The complex interactions between plants, soils, and climates in semi-arid ecosystems make it difficult to define specific ecohydrological optimization mechanisms that underlie observed landscape-scale patterns in vegetation structure. There remains a need to clarify the manner by which vegetation self-organizes across scales within semi-arid landscapes, and how regional, landscape, and individual-based patterns of vegetation interact with their accompanying climates and soils. Such clarification necessitates the development of conceptual models capable of interpreting and predicting spatial pattern formation in savannas (and similar dry woodland ecosystems) as well as metrics for assessing optimization or organization of patterns as one scales from individual canopies to landscapes and beyond. Here we highlight some recent efforts to link the surface hydrological cycle to the dynamics of vegetation pattern, which we have characterized through both observations and modeling across a suite of semi-arid ecosystems. Our study area is a regional rainfall gradient in southern Africa (the Kalahari Transect). We use this natural gradient in vegetation structure and climate to reveal mechanisms of tree/grass interactions inferred from a combination of field observations and remote-sensing data coupled to models of surface energy/water balance.

Results/Conclusions At the regional scale, we find that ``down-regulation'' of grass production in dry seasons  reduces  tree water stress. In contrast, for wet years, a greater amount of water is removed from the upper soil layers with the increased dynamic grass cover, and therefore less water is lost from the base of the overall root zone. At smaller, canopy scales, we demonstrate pronounced changes in tree-grass interaction across a regional rainfall gradient. Importantly, we show that interaction in different parts of the Kalahari Transect can change in sign and magnitude over time, depending on both the average rainfall and inter-annual variability. For example, in the driest areas of the Kalahari the potential mutualism of larger trees supplying regeneration sites of increased moisture in an overall dry environment turns on and off depending on the climate. While our results suggest a general hypothesis that many semi-arid ecosystems tend to self-organize with respect to optimizing water use, a key difficulty arises in assessing which metrics of optimization are relevant at different spatial and temporal scales of interest. Regardless, the simulated linkages between vegetation structure and water availability for both trees and grasses emphasize the importance of assessing biophysical function of Kalahari savannas through ecohydrological techniques.