Background/Question/Methods   Mediterranean-climate shrublands are expected to experience strong effects from climate change, including our focal regions of California and the Cape region of South Africa. Shrub species in these systems are affected directly by climate variation via effects on recruitment, growth rates and mortality. At least equally importantly, these species and their biomes will be affected indirectly by climate through its effects on fire. Generally these shrubland biomes experience high-intensity fires on a decadal time scale. Previous studies have identified relationships between weather patterns and fire in these systems, primarily by focusing on summaries of fire in a region such as number of large fires and area burned. That work has highlighted the key role played by synoptic weather patterns and has enabled coarse projections of future fire frequency. In our work we complement this broader-scale work by using spatiotemporal statistical models to relate fire occurrence records, and thus return times, to weather and environmental factors that vary in space and time at a finer scale. This approach aims to exploit smaller-scale variability to illuminate what factors are important in determining intra-regional differences in fire frequency and thus future patterns in fire regimes and potential effects on species and biomes.

Results/Conclusions   We have directly observed the effect of climate on shrub demography and hence potentially fuel load. Through field experiments and observations of natural populations across an aridity gradient in South Africa, we have identified significant effects of precipitation (positive, P<0.01) and temperature (mixed results, P<0.05) on recruitment and growth of representative shrub species in genus Protea. An exploratory experiment on California chaparral shrubs showed an effect of soil temperature on seedling survival consistent with these results (P<0.05). Our statistical fire models identified potentially important indirect effects of climate via fire. In the Cape region, we found a recent shortening of fire return times; warming projected by climate models for the region would likely further increase fire frequency, depending also on global circulation patterns as captured in the Antarctic Oscillation. Nonetheless field data on plant growth, together with patterns in a remotely sensed vegetation index, suggest fuel limitation could begin to restrict fire frequency. In the more productive California chaparral, it appears less likely that aridity-induced fuel starvation would counteract the effects of warming on fire return times, and an accelerated fire regime may lead to contraction of chaparral and population failures of obligate reseeding species in some areas.