Landscape analysis and ungulate movements in the Greater Yellowstone Region: Integrating remote sensing, modeling and visualization technologies in support of conservation research

The migration of large mammals over large distances is one of the most prominent yet threatened ecosystem processes. It is driven by landscape gradients across large scales and constrained by landscape heterogeneity at small scales. Remote sensing and landscape modeling can describe and predict these landscapes in the context of global dynamics. The challenge then is to build from this toward predictions of wildlife movement response to global change. We argue that since movement is dynamic that a statistical dynamic approach is needed. We combined kernel-density-estimator (KDE) and resource-selection-function (RSF) approaches within a novel statistical framework to form a method that we term Selective Computational Diffusion. We applied the method to short-term dynamics of bison distribution in Yellowstone National Park and showed that in this case study, the method was superior to conventional KDE and RSF examples. We will illustrate example applications of the approach through visualization of predicted and observed bison movement at fine scales over large landscapes.