Divergent responses of soil heterotrophic respiration to N deposition

Numerous studies have demonstrated that soil heterotrophic respiration (<i>R</i>_H) is significantly influenced by the atmospheric nitrogen (N) deposition. However, the response ratios (Δ<i>R</i>_H) exhibit substantial spatial heterogeneity worldwide. Here, based on a hierarchical meta-analysis of 912 paired observations from global N addition experiments, we show that this heterogeneity in Δ<i>R</i>_H can be explained by the divergences in the ecological characteristics of the sample sites. The mean annual precipitation (MAP) and mean annual temperature (MAT) at the sample sites exhibited linear negative effects on Δ<i>R</i>_H, especially in regions with high MAP (&gt; 700 mm) or MAT (&gt; 17 °C). Δ<i>R</i>_H significantly increased in alkaline soils (pH &gt; 7.5) but decreased in acidic soils (pH &lt; 5.5). Furthermore, Δ<i>R</i>_H significantly increased in soils with high total organic carbon (TOC &gt; 15 g/kg) or low available N contents (NH4+ &lt; 19 mg/kg or NO3− &lt; 15 mg/kg), but decreased in soils with high total N (TN &gt; 11 g/kg). Additionally, N addition altered soil properties, such as soil microbial biomass, pH, TOC, TN, C:N ratio, NH4+, and NO3−. The alternations in soil microbial biomass (ΔBiomass) and pH linearly affected Δ<i>R</i>_H at the global scale, whereas others linearly impacted Δ<i>R</i>_H only under certain threshold levels. Structural equation modeling demonstrated that the ΔBiomass, which directly and ultimately drives the variation in Δ<i>R</i>_H, was simultaneously modulated by climatic conditions, initial soil properties, and their alternations after N addition. Moreover, these factors significantly interacted with each other, and their relative importance to Δ<i>R</i>_H varied across sample sites. Using machine learning, we predicted the global Δ<i>R</i>_H following a 10% increase in atmospheric N deposition and observed significant heterogeneity (–32.3% to +142%). Our findings suggest that when accurately evaluating soil <i>R</i>_H to N deposition, the levels of sample site characteristics and their interactions should be comprehensively considered.