Modern soil system constraints on reconstructing deep-time atmospheric CO2

Paleosol carbonate-based estimates of paleo-atmospheric CO2 play a prominent role in constraining radiative-forcing and climate sensitivity in the deep-time. Large uncertainty in paleo-CO2 estimates made using the paleosol-carbonate CO2-barometer, however, arises primarily from their sensitivity to soil-respired CO2 (S(z)). This parameter is poorly constrained due to a paucity of soil CO2 measurements during carbonate formation in modern soils and a lack of widely applicable proxies of paleo-soil CO2. Here the δ13C values of carbonate and soil organic matter (SOM) pairs from 130 Holocene soils are applied to a two-component CO2-mixing equation to define soil order-specific ranges of soil CO2 applicable for constraining S(z) in their corresponding paleosol analogs.Equilibrium carbonate–SOM pairs, characterized by Δ13Ccarb–SOM values of 12.2–15.8‰, define a mean effective fractionation of 14.1‰ and overall inferred total soil CO2 contents during calcite formation of <1000–10,000 ppmv. For those Aridisols and Alfisols, characterized by a net soil-moisture deficit, and their paleosol analogs (Calcisols and Argillisols), a best estimate of S(z) during calcite formation is 1500–2000 ppmv (range of 500–2500 ppmv). Overall higher values (2000–5000 ppmv) are indicated by the subset of these soils characterized by higher moisture content and productivity. Near atmospheric levels (400 ± 200 ppmv) of estimated S(z) are indicated by immature soils, recording their low soil productivity. Vertisols define the largest range in total soil CO2 (<1000 to >25,000 ppmv) reflecting their seasonally driven dynamic hydrochemistry. A S(z) range of 1000–10,000 ppmv is suggested for paleo-Vertisols for which calcite precipitation can be constrained to have occurred in an open system with two-component CO2 mixing, with a best estimate of 2000 ppmv ± 1000 ppmv appropriate for paleo-Vertisols for which evidence of protracted water saturation is lacking. Mollisol pairs define a best estimate of S(z) of 2500 ppmv (range of 600–4000 ppmv) for late Cretaceous and Cenozoic analogs.Non-equilibrium pairs with Δ13C values >16‰ make up 51% of the dataset, lending support to the hypothesis that pedogenic carbonate precipitation occurs during periods of low productivity in a soil atmosphere with a large component of atmospheric CO2. Predictable scaling between estimated soil CO2 and the difference in δ13C between measured pedogenic carbonate and that predicted to have formed from soil-respired CO2 (inferred from measured SOM) can be used to further constrain appropriate ranges of S(z) for reconstruction of paleo-atmospheric pCO2. Soil CO2 estimates are poorly correlated to mean annual precipitation likely reflecting that for carbonate-bearing soils, where moisture limits CO2 production, total soil CO2 is most strongly influenced by actual evapotranspiration.