Clumped isotope paleothermometry using pedogenic carbonates is a powerful tool for investigating past climate changes. However, location-specific seasonal patterns of precipitation and soil moisture cause systematic biases in the temperatures they record, hampering comparison of data across large areas or differing climate states. To account for biases, more systematic studies of carbonate forming processes are needed. We measured modern soil temperatures within the San Luis Valley of the Rocky Mountains and compared them to paleotemperatures determined using clumped isotopes. For Holocene-age samples, clumped isotope results indicate carbonate accumulated at a range of temperatures with site averages similar to the annual mean. Paleotemperatures for late Pleistocene-age samples (ranging 19-72 ka in age) yielded site averages only 2 degrees C lower, despite evidence that annual temperatures during glacial periods were 5-9 degrees C colder than modern. We use a 1D numerical model of soil physics to support the idea that differences in hydrologic conditions in interglacial versus glacial periods promote differences in the seasonal distribution of soil carbonate accumulation. Model simulations of modern (Holocene) conditions suggest that soil drying under low soil pCO2 favors year-round carbonate accumulation in this region but peaking during post-monsoon soil drying. During a glacial simulation with lowered temperatures and added snowpack, more carbonate accumulation shifted to the summer season. These experiments show that changing hydrologic regimes could change the seasonality of carbonate accumulation, which in this study blunts the use of clumped isotopes to quantify glacial-interglacial temperature changes. This highlights the importance of understanding seasonal biases of climate proxies for accurate paleoenvironmental reconstruction. Reconstructing the amount of temperature change associated with past climate changes for individual regions is important for understanding their climate vulnerability. Carbonate minerals developed naturally in desert soils record past temperatures in the numbers of their rare isotopes, called clumped isotopes. However, the temperature recorded in soil minerals is linked to the time of year they form, which varies greatly from winter to summer, so understanding the timing is key to interpreting past climate. We measured underground temperatures in the southern Rocky Mountains, compared them to mineral temperatures from young soils, and found that they record mean annual soil temperature. In contrast, temperatures recorded by soil minerals during the last ice age were only 2 degrees C colder than young soil temperatures, despite evidence that ice age air temperatures were 5-9 degrees C colder. We performed numerical modeling to predict the seasonal timing of soil carbonate accumulation under interglacial and glacial climate states and found that carbonate likely forms year-round during interglacial states but forms during the summertime under glacial conditions due to delayed melting of snow under colder temperatures. This lowers the difference between glacial and interglacial temperatures, which is important to account for when quantifying past climate change for the region. Clumped isotope temperatures for soil carbonate are biased to different seasons in different regions and time periods depending on climate In the San Luis Valley, USA, monitoring, modeling, and isotope results suggest carbonate accumulation throughout the year in the Holocene In the glacial late Pleistocene, clumped isotopes and soil modeling indicate longer snow cover shifted carbonate accumulation to the summer

The clumped and stable isotope (Delta(47), delta O-18, and delta C-13) composition of pedogenic (soil) carbonates from cold, arid environments may be a valuable paleoclimate archive for climate change-sensitive areas at high latitudes or elevations. However, previous work suggests that the isotopic composition of cold-climate soil carbonates is susceptible to kinetic isotope effects (KIE). To evaluate the conditions under which KIE occur in cold-climate soil carbonates, we examine the Delta(47), delta O-18, and delta C-13 composition of soil carbonate pendants from Antarctica (Dry Valleys, 77 degrees S), the High Arctic (Svalbard 79 degrees N), the Chilean and Argentinian Andes, and the Tibetan plateau (3800-4800 m), and compare the results to local climate and water delta O-18 records. At each site we calculate the expected equilibrium soil carbonate Delta(47) and delta O-18 values and estimate carbonate Delta(47) and delta O-18 anomalies (observed Delta(47) or delta O-18 minus the expected equilibrium Delta(47) or delta O-18). Additionally, we compare the measured carbonate delta C-13 to the expected range of equilibrium soil carbonate delta C-13 values. To provide context for interpreting the Delta(47) and delta O-18 anomalies, the soil carbonate results are compared to results for sub-glacial carbonates from two different sites, which exhibit large Delta(47) anomalies (up to -0.29 parts per thousand). The Antarctic and 4700 masl Chilean Andes samples have negative Delta(47) anomalies and positive delta O-18 anomalies consistent with KIE due to rapid bicarbonate dehydration during cryogenic carbonate formation. In contrast, the lower elevation Chilean Andes, Argentinian Andes, Tibetan Plateau and High Arctic results are consistent with equilibrium, summer carbonate formation. We attribute the differences in Delta(47) and delta O-18 anomalies to variations in inter-cobble matrix grain size and its effects on the effective soil pore space, permeability (hydraulic conductivity), moisture, and bicarbonate dehydration rate. The Antarctic and 4700 masl Chilean Andean soils have coarse-grained matrices that facilitate rapid bicarbonate dehydration. In contrast, the lower elevation Chilean Andes, Argentinian Andes, High Arctic and Tibetan Plateau soils have finer-grained matrices that decrease the soil pore space, soil permeability and CO2 gas flux, promoting equilibrium carbonate formation. The sub-glacial carbonate samples yield highly variable Delta(47) and delta O-18 anomalies, and we propose that the differences between the two glacier sites may be due to variations in local sub-glacial drainage conditions, pCO(2), and pH. Our findings suggest that carbonates from soils with coarse-grained matrices may exhibit KIE in cold climates, making them poor paleoclimate proxies. Soils with fine-grained matrices are more likely to yield equilibrium carbonates suitable for paleoclimate reconstructions regardless of climate. Paleosol matrix grain size should therefore be taken into account in the evaluation of carbonate stable and clumped isotope values in paleoclimate studies. (C) 2018 Elsevier Ltd. All rights reserved.