Figure 7: Clumped isotope derived estimates of temperature and δ18Owater for a travertine terrace in Ainet, Austria. Solid symbols represent estimates derived from material-specific calibrations, while lighter symbols represent estimates derived from our composite freshwater calibration. Gray bands represent the range of modern stream values measured in May, July, and October (Boch et al., 2005) and data for GISP2 ice core record is from Stuiver and Grootes (2000). Values are broadly consistent with modern temperatures and δ18Owater of nearby streams. Seasonally enhanced percolation of meteoric waters and melting of the last glacial Alpine glaciers during initiation of the pronounced Bølling-Allerød warm period are hypothesized to have recharged groundwater via new flow routes (slope failure/graben and faults) opening during the melting phase and promoting the relatively short lived discharge and calcium carbonate precipitation event.
3.6.3 Application to Paleoaltimetry
Following on stable isotope and leaf margin-based paleoaltimetry (Blisniuk & Stern, 2005; Gregory-Wodzicki, 2000; McElwain, 2004; Rowley & Garzione, 2007; Wolfe et al., 1998), clumped isotope analyses have been used to constrain the tectonic history of a region (Ghosh, Garzione, et al., 2006; Huntington & Lechler, 2015; L. Li et al., 2019; Quade et al., 2013). This proxy relies on the premise that lake water temperature is directly related to air temperature; therefore, as basins undergo surface uplift as a result of large-scale tectonic processes (e.g., crustal shortening and thickening, convective removal of lower lithosphere, etc.), the ambient air and water temperature should decrease, as governed by the local lapse rate (Ghosh, Garzione, et al., 2006; Huntington & Lechler, 2015; L. Li et al., 2019; Quade et al., 2013). Reconstructed δ18Owater can provide additional constraints on paleoelevation, because the stable isotope compositions of meteoric and surface waters decreases as altitude increases (Chamberlain & Poage, 2000; Poage & Chamberlain, 2001; Rowley & Garzione, 2007). To assess the impact of our calibrations on a published paleoelevation reconstruction for the Tibetan Plateau, we recalculate temperatures and δ18Owater estimates from L. Li et al. (2019) using the micrite, composite, and Anderson et al. (2021) calibration for lacustrine samples in the Nangqian Basin. We used our clumped-isotope derived results to compare estimates of elevation changes during the late Eocene following the methodology used in the original publication.
Overall, the micrite calibration equation derives similar temperatures for unaltered carbonates to those derived in the original publication using the Kelson calibration (Supplementary Table 9). Mean temperatures for Units 1, 3, and 4 using the new calibration derived in this study are 25.1 ± 3.0, 39.0 ± 3.7, and 29.0 ± 3.5°C. These values are 0.2°C higher, 1.9°C lower, and 0.7°C lower than published results, respectively. The general calibration estimates slightly cooler values than our material-specific derived results (1.4, 0.2, and 1.1°C lower for Units 1, 3, and 4, respectively). The Anderson calibration projects even cooler values, with Units 1, 3, and 4 estimates being 3.4, 1.2, and 3.4°C lower, respectively.
In the new calibration, samples from Unit 2 yield high temperatures between 44.2 and 67.8°C, which agree with previous observation and interpretation of deep burial diagenetic alteration of the Unit 2 samples (L. Li et al., 2019). We also note that the previous interpretation for the new temperatures of Unit 1 still holds, such that the lower temperatures in the Late Cretaceous represent either less heating of lake water or possibly higher surface elevation in the Nangqian Basin.
Late Eocene (38-37 Ma) carbonates analyzed from Unit 4 were suggested to be deposited near-surface and their Δ47-temperature values were used to reconstruct δ18Owater values and paleoelevation (L. Li et al., 2019). We calculate similar values using our calibration for late Eocene highland water δ18O values to be -9.4‰ and -9.8‰, respectively, compared to -9.1‰ and -9.8‰ in the original study. Following the method in L. Li et al. (2019) where the authors use different modeling approaches to consider dominant regional moisture sources, the paleo-water δ18O values correspond to a mean hypsometric paleoelevation of 2.8±1.1 km and 3.1±1.1 km, which is very similar to the reconstruction that was published (2.8±1.1 km and 3.2±1.1 km), and ~1.3 km lower than the modern hypsometric mean elevation (4.2 km) of the Nangqian Basin. Given the similarities in projected water temperatures between the material-specific and composite regressions for the sample used to reconstruct paleoelevation, it is unsurprising that the projected elevation estimates are identical (Table 4). The cooler temperatures derived using Anderson calibrations result in more depleted water δ18O values (-9.5‰ and -10.0‰) and slightly higher estimates of paleoelevation (2.9±1.1 km and 3.2±1.1 km), but still within error of the original publication and findings in this study.
The basin floor paleoelevation of the Nangqian Basin was estimated from the T-Δ47 values of late Eocene lacustrine carbonates in unit 4. The newly calculated mean T-Δ47 value of these samples is 29.5°C, 0.5°C higher than what was published. Following the method in Li et al. (2019), this mean T-Δ47 value is 12.5°C higher than the estimated warm-season lake surface water temperature (~17oC) for an elevation at 3.8 km. Benthic foraminiferal records indicate that roughly 6°C of the temperature decrease can be attributed to the post-Eocene global cooling (Hansen et al., 2008), while the remaining 6.5°C reflects paleoelevation increase of the basin floor after the Eocene. Following the method in L. Li et al. (2019), this indicates 1.1 ± 0.3 km of post–late Eocene elevation increase, if a lapse rate of –6.1 ± 1.0 °C/km is applied for lake surface-water temperature on the Tibetan Plateau (Huntington et al., 2015). Therefore, the late Eocene elevation of the Nangqian Basin floor is estimated as 2.7 km above sea level, which is 1.1 km lower than the modern elevation of 3.8 km.
4 Conclusions
In order to confidently use proxies to characterize and understand past environments, it is necessary to have a solid understanding of modern systems. Thus, in this work, we present an extensive composite dataset of 135 clumped isotope samples of terrestrial freshwater carbonates from 96 sites, and derive relationships between modern water temperatures and Δ47. These freshwater Δ47-temperature calibrations are well constrained, encompass a variety of types of natural lacustrine, fluvial, and spring carbonates, and span a broad range of temperatures, elevations, and latitudes. As the carbonates presented in this study are taken directly from modern freshwater settings, they are more representative of real-world systems, and may, in some circumstances, be more appropriate for application to reconstruct paleotemperatures, than synthetically-derived Δ47-temperature carbonate calibrations. However, we note that since our approach utilizes in-situ lake water surface temperature data, there is an added uncertainty in the timing of carbonate formation temperature and calcification timeframe for each of our calibration samples.
Our results show a convergence of slopes but differences in the intercepts of the Δ47-temperature relationship between freshwater carbonate groups. Specifically an ANCOVA analysis shows that material specific calibrations based on grouping freshwater carbonates (biogenic, biologically mediated, micrite, and travertine) have statistically indistinguishable slopes between other freshwater carbonate groups and recently published calibration studies, but in some cases, where there is strong evidence for biogenicity, detects differences in intercepts.
Utilizing a single, composite calibration does not always yield the most accurate results, while in most cases, implementing material-specific calibrations reduces the magnitude of residuals (offsets between Δ47-derived temperatures/water δ18O and measured temperature/water δ18O) and RMSE, and thus, can provide more robust estimates of temperature. Water δ18O values derived from utilizing material specific calibrations can recover independently measured water δ18O values accurately, with 39% and 74% of lakes being within 1‰ and 2‰ of measured water δ18O, respectively, reflecting a 7% and 10% improvement relative to our composite freshwater calibration.
Additionally, we explore the application of the new calibration relationships reported in this study by examining three case studies using Δ47 measurements of freshwater carbonates. First, we utilize the biologically-mediated and composite calibrations to explore hydroclimatic changes from the LGM and deglacial at Lake Surprise in the Western US and derive cooler temperatures using material-specific, composite, and Anderson calibrations, but with similar results for evaporation and precipitation rates. Second, we utilize the micrite calibration to estimate the Eocene paleoelevation of the Nangqian Basin within the Tibetan Plateau and find that our material-specific derived elevations are in agreement with the original publication, but elevations derived from the composite and Anderson calibration suggest a higher degree of uplift. We also present a new dataset from a travertine sequence in Austria and show that the origin of the fluid that allowed for carbonate formation was likely derived from groundwater sources due to increased percolation of meteoric water and glacial melt. Overall, this work provides a basis for more accurate reconstructions of terrestrial paleoclimate, paleohydrology, and paleoaltimetry using freshwater archives, and opens the door to more robust understandings of paleoenvironmental processes.