Mapping Grazing-Induced Degradation in a Semi-Arid Environment: A Rapid and Cost Effective Approach for Assessment and Monitoring

2009 
Improved techniques for measuring and monitoring the state of biodiversity are required for reporting on national obligations to international and regional conservation institutions. Measuring the extent of grazing-related degradation in semi-arid ecosystems has proved difficult. Here we present an accurate and cost-effective method for doing this, and apply it in a South African semi-arid region that forms part of a globally significant biodiversity hotspot. We grouped structurally and functionally similar vegetation units, which were expert-mapped at the 1:50,000 scale, into four habitat types, and developed habitat-specific degradation models. We quantified degradation into three categories, using differences between dry and wet season values of the Normalized Difference Vegetation Index (NDVI) for the three succulent karoo habitats, and the difference between maximum and mean NDVI values for the subtropical thicket habitat. Field evaluation revealed an accuracy of 86%. Overall, degradation was high: 24% of the study area was modeled as severely degraded, and only 9% as intact. Levels of degradation were highest for bottomland habitats that were most exposed to grazing impacts. In sharp contrast to our methods, a widely used, broad-scale and snapshot assessment of land cover in South Africa was only 33% accurate, and it considerably underestimated the extent of severely degraded habitat in the study area. While our approach requires a multidisciplinary team, and in particular expert knowledge on the characteristics and spatial delimitation of vegetation types, it is repeatable, rapid, and relatively inexpensive. Consequently, it holds great promise for monitoring and evaluation programs in semi-arid ecosystems, in Africa, and beyond.
    • Correction
    • Source
    • Cite
    • Save
    • Machine Reading By IdeaReader
    62
    References
    36
    Citations
    NaN
    KQI
    []