An empirical evaluation of the area and isolation paradigm of metapopulation dynamics

Much of metapopulation theory assumes that the persistence of individual populations in a metapopulation, and persistence of the metapopulation as a whole, is best modeled by the area of habitat patches and their isolation. Estimates of isolation typically include a measure of geographic distance and a measure of either population size or patch area. This “area and isolation paradigm” assumes a functional relationship between the area of a patch and its extinction probability, and between isolation of a patch and its colonization probability. Although these assumptions are fundamental to use of incidence function models of metapopulation dynamics, the assumptions have been validated in only a small number of studies. We tested the ability of area and isolation to predict extinction and colonization patterns using multiple-year occupancy data for 10 species from three taxonomic groups (butterflies, amphibians, and birds). We examined 13 potential models of metapopulation dynamics. All models included four basic parameters: occupancy during the first year of the survey, probability of extinction, probability of colonization, and single-visit detection probability. In eight models, each parameter was either constant or time-dependent. Five models included a patch-level covariate of extinction probability (patch area or population size), colonization probability (connectivity, the inverse of isolation), or both. Extinction patterns generally were predicted more effectively as a function of local population size than as a function of patch area, a constant probability of extinction, or a time-dependent probability of extinction. In most cases, inclusion of connectivity as a patch-level covariate did not improve predictions of colonization patterns. We estimated single-visit detection probabilities for all species in our analyses, thus providing evidence-based guidelines for the refinement of future monitoring protocols.