Distribution Modeling

Our ultimate goal is to develop scientifically valid, accurate, and precise geographic range maps based on recorded occurrences and predictive models. We consider range maps as hypotheses for testing. We’re using a variety of modeling methods to account for wide variation across species in sample sizes, data quality, and ecological knowledge. We view habitat or distribution models as tools to assist in delineating a species' range, maps that may take into account additional factors that are not readily modeled by computer programs, such as geographic barriers and history of disturbance.

krat1_200w.jpg

Here we present draft maps for the endangered Stephens' kangaroo rat (Dipodomys stephensi) generated by means of three different modeling approaches that differ in logical structure and complexity. Future efforts will consider procedures for (1) weighting and combining predictions from multiple methods, (2) verifying maps by stratified sampling, and (3) integrating additional factors such as disturbance history, geographic barriers, and “expert opinion.”

Stephens’ kangaroo rat is an endangered heteromyid rodent with a small geographic range in southwestern California. It is closely associated with sparse annual grasslands or coastal sage scrub, abundant patches of bare ground, deep well-drained loamy soils, and gentle slopes. The species’ known geographic range was significantly extended in 1997 with the discovery of a disjunct population in the Ramona Grasslands. Other areas that might host undiscovered populations of potentially suitable habitat have not been surveyed.