No abstract

In contrast to active tectonic settings, little is known about the potential feedback between surface processes and climate change in tectonically inactive cratonic regions. Here, we studied the driving forces of erosion and landscape evolution in the Kruger National Park in South Africa using cosmogenic nuclide dating. 10Be‐derived catchment‐wide erosion rates (~2 and ~10 mm ka−1) are similar in magnitude to erosion and rock uplift elsewhere in South Africa, suggesting that (1) rock uplift is solely the isostatic response to erosion and (2) the first‐order topography is likely of Cretaceous age. The topographic maturity is promoted by widespread exposure of rocks resistant to erosion. Our data, however, suggest that local variations in rock resistance lead to transient landscape changes, with local increases in relief and erosion rates.

Sediment source fingerprinting using environmental magnetism has successfully differentiated between sediment sources in several studies in the Eastern Cape Province of South Africa. The method was applied in this study to the near-natural landscape of southern Kruger National Park (Mpumalanga Province) to trace sediment and determine sediment yields by lithology in four reservoir catchments that were underlain by igneous, metamorphic, and sedimentary rocks. The Park area has no history of cultivation and is a conservation area, so catchment sources were dominated by underlying lithologies. One sediment core was collected in the assumed deepest area of the reservoir. Source discrimination and apportionment were estimated using a common statistical protocol that includes a Mann-Whitney U or Kruskal-Wallis H test, mass conservation test, discriminant function analysis, and an (un)mixing model. A contribution from each lithology-defined source was estimated.Sediment yield by lithology was estimated by using published catchment areaspecific sediment yields in combination with the (un)mixing model. Underlying lithology determined vegetation type and density, and vegetation appeared to play a crucial role in protecting soils and reducing erosion. Proximity to reservoir, that is, travel distance for eroded sediment, and connectivity were also important factors controlling the relative contribution from each potential source. The contributing area for sediment was found to be dynamic through time and was probably dependent on runoff and temporal variations in vegetation cover.

No abstract

Sediment source fingerprinting using environmental magnetism has successfully differentiated between sediment sources in different regions of South Africa. The method was applied in the natural landscape of the Kruger National Park to trace sediment sources delivered to four reservoirs (Hartbeesfontein, Marheya, Nhlanganzwani, Silolweni) whose contributing catchments were underlain by a range of igneous, metamorphic, and sedimentary rocks. This research attempted to evaluate the impact of vegetation, lithology, and particle size controls on the ability of magnetic signatures to discriminate between lithology-defined potential sources. Potential source samples were collected from each lithology present in all catchments, except for the Lugmag catchment where the lithology was uniform, but the vegetation type varied significantly between woodland and grassland. One sediment core was taken in each of the four catchment reservoirs where there was more than one lithology present in order to unmix and apportion contributing sediment sources. Sampling time in the field was often restricted to short periods, dependent on anti-poaching activities and movement of free-roaming wildlife across the Park. This occasionally led to the sub-optimal collection of enough source samples to capture source signature variability. Mineral magnetic parameters were unable to discriminate between vegetation-defined sediment sources in the Lugmag catchment (homogenous underlying lithology) but were able to discriminate between lithology-defined sediment sources (to varying degrees) in the other four catchments. The contributions of each lithology-defined sediment source were estimated using a straightforward statistical protocol frequently used in published literature that included a Mann-Whitney U or Kruskal-Wallis H test, mass conservation test, discriminant function analysis, and an (un)mixing model. A contribution from each lithology source to reservoir sediment was estimated. Connectivity was a significant factor in understanding erosion in each of the catchments. Both longitudinal (e.g., drainage density) and lateral connectivity (e.g., floodplain - river) were important. Travel distance of eroded sediment to reservoirs was also an essential element in two of the four catchments. There are no defined floodplains, so channel bank soils are very similar to the catchment soils. Therefore, channel bank storage potential would be similar to the storage potential within the catchment. Vegetation played a crucial role in protecting soils, by reducing ii erosion potential as well as trapping and storing sediment, thereby interrupting lateral connectivity. Underlying geology and soils are determining factors of vegetation type and density. A published study estimated catchment area-specific sediment yields for different KNP catchments, including the Hartbeesfontein, Marheya, Nhlanganzwani and Silolweni catchments. The published data was used in combination with the (un)mixing model source contribution estimates of this thesis to determine specific sediment yields by lithology, i.e., for each catchment source. The polymodal particle size characteristics of the sample material led to an investigation into particle size controls on the ability of magnetic signatures to discriminate between potential sources. Due to time constraints, only the Hartbeesfontein and Marheya catchments were tested for grain size differences. For each catchment, one bulk sample was created for each lithology source. This bulk sample was divided into 10 subsamples. The samples were then fractionated into four particle size fraction groups: coarse (250 – 500 μm), medium (125 – 250 μm), fine (63 – 125 μm), and very fine (<63 μm). Reservoir samples were also bulked to create 10 down-core samples for each reservoir, and the samples were also fractionated into the four fraction groups. The same statistical protocol was applied to the fractionated samples and contribution estimates were obtained by lithology for each particle size fraction group. The goodness of fit and uncertainty of the (un)mixing model varied in each catchment, with the two measures of accuracy often showing an inverse relationship. The fractionated modelling estimated the same primary source in the two catchments as in the unfractionated modelling. However, additional information on the secondary and tertiary sources was obtained. Connectivity remained a significant factor in interpreting the results of the fractionated analysis. Specific sediment yields were estimated for each catchment source per particle size fraction group. These sediment yields provided a deeper understanding of sediment transport through a catchment and which particle size groups are most important in catchment erosion. An original contribution to research was made by estimating source contribution estimates for the four reservoirs, quantifying sediment yields for each catchment lithology and then for each catchment lithology by particle size. Mineral magnetic tracing of the catchments was applied for the first time in this region of South Africa.