A Geospatial Approach to Wildlife and Wilderness Management

The populations in the western United States have increased significantly over the last few decades resulting in increased development pressures on undeveloped and pristine lands. Population growth and increased human activities have also changed the overall ecosystems more rapidly and extensively in recent years than in the past, emphasizing the need for protecting natural land and ecosystems. This dissertation conducts analyses to highlight the importance of protected areas and wildlife habitat. Chapter 1 explores the use of Poisson and negative binomial regression models to examine winter habitat use by mountain goats in the Kenai Mountains of South-Central Alaska. Using GPS collared locations data, these models produce parameter estimates similar to discrete choice models, popular in resource selection studies, but with less computational complexity. The potential serial (or spatial) correlations present in the data are controlled for by incorporating spatial effects in a Gaussian conditional autoregressive framework. The results support the existing literature on mountain goat habitat use with most of the terrain features statistically significant across individual goat models. The distance to the nearest escape terrain is found to be the most consistent and highly significant determinant of goat habitat selection, where individual goats tend to increase the number of visits by 37.5 to 71.6 percent more to a particular location that is 100m closer to escape terrain. The statistical significance of spatial parameter highlights the importance of neighborhood effects in habitat selection by mountain goats. Chapter 2 seeks to achieve two objectives. First, using regression based test of equivalence, the predicted habitat use by mountain goats are compared for two different sources of data. For mountain goats, the findings suggest that the predicted habitat uses from GPS collar and aerially surveyed locations are equivalent. Second, using the aerially surveyed goat locations data, the Bayesian inference techniques with respect to negative binomial regressions are employed to explore the effects of winter recreation on mountain goat habitat selection. In addition to the landscape features, the model comparison based on Bayes factor suggests that human recreation is an important factor affecting goats habitat use. Goats tend to avoid areas with higher human recreational activities represented by ski-tracks, and hiking trails; increase in the ski-tracks area by one square km within a buffer of 5 km reduces goat count by approximately 2.5 percent at a particular geographic location. This highlights the potential conflict between human recreation and winter habitat and provides an input into policy discussions on conservation of mountain goat habitat. Chapter 3 of this dissertation analyzes the statistical properties of estimated parameters in the hedonic studies with special reference to the distance variables used to represent the proximity to environmental amenities or dis-amenities. Using Monte Carlo simulations, the findings suggest that when there are single locations available for each of the landmarks in the study area, including two or more distance variables in regression results in inconsistent parameter estimates associated with the distance variables. Nevertheless, when multiple locations are available for each of the landmark types and the distance variables capture the proximity to the nearest landmark location of each type, the parameters are consistently estimated. These findings are important in studies focusing on the estimation of welfare values based on distance variables. The major objective of Chapter 4 is to examine the effects of proximity to wilderness areas on residential property values in Yavapai County, Arizona by utilizing geographic information system and recently developed spatial models. On average, a residential property located one mile closer to a nearest wilderness area is estimated to have values $4730 and $5024 higher based on Euclidean and road distance as a measure of proximity respectively