Performance of Greater Sage-Grouse Models for Conservation Assessment in the Interior Columbia Basin, U.S.A.

Abstract: Valid modeling of habitats and populations of Greater Sage-Grouse ( Centrocercus urophasianus) is a critical management need because of increasing concern about population viability. Consequently, we evaluated the performance of two models designed to assess landscape conditions for Greater Sage-Grouse across 13.6 million ha of sagebrush steppe in the interior Columbia Basin and adjacent portions of the Great Basin of the western United States (referred to as the basin). The first model, the environmental index model, predicted conditions at the scale of the subwatershed (mean size of approximately 7800 ha) based on inputs of habitat density, habitat quality, and effects of human disturbance. Predictions ranged on a continuous scale from 0 for lowest environmental index to 2 for optimal environmental index. The second model, the population outcome model, predicted the composite, range-wide conditions for sage grouse based on the contribution of environmental index values from all subwatersheds and measures of range extent and connectivity. Population outcomes were expressed as five classes (A through E) that represented a gradient from continuous, well-distributed populations (outcome A) to sparse, highly isolated populations with a high likelihood of extirpation (outcome E). To evaluate performance, we predicted environmental index values and population outcome classes in areas currently occupied by sage grouse versus areas where extirpation has occurred. Our a priori expectations were that models should predict substantially worse environmental conditions ( lower environmental index) and a substantially higher probability of extirpation ( lower population outcome class) in extirpated areas. Results for both models met these expectations. For example, a population outcome of class E was predicted for extirpated areas, as opposed to class C for occupied areas. These results suggest that our models provided reliable landscape predictions for the conditions tested. This finding is important for conservation planning in the basin, where the models were used to evaluate management of federal lands for sage grouse.