Modeling bat community structure and species distribution across fragmented landscapes within the upper Wabash River basin
Abstract
Bats are the second most speciose order of mammals in the world and perform a myriad of ecosystem functions. Population declines for many bat species are associated with rapid, human-induced ecosystem changes. Within this dissertation, I relate differences in community structure, probability of species occurrence, and roost tree selection to environmental variables measured at multiple spatial scales across a gradient of agricultural intensity. To accomplish this, I explored more flexible statistical models for identifying bat echolocation calls to improve species identification accuracy. I found that more flexible statistical models such as neural networks offer improvements over traditional linear discriminant analysis when identifying bat species by their echolocation calls. I then developed models to relate changes in bat community structure to both important environmental variation and evolved species traits. For these analyses, I demonstrated a strong relationship between evolved species traits and an apparent tradeoff between two landscape-scale environmental measures: forest and urban area. I also constructed species-specific occurrence models using hierarchical Bayesian methods that accommodated my study's spatially hierarchical sampling design. Here, I again demonstrated a tradeoff between urban and forest environmental factors. However, I also demonstrated specific relationships for species at different spatial scales. I examined forest interactions at a finer scale by studying bat roosting habits within forests. I applied discrete-choice analysis to elucidate summer roost selection criteria by three species of bat at two spatial scales within a typical home range. Here, I found bats selected older trees in forest areas with distinct overstory and midstory canopies. Finally, I explored the potential contribution of bats to forest ecosystems through the creation of nutrient peaks surrounding their roost sites. I concluded that large Myotis sodalis maternity roosts could cause a temporary nutrient peak surrounding their roost during seasonal periods corresponding to parturition and lactation.
Degree
Ph.D.
Advisors
Swihart, Purdue University.
Subject Area
Ecology|Forestry
Off-Campus Purdue Users:
To access this dissertation, please log in to our
proxy server.