The invasion of pteronia incana (blue bush) along a range of gradients in the Eastern Cape Province : it's spectral chacteristics and implications for soil moisture

摘要

Extensive areas of the Eastern Cape Province have been invaded by Pteronia incana (Blue bush), a non-palatable patchy invader shrub that is associated with soil degradation. This study sought to establish the relationship between the invasion and a range of eco-physical and land use gradients. The impact of the invader on soil moisture flux was investigated by comparing soil moisture variations under grass, bare and P. incana invaded surfaces. Field based and laboratory spectroscopy was used to validate P. incana spectral characteristics identified from multi-temporal High Resolution Imagery (HRI). A belt transect was surveyed to gain an understanding of the occurrence of the invasion across land use, isohyetic, geologic, vegetation, pedologic and altitudinal gradients. Soil moisture sensors were calibrated and installed under the respective surfaces in order to determine soil moisture trends over a period of six months. To classify the surfaces using HRI, the pixel and sub-pixel based Perpendicular Vegetation Index (PVI) and Spectral Mixture Analysis (SMA) respectively were used. There was no clear trend established between the underlying geology and P. incana invasion. Land disturbance in general was strongly associated with the invasion, as the endemic zone for the invasion mainly comprised abandoned cultivated and overgrazed land. Isohyetic gradients emerged as the major limiting factor of the invasion; a distinct zone below 619mm of mean annual rainfall was identified as the apparent boundary for the invasion. Low organic matter content identified under invaded areas was attributed to the patchy nature of the invader, leading to loss of the top soil in the bare inter-patch areas. The area covered by grass had consistently higher moisture values than P. incana and bare surfaces. The difference in post-rainfall moisture retention between grass and P. incana surfaces was significant up to about six days, after which a near parallel trend was noticed towards the ensuing rainfall episode. Whereas a higher amount of moisture was recorded on grass, the surface experienced moisture loss faster than the invaded and bare surfaces after each rainfall episode. ii There was consistency in multi-temporal Digital Number (DN) values for the surfaces investigated. The typically low P. incana reflectance in the Near Infrared band, identified from the multi-temporal HRI was validated by field and laboratory spectroscopy. The PVI showed clear spectral separability between all the land surfaces in the respective multi-temporal HRI. The consistence of the PVI with the unmixed surface image fractions from the SMA illustrates that using HRI, the effectiveness of the PVI is not impeded by the mixed pixel problem. Results of the laboratory spectroscopy that validated HRI analyses showed that P. incana’s typically low reflectance is a function of its leaf canopy, as higher proportions of leaves gave a higher reflectance. Future research directions could focus on comparisons between P. incana and typical green vegetation internal leaf structures as potential causes of spectral differences. Collection of spectra for P incana and other invader vegetation types, some of which have similar characteristics, with a view to assembling a spectral library for delineating invaded environments using imagery, is another research direction.