Linking changing climate, productivity, and fire in the Cape Floristic Region: A spatio-temporal Bayesian analysis of fire frequency

Recent studies have demonstrated strong correlations between warming temperatures and increased fire frequencies. Changes in fire regime would strongly affect the future of the mediterranean-climate shrublands of the Cape Floristic Region of South Africa, since the persistence of the biome and its extraordinarily high plant biodiversity depend on regular fire recurrence. Acceleration of fire return times here would alter relative abundances of species, shift biome boundaries, and alter water cycling. We use an extensive 26-year record of fires on public lands in these shrublands (>1500 fires over ~40,000 km²), together with high-resolution records of daily temperature and precipitation, to explore temporal and spatial trends in climate and fire regime. Using Bayesian spatio-temporal event models, we link these data sets to associate patterns in fire frequency and seasonality with drought dynamics, vegetation productivity, and global-scale climate cycles (e.g. SOI). Preliminary results show that the region has experienced significant acceleration in fire frequency from 1980-2005, and that this acceleration is associated with changes in mean and extreme climate characteristics (e.g. soil moisture availability, maximum drought stress). Further, we find that these regional changes exhibit meaningful patterns at the subregional scale (east-west gradient in magnitude of changes) and the global scale (linkage to global sea-surface temperature cycles). In our ongoing work we are exploring connections between fire and modeled soil moisture, vegetation productivity, water use efficiency and developing dynamic coupled models of these relationships, which we plan to validate using both field and satellite-derived productivity observations.