Alexandrea Arnold^1,3*, John Mering⁴, Lauren Santi^2-4, Cristian Román-Palacios³, Huashu Li⁵, Victoria Petryshyn⁶, Bryce Mitsunaga⁴, Ben Elliott⁴, John Wilson⁴, Jamie Lucarelli^3-4, Ronny Boch⁷, Daniel Ibarra⁸, Lin Li¹¹, Majie Fan⁹, Darrell Kaufman¹⁰, Andrew Cohen¹¹, Rob Dunbar¹², James Russell⁸, Stefan Lalonde¹³, Priyadarsi D. Roy¹⁴, Martin Dietzel⁷, Xingqi Liu⁵, Fengming Chang¹⁵, Robert A. Eagle^1,3,13,16, and Aradhna Tripati^1-4,13,16

¹Department of Atmospheric and Oceanic Sciences, University of California, Los Angeles, California, USA, Math Science Building, 520 Portola Plaza, Los Angeles, CA 90024, USA

²Institute of the Environment and Sustainability, University of California, Los Angeles, LaKretz Hall, 619 Charles E Young Dr E #300, Los Angeles, CA 90024, USA

³Center for Diverse Leadership in Science, University of California, Los Angeles, California, USA, 595 Charles E Young Dr E, Los Angeles, CA 90024, USA

⁴Department of Earth, Planetary, and Space Sciences, University of California, Los Angeles, 595 Charles E Young Dr E, Los Angeles, CA 90024, USA

⁵College of Resource Environment & Tourism, Capital Normal University, Beijing 100048, China

⁶Environmental Studies Program, University of Southern California, Los Angeles, CA, 3454 Trousdale Pkwy CAS 106, Los Angeles, CA 90089-0740

⁷Institut für Angewandte Geowissenschaften, Technische Universität Graz, Rechbauerstraße 12, 8010 Graz, Austria

⁸Earth, Environmental, and Planetary Sciences, Brown University, 324 Brook Street, Providence, RI, 02906

⁹Department of Earth and Environmental Sciences, University of Texas at Arlington, Arlington, Texas 76019, USA

¹⁰School of Earth and Sustainability, Northern Arizona University, S San Francisco St, Flagstaff, AZ 86011

¹¹Department of Geosciences, University of Arizona, 1040 E. 4th Street Tucson, AZ 85721

¹²Department of Earth System Science, Stanford University, 473 Via Ortega, Building Y2E2, Stanford, CA 94305, USA

¹³University of Brest, Brest, France, UMR 6538 Laboratoire Géosciences Océan, Institut Universitaire Européen de la Mer, Technopôle Brest-Iroise, Place Nicolas Copernic, Plouzané 29280, Brest, France

¹⁴Instituto de Geología, Universidad Nacional Autónoma de México (UNAM), Ciudad Universitaria, 04510 Ciudad de México, México

¹⁵Key Laboratory of Marine Geology and Environment, Institute of Oceanology, Chinese Academy of Sciences, Qingdao 266071, China

¹⁶School of Earth Sciences, School of Geographical Sciences, University of Bristol, Bristol, BS8 1QU, United Kingdom

Corresponding author: Alexandrea Arnold (ajarnold1@g.ucla.edu) and Aradhna Tripati (atripati@g.ucla.edu)

Key Points:

Abstract

Lacustrine, riverine, and spring carbonates are archives of terrestrial climate change and are extensively used to study paleoenvironments. Clumped isotope thermometry has been applied to freshwater carbonates to reconstruct temperatures, however, limited work has been done to evaluate comparative relationships between clumped isotopes and temperature in different types of modern freshwater carbonates. Therefore, in this study, we assemble an extensive calibration dataset with 135 samples of modern lacustrine, fluvial, and spring carbonates from 96 sites and constrain the relationship between independent observations of water temperature and the clumped isotopic composition of carbonates (denoted by Δ₄₇). We restandardize and synthesize published data and report 159 new measurements of 25 samples. We derive a composite freshwater calibration and also evaluate differences in the Δ₄₇-temperature dependence for different types of materials to examine whether material-specific calibrations may be justified. When material type is considered, there is a convergence of slopes between biological carbonates (freshwater gastropods and bivalves), micrite, biologically-mediated carbonates (microbialites and tufas), travertines, and other recently published syntheses, but statistically significant differences in intercepts between some materials, possibly due to seasonal biases, kinetic isotope effects, and/or varying degrees of biological influence. Δ₄₇-based reconstructions of water δ¹⁸O generally yield values within 2‰ of measured water δ¹⁸O when using a material-specific calibration. We explore the implications of applying these new calibrations in reconstructing temperature in three case studies.

1 Introduction

Paleoenvironmental reconstructions from freshwater sediments can be used to enhance our understanding of ecosystem and climate change both within the aquatic systems and their proximal terrestrial environments (e.g., Brenner et al., 1999; Xu et al., 2006)). Carbonate-bearing sediments deposited in freshwater systems are widespread and are sensitive to changes in the local environment, tectonic setting, and hydrological conditions, and thus provide a promising archive of continental paleoclimatic information (Arenas-Abad et al., 2010; Gierlowski-Kordesch, 2010; Hren & Sheldon, 2012). However, quantitative terrestrial temperature proxies that can be applied to carbonate sediments are relatively scarce. Of the multiple proxies that have been used to estimate terrestrial temperatures with varying degrees of uncertainty, including soil carbonates, speleothems, fracture veins, ostracods, trace element ratios in lacustrine sediments, tree rings, leaf margin analysis, pollen, biomarkers, and noble gasses in groundwater (Boch et al., 2019; Esper et al., 2018; Gallagher & Sheldon, 2013; Kaufman et al., 2020; Meckler et al., 2021; Powers et al., 2010; Stute & Schlosser, 2000; Wilf, 1997; Wrozyna et al., 2022), only the first five types of proxies are carbonate-associated.

However, work has shown that multiply-substituted carbonate “clumped” isotope thermometry presents a promising proxy for reconstructing temperature, based on the thermodynamic exchange of isotopes between isotopologues of carbonate-containing groups (Ghosh, Adkins, et al., 2006; Hill et al., 2014; Schauble et al., 2006; Tripati et al., 2015) with widespread applicability in paleoclimatic, paleohydrological, and paleoelevation contexts (e.g. Csank et al., 2011; Eagle et al., 2013; Hren et al., 2013; Huntington et al., 2010, 2015; Huntington & Lechler, 2015; Santi et al., 2020; Tripati et al., 2010, 2014). Theoretical calculations indicate that measurements of clumped isotopes can be used for paleothermometery because at equilibrium, the abundance of the multiply-substituted isotopologue¹³C¹⁸O¹⁶O₂in carbonates is related solely to the formation temperature of the mineral (Ghosh, Adkins, et al., 2006; Hill et al., 2014; Schauble et al., 2006; Tripati et al., 2015), with cooler temperatures favoring enhanced “clumping” of heavy isotopes within the mineral lattice (e.g., the forward reaction in Equation 1).

Ca¹²C¹⁸O¹⁶O₂+ Ca¹³C¹⁶O₃ → Ca¹³C¹⁸O¹⁶O₂+ Ca¹²C¹⁶O₃ (1)

We report the abundance of mass-47 CO₂ liberated from carbonate minerals digested in phosphoric acid (δ₄₇) compared to a randomized (stochastic) distribution of clumping in a sample. This excess of¹³C¹⁸O¹⁶O is denoted as Δ₄₇ in Equation 2, where Rⁱ= (mass i/mass 44):

Δ₄₇ (‰) = [(R⁴⁷/(R⁴⁷_stochastic) - 1) - (R⁴⁶/(R⁴⁶_stochastic) - 1) - (R⁴⁵/(R⁴⁵_stochastic) - 1)] × 1000 (2)

An advantage of carbonate clumped isotope derived temperature estimates is that they are independent of the¹⁸O/¹⁶O ratio (δ¹⁸O) of the precipitating fluid, as the relevant isotope exchange reaction (Equation 1) takes place within a single phase. Carbonate δ¹⁸O ratios are simultaneously measured during clumped isotope analysis and can be combined with temperature estimates obtained from Δ₄₇ analysis to calculate δ¹⁸O values of water at the time of carbonate formation (Epstein et al., 1953; Urey, 1947; Vasconcelos et al., 2005).

Clumped isotopes have been previously used to constrain past lake and river water temperature (Cheng et al., 2022; Horton et al., 2016; Hudson et al., 2017; Huntington et al., 2010, 2015; Kele et al., 2015; H. Li et al., 2021; Petryshyn et al., 2015; L. M. Santi et al., 2020; Wang et al., 2021). The additional temperature constraint provided from Δ₄₇ measurements allows for a calculation of the δ¹⁸O of meteoric waters, which can provide constraints on past hydrology. The Δ₄₇-temperature and Δ₄₇-derived water δ¹⁸O in freshwater carbonates and other types of terrestrial archives have in turn been used to evaluate process depiction in climate models (Cheng et al., 2022; Eagle, Risi, et al., 2013; L. M. Santi et al., 2020; Tripati et al., 2014), constrain hydrologic parameters (L. M. Santi et al., 2020), and to constrain paleoaltimetry (e.g., Ghosh, Garzione, et al., 2006; Huntington et al., 2010, 2015; Ingalls et al., 2017; L. Li et al., 2019).

However, the accuracy of these reconstructions is fundamentally underpinned by the calibration(s) used for calculations. The body of literature for clumped isotope measurements of modern lacustrine and riverine samples is limited. Only four studies have reported Δ₄₇-T calibrations (Anderson et al., 2021; Kato et al., 2019; Kele et al., 2015; H. Li et al., 2021), with three calibrations being solely derived from freshwater sediments (H. Li et al., 2021: n=33, Kele et al. 2015: n = 24, Kato et al. 2019: n=33). Most clumped isotope studies of freshwater carbonates have analyzed a small number of samples (Anderson et al., 2021; Grauel et al., 2016; Horton et al., 2016; Hudson et al., 2017; Huntington et al., 2010, 2015; Kato et al., 2019; Kele et al., 2015; H. Li et al., 2021; Petryshyn et al., 2015; L. M. Santi et al., 2020; Wang et al., 2021). Of these 11 studies reporting data for modern freshwater carbonates, five report new data for <5 samples, while eight have data for <11 samples, and the remaining four have data consisting of 25-33 samples. The smaller size of these prior datasets on clumped isotope compositions means that the community has not been able to explore in detail the potential for material-specific calibrations, possible influences of seasonal and temperature bias in carbonate formation, and non-equilibrium kinetic or pH dependent effects which could differentially affect different types of carbonates.

Recent work to constrain the Δ₄₇-temperature relationship has suggested that all carbonates have the same temperature dependence (i.e., calibration slope), but the calibration intercept may differ due to mineral-specific phosphoric acid fractionation factors (Müller et al., 2019; van Dijk et al., 2019). Recent efforts have advanced practices and improved interlaboratory consistency (Anderson et al., 2021; Bernasconi et al., 2021; Daëron et al., 2016; Petersen et al., 2019; Upadhyay et al., 2021), and use of consistent approaches for standardization and isotope ratio calculations have reduced many offsets (e.g., Petersen et al., 2019). Petersen et al. (2019) reprocessed 14 calibration studies, encompassing synthetic and natural carbonates, using updated parameter values for clumped isotope calculations and found that use of the IUPAC parameters with identical data processing resulted in increased agreement between calibration lines and a convergence of slopes within many of the individual studies. From this synthesis, the authors proposed a ‘universal’, in which one calibration is derived and applied for all carbonate types. However, it has been acknowledged that it is uncertain if only one relationship exists, since despite improvements in data processing and standardization procedures, differences are still observed (Petersen et al., 2019). Recently, the “InterCarb” project defined new, community-based standard values for carbonate clumped isotope standardization, as well as proposed a new reference frame (I-CDES), which has been shown to help resolve inter-laboratory differences (Bernasconi et al., 2021). The application of carbonate standardization and newly defined carbonate standard values, in concert with developments in data handling procedures (Daëron, 2021), to reprocessed data from four older calibration studies has been shown to help resolve the disagreement between their derived calibration lines (Anderson et al., 2021).

A few studies have suggested statistically significant differences between calibrations using different types of carbonates (Davies & John, 2019; Eagle, Eiler, et al., 2013; Henkes et al., 2013; Kele et al., 2015; Kimball et al., 2016; Müller et al., 2019). Factors such as seasonality of carbonate growth, the ecology of shell forming organisms, and kinetic isotope effects relating to different processes in mineral formation have been shown to broadly influence empirical relationships between Δ₄₇ measurements and environmental parameters, including in corals (eg. Ghosh, Adkins, et al., 2006; Kimball et al., 2016; Saenger et al., 2012; Spooner et al., 2016), echinoids (Davies & John, 2019), terrestrial gastropods (eg. Dong et al., 2021; Eagle, Eiler, et al., 2013; Zaarur et al., 2013), as well as synthetic carbonates (eg. Tang et al., 2014; Tripati et al., 2015), with studies of slow growing cave carbonates indicating that many if not most carbonates could express a degree of kinetic isotope effect (Daëron et al., 2019; Kluge et al., 2014; Tripati et al., 2015).

In this work, we discuss clumped isotope data from 135 samples collected from 96 sites in modern lakes, rivers, and springs. At present, due to limited freshwater calibration data, it is unresolved whether the differences in calibration relationships between different types of freshwater carbonate materials exist, and whether synthetically-derived regression parameters are appropriate to use in freshwater samples that are field-collected. The recently published syntheses from Petersen et al. (2019) had no freshwater carbonates, and Anderson et al. (2021) had 16 carbonates (tufas and travertines) of which 7 were between T < 10 ^oC, and 6 were T > 30^oC. This study augments the literature with 159 new clumped isotope measurements for 25 sites, and additionally incorporates measurements from published datasets, including samples from 59 sites that have been recalculated on the new reference frame (Bernasconi et al., 2021). It includes pairs of measurements from 12 sites that have been recently reanalyzed (Anderson et al., 2021) and are recalculated on the new reference frame.

This dataset allows us to investigate clumped isotope signatures in travertines, micrites, biotic (freshwater mollusks), and biologically mediated (tufas and microbialites) phases of freshwater carbonate in order to provide a foundation for intercomparison and calibration of carbonate clumped isotope results from freshwater systems. Sample localities within this study are geographically diverse, and include equatorial, mid-latitude, and polar sites at a variety of elevations and climates. We evaluate the seasonality of carbonate formation, and present a composite freshwater calibration and material-specific calibrations. We assess the potential of this proxy to robustly reconstruct water temperature and source water δ¹⁸O in both modern and past contexts.

2 Materials and Methods

2.1 Sample and site selection

Samples included in this study are from modern lakes (including playas), rivers, and springs, from geographically and climatically diverse settings (Figure 1; Table 1). Carbonate materials included in this dataset were selected to represent modern lacustrine surface-water conditions. For lakes, biogenic and biologically-mediated samples selected for analysis are types that grow nearshore, or occupy the photic zone above the thermocline; we also selected endogenic carbonate because formation occurs in surface waters, where evaporation and photosynthesis have the strongest effect on water chemistry (Gierlowski-Kordesch, 2010; Hren & Sheldon, 2012; Platt & Wright, 2009).

For this work, we only included localities where modern surface water temperatures have been measured (Supplementary Table 1). For sites where multiple years of hydrographic data are available, we report temperature variability of one standard deviation of the monthly average temperatures during the typical season of carbonate formation for each type of carbonate (see Section 3.1). In sites where the average modern water temperature during the interval of carbonate growth was limited to data from a short time interval (i.e., less than one year), given that there is year-to-year variability in temperatures, we report a temperature uncertainty of two standard deviations of the available measurements. Some sites within our study are less studied, and if there was limited temperature information for the water body, such as a ‘summer’ average, or the site has a measured temperature value but uncertainty was not reported, we use the average standard deviation for well constrained sample sites in our study (± 2°C).