Date of Graduation

Fall 2011

Degree

Master of Science in Geospatial Sciences

Department

Geography, Geology and Planning

Committee Chair

Xin Miao

Keywords

species distribution modeling, predictive mapping, logistic regression, Joshua tree, remote sensing

Subject Categories

Botany | Natural Resources and Conservation | Remote Sensing

Abstract

The Joshua tree is an indicator species that is essential to the Mojave Desert ecosystem. Recent research has revealed threats facing future Joshua tree survivorship, and monitoring the species has been identified as important under the Clark County (NV) Multiple Species Habitat Conservation Plan. Using techniques from species distribution modeling, the author utilized environmental predictor variables to create a logistic regression on a sample of Joshua tree presence-absence locations. The output of this model was a map of habitat suitability ranged from 0.0 to 1.0 across Clark County. Though not as typical, increased resolution and availability has raised interest in the use of remotely sensed imagery in distribution models. Subsequently, regression models with and without predictors derived from imagery were compared in order to assess the potential benefits of using remotely sensed imagery in the Joshua tree model. Using calibration plots, confusion matrices, and receiver operating characteristic analysis, evaluation of discrimination ability and prediction reliability revealed that the model with imagery predictors outperformed the model without imagery predictors. As a supplement to the suitability map, thresholds were applied to the predictions of the final model in order to create a classified habitat map to assist in conservation efforts. The results of this research can serve as an elementary step in producing a conservation strategy for Joshua tree habitat in Clark County while also providing further evidence of the validity of using data from remote sensors as a component of species distribution modeling.

Copyright

© Joseph Michael Sturgis

Campus Only

Share

COinS