|Title||A GIS framework for modelling wildlife species distributions|
|Year of Publication||1998|
|Keywords||evidence, expert system, GIS, inference, scale, wild pigs, wildlife modeling|
Maps of wildlife species distributions are a fundamental display of data in biogeography, and increasingly GIS methods are used to develop models of distributions. This dissertation examines some of the major issues in constructing predictive maps of species, focusing on the capability of GIS to relate environmental factors to distributions through logical or mathematical inference. The dissertation is structured in three parts. The first part considers how a variety of data sources may be aggregated to build up a picture of a distribution, using the example of the orange-throated whiptail, a lizard species living in southern California. It discusses how structuring these data on a hierarchy of spatial scales can lead to new inferences about distributions and habitat relationships. The second and third sections elaborate this theme of data availability and spatial scale in distribution modelling, using the example of the feral pig in central California. The second section presents a case study of developing an expert system to predict relative pig abundance at a regional scale. It illustrates how an expert system provides a formal treatment of aggregation of evidence, and how increasing the degree of interaction with a GIS can lead to elicitation of better models from domain experts. The third section presents a habitat model for the feral pig at a local scale. The grain size of this model is very finely resolved with respect to the home range of a pig, so this model integrates habitat elements over the home range size of the animal to create a spatially sensitive model of habitat quality. This model is tested against observation data at a number of different spatial scales, the results illustrating that it is important to recognize the spatial scale of a habitat model when it is applied.