OOS 4-8
Modeling density in stratified populations using hierarchical spatial capture-recapture

Monday, August 10, 2015: 4:00 PM
316, Baltimore Convention Center
Sarah J. Converse, Patuxent Wildlife Research Center, US Geological Survey, Laurel, MD, USA
J. Andrew Royle, , USGS Patuxent Wildlife Research Center, Laurel, MD, USA

Understanding how the abundance and dynamics of populations respond to factors such as habitat quality, management actions, or disturbances is of primary interest to ecologists. Investigating these responses frequently involves implementation of capture-recapture experiments that are replicated in time and space. However, when capture methods are passive, there is a question about the population under study. Because of animal movements, capture grids sample an area that is larger than the grid itself. The potential for differential movements across replicates can lead to a problem whereby inference based on abundance estimates is dubious. We developed a spatial capture-recapture model of general use in investigating factors influencing variation in population density over time and space. The model links spatial and temporal replicates of capture-recapture experiments using hierarchical structures.  The analysis requires Bayesian methods to achieve the inclusion of individual effects in the spatial capture-recapture models; input data are spatially-referenced individual encounter histories. We applied the model to data from live-trapping grids implemented as part of an experiment to examine the response of small mammal populations to forest fuel reduction treatments and natural disturbances. 


Using the model we constructed, we confirmed previously-documented positive response of Peromyscus spp. to both thinning and wildfire; in particular, densities showed a strongly positive response to both disturbance types. Our model addresses the need for an analytical approach allowing estimation and modeling of the population state variables, i.e., densities, arising from capture-recapture experiments, while accounting for the differential movements of individuals. The approach we demonstrate is synthetic, in that densities are both estimated and modeled in a single integrated analysis. Our model provides a path to robust inference about density estimates from spatially- and temporally-replicated capture-recapture experiments.