TRANSFORMATION OF RAINFALL δ18O VARIABILITY BY THE CAVE AND KARST ENVIRONMENT: RESULTS FROM LONG-TERM CAVE MONITORING IN CENTRAL TEXAS

Reconstructions of past variability in hydroclimate from speleothem δ¹⁸O records are commonly based on the idea that speleothem δ¹⁸O values directly reflect the δ¹⁸O values of precipitation. Understanding the way in which the cave and karst environment alters the isotopic composition of precipitation, therefore, is pertinent to distinguishing between climate and non-climate related speleothem δ¹⁸O variability. To address this, we present results from a long-term (>10 years) cave-monitoring project in which samples of Austin-area rainfall and Natural Bridge Caverns cave drip water were routinely collected (~4 to 6 weeks) over an interval spanning several cycles of drought and above average wet conditions. This study focuses on samples collected from two drip sites (NBSB and NBWS) that are interpreted to be supplied by conduit flow paths that should be the most sensitive to climatic variability. Drip water δ¹⁸O values at two conduit drip sites exhibit a ranged of 2.6‰ (from -5.2 to -2.6‰), although the difference between 1^st and 3^rd quartiles is only 0.4‰. Drip water δ¹⁸O variability is muted relative to rainfall δ¹⁸O variability, which ranges from -12.6‰ to -2.8‰ with a difference between 1^st and 3^rd quartiles of 2.1‰. The average drip water δ¹⁸O value of both sites (-4.3‰) is similar to the weighted average rainfall δ¹⁸O value (-4.2‰), and the correlation between drip water and rainfall δ¹⁸O values is poor (R² of 0.18 and <0.01 for NBSB and NBWS, respectively). These results suggest that mixing of infiltrating water occurs within the vadose zone above the cave, which drives the homogenization of dripwater δ¹⁸O values. Furthermore, similarity between average dripwater δ¹⁸O values and weighted average rainfall δ¹⁸O values suggests that dripwater δ¹⁸O are not biased toward a particular season or affected by evaporation, as both processes would result in an offset between dripwater and rainfall δ¹⁸O values. Overall, our result suggests that speleothem δ¹⁸O values, in the absence of non-equilibrium fractionation, do not preserve high-resolution (e.g., seasonal, inter-annual) variations in rainfall δ¹⁸O values, and, instead, likely reflect long-term (e.g., multi-decadal, centennial) shifts in average rainfall δ¹⁸O values.