Nice weather for bettongs: using weather events, not climate means, in species distribution models

Current applications of species distribution models (SDM) are typically static, in that they are based on correlations between where a species has been observed (ignoring the date of the observation) and environmental features, such as long-term climate means, that are assumed to be constant for each site. Because of this SDMs do not account for temporal variation in the distribution of suitable habitat across the range of a species. Here, we demonstrate the temporal variability in the potential geographic distributions of an endangered marsupial, the northern bettong Bettongia tropica as a case study. Models of the species distribution using temporally matched observations of the species with weather data (includ- ing extreme weather events) at the time of species observations, were better able to defi ne habitat suitability, identify range edges and uncover competitive interactions than models based on static long-term climate means. Droughts and variable temperature are implicated in low densities and local extinctions of northern bettong populations close to range edges. Further, we show how variable weather can infl uence the results of competition with the common rufous bettong Aepy- prymnus rufescens. Because traditional SDMs do not account for temporal variability of suitable habitat, static SDMs may underestimate the impacts of climate change particularly as the incidence of extreme weather events is likely to rise.