Controlling woody plant encroachment in a southern African savanna.

Doctoral Degree. University of KwaZulu-Natal, Pietermaritzburg.

Woody plant encroachment is considered one of the most extensive forms of degradation affecting savannas in arid and semi-arid ecosystems. Thus, reactive interventions such as chemical and mechanical controls, as well as fire application remains the norm in management of woody plant encroachment. I conducted a series of woody plant control experiments at the Agricultural Research Council’s Roodeplaat experimental ranch, situated in Gauteng Province, South Africa. The first experiment was a tree-thinning study at two savanna sites that differ in soil texture and woody species. Site 1 was on previously cultivated clay-dominated soils characterized by severe soil erosion and was encroached by Vachellia tortilis. Site 2 has never been cultivated and was on sandy soils with several woody species. At each site, 24 30 m × 30 m plots separated by 5 m wide fire breaks were established. Trees were removed to the approximate equivalents of 0% (control-no removal), 10%, 20%, 50%, 75% and 100% (complete removal of trees), followed by herbicide application on half of the stumps for each plot. I also investigated the effectiveness of Tree Poppers® (weed wrench) as a low-cost mechanical control tool to physically uproot seedlings and saplings of woody plants. To examine the effectiveness of the Tree Popper®, I used eight dominant tree species that were grouped into three height classes (0-49 cm, 50-99 cm, 100-150 cm) of ten seedlings and saplings per species per height class. In addition, investigated the effects of five years of annual burning on vegetation dynamics in a Vachellia karroo woodland. To determine the effects of annual burning on vegetation dynamics, plots (0.25 ha) established in 2013 were used. These studies are summarized below: (1) I determined the combined effects of tree species, tree thinning, stump diameter and herbicide application on resprouting patterns of woody plant species (Dichrostachys cinerea, Ehretia crispa, E. rigida, Gymnosporia buxifolia, Pappea capensis, Searsia lancea, S. caffra, Vachellia karroo, V. nilotica, V. robusta, V. tortilis and Ziziphus mucronata) that encroach study site 1. All the tree species in this study resprouted after cutting. Herbicide application significantly reduced the resprouting ability of D. cinerea, E. rigida, V. robusta and Z. mucronata. Tree removal positively influenced the resprouting ability and vigour of E. crispa only. The diameter of stumps was an important factor in determining resprouting ability, with shoot production decreasing with increasing stump diameter. The findings from this study suggest that woody plants are more likely to resprout and survive as juveniles than as adults after cutting. (2) I found no significant differences in the number of seedlings and saplings uprooted by Tree Popper®. However, there were significant differences in the number of juveniles uprooted using a Tree Popper® with a few individuals of Vachellia species uprooted. The effectiveness of the Tree Popper® may be due to differences in plant morphological structure, particularly the root system. The Tree Popper® is not an effective tool for controlling the Vachellia species used in this study. However, communal ranchers may mechanically control shallow-rooted tree seedlings with the Tree Popper® but not deep-rooted ones, specifically Vachellia species. (3) In the tree thinning experiment, I determined the effects of different tree thinningintensities on grass species-richness, composition, cover, β diversity, and soil fertility. I found that tree thinning did not have any significant effects on grass species-richness in either study site. However, we found a clear separation of different grass species among the treatments over the study period. Different levels of tree thinning increased the abundance of two dominant grass species (i.e. Digitaria eriantha and Panicum maximum) in both study sites, particularly in moderate (50%) and high removal (75% and 100%) treatments. However, the nitrophilous grass (i.e. P. maximum) will likely decline in abundance with time, particularly in the 100% thinning treatment because the ecological process that is responsible for N-fixation is no longer existent. Contrastingly, I found no evidence that tree thinning affects the amunt of soil cover. In addition, tree thinning did not have a significant impact on soil fertility in either study site. I recommend maintaining a stand density of 50% in rangeland affected by woody plant encroachment. In this study, 50% thinning created an opportunity for different palatable grass species to increase in abundance, which may help to increase forage production. (4) I determined the effects of different tree removal-intensities on grass production, tree-seedling establishment and growth, and the growth of the remaining large trees. In site 1, tree-removal treatments (i.e. 75 and 100%) significantly reduced grass biomass production after the first growing season, with no effect after the second season. In site 2, tree removal significantly increased grass biomass production. I found no significant effect of tree removal on tree seedling establishment in site 1. In site 2, tree removal had a significantly negative effect on overall tree seedling establishment. In both sites, there were no significant differences in tree seedling growth. Moderate (50%) to high (75%) removal of trees had a positive effect on the growth of remaining large trees in both study sites. I found that increased and/or diminished grass biomass production plays a vital role on tree seedling recruitment. Reduced tree competition facilitates the growth of the remaining large trees. An implication of these findings is that regardless of the substantial costs of woody plant control, the recovery of key ecosystem services such as an increased forage production may not be realised. However, this may be system-specific. In other systems, the absence of management interventions such as tree removal may compromise provision of ecosystem services and ecosystem functioning. (5) In the fire experiment, I investigated the effects of five years of annual burning on the density of young and adult Vachellia karroo plants. This study also aimed to investigate the effects of annual burning on tree growth (i.e. height, stem diameter and canopy size). The results supported the “fire-trap” paradigm by demonstrating substantially higher densities of young plants in the burned plots than in the unburned plots. In addition, the recruitment of young plants and saplings into adult trees was significantly higher in the unburned plots than in the burned plots. V. karroo populations substantially increased in growth (height and basal diameter) in the unburned plots. Different grass species changed in abundance in response to annual burning. However, I found no significant changes in grass species diversity and richness between the treatments. I found that the removal of the grass layer by fire and repeated topkill increased the number of young V. karroo individuals. Annual burning limited V. karroo juveniles and saplings from reaching an adult size class that may have detrimental effects on the herbaceous layer. I demonstrated that grass species composition is more prone to fireinduced changes than species diversity and richness in our study area. In conclusion, I show that managers of savanna rangelands may use annual burning to achieve specific vegetation structural objectives. This thesis demonstrated that mechanical- and chemical -control, as well as fire application influences the structure and functioning of savannas. By creating gaps that promote grass production, these management practices may assist increase the economic viability of savanna ecosystems. However, despite the popular belief that reduced tree densities promote ecosystem functions, this thesis demonstrates that the impact of control techniques (especially tree thinning) on forage production vary across savanna sites. This thesis also shows that management with prescribed annual fire reduced woody plant encroachment across the 5-year study, suggesting that fire management can be beneficial and should be explored as a management method.