Deciphering the stability of grassland productivity in response to rainfall manipulation experiments

Abstract

Aim:Rainfall manipulation experiments are essential tools for deciphering the mechanisms leading to variation in ecosystem stability across sites. Here, we gathered articles reporting results of experimental droughts on the above-ground biomass of grasslands to identify which indices have been used to assess stability, to evaluate the overall grassland responses to drought and to quantify the relative importance of drought characteristics and climatic conditions for explaining variation in stability. Location: Global.Time period: 1989-2018. Major taxa studied: Grasslands.Methods: We used meta-analytical approaches to evaluate overall grassland stability in terms of resistance, recovery and resilience, and multi-model inference to assess the relative importance of different moderators on explaining the variability of those three stability properties.Results: Numerous indices of stability have been used, but they are inadequate for comparisons across sites. After applying standardized indices, we found that grasslands were resilient (biomass remained unchanged 1 year after drought) and exhibited a trade-off between low resistance (biomass was lost during drought) and high recovery (new biomass was produced after drought). Overall, climatic conditions and drought characteristics (intensity, duration and frequency) were not important to explain the differences in stability observed across grasslands. Main conclusions:Grasslands are resilient, but if drought events last > 1 year, there might be long-term declines of biomass production owing to incomplete recovery.Despite the hundreds of experiments conducted in grasslands across the globe, the results are still inconclusive because of four important shortcomings: 50% of the studies have failed to create drought; 81% have not included recovery and resilience, assessing only resistance; 87% have not applied quantitative indices to assess stability; and < 1% of the studies were conducted on tropical grasslands. We discuss how to overcome those limitations to improve our ability to ensure stable grassland productivity under climate change. | 559MATOS eT Al.

Critical transitions in rainfall manipulation experiments on grasslands

“…Rainfall manipulation experiments are a powerful approach to test the existence of such critical thresholds, because they allow the assessment of vegetation response to precipitation reduction far below the current ranges (i.e., extreme droughts). Hundreds of rainfall manipulation experiments have been conducted across the world (Hoover, Wilcox, & Young, 2018), particularly in grasslands (Matos, Menor, Rifai, & Rosado, 2019), because it is relatively easier to install rain-out shelters in such open environments ( Figure 2), compared to shrublands and forests. Furthermore, grassland responses are usually quicker and easier to monitor (e.g., Hoover et al, 2018;Ruppert et al, 2015).…”

“…Actually, most of them have failed to subject grasslands to extreme drought treatments and to monitor biomass production after drought cessation (Estiarte et al, 2016;Gao et al, 2019;Matos et al, 2019). Additionally, most studies have assessed resilience using metrics (Ingrisch & Bahn, 2018) that disregard the possibility that alternative stable states may exist in the system (Kéfi et al, 2019;van Nes et al, 2016).…”

“…In rainfall manipulation experiments conducted to date, resilience has often been assessed by measuring how close the system is from its initial state (Ingrisch & Bahn, 2018;Kéfi et al, 2019;Smith, Wilcox, Power, Tissue, & Knapp, 2017). This is usually done by comparing ANPP values obtained before drought (ANPP1), immediately after drought cessation (ANPP2) and after the recovery period (ANPP3; Figure 5a,b; Matos et al, 2019). The closer ANPP3 values are from ANPP1, the more resilient the system is (i.e., in Figure 5a ANPP1 = ANPP3, thus RS = 0).…”

“…Even though some of those experiments have reported significant structural changes in species composition and abundances, they have not been designed to specifically test for alternative stable states of productivity (Kreyling, Jentsch, & Beier, ; Luo, Jiang, Niu, & Zhou, ; Smith, Knapp, & Collins, ). Actually, most of them have failed to subject grasslands to extreme drought treatments and to monitor biomass production after drought cessation (Estiarte et al, ; Gao et al, ; Matos et al, ). Additionally, most studies have assessed resilience using metrics (Ingrisch & Bahn, ) that disregard the possibility that alternative stable states may exist in the system (Kéfi et al, ; van Nes et al, ).…”

“…When conducting tests for hysteresis, researchers must be aware of the effects of interannual variations in precipitation on drought treatments (Matos et al, ). For instance, if the experiment is conducted in a grassland with MAP = 1,200 mm and during the first experimental year—that is when precipitation should be decreased by 10%—a natural drought event reduces total precipitation to 1,000 mm, then the actual precipitation received in the drought treatments would be 900 mm, instead of 1,080 mm.…”