Empirical validation of landscape resistance models: insights from the Greater Sage-Grouse (Centrocercus urophasianus)

Context The ability of landscapes to impede species' movement or gene flow may be quantified by resistance models. Few studies have assessed the performance of resistance models parameterized by expert opinion. In addition, resistance models differ in terms of spatial and thematic resolution as well as their focus on the ecology of a particular species or more generally on the degree of human modification of the landscape (i.e. landscape integrity). The effect of these design decisions on model accuracy is poorly understood. Objectives We sought to understand the influence of expert parameterization, resolution, and specificity (i.e. species-specific or landscape integrity) on the fit of resistance model predictions to empirical landscape patterns. Methods With genetic and observational data collected from Greater Sage-Grouse (Centrocercus urophasianus) in Washington State, USA, we used landscape genetic analysis and logistic regression to evaluate a range of resistance models in terms of their ability to predict empirical patterns of genetic differentiation and lek occupancy. Results We found that species-specific, fine resolution resistance models generally had stronger relationships to empirical patterns than coarse resolution or landscape integrity models, and that the expert models were less predictive than alternative parameterizations. Conclusions Our study offers an empirical framework to validate expert resistance models, suggests the need to match the grain of the data to the scale at which the species responds to landscape heterogeneity, and underscores the limitations of landscape integrity models when the species under study does not meet their assumptions.