Precipitation is a key driver of litter decomposition in arid/semiarid regions; where soils are poor in organic matter, and thus re-incorporation of litter is key for soil nutrient accumulation and soil structure. It remains unclear, though, whether litter decomposition responds symmetrically to precipitation variation (e.g., if precipitation surpluses produce a stimulatory effect of a similar magnitude, but opposite direction to inhibitory effects of precipitation deficits), and whether litter decomposition and litter nutrient dynamics in arid and semiarid ecosystems that differ in climate show similar responses to precipitation. We set up a 5-and-a-half-year experiment that manipulated rainfall along a gradient (7 treatments): increases by 20%, 40%, and 60%, background precipitation, and reductions by the same 3 percentages. We applied such experiment in two sites with different pattens of precipitation (Urat: arid; and Naiman: semiarid) in Inner Mongolia to elucidate our questions. Litter mass loss and all nutrients that we measured (carbon, nitrogen, phosphorous, potassium, plus lignin) decomposed faster at the highest level of surplus precipitation, and more slowly in the two largest precipitation reductions. This indicates that these levels of precipitation constitute thresholds (value of precipitation beyond which ecosystem function is critically altered). Litter decomposition in the semiarid site was faster and more complete, but decomposition in the direr Urat was more efficient per unit cumulative rainfall. Thus, site specific effects played an important role in decomposition. Reductions in precipitation decreased the loss of C, N, P, K, and lignin from litter; and clear precipitation thresholds in the dynamic of these nutrients in litter were observed. Overall, this indicated the importance of precipitation limitation at controlling nutrient release. Our study highlights the importance of long-term studies on litter decomposition in environments with slow decomposition rates, and the importance of taking into account mechanistic effects of water availability on decomposition.

降水是干旱/半干旱地区凋落物分解的关键驱动因素；土壤有机质贫乏，因此枯枝落叶的重新融入是土壤养分积累和土壤结构的关键。然而，目前尚不清楚凋落物分解是否对降水变化对称地响应（例如，降水过剩是否产生类似程度的刺激效应，但与降水不足的抑制作用相反的方向），以及干旱中的凋落物分解和凋落物养分动态是否相同。气候不同的半干旱生态系统对降水表现出相似的反应。我们进行了一项为期 5 年半的实验，沿梯度控制降雨量（7 种处理）：背景降水量增加 20%、40% 和 60%，并减少 3 个百分点。我们在内蒙古两个具有不同降水模式的地点（乌拉特：干旱；奈曼：半干旱）应用了这样的实验来阐明我们的问题。我们测量的凋落物质量损失和所有养分（碳、氮、磷、钾和木质素）在剩余降水量最高的情况下分解得更快，而在两次最大降水量减少的情况下分解得更慢。这表明这些降水水平构成了阈值（超过该降水值，生态系统功能就会发生严重改变）。半干旱地区的凋落物分解更快、更完全，但在干旱的乌拉特地区，每单位累积降雨量的分解效率更高。因此，位点特异性效应在分解中发挥了重要作用。降水量减少减少了凋落物中碳、氮、磷、钾和木质素的损失；并观察到凋落物中这些养分动态的明确降水阈值。总体而言，这表明降水限制对于控制养分释放的重要性。我们的研究强调了对分解速度慢的环境中的凋落物分解进行长期研究的重要性，以及考虑可用水对分解的机械影响的重要性。