Table 4: Comparison of recalculated Δ47-based proxy reconstructions from this study to published values. Top: Comparison of water temperature, water δ18O, precipitation, and weighted evaporation anomalies at Lake Surprise, CA to published values from Santi et al. (2020). Modern MAAT is 9.2 ± 1.0°C from Cedarville, CA and modern δ18O used for anomaly calculations is -14.6 ± 0.6‰. Modern precipitation used to calculate the anomaly is 630 mm/year, including contributions from rainfall and snowmelt, and modern evapotranspiration rates are 528 mm/year. Bottom: Comparison of clumped isotope derived water temperature, water δ18O, and elevation estimates for samples run for clumped isotope analysis from the Nangquian Basin to published values from L. Li et al. (2019).
3.6.2. Origin of travertine and tufa deposits in Ainet, Austria
We apply our calibrations to study the formation of travertines and tufas in a mountainous region of the Alps. To demonstrate this, we examined a lateglacial sequence of travertine and calcareous tufa from Ainet, Austria that was first described in Boch et al. (2005). It represents the only aragonitic travertine sequence known in the Eastern Alps. This sequence of ~2.7 m thickness formed over the course of ~1,000 years, following the rapid initiation of warming during the Bølling-Allerød (Figure 7). Carbonate deposition within the travertine sequence alternated between aragonite and calcite layers (on mm-scale), which was hypothesized to represent differences in seasonality, with aragonite precipitation occurring during the warm season and calcite precipitation occurring during cooler intervals. Following the deposition of the compact and aragonite dominated travertine (~2.5 m), this sequence was then capped by a highly porous and calcareous tufa layer (~0.2 m) consisting of calcite solely.
Here, we constrain formation temperature and δ18Owater for these differentiated calcium carbonates using our biologically-mediated and travertine calibrations from this study. Temperature estimates for the tufa and travertine sequence range from 9.0 to 15.1°C, with an average value of 12.7°C for the travertine terrace (Supplementary Table 8). Modern water temperatures taken in May, July, and October range from 6.6 - 12.2°C, similar to our Δ47-derived estimates. Although the initial study suggested a seasonal control on aragonite and calcite formation, with aragonite being precipitated in the warmer months and calcite being precipitated in the cooler months, we do not resolve a clear relationship between temperature and mineralogy within this limited dataset (Figure 7).
Our Δ47-temperatures and δ18Owater values support the hypothesis that the travertine sequence did not have a hydrothermal origin (thermal water discharge), but are consistent with being derived from rapid CO2 degassing from groundwater discharge of meteoric origin, with sufficient time for dissolved inorganic carbon equilibration to occur. We observe consistency between modern δ18Owater values measured from a series of nearby streams (-11.5 to -12.1‰ VSMOW) and δ18Owater estimates derived from clumped isotope analysis (-11.4 to -12.2‰ VSMOW). Given this consistency in δ18Owater, we suggest recharged meteoric groundwater (seasonal rainfall, snowmelt) and eventually some contribution from ice melting due to a rapid increase in temperatures during the Bølling-Allerød is likely to have been the surface dominated paleofluid source for carbonate precipitation here. The calcium carbonate source of the freshwater carbonate precipitates is probably manifested in local marble lenses within the prevalent metamorphic rocks and additional CO2 from underground might have been dissolved in the percolating groundwaters favored by deep-seated slope displacements (as a consequence of glacial ice melting) and faults that have been locally detected (Boch et al., 2005).
The analysis of this travertine sequence illustrates the importance of an appropriate calibration selection. While application of our composite freshwater carbonate calibration would yield temperature values 2.5 - 2.7°C higher than modern streamflow, use of the material-specific travertine calibration yields formation temperatures more similar to modern stream temperatures (Supplementary Table 8), which are more likely to be correct for carbonates forming in an interval of distinct relative warmth in the last glacial period. The δ18Owater values reconstructed from the material-specific calibration are within error of modern groundwater values, measured from spring-fed streams, while reconstructed δ18Owater values are 0.5 to 0.6‰ higher when using the composite calibration relative to the material-specific calibration. We note the Anderson et al. (2021) calibration estimates the formation temperature of the calcareous tufa sample to be 9.1°C, 1.6°C colder than the temperatures calculated using the biologically mediated calibration. The cooler temperature estimated by Anderson et al. (2021) results in δ18Owater that are 0.4‰ more depleted relative to the biologically mediated calibration. Reconstructed temperatures and δ18Owater values for the travertine sequence using the Anderson et al. (2021) calibration result in temperature values that are similar to our travertine calibration (0 - 0.2°C higher) and identical δ18Owater values. Given that the travertine calibration is indistinguishable from the Anderson et al. (2021) calibration according to an ANCOVA test (slope: p = 0.6320; intercept: p = 0.4440), the similarities in temperatures and δ18Owater values between the two calibrations is unsurprising.