Patch centrality affects metapopulation dynamics in small freshwater ponds

Despite advances in metapopulation theory, recent studies have emphasized the difficulty in understanding and accurately predicting dynamics in nature. We address this knowledge gap by coupling 4 years of population data for the freshwater zooplankter Daphnia pulex, inhabiting 38 newly established ponds in Upstate New York, with (i) a spatially explicit stochastic model and (ii) a deterministic model where we have averaged the spatial dependencies. By modifying the identity of ponds stocked/removed in our model, we examine the effects of network structure on metapopulation dynamics and local occupancy patterns. From these modeling exercises, we show that the centrality of ponds (stocked or removed) has contrasting effects on metapopulation persistence when selecting ponds to initially stock versus preserve. The pond network was not robust to the removal of centrally located ponds as the simulated removal of these ponds resulted in rapid collapse of the metapopulation. However, when initially founding a metapopulation, the location of patches did not influence occupancy dynamics. Because stochastic simulations can be computationally expensive, we also introduce a quantity for use in a simple differential equation model that encompasses all spatial information in a single variable. Using this quantity, we show how the output of our simple differential equation model matched the quasi-steady state of the stochastic simulations in networks characterized by high connectivity. The method we use is general enough to be applied in other systems and can provide insights for habitat conservation and restoration efforts including how network structure can drive spatiotemporal metapopulation dynamics.