Rainfall Manipulation Experiments in Semi-arid Grassland: What Have We Learned?

There is considerable evidence suggesting that worldwide, precipitation regimes (amount, variability, and extremes) have already been altered, and will continue to be to an even greater extent in the future. The ecosystem consequences of these alterations remain largely unresolved and poorly quantified. This is particularly problematic for water-limited, grass-dominated ecosystems such as grasslands and savannas. To generate the knowledge and understanding necessary to cope with a future with more variable precipitation regimes, we have conducted experimental manipulations of rainfall timing, mean rainfall event size, overall pattern, and rainfall extremes (drought and deluges) in a semi-arid native grassland. We have found that above-ground net primary production (ANPP) and other C cycling processes in semi-arid grassland can be as responsive to changes in rainfall pattern as these ecosystems are to changes in amount. For example, deluges (large single events) may even increase ANPP in this grassland relative to several smaller rain events. Furthermore, there is evidence that large event sizes may compensate for expected reductions in ANPP caused by drought. Finally, our research suggests that responses of C₄ and C₃ grasses to experimentally imposed extreme drought are inconsistent with expectations based on ecophysiological theory and their distributional responses to climatic means. This suggests that our understanding of dryland responses to nominal variation in rainfall may not be predictive of responses to extremes.