Year of Award

2011

Document Type

Thesis

Degree Type

Master of Science (MS)

Degree Name

Geography (Cartography and GIS Option)

Department or School/College

Department of Geography

Committee Chair

Anna Klene

Commitee Members

Solomon Dobrowski, Zachary Holden, Susan Rinehart

Keywords

GIS, Habitat modeling, Rare plant, Variable selection

Abstract

Habitat suitability modeling is widely-used in both biogeography and ecology to characterize the biophysical requirements and distribution of plant and animal species. Many of these modeling efforts use different variants of essentially the same topo-climatic variables (elevation, slope, aspect, precipitation, and temperature). However, these commonly used variables may not sufficiently explain the distribution of rare-plant species, which may have additional habitat needs. The aim of this project was to determine guidelines for selection of variables to include in statistical modeling efforts to predict suitable rare-plant habitat. Additionally, how background extent, data resolution, sample size, and various ranking criteria effect environmental variable selection were considered. For this case study, Broad-fruit Mariposa (Calochortus nitidus Dougl.) a rare-plant species found within the 2.2 million acre Nez Perce National Forest of north-central Idaho was used. The study area is dominated by mountainous terrain, with elevations ranging from 500 to 2800 m (~1500 to 9000 ft). The widely used MAXENT model and additional methods were used to statistically determine the relative importance of more than 30 environmental variables considered in the analysis and yield recommendations about the most effective way of utilizing these often highly correlated variables. Study area extent and the sample size of occurrence data had by far the greatest impact. Sensitivity to these factors resulted in variables being ranked differently, but the majority of the models ranked elevation, May precipitation, vegetation type, April minimum temperature, NDMI, September precipitation, and July maximum temperature as highly important for Broad-fruit mariposa. Vegetation type, NDMI and NDVI tended to be ranked highest when modeled at the 30×30 m resolution, suggesting that these fine resolution datasets may be extremely valuable in predicting the habitat of Broad-fruit mariposa.

Share

COinS
 

© Copyright 2011 Thor Burbach