Nonlinear carbon cycling responses to precipitation variability in a semiarid grassland

Changes in precipitation amount and variability would profoundly affect carbon (C) cycling in arid and semiarid grasslands. However, compared to the effects of precipitation amounts, little is understood about the impacts of precipitation temporal variability on terrestrial C cycling. To explore relationships between precipitation variability and C cycling processes and the underlying mechanisms, we conducted a 3-year field experiment and a 12-year model simulation, in which the constant seasonal precipitation amount was temporally manipulated with four and six levels of precipitation variability, respectively, in a semiarid grassland. Based on the manipulative experiment, we found various nonlinear relationships between C cycling processes and the coefficient of precipitation variability (P-cv), including a trinomial relationship for soil respiration (tipping point: 3 and 4.3), convex relationships for gross and net ecosystem production (peak at 3.5) as well as belowground biomass (peak at 4) and a nonlinear negative relationship for ecosystem respiration (peak at 2.5). Such relationships were regulated by seasonal averaged soil water content (SWC), early-growing season precipitation amount, soil inorganic nitrogen availability (SIN), and both SWC and SIN, respectively. However, these results from the manipulative experiment did not match those from the model simulation, in which ecosystem C cycling processes, dominated only by SWC responses, showed positive linear responses to P-cv. Our results mirror that the nonlinear responses of grassland C cycling to precipitation variability as regulate by SWC and SIN should be incorporated into models to forecast future ecosystem shifts under climate change. (C) 2021 Elsevier B.V. All rights reserved.