Understanding the variation in population abundance or density for species is almost always a factor in making sound management and conservation decisions, as well as in answering basic ecologic questions. To determine abundance, an increasing number of studies are using trapping arrays (e.g., hair snares, camera traps, mist-nests, etc.) for capture-recapture estimation of populations. A new class of models, spatially explicit capture recapture models, has emerged that make use of the spatial information from capture locations of individuals obtained by a trapping array. By assuming that each individual has an activity center, these models can account for heterogeneity in detection as a function of the distance from the activity center to a trap. While these models offer much promise, no empirical tests have been done to evaluate the estimates of these models, which may be biased.Here, we present a study on raccoons (Procyon lotor), where a known subset of the population (131 animals) was live captured and marked with numeric tags attached to a neck collar prior to the camera trapping study. Then, 20 camera traps were deployed on a 2859 hectare barrier island off the coast of North Carolina to photo-capture individuals.
Results/Conclusions
The relative bias for the estimated abundance is small, and the 95% credible interval includes the known 131 individuals. Based on this unique dataset, we provide an empirical evaluation of spatial capture recapture models and discuss interesting potential extensions to the model. These extensions include the incorporation of radio-telemetry data collected from 37 individuals in the population and modeling misclassified data (either from individuals incorrectly identified or individuals that were not marked and cannot be identified). The results of this study are widely applicable given the increasing number of studies using non-invasive capture recapture techniques, particularly on rare and elusive species.
