<p>Structural diversity of water-limited savanna and ecosystem response to climatological variability

Determining the sensitivity of ecosystem response to environmental variability is fundamental to understand the capacity of ecosystems to resist and recover from change. Despite limitations to infer about biotic mechanisms, remote sensing has the advantage of high temporal resolution, over long records, to address fundamental challenges in ecology in an environmental context. The degree of structural diversity in water-limited savanna changes across a climatological gradient over the sand covered area in interior southern Africa, between 29° S to the equator and 14° E to 28° E. To determine the coupling between structural diversity and ecosystem function in water-limited ecosystems, we use a 15-year record of daily precipitation, and satellite data for Normalized Difference Vegetation Index (NDVI), matching the same time period (2001-2015), along the Kalahari sand sheet. Savanna ecosystems show a varied temporal sensitivity to environmental fluctuation going from 200 mm to 750 mm of mean annual precipitation. Increasing grass cover results in higher sensitivity to intra-annual dynamics, while an increased woody cover reflects purely annual patterns. Lagged correlations suggest a similar sensitivity at the extreme ends of the gradient, but highest in the middle, between 400 mm and 450 mm of mean annual precipitation. The vegetation in this region, described as a wooded grassland, has contributions to net primary productivity more equally shared by trees and grasses. In particular, the role of rainfall seasonality, magnitude, and frequency is examined in the different savanna, against different indices of ecosystem stability. Finally, using the savanna example to better understand trade-offs, such as how resistance versus recovery, and similarly, stress avoidance versus resource use optimization under a gradient of conditions, highlights the importance of structural diversity in determining ecosystem response to environmental variability.