Developing sustainable rangeland management strategies requires solution-driven research that addresses ecological issues within the context of regionally important socioeconomic concerns. A key sustainability issue in many regions of the world is conserving habitat that buffers animal populations from climatic variability, including seasonal deviation from long-term precipitation or temperature averages, and that can establish an ecological bottleneck by which the landscape-level availability of critical resources becomes limited. We integrated methods to collect landscape-level animal occurrence data during severe winter conditions with estimation and validation of a resource selection function, with the larger goal of developing spatially explicit guidance for rangeland habitat conservation. The investigation involved greater sage-grouse (Centrocercus urophasianus) that occupy a landscape that is undergoing human modification for development of energy resources. We refined spatial predictions by exploring how reductions in the availability of sagebrush (as a consequence of increasing snow depth) may affect patterns of predicted occurrence. Occurrence of sage-grouse reflected landscape-level selection for big sagebrush, taller shrubs, and favorable thermal conditions and avoidance of bare ground and anthropogenic features. Refinement of spatial predictions showed that important severe winter habitat was distributed patchily and was constrained in spatial extent (7–18% of the landscape). The mapping tools we developed offer spatially explicit guidance for planning human activity in ways that are compatible with sustaining habitat that functions disproportionately in population persistence relative to its spatial extent or frequency of use. Increasingly, place-based, quantitative investigations that aim to develop solutions to landscape sustainability issues will be needed to keep pace with human-modification of rangeland and uncertainty associated with global climate change and its effects on animal populations.