Through litter decomposition enormous amounts of carbon is emitted to the atmosphere. Numerous large-scale decomposition experiments have been conducted focusing on this fundamental soil process in order to understand the controls on the terrestrial carbon transfer to the atmosphere. However, previous studies were mostly based on site-specific litter and methodologies, adding major uncertainty to syntheses, comparisons and meta-analyses across different experiments and sites. In the TeaComposition initiative, the potential litter decomposition is investigated by using standardized substrates (Rooibos and Green tea) for comparison of litter mass loss at 336 sites (ranging from -9 to +26 °C MAT and from 60 to 3113 mm MAP) across different ecosystems. In this study we tested the effect of climate (temperature and moisture), litter type and land-use on early stage decomposition (3 months) across nine biomes. We show that litter quality was the predominant controlling factor in early stage litter decomposition, which explained about 65% of the variability in litter decomposition at a global scale. The effect of climate, on the other hand, was not litter specific and explained <0.5% of the variation for Green tea and 5% for Rooibos tea, and was of significance only under unfavorable decomposition conditions (i.e. xeric versus mesic environments). When the data were aggregated at the biome scale, climate played a significant role on decomposition of both litter types (explaining 64% of the variation for Green tea and 72% for Rooibos tea). No significant effect of land-use on early stage litter decomposition was noted within the temperate biome. Our results indicate that multiple drivers are affecting early stage litter mass loss with litter quality being dominant. In order to be able to quantify the relative importance of the different drivers over time, long-term studies combined with experimental trials are needed.

The allocation and stoichiometry of plant nutrients in leaves reflect fundamental ecosystem processes, biotic interactions, and environmental drivers such as water availability. Climate change will lead to increases in drought severity and frequency, but how canopy nutrients will respond to drought, and how these responses may vary with community composition along aridity gradients is poorly understood. We experimentally addressed this issue by reducing precipitation amounts by 66% during two consecutive growing seasons at three sites located along a natural aridity gradient. This allowed us to assess drought effects on canopy nitrogen (N) and phosphorus (P) concentrations in arid and semiarid grasslands of northern China. Along the aridity gradient, canopy nutrient concentrations were positively related to aridity, with this pattern was driven primarily by species turnover (i.e., an increase in the relative biomass of N- and P-rich species with increasing aridity). In contrast, drought imposed experimentally increased N but decreased P concentrations in plant canopies. These changes were driven by the combined effects of species turnover and intraspecific variation in leaf nutrient concentrations. In addition, the sensitivity of canopy N and P concentrations to drought varied across the three sites. Canopy nutrient concentrations were less affected by drought at drier than wetter sites, because of the opposing effects of species turnover and intraspecific variation, as well as greater drought tolerance for nutrient-rich species. These contrasting effects of long-term aridity vs. short-term drought on canopy nutrient concentrations, as well as differing sensitivities among sites in the same grassland biome, highlight the challenge of predicting ecosystem responses to future climate change.

Extreme drought decreases aboveground net primary production (ANPP) in most grasslands, but the magnitude of ANPP reductions varies especially in C3‐dominated grasslands. Because the mechanisms underlying such differential ecosystem responses to drought are not well resolved, we experimentally imposed an extreme 4‐yr drought (2015–2018) in two C3 grasslands that differed in aridity. These sites had similar annual precipitation and dominant grass species (Leymus chinensis) but different annual temperatures and thus water availability. Drought treatments differentially affected these two semiarid grasslands, with ANPP of the drier site reduced more than at the wetter site. Structural equation modeling revealed that community‐weighted means for some traits modified relationships between soil moisture and ANPP, often due to intraspecific variation. Specifically, drought reduced community mean plant height at both sites, resulting in a reduction in ANPP beyond that attributable to reduced soil moisture alone. Higher community mean leaf carbon content enhanced the negative effects of drought on ANPP at the drier site, and ANPP–soil‐moisture relationships were influenced by soil C:N ratio at the wetter site. Importantly, neither species richness nor functional dispersion were significantly correlated with ANPP at either site. Overall, as expected, soil moisture was a dominant, direct driver of ANPP response to drought, but differential sensitivity to drought in these two grasslands was also related to soil fertility and plant traits.