Forging a Bayesian link between habitat selection and avoidance behavior in a grassland grouse.

Habitat selection is a basic aspect of the ecology of many species, yet often the term is conflated or confused with both habitat preference and habitat use. We argue that each term fits within a conceptual framework that can be viewed in Bayesian terms and demonstrate, using long-term data on occupancy patterns of a grassland grouse, how prior probability profiles can be estimated. We obtained estimates by specifically focusing on whether and to what extent the Lesser Prairie-Chicken (Tympanuchus pallidicinctus) avoids anthropogenic features such as roads, powerlines, petroleum wells, fences, and buildings, in two study areas, one with denser and one with sparser incidence of features. Grouse strongly avoided large features such as outbuildings and tended to avoid tall features such as powerlines; by contrast, grouse did not or only slightly avoided low, unobtrusive features such as fences. We further examined co-location of pairs of anthropogenic features and found that certain features were avoided so strongly that avoidance distance may be shorter for other features; that is, birds were "pushed toward" some features because they are "pushed away" from others. In each case, our approach points toward a means to incorporate avoidance behavior directly into analytic studies of habitat selection, in that data on use (the posterior, as it were) could be used to infer the selection process provided data on preference (the prior, as it were) could be obtained.