Get reliable science

The Grassland biome of South Africa has been identified as critically endangered and the biome in South Africa most requiring conservation attention through the implementation of efficient, sustainable systematic conservation plans. The ability to predict where land-cover transformation as well as information on the occurrence and severity of current land cover transformation activities, as threats to biodiversity, is required as part of the systematic conservation planning process. Neke & du Plessis (2004) predicted land cover transformation and the severity of the impact on biodiversity in the Grassland biome. This model was based on potential land use suitability models and land cover information for the 1994/5 season extracted from the National Land Cover database (NLC1994). These predictions were tested by assessing actual land cover change in the Grassland biome using observed differences in grassland land cover between the NLC1994 and NLC2000 databases. Methodology Because of differences in format and land-cover classification between the original datasets, both NLC1994 and NLC2000 had to be modified before any analyses could be carried out. These differences exist because different techniques were used to collate the respective datasets, thus introducing the potential for significant mapping error in the original datasets and more significantly erroneous results with respect to landcover change detection. The implications of this were presented in the discussion. Both datasets were spatially resampled and class-standardised and it was felt that this would significantly reduce any the impact of any such existing errords in the original datasets. Thereafter landcover information for the Grassland biome was be extracted and the comparative landcover analyses executed. The analyses carried out included: • Landcover change per landcover class within the Grassland biome with emphasis on the Grassland landclass losses and gains • An assessment and comparison of the relative fragmentation of the remaining grassland patches in both datasets • An assessment of current grassland habitat degradation • The comparison of the predicted land cover change as given by Neke & du Plessis (2004) against the observed grassland changes • The creation of a new Grassland Transformation threat map reflecting current land cover change threats, and including information pertaining to the threats to Grassland biodiversity posed by invasive alien plants, road effects, urban areas and soil erosion hazards. Results and Discussion 25% of the remaining grassland patches underwent transformation to other land classes. Grassland clearing for cultivation, bush encroachment and bushland vegetation regeneration were the main causal factors behind the observed grassland losses. However, grassland vegetation regeneration on formerly cultivated land, bush clearing and reclassification of degraded lands as grasslands in the NLC2000 dataset contributed to a net 2% gain in area of the grassland land class. The remaining grassland patches are more fragmented than they were in NLC1994, the average patch size (NLC2000) is three times smaller and the total number of grassland patches has increased (also by a factor of 3) and the remaining grassland patches are more isolated. The largest, least fragmented grassland patches occur along and to the west of the Great Escarpment as it traverses the Grassland biome. Most of the predictions of grassland transformation were realised, however the model used by Neke & du Plessis (2004) consistently underestimated and in some cases failed to predict the occurrence of grassland transformation in the central interior of the Grassland biome. Current, measurable human activities that act as grassland transformation agents were incorporated to create a threat map showing the extent and severity of land-cover transformation activities within the biome; grassland bird species richness information was then incorporated into this map to create biodiversity transformation threat map. This map was used to show the location and severity of the impacts of human transformation activities on grassland biodiversity. Both transformation threat map reflect the current situation across the biome today and were compared against the Potential transformation threat map produced by Neke & du Plessis (2004). The human transformation threat map confirmed the inability of the Neke & du Plessis model to make correct predictions of land cover change away from the eastern, 7 high altitude boundary of the biome. Given that the biome is defined by its climatic characteristics, the incorporation of global climate change effects would further refine the results gained, and perhaps provide more accurate predictions. As aforementioned, there are however factors existing within the original datasets used in this analysis that may have affected the accuracy of the landcover change analyses. These factors are centred on the potential effects of mapping errors within either of the NLC datasets. The delineation of landclass boundaries in the NLC1994 dataset is one such factor- placing a line over what is in reality a gradient of changing vegetation, is a subjective exercise and depends entirely on the technician involved this in itself may have introduced a fair amount of error in the mapping process. When coupled with the automated classification techniques used, for the most part, for the NLC2000 dataset, it becomes apparent that it is highly unlikely that even in the absence of actual landcover change the same boundaries would be drawn between two landclasses in the same area. This would provide false positive results for landcover change where in fact this is as a result of mapping errors. This is acknowledged and included in the interpretation of the results and it is felt that in spite of this, all possible steps were taken to minimize the impact of these effects on the reslults. The analysis allowed the identification of the current land cover transformations leading to grassland loss. However, land-cover change is only the physical expression of the complex interactions between socio-economic factors. To create effective and sustainable conservation plan for the Grassland biome, with an aim to reducing habitat loss requires an action plan to address these factors as the ultimate drivers of land cover change.