Differences in Niche Space between Migratory and Resident Ungulates- Predicting Migratory and Resident Mule Deer (Odocoileus hemionus) Behavior through Remotely Sensed Plant Phenology, Snow, and Landcover

Migration is a valuable life history strategy across a range of taxa. Potentially due to substantial anthropogenic change to landscape dynamics, there has been an increasing prevalence of migratory individuals forgoing migration and remaining resident throughout the year. Efforts to conserve, and restore migratory behavior require a better understanding of what the landscape dynamics are that promote the ecological tradeoff between migration and residence. Partial migrant populations such as mule deer provide a valuable opportunity to study ecological tradeoff influencing migration and residence. By studying GPS-collar data of 42 mule deer within Utah, USA, we classified the seasonal niches of mule deer as either seasonal migratory niches or residence niche space, given remotely sensed data on the forage quality, land cover, snow cover, and development of used niche space. We classified mule deer niches using a random forest model, then projected the model across Utah to visualize where winter migratory, summer migratory, and resident niches mostly likely occur. Our random forest model accurately classified summer migratory, winter migratory, and resident niches with a prediction accuracy of 97.8% and a cross-validated accuracy of 81.2%. Proximity to development was the most important variable in classifying between summer migratory, winter migratory, and resident niche space, where resident niche space was most often closer to developed areas. The methods and visualizations we present provide a relatively intuitive analysis that demonstrates how remote sensing can improve efforts to conserve migratory behavior of wide-ranging species such as mule deer both by recognizing the environmental features important to promoting migration as well as by visualizing critical habitat for migrants.