Simple additive simulation overestimates real influence: altered nitrogen and rainfall modulate the effect of warming on soil carbon fluxes.

Experiments and models have led to a consensus that there is positive feedback between carbon (C) fluxes and climate warming. However, the effect of warming may be altered by regional and global changes in nitrogen (N) and rainfall levels, but the current understanding is limited. Through synthesizing global data on soil C pool, input and loss from experiments simulating N deposition, drought and increased precipitation, we quantified the responses of soil C fluxes and equilibrium to the three single factors and their interactions with warming. We found that warming slightly increased the soil C input and loss by 5% and 9%, respectively, but had no significant effect on the soil C pool. Nitrogen deposition alone increased the soil C input (+20%), but the interaction of warming and N deposition greatly increased the soil C input by 49%. Drought alone decreased the soil C input by 17%, while the interaction of warming and drought decreased the soil C input to a greater extent (-22%). Increased precipitation stimulated the soil C input by 15%, but the interaction of warming and increased precipitation had no significant effect on the soil C input. However, the soil C loss was not significantly affected by any of the interactions, although it was constrained by drought (-18%). These results implied that the positive C fluxes-climate warming feedback was modulated by the changing N and rainfall regimes. Further, we found that the additive effects of [warming × N deposition] and [warming × drought] on the soil C input and of [warming × increased precipitation] on the soil C loss were greater than their interactions, suggesting that simple additive simulation using single-factor manipulations may overestimate the effects on soil C fluxes in the real world. Therefore, we propose that more multifactorial experiments should be considered in studying Earth systems.