Broad‐scale occurrence of a subsidized avian predator: Reducing impacts of ravens on sage‐grouse and other sensitive prey

Expanding human enterprise across remote environments impacts numerous wildlife species. Anthropogenic resources provide subsidies for generalist predators that can lead to cascading effects on prey species at lower trophic levels. A fundamental challenge for applied ecologists is to disentangle natural and anthropogenic influences on species occurrence, and subsequently develop spatially explicit models to help inform management and conservation decisions. Using Bayesian hierarchical occupancy models, we mapped the broad‐scale occurrence of common ravens Corvus corax as a function of natural and anthropogenic landscape covariates using >15,000 point count surveys performed during 2007–2016 within the Great Basin region, USA. Raven abundance and distribution is substantially increasing across the American west due to unintended anthropogenic resource subsidies. Importantly, ravens prey on eggs and chicks of numerous species including greater sage‐grouse Centrocercus urophasianus, an indicator species whose decline is at the centre of national conservation strategies and land‐use policies. Anthropogenic factors that contributed to greater raven occurrence were: increased road density, presence of transmission lines, agricultural activity, and presence of roadside rest areas. Natural landscape characteristics included lower elevations with greener vegetation (Normalized Difference Vegetation Index), greater stream and habitat edge densities, and lower percentages of big sagebrush Artemisia tridentata spp. Interactions between anthropogenic sources of nesting substrate and food subsidies suggested that raven occurrence increased multiplicatively when these resource subsidies co‐occurred. Overall, the average probability of raven occurrence estimated within sagebrush ecosystems of the study area was c. 0.83. Synthesis and applications. We demonstrate how anthropogenic factors can be disentangled from natural effects when making spatially explicit predictions of subsidized predators occurring across expansive landscapes. This approach can guide management decisions where subsidized predators overlap sensitive prey habitats. For example, we identify areas where elevated raven occurrence coincides with breeding sage‐grouse concentration areas and appears to be largely driven by anthropogenic factors. Management applications could focus on reducing raven access to anthropogenic subsidies in these areas, while prioritizing habitat improvements for sage‐grouse elsewhere. Our approach is applicable to other species where widespread survey data are available.