Print PageMetapopulation theory has commonly been used to address conservation questions, especially in view of increasing loss, alteration, and fragmentation of habitat due to land use change that can jeopardize the persistence of species. Crucial to metapopulation dynamics and species persistence is connectivity among habitat patches on the landscape, which is necessary for rescue or recolonization of populations following local extinction. Graph theoretic approaches are increasingly being utilized to answer questions concerning connectivity and flow among habitat patches and populations. Metapopulations of amphibians are often viewed from a ponds-as-patches perspective. Taking this perspective, we studied wood frogs in east-central Missouri where they are at the edge of their Midwest distribution, and are a species of conservation concern. Previous research has shown that Missouri wood frog populations are stochastic in time and space. Using 5 years of egg mass survey data, we assessed the demographic connectivity, spatial structure, and breeding pond persistence of wood frogs in detail at a conservation area in east-central Missouri using a combination of graph theory and occupancy modeling. We constructed a demographic network model for each season as well as an average model for the 5 years. Based on the average network model we designated ponds as sources, sinks, or pseudo sinks.
Results/Conclusions
We found that 34 ponds have been used for breeding since 2005. Connectivity among ponds was generally high, but decreased drastically in drought years. Of the 34 breeding ponds, only 15 harbor populations large enough to serve as sources. The remainder of the ponds are assumed to act as sinks or pseudo-sinks, whereby populations are generally too small or too isolated to contribute individuals to other breeding ponds. Using multi-season occupancy models, we estimated pond-specific probabilities of persistence over the 5-year study period as a function of environmental and site covariates, and found persistence to be significantly influenced by the average level of self-recruitment at a pond. Pond self-recruitment is a component of our population graph analysis, and is a function of average number of clutches per year and spatial proximity to other breeding ponds. Thus we have linked a spatially-explicit connectivity model with a temporal colonization-persistence model. Our results provide working models for interpreting patterns of occupancy and gene flow, and can serve as an initial guide for future habitat management and restoration for this vulnerable species in Missouri.
