Estimation of Probability of Habitat Use of Roosevelt Elk on the Olympic Peninsula

Estimating the probability of habitat use for a particular species is crucial to the direct management and conservation of that species. Without knowledge of habitat preferences, managers cannot effectively focus efforts on vital resources or landscape types. However, modelling probability of habitat use can be done in several ways which leaves room for variation and uncertainty in the estimates produced by each method. This study is an examination of the variation between two estimates of probability of habitat use while focusing on a particular subspecies of elk that inhabits a unique ecosystem relative to other elk subspecies. I modeled elk resource selection using both an occupancy framework and a zero-inflated Poisson regression. This project is essentially a comparison between using logistic regression to model habitat use and generalized linear regression. Occupancy modelling gives the same weight to every location where the target species is detected, no matter the frequency of use. Whereas a generalized linear model is a count-based approach that determines the relationship between the number of sightings at particular survey sites and associated habitat variables. Understanding the differences in estimates that are produced by each method can help future researchers decide to implement one over the other for their particular application.