Terrestrial plant and soil respiration, or ecosystem respiration (R eco), represents a major CO 2 flux in the global carbon cycle. However, there is disagreement in how R eco will respond to future global changes, such as elevated atmosphere CO 2 and warming. To address this, we synthesized six years (2007-2012) of R eco data from the Prairie Heating And CO 2 Enrichment (PHACE) experiment. We applied a semi-mechanistic temperature-response model to simultaneously evaluate the response of R eco to three treatment factors (elevated CO 2, warming, and soil water manipulation) and their interactions with antecedent soil conditions [e.g., past soil water content (SWC) and temperature (SoilT)] and aboveground factors (e.g., vapor pressure deficit, photosynthetically active radiation, vegetation greenness). The model fits the observed R eco well ( R2=0.77). We applied the model to estimate annual (March-October) R eco, which was stimulated under elevated CO 2 in most years, likely due to the indirect effect of elevated CO 2 on SWC. When aggregated from 2007 to 2012, total six-year R eco was stimulated by elevated CO 2 singly (24%) or in combination with warming (28%). Warming had little effect on annual R eco under ambient CO 2, but stimulated it under elevated CO 2 (32% across all years) when precipitation was high (e.g., 44% in 2009, a 'wet' year). Treatment-level differences in R eco can be partly attributed to the effects of antecedent SoilT and vegetation greenness on the apparent temperature sensitivity of R eco and to the effects of antecedent and current SWC and vegetation activity (greenness modulated by VPD) on R eco base rates. Thus, this study indicates that the incorporation of both antecedent environmental conditions and aboveground vegetation activity are critical to predicting R eco at multiple timescales (subdaily to annual) and under a future climate of elevated CO 2 and warming.
