Enhancing the area–isolation paradigm: habitat heterogeneity and metapopulation dynamics of a rare wetland mammal

Conservation of species in fragmented landscapes often is guided by spatially realistic metapopulation theory. However, convincing cases of metapopulation dynamics are uncommon, especially for vertebrates. Moreover, there is concern that the patch area and isolation paradigm for metapopulations is an oversimplification for heterogeneous landscapes. We tested predictions from metapopulation theory for a rare wetland mammal (round‐tailed muskrat, Neofiber alleni) and asked whether it was necessary to use a habitat‐informed version of the area–isolation paradigm that included patch quality and matrix heterogeneity. In each of two years, we surveyed 457 isolated wetlands in central Florida, USA, for presence–absence of Neofiber and evaluated logistic regression models of patch occupancy, extinction, and colonization. We documented metapopulation dynamics in which patch occupancy was constant between years (26% of patches occupied) due to balanced local extinctions (n = 45) and recolonizations (n = 46). Neofiber was both habitat and dispersal limited. Local extinctions were related negatively to patch area, patch quality (cover of maidencane grass, Panicum hemitomon), and distance to nearest roadside ditch. Patch colonization depended on patch area, patch quality, and spatial connectivity to potential source wetlands. Despite the importance of patch quality, Neofiber did not exhibit a habitat‐tracking metapopulation on an annual time scale. Cost–distance modeling suggested effective distances that included high costs for moving through forested matrix habitats generally were better than Euclidean distances for predicting patch colonization and occupancy. Two dominant land uses were tied to turnover dynamics: cattle grazing decreased habitat quality of wetlands, and presence of pine (Pinus spp.) plantations decreased functional connectivity. The simple area–isolation paradigm was not adequate for characterizing spatial dynamics of the Neofiber metapopulation. Nevertheless, we contend that the metapopulation approach remains a useful conservation framework for many species if landscape heterogeneity is embraced and explicit effects of land‐use practices on turnover processes are considered.