Spatially explicit metapopulation models for informing conservation
Howell, Paige Elizabeth
MetadataShow full item record
For many species, populations exist as highly fragmented subpopulations linked by dispersal. To manage for long-term metapopulation viability effectively, information is needed about the factors influencing local subpopulation dynamics and connectivity among subpopulations. The objectives of my dissertation were to 1) improve the linkages between metapopulation and landscape ecology by developing spatially-explicit dynamic metapopulation models allowing for inference about local and landscape-level processes, 2) expand on existing metapopulation models by modeling spatio-temporal variation in density, 3) evaluate hypotheses regarding the effects of patch hydroperiod, landscape structure, and density-dependence on metapopulation dynamics using statistical models, and 4) provide management recommendations to enhance the viability of the Chiricahua leopard frog (Lithobates chiricahuensis). Colonization rate was influenced by patch hydroperiod, elevation and the spatial distribution of streambeds. Patch-specific growth rates were density-dependent and influenced by hydroperiod. The proportion of occupied ponds increased initially from the reintroduction of tadpoles into three ponds in 2003 to 18 (95% CI; 12, 33) of the 274 available ponds occupied in 2017. Metapopulation extinction risk over a 25-yr time horizon (2018-2043) with static environmental conditions was predicted to be low (7%) if invasive predator control continues and permanent ponds are maintained. However, under a scenario of increasing drought conditions, extinction risk is substantially higher, particularly in the most pessimistic scenario where some ponds fail and there is no management (40%). Results from my dissertation illustrate the utility of spatially-explicit statistical models for understanding the processes underlying metapopulation dynamics and forecasting metapopulation viability, while formally accounting for uncertainty.