Early Last Interglacial environmental changes recorded by speleothems from Katerloch (south‐east Austria)

In the European Alps, the Last Interglacial (LIG, ~129–116 ka) has been primarily studied using pollen preserved in mires and lake sediments. These records document the vegetation succession across the LIG, but are poorly constrained chronologically. Here, we present a precisely dated stable isotope record for the early LIG (129.6 ± 0.4 to 125.0 ± 0.8 ka) based on two stalagmites from Katerloch, a cave located on the south‐eastern side of the Alps. The onset of the interglacial is marked by a sharp rise in the oxygen isotope values at 129.4 ± 0.4 ka, consistent within dating uncertainty with the timing of Termination II as recorded by other Alpine speleothems. Carbon isotope values show an equally prominent drop at Termination II and the establishment of a forest ecosystem. Although concentrations are low, pollen from these stalagmites provide insights into how the local vegetation changed across the first five millennia of the LIG. The spectra indicate a vegetation optimum recorded by the occurrence of warm‐demanding taxa typical of the Eemian mesocratic phase. By combining stable isotopes and pollen data, we propose an absolutely dated chronological framework for peri‐Alpine pollen successions from lacustrine sediments covering the first half of the LIG.


INTRODUCTION
The Last Interglacial (LIG) was the time period about~129-116 thousand years (ka) before present when global temperatures were as high as or even higher than in the current Interglacial (Kukla et al., 2002;Otto-Bliesner et al., 2013, greenhouse gas concentrations were higher than during pre-industrial times (e.g. CO 2 up to 285 p.p.m.; Landais et al., 2013), sea level was higher (~about 6-9 m above present sea level) and highlatitude ice sheets were smaller (Dutton et al., 2015;Rovere et al., 2016). Today, the climate is being overprinted by Anthropocene warming, and the modern global temperature is now approaching the warmth of the LIG (Bova et al., 2021). Although the LIG is not a perfect model for the Holocene, mainly due to a different orbital configuration (Yin and Berger, 2015), there is a growing interest in the paleoclimate community to study and document this most recent interglacial before the Holocene as a 'test bed' for climate model projections in a purely natural climate forcing scenario (Lunt et al., 2013;Fischer et al., 2018;Otto-Bliesner et al., 2021). Mountain regions are ideal sites to investigate the magnitude and impact of climate warming because there are clear indications that the warming rate is amplified with elevation (Pepin et al., 2015). Examples of this include the alarming retreat rate of Alpine glaciers and thawing permafrost (e.g., Zekollari et al., 2019;Sommer et al., 2020). In addition, warming of mountain regions has been shown to result in an acceleration of biodiversity change in Alpine ecosystems (Steinbauer et al., 2018).
In the foreland of the European Alps, the LIG has been studied using pollen and plant macro remains preserved in mires and lake sediments (e.g. Woillard, 1979;Drescher-Schneider, 2000;Müller et al., 2005;Pini et al., 2010). These records provide the most detailed and comprehensive record of vegetation changes during the LIG, and include attempts to quantify temperature (Guiot et al., 1989;Kühl and Litt, 2007;Brewer et al., 2008). These studies suggest rather stable temperatures during the LIG with an early optimum followed by a slight cooling. Some of these data were compared to paleoclimate model output (Kaspar, 2005) indicating higher temperatures than today and a W-E gradient in winter temperature with increasing anomalies towards eastern Europe. Although well documented, the timing of these vegetation changes is poorly constrained chronologically (Govin et al., 2015). These chronologies are either unconstrained by radiometric dating (Drescher-Schneider, 2000) or based on correlation [e.g. with orbitally tuned marine records (Müller et al., 2005) or regional speleothem data (Sirocko et al., 2005)]. One of the very few direct age constraints is a 230 Th date (128 ± 4 ka) from a peat layer in lacustrine sediments from Füramoos in Bavaria, but no analytical data were reported in this study (Müller et al., 2003). Varved, i.e. annually laminated, sediments spanning the LIG are rare, and only the record from Lago Grande di Monticchio in central Italy (Brauer et al., 2007) seems to cover the full duration of the interglacial. A revised chronology using tephra layers resulted in a mean age uncertainty of 5% (Wulf et al., 2012). Little data are available on non-biological proxies from LIG lake sediments in Europe. Bulk carbonate samples from lake Mondsee in Austria record a two-step increase in δ 18 O at the onset of the LIG followed by rather constant values during the course of the LIG, but the chronology is based on pollen biostratigraphy only (Drescher-Schneider and Papesch, 1998).
Speleothems offer superior age control and can contain sufficient pollen to provide information on the vegetation outside the cave (McGarry and Caseldine, 2004;Dredge et al., 2013;Festi et al., 2016). An increasing number of speleothem records spanning (part of) the LIG have has been reported from the Alpine realm (Spötl et al., 2002(Spötl et al., , 2007Meyer et al., 2008;Häuselmann et al., 2015;Moseley et al., 2015;Johnston et al., 2021) and their environmental interpretation is largely based on stable isotope proxies. More recent speleothem studies from this mountain range attempted to quantify temperature variations during the LIG (Johnston et al., 2018;Wilcox et al., 2020), while only one study looked into pollen preserved in stalagmites to reconstruct the paleo-vegetation in the Western Alps (Luetscher et al., 2021). In this paper, we present stable isotope data along with a pollen record from two speleothems overlapping in age from a cave from the south-eastern fringe of the Alps. Constrained by 230 Th dates, the isotope record is first compared to Alpine speleothems and then with other European records. Subsequently, both the isotope and pollen record are compared to peri-Alpine lake sediments lacking radiometric age to provide, for the first time, a chronology for their respective pollen zones during the first half of the LIG.

Site description
Katerloch Cave is located at the south-eastern fringe of the Alps, in the Austrian province of Styria (Fig. 1). This region is also known as the 'Styrian Karst Province', and the cave host rock is a Devonian limestone (Schöckelkalk). The cave opens at an altitude of 901 m a.s.l. (47.25316°N, 15.55064°E) 20 km north of Graz. The cave developed in a forested ridge dominated by spruce [Picea abies (L.) H. Karst], beech (Fagus sylvatica L.), maple (Acer pseudoplatanus L.), ash (Fraxinus excelsior L.), whitebeam [Sorbus aria (L.) Crantz] and hornbeam (Carpinus betulus L.) (Pratl, 1971). The cave follows the general dip of the host limestone and comprises a series of halls and narrow connections in between. The large entrance hall is followed by a 50-m shaft (Eulenschacht) which connects to the underlying Marteldom. The entrance area is also now connected via a narrow artificial shaft to the speleothem-rich chambers: the Phantasiehalle (where our two stalagmites were retrieved), Zauberreich and the Halle der Einsamkeit. Adjacent to the Zauberreich is the deepest known section of the cave, Seeparadies. The explored maximal vertical distance from the  (1) and other locations mentioned in the text: Lake Mondsee (2), Samerberg (3), Eurach (4), Schneckenloch and Hölloch caves (10 km apart) (5), Melchsee-Frutt and Sieben Hengste cave systems (35 km apart) (6), Cesare Battisti (7), Bigonda cave (8), Azzano Decimo (9), Lake Fimon (10), Baradla cave (11), Corchia cave and Tana Che Urla (5 km apart) (12) and deep-sea core MD01-2444 from the Iberian Margin (13). (C) Google Earth view of the location of Katerloch on a south-facing forested ridge. [Color figure can be viewed at wileyonlinelibrary.com] cave entrance to the deepest point is 135 m and the total length is slightly over 1 km (Boch et al., 2009).
In terms of climate the cave site receives Atlantic moisture from the west and north-west and is also under the influence of Mediterranean air masses from the south (which are most pronounced during spring and autumn, including local summer thunderstorms). During winter, the North Atlantic Oscillation influences the regional climate (Boch et al., 2009). There is an intermittent snow cover at the site between winter and early spring. The mean annual air temperature measured at the cave entrance is 8°C (2006 to 2008) and~4°C in Phantasiehalle (Boch et al., 2011), reflecting a cold-trap behavior of the cave which shows a descending geometry lacking a lower entrance.
Katerloch cave is a well-studied and well-monitored cave (Boch, 2008;Boch et al., 2009Boch et al., , 2011. The cave's micrometeorology has been modified to some extent because two short tunnels were blasted during show cave development in the 1950s. This probably led to an intensification of the cave ventilation by enhanced winter cooling due to the descending geometry of the cave. A comparison of stable isotope data of modern, late (i.e. pre-show cave) and early Holocene calcite suggests, however, that the effect of the modified cave micrometeorology on the stable isotope composition of calcite is minor. Late Holocene (0.6-2.5 ka) specimens show mean δ 18 O and δ 13 C of −6.1 and −7.5‰, respectively (Boch, 2008), i.e. the C isotope values of post-1950s calcite is actually lower than that of pre-show cave calcite, arguing against enhanced CO 2 degassing and concomitant kinetic isotope fractionation in recent decades. Early Holocene (7-10 ka BP) speleothems yielded mean δ 18 O and δ 13 C values of −6.3 and −7.9‰ (stalagmite K1) and −6.3 and −8.5‰ (stalagmite K3), respectively (Boch et al., 2009), similar to those of the late Holocene samples.

Sampling and petrography
Stalagmites K2 and K4 were found broken in the same cave room (Phantasiehalle). Their top parts are missing. The stalagmites were cut in half with a diamond blade saw, polished and scanned (Fig. 2). Thin sections were cut along the growth axes of the stalagmites, polished and analyzed using a Nikon Eclipse polarizing microscope.

Th dating
Nine (K4) and eight (K2) samples were hand-drilled along the speleothem growth axis for 230 Th dating (Table 1). Between 120 and 180 mg of calcite was used given the low U concentrations (between 80 and 140 p.p.b. for K4 and between 60 and 100 p.p.b. for K2). The samples were prepared following the chemical procedure described by Edwards et al. (1987). The measurements were performed on a Thermo Fisher Neptune Plus MC-ICP-MS at the Xi'An Jiatong University in China, using the technique described by Cheng et al. (2013).

Stable isotopes
Subsamples were microdrilled along the extension axis of both stalagmites at a resolution of 5 mm using a handheld drilling device to obtain ca. 0.3 mg of calcite. Between 250 and 278 stable isotope measurements (SI) were carried out along the growth axis of each stalagmite at a resolution of 5 and 2.5 mm where growth rate slowed down in one section of K2. The isotope analyses were performed using a Delta V Plus isotope ratio mass spectrometer linked to a Gasbench II (both Thermo Fisher, Bremen, Germany) (Spötl, 2011) at the University of Innsbruck. Calibration of the instrument was accomplished using international reference materials and the results are reported relative to VPDB. Long-term precision on the 1-sigma level is 0.06 and 0.08‰ for δ 13 C and δ 18 O, respectively.

Pollen analyses
A total of 14 calcite samples from both stalagmites were processed for pollen, whereby the sample weight ranged from 167 to 460 g per sample for a total of about 4.6 kg of calcite. Pollen and microfossil extraction followed a protocol developed for calcite samples with low pollen content, which allows us to avoid decantation steps preventing the potential loss of microfossils. This was achieved using a combination of filtration and evaporation steps as well as avoiding acetolysis. Pollen extraction included the following steps: (1) weighing, (2) cleaning using HCl (10%; 3-5 s), (3) cleaning with double distilled water, (4) drying at 40°C, (5) weighing, (6) adding HCl (10%) to dissolve carbonates, (7) adding double distilled water (5-10× the amount of HCl), (8) filtering using a 7-μm filter, (9) transferring the content of the filter into a sample tube, (10) adding a few drops of ethanol (96%), (11) adding one drop of glycerol and (12) evaporation at 95°C. To control eventual contamination sources, a blank consisting of double distilled water was added to every batch of samples prepared. Pollen samples were mounted in glycerol, stained with fuchsine and the complete content of microfossils was analyzed. Pollen identification was performed by transmitted-light microscopy at magnifications of 400× and 600× using standard identification keys (Moore et al., 1991;Faegri et al., 1992;Beug, 2004) and a pollen atlas (Reille, 1992). Where the identification of morphological features was not possible, the pollen grains were classified as 'indeterminata'. Pollen counts were plotted using the C2 software (Juggins, 2007)

Petrography
The fabric of the two stalagmites consists of coarsely crystalline columnar calcite. Macroscopic lamination is noticeable, consisting of white, porous laminae alternating with translucent and more compact laminae. The same alternation is also observed microscopically in thin sections. No petrographic evidence of hiatuses was observed.

230
Th dating and age model Uranium concentrations of both stalagmites are low, but the detrital 232 Th contamination is also low, resulting in relative age uncertainties of 0.37-0.57% (Table 1).
OxCal and a Poisson-process deposition model (Ramsey and Lee, 2013) was used to establish depth-age relationships for both stalagmites (Supporting Information Figs. S1 and S2). The average growth rate is 0.45 mm a −1 for K4 and regular along the whole record, while the average growth rate for K2 is 0.42 mm a −1 , but with one slower growth section between 127.6 ± 0.5 and 126.2 ± 0.5 ka. The recorded growth intervals of both stalagmites are relatively short: K2 (102 cm) formed from ca. 125.0 ± 0.7 to 128.6 ± 0.5 ka, while K4 (126 cm) covers the period from ca. 126.8 ± 0.5 to 129.6 ± 0.4 ka.

Stable isotopes
Stalagmite K4 started growing 129.6 ± 0.4 ka ago and records the onset of the LIG. The record begins with values of −10‰ for δ 18 O and −6‰ for δ 13 C (Fig. 3). Shortly after, there is a rapid rise in δ 18 O of about 2.5‰ centered at 129.4 ± 0.4 ka and a slightly delayed drop in δ 13 C of 4‰. A small drop in δ 18 O (by~0.5‰) occurs between 128.9 and 128.6 ka, followed by a gradual increase towards a first maximum at 128.4 ka. The remainder of the δ 18 O record shows an overall slightly increasing trend with superimposed small-scale variations of <~0.5‰ (Fig. 4). The δ 13 C values depict no long-term trend. The isotope record ends at 126.7 ± 0.5 ka, and the uppermost part of stalagmite K4 is missing.
Stalagmite K2 grew between 128.6 ± 0.5 and 125.0 ± 0.7 ka. During this time period, carbon isotope values are stable, in agreement with those of stalagmite K4, and lack a long-term trend (Fig. 3). δ 18 O values increase gradually, also consistent with stalagmite K4.

Robustness and significance of the stable isotope record
The oxygen isotopic composition of drip waters and calcite precipitates in caves is controlled by a combination of several factors. These include the oceanic moisture source (icevolume effect, sea-surface temperature), the trajectories of the air masses, the altitude of cloud condensation, evapotranspiration in the catchment and the temperature in the cave (Rozanski et al., 1992;McDermott, 2004;Lachniet, 2009 (Sodemann and Zubler, 2010) and also reflect the cave air temperature (Boch et al., 2009). The overall oxygen stable isotope pattern from Katerloch stalagmites (K2 and K4) is therefore similar to that of speleothems from other parts of the Eastern (Moseley et al., 2015), Central (Meyer et al., 2008) and Western Alps (Wilcox et al., 2020;Luetscher et al., 2021), which also receive predominantly Atlantic-derived moisture, and where δ 18 O primarily reflects atmospheric temperature.
During winter, the North Atlantic Oscillation (NAO) controls the temperature and precipitation variability in the southeastern Alpine region (Casty et al., 2005;Efthymiadis et al., 2007). Because of its southerly location, Katerloch also receives moisture from the Western Mediterranean Sea which results in slightly enriched δ 18 O values of dripwater compared to sites on the northern side of the Alps. The average interglacial δ 18 O values of the Katerloch speleothems are, however, comparable to those of speleothems in Cesare Battisti cave and Bigonda cave in the Italian Alps (Johnston et al., 2018(Johnston et al., , 2021 and Baradla cave in north-eastern Hungary (Demény et al., 2017). Boch et al. (2011Boch et al. ( , 2009 performed Hendy tests on calcite from the top of actively growing stalagmites as well as on calcite precipitated on glass plates. They observed a distinct enrichment in 13 C with increasing distance from the central axis, suggesting some kinetic fractionation. By contrast, the oxygen isotope values were rather stable and a comparison of the measured isotope values and the expected (equilibrium) values, using the temperature-dependent fractionation factor of Friedman and O'Neil (1977), supports year-round modern calcite precipitation close to O isotopic equilibrium (Boch, 2008). Although subject to kinetic fractionation, the main control on large carbon isotopic variation in Katerloch speleothems is soil bioproductivity.
The mean stable isotope values for the LIG (excluding the initial rise) and for the time interval where the two records (K2 and K4) overlap (from 126.8 to 128.6 ka) are the same for stalagmites K2 and K4 (−7.6‰ for δ 18 O and −9.7‰ for δ 13 C). Both δ 18 O and δ 13 C values of the LIG samples are lower than those of present-day and Holocene samples by about 1.3-1.5 and 0.8-2.2‰, respectively. Speleothems from Cesare Battisti Table 2. Pollen samples arranged according to their distance from the top (DFT) of the stalagmites, with their respective top and bottom ages, the duration covered by each sample according to the mean growth rate (with 2σ confidence interval) and the pollen concentration and flux.

LIG Alpine speleothem records
Lower δ 18 O values in Katerloch south-east of the Alps highlight a different control of the O isotopic composition than for sites north of the Alps (Fig. 5). In the latter caves, LIG values are generally higher than those of the Holocene, reaching highest values during the climate optimum in the first half of the LIG. Apart from the local altitude effect (Sieben Hengste~1700 m a.s.l.; Melchsee-Frutt~2000 m a.s.l., Schneckenloch-Hölloch 1200 m a.s.l.) and the correction for the ice-volume effect (Duplessy et al., 2007) the most likely explanation for the lower δ 18 O values at Katerloch during intervals warmer than the Holocene is a higher input of Atlantic-derived moisture.
Katerloch stalagmite K4 precisely captured Termination II at 129.4 ± 0.4 ka with the climate transition lasting only about 100 years (according to the age model). The timing of Termination II is in good agreement with other precisely dated Alpine speleothems (Fig. 5). A stalagmite from Melchsee-Frutt in central Switzerland recorded it at 129.6 ± 0.8 ka (Wilcox et al., 2020), and a stalagmite from the nearby Sieben Hengste cave system at 129.7 ± 0.8 ka (Luetscher et al., 2021), consistent within uncertainty with the Katerloch chronology, while stalagmites from Schneckenloch and Hölloch caves in western Austria suggest a slightly older age of 130.9 ± 0.9 and 130.7 ± 0.9 ka, respectively (Moseley et al., 2015). This difference could be partly explained by a larger detrital Th content in both the Schneckenloch and Hölloch speleothems, increasing the dating uncertainty and slightly shifting these ages towards older values. The magnitude of the oxygen isotope rise at Termination II is comparable between the five sites (~2‰). Following the termination there is a small drop in δ 18 O (−0.5‰) from 128.8 ± 0.4 until 128.4 ± 0.5 ka, which could correspond to a cooling event also identified in the Schneckenloch and Hölloch records between 129.1 ± 0.6 and 128.5 ± 0.5 ka (−0.5‰) and attributed to a cold event because of meltwater input into the North Atlantic (Moseley et al., 2015). In the Melchsee-Frutt record, a later intra-LIG cooling event was identified between 125.8 ± 0.5 and 124.6 ± 1.0 ka based on fluid inclusion and stable isotope data (Wilcox et al., 2020). The Katerloch record lacks evidence of a clear-cut cooling event but, given that the record ends at 125.4 ± 0.4 ka, it might have not been recorded.
With respect to the carbon isotopic composition, the most prominent feature is the massive decrease in δ 13 C (by 4‰), reflecting a rather rapid (~250 years) recovery of the vegetation and associated soil development after the end of the penultimate glacial period. For the remainder of the LIG the flat δ 13 C values suggest a stable forest ecosystem at the study site. The Sieben Hengste stalagmite recorded an even more abrupt drop in δ 13 C lagging Termination II by~90 years (Luetscher et al., 2021). In contrast to Katerloch, this Western Alpine speleothem never reached negative δ 13 C values, documenting a strong rock-buffering of the C isotope signal or less soil/vegetation activity due to the high elevation. This is also the case for the stalagmites from the nearby Melchsee-Frutt cave system, which even lack a clear expression of early LIG vegetation signal and show a gradual δ 13 C decrease from 132 ka across the entire LIG (Wilcox et al., 2020). A stalagmite from Schneckenloch does record a marked drop in δ 13 C, but the age model suggests an earlier timing (~132 ka; Moseley et al., 2015). No distinct signal is recorded by the coeval stalagmite from the nearby Hölloch cave, underscoring the strong site-specific variability of the stable C isotope signal even across major climate shifts.
Lower δ 13 C values of Katerloch speleothems during the LIG are attributed to milder (and/or shorter) winters and/or a denser vegetation and/or a higher soil bioproductivity, consistent with the general climate interpretation of the LIG in the greater Alpine realm.

Katerloch and other European records
The major temperature rise associated with Termination II in our record is synchronous with the sea-surface temperature increase on the Iberian Margin (Martrat et al., 2007;Tzedakis et al., 2018). After the termination, an Atlantic cold event at 128.5 ka, identified as C28 (Tzedakis et al., 2018), might have been associated with a small negative excursion in δ 18 O (−0.5‰) recorded in stalagmite K4 (Fig. 6). This cold event is also observed in Melchsee-Frutt (Wilcox et al., 2020) and Corchia (Tzedakis et al., 2018). While the structure of Termination II in Katerloch is comparable to other Alpine speleothems records, it is different from Corchia cave speleothems (Drysdale et al., 2009;Tzedakis et al., 2018) which show a more gradual transition (Fig. 6) as changes in δ 18 O from this record primarily reflect variations in rainfall amount. The nearby Tana Che Urla cave reveals a similar structure of Termination II within age error. These two central Italian speleothems have recorded two events of reduced moisture at ca. 129.6 and 126.0 ka, corresponding within error to prior growth of stalagmite K4 for the first one and slower growth rate in K2 for the second. After 128.4 ± 0.5 ka the Katerloch record shows a gradual warming trend until at least 125.4 ± 0.4 ka (when the record ends), consistent with other central European speleothems (e.g. Meyer et al., 2008;Demény et al., 2017;Pawlak et al., 2021) and pollen records (Sirocko et al., 2007;Brewer et al., 2008). Except for the first part of Termination II, the record from Bigonda cave in the Italian Prealps is very similar to Katerloch (Fig. 6), in terms of both structure and mean δ 18 O isotope values. With comparably low δ 18 O values during early LIG, both records corroborate the hypothesis of a northward shift in the ITCZ (Johnston et al., 2021) resulting in moisture of lower δ 18 O values from the Atlantic and a smaller influence of moisture derived from the Mediterranean and the Adriatic Sea.

Reliability of the pollen data
Pollen counts obtained from the Katerloch speleothems are lower than those obtained from sediments (e.g. lakes), but nevertheless yielded pollen spectra typical of the end of the penultimate glaciation and the beginning of the LIG, with a pollen assemblage dominated by Pinus sylvestris/mugo-type, Quercus, Ulmus, Tilia and the peculiar exceptional abundance of Hedera (Fig. 7). The pollen concentration is low in all Katerloch samples compared to the only other currently available Alpine speleothem LIG record from Sieben Hengste Cave (1-20 grains g −1 ; Luetscher et al., 2021), and the Holocene speleothem record from Milchbach cave (4-564 grains g −1 ; Festi et al., 2016), also located in Switzerland. However, the palynomorph flux into the cave is slightly higher in Katerloch than in Sieben Hengste, reaching up to 73-82 pollen grains ka -1 cm −2 . Given that the growth rate of the Katerloch stalagmites is significantly higher than that of the stalagmite from Sieben Hengste cave (~0.49 vs.~0.007 mm a −1 ) rather large sample sizes could be used to obtain enough pollen grains without compromising the temporal resolution. Despite the large amount of calcite dissolved, some samples yielded pollen sums that are unsuitable for a reliable calculation of percentage values of pollen taxa. To partially overcome this issue we merged samples into macro-samples to obtain, where possible, a pollen sum of about 100 pollen grains. To this end, we merged adjacent samples from the same speleothem and obtained six macro-samples ( Fig. 7; Supporting Information Table S1). Three macro-samples are still below the threshold of 100 pollen grains and these percentage values must be interpreted with caution. This applies in particular to macro-sample K4/0-59 which has a pollen sum of only 46. Despite this limitation, the spectra of these six macro-samples provide important and well-dated insights into the vegetation response during the first part of the LIG.

Early LIG vegetation response at the south-eastern fringe of the Alps
Pre-interglacial vegetation (before 129.4 ± 0.4 ka) was characterized by a forest-steppe including P. sylvestris/mugo, P. cembra and Betula as well as shrubs and xerophytic elements such as Artemisia, confirming that the area was not glaciated. The presence of the vegetation in this phase is well reflected by the low carbon isotope values, which are lower compared to speleothems from other caves in the Alps (Fig. 5). These other caves are located at higher altitudes and their catchments were affected by glaciations during the penultimate glacial maximum. This cold climate is also recorded in two pollen records located south of the Alps, Lake Fimon and Azzano Decimo, by their respective pollen zones FPD10 and AZ54 (Table 3). At Fimon, the vegetation in this period was characterized by the dominance of herb pollen, xerophytes and Pinus, while at Azzano Decimo an open pine forest including P. cembra (arolla pine) was present (Pini et al., 2009(Pini et al., , 2010. North of the Alps, the pre-interglacial phase correlates with LPAZ MO1 of Lake Mondsee and DA1 of Samerberg and Eurach which are  (Moseley et al., 2015;Wilcox et al., 2020;Luetscher et al., 2021), (B) speleothems from Hungary (Demény et al., 2017) and NE Italy (Johnston et al., 2018(Johnston et al., , 2021, (C) from Katerloch (this study), and (D) from Corchia Cave (Tzedakis et al., 2018) and Tana Che Urla (Regattieri et al., 2014). (E) Percentage of temperate (Mediterranean and Eurosiberian) pollen from deep-sea core MD01-2444 from the Iberian Margin, and (F) sea-surface temperatures (Uk'37) from the same core updated by Tzedakis et al. (2018). The red bar indicates the warming phase during Termination II, the blue bar meltwater event C28 and the green bar the LIG climax vegetation, with the highest pollen concentrations identified in core MD2444 from the Iberian Margin and in the Katerloch speleothems. [Color figure can be viewed at wileyonlinelibrary.com] characterized by a treeless vegetation (Beug, 1979;Grüger, 1979;Drescher-Schneider and Papesch, 1998).
A possible rise in temperatures starting at 129.4 ± 0.4 ka (Termination II) is indicated by a shift in the stable isotopic composition and in the pollen record by the expansion of pine-dominated forests with the occurrence of warmdemanding elements starting with Corylus and Hedera. The occurrence of ivy suggests its local presence near the cave (and/or at the cave entrance) already in this period, and points to mild winters (mean temperature of the coldest month above −2°C) as indicated by the modern ecological requirements (Iversen, 1944). North of the Alps at Mondsee, this period corresponds to a reforestation phase that started with the spread of Juniperus (juniper) followed by Pinus in LPAZ MO2, whose upper limit saw a steep rise in isotope values of the bulk carbonate (Drescher-Schneider and Papesch, 1998). Also in Samerberg and Eurach this phase corresponds to the afforestation phase with Pinus, Betula and Juniperus (DA2). South of the Alps, in the Lake Fimon record, this phase is represented by the expansion of pine forests (PAZ FPD11a), which was correlated with the first increase in the oxygen isotope values in a stalagmite from Corchia cave at 132.5 ± 2.5 ka (Drysdale et al., 2009).
In both stalagmites K2 and K4, samples with the largest pollen concentration and influx cover the same period between 128.6 and 128.0 ka. This corresponds to the early growth of stalagmite K2 and a slight increase in δ 18 O values of stalagmite K4. This period is considered as the LIG climax based on data from Sieben Hengste (Luetscher et al., 2021) as well as the highest percentage of temperate (Mediterranean + Eurosiberian) pollen in deep-sea core MD01-2444 from the Iberian Margin (Fig. 6). The Katerloch pollen record suggests a relatively rapid vegetation response to the temperature increase registered by the δ 18 O signal, reflecting a quick expansion of warm-demanding deciduous trees (Quercus, Ulmus, Tilia) with exceptionally abundant Hedera. In contrast, the palynological succession of the Sieben Hengste stalagmite and its δ 18 O record reveal an up to 3-ka-long lag between the  Table S1 for details on sample merging. Only main pollen types and spores are presented. In the main diagram: in green AP = arboreal pollen (trees and shrubs); in yellow NAP = non-arboreal-pollen (herbs). The width of the sample bars corresponds to the timespan covered by each sample. [Color figure can be viewed at wileyonlinelibrary.com] Table 3. Proposed palynological correlation between Katerloch samples and chronology with pollen zones/diagram phases of coeval Eastern Alpine sedimentary records located north (Eurach, Samerberg, Mondsee) and south of the Alps (Lake Fimon, Azzano Decimo). Chronological overlap is because samples originate from two speleothems and the macro samples presented here have been merged according to their speleothem of origin (see text and Supporting Information Table S1).  -Schneider and Papesch, 1998;Drescher-Schneider, 2000). In general, at Katerloch this phase corresponds to LPAZ MO 3 and 4. However, in pollen diagrams from Gondiswil, Eurach and Samerberg, the Ulmus and Quercus peaks coincide as in Katerloch (Beug, 1979;Grüger, 1979;Zagwijn, 1996). In Lake Fimon, south of the Alps, the onset of the LIG coincided with a sharp rise in humid temperate taxa (Quercus and Corylus; PAZ FFD11b), dated at 132.5 ± 2.5 ka according to the main shift of δ 18 O at Corchia cave (Pini et al., 2010). At Azzano Decimo the onset of the LIG occurred within PAZ AZ55, but the lower limit of this zone is represented by a hiatus.
Finally, the timing of the highest pollen concentration and flux in the Katerloch samples (128.6-128.0 ka) is synchronous with the highest abundance of temperate tree pollen retrieved from core MD2444 on the Iberian Margin (Tzedakis et al., 2018). The higher pollen flux at Katerloch in this period probably reflects an increase in pollen productivity driven by the warm climate that allowed the development of a dense forest vegetation.
As already mentioned, the abundance of Hedera pollen is one of the distinctive features of the Katerloch record (Fig. 7). This pollen type is never dominant in sedimentary records but it is recognized as a distinctive feature of LIG vegetation spectra. Considering the exceptional abundance of Hedera in both Katerloch and Sieben Hengste speleothem pollen records, this pollen might be a typical feature of pollen records from LIG speleothems in the Alpine region. The first occurrence of ivy in Sieben Hengste cave is recorded around 127 ka, and the highest values around 125 ka (Luetscher et al., 2021). In Katerloch, the ivy maximum occurred at 128.4 ± 0.5 ka and led the Sieben Hengste record by~2 ka. This is again probably related to a difference in altitude, and to the fact that Sieben Hengste is located north of the Alps and in an area that was glaciated during the penultimate glacial maximum. Irrespective of the timing, ivy is clearly better represented in speleothems than in lake sediments, suggesting that this taxon was an important part of the LIG vegetation.

CONCLUSIONS
The Katerloch record provides a replicated, precisely dated, high-resolution stable isotope record of the beginning and the warmest phase of the LIG in the Alps.
Termination II recorded by stalagmite K4 is marked by a rapid rise of oxygen isotope values (2‰) precisely dated at 129.4 ± 0.4 ka and, with a brief delay, the carbon isotope values decrease significantly (4‰) reflecting vegetation regrowth. The timing of Termination II is coherent with other precisely dated speleothems from the Alps.
After a short cooling event (128.8 ± 0.4 to 128.4 ± 0.5 ka) the oxygen isotope data show a gradual warming trend until 126.7 ± 0.5 ka when the K4 record ends. Stalagmite K2 grew simultaneously from 128.6 ± 0.5 to 126.7 ± 0.5 ka, and during this period the mean isotope values of both carbon and oxygen are the same for the two records. The oxygen values from K2 slowly increase until 125.4 ± 0.4 ka (end of the K2 record) and the carbon isotope values are rather stable, lacking a long-term trend.
The comparably low oxygen stable isotope values of Katerloch suggest reduced advection of Mediterranean moisture to the southern fringe of the Alps and probably stronger westerlies during the LIG compared to the Holocene. This observation is supported by other speleothem records from the Southern Alps (Johnston et al., 2018(Johnston et al., , 2021.
Pollen spectra extracted from the Katerloch speleothems capture the vegetation optimum between 128.6 ± 0.5 and 128.0 ± 0.4 ka by the occurrence of warm-demanding taxa typical of the Eemian mesocratic phase, such as Quercus, Ulmus, Tilia, as well as Ilex and Hedera.
The carbon isotope data and the pollen spectra record a vegetation response delay to rising temperatures during Termination II within about 250 years (soil bioproductivity) and 1000 years (arrival of warm-demanding taxa), respectively.
Although the pollen concentration is low in these speleothems, useful information on vegetation changes can still be extracted, yet this study illustrates the challenge of finding speleothems that contain sufficient pollen and are clean enough to allow precise dating. Fast growing speleothems are an advantage given that more material can be sampled without compromising the temporal resolution. Our findings highlight that more speleothems should be tested for pollen presence and these data should be integrated in the vegetation reconstruction based on pollen records from sediments. These will provide a wider perspective on the past vegetation, allowing the acquisition of better information on the occurrence and importance of taxa that are poorly represented in the sediments, e.g. Hedera. Clearly records with higher pollen concentrations are critically required to provide an absolutely dated chronological framework spanning the entire LIG in the greater Alpine realm.

Supporting information
Additional supporting information can be found in the online version of this article.This article includes online-only Supplemental Data. Figure S1. Time-depth models for stalagmite K2 from Katerloch produced using OxCal 4.3. The age model was calculated using the P-sequence with a k-value of 0.1 cm −1 , an interpolation rate of 1 cm −1 , and k varying between 10 −2 and 10 2 . Figure S2. Time-depth models for stalagmite K4 from Katerloch produced using OxCal 4.3. The age model was calculated using the P-sequence with a k-value of 0.1 cm −1 , an interpolation rate of 1 cm −1 , and k varying between 10 −2 and 10 2 . Table S1. Merged pollen samples for the percentage pollen diagram, with their respective top and bottom ages, the duration covered by each sample according to the mean growth rate (with 2σ confidence interval) and the pollen concentration and flux. †Based on minimum growth rate of 0.1 mm a −1 . Appendix S1. Stable isotope data for stalagmite K2 and K4, with their respective age model and pollen counts for K2 and K4.
Xue for running additional 230 Th ages, Tanguy Racine for his help during fieldwork, Gabriella Koltai for her help in the 230 Th laboratory, and Ronny Boch, Ruth Drescher-Schneider, Roberta Pini and Erwan Messager for discussion. Comments by Kale Sniderman and an anonymous reviewer helped to improve the clarity of the paper.

Data Availability Statement
The stable isotope data, the U/Th data and the pollen counts that support the findings of this study are available in the Supporting Information (Appendix S1).
Conflict of interest-The authors declare that there are no conflicts of interest.