Using dynamic occupancy models to inform climate change adaptation strategies for California spotted owls

Abstract

Management strategies intended to reduce the impacts of future climate change on montane species ought to consider fine-scale variation in microclimate (e.g. because of topography) and vegetation conditions that create climate refugia. Identifying potential refugia and quantifying their importance is facilitated by analytical approaches that explicitly model associations between dynamic population processes and environmental conditions. We developed a population model to project future site occupancy dynamics in spotted owls Strix occidentalis under climate change and forest-management scenarios. The model was parameterized with statistical relationships derived from dynamic occupancy analyses applied to 20 years of spotted owl presence/nondetection data in the Sierra Nevada, California, USA. This approach allowed us to link colonization and extinction processes with spatial and temporal variation in vegetation characteristics, microclimate and weather while accounting for imperfect detection. Occupancy analyses indicated that greater prevalence of closed-canopy forest in owl sites was associated with lower extinction and higher colonization rates. Both rates increased following consecutive warm summers, but warm summer temperatures had a greater effect on extinction than colonization. Following the warmest summers, extinction was most likely to occur in owl sites with cool microclimates and relatively little closed-canopy forest, but extinction became considerably less likely in cool microclimates when greater amounts of closed-canopy forest were present. By comparison, extinction probability was lower at owl sites with warmer microclimates and relatively invariant to the amount of closed-canopy forest present. Simulations of our population model suggested that managing for greater amounts of closed-canopy forest may ameliorate the potential negative effects of warm summer temperatures by the mid-century, especially at high elevations. The model also allowed us to identify sites where potential management efforts would be most effective. Synthesis and applications. By modelling environmental processes influencing site colonization and extinction dynamics, we were able to identify features of potential climate change refugia for a montane species of management interest that could be promoted by land managers to increase species persistence. We also provided a general, process-based population modelling approach for projecting changes in animal populations and developing habitat-based climate change adaptation strategies using presence/nondetection data.