant drivers of larval dispersal and reserve connectivity. To a lesser extent, larval behavior and mortality modified dispersal and connectivity. Over a 21 day larval duration, particles dispersed a mean distance of 2 to 75 km over an area covering 2 to 471 km2. Local retention of passive surface particles was typically small in magnitude (median <1%) such that immigration exceeded local retention (i.e., reserves demographically "open"). Over 5 years, ~40% of the 90 possible inter-reserve connections occurred, but the magnitude of connections was highly variable and often asymmetrical. The presence of spatiotemporal variation in adult demographic rates and connectivity among reserves suggests that this reserve network is particularly amenable to application of metapopulation concepts.;Oyster demographic rates and larval connectivity were integrated within a metapopulation matrix model to (1) assess reserve network self-sustainability, (2) quantify reserve relative importance to the network (i.e., source-sink status), (3) assess the efficacy of stock enhancement to improve network sustainability, and (4) evaluate whether increasing the number or size of reserves in the network promoted greater network connectivity. Source reserves provided a metapopulation 'rescue effect' whereby larval subsides from sources increased the population size of the subsidized reserve(s), but the effect was insufficient for network sustainability. The inability of this reserve network to be self-sustaining despite stock enhancement was due primarily to limited connectivity. Increasing the number of reserves in the network tended to promote greater connectivity, although as the number and size of reserves in the network continued to expand, larval connectivity was equivalent, suggesting that a Few Large and Several Small reserves may be the optimal network design. Marine reserves can be effective at improving demographic rates, but designation of multiple reserves does not guarantee a self-sustaining network without consideration of metapopulation dynamics in the design.