PARENT SESSION
Oral Session #29: Spatial Ecology and Disturbance Ecology.
Presiding: J. Ludwig
Tuesday, August 6. 8:00 AM to 11:30 AM. Coconino Meeting Room, TCC.

A model based approach for spatially interpolating count data using local environmental and population variability.

Farnsworth, Matthew^*,1, Reich, Robin¹, ¹ Colorado State University, Fort Collins, Colorado

ABSTRACT- Estimating the abundance of a species in locations where sample data is unavailable can be achieved through the application of spatial interpolation. Frequently, model-based interpolation is carried out using methods such as trend surface modeling or Kriging by developing models that are based on the spatial structure of the entire population. However, as a consequence of non-linear variations in environmental conditions animal and plant populations are distributed in a discontinuous manner at most spatial scales. As a result, assumptions regarding the existence of a global gradient in the distribution of a species are often incorrect, obviating the use of interpolation procedures based on a single global model. We use nine years of mule deer (Odocoileus hemionus) count data collected across roughly 1500 km² to develop a surface of estimated density based on local variations in both the environment and deer survey data. Our method is implemented through a variable-sized moving window function in which multiple candidate models are fit to the data within a specified neighborhood, which is a localized subset of all deer count stations. Within each neighborhood, models based on the local spatial structure of the deer population and environmental covariates are assessed for their predictive ability. The best predictive model within a neighborhood is used to carry out the interpolation. The result is a flexible modeling approach that takes advantage of local conditions to carry out the interpolation when inferences from global models are not applicable.

KEY WORDS: species distribution models, spatial interpolation