In this study, we scrutinise the GLASOD assessments for the African continent for reasons of data availability and policy-relevance; the social and economic impact of land degradation seems to be most severe in Africa. First, we test the GLASOD assessments for their consistency by comparing expert judgments on the status of soil degradation for similar combinations of land and land use. Secondly, we evaluate the reproducibility of expert judgments by estimating an ordered logit model that relates degradation classes to easily-available information on explanatory variables, so as to make land degradation assessments at unvisited sites. Thirdly, we analyze the impact of the land degradation on food production in a cross-sectional analysis relating GLASOD assessments to crop production data at sub-national level. To account for climatic variability we express productivity as a ratio of actual to potential yield, while soil fertility appears explicitly.
Furthermore, we analyze the association of the degradation-productivity relationship with prevalence of malnutrition and fertilizer usage. So, how good is GLASOD? We find that the experts were not very consistent in assigning soil degradation classes to similar sites, possibly because they had different concepts of the degrees of degradation - these differences are likely to be more pronounced when experts come from different countries and have dissimilar experience of land degradation. Because of this lack of consistency, it is difficult to reproduce expert judgments with a parametric model approach. The findings confirm the results in other studies where deviating trends of expert assessments make it necessary to use country dummies in the qualitative response models to correct for interpretation differences among the international forum of experts.
The findings on the relationship between yields and land degradation were counter-intuitive: yields increase for higher levels of land degradation. Apparently, more intensive cultivation without appropriate soil protection measures causes higher degradation levels but does not necessarily reduce productivity. Moreover, yields on more productive but severely degraded soils are largely maintained by application of fertilizer. Better soils also seem to resist the impact of the lower levels of degradation without the need for fertilizers, yet, yields deteriorate rapidly for the more degraded areas with poorer soils. The high prevalence of malnutrition in areas with declining yields on the poor and highly degraded soils is alarming indeed.
We conclude that the GLASOD expert assessments are not very reliable. However, our verdict should not be too harsh. With slender resources, and in a very short time, a global assessment was completed that clearly depicted, for the first time, the extent and degree of land degradation. Its limitations were made clear by the authors and, in spite of these limitations, GLASOD underpinned environmental policy discussions - it has been the only information available. Improved methods of assessment of land degradation are now needed to provide decision makers with the appropriate information for the development of sound environmental policies and it is likely that any new global assessment will have to resort, in some degree, to expert judgments - so the lessons learned from this GLASOD analysis will be valuable.