Tracking rainfall variations in the late Pleistocene using U isotopes in dated secondary soil minerals

Past changes in rainfall are an important indicator of variations in atmospheric circulation. However, there are very few approaches that are uniquely sensitive to past changes in rainfall and use common materials that can be accurately dated. Secondary minerals in arid soils form directly from rainfall and can precipitate continuously over hundreds of thousands of years. Previous studies have used stable isotope measurements of soil carbonates to look at a variety of processes such as precipitation source and amount, evaporation, temperature, and vegetation. Often, the challenge with stable isotope approaches is to distinguish between these factors. In addition, stable isotope studies have not traditionally used high-resolution analytical techniques to capture the temporal variations within a given sample. The development of alternative isotopic systems that more directly reflect rainfall would provide a complimentary tool to more traditional strategies. In order to evaluate whether widely observed variations in the calculated initial U isotopic composition of dated soil minerals (e.g. (²³⁴U/²³⁸U)₀) reflect changes in past rainfall, we sampled modern soil pore waters, soils and dust from a rainfall gradient in Fish Lake Valley, NV. In situ ion microprobe (SHRIMP-RG) techniques were used to determine the ²³⁰Th-U ages and (²³⁴U/²³⁸U)₀ from uranium-rich soil opal collected at three field sites in western North America. Modern pore waters in desert soils from Fish Lake Valley, NV show systematic decreases in (²³⁴U/²³⁸U) with increasing rainfall. This is attributed to increasing infiltration flux and chemical weathering at higher rainfall rates, as the eolian influence should likely remain the same across the rainfall gradient. Although changes in the uranium concentration and isotopic composition of dust through time may also influence the (²³⁴U/²³⁸U) of soil water, these changes are likely minimal relative to changes in infiltration flux. The variations in (²³⁴U/²³⁸U)₀ obtained from ²³⁰Th-U dating of soil opal span approximately the last 5-60 kyrs. Based on the modern pore water data, we interpret low (²³⁴U/²³⁸U)₀ to indicate periods of higher rainfall, and high (²³⁴U/²³⁸U)₀ to indicate periods of reduced rainfall. We observe nearly synchronous shifts during the last glacial-interglacial transition along the latitudinal transect of our field sites. Our initial results show that the transition from MIS 3 to 2 was likely characterized by increasing precipitation, with peak rainfall at the beginning of MIS 2. This corresponds to generally decreasing SST off of the California coast and rainier periods in speleothem records from the southwestern United States. Additionally (²³⁴U/²³⁸U)₀ from our soil opal records co-varies with fluctuations in lake levels of large pluvial lake systems in western North America. The synchronous signal of increased rainfall during MIS 2 is consistent with a southward-shift in westerly storm tracks at the LGM. Our approach demonstrates an alternative method for increasing the spatial coverage and chronology of climate records in arid regions.