Hierarchical Spatial Capture–Recapture Models for Estimating Density from Trapping Arrays

Abstract

Much of the theory and methodology underlying inference about population size is concerned with populations that are well-defined in the sense that one can randomly sample individuals associated with some location or area and, usually, uniquely identify them. However, individuals within populations are spatially organized; they have home ranges or territories, or some sense of “place,” within which they live and move about. The juxtaposition of this place with a trap or array of traps has important implications for sampling design, modeling, estimation and interpretation of data that result from trapping data. In particular, this juxtaposition induces two general problems. First, for most populations, the spatial area over which individuals exist (and are exposed to capture) cannot be precisely delineated, and movement of individuals onto and off of a putative sample unit results in a form of non-closure, which has a direct effect on our ability to interpret the estimates of population size, N, from closed population models. The second problem is that this juxtaposition induces heterogeneity in capture probability as a result of variable exposure of individuals to capture. Certain individuals, e.g., those with territories on the edge of a trapping array, might experience little exposure to capture, perhaps only coming into contact with one or two traps. Conversely, individuals whose territories are located squarely in the center of a trapping array might come into contact with many traps. As such, these individuals should experience higher probabilities of capture than individuals of the former type.