A BAYESIAN STATE-SPACE FORMULATION OF DYNAMIC OCCUPANCY MODELS

摘要: Species occurrence and its dynamic components, extinction colonization probabilities, are focal quantities in biogeography metapopulation biology, for species conservation assessments. It has been increasingly appreciated that these parameters must be estimated separately from detection probability to avoid the biases induced by non-detection error. Hence, there is now considerable theoretical practical interest occupancy models contain explicit representations of dynamics such as extinction, colonization, turnover well growth rates. We describe a hierarchical parameterization analogous state-space formulation time series, where model represented two one partially observable process another observations conditional on process. This naturally allows estimation all conventional approach models, but addition, yields great flexibility extensibility, e.g., modeling heterogeneity or latent structure parameters. also highlight important distinction between population finite sample inference; latter much more precise estimates particular at hand. Finite can easily obtained using representation difficult obtain under likelihood-based estimation. use R WinBUGS apply examples. In standard analysis European Crossbill large Swiss monitoring program, we fit with year-specific Estimates varied greatly among years, highlighting irruptive species. second example, analyze route Cerulean Warblers North American Breeding Bird Survey (BBS) allowing site-specific The results indicate relatively low stable distribution which contrast analyses counts individuals same survey declines. discrepancy illustrates inertia relative actual abundance. Furthermore, reveals declining patch survival probability, increasing turnover, toward edge range species, consistent perspectives genesis edges. Given detection/non-detection data, described here have potential study distributions dynamics.