Background/Question/Methods Grasslands and shrublands cover >40% of Earth's land surface and play an important role in global carbon cycling. These ecosystems are often plagued by short-term (one-year) droughts that are predicted to intensify with climate change. Previous experiments suggest that these ecosystems vary widely in their sensitivity to drought—possibly due to differences in study methods and because extreme drought is rarely imposed. Our research sought to better understand the effects of extreme drought on these ecosystems by 1) more accurately quantifying the effects of single year, extreme drought on aboveground net primary production (ANPP), 2) determining if the mean drought effect differed between grassland and shrubland ecosystems globally, and 3) reducing uncertainty in the magnitude of drought responses resulting from differences in experimental design. To achieve these goals, we used data from a coordinated distributed experiment across a broad range of grassland and shrubland sites globally: the International Drought Experiment or IDE. IDE sites constructed infrastructure that uses a passive approach to impose an extreme, 1-in-100-year drought based on each site’s historic precipitation record. Sites measured annual ANPP, a fundamental component of the global carbon cycle, as a metric for drought-induced impact to ecosystem function in these grasslands and shrublands.
Results/Conclusions In total, 94 of 136 sites met our criteria for inclusion. These criteria included deploying infrastructure in replicated plots capable of imposing a site-specific, historically-based extreme drought for a full growing season and following all other IDE protocols. The 94 sites in this analysis were arrayed across six continents in 20 countries and spanned broad climatic and edaphic gradients. On average, IDE sites reduced precipitation by 52%, which decreased aboveground net primary production (ANPP) by ~26% regardless of ecosystem type. However, there was substantial variation across sites, from catastrophic ( >80%) loss in ANPP to complete drought resistance (no change in ANPP). Our experimental approach resulted in an unprecedented number of sites (n = 40) experiencing a statistically extreme one-year drought, with less severe drought imposed in the remaining sites. Differences in drought severity explained 44% of variation in ANPP responses among ecosystems. When only those ecosystems exposed to historically extreme drought were considered, ANPP decreased by 40%, but variation in drought responses remained high. Our results demonstrate that ecosystem sensitivity to short-term drought varies dramatically, even with standardized approaches and comparably extreme droughts. Such inherent variability will challenge local-scale drought mitigation efforts as the climate changes.