Distribution and site selection of Le Conte's and Crissal thrashers in the Mojave Desert: A multi-model approach
Information on the distribution and habitat requirements of a species are critical components to the development of meaningful conservation plans. Such knowledge, however, is particularly difficult to obtain for species that are elusive and occur at low densities, such as the Le Conte's (Toxostoma lecontei) and Crissal (Toxostoma crissale) thrashers. In association with a regional conservation plan, I evaluated the distribution and habitat selection of these thrashers within Clark County, Nevada in the eastern Mojave Desert. I used a call-broadcast approach to sample 432 stratified random locations, detecting Le Conte's thrashers at 45 locations and Crissal thrashers at 41 locations. To model suitable habitat and predict thrasher occurrence, I used site-specific and landscape level information to create models that represented habitat data at two spatial scales. At each of these spatial scales, I measured variables corresponding to five environmental categories; plant assemblages, substrate, landform features, climate, and human disturbance. For analyses, I used logistic regression and assessed resulting models using an information theoretic approach. Inclusions in the best-fit model sets were determined using an Akaike Information Criterion approach. Model-averaging was used to determine the best possible parameter estimates for predicting thrasher presence from the complete sets of best-fit models. Results from the models indicated that Le Conte's thrashers occur within areas of little topographic relief such as valley bottoms near dry lake beds (playas). This pattern was strongly evidenced by the negative relationship between these thrashers and slope, in that they were never observed on slopes greater than 5 degrees, and by the disassociation with mountainous habitat and higher-elevation plant assemblages. The site-specific (ecological model) supported this broad pattern in identifying strong positive associations with playas and saltbush assemblages (specifically, Atriplex polycarpa and A. canescens). Positive associations were also determined for three other plant assemblages: wash vegetation, cholla, and Mojave mixed scrub (dominated by Yucca schidigera). The landscape model confirmed the important relationship of saltbush and wash vegetation. Crissal thrashers presence showed a strong negative relationship with creosote-bursage, shadscale, and creosote-sparse Joshua tree plant assemblages and with a principal component describing climatic patterns associated with decreasing temperatures and increased precipitation at higher elevations. Two plant assemblage categories, riparian and wash vegetation, and a principal component describing latitudinal patterns in climate were positively associated with this thrasher. The landscape model for the Crissal thrasher identified the same variables and relationships as the site-specific model. Suitable habitat for both species were predicted in ArcGIS using model average coefficients derived from best-fit landscape models. The predictive maps greatly improved on existing habitat models for these species within Clark County, and provide tools for conservation planning.