Wildfires in the Cape floristic region : exploring vegetation and weather as drivers of fire frequency

Includes bibliographical references (leaves 68-76).This study assessed the spatial and temporal patterns of wildfires in the Cape Floristic Region (CFR). It focused on the factors that influence fire frequency; namely vegetation age, ignition sources and weather conditions. This work was done to aid decisions on fire management in fynbos nature reserves. Fire intervals were extracted from historical fire records in four reserves in the CFR. The study sites were the Cederberg and Hottentots-Holland (western) reserves and the Swartberg and Outeniqua (eastern) reserves, and fire records were used from 1970 to 2007. A non-parametric technique and smoothing methods were used to highlight patterns in the extracted fire intervals. Comparisons of fire frequency were made between the study sites to analyse spatial patterns of burning. The impact of anthropogenic ignitions on fire frequency was analysed to explore the effect of people on fire patterns. The relationship between fuel age and fire size was analysed to determine the influence of vegetation age on fire patterns. Two novel methods were described in this thesis. The first method developed a technique to analyse temporal patterns in fire frequency while avoiding the impacts of temporal autocorrelation. The second method analysed the relationship between weather condition and fire events by utilising self-organising maps of synoptic states. A temporal change in the frequency of these synoptic states was tested for over the recording period. Synoptic states were used to produce two regional fire risk indicators for the CFR. I found that fire intervals in western study areas of CFR were shorter than fire intervals in eastern study areas. The effect of anthropogenic ignition sources shortened fire intervals in all study sites; however, this was relative to the natural fire frequency of each study site. Prescribed burning as a form of fire management contributed relatively little to the total area burned in all study sites. Fuel age has a significant correlation to fire size in only the drier (Swartberg) study site. A decreasing trend in fire return intervals was found in three study sites; Cederberg, Hottentots-Holland and Outeniqua. Synoptic states characteristic of the southern-most extent of a tropical easterly wave low were correlated to frequency fire events in the western study areas. Fires in the eastern study areas were correlated to a synoptic state characteristic of a tropical temperate trough. Easterly wave lows are associated with strong atmospheric convection whereas tropical temperate troughs are associated with pre-frontal conditions and strong, hot and dry winds. The frequencies of these synoptic states were shown to have increased in recent decade. The factors influencing fire frequency in the western CFR are predominantly sources of ignition, while the availability of fuels and suitable weather conditions restrict fires in the eastern CFR. Fire frequency has increased in the study sites where weather exerts the dominant control and this is due to the increase in synoptic states that promote wildfires. Historical records show that fire management has had little impact on the total area burned, thus fire management under climate change is unlikely to influence fire frequency