The factors that influence the invasion of hillslopes by the shrub Pteronia incana in the communal rangelands of Ngqushwa (formerly Peddie) district were investigated. Mgwalana, one of the catchments in the district where encroachment by the species is widespread, was chosen. The study combined field observations with image analysis based on high resolution infrared imagery. The catchment was flown and high resolution infrared images (1mx1m) were taken using a Kodak DCS420 digital, colour-infrared camera. The images were analysed using Idrisi32 and Kilimanjaro GIS versions. The ability of different vegetation indices to separate P. incana from the other cover types was investigated. Field observations of the degree of P. incana invasion in relation to, inter alia, soil surface conditions, slope angle and visible forms of erosion were made. A Digital Elevation Model (DEM) of 20m spatial resolution was used to derive terrain parameters. The presence and absence of P. incana in relation to slope gradient and aspect were determined. The combined influence of the two terrain parameters and land use on the invasion was also investigated. The Topographic Wetness Index (WI), a component of the TOPMODEL was derived from the DEM and its relationship with the spatial distribution of P. incana was explored. Soil moisture dependencies for P. incana and grass species as well as surrogates for runoff under the shrub and adjacent bare areas were determined in the field. A high level of classification accuracy confirmed the reliability of digital camera imagery for spatial analyses. Distinct spectral separability for the surface vegetation cover types was achieved by means of the Perpendicular Vegetation Index (PVI) as opposed to the ratio based vegetation indices (NDVI, SAVI and MSAVI). The absence/presence of P. incana was noted to be strongly influenced by slope angle and aspect. The probability for P. incana occurrence increased with slope steepness and southerly slope orientation. Abandoned and grazing lands were identified as the main invasion hotspots. Blanket invasion of the former signified the high susceptibility of abandoned land to P. incana invasion. The combined influence of land use, slope gradient and aspect was also noted to have promoted the invasion. This is borne out by the concentration of the invasion on abandoned steep slopes with a southerly orientation. Local topographic variations were identified as having a strong bearing on P. incana spatial distribution. The topographically driven WI confirmed this relationship, such that P. incana was associated with the low WI values of convexities. Differences in the moisture dependencies between P. incana and grass species were demonstrated by the greater rooting depth of the former. During field surveys, soil surface crusting was noted as inherent to P. incana patchiness. The coupling between local topography and soil surface crusting underpins soil moisture variability at hillslope and patch scales respectively. This in turn determines the competition between P. incana and grass species and the eventual replacement of the latter by the former. A close spatial correlation between fully established P. incana and severe forms of soil erosion was observed. Loss of patchiness and expansion of inter-patch bare areas promote runoff connectivity erosion. As most of the runoff becomes run out, hillslopes tend towards dysfunctional systems. Greater soil moisture storage after rainstorms under P. incana tussocks than the adjacent bare areas signifies the shrub’s water harvesting capabilities. The tussocks could thus serve as a starting-pointbuilding- block for the rehabilitation of dysfunctional hillslope systems.
Identifer | oai:union.ndltd.org:netd.ac.za/oai:union.ndltd.org:rhodes/vital:4833 |
Date | January 2004 |
Creators | Kakembo, Vincent |
Publisher | Rhodes University, Faculty of Science, Geography |
Source Sets | South African National ETD Portal |
Language | English |
Detected Language | English |
Type | Thesis, Doctoral, PhD |
Format | 229 leaves, pdf |
Rights | Kakembo, Vincent |
Page generated in 0.0016 seconds