EQUILIBRIUM RESPONSES OF SOIL CARBON TO CLIMATE-CHANGE - EMPIRICAL AND PROCESS-BASED ESTIMATES

We use a new version of the Terrestrial Ecosystem Model (TEM), which h as been parameterized to control for reactive soil organic carbon (SOC ) across climatic gradients, to evaluate the sensitivity of SOC to a 1 degrees C warming in both empirical and process-based analyses. In th e empirical analyses we use the steady state SOC estimates of TEM to d erive SOC-response equations that depend on temperature and volumetric soil moisture, and extrapolate them across the terrestrial biosphere at 0.5 degrees spatial resolution. For contemporary climate and atmosp heric CO2, mean annual temperature explains 34.8% of the variance in t he natural logarithm of TEM-estimated SOC. Because the inclusion of me an annual volumetric soil moisture in the regression explains an addit ional 19.6%, a soil moisture term in an equation of SOC response shoul d improve estimates. For a 1 degrees C warming, the globally derived e mpirical model estimates a terrestrial SOC loss of 22.6 10(15) g (Pg), with 77.9% of the loss in extra-tropical ecosystems. To explore wheth er loss estimates of SOC are affected by the spatial scale at which th e response equations are derived, we derive equations for each of the eighteen ecosystems considered in this study. The sensitivity of terre strial SOC estimated by summing the losses predicted by each of the ec osystem empirical models is greater (27.9 Pg per degrees C) than that estimated by the global empirical model; the 12.2 Pg loss (43.7%) in t ropical ecosystems suggests that they may be more sensitive to warming . The global process-based loss of SOC estimated by TEM in response to a 1 degrees C warming (26.3 Pg) is similar to the sum of the ecosyste m empirical losses, but the 13.6 Pg loss (51.7%) in extra-tropical eco systems suggests that they may be slightly less sensitive to warming. For the modelling of SOC responses, these results suggest that soil mo isture is useful to incorporate in empirical models of SOC response an d that globally derived empirical models may conceal regional sensitiv ity pf SOC to warming. The analyses in this study suggest that the max imum loss of SOC to the atmosphere per degrees C warming is less than 2% of the terrestrial soil carbon inventory. Because the NPP response to elevated CO2 has the potential to compensate for this loss, the sce nario of warming enhancing soil carbon loss to further enhance warming is unlikely in the absence of land use or changes in vegetation distr ibution.