A speleothem record of seasonality and moisture transport around the 8.2 ka event in Central Europe (Vacska Cave, Hungary)

Abstract

A stalagmite was collected in northern Hungary from the Vacska Cave, where monitoring and ventilation-based site selection had been conducted. The stalagmite covers the 10–8 ka (relative to AD 1950) period, including the so-called 8.2 ka event, and showed preceding signs of climate change that were evaluated by petrographic observations, ¹⁴C activities, Sr concentrations, and stable isotope compositions of calcite and inclusion-hosted water. Comparisons of speleothem records show that isotope peaks at ca. 8.5 ka are related to a regional climate anomaly, rather than to a continental-scale event. In accordance with regional proxy records, the 8.2 ka event was associated with a series of temperature and precipitation amount changes, starting with cooling and a reduction in the winter-to-summer precipitation ratio, and then becoming a humid and warm phase at 8.15 ka. X-ray diffraction-based crystallinity parameter (FWHM) values provided evidence for diagenetic alteration of the stable oxygen isotope compositions of inclusion waters. Nevertheless, the stable hydrogen isotope compositions of inclusion waters and the oxygen isotope values of the host calcite revealed elevated d-excess values, and therefore increased Mediterranean moisture contribution during the 8.2 ka event, which indirectly indicate the southward displacement of moisture transport from the Atlantic Ocean.

Introduction

The pioneering multiproxy analysis of the Greenland GISP2 ice core in Alley et al. (1997) detected a prominent climate anomaly between 8 ka and 8.4 ka (all ages in this paper are reported relative to AD 1950) and revealed its global significance. Ever since, the “8.2 ka event” has become common knowledge in paleoclimate research. Hundreds of studies have covered the appearance, magnitude, and duration of climate-proxy signals in various marine and terrestrial records, as well as conducted model calculations to analyze global and local temperature, precipitation, and seasonality changes associated with the 8.2 ka event. The 8.2 ka event is best observed in Greenland ice cores, where stable oxygen isotopic (δ¹⁸O) records show an approximately −2.5‰ peak (Thomas et al., 2007). The negative δ¹⁸O change is generally attributed to the sudden cooling caused by low δ¹⁸O freshwater discharge from lakes Agassiz and Ojibway (in the area of the Laurentide icesheet) to the North Atlantic Ocean at 8.470 ka (Barber et al., 1999). The freshwater discharge not only changed the oxygen isotopic composition of the moisture source, but also disturbed the North Atlantic thermohaline circulation, resulting in a weakened Atlantic Meridional Overturning Circulation (AMOC) (Alley et al., 1997; Barber et al., 1999). Although this early freshwater discharge caused changes in North Atlantic Ocean current strengths (Tegzes et al, 2014), the AMOC was not severely affected, and only the 8.2 ka event sensu stricto was associated with significant AMOC reduction (e.g., Shi et al., 2020). The AMOC reduction, in turn, resulted in southward displacement of the Intertropical Convergence Zone (ITCZ) and increased the heat and moisture transport to the European continent, as reflected by speleothem stable isotope records (e.g., Sha et al., 2019; Demény et al., 2021a). The movement of the ITCZ dispersed the effects of the 8.2 ka climate change worldwide, producing a global signal (Cheng et al., 2009; Morill et al., 2013; Parker and Harrison, 2022). As such, the 8.2 ka δ¹⁸O peak may be used as a time marker to verify age-depth models (Benson et al., 2021), and its analysis also can provide information on seasonal temperature and precipitation changes (Prasad et al., 2009; Morill et al., 2013; Peckover et al., 2019). High-resolution analyses of several Greenland ice core δ¹⁸O records led Thomas et al. (2007) to suggest a dual structure of the δ¹⁸O peak, with a “whole event” from 8.247–8.086 ka and a “central event” from 8.212–8.141 ka. Brouard et al. (2021) reviewed 597 ¹⁴C records from marine and continental archives and determined that a freshwater discharge occurred at 8.220 ka and a subsequent lake dam breakup occurred at 8.160 ka. These studies indicate that the 8.2 ka event potentially had a fine structure that may be associated with complex changes in precipitation, as well as in temporal and spatial temperature variation (e.g., Seppä et al., 2007; Prasad et al., 2009).

Due to age uncertainties, it is sometimes difficult to determine if an individual δ¹⁸O change in the paleoclimate record is related to the 8.2 ka event or if it represents a preceding or subsequent event, or even a broad period of climate change (Rohling and Pälike, 2005). Parker and Harrison (2022) provided a global compilation of well-dated speleothem records that were sampled at sufficient resolution (≤ 30 years). The global average start (8.223 ka) and end (8.062 ka) dates of the 8.2 ka δ¹⁸O peaks (Fig. 1A) fit the ages reported by Thomas et al. (2007), suggesting a globally coherent signal. However, some of the negative speleothem δ¹⁸O peaks included in this calculation would fall outside the 8.2 ka period if the reported age uncertainties were considered. One of the most prominent records is the Jeita Cave (Lebanon) δ¹⁸O record (Cheng et al., 2015; Fig. 1B), which is exceptionally well dated (63 age dates for the period of 12.180–1.210 ka and 5 ages for the 9–8 ka period, with an average age uncertainty of 42 years; Cheng et al., 2015). The Jeita δ¹⁸O record is also exceptional because it contains strongly negative δ¹⁸O shifts before and after the 8.2 ka event, which would indicate that the 8.2 ka peak of the Greenland ice core records (e.g., NEEM, Gkinis et al., 2021; Fig. 1C) would instead fit a positive shift in the Jeita record within the age uncertainties (Fig. 1). This observation raises the following questions: (1) is the 8.2 ka event represented exclusively by negative δ¹⁸O shifts in Europe and the Mediterranean, and (2) are pre- and post-8.2 ka peaks also related to the multiple freshwater discharge events in the North Atlantic?

Figure 1. (A) List of δ¹⁸O records compiled by Parker and Harrison (2022) in their Table S1, with the time intervals of their δ¹⁸O peaks, as well as (B) the Jeita Cave (Cheng et al., 2015) and (C) the NEEM ice core (Gkinis et al., 2021) δ¹⁸O records. The age uncertainty of the Jeita record is also shown with arrows. Shaded bar shows the interval of the “whole event” from 8.247–8.086 ka defined by Thomas et al. (2007). The δ¹⁸O values are all in ‰ relative to VPDB (B) and VSMOW (C).

In this study, we present stable isotope and trace element records of a stalagmite from northern Hungary that covers the 10–8 ka period and indicates a climate change event at ca. 8.5 ka (Demény et al., 2022). Conventional high-resolution stable carbon and oxygen isotope analyses of stalagmite carbonate are supplemented with petrographic observations, stable hydrogen and oxygen isotope analyses of inclusion-hosted waters, X-ray diffraction-based crystallinity measurements, Mg-P-Sr concentration determinations, and AMS ¹⁴C analyses to determine temperature and precipitation changes during the 9–8 ka period, as well as to derive moisture transport information of the 8.2 ka event.

Cave site description

The Pilis Mountains consist primarily of Triassic limestones (Dachstein Limestone) and dolostones (Haas, 2001), but small outcrops of Oligocene Hárshegyi Sandstone are also found (Deák et al, 2007) in the vicinity of the Ariadne cave system. The Dachstein Limestone Formation is >1 km thick. The Vacska Cave (Fig. 2) is a part of the Ariadne cave system that contains seven caves, which are now interconnected due to cave exploration works. The Vacska Cave (47°41′N, 18°50′E) is 9298 m long and 206 m deep, with an entrance at 459 m asl.

Figure 2. Location map of the Vacska Cave with the site (Fennkőháti chamber) of the V-03 stalagmite.

The Ariadne cave system is in a densely forested area dominated by deciduous trees. Cave monitoring by temperature and CO₂ content, as well as drip-water pH, chemistry, and stable isotope compositions were conducted in 2019 and 2020 (Czuppon et al., 2022a, b). The local average annual surface air temperature between 2008 and 2020 was 10.4°C (measured directly above the cave) whereas the cave air temperature near the study site (Fennkőháti chamber, Fig. 2) was 8.8 ± 0.01°C (Czuppon et al., 2022a). The mean annual precipitation amount between 1950 and 2019 was 645 mm (min = 360 mm, max = 1200 mm). During this period, the most arid month was March (37 ± 24 mm), and the most humid month was June (76 ± 41 mm) (Supplementary Fig. 1).

The annual amount-weighted averages of the stable hydrogen and oxygen isotope compositions of precipitation at a nearby station (at Piliscsaba, ~8 km SE from the cave) were δ²H = −60.1‰ and δ¹⁸O = −8.6‰ in 2019 and 2020, respectively (Czuppon et al., 2022a). The study site's drip water had slightly lower isotope compositions (δ²H = −70.7 ± 0.8‰, δ¹⁸O = −10.2 ± 0.13‰), indicating a 70–75% contribution of winter half-year precipitation to the drip water (Czuppon et al., 2022a). The CO₂ concentration in the cave air varied between 550 ppmv (parts per million by volume) in winter and 9700 ppmv in summer during the 2019–2020 monitoring period (Czuppon et al., 2022a). In general, the entire cave system is well ventilated. Czuppon et al. (2022b) searched for the least-ventilated part of the cave system using comprehensive stable isotope analysis of freshly precipitated carbonate on active speleothem surfaces. Although strongly positive C and O isotope correlations were found for many locations, indicating kinetic isotope fractionation due to ventilation, the deepest parts of Vacska Cave were found to be intact. As a result, a stalagmite (called V-03) that showed appropriate U and low Th concentrations (Czuppon et al., 2022b) was selected from this sector for detailed analyses (Supplementary Fig. 2).These results are presented in this study.

Methods

Petrographic analysis and lamina counting were conducted on polished thin (~100 μm) sections using a Nikon Eclipse E600 POL optical microscope.

For the X-ray diffraction (XRD) analyses, 40 mg of finely powdered calcite samples were used in a suspension of 1 cm³ ethanol and were dried on a 4 × 3 × 0.1 cm metal sample holder. The XRD measurements were conducted using a Rigaku Miniflex600 powder diffractometer, with a 35 min analysis time, 2–70° 2Θ range, 0.05° step size, 2°/min rate, Cu X-ray source, graphite monochromator, and scintillation counter. Full width at half maximum (FWHM) values were determined for the 2Θ = 29.39 peak. The Philips X'pert Software (1.2a) was used for FWHM value calculations.

Sr concentrations were determined using a Spectro xSORT HH03 handheld energy-dispersive X-ray fluorescence (XRF) instrument, which consists of a Rh X-ray tube, two filters, and a silicon drift detector (SDD). Two spectra were recorded for optimum excitation of the X-ray tube values during measurements with 50 kV 0.05 mA and 15 kV 0.12 mA. The “FP- Mining” calibration developed by Spectro was used for the measurements. The instrument was slid along the surface of the stalactite in mm increments and provided information from a sample area of ~3 mm.

XRF analyses also were performed on stalagmites that previously had been measured by laser ablation inductively coupled mass spectrometry (LA ICP-MS). LA ICP-MS analyses were conducted on two stalagmites (Kiskőhát: Siklósy et al., 2009; BAR-II L: Demény et al., 2017) using a Perkin-Elmer ELAN 6100 DRC ICP-MS coupled with a LAMBDA PHYSICS excimer laser (193 nm) at the University of Lausanne (Siklósy et al., 2009). The XRF-based and LA ICP-MS results are shown in Supplementary Figure 3. Although there are some sections where the two records do not match (BAR-II L: ~16–190 mm; Kiskőhát: ~202–228 mm; Supplementary Fig. 3), most of the records are similar, and the major patterns are reproduced, supporting that XRF analyses can be used to detect major Sr concentration changes. The Sr concentration range of the V-03 stalagmite studied in this paper exceeds the XRF–ICP-MS difference, thus, the XRF-based Sr ppm variation may reflect primary, stalagmite-related features beyond analytical uncertainty.

Calcite samples were drilled from a polished stalagmite surface using a 0.6-mm drill bit. The entire stalagmite was sampled at a resolution of ~1 mm, but the section of 39.5–59.9 mm dft (distance from top) was sampled by drilling a trench with an average resolution of 0.1 mm (ranging from 0.04–0.2 mm, depending on the density of calcite laminations). Stable carbon and oxygen isotope compositions were determined using an automated carbonate preparation device (GASBENCH II) and a Thermo Finnigan delta plus XP continuous-flow mass spectrometer at the Institute for Geological and Geochemical Research, Research Centre for Astronomy and Earth Sciences (Budapest, Hungary). Three laboratory standards, calibrated using the NBS-18, NBS-19, and LSVEC reference materials (provided by the International Atomic Energy Agency), were used for sample standardization. These isotope compositions are expressed as δ¹³C and δ¹⁸O values (in ‰) relative to the Vienna Pee Dee Belemnite (VPDB). As a test of external precision, the Harding Iceland Spar (Landis, 1983) sample was measured as unknown and yielded δ¹³C and δ¹⁸O values of −4.82 ± 0.02‰ and −18.57 ± 0.06‰, respectively (n = 57 in the year of the V-03 analyses). These values agree with the published δ¹³C and δ¹⁸O values of −4.80‰ and −18.56‰, respectively (Landis, 1983) (differences between the measured and published values indicate external accuracies).

Inclusion-hosted water was extracted by crushing speleothem chips in stainless steel tubes, followed by measurements of the stable hydrogen and oxygen isotope compositions with a model LWIA-24d (Los Gatos Research Ltd.) liquid water isotope analyzer, as described by Demény et al. (2016). The isotope compositions are expressed in conventional δ²H and δ¹⁸O values (as δ¹⁸O_fi for fluid inclusions to distinguish from the calcite's δ¹⁸O_cc) relative to Vienna Standard Mean Ocean Water (VSMOW). Analytical accuracies are about ± 2‰ for δ²H and ± 0.5‰ for δ¹⁸O_fi on the basis of sample measurements with known isotopic compositions (Demény et al., 2021b).

Twenty accelerator mass spectrometry (AMS) ¹⁴C analyses were conducted for the entire stalagmite section. First, 10–20 mg of carbonate samples were treated with 85% orthophosphoric acid (H₃PO₄) in a vacuum-tight digestion reactor and kept overnight at 70°C to release the carbon as CO₂. Next, the liberated CO₂ was purified and trapped cryogenically with a customized vacuum line (Molnár et al., 2013a). AMS graphite targets were prepared from the CO₂ samples by a customized sealed tube graphitization method (Rinyu et al., 2013). The ¹⁴C/¹²C ratios were determined at the EnvironMICADAS ¹⁴C facility (Molnár et al., 2013b), Hertelendi Laboratory of Environmental Studies at the Institute for Nuclear Research, (Debrecen, Hungary).

Uranium series dating analyses were performed on 14 samples at the Isotope Climatology and Environmental Research Centre at the Institute for Nuclear Research (Debrecen, Hungary). The analyzed parts of the samples were completely dissolved in diluted (~10%) hydrochloric acid (HCl). The samples were then spiked with a ²²⁹Th/^233,236U tracer to remove organic material, after which they were converted to nitric form by boiling nitric acid (HNO₃). The U/Th separations were conducted with extraction chromatography on UTEVA resin. The final fractions were measured using a Thermo Scientific Neptune Plus MC-ICPMS. The uncertainties of the ages were calculated with the Monte Carlo method.

The calcite samples used earlier for fluid inclusion analyses were powdered, and batches were selected for trace element analyses to ensure there was sufficient material. The powdered samples were divided into two parts for two parallel measurements. The samples (~100–300 mg) were weighed with an analytical scale into glass beakers to which 3 ml of concentrated HCl (37%, Suprapur grade) was slowly added. After 30 minutes, the solution was transferred into a volumetric flask and topped up to 20 ml with ultrapure water (SUEZ, 18.2 MΩ/cm). The trace element concentrations of the sample solutions were measured with a Spectro Genesis Inductively Coupled Plasma Optical Emission Spectroscopy (ICP-OES) simultaneous spectrometer, with axial plasma observation system. The instrument was calibrated with a multielement standard (LobaChemie, Multielement Standard Solution for ICP 33 components 100 mg/l).

The studied stalagmite

The V-03 stalagmite was collected near the Fennkőháti chamber (Fig. 2). The stalagmite formed on a flowstone layer that was later broken either by rock movements or during cave exploration. It is 15.2 cm from the flowstone base, with a diameter of ~7.5 cm (Supplementary Fig. 2). The stalagmite is composed of finely laminated clear calcite with no visible hiatus. The laminae have flat tops that are ~3 cm wide. No pores are visible in the axial part, although some pores appear at the flanks. The lamination is expressed by thin, brownish layers, which likely contain organic material, and the alternation of inclusion-poor and inclusion-rich layers (Fig. 3). The stalagmite's fabric is dominated by columnar calcite with a varying elongation rate (Supplementary Fig. 2).

Figure 3. Optical microscope images of the 45–85 mm dft (distance from the top, in mm) interval with one nicol (1N) and crossed nicols (+N), as well as the stable C and O isotope compositions (in ‰ relative to VPDB). White arrow shows a fabric change where fluid inclusion trails are shifted, indicating drifting growth axis. Subsections with low hydrogen and oxygen isotope compositions of inclusion water are also shown. The hiatus at 48.0 mm dft is marked by arrows. δ¹³C and δ¹⁸O records: gray lines = low-resolution sampling; black lines = high-resolution sampling.

The present study focuses on the part that is 45–85 mm from the top of the stalagmite and is covered by one thin section. This section is dominated by elongated, compact columnar calcite (Fig. 3). From ~81 mm to ~78 mm dft (distance from top), the growth direction is shifted, and then growth continues in the original direction. This shift is expressed by the calcite crystals, as well as the fluid inclusion distribution (Fig. 3). Large inclusions, representing inter-crystalline voids, appear in several layers at 73 mm, 67 mm, and 52 mm dft. At 48.0 mm dft, there is a hiatus with flat crystal terminations, followed by new crystal formation. The surface of the hiatus is sharp, possibly suggesting erosion, but the “mesa” and “valley” forms (Railsback et al., 2013), which would indicate strong erosion, are absent. The surface is composed of pure and clean calcite, lacking the detrital material that would have been deposited if the surface underwent prolonged dry exposure. Thus, the hiatus is considered to be a short-term growth stop with dissolution.

Results

Fourteen U-Th age determinations were conducted on the section 15–145 mm dft (Supplementary Table 1). Excluding one outlier (at 55 mm dft; Supplementary Fig. 2), the age-depth relationship was established using the StalAge algorithm (Scholz and Hoffmann, 2011) (Supplementary Fig. 2). The ages range between 10,220 ± 165 years at 141 mm dft and 8140 ± 130 years at 16 mm dft. The 2σ uncertainties range from 125–270 years, with an average of 166 years. Although the StalAge algorithm cannot handle the hiatus detected at 48.0 mm dft, the distribution of ages indicates that the hiatus did not span more than some decades, and thus was not investigated further.

To determine the recent contribution of rock-related carbon (“dead carbon proportion” [dcp]) at the study site, the top layer of a recently forming stalagmite (V-01) collected in the Vacska Cave, representing the 2008–2020 period (Demény et al., 2021c), also was analyzed (Supplementary Table 2). The average atmospheric CO₂ radiocarbon activity value, which was calculated from the datasets of Hua et al. (2013) and Major et al. (2018), yielded a percentage modern carbon (pMC) value of 102.09. This value, together with the measured pMC value of the V-01 stalagmite (97.19; Table 1), yields a dcp value of 4.8%. The V-03 ¹⁴C data increase from 29.93 ± 0.12 pMC to 37.89 ± 0.14 pMC without a major jump (Supplementary Table 2). The V-03 ¹⁴C ages were corrected for a set of potential dcp values from 4–10%. The dcp-corrected ¹⁴C ages were calibrated using OxCal 4.1 (Bronk Ramsey, 2001) and the IntCal20 dataset (Reimer et al., 2020). Age-depth relationships were modeled using the P_Sequence function (Bronk Ramsey, 2008). The best visual match of the U-Th ages and the StalAge age-depth relationship was observed with the AMS ¹⁴C deposition chronologies considering dcp corrections from 5–8% and (Fig. 4). This dcp range is usual among similar karstic environments in the region (Molnár et al., 2006, 2016; Demény et al., 2019b). Most of the U-Th ages plot within the 5% and 8% dcp age-depth models (Fig. 4), except the 40–70 mm dft section, in which elevated dead carbon contribution can be assumed.

Figure 4. U-Th ages (with 2σ uncertainties), and AMS ¹⁴C-based age models assuming 5% (blue shading) and 8% (brown shading) dead carbon proportion (dcp), the StalAge-based age-depth model (dashed line), and lamina-counting results (solid line) for the V-03 stalagmite. See text for details.

The younger part of the stalagmite (~7 cm dft) showed clearly visible lamination under the binocular and optical microscopes. As such, this section was used for lamina counting, under the assumption that a single lamina would represent an annual layer. Laminae were counted starting at 71.0 mm dft at a U-Th dating spot that yielded 8450 ± 160 (2σ) years Within the topmost 71-mm lamina, counts were conducted in 8 subsections, and the lamina numbers were subtracted. The lamina-based age model, starting from 8450 years (Fig. 4) closely follows both the StalAge-based and the ¹⁴C-based age-depth relationships that support the use of the StalAge model ages. However, lamina counting was not used to improve the U-Th age-based StalAge model, because detection of annual laminae is not definitively proven.

The XRF-based Sr concentrations and stable C-O isotope compositions of the stalagmite calcite (Supplementary Table 3) are shown in Figure 5, along with the U-Th ages and the StalAge age-depth model curve. Compared to the 0–100 mm section, the δ¹³C record shows strong fluctuation below 100 mm dft, with several positive shifts (between ~140–150 mm, at ~123 mm, and at ~100–106 mm), and a negative peak at 116 mm dft. The δ¹⁸O_cc record shows smaller variations in the 0–40 mm and 90–140 mm dft sections, whereas the 40–90 mm dft section is characterized by stronger δ¹⁸O_cc fluctuation. The lowermost δ¹⁸O_cc value in this section is at 52.17 mm dft (8190 ± 105 years), so the peak is tentatively attributed to the 8.2 ka peak (Fig. 5). A similar shift is visible at 82.7 mm dft (corresponding to a StalAge model age of 8524 years). The Sr concentration scatters between 32 ppm and 46 ppm.

Figure 5. Sr concentration (in ppm), with stable carbon and oxygen isotope compositions of stalagmite calcite (in ‰ relative to VPDB), U-Th ages, and the StalAge-based age-depth model (with 2σ uncertainty ranges). The shaded bar marks the low-δ¹⁸O peak tentatively related to the 8.2 ka event.

A short-term hiatus is indicated by a break in the records at 48 mm (Fig. 5). Just before the hiatus, the δ¹³C values strongly fluctuate and the Sr concentrations are low. Directly after the hiatus, the δ¹³C values decrease and the Sr concentrations are higher than they were before the hiatus.

The calcite and fluid-inclusion isotope compositions (Supplementary Table 4) are plotted as a function of age in Figure 6. High-resolution drilling was conducted in the 8.105–8.252 ka (39.5–59.9 mm dft) section to determine if additional peaks were detectable at a resolution of ~0.1 mm (ca. 1 year). The high resolution δ¹³C-δ¹⁸O records reproduced the low-resolution records without significant deviations (Figs. 3, 6). The patterns of stable H and O isotope compositions of inclusion-hosted water (δ²H, δ¹⁸O_fi) are different from those of the calcite records (Fig. 6). The δ¹⁸O_fi values fluctuate between −12.8‰ and −8.6‰, and one sampling section yielded a very low δ¹⁸O_fi value (−15.8‰) at 8.196 ka. This sample overlaps with the C-O isotope sampling section containing the lowest δ¹⁸O_cc value. The δ²H values range from −83.1‰ to −66.6‰ and are not correlated with the other isotope records. The lowermost δ²H values were obtained for the section of 8.521–8.433 ka. The 8.2 ka event's period is not associated with an appreciable δ²H shift; only a slight indication of a negative δ²H shift appears compared to the preceding and subsequent samples.

Figure 6. Stable hydrogen, with carbon and oxygen isotope compositions of the V-03 stalagmite calcite (δ¹³C and δ¹⁸O_cc, in ‰, relative to VPDB) and inclusion-hosted water (δ²H and δ¹⁸O_fi, in ‰, relative to VSMOW) and published records. Inclusion-hosted water compositions (in ‰, relative to VSMOW) for the Père Noël Cave (Belgium) from Allan et al. (2018); inclusion-hosted water compositions for the Milandre Cave (Switzerland) from Affolter et al. (2019); cave ice data from the Scărișoara Cave from Perșoiu et al. (2017). Calcite δ¹⁸O records (in ‰ relative to VPDB) from the Limnon, Hoti, Qunf, and Jeita caves are from Peckover et al. (2019), Neff et al. (2001), Fleitmann et al., 2007, and Cheng et al. (2015), respectively. The δ¹⁸O record of the NEEM ice core (blue line = 10-year moving average) is from Gkinis et al. (2021).

FWHM values were determined for the 2Θ = 29.4° calcite XRD peak and are listed in Supplementary Table 4. The FWHM values are positively correlated with the δ¹⁸O_fi values (r = 0.62, p = 0.02; Fig. 7; excluding the sample with δ¹⁸O_fi value = −15.8‰ would not change the correlation parameters) and show no relationship with the δ²H data (r = −0.08, p = 0.78).

Figure 7. δ¹⁸O deviations (in ‰; Δ¹⁸O_{inclusion–drip water} = δ¹⁸O_fi–δ¹⁸O_{drip water}) of inclusion-hosted water compositions from local drip-water δ¹⁸O values as a function of full width at half maximum (FWHM) values.

The ICP-MS-based trace element concentrations and stable isotope compositions of subsamples crushed for fluid inclusion analyses are listed in Supplementary Table 5. Because these materials were subsamples of other analyses, they are not compared with other, distance-controlled data. However, their internal relationships were investigated. The Mg contents range from ~900 ppm to ~1200 ppm, the P contents from 27–69 ppm, and the Sr concentrations from 23–29 ppm. The correlations are also given in Supplementary Table 5. The most significant correlation is between the δ¹³C values and P concentrations, with r = −0.82 and p = 0.00.

Discussion

General interpretations of chemical and isotopic proxies

This study utilizes Sr and P concentrations, stable carbon and oxygen isotope compositions of calcite, and stable hydrogen and oxygen isotope compositions of inclusion-hosted water, the basic interpretations of which are briefly summarized here.

The Sr content of speleothem calcite is determined by the Sr content of drip water, the partition coefficient between carbonate and water, and the carbonate precipitation rate (Huang and Fairchild, 2001). The Sr content of drip water is governed by rock weathering in the soil zone, the degree of carbonate rock dissolution, prior calcite precipitation (PCP) along the water migration route, and dilution due to increased precipitation and water seepage (Sinclair, 2011; Fairchild and Baker, 2012; Borsato et al., 2016). PCP is associated with a coupled increase in the δ¹³C and δ¹⁸O values of the precipitating carbonate (Fairchild and Baker, 2012; Steponaitis et al., 2015; Dreybrodt and Fohlmeister, 2022), so the presence of a positive Sr–δ¹³C–δ¹⁸O correlation may indicate PCP, whereas the absence of a positive δ¹³C-δ¹⁸O correlation may suggest the opposite. Because neither a significant change in growth rate nor a positive δ¹³C-δ¹⁸O correlation is observed in the V-03 stalagmite, the Sr content variation may be related to hydrological conditions such as precipitation amount (dilution effect) and changes in water residence time.

The stable carbon isotope composition of speleothem carbonate is affected by various factors, such as vegetation type, biogenic activity in the soil zone, carbonate rock dissolution, PCP, and kinetic fractionations during speleothem deposition (see Fairchild and Baker, 2012, and references therein). In the absence of coupled δ¹³C, δ¹⁸O and Sr content changes, PCP and ventilation-driven kinetic fractionation (Mickler et al., 2006; Steponaitis et al., 2015) can be excluded. Due to the short time period covered by the studied stalagmite, significant changes in vegetation type (¹³C-enriched C₄ or ¹²C-enriched C₃ plants) also can be dismissed. Biogenic activity in the soil manifests in the amount of ¹²C in the dissolved CO₂ and, therefore, in the precipitating carbonate. Biogenic activity increases under humid conditions, so the δ¹³C values may indicate variations in the amount of precipitation. Additionally, organic ligands in the drip water may transport phosphorus, which is absorbed on the surface of the precipitating carbonate (Fairchild et al., 2001). Negative δ¹³C-P relationships (i.e., less biogenic ¹²C in the DIC, higher δ¹³C in the speleothem carbonate, and less phosphorus transported by the drip water) have been found in speleothems from southern (Demény et al., 2019a) and northern (Demény et al., 2019b) Hungary, suggesting that a negative δ¹³C-P relationship may indicate paleohydrological control. Batches of carbonate samples used for analyses of inclusion water contents also were selected to determine P, Mg, and Sr concentrations (Supplementary Table S5). The δ¹³C values of the sample batches have no relation to the Mg contents, but they show a weak positive correlation with the Sr concentrations and a negative correlation with the P concentrations. This is in accordance with the assumption that humid conditions with elevated precipitation amounts result in drip water dilution (lower Sr content) and improved soil activity (lower δ¹³C, higher P), which were also observed in two other Hungarian caves (Demény et al., 2019a, b).

The oxygen isotope composition of speleothem calcite is determined by the drip water composition, formation temperature, and the temperature dependence of calcite-water oxygen isotope fractionation (McCrea, 1950, as the first study; Johnston et al., 2013, for empirical speleothem data). Additionally, site-related fractionation processes, such as ventilation-driven degassing and evaporation, or PCP may cause kinetic fractionation, affecting the δ¹³C and δ¹⁸O values of speleothem calcite. However, the kinetic fractionations may result in positive δ¹³C-δ¹⁸O correlations (Mickler et al., 2004, 2006). Positive δ¹³C-δ¹⁸O correlations also may be caused by climatic factors. For example, a warm and dry period may result in elevated δ¹³C values due to a decrease in biogenic activity, and higher δ¹⁸O values in the drip water. The drip water oxygen isotope composition also may be affected by several factors, including changes in vapor source composition, moisture transport trajectory, seasonality (winter/summer precipitation ratio), air temperature, and precipitation amount (e.g., Lachniet, 2009). Kern et al. (2019) compiled speleothem and precipitation composition data, as well as determined the effects of regional temperature variation and precipitation amount. The δ¹⁸O value of local precipitation depends primarily on the precipitation amount in the Mediterranean, whereas the Carpathian Basin is characterized by temperature-dependence of the δ¹⁸O_water values. The δ¹⁸O-temperature relationship is usually higher (~0.6‰/°C, Kern et al., 2019; Demény et al., 2021b) than that of calcite-water oxygen isotope fractionation (−0.23‰/°C; Tremaine et al., 2011). As such, the δ¹⁸O_cc value is mainly governed by surface-air temperature changes, provided that the calcite is formed under equilibrium conditions. Under arid conditions, however, the precipitation water may experience evaporation and ¹⁸O-enrichment that may be transferred to the speleothem site. This is why the δ¹⁸O values should be interpreted in combination with other proxy data.

In theory, the stable hydrogen and oxygen isotope values of fluid inclusion-hosted water should directly reflect drip water compositions, provided that late-stage alterations have not modified the original compositions.

Diagenetic alterations versus primary signals

In general, the δ²H values are regarded as intact regarding diagenesis due to the lack of hydrogen in the host calcite. However, because both the host calcite and the inclusion-hosted H₂O contain oxygen, diagenetic alteration may change the original δ¹⁸O_fi value (Demény et al., 2016), leading to independent δ²H and δ¹⁸O_fi patterns. As detected by earlier analyses, calcite recrystallization is reflected by FWHM values (Demény et al., 2016). The width of the 2Θ = 29.39 XRD peak is determined by the calcite-crystal domain size and lattice strain, but significant lattice deformation is not expected in the case of speleothems, consequently the FWHM changes may indicate crystal size variation (Demény et al., 2016). Furthermore, the sizes of calcite crystallites within the speleothem are determined by the original domain size of the carbonate precipitate and subsequent crystallization processes.

To investigate if diagenetic alteration may have influenced the δ¹⁸O_fi values, crystallinity measurements were conducted on the same samples that underwent fluid inclusion analyses. The FWHM values are positively correlated with the δ¹⁸O_fi data. However, to quantify the δ¹⁸O_fi shifts during re-crystallization and to make the changes comparable to other occurrences, the differences between the inclusion-hosted water oxygen isotope compositions and the local drip water δ¹⁸O value (−11.2‰; Czuppon et al., 2022a) were plotted as a function of FWHM values (Fig. 7). The data of Demény et al. (2016), which were gathered for a flowstone deposit of the Béke Cave (northeastern Hungary), are also shown in Figure 7. The slope of the V-03 data is very similar to the slope found for the Béke Cave flowstone (Demény et al., 2016), indicating a general diagenetic process.

The diagenetic alteration detected by FWHM analysis raises the question of whether the δ¹⁸O_cc data also were affected, which may bias the δ¹⁸O_cc-based temperature calculations. Assuming that the δ¹⁸O_cc values are primary compositions related to equilibrium calcite-water oxygen isotope fractionation and that the δ²H values reflect the composition of the drip water from which the calcite precipitated, combining these data would yield reliable formation temperatures. Using a local meteoric-water line equation, the water δ¹⁸O values (δ¹⁸O_w) can be estimated from the δ²H data, while the formation temperature is given by the δ¹⁸O_cc-δ¹⁸O_w fractionation and the relevant fractionation equation (Demény et al., 2021b). The fractionation equation of Johnston et al. (2013) was applied in this study because it also contains the data of Tremaine et al. (2011). The local meteoric water line (LMWL) equation determined for the northern part of the Carpathian Basin (δ²H = 7.6⋅δ¹⁸O + 6.0; Czuppon et al., 2018) was selected. This equation is very close to that of southern Hungary (δ²H = 7.8⋅δ¹⁸O + 6.1; Demény et al., 2021b, based on the 2004–2017 data of Fórizs et al., 2020), so the uncertainty of LMWL determination can be considered negligible. The formation temperatures obtained by this procedure are shown in Figure 8 as the δ¹⁸O_cc-based dataset.

Figure 8. (A) Calculated temperatures (in °C) using the δ¹⁸O_cc-based and the δ²H-based methods (Demény et al., 2021b). (B) Calculated d-excess values for the V-03 stalagmite using formation temperatures of 8°C, 9°C, and 10°C. (C) Reported d-excess values for the Scărișoara Cave ice (Perșoiu et al., 2017).

Another approach for paleotemperature calculations is based on the temperature-dependence of the hydrogen isotope compositions of precipitation water, for which the δ²H-T gradient must be known (Demény et al., 2021b). The local gradient is 5‰/°C (Demény et al., 2021b). The δ²H shift from the recent drip water composition (−71‰; Czuppon et al., 2022a), together with the δ²H-T gradient, yields the temperature shift from the recent condition (8.8°C; Czuppon et al., 2022a), thus providing past temperatures. The δ²H-based formation temperatures are also shown in Figure 8. The δ²H-based procedure yielded generally reasonable temperatures between 7.8–9.7°C, except for three values ca. 8.5 ka. In contrast, the δ¹⁸O_cc-based temperatures are too low (ca. 8.5 ka) and unrealistically high (8.25–8.1 ka). This may indicate diagenetic alteration in the δ¹⁸O_cc values. However, without evidence of mineralogical change (e.g., aragonite to calcite transformation), strong recrystallization (e.g., appearance of mosaic fabric instead of columnar calcite), deposition of secondary carbonate (e.g., veins, pore fillings), or a diagenesis-sensitive porous texture, significant diagenetic alteration is not expected in a dense and compact calcite formation such as the V-03 stalagmite. Other causes of these results must be considered.

Seasonal signals in the inclusion-based stable isotope data

One possibility for the explanation of different calculated temperature records is the difference of seasonal signals in the δ²H and δ¹⁸O_cc records. The winter and summer laminae may have formed under different conditions, resulting in a difference in the efficiency of inclusion-water entrapment and a seasonal bias in the δ²H values. To investigate this possibility, high-resolution δ¹³C and δ¹⁸O_cc analyses were conducted on the 8.25–8.1 ka section (Supplementary Fig. 4).

Two subsections (8.143–8.145 ka and 8.175–8.179 ka) yielded data at a resolution of 0.3 year, which may record winter and summer signals separately. The δ¹⁸O fluctuations in these subsections are 0.3–0.5‰. Considering mixing during sampling, we can safely estimate that the difference in winter and summer δ¹⁸O_cc values is ~0.5–1‰. This shift in the δ¹⁸O_cc values would change the calculated temperatures by 2.2–4.5°C, bringing the δ¹⁸O_cc-based temperature record close to the δ²H-based one. This suggests that the low temperatures around 8.5 ka are due to winter-biased compositions, whereas the high temperatures between 8.25–8.1 ka are due to summer-biased values. This hypothesis may also explain the low δ²H peak at ca. 8.5 ka assuming an elevated amount of winter precipitation, as well as the absence of a strong negative δ²H peak at 8.2 ka, assuming increased summer precipitation and inclusion formation. High-resolution sampling was performed on half a stalagmite, and thin sections were produced from another half, precluding direct comparison with microscopic textures. Thus, the seasonality effect remains speculative. These considerations indicate that seasonal bias in inclusion formation is a potential explanation for the unrealistic formation temperatures calculated. These assumptions would also mean that, although the δ¹⁸O_fi values are diagenetically altered, the buffering capacity of the stalagmite calcite may help to preserve the δ¹⁸O_cc values and that biased formation-temperature data may not necessarily mean diagenetic alteration occurred.

The change of hydrological conditions around 8.5 ka is further suggested by the growth axis drift (assumed on the basis of inclusion trail shifts, Fig. 3) that, among other reasons, physical movement of the stalagmite or blockage of the drip water outflow, may indicate a change in drip-water flow from the feeding stalactite. However, the δ²H variations are not associated with systematic changes in the humidity-proxy δ¹³C and Sr records (Fig. 5), suggesting that precipitation variations do not indicate an annual decrease or increase, but rather a seasonal signal (i.e., an elevated amount of winter precipitation around 8.5 ka).

8.5 ka climate change event or local anomaly?

As we have shown, the V-03 stalagmite demonstrates a strongly negative δ²H peak at 8.5 ± 0.1 ka, which is associated with a less-pronounced δ¹⁸O_cc peak at 8.53 ka (Fig. 6). It is tempting to associate these isotope changes with the Laurentide Ice Sheet collapse and consequent freshwater discharge to the North Atlantic Ocean at 8420 ± 80 years (Tegzes et al., 2014), which may have induced climate change processes on the European continent. These climate change effects may be indicated by the productivity and therefore the temperature and hydroclimate indicator TOC% of chlorines in the annually resolved record of Eifel maar lakes (Sirocko et al., 2021), the record of which shows a sharp rise at 8.52 ± 0.12 ka. Stable hydrogen isotope data that may be directly compared with the V-03 δ²H record are scarce. Fluid inclusion data are available for speleothems from the Père Noël Cave (Belgium; Allan et al., 2018) and the Milandre Cave (Switzerland; Affolter et al., 2019), while cave ice data are available from the Scărișoara Cave (Perșoiu et al., 2017) (Fig. 6). Although the δ²H change is minor in the Milandre record, it shows a broad negative peak between 8.75 ± 0.02 ka and 7.9 ± 0.04 ka, with a δ²H change of −7‰ (Affolter et al., 2019). The Père Noël Cave record contains a sudden δ²H shift from about −24‰ to −39‰ at 8.6 ka, and a slow increase to −33‰ at 7.9 ka (with an average error of ± 0.04 ka for the 7.9–8.6 ka section). The Scărișoara record has distinct negative δ²H peaks at ca. 8.6 ka, 8.4 ka, and 8.25 ka (Fig. 6) (with an average error of ± 0.34 ka for the 8.2–8.6 ka period).

To investigate if this presumed 8.5 ka climate change event appears in European speleothem records, the δ¹⁸O_cc data of the SISAL database (Atsawawaranunt et al., 2018) were gathered, and the relative δ¹⁸O_cc changes at 8.5 ka were plotted on a map (Fig. 9A). The distribution of the isotope shifts does not reveal a systematic pattern, apart from the fact that similar δ¹⁸O changes occur in the Vacska, Limnon (Greece), and Jeita (Lebanon) caves during the 8.6–8.4 ka period. This areal distribution suggests that the “8.5 ka event” (Demény et al., 2022) is only a local or regional phenomenon that may be indirectly related to the North Atlantic freshwater discharge but is a short-term climate anomaly.

Figure 9. δ¹⁸O_cc changes (in ‰) at (A) 8.5 ka and (B) 8.2 ka relative to the preceding 100 years of data in the individual records. The records used for the maps (shown in Supplementary Fig. 5) were gathered from the SISAL database (Atsawawaranunt et al., 2018), excepting the Kaite cave record (Domínguez-Villar et al., 2017) and the Vacska-03 record (this study). The numbers before cave names are SISAL entity codes. (11) Cheng et al. (2015). (53) and (66): Daley et al. (2013). (58): Fohlmeister et al. (2013). (100): Boch et al. (2009). (117): Fohlmeister et al. (2012). (123): Luetscher et al. (2011). (129): Frisia et al. (2006). (135) and (215): Ait Brahim et al. (2019). (137): Baker et al. (2017). (141): Fleitmann et al. (2009). (215): Ait Brahim et al. (2019). (261): Rossi et al. (2018) and Kilhavn et al. (2022). (262): Breitenbach et al. (2019). (292): Peckover et al. (2019).

Temperature and precipitation changes during the 8.2 ka event

The δ¹⁸O_cc values depend on both formation temperature and water composition. Surface temperatures are generally positively correlated with δ¹⁸O_w values (about +0.6 ‰/°C; Demény et al., 2021b), while formation temperature increase causes a −0.23‰/°C change in speleothems (Tremaine et al., 2011). As a result, the net δ¹⁸O_cc change with temperature is ~0.37‰/°C. The average δ¹⁸O_cc value for the 8.25–8.5 ka period is −9.0 ± 0.4‰, and the lowest δ¹⁸O_cc value at 8.18 ka is −9.8‰. The −0.8‰ shift corresponds to a 2.1°C cooling, provided that the recent δ¹⁸O-temperature relationship is valid. It should be noted, however, that this temperature change is an estimated maximum, because it was calculated using the lowest δ¹⁸O_cc value. Within the negative δ¹⁸O_cc peak, the δ¹³C values first increase to −8.6‰ at 8.23 ka and then decrease to −10.2‰ at 8.18 ka when the δ¹⁸O_cc record reaches its minimum. This observation suggests that the first part of the 8.2 ka event is associated with strong fluctuations in soil activity. The Sr contents are low (Fig. 5), suggesting either dilution, and therefore a higher amount of precipitation, or decreased weathering in the soil zone. The δ²H values are not shifted in a negative direction, indicating an elevated relative contribution of summer precipitation. In summary, the first part of the 8.2 ka event period is characterized by general cooling, a relatively elevated precipitation amount in the summer, and fluctuating soil activity, which are unfavorable climate conditions for vegetation. The strong δ¹³C fluctuation and the decreased soil-zone weathering indicate changing dry/wet conditions during the cooling period. In the 40–70 mm dft (ca. 8.38–8.1 ka) section of the V-03 stalagmite, the U-Th age model would fit a ¹⁴C-based age model with >8% dcp, whereas the other parts of the U-Th-based StalAge model curve would fall between the 5% and 8% dcp ¹⁴C model curves (Fig. 4). The elevated dead carbon contribution would be consistent with positive δ¹³C shifts and generally lower Sr concentrations in the 40–70 mm dft section in case of soil activity decrease (i.e., less biogenic carbon contribution) and reduced weathering in the soil zone due to cooling and/or drying, but with transient humid conditions (water-filled karstic system precluding exchange with atmospheric CO₂; Noronha et al., 2014), again suggesting fluctuating precipitation amounts.

Just before the hiatus at 48.0 mm (corresponding to a model age of 8.15 ka), the δ¹³C and δ¹⁸O_cc values rise markedly, indicating dry conditions, before finally resulting in a growth stop. This short period may be followed by a sudden increase in precipitation amount, likely with a decrease in pH levels (i.e., dissolution surface on the stalagmite) due to elevated biogenic CO₂ dissolution. The increase in precipitation amount, as well as the more favorable soil conditions are indicated by the generally low δ¹³C values after 8.15 ka and the slightly elevated Sr concentrations, the latter related to enhanced weathering in the soil zone.

Similar temperature/precipitation changes have been inferred in lake deposits (Pál et al., 2016), as well as in speleothems and alluvial fan sediments (Peckover et al., 2019) in Southeast Europe. Pál et al. (2016) investigated seasonal temperature and precipitation amount changes using pollen concentrations, biogenic silica percentage determinations, and diatom distributions in lake sediments in the Retezat Mountains (Southern Romania). Microcharcoal pieces, Poaceae, Corylus, Quercus, Ulmus, and Picea pollen show peaks at 8.3–8.25 ka (with an error of about ± 0.1 ka), followed by a sharp peak of Carpinus betulus and planktonic diatom taxa at 8.15 ka These pieces of information would collectively suggest that the 8.2 ka cooling was associated with alternation of dry/warm summers, and then precipitation amount and winter temperature increased (Pál et al., 2016). Similar temperature and precipitation changes are inferred from the combined analyses of a speleothem (growth rate, δ¹³C, δ¹⁸O, and trace element concentrations) and alluvial fan sediments in Greece by Peckover et al. (2019). The combined evaluation of these proxy records led Peckover et al. (2019) to conclude that the 8.2 ka event was cold and dry, with “flashier” summer rainfall, and a humid phase after 8.1 ka. These precipitation and temperature changes are in agreement with the coupled δ²H–δ¹³C–δ¹⁸O_cc–Sr changes in the V-03 stalagmite results reported in this paper, as well as the growth stop and short-term dissolution at 8.15 ka.

Moisture transport changes reflected by d-excess data

Moisture transport trajectories may be inferred from d-excess values (d = δ²H−8⋅δ¹⁸O) because moisture originating from the Atlantic Ocean has generally lower d-excess values than moisture that comes from the Mediterranean Sea (Gat, 1980; Gat and Carmi, 1987). The δ¹⁸O_w data can be calculated from the δ¹⁸O_cc values, the estimated formation temperatures, and the calcite-water oxygen isotope fractionation in Johnston et al. (2013), after which the d-excess values may be computed from the measured δ²H and calculated δ¹⁸O_w values. Formation temperatures were estimated to be 8–10°C (recent cave temperature is 8.8°C). The obtained d-excess values range from 3.8–23.0‰, with remarkably low and high values around 8.5 ka and 8.2 ka respectively (Fig. 8B). Assuming the same seasonal δ¹⁸O_cc bias as discussed previously, the range would become 6.7–15.3‰, which is similar to the range of 9–14‰ obtained from recent precipitation monitoring data (Czuppon et al., 2018). These calculations indicate that the d-excess values would fit the observed range when using the hypothetical summer precipitation bias.

The uncorrected d-excess values seem unrealistic, so the δ¹⁸O_cc correction must be assumed to arrive at a reasonable range. Nevertheless, the calculations may indicate that the 8.6–8.4 ka period is dominated by low d-excess values, whereas the 8.3–8.1 ka period is characterized by elevated d-excess values. The d-excess record of the Scărișoara cave ice (Perșoiu et al., 2017) also indicates elevated values at 8.2 ka (Fig. 8C), which were attributed to an increased contribution of Mediterranean moisture. It is important to note that low d-excess peaks appear at ca. 8.6 ka and 8.4 ka in the Scărișoara record, which may be correlated with the low d-excess values of the V-03 stalagmite in this study.

Elevated d-excess values and the inferred Mediterranean moisture contribution are indirect indicators of moisture-transport changes. Moisture trajectory shifts and varying degrees of Mediterranean moisture contribution have been inferred for the Katerloch Cave (Austrian Alps), where speleothem records spanning the 8.2 ka event have been gathered (Boch et al., 2009). Parker and Harrison (2022) assumed that global transfer of the low δ¹⁸O signal associated with the 8.2 ka event was due to the southward shift of the Intertropical Convergence Zone (ITCZ), which affected climate conditions all over the world. The southward movement of the ITCZ would also would have induced a north-south displacement of moisture transport from the northern part of the North Atlantic Ocean to the southern regions. Although model calculations indicate a weakening of the low δ¹⁸O signal below 40°N latitude (Wiersma et al., 2011), the speleothem δ¹⁸O records show negative δ¹⁸O shifts, even in Morocco, until the negative δ¹⁸O signal increasingly diminishes towards the East (Fig. 9B). The drawbacks in model calculations also appear in the duration of the 8.2 ka event. Wiersma et al. (2011) estimated it to last around 60 years, whereas the speleothem records generally suggest a longer duration of about 160 years.

Other than the southward shift of moisture transport and increased Mediterranean contribution to the precipitation in East-Central Europe (Perșoiu et al., 2017, and this study), we also suggest that the amount of precipitation changed seasonally, with a relatively increased summer contribution. This is in accordance with the positive δ¹⁸O changes in the eastern part of the Mediterranean, where the oxygen isotope composition of rainwater, and therefore of speleothems, is primarily governed by precipitation amount, as shown by the compilation of regional speleothem records in Kern et al. (2019).

Conclusion

A stalagmite was collected in northern Hungary from a carefully selected site (Vacska Cave), where ventilation was negligible compared to other parts of the cave system, and that was monitored for two years for physical cave parameters and drip water compositions. Fourteen U-Th age dates suggest that the stalagmite covers the period of 10.6–8.0 ka (relative to AD 1950). In this study, the 8–9 ka section was thoroughly investigated using petrographic observations, AMS ¹⁴C measurements, XRF-based Sr concentration measurements, and stable H, C, and O isotope analyses of stalagmite calcite and inclusion-hosted water. A δ¹⁸O_cc peak with an approximately −0.8‰ shift between 8.25–8.1 ka was observed, which may be related to the 8.2 ka event. Additionally, negative δ¹⁸O_cc and δ²H peaks appear at ca. 8.5 ka Using a compilation of European and eastern Mediterranean speleothem data from the SISAL database (Atsawawaranunt et al., 2018), the “8.5 ka event” was actually determined to be a local or regional climate anomaly that may have had indirect connections with North Atlantic freshwater discharge.

Stable oxygen isotope compositions of inclusion-hosted water were evaluated for a paleoclimate interpretation, but their correlation with the carbonate crystallinity indicator FWHM of the main X-ray diffraction peak instead indicated that diagenetic alteration occurred, which superseded the primary climate-related information. The petrographic and geochemical observations for the 8.25–8.1 ka period revealed a series of temperature and precipitation changes during the 8.2 ka event. After the event's peak conditions (strong cooling and drying), the local climate became humid and warm, with a sudden and sharp change at 8.15 ka. The 8.2 ka event peak was associated with relatively increased d-excess values, indicating increased Mediterranean moisture contribution. This change in the origin of the moisture is in agreement with the southward displacement of heat and moisture-transport trajectories during the 8.2 ka event, as inferred by previous studies and shown by the regional distribution of 8.2 ka δ¹⁸O peaks of speleothem records gathered from the SISAL database.