Ma/Ca Ratio in Stalagmites as a Potential Palaeo‐Temperature Indicator

Mg/Ca ratio in stalagmites primarily serves as an indicator of past hydrological conditions rather than temperature variations. Here, we present an Mg/Ca ratio record in a stalagmite from northern China, spanning ice age termination II. A unique finding is that Mg/Ca ratio shows a moderately strong positive correlation with cave air temperature obtained from clumped isotopes data in the same stalagmite. The stable hydrogeochemistry of source water and high temperature sensitivity of Mg partition coefficient plausibly explain this temperature dependence. Furthermore, we establish an Mg/Ca‐temperature calibration for stalagmites from mid‐latitudes, encompassing a temperature range between ∼2 and ∼15°C, where Mg/Ca = 14.221 × e0.073×T (R2 = 0.40). Applying this function to Mg/Ca ratio from similar stalagmites could tentatively provide a quantitative regional temperature record over millennial and centennial timescales. This research highlights Mg/Ca ratio in stalagmites as a potential palaeo‐temperature proxy and introduces, for the first time, a tentative Mg/Ca‐temperature calibration function for speleothems.


Introduction
Palaeo-temperature reconstructions provide valuable insights into Earth's past climate conditions, aiding in understanding of current global warming.However, compared to ocean sediments and ice cores (Jouzel et al., 2007;Martrat et al., 2007), temperature proxies in terrestrial archives are relatively scarce.Common continental temperature records come from loess (Lu et al., 2019;Thomas et al., 2017), peat (Mauquoy et al., 2004) and lake sediments (Heiri et al., 2011).Stalagmites are widely regarded as one of the most reliable sources for understanding past climate changes, because they can be precisely dated by radiometric methods up to 640,000 years ago (Cheng et al., 2016).The isotopic and noble gas compositions of fluid inclusions, clumped isotopes (Δ 47 ) as well as biomarkers in speleothems have been explored to derive temperature-related information (Affek et al., 2008;Duan, Wang, et al., 2022;Ghadiri et al., 2018;Honiat et al., 2023;Kluge et al., 2008;Løland et al., 2022;Wassenburg et al., 2021).However, these temperature proxies, especially fluid inclusions and biomarkers, require a large amount of subsample for analysis, which limits high-resolved temperature estimations using speleothems.
Mg/Ca ratio in carbonates is one of the most regularly used trace element proxies for palaeoclimate reconstructions.As Mg/Ca ratio of seawater is commonly constant over timescales shorter than 1,000 ka, Mg/Ca ratio in marine biogenic carbonates, such as foraminifera and corals, has been widely used for estimating past seawater temperature (Barker et al., 2005 Supporting Information may be found in the online version of this article.Müller, 2012;Mitsuguchi et al., 1996).However, Mg/Ca ratio in speleothems primarily serves as an indicator of past hydrological conditions (Cruz et al., 2007;Duan et al., 2012;Fairchild & Treble, 2009;Fairchild et al., 2000;Huang & Fairchild, 2001;Huang et al., 2001;Sinclair et al., 2012;Treble et al., 2003;Tremaine & Froelich, 2013;Wolf et al., 2023).That is because the significant variations in drip-water Mg/Ca ratio typically surpass the temperature-dependence of Mg partitioning into speleothems (Fairchild & Treble, 2009).In this context, temperature is deemed capable of solely explaining seasonal fluctuations in Mg/Ca ratio (Carlson et al., 2018;Fairchild et al., 2000;Roberts et al., 1998).Nevertheless, a limited number of studies have indicated that Mg/Ca ratio in speleothems can also serve as a plausible palaeo-temperature indicator.For instance, Mg/Ca ratios in a stalagmite from central China show lower values during the relatively cold marine isotope stage (MIS) 5d and higher values during relatively warm MIS 5c (Zhou et al., 2011).However, this inference lacks independent verification from a coeval temperature record.A recent study revealed that Mg concentration in a subaqueous speleothem from Italy aligns with the regional sea surface temperature over the past 350,000 years (Drysdale et al., 2020).The authors emphasized that the subaqueous speleothem formed in a cave pool characterized by a limited range of Mg/Ca ratios in its source water, distinct from the stalagmites.Hence, in the case of subaqueous speleothems, the influence of cave air temperature on the partitioning of Mg into calcites outweighs the impact of pool-water Mg/Ca ratio (Drysdale et al., 2019(Drysdale et al., , 2020)).
In this paper, we present the records of Mg/Ca and Sr/Ca ratios and δ 13 C from a stalagmite in Xinglong cave, northern China, covering the most portion of ice age termination II (T-II).The records of high-resolution δ 18 O and Δ 47 -cave air temperature have been reported previously (Duan et al., 2019;Duan, Wang, et al., 2022).We found a moderately strong positive correlation between Mg/Ca ratio and Δ 47 -cave air temperature, and discussed the potential mechanisms.We also tentatively derived an Mg/Ca-temperature calibration for speleothems.
Stalagmite XL-4 was found broken in the middle part of a collapse slope in the gallery.It grew from 133.4 ± 0.3 to 126.6 ± 0.3 ka BP.It exhibits a standard cylindrical shape with a length of 170 mm and a diameter of 100 mm (Figure S1b in Supporting Information S1) (Duan et al., 2019;Duan, Wang, et al., 2022).

Methods
The methods for 230 Th dating, stable isotopes and clumped isotopes analysis have been extensively elucidated in previous studies (Duan et al., 2019;Duan, Wang, et al., 2022) and will not be reiterated in this study.

Elements Concentration Measurements for Stalagmite
A ∼2 mm-thick and ∼1.5 cm-wide slab was extracted from the central portion of XL-4, along its growth axis.Two transects, spaced 5 mm apart and aligned parallel to the growth axis, were employed for elements (Ca, Mg, Sr) concentration measurements.The analysis was performed by using an ELEMENT XR sector field inductively coupled plasma mass spectrometry (Thermo Scientific, USA) in conjunction with an Analyte G2 193 nm ArF excimer laser ablation system.ARM-3 (glass) was used for external calibration (Wu et al., 2019(Wu et al., , 2021) ) and MACS-3 was used for data quality control.This methodology follows a similar approach as outlined by Wu et al. (2018) (Text S1 in Supporting Information S1).This work was performed at the State Key Laboratory of Lithospheric Evolution, Institute of Geology and Geophysics, Chinese Academy of Sciences.

Field Work
Daily air temperature and precipitation outside Xinglong cave were monitored from October 2017 through November 2019, and again August 2022 through December 2023.Drip water was collected by placing 500-mL high-density polyethylene (HDPE) bottles at drip sties for ∼18 hr.A total of 39 drip water samples were intermittently collected at eight sites between October 2017 and December 2023 (Figure S2 in Supporting Information S1).Aliquots for elements analysis were acidified with super-pure nitric acid (HNO 3 ) on-site and stored in Milli-Q water-rinsed 125-mL HDPE bottles.Systematic monthly monitoring has been performed since July 2023 at drip sites XL-1, XL-2, XL-3, and XL-5, involving measurements of drip rates, cave air temperature, and collections of drip water and modern calcite (Figure S3 in Supporting Information S1).Drip rate snapshots were estimated by counting drops over 1 min and reported as drips/minute.Cave air temperature was measured using a handheld temperature probe.Modern calcite was collected by placing frosted and cleaned 5 × 5 cm glass plates under the four drip sites.The plates were removed and replaced during drip water collection.Only calcite deposited at sites XL-2 and XL-3 between 17 May and 13 July 2023 yielded sufficient quantities for elements concentration measurements.

Elements Concentration Measurements for Drip Water and Modern Calcite
All calcite deposits were scraped from the plates.Calcite powders weighing 17.1 mg from XL-2 and 8.0 mg from XL-3 were dissolved in 14 mol/L nitric acid, and then diluted to 15.0 g for XL-2 and 13.3 g for XL-3 with 2% nitric acid.Prior to elements analysis, each drip water and modern calcite solution was filtered through a 0.45micron polyethersulfone membrane filter.An iCAP PRO inductively coupled plasma emission spectrometer from Thermo Fisher Scientific was used to determine elements (Ca, Mg, Sr) of the samples with a relative standard deviation of less than 2%.This work was performed at Key Laboratory of Cenozoic Geology and Environment, Institute of Geology and Geophysics, Chinese Academy of Sciences.

Timeseries of Mg/Ca and Sr/Ca Ratios in XL-4
Mg/Ca and Sr/Ca ratios versus depth for the two parallel transects have a good replicability (Figure S4 in Supporting Information S1).Transect 1 was used to produce the timeseries employing the age-depth model established in previous study.This model incorporates 15 230 Th dates with errors (2σ) ranging from 222 to 330 years (Figure S5 in Supporting Information S1) (Duan et al., 2019).Mg/Ca and Sr/Ca ratios records consist of ∼3,800 independent points, with an average resolution of ∼2 years.Mg/Ca and Sr/Ca ratios range from 19.6 to 52.4 mmol/mol and 0.09 to 0.36 mmol/mol, respectively.Locally estimated scatterplot smoothing (LOESS) (Cleveland, 1979) regression with a 100-year-smoothing window was applied to discern the main patterns.The "smoothed" timeseries are used throughout this work.There is no significant correlation between Mg/Ca and Sr/ Ca ratios (Figure S6 in Supporting Information S1).One of the most outstanding characteristics of Mg/Ca ratios is the step-wise increment from 132.5 to 127.8 ka BP, characterized by a sudden surge of 14 mmol/mol occurring between 129.3 and 129.1 ka BP.In contrast, Sr/Ca ratios exhibit a progressive rise from 133.4 to 131.5 ka BP, maintaining relatively high values until 128.4 ka BP, followed by a subsequent decrease toward 126.6 ka BP (Figure S6a in Supporting Information S1).Δ 47 -cave air temperature record of XL-4 reveals a stepwise increase from 133.4 to 127.4 ka BP, with amplitude of 8.6 ± 2°C (Figure 1a) (Duan, Wang, et al., 2022).The substantial temperature change derived from the same stalagmite facilitates a direct evaluation of the reliability of Mg/Ca ratio as a temperature indicator.A moderately strong positive correlation exists between Mg/Ca ratio and Δ 47 -cave air temperature records (r = 0.63, p < 0.01) (Figure 1).Generally, Mg/Ca ratio aligns with Δ 47 -cave air temperature, both exhibiting a long-term increasing trend from T-II to Last Interglacial (LI) (Figure 1a).Notably, Mg/Ca ratio accurately captures the abrupt rise in cave air temperature during the transition into LI, occurring between ∼129.3 and ∼129.1 ka BP (Figures 1a and  2a).Nevertheless, there are some mismatches between them (Figures 1a and 2a).For instance, a significant positive excursion in Mg/Ca ratio is observed at beginning of the record from 133.4 to 132.5 ka BP, in contrast to the contemporaneous steady rise in cave air temperature.Additionally, another substantial positive excursion in Mg/Ca ratio occurred between 128.3 and 127.6 ka BP, diverging from the concurrent high variability of cave air temperature.This discrepancy results in misalignment between the "peaks" of the two records.

Discussion
In short, this study presents the first instance of Mg/Ca ratio in a speleothem demonstrating a positive correlation with temperature within the same chronological framework, notwithstanding certain disparities between them.In contrast, Mg/Ca ratio in speleothems is regularly interpreted as a proxy for past hydrologic conditions rather than temperature changes.This is primarily because the substantial variability in drip-water Mg/Ca ratio tends to overshadow the influence of temperature on Mg partitioning (Cruz et al., 2007;Duan et al., 2012;Fairchild & Treble, 2009;Fairchild et al., 2000;Huang & Fairchild, 2001;Huang et al., 2001;Sinclair et al., 2012;Treble et al., 2003;Tremaine & Froelich, 2013;Wolf et al., 2023).The qualitative relationship between Mg/Ca ratio and hydrologic conditions can be summarized as follows: When recharge diminishes, extended water residence time can enhance prior calcite precipitation (PCP) within the epikarst and/or on the cave ceiling, leading to higher Mg/ Ca ratio in both the solution and speleothems (Cruz et al., 2007;Duan et al., 2012;Fairchild & Treble, 2009;Fairchild et al., 2000;Huang & Fairchild, 2001;Huang et al., 2001;Treble et al., 2003;Tremaine & Froelich, 2013;Wolf et al., 2023).Similarly, enhanced PCP can also increase Sr/Ca ratio, establishing a positive correlation between Sr/Ca and Mg/Ca ratios (Cruz et al., 2007;Fairchild & Treble, 2009;Fairchild et al., 2000;Huang & Fairchild, 2001;Huang et al., 2001;Sinclair et al., 2012;Treble et al., 2003;Tremaine & Froelich, 2013).Moreover, prolonged water residence time in the karst aquifer intensifies water-rock interaction, potentially increasing Mg concentration in the solution, particularly in Mg-rich dolomitic bedrock due to incongruent dissolution of dolomite (IDD) (Fairchild et al., 2000;Roberts et al., 1998).IDD elevates Mg/Ca ratio but decreases Sr/Ca ratio, as dolomite typically contains lower Sr content compared to calcite.Consequently, IDD results in a negative correlation between Mg/Ca and Sr/Ca ratios in speleothems (Fairchild et al., 2000;Roberts et al., 1998).Therefore, the absence of co-variation between Mg/Ca and Sr/Ca ratios in XL-4 (Figure S6 in Supporting Information S1) implies that PCP and/or IDD are not the primary drivers of Mg/Ca ratio variability.Additionally, the positive co-variation of Mg/Ca ratio and δ 13 C typically observed in most stalagmites is attributed to PCP (Arienzo et al., 2017;Fairchild & Treble, 2009;Johnson et al., 2006).In contrast, a weak inverse correlation is noted between XL-4 Mg/Ca ratio and δ 13 C (r = 0.40, p < 0.01) (Figure 2b).Specifically, within the interval from 133.4 to 128.4 ka BP, a moderately strong negative correlation emerges (r = 0.60, p < 0.01).Therefore, we further propose PCP is not a major driver of Mg/Ca ratio variability.Nevertheless, the positive covariations observed between Mg/Ca ratio and δ 13 C from 128.4 to 126.6 ka BP suggest a notable influence of PCP.(Scholz & Hoffmann, 2011), following the previous study (Duan et al., 2019).Dating errors (2σ) are indicated at the bottom.(a) Timeseries of Mg/Ca ratio (top) and Δ 47 -cave air temperature (bottom) (Duan, Wang, et al., 2022).Yellow bar indicates the Termination-II transition.(b) Scatterplot for Mg/Ca ratio and Δ 47 -cave air temperature (Duan, Wang, et al., 2022) and their correlation coefficient.Both Mg/Ca ratio and Δ 47 -cave air temperature data are smoothed using a LOESS (Cleveland, 1979) regression with a 100-year-window.
Meanwhile, an elevation in temperature can induce an augmented biogenic CO 2 presence in soil, leading to decreased δ 13 C values in speleothems.That is because biogenic CO 2 is characterized by 13 C-depletion (Arienzo et al., 2015(Arienzo et al., , 2017;;Genty et al., 2003).Consequently, the inverse relationship between Mg/Ca ratio and δ 13 C serves to reinforce the temperature dependency of Mg/Ca ratio in XL-4.A moderately strong negative correlation is observed between Mg/Ca ratio and δ 18 O (r = 0.67, p < 0.01) (Figure 2c).Previous studies have consistently shown that Chinese speleothem δ 18 O values are higher during relatively cold deglacial periods and lower during relatively warm interglacial periods (Cheng et al., 2009(Cheng et al., , 2016;;Duan, Cheng, et al., 2022;Duan et al., 2019;Wang (Duan, Wang, et al., 2022) smoothed using a LOESS (Cleveland, 1979) regression with a 100-year-window.(b) δ 13 C record. (c) δ 18 O record (Duan et al., 2019).Yellow bar marks the Termination-II transition.et al., 2008).Thus, the inverse co-variation of Mg/Ca ratio and δ 18 O serves as indirect evidence for the temperature-dependence of Mg/Ca.This study furnishes evidence supporting the viability of utilizing Mg/Ca ratio in a stalagmite as a palaeotemperature proxy over millennial and centennial timescales, albeit requiring validation through additional stalagmite records in the future.

Mechanisms for Temperature-Dependence of Mg/Ca Ratio in XL-4
Trace elements incorporation into carbonate can be elucidated through the concept of the distribution coefficient (D X ), which is defined as: where (X/Ca) CaCO3 and (X/Ca) aq is the trace-element-to-Ca-ratio of carbonate and source water, respectively.A positive relationship between D Mg and temperature has been demonstrated in numerous laboratory and cave studies (Carlson et al., 2018;Day & Henderson, 2013;Gascoyne, 1983;Huang & Fairchild, 2001;Huang et al., 2001;Mucci, 1987;Mucci & Morse, 1983;Wassenburg et al., 2020).In theory, the prevalence of thermal forcing over hydrological influences on Mg/Ca ratio implies that Mg/Ca ratio in drip water recharging this stalagmite does not change significantly over time and/or that its temperature-dependent slope of D Mg may be significantly steeper than other speleothems (Drysdale et al., 2019(Drysdale et al., , 2020)).
Based on drip water collected monthly from July to December 2023, the coefficients of variation (CV, 1σ) for Mg/ Ca and Sr/Ca ratios are as follows: 0.5% and 1.1% for XL-1, 13.8% and 6.9% for XL-2, 0.7% and 0.6% for XL-3, and 5.6% and 5.9% for XL-5, respectively (Figure S3 in Supporting Information S1).Except for XL-2, CV values for Mg/Ca and Sr/Ca ratios are comparable to or even lower than pool water within an Italian cave (4.7% for Mg/ Ca ratio and 6.6% for Sr/Ca ratio) (Drysdale et al., 2019(Drysdale et al., , 2020)).This suggests that the seasonal variations of drip water Mg/Ca and Sr/Ca ratios at most sites are limited and even comparable to stable pool water.Drip rates of XL-1 and XL-3 are higher during rainy season and lower during dry season, roughly following precipitation changes.Drip rates of site XL-5 remain constant throughout monitored interval (Figure S3 in Supporting Information S1).Drip rates at site XL-2 are not considered due to nearby construction activities.Therefore, although Mg/Ca and Sr/Ca ratios of drip water at most sites are relatively invariant, their drip rates are not necessarily constant.Their discharge modes are site-specific and drip water may be a mixture of integrated percolation water over a long period.Average (±1σ) of Mg/Ca and Sr/Ca ratios of drip water collected intermittently across eight drip sites between October 2017 and December 2023 is 0.999 ± 0.126 mol/mol and 0.211 ± 0.028 mmol/mol respectively (Figures S2a and S2b in Supporting Information S1).CV values for Mg/Ca and Sr/Ca ratios both are ∼13%, indicating the insignificant spatial variations of Mg/Ca and Sr/Ca ratios of percolation water.The relatively stable hydrogeochemistry of drip water is plausibly ascribed to the considerable thickness of overlying bedrock (∼120-m thick) (Duan et al., 2016(Duan et al., , 2019;;Duan, Wang, et al., 2022).Although there are some fissures within bedrock, they are not located above stalagmite XL-4 and the monitored drip sites.
XL-4 exhibits a uniform cylindrical morphology with a flat apex and a constant diameter of ∼100 mm.The longitudinal section reveals subhorizontal growth layers with approximately equal width (Figure S1b in Supporting Information S1).Considering the thick, fissure-free bedrock above XL-4, its morphology suggests that XL-4 formed under relatively homogeneous conditions.It is probable that it was primarily recharged by wellmixed stored groundwater characterized by relatively stable hydrogeochemistry (Dreybrodt & Romanov, 2008;Frisia, 2015;Martín-Chivelet et al., 2017).However, this argument remains speculative, relying on theoretical models from prior studies rather than conclusive empirical evidence.
If the hydrogeochemistry of drip water recharging XL-4 remained constant, we need to address why Sr/Ca ratio record exhibits an inverted-U shape.A weak positive correlation exists between Sr/Ca ratio and growth rate (r = 0.45, p < 0.01).Notably, a moderately strong positive correlation emerges during 133.4-128.5 ka BP (r = 0.54, p < 0.01) (Figure S7 in Supporting Information S1).This indicates that the inverted-U pattern of Sr/Ca ratio record is primarily shaped by variations in growth rate (Huang & Fairchild, 2001;Mucci & Morse, 1983;Treble et al., 2003), rather than drip water Sr/Ca ratio.Nevertheless, from 128.5 to 126.6 ka BP, Sr/Ca ratio shows
a step-wise decreasing trend, in contrast to a sharp decrease in growth rate at ∼128.5 ka BP and subsequent inconspicuous decreasing trend (Figure S7 in Supporting Information S1).One plausible explanation is that growth rate is calculated using age-depth model with age errors of ∼250 years (Figure S5 in Supporting Information S1); thus the actual peak of growth rate may not be as sharp as depicted.
The high temperature sensitivity of D Mg within a subaqueous speleothem has been attributed to its exceptionally slow growth rate (Drysdale et al., 2020).The average growth rate of XL-4 is ∼0.029 mm/year, significantly lower than the adjacent stalagmite (∼0.124 mm/year) (Duan et al., 2016).Consequently, the pronounced temperature sensitivity of D Mg in XL-4 is likewise conceivable.To check this, we calculated D Mg values (Equation 1), assuming stable Mg/Ca ratio in drip water recharging XL-4, equivalent to the modern average of 0.999 ± 0.126 mol/mol.D Mg ranges from 0.020 to 0.052, with mean of 0.031 ± 0.008.The relationship between D Mg and cave air temperature can be expressed by either exponential regression (Equation 2) or linear regression (Equation 3) as follows (Figure 3): The modern D Mg of calcites deposited on glass plates at drip sites XL-2 and XL-3, where cave air temperature is 13.0°C, are 0.023 and 0.027, respectively.Both data sets of D Mg -temperature fall below the regression lines (Figure 3).This discrepancy is likely due to the glass substrate potentially not being truly representative of natural Green line is from Day and Henderson (2013).
calcite growth, particularly concerning the effects of crystal growth and morphology on D Mg (Drysdale et al., 2020).The slope of temperature dependence of D Mg in XL-4 is 0.0024/°C (Equation 3), comparable to subaqueous speleothem (0.0022/°C) (Drysdale et al., 2020), yet approximately an order of magnitude greater than those reported in other studies (Figure 3) (Carlson et al., 2018;Day & Henderson, 2013;Gascoyne, 1983;Huang & Fairchild, 2001;Huang et al., 2001;Mucci, 1987;Mucci & Morse, 1983;Rimstidt et al., 1998;Wassenburg et al., 2020).Given drier climate conditions during weak monsoon interval (WMI) within T-II, in comparison to present (Cheng et al., 2009(Cheng et al., , 2016;;Duan, Cheng, et al., 2022;Duan et al., 2019;Wang et al., 2008), it is highly improbable that drip water Mg/Ca ratios would be equivalent to or lower than modern values.For instance, if we consider a hypothetical drip water Mg/Ca ratio of 1.199 mol/mol, which is 0.20 mol/mol higher than modern values, a D Mg of 0.017 is necessary to achieve the lowest Mg/Ca ratio of 20.31 mmol/mol of XL-4 during T-II (Figure S8 in Supporting Information S1).Consequently, the actual D Mg during cold WMI would be smaller than currently estimated 0.020.This, in turn, suggests that the slope of 0.0024/°C represents a conservative minimum value.Thus, the stronger temperature dependence of D Mg , compared to other calcites, is another potential explanation for the dominance of thermal forcing over hydrological influences on XL-4 Mg/Ca ratio (Drysdale et al., 2019(Drysdale et al., , 2020)).
In summary, our results suggest that Mg/Ca ratio in stalagmites can serve as a potential palaeo-temperature proxy, provided that the stalagmite formed under specific conditions.Taking XL-4 as an illustrative case, we propose a set of discerning criteria for stalagmite selection.Firstly, the overlying bedrock should be sufficiently thick to ensure that the percolation water reaching cave is well-mixed.The stalagmite should be replenished by stored groundwater, characterized by relatively stable Mg/Ca ratio over time.Secondly, the stalagmite should have a slow growth rate, because the kinetics of slow carbonate precipitation could amplify the temperature dependency of D Mg (Drysdale et al., 2020).Thirdly, stalagmites conforming to a uniform cylindrical morphology with a flat apex and constant diameter, revealing subhorizontal growth layers of approximately equal width in its longitudinal section.Finally, Mg/Ca ratio demonstrates no correlation with Sr/Ca ratio but exhibits a moderately strong negative correlation with δ 13 C and δ 18 O.

Mg/Ca-Temperature Calibration for Stalagmites
Numerous investigations have been undertaken to establish the calibration for planktic foraminiferal Mg/Ca ratio relative to seawater temperature (Lea et al., 1999;Vázquez Riveiros et al., 2016).However, the Mg/Ca-temperature calibration for speleothems has not been reported, primarily due to the absence of independent absolute temperature reconstructions derived from the same speleothem.We propose a quantitative calibration of Mg/Ca ratio in relation to temperature for speleothems, utilizing the data of Mg/Ca ratio and Δ 47 -cave air temperature from XL-4.The exponential regression is expressed as follows (Figure S9 in Supporting Information S1): Mg/Ca (mmol/mol) = 14.221 × e 0.073×T (°C) R 2 = 0.40,p < 0.01) This equation is plausibly applicable to the cases in mid latitudes, encompassing temperature range between ∼2 and ∼15°C.The temperature sensitivity coefficient for Mg incorporation into calcite is 0.073 (Equation 4), comparable to the integrated planktic foraminiferal Mg/Ca-temperature calibrations (0.065 and 0.084) (Vázquez Riveiros et al., 2016).If stalagmites meet the aforementioned criteria, the high-resolution Mg/Ca ratio data from them could tentatively provide a quantitative record of regional temperature by employing Equation 4.However, it is crucial to acknowledge the existence of discrepancies between Mg/Ca ratio and cave air temperature (Figure 1).Consequently, uncertainties may emerge in temperature estimations when there were significant changes in hydrological conditions.For instance, in segments demonstrating positive co-variations of Mg/Ca ratios with δ 13 C and δ 18 O induced by PCP associated with drought, temperature estimations may potentially exceed the true values.

Conclusions
Here, we present a stalagmite Mg/Ca ratio record from northern China, encompassing T-II.There is a moderately strong positive correlation between Mg/Ca ratio and cave air temperature derived from Δ 47 data within the same stalagmite.This observation marks a departure from prior investigations, wherein Mg/Ca ratio in speleothems primarily served as an indicator of hydrological conditions.The temperature dependence of Mg/Ca ratio is plausibly attributed to the relatively stable hydrogeochemistry of the source water and the remarkably high temperature sensitivity of D Mg .Furthermore, we tentatively established an Mg/Ca-temperature calibration function for stalagmites in mid-latitudinal regions.
; Elderfield & Ganssen, 2000; Elderfield et al., 2012; Evans & • A Chinese stalagmite at termination II shows moderately strong positive correlation between Mg/Ca ratio and clumped isotopes-temperature • Stable drip water hydrogeochemistry and high temperature sensitivity of Mg partition coefficient may explain this temperature dependence • We tentatively establish, for the first time, an Mg/Ca-temperature calibration function for speleothems Supporting Information:

Figure 3 .
Figure 3.Comparison of D Mg versus temperature from this study (blue dots and lines) and published cave/cave analog studies.D Mg values are calculated under the assumption that Mg/Ca ratio in drip water recharging XL-4 remains invariant over time, equal to the modern mean of 0.999 ± 0.126 mol/mol.The exponential and linear regressions are depicted by a light blue solid line and a dark blue dashed line, respectively.Thin lines illustrate the regressions adjusted for drip water Mg/Ca ratio, shifting by ±1σ, with values of 0.873 and 1.125 mol/mol.Two blue circles denote data sets of modern D Mg -temperature in Xinglong cave.Purple squares and the related regression line are from Drysdale et al. (2020).Pink line is from Wassenburg et al. (2020).Black line is from the data of Rimstidt et al. (1998) as published inDay and Henderson (2013).Green line is fromDay and Henderson (2013).