Using Triple Oxygen Isotopes of Pedogenic Carbonate to Identify Ancient Evaporation: First Steps from Modern Soils

The stable isotope composition of soil carbonates is commonly used to reconstruct continental paleoclimates, but its utility is limited by an incomplete understanding of how soil carbonates form. In particular, it is often unclear if the parent soil water has been enriched in ¹⁸O due to evaporation, muddying our ability to infer meteoric water δ¹⁸O from paleosol carbonates. Here we demonstrate the potential use of triple oxygen isotopes (termed ∆’¹⁷O) to account for evaporation and identify formation process through a study of modern soil carbonate isotope values.  Evaporation results in a decreased slope in the relationship between δ¹⁷O and δ¹⁸O and deviations from the global meteoric water line, such that ∆^’17O values in soil water and resulting carbonate decrease with increased evaporation. We report ∆^’17O values of CO₂ derived from soil carbonates and measured as O₂ on a mass spectrometer, with 1-4 replicates per soil carbonate. We find a step-like relationship between ∆’¹⁷O in globally distributed Holocene soil carbonate samples and aridity, where aridity is defined using the aridity index (AI, mean annual precipitation/potential evapotranspiration). Low ∆^’17O values occur in hyper-arid climates (AI < 0.05), with mean ∆^’17O = -0.164 ‰, SD = 0.004 ‰. A transition, or step, occurs in arid climates (AI from 0.05 to 0.2), with ∆^’17O values that range from -0.129 ‰ to -0.165 ‰, and mean ∆^’17O of -0.148 ‰, SD = 0.010‰. High ∆^’17O values occur in semi-arid through humid climates (AI >0.5) with mean ∆^’17O of -0.135 ‰, SD = 0.008 ‰.  The lowest observed ∆^’17O values are consistent with extensive evaporation – for context, the ∆^’17O values are similar to those measured in lacustrine carbonates from closed lake basins. The highest ∆^’17O values are consistent with little soil water evaporation. We interpret the step-like pattern in ∆’¹⁷O values as an indication of the threshold in the importance of evaporation vs. transpiration in soil dewatering. This data highlights the potential to use ∆^’17O to identify the extent of evaporation in paleosol carbonates. Eventually, we hope that this novel technique will lead to quantitative accounting of evaporation in soil water and improved reconstructions of meteoric water δ¹⁸O from soil carbonates. The ability to constrain the evaporative conditions of soil carbonate formation will also aid interpretations of δ¹³C (including pCO₂ reconstructions) and clumped isotope-based temperatures. These efforts will ultimately aid in our ability to integrate paleoclimate data from soil carbonates with data from other terrestrial records.