Catastrophic vegetation dynamics and soil degradation in semi-arid grazing systems.

When vegetation is drastically reduced as a result of drought or an increase in herbivore numbers, it does not simply recover if periods with normal rainfall follow or if herbivores are removed. These are commonly recognized catastrophic phenomena of semi-arid grazing systems in general and of the African Sahel in particular. The main aims of this thesis are to provide an effective explanation of the catastrophic properties of vegetation dynamics in these systems and to predict under which conditions they might be expected.We start with a description of Sahelian rangeland vegetation dynamics, to reveal its catastrophic properties. This exercise appeared a very useful first step in the growth of our ideas about catastrophic vegetation dynamics because: 1) it translated rather vague concepts into a verifiable format by deducing hypotheses about the conditions under which catastrophic vegetation dynamics might be expected, and 2) it generated the notion that soil degradation could somehow be an important factor attributing to catastrophic vegetation dynamics in semi-arid grazing systems. This is in contrast with models that emphasize herbivore feeding characteristics or plant competition as possible mechanisms underlying catastrophic vegetation dynamics. We tested the hypothesis that soil degradation, i.e. soil erosion by run-off and wind and the consequent loss of water and nutrients, is sufficient to explain catastrophic vegetation dynamics by mathematical modelling.Our model studies indeed show that soil degradation can effectively explain the catastrophic properties of semi-arid grazing systems. Soil degradation can cause a positive feedback between reduced resource (soil water and nutrients) availability and reduced vegetation biomass which may lead to collapse of the system. This positive feedback loop can be triggered by grazing. We argue on the basis of a large body of literature that this is an important mechanism causing catastrophic vegetation dynamics in semi-arid grazing systems. Furthermore, our model studies predict for which site-specific properties catastrophic vegetation dynamics may be expected, that is on loamy or clayey soils in case of water-limited vegetation biomass production, and on sandy soils in case of nutrient-limited biomass production. This is because sandy soils have higher water infiltration rates but are more vulnerable to nutrient loss through erosion than loamy or clayey soils.Based on our models, we hypothesized that the removal of aboveground herbaceous biomass would lead to a reduced soil water content and biomass production because of reduced water infiltration and increased run-off. We tested this hypothesis in a semi-arid savanna in Tanzania (East Africa). Indeed, as a consequence of biomass removal, a reduction in soil water content and biomass production occurred. But it appeared that increased loss of soil water through increased soil evaporation as a consequence of litter removal ultimately outbalanced all other effects on soil water content. Several factors might have contributed to the importance of increased soil evaporation, overriding that of reduced water infiltration and increased run-off. The soil in the research area was a sandy loam, with higher water infiltration rates than soils with a lower percentage sand and higher perentage clay, while rainfall primarily occurred in light showers. Thus, under these conditions, when the positive feedback between reduced water infiltration and reduced biomass does not operate, another positive feedback that is between increased soil evaporation and reduced biomass may become prominent.We further hypothesized that at a certain range of herbivore impact small initial differences in plant cover and amount of soil resources can magnify to alternative states which persist in time due to positive plant-soil feedbacks. We tested this hypothesis in a semi-arid grazing system in Burkina Faso (West Africa), where we studied vegetation patchiness along a gradient of herbivore impact. Indeed, the occurrence and likely persistence of a spatial pattern of vegetated patches alternating with bare soil at a certain range of herbivore impact could be explained by the positive plant-soil feedback between vegetation biomass and water infiltration.We stress the general applicability of our models by comparing catastrophic vegetation dynamics of the semi-arid grasslands of the African Sahel with the arctic salt marshes along the Hudson Bay in Canada. We argue that in both systems, an increase of herbivory triggered a catastrophic vegetation shift, which was ultimately caused by a positive plant-soil feedback, leading to desertification.One of our model assumptions was that herbivore density is not regulated by vegetation biomass. In the general discussion, I investigated the influence of a positive feedback between vegetation biomass and water infiltration on the dynamics of a plant-herbivore system, where herbivore density depends on vegetation biomass. As a consequence of the positive feedback and if herbivore reproduction is efficient, I predict that the plant-herbivore system could destabilize and collapse. In this chapter I also stress the practical relevance of our studies as our approach may finally lead to objective ecological criteria on which pastoral managers can base their decision how to evade the hazard of degradation of their rangelands.I highlight three topics which deserve more priority on the reseach agenda concerning semi-arid grazing systems in the near future. Hereby, I want to stress that it is important to put experimental and empirical studies into a clear theoretical framework, whereby mathematical modelling should play an important role. The three topics are:spatial heterogeneity and vegetation pattern formation,facilitation and competition between functional plant groups within the herbaceous layer andthe effects of positive plant-soil feedbacks on herbivore dynamics