Elevated CO₂ generally stimulates plant productivity, but herbaceous plants have only a limited capacity to sequester C so that the ecosystem C sequestration will be governed by soil C accumulation. At the global scale, extra soil C sequestration constitutes a negative feedback on the ongoing atmospheric CO₂ rise. Measuring C sequestration directly is notoriously difficult for methodological reasons, but modeling and carbon isotope techniques can overcome some of these limitations. We have investigated soil C sequestration in a Spring wheat ecosystem exposed for five years to elevated CO₂. Combining C isotope data with the Rothamsted soil C model (“RothC”), we partitioned net C sequestration rates into changes in new C fixed over the experimental period (C_new) and old, pre-experimental C (C_old). Elevated CO₂ plots were exposed to CO₂ depleted in ¹³C, allowing to quantify soil C_new inputs throughout the experiment.

Results/Conclusions
Isotope mass balance indicated a net accumulation of 396 g C_new m^-2 over five years, of which ~ 26 g C m^-2 were in soil microbial biomass. While this sets an upper constraint to soil C sequestration, estimates of decomposition of C_old are required to infer on net soil C sequestration. We obtained these running a modified version of RothC that could be driven with measured soil moisture and temperature data. RothC accurately reproduced the measured C pools and fluxes. Inverse modeling predicted gross C inputs of 1000±100 g C m^-2 under elevated CO₂. Over the same period, an estimated 720 g C m^-2 of C_old were decomposed. The extra plant productivity under elevated CO₂ let to an extra C_new accumulation of ca. 40 g C m^-2 relative to ambient conditions. However, when taking into account the accelerated decomposition of C_new and C_old due to higher soil moisture in elevated CO₂, CO₂ enrichment resulted in a net loss of ~30 g C m^-2 relative to ambient conditions. A similar analysis of C sequestration for extensively managed calcareous grassland exposed to elevated CO₂ for six years showed a very similar picture for both ecosystems.

In conclusion, our results indicate that increased soil moisture under elevated CO₂ (due to lower stomatal conductance) can reduce the soil carbon balance of elevated CO₂ plots from a sink to a source (relative to ambient conditions), despite higher plant production in elevated CO₂ plots. This underlines the paramount importance of indirect effects of soil moisture changes under elevated CO₂.