| Summary: | Forests and savannas in a savanna-forest mosaic are maintained by positive feedback loops with fires, promoting fires in savannas and excluding fires in forests. Occasional extreme fires do however burn into forest edges and cause extensive mortality. So, while our understanding of bistability in these systems has advanced, our understanding of biome-level change has remained rather static. Very few studies address the issue of forest to savanna transitions following extreme fire events. This study is posed to address the criticisms surrounding the bistability of savanna and forest vegetation and present evidence for catastrophic regime shifts in these mosaic systems. What sets it apart from previous research is the boundary between vegetation types is naturally occurring, not the result of logging within forests and the subsequent damage by fire. In HiP, forest patches naturally abut savannas, the system has its native herbivores, and human impacts are low. In my thesis I explored the circumstances causing a catastrophic regime shift or recovery of a resilient forest boundary. I initially set out to determine the drivers of vegetation distributions and dynamics in Hluhluwe-iMfolozi Park (HiP). Next I documented the aftermath of an extreme fire in thicket, forest and savanna vegetation focusing on the recovery of vegetation and evidence for catastrophic regime shifts. And lastly, I explored the association between bark thickness and commonly measured traits in order to understand trait evolution in response to fire regime and other selective pressures such as herbivory and drought. Using aerial photographs from six time periods between 1937 and 2013, I mapped vegetation changes in Hluhluwe iMfolozi Park. Using a Generalized Additive Model (GAM) I built a Habitat Suitability Index (HSI) map based on vegetation distribution maps and topographic variables related to fire behaviour. I investigated transitions between time periods based on the HSI map, as well as the effects of neighbourhood on transition probabilities. Forest distributions in HiP have not remained static over time and have expanded into areas that were once savannas. The habitat suitability index, using topographic predictors associated with fire behaviour, relates to the expansion and contraction of forest vegetation. The expansion and contraction dynamics are however more nuanced, with the in situ vegetation neighbourhood playing a large role. This is a dynamic system where both forest and savanna boundaries can and have changed considerably. The mechanism proposed here allowing for forest establishment is a local scale modification of fire regime creating pockets where fires are most likely to peter out resulting in patches with low fire return intervals. Fire sensitive vegetation establishes and persists in areas where the fire return interval is lower due to a topographic hindrance on fire spread. As opposed to widely available correlative approaches, a mechanistic understanding of fire spread and behaviour on complex terrain could advance our understanding of savanna-forest coexistence and provide insight into the presence and persistence of enigmatic forest patches in a fire-prone ecosystem. I used a fire spread model to predict fire behaviour (rate of spread) based on different wind directions and speeds in Hluhluwe Game Reserve on a homogeneous fuel layer to focus explicitly on the interaction between fire, topography and wind. Fire behaviour predictions were then compared to the earliest records of forest distributions, captured in aerial photographs in 1937. Wind direction used in the prediction of fire behaviour had a significant effect on the distribution of ‘fire shadows’. Large portions of the landscape show lower rates of spread. When predictions are based on actual fire season conditions these areas are generally occupied by forests. Fire refugia are important for the long term persistence of forests in fire prone landscapes. The firestorms in HiP crossed the boundary from savannas to closed woody vegetation with ease, switching from highly flammable grass fuels to forest leaf litter, understorey shrubs and herbs, and woody biomass, causing widespread mortality and topkill of adult trees. Extreme fires provide the opportunity for a biome switch from forest and thicket systems to savanna systems, even within these refugia. However, the timing of subsequent fires is vital for the colonisation and establishment of flammable grass clades and a savanna fire regime. Fire suppression within these mosaics may lead to a loss of resilience of forest margins as the seed bank of these pioneer species is depleted over time. In the future, chance fires which are not suppressed will lead to larger losses of forest area as shade intolerant pioneer species are not readily available to reclaim forest edges. There are clear differences in the drivers between high biomass forests and low biomass savannas, despite relatively similar net primary productivity. Much of the earlier trait based literature was assembled in temperate deciduous forests. If forests and savannas are ASS, they should persist long enough for discrete traits to diverge in the two systems with contrasting fire responses at the centre of these trait differences. Disturbance regime, as opposed to climate, is gaining recognition as a global driver of ecosystem functioning. The plant functional traits (PFT) governing fitness of forest and savanna trees differ. Within savanna trees, PFTs differ between demographic stages and the disturbance regime (type, history, severity, intensity of disturbance). Disturbance prone trees have developed a number of strategies, including clonal spread and belowground bud banks allowing basal resprouting. The difficulty going forward is to assimilate these traits into existing global frameworks. We explored the association between bark thickness and commonly measured traits in order to understand trait evolution in response to fire regime and other selective pressures such as herbivory and drought. Forest margins appear to be far more resilient, where they occur in mosaics with savannas in Africa, than forest interiors in the neotropics. Understanding the nature and extent of these impacts in African savannas is crucial due to the direct dependence of a large marginalized, vulnerable, rural population on savanna ecosystem services for agriculture and grazing. As climate changes, there is the potential for surprises such as more firestorms, as fire regimes are closely coupled to climate. An increase in these extreme events would allow more opportunities for catastrophic regime shifts and losses of forest vegetation. How does one reconcile lengthening fire seasons and more extreme fires with an increase in woody biomass? And what predictions can we make for the future of African savannas and forests?
|
|---|
