Climate warming will affect terrestrial ecosystems in many ways, and warming-induced changes in terrestrial carbon (C) cycling could accelerate or slow future warming. So far, warming experiments have shown a wide range of C flux responses, across and within biome types. However, past meta-analyses of C flux responses have lacked sufficient sample size to discern relative responses for a given biome type. For instance grasslands contribute greatly to global terrestrial C fluxes, and to date grassland warming experiments provide the opportunity to evaluate concurrent responses of both plant and soil C fluxes. Here, we compiled data from 70 sites (in total 622 observations) to evaluate the response of C fluxes to experimental warming across three grassland types (cold, temperate, and semi-arid), warming methods, and short (≤3 years) and longer-term (>3 years) experiment lengths. Overall, our meta-analysis revealed that experimental warming stimulated C fluxes in grassland ecosystems with regard to both plant production (e.g., net primary productivity (NPP) 15.4%; aboveground NPP (ANPP) by 7.6%, belowground NPP (BNPP) by 11.6%) and soil respiration (Rs) (9.5%). However, the magnitude of C flux stimulation varied significantly across cold, temperate and semi-arid grasslands, in that responses for most C fluxes were larger in cold than temperate or semi-arid ecosystems. In semi-arid and temperate grasslands, ecosystem respiration (Reco) was more sensitive to warming than gross primary productivity (GPP), while the opposite was observed for cold grasslands, where warming produced a net increase in whole-ecosystem C storage. However, the stimulatory effect of warming on ANPP and Rs observed in short-term studies (≤3 years) in both cold and temperate grasslands disappeared in longer-term experiments (>3 years). These results highlight the importance of conducting long-term warming experiments, and in examining responses across a wide range of climate.
