Triple oxygen isotope compositions of globally distributed soil carbonates record widespread evaporation of soil waters

The stable isotopic composition of pedogenic carbonates is central to many studies of past climate and topography, providing a basis for our understanding of Earth's terrestrial history. A core assumption of many applications of oxygen isotope values (δ18O) of pedogenic carbonate is that they reflect the δ18O value of precipitation (rain/snow). This assumption is violated if soil carbonates form in evaporated soil waters. In this work, we develop a means to identify evaporation in ancient soils using the triple oxygen isotope composition (16O-17O-18O) of pedogenic carbonates. Both theoretical predictions of isotope kinetics during evaporation and studies of triple oxygen isotopes in other geological materials show that the deviation in the relationship between δ17O and δ18O from a reference line, evaluated using the parameter Δ'17O, is sensitive to evaporation. As a first step in developing the use of Δ'17O in ancient pedogenic carbonates, we report Δ'17O values from 47 near-modern pedogenic carbonate samples from globally distributed environments that vary in aridity (hyper-arid to humid). The Δ'17O values of pedogenic carbonate range from -154 to -60 per meg (as CaCO3, measured via O2, VSMOW-SLAP), corresponding to calculated soil water values of -66 to +27 per meg (VSMOW-SLAP) (using a carbonate-water triple oxygen isotope fractionation exponent of 0.5250 and clumped isotope-derived carbonate growth temperatures). The Δ'17O values indicate that evaporative modification of soil water from which pedogenic carbonate forms is common, especially in arid environments. Arid environments host pedogenic carbonates formed from soil waters ranging from highly to minimally evaporated, while humid environments host pedogenic carbonates formed from waters that are only minimally evaporated. The variability in Δ'17O within environments classified by the same aridity may relate to the fact that pedogenic carbonates record soil conditions only during times of carbonate mineralization, which may deviate from annual conditions. Thus, Δ'17O may be useful in understanding the specific circumstances of pedogenic carbonate formation but may not provide incontrovertible evidence of the magnitude of environmental aridity. Evaporative modification of δ18O values of pedogenic carbonates can be detected with Δ'17O, thereby improving estimates of δ18O of unevaporated waters. Our data show that evaporation must be (re)considered for all paleoclimate inferences based on the δ18O of pedogenic carbonate. The addition of Δ'17O will re-energize paleoclimate studies that use (or have avoided using) δ18O of pedogenic carbonate.