| This work examines how water availability affects vegetation structure and vegetation-atmosphere exchange of water, carbon, and energy for a savanna ecosystem. The study site is the Kalahari Transect (KT), in southern Africa, which follows a north-south decline in mean annual rainfall from ∼1600 mm/yr to ∼250 mm/yr between the latitudes 12°–26°S. Eddy covariance (EC) flux measurements taken over a time frame of 1–9 days at four sites along the transect during the wet (growing) season revealed that the ecosystem water use efficiency for the sites, defined as the ratio of net carbon flux to evapotranspiration, decreased with increasing mean annual rainfall. EC data were used to parameterize a large eddy simulation model, which was applied over a heterogeneous remotely-sensed surface. Water availability for the vegetation was found to affect the relative controls (structural vs. meteorological) on the spatial distribution of vegetation fluxes. When the spatial distribution of vapor pressure deficit, D, was most predictable (i.e. non water-limiting conditions) it was unimportant in shaping the distribution of the vegetation fluxes, while at times when D was least predictable (i.e. water-limiting conditions) it was most important. This observation is explained by the relative degree of vegetation-atmosphere coupling and the complexity of the non-local effects on D , both of which are dependent upon water availability.; Based upon the differing ways in which trees and grass respond to interannual variability in rainfall, a new method was developed to estimate fractional tree, grass, and bare soil cover from a synthesis of satellite and ground-based data. This method was applied to the KT where it was found that tree fractional cover declines with mean annual rainfall, while grass fractional cover peaks near the middle of the gradient. A soil moisture model applied to this data indicated a shift from nutrient- to water-limitation from the mesic to arid portions of the transect. The spatial extent of the tree and grass cover in the water-limited portion of the transect could be explained by separately considering the water demand stress on the individual cover fractions, while minimizing the amount of water lost as leakage from the root zone. |
