OOS 4-9
Estimating abundance when landscape structure determines patterns of both space-use and density

Monday, August 10, 2015: 4:20 PM
316, Baltimore Convention Center
Christopher Sutherland, Natural Resources, Cornell University, Ithaca, NY, USA
Angela K. Fuller, Department of Natural Resources, New York Cooperative Fish and Wildlife Research Unit, Cornell University, Ithaca, NY, USA
J. Andrew Royle, , USGS Patuxent Wildlife Research Center, Laurel, MD, USA

The configuration of habitat and structure of landscapes influence animals in two ways: the distribution of home-ranges across the landscape (‘second-order’ resource selection), and the manner by which areas within a home range is utilized (‘third-order’ resource selection). The advent of spatially explicit capture-recapture models (SCR) is a relatively new, and extremely powerful, approach that provides a spatial context for abundance estimation and allows simultaneous modeling of spatial patterns of density and space-use. Independently, SCR has been extended to incorporate (1) spatial structure in density, as a function of spatial covariates, and (2) spatial structure in detectability by admitting landscape associated asymmetries in home range geometry.

We investigate the utility of SCR to simultaneously detect spatial structure in in both model components using simulations. We simulated spatial encounter data in a highly structured ecological network (a river system) with increasingly strong spatial structure in both density (ranging from uniform, to linear distribution of individuals), and detectability (from Euclidean/symmetric to linear home ranges).


Fitting standard (null) SCR models to the simulated data, we find that  mis-specification of either model component (density or detection), results in biased estimates of abundance – a bias that becomes more severe with increasing dependencies between second- and third-order resource selection and landscape structure. However, we also find that the SCR model framework is flexible enough to accommodate spatial structure in both component models and is able to reliably characterize landscape-density and landscape-space relationships.

In studies of wild living populations that exist in highly structured landscapes that are becoming more fragmented, accommodating such structure in the estimation of population size is increasingly important. Here we demonstrate this is the context of a riparian system, a canonical example of an ecological network with encouraging results. The continued trend towards the integration of biologically realistic representations of systems in SCR is promising for wildlife monitoring and conservation.