| Summary: | The spatial and temporal patterns of fire in Africa south of the equator are described and
investigated to test hypotheses on the main drivers of fire. In particular, the relative
importance of climate, vegetation, and human activities are assessed and related to
current theory on savanna fire regimes and their response to global change.
This thesis makes use of newly-released burned area and fire radiative power data
derived from satellite imagery. This information was integrated with data on climate,
vegetation, and human activities, as well as long-term field data on fire, to test hypothe-
ses about savanna fire regimes. Regression tree techniques and mixed effects models
were used to determine the main drivers of spatial and temporal variability in fire, and
feedbacks between fire and tree cover were explored at a range of spatial scales. New
datasets on fire size and fire number were also developed to test theories on the effect of
human activities on ignition probability and fire spread.
The results show that fire in these grassy ecosystems is very responsive to climatic
factors such as rainfall but there are large areas where human activities play an equally
significant role. The effect of people was always to reduce the total area burned, although
the number of individual fires increased with population density up to about 10 people
per km2. Areas inhabited by people were also less susceptible to year-to-year
fluctuations
in annual burned area caused by climatic variability, possibly because of the reduced
probability of large, runaway fires in these highly-impacted landscapes. Both of these
findings are important for modelling of fire and carbon emissions in Africa.
Evidence of negative feedbacks between tree cover and fire were found. When the
percentage tree cover is less than 40% a range of factors control the extent of fire in a
system; but above 40% tree cover fire is never a large component of the landscape. On
the other hand, tree cover was shown to have a bi-modal distribution, and examples of
intermediate tree cover (40-70%) were rare. Over a wide range of mean annual rainfall
(1000-2000mm) vegetation can exist either in a "high tree-cover, low fire" or a "high
fire, low tree cover" state. This research challenges some assumptions about how savanna fires will respond to
global change, and provides important clues to help disentangle the mechanisms driving
fire in Southern Africa. Importantly, it highlights that fire in southern Africa, and in
other highly-utilised savanna systems, can not be understood outside the context of
human use.
|
|---|
