Cryogenic cave carbonates in the Dolomites (Northern Italy): insights into Younger Dryas cooling and seasonal precipitation

. In the European Alps, the Younger Dryas (YD) was characterized by the last major glacier advance with equilibrium 10 line altitudes being ~220 to 290 m lower than during the Little Ice Age and also by the development of rock glaciers. Dating of these geomorphic features, however, is associated with substantial uncertainties leading to considerable ambiguities on the internal structure of this stadial, the most intensively studied one of the last glacial period. Here we provide robust physical evidence based on precise 230 Th-dated cryogenic cave carbonates (CCC) coupled with thermal modelling indicating that early YD winters were only moderately cold in the Southern Alps, challenging the commonly held view of extreme YD seasonality. 15 Our data argue for a negative temperature anomaly of ≤ 3°C in mean annual air temperature at the Allerød-YD transition in a mountain cave (Cioccherloch, 2274 m a.s.l.) in the Dolomites of northern Italy. Our data suggest that autumns and early winters in the early part of the YD were relatively snow-rich, resulting in a stable winter snow cover. The latter insulated the shallow subsurface in winter and allowed the cave interior to remain close to the freezing point (0°C) year-round, promoting CCC formation. The main phase of CCCs precipitation at ~12.2 ka BP coincides with the mid-YD transition recorded in other 20 archives across Europe. Based on thermal modelling we propose that CCC formation at ~12.2 ka BP was most likely associated with a slight warming of approximately +1°C in conjunction with drier autumns and


Introduction 25
The last glacial period in the Northern Hemisphere (from 119 to 11.7 thousand years (ka) BP; Rasmussen et al., 2014) was characterized by abrupt climate shifts from cold and commonly arid stadials to mild and more humid interstadials. The youngest of these stadials is known as the Younger Dryas (YD or Greenland Stadial (GS) 1, ~12.8 to 11.7 ka BP;Rasmussen et al., 2014) and was a period when Northern Hemisphere temperatures returned to near-glacial levels, interrupting the last Termination. 30 https://doi.org/10.5194/cp-2020-107 Preprint. Discussion started: 25 August 2020 c Author(s) 2020. CC BY 4.0 License.
The YD is among the most extensively studied periods in the late Quaternary due to the availability of high-resolution palaeoclimate records such as ice, marine and lacustrine sediment cores. Still, the forcing mechanism(s) for this cold episode remain debated (Alley, 2000;Baldini et al., 2018;Brauer et al., 2008;Broecker et al., 2010;Renssen et al., 2015). The most widely accepted model invokes a near-shutdown of the Atlantic Meridional Overturning Circulation (AMOC) as a result of catastrophic meltwater injection into the North Atlantic Ocean (e.g. Broecker et al., 1989), and the concomitant large-scale 35 reorganization of the westerlies due to extensive winter sea ice formation (e.g. Bakke et al., 2009;Brauer et al., 2008). The resulting southward displacement of the polar front and the westerlies led to cold, almost Siberian-like conditions in N and NW Europe during prolonged winters (e.g., Broecker, 2006). In a recent study using a European-wide compilation of plant indicator species Schenk et al. (2018), however, suggested that YD summers remained relatively warm despite the AMOC shutdown with temperature decreases of 4.3 o C in NW Europe and 0.3 o C in eastern Europe relative to the preceding Bølling-40 Allerød interstadial (Greenland Interstadial (GI) 1). Using climate model simulations, Schenk et al. (2018) attributed relatively warm summers to atmospheric blocking induced by the Fennoscandian Ice Sheet preventing the penetration of cold westerly air masses entering Europe during the short summers. In contrast, blocking was almost absent during YD winters (Renssen et al., 1996). Overall, proxy data suggest that the YD climate was dominated by high seasonality and continentality across Europe with large meridional summer temperature gradients (e.g. Heiri et al., 2014a). Changes in winter climate were likely 45 disproportionally larger (e.g. Broecker, 2006), but quantitative understanding of the YD climate remains heavily biased towards the summer given the scarcity of winter proxy data.
Speleothem-based palaeotemperature reconstructions from the Jura Mountains in northern Switzerland suggest a large drop of the mean annual air temperature (MAAT) of up to -10°C during the YD (Affolter et al., 2019;Ghadiri et al., 2018), while preliminary results of a similar study from a cave in western Austria suggest a much smaller difference (5.5°C) (Luetscher et 50 al., 2016). Rock glacier records from the SE Swiss Alps argue for an even smaller cooling of only up to 3-4°C (Frauenfelder et al., 2001).
More recently, evidence for a time-transgressive climate shift mid-way through the YD has been reported from lacustrine sediments (e.g. Bakke et al., 2009;Brauer et al., 2008;Schlolaut et al., 2017), speleothems (Baldini et al., 2015;Bartolomé et al., 2015;Rossi et al., 2018) and marine sediments in Europe (Naughton et al., 2019). This climate shift has been attributed to 55 a gradual northward movement of the polar front driven by the resumption of the AMOC and concomitant sea-ice retreat in the North Atlantic. The earliest indication of a climate shift during the mid-YD is recorded by a stalagmite from the Pyrenees, showing a gradual transition from dry to wet conditions starting at 12.45 ka BP (Bartolomé et al., 2015). It took about three hundred years for this shift to propagate to central and finally to northern Europe, where it is documented as a rapid change in certain proxy properties (Bakke et al., 2009;Lane et al., 2013). While many records from SW (Baldini et al., 2015;Bartolomé 60 et al., 2015;Naughton et al., 2019;Rossi et al., 2018) and N Europe indicate that the first half of the YD was colder and drier than the second one, biomarker data from lacustrine sediments of the Gemündener Maar in W Germany suggest the opposite trend (Hepp et al., 2019). https://doi.org/10.5194/cp-2020-107 Preprint. Discussion started: 25 August 2020 c Author(s) 2020. CC BY 4.0 License.
In the Alps, climate information about the YD has been traditionally derived from studies of lake sediments (e.g. Grafenstein et al., 1999;Heiri et al., 2014;Lauterbach et al., 2011) andglacier reconstructions (e.g. Ivy-Ochs et al., 2009;Kerschner et al., 65 2000;Kerschner and Ivy-Ochs, 2008;Moran et al., 2016), complemented by a few cave records (Luetscher et al., 2016;Wurth et al., 2004). Studies on the internal structure of the YD, however, are rare and compromised by poor dating resolution, limiting our understanding of the mechanism(s) of the proposed mid-YD climate shift. Alpine paleoglacier records suggest an early YD glacier maximum between about 13.5 and 12.0 ka BP attributed to a combination of low temperatures and enhanced precipitation differences between the northern, central and southern part of the Alps (Kerschner et al., 2016;Kerschner and 70 Ivy-Ochs, 2008). Equilibrium line altitude reconstructions show that the inner zone of the Alps received ~20-30% less precipitation than today, mostly due to a decrease in winter precipitation, while annual precipitation in the Southern Alps was probably similar to modern values (Kerschner et al., 2016). Paleoglaciers from the Southern Alps show evidence of a double response, whereby the outermost and innermost moraines stabilized at ~12.3±0.7 ka and before 11.2±0.8 ka BP, respectively (Baroni et al., 2017;Ivy-Ochs et al., 2009). 75 In this study we focus on the Dolomites in the Southern Alps and provide seasonally resolved insights into the climate of the YD from cave sediments anchored by a precise 230 Th chronology. Rather than examining stalagmites commonly used in speleothem-based palaeoclimate research, we utilize a rather novel speleothem variety which provides a uniquely robust temperature control: coarsely crystalline cryogenic cave carbonate (CCC for short). The leading genetic model envisages CCC formation under degrading permafrost conditions (e.g., Žák et al., 2018). Water ingresses into the cave when the seasonally 80 thawing active layer of permafrost intersects the ceiling of the cave chamber, while most of the chamber is still within the permafrost, resulting in cave ice formation. Further climate warming leads to progressive degradation of permafrost and the cave air temperature slowly rises to 0°C. Drip water creates meltwater pools in the cave ice bodies that freeze slowly triggering the precipitation of CCC. Regardless of the details of this model, the key point is that CCC form within perennial cave ice at temperatures very close to 0°C (Žák et al., 2018). 85 Using a strategically selected high-alpine cave whose paleothermal regime is assessed using heat-flow modelling, we use CCC data to argue against strong winter cooling during the early YD, provide evidence of a maximum of 1-2°C warming at the mid-YD transition, and show that autumns and winters became drier in the second half of the YD.

Study site
Cioccherloch is a single-entrance cave opening at 2245 m a.s.l. on the karst plateau of the Sennes region in the Dolomites. 90 Given that the early YD polar front was supposedly located south of the main crest of the Alps , this site in the Southern Alps is ideal to trace the northward migration of the polar front (Fig. 1). The cave has approximately 250 m of passages. The entrance shaft is 20 m deep and intersects a subhorizontal cave level. A firn and ice cone is present at the bottom of this shaft fed by winter snow sliding down the shaft. Separated by a narrow squeeze which was excavated by cavers, a separate branch of the cave 60 m long and up to 10 m high descends from this cave level to approximately 55 m below the surface. Near the lower end of this gallery CCC were discovered for the first time in the Dolomites (Fig. 1). The air temperature at the CCC site monitored over a 1-year-period averages 2.5°C with minimum values (2.2°C) in January to March and maximum values (2.7°C) in November. These data show that the cave chamber is in thermal equilibrium with the outside MAAT at this elevation, obtained from nearby weather stations at Rossalm (2340 m a.s.l.) and Piz la Ila (2050 m a.s.l.) located less than 10 km from the study site. The mean air temperature is 2.2°C and 3.3°C at Rossalm and Piz la Ila, respectively (2015-100 2018; data source: Hydrographisches Amt, Autonome Provinz Bozen -Südtirol). The majority of snowfall in the Dolomites occurs from January to April with average snow heights of 4.2 and 3.4 m at Rossalm (2012-2019) and Piz la Ila (1999-2014), respectively. Autumn to early winter (September to December) snowfall amounts to an average of 1.0 and 0.8 m at the two weather stations.

Field work
CCC occurrences were mapped and samples were collected from five distinct heaps labelled A to E (Suppl. Fig. 1). In addition, small in-situ stalagmites were taken from the same chamber. Cave air temperature was recorded on an hourly basis using a Hobo Temp Pro v2 logger (Onset) between August 2016 and July 2017.

Morphological characterization 110
CCC samples were cleaned in an ultrasonic bath prior to documentation and laboratory analyses. Individual morphologies were examined using a Keyence VHX-6000 digital microscope.

Stable isotope analyses
CCCs samples were analyzed for their stable oxygen and carbon isotope composition using isotope ratio mass spectrometry (Spötl and Vennemann, 2003). In addition, two larger CCC particles were cut in half and micromilled at 0.1 to 0.3 mm 115 resolution. The results are reported relative to the VPDB standard with a long-term precision better than ±0.08‰ (1σ) for both δ 13 C and δ 18 O.

230 Th dating
Seventeen individual CCC particles were selected for 230 Th dating. 15-20 mg of calcite was drilled using a handheld drill from 15 crystals in a laminar flow hood. Two skeletal CCC crystals were analyzed as a whole as they were too small for aliquots to 120 be drilled from them. Growth layers of a stalagmite (Cioc1) collected next to the CCC spots were drilled at three discrete horizons (2, 24 and 50 mm from the top) and prepared for analyses. https://doi.org/10.5194/cp-2020-107 Preprint. Discussion started: 25 August 2020 c Author(s) 2020. CC BY 4.0 License.
Ages were determined by measuring U and Th isotope ratios on a multi-collector inductively coupled mass spectrometer after their chemical separation following Edwards et al. (1987) and Cheng et al. (2013). Analyses were performed at Xian Jiaotong University (China). 2σ uncertainties for U and Th isotopic measurements include corrections for blanks, multiplier dark noise, 125 abundance sensitivity, and contents of the same nuclides in the spike solution. Decay constants for 230 Th and 234 U were reported by Cheng et al. (2013). Corrected 230 Th ages assume an initial 230 Th/ 232 Th atomic ratio of (4.4 ±2.2) ×10 -6 and 232 Th/ 238 U value of 3.8 as the value for material at secular equilibrium with the bulk earth. Final ages are given in years BP (before 1950 AD).

Thermal modelling
Heat conduction from the surface to 70 m depths was modelled using a 1d heat-flow model ( 130 https://zenodo.org/record/3982221). This model considers conductive heat transfer only and solves the heat equation utilizing finite differences as space discretizations alongside a forward Euler time-stepping scheme, stabilized with a diffusion-type Courant-Friedrichs-Lewy condition. The relationship between heat flow Q, thermal conductivity c, and geothermal gradient dT/dz is given by Eq. (1): The model assumes a homogenous host rock with no internal heat generation. Thermal diffusivity of the limestone was set to 1.2 x 10 -6 m 2 /s (Hanley et al., 1978) to account for some air-filled porosity. The ground heat flux (dT/dz) was set to 0, to account for the presence of the shaft in the entrance zone of the cave, allowing exchange with the ambient air in this upper part of the cave suppressing the minor effect of the geothermal heat flux. This assumption is supported by the agreement between the temperature at the end of the CCC-bearing cave gallery and the ambient MAAT. Modeling results are shown as MAAT 140 against depth below the surface.

CCC morphology
CCC occur as loose crystals and crystal aggregates in small heaps on and partly underneath five breakdown blocks (Suppl. Morphologies include amber-colored crystals and crystal aggregates of rhombic, raft, beak-like and split crystal habits (Suppl. confirming their precipitation in slowly freezing of pockets of water enclosed in cave ice. A beak-like crystal, 4.5 mm in diameter, revealed ≤1.3‰ and ≤1.2‰ intra-crystalline variability in δ 13 C and δ 18 O, respectively, while a 5 mm large rhombohedral crystal shows ≤0.4‰ and ≤0.2‰, respectively. Holocene stalagmites from the same chamber exhibit distinctly different stable isotope values with δ 13 C and δ 18 O values ranging from -7.6 to -2.0‰ and from -9.1 to -7.4‰, respectively 155 ( Fig. 2).

230 Th dating
The 238 U concentration of CCC samples varies from 0.8 to 2.2 ppm ( In contrast to CCCs, the 238 U concentration of stalagmite Cioc1 is much lower (~0.5 ppm). 230 Th/ 232 Th atomic ratios vary between 34×10 −6 and 1570×10 −6 ( Table 1). The resulting 230 Th ages demonstrate that stalagmite growth commenced during the Late Glacial at 14.98 ±0.14 ka BP. Petrographic analysis provides strong evidence for a growth interruption after which calcite deposition re-started in the mid-Holocene at 5.88 ±0.15 ka BP and continued until 1.32 ±0.27 BP ka.

Thermal modelling 170
We performed a series of model runs covering possible climate scenarios for the YD (Table 2) to explore the relationship between atmospheric temperature changes and the temperature 50 m below the surface at this high-elevation site. We define the thermal boundary conditions for CCC formation as -1 to 0°C ("CCC window" -see discussion) whereby the likelihood of CCC formation is highest between -0.5 and 0°C.
Paleotemperature estimates used in these computations are based on regional annual and summer air temperature 175 reconstructions. Three of the models (scenarios 2c, 2e and 3c) also consider the insolating effect of a winter snow cover, expressed as ∆Ts, i.e. the buffering of the winter temperature at the ground surface by the snow cover (following Zhang, 2005; Table 2).

Scenario 1 -Allerød interstadial
In this scenario we simulated an interstadial, similar to the 1000 year-long Allerød preceding the YD, in order to precondition 180 the thermal regime of the subsurface (Table 2). We assumed that no permafrost was present at the site at the start of the https://doi.org/10.5194/cp-2020-107 Preprint. Discussion started: 25 August 2020 c Author(s) 2020. CC BY 4.0 License.
interstadial (hence a ground temperature of 1°C at the surface). The experiment was forced with a MAAT 2°C lower than present day (i.e., ΔMAATModern-Allerød=-2°C) consistent with regional proxy data (e.g., Ilyashuk et al., 2009). After 1000 years a temperature of 0.5°C is reached at 50 m depth (Suppl. Fig. 2). The model output was used as the initial setup for all scenario 2 models. 185

Scenario 2 -Stadial conditions during the early YD
In the next five experiments (2a -2e) we explored the timing of perennial cave ice development and tested whether water pockets in cave ice could have experienced slow freezing during an early YD characterized by cold stadial conditions. As the mid-YD transition was determined at 12,240 ± 40 varve years BP at Meerfelder Maar, Germany , scenario 2 models were run for 610 years (i.e. from 12.85 to 12.24 ka BP). 190 Scenario 2a was forced with a ΔMAATModern-early YD of -9°C (i.e. ΔMAATAllerød-early YD = -7°C; -20°C in January; Table 2), consistent with speleothem-based paleotemperatures from northern Switzerland (Affolter et al., 2019;Ghadiri et al., 2018).
The results show that the atmospheric cooling rapidly propagates into the subsurface resulting in the development of permafrost down to 50 m depth in less than 50 years (Fig. 3a, Suppl. Fig. 3a).
In scenario 2b we applied a less dramatic atmospheric cooling characterized by ΔMAATModern-early YD of -5°C (i.e. 195 ΔMAATAllerød-early YD = -3°C; -13°C in January; Table 2). This cooling amplitude has been suggested by stalagmite fluid inclusion data from western Austria (Luetscher et al., 2016). Modelling results show that the cave 50 m below the surface cools to -1°C in about 100 years. Further cooling would lead to -2.0°C about 100 years later (Fig. 3b, Suppl. Fig. 3b).
The third stadial experiment (2c) investigates the influence of an autumn/early winter snow cover (Table 2) insulating the ground from cold air during winter, assuming a buffering of the winter temperatures by the snow cover of 5°C (i.e. ∆T of 200 5°C). The presence of a snow pack delays the cooling of the ground and results in temperatures at 50 m depth between 0.5 and -1°C within 150 years from the start of the atmospheric cooling (Fig. 3c, Suppl. Fig. 3c).
Scenarios 2d and 2e consider an even smaller drop in MAAT (ΔMAATModern-early YD = -4.5°C i.e. -12°C in January; Table 2), as suggested by rock glacier records from the Southern Alps (Frauenfelder et al., 2001). With no winter snow cover (2d) the temperature at the CCC site reaches -0.5°C after ~60 years and drops below -1°C (i.e. it leaves the "CCC window") after 100 205 years (Fig. 3d, Suppl. Fig. 3d). On the other hand, if snow insolates the ground in winter and buffers the winter cold by 4.7 °C (i.e. ∆T of 4.7°C; Table 2), temperatures at 50 m depth stay above -0.8°C for an extended period of time (Fig. 3e, Suppl. Fig.   3e).

Scenario 3 -Stadial conditions during the late YD
Three experiments were designed to explore the impact of climate change at the mid-YD transition, starting at 12.24 ka BP 210 , on the subsurface thermal regime at this high-alpine site (Table 2). In the first experiment (3a) we examined how permafrost conditions would change with increasing aridity in autumn compared to the early YD (2e). As the insulating effect of the winter snowpack is reduced, the depth zone of the CCC site experiences rapid cooling and approaches -1.8°C https://doi.org/10.5194/cp-2020-107 Preprint. Discussion started: 25 August 2020 c Author(s) 2020. CC BY 4.0 License.
after 100 years, leading to permafrost development, inconsistent with stable conditions near 0°C required for CCC formation (Fig. 4a, Suppl. Fig. 4a) 215 We applied a +1°C change in MAAT represented by a 2°C rise in January temperatures with (3c) and without (3b) snow cover (Table 2). We kept the climate warming at 1°C (i.e. ΔMAATAllerød-late YD= -2°C), because a larger warming would result in a climate similar to the preceding Allerød. The results of scenario 3b demonstrate that even though winters become slightly less cold, the subsurface at 50 m depth would nevertheless cool from -1.3°C to -1.5°C due to the lack of a winter snow cover. This scenario is not compatible with CCC formation as it leads to permafrost aggradation (Fig. 4b, Suppl. Fig. 4b). In contrast, even 220 the presence of a moderate snow cover (3c) would allow the subsurface at 50 m depth to slowly warm to -1°C after 75 years of the start of the atmospheric warming (Fig. 4c, Suppl. Fig. 4c), creating favorable conditions for slow freezing of liquid water pockets in the ice introduced by dripping water.

0°C conditions in the shallow subsurface 225
CCCs form in slowly freezing water pockets enclosed in cave ice when the cave interior temperature is very close to the 0°C isotherm (e.g., Žák et al., 2018). Although the deposition of CCCs in many cases mark climate transitions (e.g., Richter and Reichelmann, 2008;Spötl and Cheng, 2014;Žák et al., 2012), the large size of some CCCs (up to 50 mm in caves elsewhere, unpublished data from our group) and 230 Th ages from their central and rim areas (unpublished data from our group) argue for very stable cave microclimate conditions for at least several years (e.g., Žák et al., 2018). 230 CCCs (Table 2) provide unequivocal evidence that perennial ice was present in the lower descending gallery of Cioccherloch during the first part of the YD. As the majority of 230 Th ages overlap within their 2σ errors, it is not possible to determine whether CCC formation took place continuously for 400-600 years or if they represent two different generations clustering at ~12.6 and ~12.2 ka BP (Fig. 5). The diversity of morphologies and the occurrence of discrete spots nevertheless supports the latter, indicating the presence of several CCC-forming water pockets/pools. Overall, CCCs in Cioccherloch record interior 235 cave air temperatures very close to 0°C from ~12.6 to ~12.2 ka BP, initiating progressive freezing of meltwater pockets in perennial ice which were created by drip water.
The air temperature in the homothermic zone of caves is in equilibrium with the MAAT of the outside atmosphere. Ice-bearing caves, however, commonly represent an exception to this rule (e.g., Perșoiu, 2018 and references therein.). Due to its descending geometry lacking a lower entrance, Cioccherloch is a cold trap, as evidenced by the snow and ice cone at the base 240 of the entrance shaft. This negative thermal anomaly, however, is restricted to the upper cave level close to the snow cone.
Today, the descending gallery with the CCC occurrences is thermally isolated from this upper level due to the narrow connection, as shown by stable cave air temperatures consistent with the ambient mean MAAT (2.4°C). Prior to cave exploration in the 1980s and 1990s, this squeeze was partly closed by rubble. Therefore, we presume that this descending gallery has been thermally decoupled from the upper cave level in the past and remained in thermal equilibrium with the 245 https://doi.org/10.5194/cp-2020-107 Preprint. Discussion started: 25 August 2020 c Author(s) 2020. CC BY 4.0 License. ambient atmosphere. As a result, CCCs in the lower gallery record changes in the thermal state of the subsurface in relation to atmospheric temperature changes. Given the lack of ventilated shafts connecting this gallery to the surface, we argue that heat exchange between the latter and the gallery occurred primarily via conduction. Additional heat transfer may involve drip water influx and minor air advection via possible small-scale fissures in the ceiling. These processes are difficult to quantify for any time in the past and no attempts were made to include them in the thermal model. Qualitatively, both processes would increase 250 the rate of temperature change in this gallery as a response to atmospheric change above the cave. By considering heat conduction only, our simulations yield quantitative constraints on the maximum duration of temperature change propagated into the shallow subsurface.
The formation of CCCs near the lower end of the descending gallery requires the 0°C isotherm to be located at 50 m below the ground surface at 2274 m a.s.l. in the Dolomites during the YD. On the other hand, stalagmite Cioc1 from the same gallery 255 provides strong evidence that subsurface conditions were favorable for speleothem growth during the Bølling-Allerød interstadial. In other words, during this interstadial this gallery was free of ice and the cave air temperature was above 0°C.
During the early YD, atmospheric cooling led to the aggradation of cave ice. Probably during the warm YD summers (from ~12.6 to ~12.2 ka BP) drip water from torrential rains and/or snowmelt created meltwater pools that subsequently underwent slow freezing cycles, requiring a cave air temperature in this gallery ~2-3°C (i.e. 2.5 ± 0.5°C) lower than today. 260 CCC deposition in caves is traditionally attributed to near-surface permafrost degradation in response to atmospheric warming (e.g. Žák et al., 2018 and references therein). However, stalagmite growth indicates that CCC formation in Cioccerloch cave does not represent a delayed response of the Bølling-Allerød interstadial warming. Instead our data suggest that CCCs in Cioccherloch may formed during transitions into cold periods. Therefore, we hypothesize that analogous to climate warmings possible "CCC windows" opened during the transition into stadials and remained open for a variably long period of time 265 depending on the local thermal conditions of the subsurface.

Magnitude of YD cooling
CCCs dated to the first and second half of the YD suggest conditions very close to 0°C for an extended period of time during this stadial. Our thermal model shows the CCC formation starting at 12.6±0.2 ka BP at this sensitive mountain site can only be reconciled by evoking a moderate atmospheric cooling at the Allerød-YD transition of -4.5 to -5°C relative to today (Fig.  270   5). Without a winter snow cover (scenarios 2b and 2d) the "CCC window" would open too early and close quickly afterwards, inconsistent with the CCC ages (Fig. 5). Scenario 2c and 2e shows that if the YD climate was characterized by a -5 to -4.5°C drop in MAAT relative today (i.e. ΔMAATAllerød-YD= -3 to -2.5°C), a thick and stable snow cover during winter is needed to prevent the cave from freezing, effectively shielding the ground from the cold stadial winters (Zhang, 2005). Scenario 2e including provides the best fit with the CCC data giving rise to a 400 year-long period characterized by a very slow cooling of 275 the subsurface with cave air temperatures near -0.8°C.
MAAT would freeze Cioccherloch rapidly even if a thin winter snowpack was present, and would result in rather abrupt development and deepening of permafrost, preventing CCC formation (Fig. 7). Such a stark cooling would in fact lead to 280 climate conditions similar to the Last Glacial Maximum (LGM), for which noble gas data from groundwater studies around the Alps suggest 7-10°C lower temperature compared to the Holocene (e.g., Šafanda and Rajver, 2001;Stute and Deak, 1989;Varsányi et al., 2011) and which would inevitably lead to the build-up of glaciers at this elevation in the Dolomites.
Our data, however, are consistent with observations from rock glaciers (Frauenfelder et al., 2001) and lake sediments in the Swiss Alps (Von Grafenstein et al., 2000) suggesting a moderate cooling at the Allerød-YD transition. A fluid inclusion-based 285 paleotemperature reconstruction using stalagmites from Bärenhöhle in western Austria also indicates a maximum temperature drop of about 5.5°C, supporting our interpretation (Luetscher et al., 2016).

Increased seasonality in the early YD
CCCs provide a uniquely robust control on cave air temperatures and consequently on the MAAT. Our data argue for a ≤3°C drop in MAAT at the Allerød-YD transition, but provide no direct information on the seasonal cycle of ambient atmospheric 290 temperatures. A recent multi-proxy-model comparison by Schenk et al. (2018) suggests persistently warm summers during the YD with a median regional cooling of 3°C and 0.3°C over NW-and E-Europe, respectively, compared to Bølling-Allerød summers. These authors also argue that previous studies using chironomids overestimated the YD cooling signal. A similar amplitude of change is suggested for July air temperatures by lake records from the Western Alps. Pollen and cladocerainferred temperature reconstructions indicate a summer cooling of 2-4°C at the Allerød-YD transition at Gerzensee (Swiss 295 Plateau, e.g., Lotter et al., 2000), consistent with a 3.5°C drop in July air temperatures reported from Maloja Pass in eastern Switzerland (Oberli et al., 2009). We therefore consider 0.3° and 4°C as minimum and maximum estimates of YD summer cooling, respectively, relative to the Bølling-Allerød. As our CCC data in conjunction with thermal modelling constrain the drop in MAAT at the Allerød-YD-transition to ≤3°C, we find that if YD summers were indeed 0.3 to 4°C colder than in the Allerød, early YD winters at 2270 m a.s.l. were no colder than -13.7°C (mean January temperature). This argues for an 300 enhanced seasonality in the Dolomites, whereby the winter-summer temperature difference increased by up to 5.7°C at the Allerød-YD transition.
Thermal modelling shows that a thin winter snowpack effectively shielding the subsurface during the cold winters is needed to account for CCC formation commencing at 12.6 ±0.2 ka at Cioccherloch. Studies in modern permafrost areas suggest that a stable winter snow cover of only ~35 cm results in a positive shift of up to 5.5°C in the mean annual ground surface 305 temperature (Zhang, 2005). Changes in the timing, duration, thickness and density of the snow cover may promote either the development or the degradation of permafrost (Zhang, 2005). A snowpack in the cold season leads to a positive temperature anomaly in the ground, whereas a summer snow cover insulates the ground from warm air and facilitates the development of permafrost. In a study of Arctic permafrost, Park et al. (2014) found that the thermal state of the underlying soil is more affected by early winter than peak winter snowfall. Therefore, we argue for a moderately humid early YD with snowfall during fall and 310 early winter. Although the spatial distribution of rock glaciers from one of the driest areas of the Swiss Alps suggests a 30-https://doi.org/10.5194/cp-2020-107 Preprint. Discussion started: 25 August 2020 c Author(s) 2020. CC BY 4.0 License. 40% reduction in YD precipitation compared to today (Frauenfelder et al., 2001), paleoglacier records point to similar amounts but a different seasonal distribution of precipitation in the Southern Alps with respect to modern day (Kerschner et al., 2016).

Climate change during the mid-YD
High-resolution speleothem (Baldini et al., 2015;Bartolomé et al., 2015;Rossi et al., 2018) and lake records (Bakke et al., 315 2009;Brauer et al., 2008;Lane et al., 2013) from W and N Europe suggest a time-transgressive change in atmospheric circulation during the YD associated with a warming of parts of Europe due to a retreat of winter sea ice and a northward migration of the polar front (e.g., Baldini et al., 2015;Bartolomé et al., 2015). At Meerfelder Maar, Germany, this so-called mid-YD transition occurred at 12.24 ±0.04 ka BP . This timing is strikingly similar to the error-weighted mean of the CCC dates from Cioccherloch (12.19 ±0.06 ka BP), suggesting that the main phase of CCC formation at this site 320 may have been related to climate change. Our thermal simulations provide important constraints on the type and magnitude of climate change during the mid-YD transition at this high-alpine site and suggest a mild warming by up to 1°C (MAAT) with a slight reduction in precipitation.
A change from moderately snow-rich to snow-poor autumns and early winters from the early to the late YD combined with a small atmospheric warming (scenario 3c) is consistent with the main advance of Alpine ice glaciers (Fig. 6) during the first 325 few centuries of the YD (Baroni et al., 2017;Heiri et al., 2014b;Ivy-Ochs et al., 2009). The subsequent glacier reduction and the parallel increase in rock-glacier activity advocate less humid conditions towards the end of the YD and the early Holocene in the Western and Eastern Alps (Ivy-Ochs et al., 2009;Kerschner and Ivy-Ochs, 2008).
Speleothem records from the Pyrenees (Baldini et al., 2015;Bartolomé et al., 2015), Cantabrian Cordiella (Rossi et al., 2018) and the Adriatic coast (Belli et al., 2017) suggest a precipitation increase in the late YD, attributed to a strengthening of the 330 westerlies. Precipitation-sensitive archives in the northern Alps, however, do not show evidence of a major change. Benthic ostracod δ 18 O records from Lake Ammersee (Grafenstein et al., 1999) and Lake Mondsee (Lauterbach et al., 2011) only show a gradual ca. 1‰ increase across the YD, arguing against a major step-wise change in the precipitation regime as would be expected as a result of the migration of the polar front across the Alps (Fig. 6). This slight increase in δ 18 O values, however, is compatible with our interpretation of a small decrease in autumn precipitation in the northern alpine catchment areas of these 335 lakes in the second half of the YD (Fig. 6) coupled with a minor (≤1°C) warming. Speleothem δ 18 O data from Hölloch cave west of the Ammersee catchment (Wurth et al., 2004) and from the Jura Mountains (Affolter et al., 2019) likewise lack isotopic evidence of a significant change in climate within the YD. While our data and paleoglacier evidence are consistent with a slight warming at the mid-YD transition, they argue for a reduction in fall and winter precipitation. This suggests that the popular model of a south-north migration of the polar front and a concomitant increase in westerly-driven precipitation (e.g., 340 Lane et al., 2013) is too simplistic at least for the greater Alpine realm, underscoring the need for regionally resolved climate models. In fact, even at the key site of Meerfelder Maar, the proposed increase in (winter) precipitation is poorly captured by most proxy data except for the abundance of Ti, which is attributed to spring snow melt .

Conclusions 345
This article presents the first record of CCC in the Dolomites which formed, in contrast to many studies from Central European caves, not during a major climate warming but within a prominent stadial. These deposits indicate sustained conditions of ~0°C between ~12.6 and ~12.2 ka BP at about 50 m below the surface, initiating the slow freezing of dripwater-induced meltwater pockets in perennial cave ice. Combined with a thermal model the high-elevation setting of this cave suggests a ≤5°C drop in MAAT compared to today, incompatible with extreme winter cooling during the YD. CCC formation during the 350 early YD requires autumn to early winter snowfall forming a sufficiently thick and stable snow cover insulating the ground from the winter cold. CCC formation during the early YD coincided with the maximum YD extent of Southern Alpine glaciers, consistent with abundant snowfall in autumn and winter and with decreased summer temperatures. Using a 0.3-4°C cooling for the short and mild early YD summers as suggested by data-model comparison studies (Heiri et al., 2014b;Schenk et al., 2018), mean January air temperatures at this alpine site were most likely not colder than -13.7°C. Seasonal temperature 355 differences between early YD summers and winters were therefore up to 5.7°C larger than during the Allerød.
The 230 Th data provide strong evidence that CCC formation at ~12.2 ka occurred in response to climate change associated with the mid-YD transition. CCC formation at this high-alpine cave advocates a mild atmospheric warming (i.e. +1°C in MAAT) and a reduction in fall precipitation in the late YD. We propose a shift from snow-rich early YD towards snow-poor late YD autumns and early winters, which is consistent with the retreat of YD glaciers in the Alps and an increase in rock glacier 360 activity.
CCCs are a novel paleoclimate archive allowing to precisely constrain permafrost thawing events in the past. Our study demonstrates that CCCs can also provide quantitative constraints on paleotemperature and seasonally resolved precipitation changes.

Code and data availability 365
The code for the 1d heat-flow model is available online (https://zenodo.org/record/3982221). Data is included in Tables 1 and   2 and additionally given in Supplementary Table 1. Affolter is thanked for providing data from Milandre cave. This work was supported by the Autonome Provinz Bozen -375 Südtirol, Amt für Wissenschaft und Forschung (grant 3/34 to C.S.) and the Tiroler Wissenschaftsförderung grant WF-F.16947/5-2019 (to G.K.).

Author contribution
G.K. and C.S. designed the study, carried out field work, performed petrographic, stable isotope analyses and heat-flow modelling. G.K. carried out uranium-series dating, supervised by H.C. G.K. wrote the paper with contributions from all co-380 authors.    Table 2: Input parameters for 1d heat conduction models simulating climate during the Allerød interstadial (1), the early YD (2a-2e) and the late YD at the study site (3a-3c). Snow ∆T ascribes the attenuation of the winter cold by the snowpack, whereas the resultant annual air temperature used as a boundary condition for the thermal model is described as mean the annual effective temperature (MAET). Modern day values are shown for comparison.