Forecasting the effect of land-use change on native and non-native mammalian predator distributions

Intensive land use can fragment continuous natural areas into smaller patches, which may be too small to support viable populations of native fauna and more susceptible to invasion by alien species. We demonstrate the utility of combining species occurrence models with land-use change models to identify areas where future development may differentially affect wildlife. Occurrence data for native (e.g., gray fox, coyote, bobcat, mountain lion, striped skunk, raccoon) and non-native (e.g., domestic dogs, domestic cats, opossums) mammalian predators were collected from 188 remotely triggered camera locations across an oak woodland and vineyard landscape in northern California. The occurrence data were used in combination with landscape variables extracted using a geographic information system to build explanatory models of predator occurrence. These statistical models were used to derive two surfaces showing relative probability of occurrence for non-native and native predators. Then, a spatially explicit land-use change model was used to examine potential future predator distributions given potential future vineyard expansion. The probability distribution models generated hypothesized low probabilities of occurrence for native predators within large vineyard blocks, but higher probabilities within isolated vineyards and also in oak woodlands. The models suggest the highest probabilities of non-native predator occurrence fell within large blocks of vineyard. Using one possible future vineyard development scenario, the distribution models illustrate areas where probability of native predator occurrence may be reduced and where non-native predators may expand due to vineyard development. This technique could be applied to prioritize acquisition of critical wildlife habitat and maintain habitat connectivity for wildlife populations.