APPLICATION OF ADVANCED DIGITAL TECHNOLOGY FOR WILDLIFE HABITAT MODELING

In recent decades, wildlife studies have begun to incorporate more structural characteristics into studies of habitat. However, most still collect habitat data primarily from field measurements, which are limited in spatial extent and force researchers to extrapolate from small-scale, ground-based measures. LiDAR offers the opportunity to objectively measure habitat features across landscape-level extents. I mapped and measured forest canopy structural diversity at 181 plots across Indiana at a series of three (25-m, 50-m, and 100-m) radii to quantify landscape heterogeneity across space, utilizing new 2016–2020 aerial LiDAR in tandem with high-resolution camera traps. I aimed to quantify the relationship between use of forest sites by seven wildlife species and habitat structural characteristics, specifically predicting species occupancy as a function of LiDAR-derived metrics. I predicted that habitat generalists like raccoons (Procyon lotor) and Virginia opossum (Didelphis virginiana) would be best predicted by the null hypothesis and that measures of gap fraction would best predict edge species such as coyote (Canis latrans), white-tailed deer (Odocoileus virginianus), and wild turkey (Meleagris gallopavo). Additionally, I predicted a positive relationship between vegetation area and both eastern cottontail rabbit (Sylvilagus floridanus) and tree squirrels (Sciuridae spp.). I expected that measures of habitat heterogeneity would be included in best models for tree squirrels and white-tailed deer, and that models for squirrels would have lower root mean square error (RMSE) values. Ultimately, structural metrics varied across radii, and best models depended on plot radius size. Measures of vertical heterogeneity were the best predictors for species like raccoon, wild turkey, and coyote, with a higher probability of occupancy for all three with increased heterogeneity. Additionally, models for eastern cottontail rabbit incorporating vegetation area indices, Gini diversity, and gap fraction demonstrated significance and low predictive error. Habitat generalists, such as white-tailed deer and Virginia opossum did not select for specific structural metrics and were best predicted by the null model. Ultimately, these results indicated that LiDAR is a promising potential tool for measuring ecologically meaningful variables at scales large enough to properly represent home range and resource use at the home-range level, filling an important gap in our understanding.