The entire JS9 stalagmite profile was analysed by a Nikon Eclipse LV100POL microscope, including microfabric structures, the appearance of calcite crystals, potential discontinuities and porosity. The analysis of speleothem microfabrics and microfabric log construction was based on the methodology proposed by Frisia (2015). The microscopic analyses were performed at the Institute of Geological Sciences at the Polish Academy of Sciences (Warsaw, Poland).

A total of 10 calcite samples (0.1–0.5 g) were collected by drilling the JS9 speleothem through its growth axes. The samples were drilled as thin as possible and had an average thickness of 2.5 ± 0.2 mm. The chemical preparation of the samples was performed at the Uranium-Series Laboratory of the Institute of Geological Sciences, Polish Academy of Sciences (Warsaw, Poland). In order to control the efficiency of the chemical procedure, a spike (233U, 236U and 229Th mixture) was initially added to the samples. In the first step of the chemical procedure, the samples were heated to decompose potential organic matter. The samples were then dissolved in nitric acid. Finally, uranium and thorium were separated from the solution by chromatographic methods using TRU Resin (Hellstrom, 2003). In addition to the regular samples, the internal standards and blank samples were processed by the same procedure. U and Th isotopic compositions of all the samples and standards were measured at the Institute of Geology of the Czech Academy of Sciences public research institution (Prague, Czech Republic) by a double-focusing sector-field inductively coupled plasma (ICP) mass analyser (Element 2, Thermo Finnigan MAT). The spectrometer settings were at a low mass resolution (m/Δm ≥ 300).

The obtained measurements were corrected for background counts and chemical blanks. The final results were reported as the activity ratios (Table 1). The final U-series ages were calculated using the latest decay constants (in yr-1): λ238 = (1.55125 ± 0.0017) × 10-10 (Jaffey et al., 1971), λ234 = (2.826 ± 0.0056) × 10-6 (Cheng et al., 2013), λ232 = (4.95 ± 0.035) × 10-11 (Holden, 1990) and λ230 = (9.1577 ± 0.028) × 10-6 (Cheng et al., 2013). The reported age errors were estimated using error propagation rules. All measurement errors were considered except for the decay constant.

A modified version of the Hellstrom algorithm was applied to correct the obtained ages (Hellstrom, 2006), with the assumption of initial contamination by 230Th, 234U and 238U isotopes. The algorithm searches for the lowest values of initial contamination by 230Th, 234U and 238U isotopes from detrital sources, which were able to correct series of ages in stratigraphic order (Błaszczyk et al., 2021). The age–depth model was calculated using the MOD-AGE algorithm (Hercman and Pawlak, 2012).

The samples for stable isotopic composition measurement were drilled by a Micro Mill with a 0.1 mm drill bit diameter. The final number of obtained samples was 290. In the first stage, the JS9 speleothem was sampled along its growth axis at a resolution of one sample per millimetre. The lower part of the stalagmite (0 to 40 mm) was additionally sampled at a resolution of one sample every 0.3 mm to minimise the difference in resolution between the lower and upper parts of the record caused by differences in growth rate. The isotopic composition of O and C was measured by a Thermo Kiel IV carbonate device connected to a Finnigan Delta^plus isotope-ratio mass spectrometer (IRMS) in dual-inlet mode. The results were normalised to three international standards, NBS 19, NBS 18 and IAEA CO 8, and were reported relative to the Vienna Pee Dee Belemnite (VPDB) international standard. The analytical precision (1σ) was better than 0.03 ‰ and 0.08 ‰ for δ13C and δ18O respectively. The reproducibility was checked by measurement of two internal standards after every 12 samples (± 0.03 ‰ for δ13C; ± 0.08 ‰ for δ18O). The analyses were performed in the Stable Isotope Laboratory (Institute of Geological Sciences, Polish Academy of Sciences) in Warsaw.

The trace element content was analysed from thin sections by an analyte excite excimer laser ablation (LA) system with a wavelength of 193 nm connected to an Element 2 ICP mass spectrometer (ICP-MS; Thermo Finnigan); a fluence of 2.44 Jcm-2 was obtained at 50 % laser power with 10 Hz pulses. The width of each line was 50 µm, and the laser speed at each scan was 5 µms-1. Additional details of the LA-ICP-MS analytical procedure followed the procedure described by Eggins et al. (1997). Measurements of near-surface trace element contents, namely, Mg, Sr, Na, Ba, P Si, Fe and Mn, were performed at medium resolution. The obtained raw data were normalised to Ca. Finally, the data were smoothed by the adjected averaging method using 10 nearby data points.

The results of petrographic studies are presented in Fig. 2. The JS9 sample is a 155 mm long columnar stalagmite with a diameter of 80 mm. Macroscopically, the JS9 stalagmite is composed of laminated calcite (Fig. 2a). The colour of the laminae changes from light cream to dark brown (Fig. 2a). Between 75 and 85 mm, the stalagmite has a grey colour. The light cream laminae between 40 and 75 mm have a zone of macroscopically visible porosity in the axial part of the stalagmite. Microscopic analysis of the appearance of calcite crystals and the identification of textural features show that most of the observed stalagmite consist of columnar polycrystals with length-to-width ratios that are usually > 10:1. “Fibre-like” calcite individuals compose each polycrystal. The overall appearance of this layer resembles spherulite, which is composed of bundles of elongated crystals bent outward (Fig. 2b, c). The polycrystals show brush extinction that converges away from the substrate when the rotating table is turned clockwise. The characteristics mentioned above are similar to those described by Frisia (2015); these characteristics indicate that the columnar radiaxial fibrous (Crf) texture is a dominant fabric in the JS9 stalagmite (Fig. 2b). Parts of stalagmites containing Crf fabric are usually separated by thin layers of small calcite crystals and detrital material – those layers are dark in cross-polarised light (Fig. 2c). The appearance of these thin layers indicates the micrite (M) fabric described by Frisia (2015). Micrite fabric is most abundant in the middle part and in the youngest layers of the JS9 stalagmite (Fig. 2c).

The results of 10 U-series dates are presented in Table 1. The reported errors are 2σ and vary from 0.8 % to 2.8 %. The analysed samples do not show any visible detrital contamination at the dissolution stage. However, four of the measured samples have a 230Th/232Th ratio lower than 300. In the case of measurement by mass spectrometry, these samples should be considered contaminated by detrital admixtures (Hellstrom, 2006). Therefore, the entire profile was corrected using a modified procedure proposed by Hellstrom (2006). The result of correction shows that the corrected ages are within the error range of the uncorrected ages (Table 1).

Based on the results of U-series dating, an age–depth model was constructed for the JS9 stalagmite (Fig. 3). According to the obtained age–depth model, deposition of the JS9 stalagmite started at 142 ± 4 ka and ended at 84 ± 3 ka. The growth rate of the JS9 stalagmite was not uniform. From 142 to 112 ka, its growth rate was relatively slow at 1.4 mmkyr-1; after 112 ka, it rapidly grew at a rate of 11.5 mmkyr-1, which ended at 109 ka, and then the growth rate slowed to 1.9 mmkyr-1. The last intensive change in the growth rate occurred after 95 ka when it increased to 4.2 mmkyr-1.

The obtained isotopic records (Fig. 4a, b) cover the interval from the late MIS 6 to MIS 5a. The mean value of the δ18O record is -7.05 ‰ (Fig. 4a), and its value varies in a range from -9 ‰ to -5.7 ‰. The δ18O signal expresses short time changes, and the average value of its amplitude is ca. 0.8 ‰ (Fig. 4a). At the end of MIS 6, an increase in δ18O values is interrupted by a 1.2 ‰ instant decline. The next important change occurs at the boundary between MIS 5d and MIS 5c – an episode of elevated δ18O values greater than -6 ‰. Finally, a short episode occurs at the boundary between MIS 5c and MIS 5b, where δ18O values decrease to -9 ‰ (Fig. 4a). The δ13C record expresses changes in its values from -1 ‰ to -9.8 ‰ (Fig. 4b). The average amplitude of δ13C values for changes over a short period is close to 1 ‰. In contrast to the δ18O record, the δ13C curve is dominated by episodes of lower and higher values. They are divided by large-scale shifts (Fig. 4b). From 143 to 139 ka, δ13C values rise to -1 ‰. From 139 to 130 ka (an interval of 9 ka), δ13C values decrease from -2 ‰ to -7 ‰ (5 ‰), and the values are low ca. -8 ‰ until 110 ka. From 110 to 107 ka, δ13C values increase from -9.3 ‰ to -2.6 ‰ (6.7 ‰) and decrease from -1 ‰ to -8.5 ‰ (7 ‰) at 101 ka. From 100 to 85 ka, δ13C values oscillate at approximately -8.2 ‰. After 85 ka, δ13C values increase to -5.6 ‰ (Fig. 4b).

The trace element content results are presented in Fig. 4c–j. Mg, Sr and Ba contents do not show clear correlations or anticorrelations (Fig. 4c–e). However, a few single Mg extremes are clearly in phase or out of phase with the Sr content. For example, at 138 ka, they are in phase, and at 122.5 ka, they are out of phase (Fig. 4c, d). Generally, Mg and Ba records are similar from 138 to 101 ka, whereas Sr and Ba contents show more similarities before 138 ka and after 101 ka. Mg and Sr contents have minimums at 101 ka, whereas the minimum Ba content occurs at 93 ka and is mimicked by a decreasing Sr content during the same period.

Na, P, Fe and Mn contents show a similar pattern (Fig. 4f–i), and they have three intervals of increased values: before 138 ka, from 106 to 98 ka and after 92.5 ka. From 98 to 92 ka, Na, P, Fe and Mn contents are lower. This trend is also similar for Ba and Sr contents, and only the Mg content has a different pattern in this interval. The Fe content has the largest number of peaks (Fig. 4h), several of which are also shown by Na, P and Mn contents (Fig. 4f, g, and i respectively). In comparison to those four records, the P content has the lowest peak amplitude (Fig. 4g).

The Si content shows a few different patterns. The most visible pattern is a short maximum at 102 ka, which transitions into a minimum almost immediately at 101 ka, but the amplitude of changes in the Si content is relatively low. However, from 122 to 102 ka, it has an increasing trend, and a similar trend can be observed for Na and P contents (Fig. 4f, g, j). Similar to the Fe, Mn, P, Na and Ba contents, the Si content has higher values before 138 ka and after 98 ka.

The meteorological data collected in Slovakia show that there is a significant dependency between the δ18O value of atmospheric precipitation and the mean annual temperature (R2 = 0.73; Holko et al., 2012). The influence of the annual amount of precipitation on the δ18O value is less obvious (R2 = 0.48; Holko et al., 2012). In the DCS region, the temperature gradient for the δ18O of precipitation is 0.36 ‰∘C-1 (Holko et al., 2012). The current δ18O precipitation value changes during the year from -16 ‰ in February to -6 ‰ in July (Holko et al., 2012). Currently, nearly 65 % of precipitation occurs during the spring and summer months (April–September). Therefore, the seepage water is biased by the seasonal effect, and its mean δ18O value is higher than expected based on the mean annual temperature.

Recently, a meta-analysis of cave drip water and precipitation monitoring records showed that the isotopic composition of dripping water is not affected by evaporation in climates with mean annual temperatures lower than +15 ∘C and aridity indices higher than 0.65 (Baker et al., 2019). The current climate of the study region fits the above-mentioned mean annual temperature and aridity index conditions. Therefore, current interglacial conditions are more conducive for δ18O signals to reflect the regional temperature conditions and isotopic composition of meteoric water. In contrast, during glacial episodes, when the local climate was more continental and the aridity index was lower, the studied δ18O record could have theoretically been affected by the evaporation effect. However, due to the low mean annual temperature, the effect should not be high.

The main factor shaping the δ18O value of western and central European speleothems during the Holocene and last interglacial period was the temperature (Moseley et al., 2015; Kern et al., 2019; Comas-Bru et al., 2020; Hercman et al., 2020). For example, stalagmites from Cobre Cave, which is located on the northern coast of Spain (Fig. 5), reflect changes in the oceanic moisture isotopic composition, which is dependent on temperature (Fig. 6; Rossi et al., 2014). Similarly, the δ18O record from Han-sur-Lesse Cave in Belgium (Fig. 6) is driven by temperature and by changes in the isotopic composition of the ocean surface (Vansteenberge et al., 2016). The main trend of the δ18O record from the Hungarian stalagmite collected in Baradla Cave (Fig. 5) reflects temperature changes (Demény, 2017).

According to the facts listed above, it can be assumed that, in the long term, the temperature effect on atmospheric precipitation is the main factor shaping the δ18O value of the DCS stalagmites.

However, in the short term, the δ18O value can be additionally affected by humidity and aridity effects, as described in the case of the Bourgeois–Delaunay Cave (Couchoud et al., 2009).

The whole temperature effect on the δ18O value of the DCS speleothems calcite is lower than the presently observed δ18O temperature gradient of local precipitation (0.36 ‰∘C-1; Holko et al., 2012). It is caused by the opposite temperature effect on calcite crystallisation (-0.18 ‰∘C-1; Tremaine et al., 2011). The expected net temperature gradient is ca 0.18 ‰∘C-1.

The δ13C value of speleothem calcite depends on the proportion of CO2 from a soil source and from a host rock source. CO2 from a soil source can reflect the changes in the isotopic composition of atmospheric CO2 and is also enriched in 12C due to biological activity. Due to that fact, a well-developed soil cover results in a lower δ13C value. The level of soil development depends on climatic conditions such as temperature and humidity. Currently, the vegetation cover over the DCS is dominated by mixed forests, including mountain-type forests and grasslands connected with mountain slope activity (Hercman et al., 2020). During the cold episodes of the last glacial period, the Slovakian landscape was dominated by boreal trees, tundra dwarf shrubs and grasslands (Feurdean et al., 2014; Jankovska et al., 2002). The beginning of the Holocene was marked by the development of temperate-type forests (Feurdean et al., 2014). Recently, a set of five Holocene speleothems from different parts of the DCS was investigated (Hercman et al., 2020). The mean δ13C value varied from ca. -8 ‰ to ca. -2 ‰ and was site-dependent (Hercman et al., 2020). Despite these values, the shape of all Holocene δ13C records from the DCS reflects the changes from boreal and tundra vegetation types to temperate and mountain forest types. This change suggests that the δ13C proxy in the DCS can be interpreted in terms of vegetation changes.

Trace elements such as Mg, Sr and Ba are transported in water solution. Their relative abundance depends on the time of water residence and on the host rock composition (Fairchild and Treble, 2009). Drier conditions result in longer water residence times. The process of dolomite rock dissolution is slower than the process of limestone rock dissolution. Therefore, during longer water residence times, the contribution of trace elements from dolomite host rock sources increases. Dolomite normally contains less Sr and Ba than calcite, resulting in higher Mg/Ca and lower Sr/Ca and Ba/Ca ratios during drier conditions (Roberts et al., 1998; Tremaine and Froelich, 2013; Rossi et al., 2014). The DCS is developed mostly in Gutenstein limestones and Ramsau dolomites (Gaál, 2016). Therefore, the dissolution of both limestones and dolomites is possible in the DCS. On the other hand, prior calcite precipitation (PCP) can also occur during dry episodes when the water residence time is longer. However, PCP results in an increase in all X/Ca ratios because Ca2+ cations are preferred during calcite crystallisation (Tremaine and Froelich, 2013). Episodes of synchronous increases in Mg/Ca and Sr/Ca ratios are not observed in the record in this study.

Elements, such as Fe, Mn and Si, may be transported as detrital particles or submicron-size colloids (Fairchild and Treble, 2009). Additionally, all elements that can be incorporated into the calcite structure can be transported as absorbed ions on the clay mineral structure. During drier periods under higher aeolian supply conditions, particles can be transported into the cave environment without water transportation (Hu et al., 2005).

Recently, a set of five speleothems of Holocene age from the DCS was investigated (Hercman et al., 2020). Holocene δ18O records from DCS speleothems reflect the same pattern (Hercman et al., 2020). Additionally, the mean value for all of these records is similar. This proves that the δ18O value from DCS speleothems reflects the final regional climatic conditions. Comparing Holocene δ18O values with the record in this study can be useful for its interpretation. A total of 95 % percent of Holocene δ18O values from the DCS are in the range from -7.6 % to -6.8 % with a mean value of -7.2 ‰ (Fig. 4a). There are five periods in which the δ18O value of the JS9 stalagmite was slightly different than the δ18O mean value of the Holocene: 143–135, 127–123, 116–113, 108–101 and 94–92 ka (Fig. 4a).

The δ13C and δ18O values in the JS9 stalagmite change dynamically during the 143–130 ka period. The short episode of elevated δ18O and δ13C values from 143 to 137 ka may be the result of drier conditions with a higher level of evaporation at the end of MIS 6 (Gascoyne, 1992; Genty et al., 2006, Couchoud et al., 2009). During the 143–137 ka time interval, Mg and Ba contents are elevated, whereas the Sr content is lower (Fig. 4c–e). Trace element contents (Mg, Ba and Sr) support this interpretation. From 143 to 137 ka, the Fe, Mn and Si contents are elevated. This result may be related to the lack of developed soil cover and increased frost erosion of the surface above the cave. The other explanation of elevated values of listed proxies is the presence of micrite (Fig. 4k). However, this explanation also leads to dryer conditions (Frisia, 2015). New data revealing CO2 concentration changes during MIS 6 (Shin et al., 2020) show a local minimum CO2 concentration between 142 and 138 ka. Therefore, the episode recorded in the JS9 stalagmite may be global in nature.

After 137 ka, the change in the negative value observed in the δ13C record (ca. 5 ‰) is ca. 2 times larger than that observed in the DCS δ13C records at the beginning of the Holocene (ca. 2.5 ‰; Hercman et al., 2020). This may reflect the more substantial change in the environment from periglacial tundra conditions to temperate forests. According to data from other speleothems, the long-term tendencies of δ13C and δ18O are clearly related to improvements in thermal conditions after the MIS 6 glacial maximum and before the MIS 6–MIS 5e transition (Pawlak et al., 2020; Moseley et al., 2015; Meyer et al., 2008; Holzkamper et al., 2004).

Termination II (T II) in the JS9 stalagmite is highlighted by rapid decrease of ca. 1.2 ‰ in the δ18O value (Fig. 4a). This decrease is in contrast to the δ18O signal recorded in speleothems from the northern rim of the European Alps (NALPS; Fig. 6). The observed positive shift in Alpine δ18O records during T II is the result of two processes: the improvement of thermal conditions and the change from winter-dominated precipitation to summer-dominated precipitation (Moseley et al., 2015; Meyer et al., 2008; Holzkamper et al., 2004; Fig. 6). Similarly, in the caves on the northern slopes of the Tatra Mountains, T II is visible as a positive change in δ18O (Pawlak et al., 2020). However, in the case of the δ18O record from Magurska Cave (Tatra Mountains, Poland), the change towards positive values is preceded by an instant decrease of ca. 2 ‰ in its value (Fig. 6). The difference between the Low Tatras and caves located on the northern slopes of the Tatra Mountains (ca. 39 km towards the NE) is that the Tatra Mountains were an important climatic barrier for moisture at that time. In the case of the JS9 stalagmite, the δ13C and trace element contents did not show any signal that could be equivalent to the rapid 1.2 ‰ negative change in the δ18O record at 130 ka (Fig. 4). Therefore, the recorded 1.2 ‰ negative shift must be caused by factors that affect only the δ18O proxy. Additionally, the δ18O value after T II remains at the lower level until the end of MIS 5e. Its average value for MIS 5e (-7.6 ‰) is ca. 0.4 ‰ lower than the average δ18O value (-7.2 ‰) for Holocene speleothems of the DCS (Hercman et al., 2020). According to the present temperature gradient in Slovakia, the -0.4 ‰ difference could be interpreted as a lower mean temperature of MIS 5e of ca. 1 ∘C in comparison to the Holocene. However, the probability of this simple interpretation is low. Other reasons, such as changes from summer-dominated precipitation to winter-dominated precipitation or more humid conditions than at present, are more probable. Generally, in central Europe, the beginning of MIS 5e is related to changes from continental to more transient climates (Demény et al., 2017; Moseley et al., 2015). Therefore, the other reason for the observed shift may be the source effect and rapid increase in the depletion of 18O moisture from the Atlantic source. The negative shift in the δ18O record observed in JS9 speleothems is similar to the change observed in Mediterranean records (Antro del Corchia, Soreq, Peqiin; Fig. 6). This type of shift in Mediterranean records can be explained as a source effect when the change in the δ18O composition of speleothems is caused by the change in the δ18O value of the Mediterranean Sea surface, which is documented by marine cores, and the change in the proportion of moisture from Atlantic and Mediterranean sources (Bar-Matthews et al., 2003; Nehme et al., 2015). These are similar to records from Antro del Corchia (CC5 stalagmite; Drysdale et al., 2005), which has an approximately 2.5 ‰ decrease in its δ18O values during T II at ca. 130 ka (Fig. 6). In the case of the JS9 stalagmite, the observed 1.2 ‰ negative change is the result of a muted response. This response must have been caused by local or regional effects, which were stronger than the thermal effect at that time. The possible effect that may have caused the lower δ18O value is the circulation effect and changes from sources of precipitation, such as the Adriatic Sea or the Black Sea, to Atlantic sources and vapour recycling over the European continent (Drysdale et al., 2005). The instant change to negative values may have been caused by source and continental effects, which overcame the temperature effect at that time.

The MIS 5e interval in the JS9 stalagmite can be divided into two parts. The first part (127–123 ka) has lower values of δ18O and δ13C (Fig. 4a, b). The Mg content during this period is lower than its average value, with a local minimum at 124 ka (Fig. 4c). Usually, low δ13C and δ18O values can be interpreted as a sign of wetter and colder conditions (Gascoyne, 1992; Genty et al., 2006; Couchoud et al., 2009). However, on a global scale, the 127–123 ka period is a time of the highest sea level and warm temperature conditions (Goelzer et al., 2016). At that time, other European records show high sensitivity to changes in the amount of precipitation. For example, the δ18O record of a stalagmite from Bourgeois–Delaunay Cave (Couchoud et al., 2009) shows millennial variability with amplitudes lower than 1 ‰ (Fig. 6). The δ18O changes are reiterated by δ13C changes (Couchoud et al., 2009). The authors' interpretation considers the influence of the amount of precipitation as the main driver of δ18O changes. Based on the above-mentioned facts recorded in JS9 stalagmite, episodes of low δ18O and δ13C values can be interpreted as wet periods with more intensive vegetation.

The second part of MIS 5e (122–115 ka) has ca. 0.3 ‰ higher δ18O values and ca. 0.5 ‰ higher δ13C values. Globally, this period is characterised by a systematic deterioration of climatic conditions, a global mean temperature decrease of approximately 1 ∘C (Goelzer et al., 2016) and a sea level decrease of approximately 30 m (Grant et al., 2012). The NGRIP record shows a 6 ‰ decrease, which reflects the changes in thermal conditions (Rasmussen et al., 2014). At the local scale, the elevated δ13C value is the response to a decline in vegetation conditions. The increase in the δ18O value is caused by aridification. Evaporation overcomes the temperature effect, which should be negative. A similar episode containing high δ18O and δ13C values at approximately 119–117 ka is observed in the records from Baradla Cave and Magurska Cave; these records are interpreted as indicating an episode of dry continental climate (Demény et al., 2017; Pawlak et al., 2020). Records from Magurska Cave, located in the Tatra Mountains, and from Baradla Cave, located in Hungary (Fig. 6), are more similar to the JS9 record during the 122–115 ka period than during the older 127–123 ka period, which suggests that the climate of the region becomes more uniform at the end of MIS 5e.

During the 108–101 ka interval, the δ18O and δ13C values of the JS9 stalagmite were elevated, and their values were the highest in comparison to the whole recorded period. The mean value of δ18O for this interval was ca. 1.3 ‰ higher than the mean δ18O value for Holocene in the DCS (Fig. 4). Elevated stable isotope values are related to elevated Fe, Mn, P and Na contents. In the case of the JS9 stalagmite, the interpretation of the 108–101 ka period as a dry interval is probable and is in accordance with δ18O, δ13C and Mg proxies. Elevated values of geochemical proxies are related to micrite microfabrics (Fig. 4k). It appears plausible that the presence of micrite fabric (M) is indicative of bio-influenced processes, as micrite layers may be associated with shifts to more positive C isotope ratios (Kaźmierczak et al., 1996). In a sample from the Nullarbor, δ13C values shift from -10.5 ‰ to -4.0 ‰ in stromatolitic-like micrite (M) layers. This phenomenon was interpreted as a possible result of microbial colonisation of the speleothem surface during a dry period (Frisia, 2015). According to all of this information, the 108–101 ka interval in the JS9 stalagmite record can be interpreted as a stable dry continental climatic period.

In contrast, the global and regional situations at that time are different. After 110 ka, the decrease in the global ocean level ended. From 108 to 101 ka, the world ocean level became elevated to 20 m in comparison to the local minimum at 110 ka (Grant et al., 2012), which suggests that the global mean annual temperature increased. This trend is also in accordance with increased insolation at that time (Berger 1978). However, the global sea level was more unstable with changes reaching 10 m. Similarly, the NGRIP record has elevated δ18O values with a short decreasing disturbance at 105 ka, which also reflects climatic instability (Fig. 6; Rasmussen et al., 2014). The climate instability at 108–101 ka is also expressed by the δ18O record from Magurska Cave (Fig. 6) and by records from the northern rim of the Alps (Boch et al., 2011). In contrast, the JS9 record expresses specific local stable conditions during the 108–101 ka period. A possible explanation is that the DCS region was constantly influenced by the continental climate at that time, in contrast to the northern Tatra Mountains and the Alps. The ca. 2 ‰ instant decrease and the lowest δ18O value in the JS9 record at 94–95 ka are not connected with any significant change in the other measured proxies. However, this episode is expressed in the NALPS δ18O records (Fig. 6) as a 1 ‰ instant drop and stalagmite growth cessation and is correlated with Greenland Stadial 23 (Mosley et al., 2020).