Seasonal and diel elk habitat selection in the Jemez Mountains of New Mexico
Roberts, Caleb P.
MetadataShow full item record
Organisms select habitat at multiple spatiotemporal scales, but many habitat selection studies fail to consider temporal scales finer than seasons. Because habitat selection is closely related to physiological and ecological needs and behavioral patterns, temporal delineations for habitat selection models should be based on behavioral patterns. Many GPS (Global Positioning System) collars come equipped with motion sensors capable of remotely measuring animal behavioral states which could circumvent costs of time, money, and stress to animals associated with collecting behavioral data. Elk (Cervus canadensis) epitomize the need for examination of finer temporal scales and the utility of GPS collar motion sensors to remotely estimate behavioral states. I addressed these gaps in the literature by asking (1) does empirical delineation of seasonal and diel temporal scales improve modeling of elk habitat selection, and (2) are motion sensors in GPS collars a viable method for remotely estimating elk behavior. I conducted this study in the Jemez Mountains of north-central New Mexico. Diel periods divided into dawn-midday-dusk-midnight intervals predicted shifts in elk foraging behavior. Compared to seasonal models, diel-scale habitat selection models detected more detail and different patterns. While seasonal models indicated consistent selection for fire history, vegetation type, and canopy cover covariates, diel models showed elk shifting preferences for these covariates throughout diel periods. At dawn and dusk, elk selected open canopy, grasslands, and herbaceous biomass whereas at midday, elk selected forested vegetation types, closed canopy, and recently burned areas. Although I utilized GPS collars with relatively unsophisticated motion sensors, I predicted active and inactive behavioral states of free-ranging elk with relative accuracy. Activity sensor hits were positively related to active behaviors (foraging, traveling) and negatively related to inactive behavior (resting). A threshold value of 6 activity sensor hits per 15 minutes categorized active versus inactive behavioral states with a 69% success rate (±12% at 95% confidence). This study emphasizes the importance of considering temporal scale and behavioral data in habitat selection studies and the ability of GPS motion sensor collars to estimate behavioral data.