An Interdisciplinary Study of a Mammoth-Bearing Late Pleistocene Sediment Succession in Lower Austria

An interdisciplinary study of a mammoth-bearing Late Pleistocene sediment succession in lower Austria

Carobene, D., Meyer, M. C., Spötl, C., Rötzel, R., Göhlich, U. B., Mandic, O., Harzhauser, M., Wimmer-Frey, I., Reimer, P. J., & Auer, F. (2020). An interdisciplinary study of a mammoth-bearing Late Pleistocene sediment succession in lower Austria. Quaternary International. https://doi.org/10.1016/j.quaint.2020.02.022

Published in: Quaternary International

Document Version: Peer reviewed version

Queen's University Belfast - Research Portal: Link to publication record in Queen's University Belfast Research Portal

Publisher rights Copyright 2020 Elsevier. This manuscript is distributed under a Creative Commons Attribution-NonCommercial-NoDerivs License (https://creativecommons.org/licenses/by-nc-nd/4.0/), which permits distribution and reproduction for non-commercial purposes, provided the author and source are cited.

General rights Copyright for the publications made accessible via the Queen's University Belfast Research Portal is retained by the author(s) and / or other copyright owners and it is a condition of accessing these publications that users recognise and abide by the legal requirements associated with these rights.

Take down policy The Research Portal is Queen's institutional repository that provides access to Queen's research output. Every effort has been made to ensure that content in the Research Portal does not infringe any person's rights, or applicable UK laws. If you discover content in the Research Portal that you believe breaches copyright or violates any law, please contact [email protected].

Download date:02. Oct. 2021

1 An interdisciplinary study of a mammoth-bearing Late Pleistocene sediment succession in Lower Austria 2 3 Daria Carobenea,*, Michael C. Meyerb, Christoph Spötlb, Reinhard Roetzelc, Ursula B. Göhlichd, Oleg 4 Mandicd, Mathias Harzhauserd, Ingeborg Wimmer-Freyc, Paula J. Reimere, Fabian Auerb 5 6 a Museum of Nature and Environment, Mühlendamm 1-3, 23552 Lübeck, Germany 7 b Institute of Geology, University of Innsbruck, Innrain 52, 6020 Innsbruck, Austria 8 c Geological Survey, Neulinggasse 38, 1030 Vienna, Austria 9 d Natural History Museum Vienna, Department of Geology and Palaeontology, Burgring 7, 1010 Vienna, 10 Austria 11 e Centre for Climate, the Environment and Chronology (14CHRONO), School of Natural and Built 12 Environment, Queen᾿s University Belfast, Belfast BT7 1NN, United Kingdom 13 *corresponding author 14 15 Abstract 16 In the Alpine foreland and the Vienna Basin loess-paleosol sequences (LPS) are common. Some of the most 17 famous LPS sites in the circum-Alpine area include Stratzing, Göttweig, Willendorf, Krems-Wachtberg, and 18 Stillfried, which cluster in a relatively small area along the Danube river in Lower Austria. LPS provide 19 detailed insights into climate-driven, terrestrial palaeoenvironmental changes that can be placed into a 20 robust chronological framework, because LPS are amenable to a range of dating techniques. Here, we 21 present a well-dated 13.6 m-thick mammoth-bearing sediment succession characterised by low-energy 22 aquatic deposits, sheet-flow deposits and sandy loess from re-deposited Miocene sediments, i.e. a 23 depositional environment that contrasts to the classical LPS sites. This new site is situated 1.6 km NNE of 24 Bullendorf in Lower Austria, where Pleistocene sediment successions with a robust chronology are rare. 25 Our multidisciplinary approach is based on optically stimulated luminescence and 14C dating, and includes 26 mammal faunal investigations and stable isotope analyses of molluscs. OSL and 14C dating suggest 27 deposition of the sediment sequence immediately before and briefly after the Last Glacial Maximum. The 28 mollusc assemblages and the mammal fauna are representative of a cold climate, characteristic of a tundra 29 steppe environment. Stratigraphic changes in δ18O of two mollusc species (Pupilla muscorum and Succinella 30 oblonga) suggest an alternating dry-cold and humid-cold climate. Oxygen isotope data of freshwater 31 gastropod shells suggest a drastic decrease in the mean growing season temperature compared to today, 32 while the carbon isotope composition is indicative of a C3 vegetation. 33 34 Keywords: Late Pleistocene, palaeoenvironment, luminescence dating, stable isotopes, Austria 35

36 1. Introduction 37 Loess deposition was widespread in the lowlands to the north and east of the Alpine ice sheet during 38 Pleistocene glaciations (Grill, 1968; Wessely, 2006; van Husen and Reitner, 2011; Terhorst et al., 2014). 39 These deposits commonly reach several metres in thickness and are composed of loess-paleosol sequences 40 (LPS) reaching back locally to the Early Pleistocene (Fink and Kukla, 1977; Thiel et al., 2011a, 2011b). LPS 41 reflect the changing climatic conditions of dust accumulation in a dry tundra or cold steppe environment 42 interrupted by more humid and milder intervals during which the dust flux was strongly reduced and syn- 43 to postdepositional pedogenesis occurred (e.g. Pye, 1995). Fluvial and fluvioglacial outwash deposits along 44 major rivers and in front of the piedmont lobes of the Alpine ice sheet form a sequence of Pleistocene 45 terraces extending into the Alpine foreland (van Husen and Reitner, 2011). Pleistocene sediments of other 46 depositional environments, such as lacustrine deposits or sediments affected by solifluction, are far less 47 common in these peri-Alpine settings. These deposits host faunal remains (most commonly mammoth 48 tusks and bones; e.g. Spötl et al., 2018) and sometimes archaeological finds (e.g. Einwögerer et al., 2006; 49 Nigst et al., 2014). Sedimentological and chronological data are available for several LPS in the Alpine- 50 Carpathian foreland and the Vienna Basin, especially for sites along the Danube corridor in the western 51 sector of the Alpine-Carpathian foreland, such as Stratzing, Göttweig, Willendorf, and Krems-Wachtberg 52 (Kukla, 1975; Fink, 1976; Zöller et al., 1994; Nigst et al., 2008; Kovanda et al., 1995; Havlíček et al., 1998; 53 Peticzka et al., 2010; Smolíková et al., 2010; Sprafke, 2016). Several of these sites, such as Willendorf (Nigst 54 et al., 2014), Krems-Wachtberg (Einwögerer et al., 2006) and Stratzing (Bednarik, 1989) are also 55 archaeologically important and collectively provide unique insights into the Upper Palaeolithic of Central 56 Europe. However, only a limited number of studies provide robust chronologies for these LPS (Terhorst et 57 al., 2014; Thiel et al., 2011a, 2011b; Nigst et al., 2014), and well-dated LPS (e.g. Frank et al., 2011) and other 58 types of sediments are rare for the loess region north of Vienna and the Danube corridor. 59 Here, we present a 13.6 m-thick mammoth-bearing sediment succession, which differs from the classical 60 LPS sites in the Danube region in terms of its depositional environment. This recently discovered site is 61 situated close to Bullendorf, in the northeastern part of Lower Austria, where continuous LPS are rare 62 (Frank et al., 2011). The sediment succession at Bullendorf is characterised, for the most part, by well- 63 bedded sandy deposits with only a few loess-like layers. We use a multidisciplinary approach including 64 sedimentology, studies of the mammal fauna and the stable isotope composition of gastropods, combined 65 with optically stimulated luminescence (OSL) and 14C dating in order to derive a palaeoenvironmental 66 reconstruction for this site at high temporal resolution. 67 68 2. Geological setting 69 The studied section is located 1.6 km NNE of Bullendorf (Lower Austria), on the northern slope of a road 70 cut of the newly constructed A5 highway (WGS84: 48° 36’ 31.5” N, 16° 40’ 11.9” E) in the Mistelbach

71 district (Fig. 1, Fig. 2). The road cut was constantly stabilised and re-vegetated by the construction company 72 and thus provided only temporary outcrops of the exposed Pleistocene deposits. The mammoth-bearing 73 sediment succession was only accessible for a few weeks during the summer of 2016. Topographically, the 74 site is located in the catchment of the Zaya River which flows in a northeasterly direction at a distance of 75 about 1.5 km from the site. The area is part of the Vienna Basin, a Neogene pull-apart basin covering a 76 great part of northeastern Lower Austria and reaching into the Czech Republic and Slovakia in the north 77 and east, respectively (Rasser et al., 2008). The oldest sediments outcropping in the vicinity of the 78 Bullendorf site are Badenian and Sarmatian (middle Miocene) in age (Fig. 2). During the middle Miocene, 79 marine limestone of the Badenian and mixed-siliciclastic carbonate deposits of the Sarmatian were laid 80 down (Harzhauser and Piller, 2004). Subsequently, during the late Miocene (Pannonian), the fluvial 81 Hollabrunn-Mistelbach Formation formed (Nehyba and Roetzel, 2004). This formation is mainly comprised 82 of gravel and sand deposited by the Danube River that today flows ~ 40 km further to the south. During the 83 Pannonian, lacustrine clays and silts were deposited in the study area, while alluvial sediments prevailed 84 again during the late Pannonian (Harzhauser et al., 2004; Wessely, 2006). 85 In post-Miocene times, especially during the Pliocene, the regional landscape evolution was dominated 86 by erosion (Wessely, 2006), giving rise to the smooth relief that characterises this part of Lower Austria 87 today. High-lying gravel beds are remains of former land surfaces (e.g. terrace gravels near Maustrenk, Fig. 88 2; Grill, 1968), while asymmetric valleys that are common in the Bullendorf area, are geomorphological 89 witnesses of Pleistocene permafrost activity. The road cuts along the A5 highway provided evidence that 90 the Pleistocene topography was also considerably different compared to the topography of today. 91 Unfortunately, due to the lack of detailed geological mapping and the temporary nature of these outcrops, 92 a reliable reconstruction of the landscape evolution is still lacking for the Bullendorf region. 93 Pleistocene sedimentation in the study area was dominated – as in the wider circum-Alpine area – by 94 loess accumulation, draping the topography (Grill, 1961, 1968; Wessely, 2006; van Husen and Reitner, 95 2011; Fig. 2). During milder and more humid climate periods of the glacials, slope processes such as 96 solifluction and sheet flow played an important role. Loess accumulation in Lower Austria was facilitated via 97 two transportation pathways: (i) long-distance transport from alluvial and glacial outwash plains, and (ii) 98 short-distance transport from regionally exposed Neogene sediments of the Foredeep and the Vienna 99 Basin. These loess deposits are often subdivided by paleosols, resulting in LPS that are commonly used for 100 litho- and chronostratigraphy (e.g. Stratzing, Göttweig, Willendorf, Krems-Wachtberg, Stillfried along the 101 Danube corridor; Fig. 1). In the study area, however, such LPS are rarely exposed and limited to few 102 outcrops near Stillfried (Fig. 1, Peticzka et al., 2010; Frank et al., 2011; Terhorst et al., 2015). However, short 103 time exposures during the construction of A5 highway opened complex loess sections with paleosols in the 104 Mistelbach area also (Smolíková et al., 2010; Havlíček and Vachek, 2016). In the wider investigation area 105 considerable tectonic movements related to the pull-apart kinematics of the Vienna Basin occurred, and

106 these movements extend into the Pleistocene (Decker et al., 2005; Weissl et al., 2017). Evidence of 107 neotectonic movements was also found in the A5 highway outcrops, ~200 m southwest of the Bullendorf 108 site. There, a dark brown paleosol is displaced by ENE-striking and SSE-dipping normal faults, striking 109 approximately parallel to the Zaya valley, which is situated ~1.5 km southeast and ~30 m lower in elevation 110 relative to this site (Fig. 2). Because of the limited outcrop situation and the potentially strong influence of 111 tectonics on the course and incision history of the Zaya river, reconstruction of the Pleistocene topography 112 at Bullendorf remains challenging. 113 114 115 3. Methods 116 3.1. Fieldwork 117 Due to the rapidly progressing construction work at the A5 highway fieldwork this work was essentially a 118 rescue excavation, and only logging and drawing of bedding architecture in the outcrop and on 119 photomosaics together with facies description and sampling for sedimentological and palaeontological 120 investigations and for dating were done. The classification of Keller (1996) was used for lithofacies 121 description, which is better suited for Pleistocene sediments than the well-known lithofacies codes of Miall 122 (1984) and which takes the specific components and textural features of glacial and periglacial sediments 123 into account. In this classification the first letter indicates the main components sand (S), fine sediment 124 (silt, clay; F) or heterolithic sediment (alternation of fine sediment and sand; H). The second letter describes 125 the texture, like lamination (l), bedding (b), pedogenic influence (p) or ripple lamination (r). 126 127 3.2 Sedimentological analyses 128 Particle size distribution was determined by wet manual sieving and automatic sedimentation analysis 129 according to DIN 4022. The <32 µm size fraction was analysed using a Micromeritics Sedigraph III 5125. 130 Samples were classified according to Müller (1961) and Füchtbauer (1959). Statistical sedimentological 131 parameters were given by the moment calculation and the method of Folk and Ward (1957). 132 The bulk and clay mineralogy of the samples were determined by X-ray diffraction (XRD). For bulk 133 mineral analysis the dried samples were ground and loaded into a sample holder as a randomly oriented 134 powder. Diffraction data were collected with a PANalytical X’Pert PRO Multi-Purpose Diffractometer 135 (goniometer PW 3050/60), Cu K radiation (40 kV, 40 mA), an automatic divergence slit, and a PIXcel 136 detector with programmable anti-scatter slit step scan mode (step size 0.013° 2, 10 s per step). The 137 samples were run from 3 – 70° 2. The semiquantitative mineralogical composition was obtained using the 138 computer program SEIFERT AutoQuan based on Rietveld analysis.

139 For clay mineral analysis the samples were treated with H2O2 in an ultrasonic bath in order to remove 140 organic matter for further disaggregation. The <2 μm fraction was separated by centrifugation. The clay

141 fraction was saturated with a 1N KCl solution and a 1N MgCl2 solution by shaking for 24 h and afterwards 142 washing with distilled water. Oriented slides of the <2 µm fraction were prepared by placing 25 mg clay in 143 suspension on a porous ceramic plate and drying at room temperature. Oriented XRD mounts were then 144 analysed in air-dried, ethylene glycol, dimethyl sulfoxide and glycerol-treated states. The clay samples were 145 run from 2-50° 2 with the same step and counting time as the bulk samples. The identification of clay 146 minerals was carried out according to Moore and Reynolds (1997). The relative percentages of the clay 147 minerals in the <2 μm fraction were determined following Schultz (1964). 148 149 3.3 OSL 150 Five samples were obtained for OSL dating by hammering stainless-steel tubes into the outcrop and 151 sealing with duct tape. Quartz in the grain size fraction 180-212 μm was physico-chemically purified from 152 the bulk sediment samples following standard procedures (Wintle, 1997). The quartz grains were mounted 153 on stainless steel discs using silicon oil. The aliquot size was chosen to be small (~ 50-100 grains per disc). 154 All OSL measurements were conducted on an automated Risø TL/OSL–DA15 Reader. Quartz OSL 155 stimulation was achieved via blue light emitting diodes (LED´s, 470 nm, 22 mW/cm2 stimulation power) and 156 signals were detected with an EMI 9235 photomultiplier tube after passing through 5 mm of Hoya U-340 157 filter to isolate the ultraviolet wavelength (Bøtter-Jensen et al., 2000). 158 The single aliquot regenerative dose (SAR) procedure was used for environmental dose (De) 159 determination (Murray and Wintle, 2000). Routine checks of SAR protocol performance were made for 160 thermal transfer, test dose sensitivity correction, dose response behaviour and feldspar contamination 161 (Murray and Wintle, 2000; Duller, 2003). Aliquots were rejected from further analyses if (1) the OSL signal 162 was weak (TN < 3 times the background signal), (2) the sensitivity-corrected zero dose value was > 5% of 163 the sensitivity-corrected natural value, (3) the repeat dose points at the end of the measurement 164 procedure were not consistent with the initial value for the same dose (> 2σ different from unity), (4) the 165 sensitivity-corrected natural OSL signal did not intercept the dose-response curve, and (5) the OSL-IR 166 depletion ratio was more than 2σ below unity, suggesting the presence of feldspar grains or inclusions 167 (Duller, 2003). Blue LED stimulation was for 100 s and the OSL signals were integrated over the initial 2 s of 168 stimulation minus a background calculated from the last 20 s of the OSL decay curve. The dose response 169 curves were fitted with a single exponential function and fitting errors were determined via 500 Monte 170 Carlo iterations. 171 For all samples, the beta dose rates were measured using a GM-25-5 low-level beta counter (Bøtter- 172 Jensen and Mejdahl, 1988), while gamma rays were measured in situ with a NaI(Tl) detector. The gamma 173 dose rates were determined using the “threshold” technique (Mercier and Falguères, 2007; Guerin and 174 Mercier, 2011), which gives an estimate of the combined dose rate from gamma-ray emitters in the U and

175 Th chains and from 40K. In addition, the gamma dose rate of sample O5 was calculated using a combination

176 of beta counting and thick-source alpha counting to obtain activity concentrations of U, Th and K, which 177 were converted to dose rates (Adamiec and Aitken, 1998). No significant or systematic differences were

178 observed between the gamma dose rates measured in these two ways for sample O5. The cosmic dose rate 179 was calculated following Prescott and Hutton (1994). An internal alpha dose rate of 0.03±0.01 Gy/ka was 180 assumed based on measurements by Jacobs et al. (2006). The field water content for all samples was close 181 to 15 wt.-% and this value was used for age calculation with a 30% uncertainty assigned to take into 182 account possible moisture and water table fluctuations during the burial period. 183 184 3.4 Radiocarbon 185 Six mammalia bones and one sample of the mammoth tusk were screened for their whole nitrogen 186 content. The Equus sample (UBA- 34810) was thought to have been coated with Mowilith® emulsion and 187 was treated with a solvent extraction in a Soxhlet distillation apparatus using a minimum of two cycles of 188 tetrahydrofuran, chloroform, petroleum spirit, acetone, methanol and lastly deionized water, similar to 189 Bruhn et al. (2001) before proceeding. Collagen extraction was done on two samples with sufficiently high 190 nitrogen using the ultrafiltration method (Brown et al., 1988) with Vivaspin® filter cleaning following Bronk

191 Ramsey et al. (2004). Samples were combusted to CO2 and processed to graphite using the zinc reduction 192 method (Slota et al., 1987) and analysed using an accelerator mass spectrometry (AMS) at the 14CHRONO 193 Centre, Queen's University Belfast. The sample 14C/12C ratio was background-corrected using 194 measurements on collagen extracted from the Hollis mammoth bone (YG 431.10, Martinez et al. 2019) and 195 normalised to the HOXII standard (SRM 4990C; National Institute of Standards and Technology). Ages were 196 calculated according to Stuiver and Polach (1977) using the AMS measured 13C/12C which accounts for both 197 natural and machine isotope fractionation. The reported error in the age was multiplied by 1.3 based on 198 reproducibility of bone standards and includes long-term variability in the background. For asymmetric 199 standard deviations the larger value is reported. Ages were calibrated using IntCal13 (Reimer et al., 2013) 200 and the CALIB 7.1 software (Stuiver et al., 2019). Calibrated age ranges are reported at two standard 201 deviations (2σ). Stable isotopes (δ13C and δ15N), % carbon and % nitrogen were measured on a Delta V 202 Advantage with Flash elemental analyser and atomic carbon to nitrogen ratios calculated (C/N). Stable 203 isotope standards IA-R041 L-Alanine (δ13C = 23.33 +/- 0.10; δ15N = -5.56 +/- 0.14), IAEA-N-2 Ammonium 204 Sulphate (δ15N = +20.3 +/- 0.2) and IAEA-CH-Sucrose (δ13C =-10.449 +/- 0.033) were analysed with the 205 unknown samples to provide two point calibration. δ13C is reported relative to VPDB and δ15N relative to 206 AIR. Reproducibility for ultrafiltered bone collagen is 0.22 ‰ for δ13C and 0.15‰ for δ15N. 207 208 3.5 Gastropods 209 All gastropod species from Bullendorf have been described by Carobene et al. (2018) in detail, and 210 include (i) Miocene species that existed exclusively in marine and brackish environments, as well as (ii)

211 terrestrial species that appeared only during the Pliocene and/or Pleistocene. In this study, we focus on the 212 isotopic analysis of autochthonous Pleistocene gastropod shells. The terrestrial Plio-Pleistocene species are 213 easily discriminated from the marine Miocene species, based on morphological criteria and their state of 214 preservation (the latter species are usually strongly abraded and fragmented). Miocene gastropods are 215 present within the reworked Miocene sediment lenses contained in unit 5 (sample C7 and B7), but not in 216 any other unit of the Bullendorf section. 217 70 gastropod shells from 12 samples were investigated, representing 10 species, 9 terrestrial and 1 218 freshwater species. Since fossil shell material may be affected by dissolution of the aragonite and 219 reprecipitation of calcite (Brennan and Quade, 1997; Balakrishnan et al., 2005a), shells were picked under a 220 stereomicroscope and examined with a scanning electron microscopy (SEM) to check for evidence of 221 dissolution and/or recrystallisation. No signs of aragonite leaching or alteration were found. Shells were 222 sampled according to three methods depending on size and specimen availability: (1) separating the 223 juvenile from the adult whorls, (2) cutting shells in random aliquots, and (3) crushing whole shells. 224 According to specimen availability, 3–6 shells of each taxon were analysed (10 species out of 14 species in

225 total; Carobene et al., 2018). Shells of different species were selected from samples B3, C1 and C3 to

226 examine the isotope variability of coeval samples. Additionally, shells of Pupilla muscorum (B2, C1, B3, C3,

227 B6) and Succinella oblonga (B3, C3, B6, C7), both present in most samples, were analysed to trace 228 palaeoenvironmental changes across the section. 229 Samples were dissolved in orthophosphoric acid at 72°C and the carbon dioxide gas was analysed in an 230 isotope ratio mass spectrometer linked to a Gasbench II. Values are reported relative to the VPDB standard. 231 Long-term precision of the δ13C and δ18O values, estimated as the 1σ-standard deviation of replicate 232 analyses, is 0.06 and 0.08 ‰, respectively (Spötl and Vennemann, 2003). 233 234 4. Results 235 4.1 Sediment description 236 Based on the sedimentary characteristics, the 13.6 m-thick succession was divided into six lithological units 237 (Figs. 3 and 4). 238 The lowermost unit 1 (~250-300 cm thickness) consists of bluish-greenish grey, clayey silt with a 239 moderate carbonate content. It is well laminated (lithofacies Fl) and contains remains of leaves 240 indeterminable to species. Bones of Equus sp. (horse, radius) and of Rangifer tarandus (reindeer, humerus) 241 were found in the lower part of the unit; one metacarpal of Rangifer tarandus was found in the upper part. 242 The top of unit 1 shows a significant erosional discontinuity. Furthermore, the sediments are penetrated by 243 numerous small faults and are probably tilted towards the southwest. This tilting might be tectonic and 244 associated with the fault system in the highway outcrop in the west (Fig. 2; see also section 2).

245 Unit 2 (~170 cm total thickness) consist of two subunits. Subunit 2a rests unconformably on unit 1 and 246 displays a lateral thickening from 50 to 120 cm towards NE (Fig. 4). It is composed of indistinctly bedded 247 silty fine sand and fine sandy silt (Hb) with frequent Pleistocene molluscs in the lowest layers (Fig. 5A). The 248 approximately 70 cm-thick subunit 2b shows 3-5 cm thick wavy-bedded, yellowish brown, fine sand 249 alternating with yellowish grey silt (Fig. 5B). Ripple cross-lamination sometimes appears in sand layers, 250 while occasional convolute-like bedding and root casts (up to ca. 4 cm long rhizoliths with diameters of a 251 few mm) occur throughout the unit (Hb, Hl, H-r, H-p; Figs. 5C, D). This subunit is poor in molluscs. Two well- 252 preserved tusks (267 x 15 cm in size) (Fig. 5E) and six vertebrae of Mammuthus primigenius were 253 discovered at the base of subunit 2b. The surfaces of the tusks are pale brown to rose-coloured, while the 254 core is whitish. 255 Unit 3 (~340 cm total thickness) comprises four subunits. Subunit 3a is 25-40 cm thick, largely unbedded 256 (massive) and composed of silt to fine sand (Sm) with a moderate carbonate content. The following 130 257 cm-thick subunit 3b consists of well layered, yellowish to brownish, silty, fine sand (Sb), characterised by a 258 high carbonate content. Bedding is laterally continuous and traceable along the entire outcrop (≥ 100 m) 259 The 100 cm-thick subunit 3c above is made up of yellowish brown to yellowish grey, spotted, layered silt to 260 fine sand (Sm-Sb; Figs. 3 and 4). It shows a higher carbonate content and contains black to dark brown, 261 poorly lithified Mn-concretions. The topping subunit 3d is 70 cm thick and comprises layered, yellowish 262 brown to yellowish grey, spotted, silty, fine sand with moderate carbonate content (Sb). As in subunits 3b 263 and 3c, layering in this subunit is laterally continuous over tens of metres (Fig. 4). 264 Unit 4 (~110 cm thickness) consists of yellowish brown, fine, unbedded, sandy silt (Sm); the carbonate 265 content is low. Abundant mollusc remains occur in the lower part. 266 Unit 5 (~200 cm thickness) is an alternation of layered silt and fine sand (Hb) with lenses of fine to 267 medium sand, gravel and reworked Late Miocene molluscs. The lowest layers comprise yellowish to 268 brownish, silty, fine sand rich in mollusc remains. 269 Unit 6 (~300 cm thickness) forms the top of the section and comprises a loess-like deposit with a high 270 amount of fine sand (Sm-Fm), but also contains a small portion of fine gravel to coarse sand (1-2 weight %). 271 The deposit is unbedded with a yellowish-brown hue and a high carbonate content. 272 273 4.2 Mammal fossils 274 The Bullendorf section yielded a few large mammal remains, which were found in two different layers. 275 Within the basal part of unit 1 fragmentary bones of a fossil horse (Equus sp.) and reindeer (Rangifer 276 tarandus) were found. An incomplete radius lacking the proximal end was identified as Equus sp. (Equidae, 277 Perissodactyla), but does not allow identification on the species level due to its incompleteness. Fossils of 278 the reindeer Rangifer tarandus Linnaeus, 1758 (Cervidae, Artiodactyla) consist of a distal fragmentary end 279 of a humerus and a right cannon bone (metacarpal III+IV), missing only the distal trochlea of metacarpal IV.

280 The right cannon bone was recovered from a slightly higher horizon in the upper part of unit 1. 281 Furthermore, a fragmentary rib and an undeterminable bone fragment were recovered from the basal part 282 of unit 1. Both allow no systematic identification other than Mammalia indet. 283 All remains of Mammuthus primigenius Blumenbach, 1799 (Elephantidae, Proboscidea) were embedded 284 in a horizon at the base of subunit 2b and distributed within a few square metres. The material comprises a 285 left and right tusk, four thoracic vertebrae, a processus spinosus of an additional thoracic vertebra, and two 286 lumbar vertebrae. Some of the vertebrae were still preserved in articulation. It is most likely that all these 287 skeletal elements stem from one individual. Because the molars of this individual were not found, the 288 systematic determination of the elephantid on species level is based on the OSL ages (see chapter 5.2) as 289 no other Mammuthus species other than M. primigenius is known from the Late Pleistocene of Europe. 290 Both tusks lack their tips but are preserved to their very basal end including the pulpa cavity. The curving of 291 the two tusks is variant and reveals intra-individual variability; while the left tusk shows the typical 292 helicoidal curving of mammoth tusks, the right tusk curves mostly only in one plane. Most of the vertebra 293 remains are sub-complete and damaged to different degrees.

294 Finally, a total of seven molar teeth of arvicolinae rodents were found in samples C3 (4 fragments) and

295 B6 (3 fragments). Due to the fragmentary conditions, only two teeth were identified at the species level (G.

296 Rabeder, pers. comm.): Microtus arvalis (Pallas, 1778) and Microtus gregalis (Pallas, 1779), in samples C3

297 and B6, respectively. 298 299 4.3 OSL tests, environmental dose distributions and optical ages 300 Dose recovery tests and a preheat plateau test (Murray and Wintle, 2003) were conducted on single

301 aliquots of quartz from sample O1 and O2 in order to determine the most appropriate measurement 302 conditions for the samples. The preheat plateau test revealed a clear plateau between 200°C and 240°C. 303 For the dose recovery tests, a laboratory dose of 25 Gy was administered to a sub-set of aliquots for which 304 the natural OSL signals have been removed via sunlight exposure. The single-aliquot regenerative-dose 305 (SAR) procedure was used to evaluate whether the known dose value can be recovered accurately, while 306 the preheat temperatures prior to measurement of the regenerative and test dose OSL signals were varied 307 systematically. The most accurate and precise results were obtained when a preheat of 200°C for 10 s was 308 applied prior to measurement of the natural (Ln) and each regenerative dose (Lx), and a preheat of 200°C 309 for 5 s prior to measurement of the test dose OSL signals (measured/given dose ratio of 0.99 ± 0.07). All 310 further OSL measurements were thus conducted with this preheat combination. 311 For all samples the natural and regenerated OSL signals are bright (typically 103 – 105 counts in the initial 312 signal channel) and OSL decay reaches background within the initial few seconds of stimulation, indicative 313 of a dominantly fast OSL component (Fig. 6).

314 The Environmental Dose (De) distributions for all samples are shown as radial plots in the supplement 315 and the corresponding ages are given in Table 1, together with the dose rate for each of the samples. All De 316 distributions cluster relatively tightly about a mean with overdispersion values (i.e. scatter in the De data 317 that are not accounted for by instrumental and measurement uncertainties) between 17% and 25%. 318 Overdispersion values of around 20% are deemed typical for well-bleached sediments (Olley et al., 2004) 319 and it is therefore plausible to assume that most of the OSL samples from Bullendorf were well bleached 320 prior to deposition. The central age model (CAM) was used for De calculation (Galbraith et al., 1999;

321 Galbraith and Roberts, 2012). For two samples (O2 and O5), however, a single high De outlier is present in

322 each of the radial plots (shown as triangles in the Supplement). The De values of O2 and O5 after rejecting

323 these two outliers did not change significantly (i.e. from 46.6 ± 2.8 Gy to 45.3 ± 2.4 Gy for O2 and from 42.0

324 ± 3.1 Gy to 41.6 ± 2.8 Gy for O5), but slightly increased in precision. We thus opt for using these slightly

325 more precise De values for age calculation. With the exception of sample O1, all optical ages overlap within 326 error (Tab. 1). 327 328 4.4 Radiocarbon 329 The horse radius and the bone fragment of an indeterminate mammal from the base of unit 1 yielded 330 acceptable values of 1.59 and 2.05 % N and corresponding collagen C/N values of 3.2 and 3.5 (Tab. 2). Their 331 calibrated radiocarbon ages overlap within their 2 sigma uncertainties, suggesting an age of about 37-38 cal 332 ka BP for these remains. The mammoth tusk showed very poor collagen preservation and could not be 333 dated using radiocarbon. 334 335 4.5 Stable isotope composition of gastropods 336 Oxygen isotope values of terrestrial gastropods range from -6.4 to -2.5‰ (m=-4.6‰; σ=1‰), while 337 carbon isotope values range from -9.2 to -3.4‰ (m=-6.8‰; σ=1.2‰) (Tab. 3). Oxygen isotope values of 338 freshwater gastropods (n=9) range from -10.6 to -8.0‰ (m=-9.7‰; σ=0.8‰) and carbon isotope values 339 range from -12.1 to -9.8‰ (m=-11‰; σ=0.6‰) (Tab. 3). 340 Data on the ontogenetic variability in the isotopic composition between juvenile and adult growth 341 stages are available for three species (six specimens) (Tab. 4). For each investigated species, a range of 342 <1‰ (σ=0.1‰) for δ18O and of <2‰ (σ=0.4‰) for δ13C was observed. 343 The species-dependent variability in stable isotope composition of terrestrial species was investigated

344 for samples C1, B3 (Pupilla alpicola/sterrii assemblage) and C3 (Galba truncatula assemblage). Species (n=5)

345 of samples C1 and B3 display δ18O values ranging from -5.8 to -2.5‰ (σ=0.9%) and δ13C values varying from

346 -9.2 to -5.8‰ (σ=1.1‰) (Tab. 3, Fig. 7A). Species (n=6) of samples C3 show δ18O values ranging from -6.4 to 347 -4‰ (σ=0.7‰) and δ13C values ranging from -8.3 to -5.2‰ (σ=0.8‰) (Tab. 3, Fig. 7B).

348 Data on stable isotope variability across the section are available for Pupilla muscorum (B2, C1, C3, B6)

349 and Succinella oblonga (B3, C3, B6, C7). From base to top, δ18O of Pupilla muscorum changes from high

350 values in B2 and C1 samples (m=-4.3‰; σ=0.8‰; m=-4.3‰; σ=0.8‰) to comparatively low values in C3 and

351 B6 samples (m=-5‰; σ=0.3‰; m=-5.1‰; σ=0.5‰) (Tab. 3, Fig. 7D). This shift is not present in δ13C, which

352 changes from comparatively 13C-enriched values (m=-7.3‰; σ=0.9‰; m=-6.6‰; σ=0.4‰) in samples B2

353 and C3 towards lower values (m=-8.4‰; σ=0.7‰; m=-7‰; σ=0.5‰) in samples C1 and B6. Succinella

354 oblonga displays 18O-enriched values (m=-3.5‰; σ=1‰; m=-4.8‰; σ=0.3‰) in samples B3 and B6 and 18O-

355 depleted ones (m=-4.8‰; σ=0.9‰; m=-3.2‰; σ=0.5‰) in samples C3 and C7. δ13C values show a systematic

356 decrease from sample B3 to sample C7 (m=-6.4‰; σ=0.5‰; m=-5.1‰; σ=1.3‰) (Tab. 3, Fig. 7E). 357 358 5. Discussion 359 5.1 Depositional environments of the Bullendorf succession 360 Classical LPS from the loess regions of Lower Austria are characterised by an alternation of loess and 361 various types of paleosols. In the context of the Austrian loess landscape, loess can be defined as an aeolian 362 deposit composed of silt particles (mainly quartz) with a modal grain size of about 30 µm that has 363 subsequently been aggregated by “loessification”, i.e. quasi-diagenetic and quasi-pedogenic processes 364 (Pécsi, 1990; Pye, 1995; Sprafke & Obreht, 2016). Paleosols in LPS of Lower Austria are usually classified 365 either as tundra-related soils that reflect short interstadials (e.g. during Marine Isotope Stage (MIS) 2) or as 366 cambic soils that formed during longer interstadials such as during MIS 3 (e.g. Therhorst et al., 2015). Loess 367 redeposited via slope wash, gravitational slope process and/or via solifluction is also locally encountered in 368 Austrian LPS. The sedimentological criteria for such redeposited or colluvial loess include a (coarse) sand 369 matrix, outsized clasts (of gravel size) or gravel layers, erosive basal contacts and/or a high degree of 370 fragmentation of molluscs shells if a malacological fauna is present (e.g. Frank et al., 2011; Händl et al., 371 2014; Meyer-Heintze et al., 2018). 372 In the case of the Bullendorf section no indications for pedogenesis (cambic soil or tundragley) were 373 recognised, and classical loess is absent in this section. Aeolian deposits that formed by redeposition of 374 local to regional Neogene sediments are presumably present in the top part of the section (i.e. unit 6). Unit 375 6 is too coarse for typical loess with 1-2 weight % of coarse sand and fine gravel (Fig. 3). This unit is thus 376 interpreted as loess or dune sand with some contribution originating from erosion and redeposition of 377 Neogene units (mainly Pannonian sand) from adjacent topographic highs (cf. Fig.2). 378 A number of sedimentological and granulometric observations suggest that units 1 to 5(Fig. 3) are also 379 different from classical loess and that the Bullendorf section is thus not directly comparable to regional LPS 380 sites. Rather, sediment units 1 to 3 appear to have been deposited in an aqueous environment. The 381 relevant sedimentological observations include (Fig. 3): (i) the silty to clayey and laminated character of 382 unit 1, (ii) ripple cross lamination in sandy subunit 2b, (iii) the laterally continuous layering over tens to >

383 100 m that is especially distinct in unit 1 and sub-units 3b, 3c and 3d (Fig. 4), (iv) a regular alternation of 384 sandy and silty beds with individual layer thicknesses of up to 10 cm (unit 1 and subuniuts 3b, 3c and 3d), 385 and (v) the absence of an erosive base for any of the subunits. 386 We exclude alluviation or sheet flow over frozen or unfrozen ground for sediment units 1 to 3, because 387 sedimentary features typical of alluvial and sheet flow deposits, such as normally graded beds with high 388 lateral variability, erosive contacts and sediment lenses that pinch out laterally or coarse-grained 389 components (Ruegg, 1983; Abdullatif, 1989; Gardner et al., 1991), are absent. In order to generate a 390 laterally continuous deposit with regular alternation of sandy and silty beds (subunits 3b, c and d) or silty to 391 clayey and laminated deposits (unit 1), a standing water body is required that allows silt and clay particles 392 to settleout. The same holds for ripple cross lamination (subunit 2b). The absence of erosional contacts 393 between these subunits also argues against alluvial or sheet flow deposition and supports (semi)continuous 394 sedimentation in an aqueous setting. The only discontinuity in the section is located between units 1 and 2 395 (Fig. 3) and is attributed to a prolonged period of erosion and/or non-deposition during the LGM. 396 The interpretation of the sedimentary environment of unit 4 (unbedded, sandy silt) and 5 (alternation of 397 layered silt and fine sand containing coarse grained lenses) is more ambiguous. Especially in unit 5 sandy 398 lenses with gravel and reworked Late Miocene molluscs point to sheet flows from adjacent Neogene hills. 399 In summary, we suggest that sediment units 1 to 3 were deposited in a low-energy, aquatic environment 400 e.g. a water-filled depression related to a fluvial system (e.g. an oxbow lake) or a (semi)closed depression of 401 local extent (Fig. 3). As outlined in section 2, the long-term course of the Zaya River has been controlled by 402 the pull-apart tectonics of the Vienna basin, and it is conceivable that the Zaya river was shifted 403 significantly to the northwest during the Late Pleistocene and was thus much closer to the Bullendorf site. 404 We suspect that neotectonic activity and the Zaya river dynamics played an important role in controlling 405 the sedimentary processes at the study site. For the topmost units (units 4 and 5) we invoke sheet-flow 406 processes as a possible depositional mechanism, while unit 6 is an aeolian deposit mainly composed of 407 reworked Neogene sediments. 408 Despite the depositional age of the individual units constrained to immediately before (unit 1) and after 409 the Last Glacial Maximum (LGM) (units 2 to 6), evidence of cryoturbation is scarce (indistinct and non- 410 pervasive convolute bedding in subunit 2b) or completely absent in this succession. 411 412 5.2 Chronology of the Bullendorf section 413 Biostratigraphic constraints of the section are provided by the fossil mammal remains. Pleistocene 414 horses are known in Central Europe throughout the Pleistocene from both glacials and interglacials. During 415 the early and middle Weichselian/Würmian the large fossil horse E. germanicus populated Central Europe 416 and was replaced only after the LGM, around 22,000-18,000 BP, by E. ferus (Koenigswald, 2002). Based on 417 the radiocarbon dating of the horse radius it most likely represents E. germanicus. Rangifer tarandus

418 inhabited Central Europe since the Middle Pleistocene during the glacial periods, while it was not present 419 during interglacials (Koenigswald, 2002). 420 From the late Marine Isotope Stage (MIS) 7 or the beginning of MIS 6 (i.e., from about 200 to 160 ka), M. 421 primigenius spread over Europe after emigrating from Asia (Kahlke, 2015) and finally disappeared in Europe 422 around 12,000 years ago (Stuart et al., 2002). The fossil record of M. primigenius is much more abundant 423 before the LGM than afterwards (Koenigswald, 2002). 424 Microtus gregalis and M. arvalis first appeared in the Middle Pleistocene (Björn, 1968). 425 The occurrence of Equus sp., R. tarandus (in unit 1), M. primigenius (in unit 2b), as well as of arvicolidae 426 (in units 3b and 4), restrict the deposition of the Bullendorf succession to the Late Pleistocene. This 427 biostratigraphic assignment is supported by radiocarbon dating of two fossil mammal remains from unit 1, 428 suggesting an age of about 36-38 ka BP for the base of the Bullendorf sediment succession.

429 OSL ages of the samples O2 to O5 are consistent with each other at the 2 sigma error level and the grand

430 mean age of these five samples is 18 ± 1.8 ka (Tab. 1). Sample O1 (25.1 ± 2 ka) is from the subunit 2b and

431 thus overlies sample O2 (18.3 ± 1.6) that was retrieved from subunit 2a. Partial bleaching of sample O1 is 432 excluded as a cause for this age discrepancy of ca. 7 ka, because (i) this sediment layer was deposited in a 433 low-energy shallow-water environment, where incomplete bleaching of the quartz OSL signal is unlikely,

434 and (ii) the De distribution of OSL sample O1 is tightly clustered around a mean De value of ~54 Gy with an 435 overdispersion value of only 17%, suggestive of a single well bleached dose population.

436 A possible explanation for O1 being older by ~28% than O2 and the rest of the OSL samples is radioactive

437 disequilibrium in the Uranium decay chain of sample O1 and/or spatial heterogeneity in the environmental

438 radiation field at the sampling site e.g. due to the vicinity of enamel and in particular bones. Sample O1 was 439 taken within ~ 1 m of the mammoth tusks and Uranium uptake by collagen enamel of the mammalian 440 bones and tusks (particularly during the early period of sediment burial) could have temporarily changed 441 the environmental dose rate and may thus have partly contributed to the observed discrepancy in OSL ages 442 (Millard and Hedges, 1996). In particular, any fossil bones or tusks that were present immediately adjacent

443 to sample O1, but were destroyed during construction of the highway, would have also impacted on the

444 environmental radiation field and thus the dose rate of sample O1. In our study the gamma dose rate was 445 determined by in-situ field gamma spectrometry. This approach captures any spatial heterogeneity in the 446 environmental radiation field if all sources responsible for the spatially heterogeneous radiation field are

447 still present in the outcrop. If bones or tusks were initially present next to sample O1 but later 448 removed/destroyed, an in-situ approach might not accurately reflect the average burial dose rate received 449 by the sample and will thus result in an inaccurate OSL age. 450 The OSL chronology supports the biostratigraphic data suggesting deposition of the sediment sequence

451 briefly after the LGM (grand mean age of 18 ± 1.8 ka). However, one OSL sample (O1) from the lower part 452 of the succession and taken from the vicinity of the mammoth tusks is slightly older than this grand mean

453 age, i.e. ca. 25 ka. Taking together, these OSL ages imply a high sedimentation rate for this sediment 454 succession and time period and constrain the age of the mammoth find at Bullendorf to ca. 18-25 ka. 455 456 5.3 Stable isotope composition of gastropods 457 Species-specific and intra-shell stable isotope composition. The low ontogenetic variability of δ18O is in 458 accordance with the range measured by Kehrwald et al. (2010) in gastropods from the LGM and no 459 systematic trend of enrichment or depletion in stable isotopes between juvenile and adult growth stages 460 was observed. Hence, ontogenetic effects are negligible in our material. Therefore, the values are regarded 461 as representative of the complete suite of shells. 462 Species-specific and intra-sample oxygen isotope composition. The observed variability in δ18O values (1- 463 2‰) of modern land snail populations living at the same place (Lécolle, 1985; Goodfriend and Magaritz, 464 1987; Yanes et al., 2011) seems to be related to factors including duration of activity, life cycle and water 465 sources, rather than species-specific offsets (Zanchetta et al., 2005; Colonese et al., 2007, 2014). It has been 466 observed that the ecological niches, occupied by different populations at the same site, can additionally 467 account for differences in the oxygen isotope composition (Goodfriend et al., 1989; Balakrishnan et al., 468 2005b; Yanes et al., 2009; Colonese et al., 2014). Inter-specific variability in δ18O of coeval species are also

469 evident at the Bullendorf site. In samples C1 and B3, Pupilla sterrii and P. loessica show the highest δ18O 470 values, whilst P. muscorum displays lower values. The differences between these species might be an 471 expression of different ecological niches. Pupilla sterrii and P. loessica are cryophilous species, usually living 472 in open dry meadows where evaporation is important. Conversely, P. muscorum has a wider range of 473 microhabitats, ranging from relatively moist to dry and sunny habitats. Likewise, the wide range of δ18O 474 values of Succinella oblonga might reflect its tendency as a generalist species.

475 In sample C3 the mean δ18O value of Columella columella is consistent with the ecological niche of this 476 hygrophilous species. The higher δ18O values measured in Trochulus hispidus, living likewise in a moist 477 habitat, could reflect differences in both temperature and/or vegetation preferences between these two 478 species. Pupilla alpicola shows slightly more 18O-depleted values compared to P. muscorum, indicating the 479 preference of this species for moister habitats. The comparatively low values of Vallonia tenuilabris might 480 reflect its preference for much drier habitats. Galba truncatula shows a wide range of δ18O values (Fig.7C). 481 Since the δ18O value of freshwater molluscs mainly depends on the isotopic composition and temperature 482 of the water in which the aragonitic shell formed (Jones et al., 2002; Leng and Marshall, 2004; Leng and 483 Lewis, 2014), the variety of aquatic habitats occupied by this species could easily explain the rather large 484 range as specimens from pond water would have dwelt in distinctly cooler water than those from shallow 485 puddles affected by evaporation. 486

487 Species-specific and intra-sample carbon isotope composition. The carbon isotope composition of food 488 ingested by land snails can be estimated applying the empirical relationship between the δ13C of vegetation

489 (C3 plants -33 to -21‰, C4 plants -17 to -9‰; Farquhar et al., 1989) and δ13C of shells devised by Stott 490 (2002) for Cornu aspersum:

13 13 491 δ Cshell = 1.35*δ Cdiet –11.73 492 Although it is debatable if this equation can be transferred to other taxa due to potential species- 493 specific isotope fractionation, it provides a first-order estimate to discriminate between coeval taxa.

494 In samples C1 and B3, δ13C-diet values of P. loessica and P. sterrii are higher than for P. muscorum.

495 Elevated δ13C values are usually associated with water-stressed C3 plants living in (semi)arid environments.

496 Succinella oblonga displays δ13C-diet values ranging from -21.1 to -19.8‰, probably also related to C3

497 plants suffering from water stress rather than a C4 plant-based diet. The δ13C-diet values of Helicopsis

498 striata reflect foraging on C3 plants as well.

499 In sample C3, the ecological demands of the species assemblage indicate cold and rather humid 500 conditions (Carobene et al., 2018). Therefore, the slightly higher δ13C-diet values of S. oblonga and P.

501 muscorum compared to samples C1 and B3 can neither be explained by increased aridity nor by a

502 predominant C4 plant community. Vallonia tenuilabris, Columella columella and Pupilla alpicola show the

503 lowest δ13C-diet values indicating a C3 plant-based diet. The rather low isotope values of Trochulus hispidus 504 might record different feeding habits due to the variety of ecological niches occupied by this species.

505 In sample C3 Galba truncatula displays δ13C values between -12.1 and -9.8‰. The δ13C value of 506 freshwater molluscs is mainly governed by the carbon isotope composition of the dissolved inorganic 507 carbon (DIC) of the ambient water (McConnaughey and Gillikin, 2008). However, dietary carbon and 508 metabolic process may also influence the carbon isotopic composition of shells. Galba truncatula has an 509 herbivorous diet, which includes grazing on algal biofilms (phototrophic biofilms δ13C ranging from -36 to -

510 19‰; Staal et al., 2007) and/or feeding on shoreline plants (C3 plants with δ13C values ranging from -33 to - 511 21‰ and macrophytes whose δ13C values range from -30 to -12‰; Leng and Marshall, 2004). The low δ13C 512 values measured in G. truncatula might reflect some contribution of carbon from submerged macrophytes

13 513 on which the gastropod fed. The higher δ C values, in contrast, might be related to CO2 exchange between 514 the atmosphere and the DIC during times of enhanced evaporation (Leng and Marshall, 2004). 515 516 5.4 Palaeoenvironmental evolution at Bullendorf 517 Available sedimentological and chronological data, fossil mammal remains and the stable isotope 518 composition of gastropods form the basis for an assessment of the palaeoenvironment during deposition of 519 units 1 to 5. 520 Unit 1. Equus germanicus is a typical faunal element of the mammoth steppe but given its presence also 521 during the last interglacial this species also inhabited forests. R. tarandus is an extant species of reindeer,

522 which today is restricted to the Holarctic and occupies taiga and tundra regions. Since R. tarandus was 523 absent during interglacials (Koenigswald, 2002), the presence of this species at Bullendorf is consistent with 524 a mammoth-steppe environment during the deposition of unit 1. 525 Unit 2. Mammuthus primigenius is the most prominent member of the cold-adopted Late Pleistocene 526 fauna which inhabited the so-called mammoth steppe biotope. In accordance with the 527 palaeoenvironmental reconstruction based on the malacofaunal composition (Pupilla alpicola/sterrii 528 assemblage; Carobene et al., 2018), the comparatively high δ18O values of Pupilla muscorum and Succinella

529 oblonga indicate dry and cold conditions in this steppe. The δ13C diet values of both species reveal that C3 530 plants most probably dominated the local vegetation.

531 Unit 3. The low δ18O values of both Pupilla muscorum and Succinella oblonga in sample C3 are probably 532 related to an increase in relative humidity rather than particularly cold temperatures (Balakrishnan and 533 Yapp, 2004; Balakrishnan et al., 2005a). This hypothesis is supported by the occurrence of the Galba 534 truncatula assemblage (Carobene et al., 2018) which indicates a shift towards rather cold but moderately 535 moist conditions and the presence of at least small water bodies. Furthermore, the presence of Microtus 536 arvalis, a generalist species which occupies a variety of open habitats such as moist meadows, pastures, 537 forest steppes and moist forests supports this ecological interpretation. The δ13C diet mean value (20.7‰)

538 of both P. muscorum and S. oblonga is slightly above the range of δ13C values of C3 plants. Nevertheless,

539 the values are too low to be the result of a contribution by C4 plants.

540 Unit 4. In sample B6, the δ18O values of both species show relatively low values. Based on the occurrence 541 of Pupilla alpicola/sterrii assemblages (Carobene et al., 2018), the values might be indicative of isotopically 542 depleted precipitation (Rozanski et al., 1993; Balakrishnan et al., 2005a) rather than an increase in relative 543 humidity. Within the mammal assemblage, the presence of Microtus gregalis, a typical loess species usually 544 associated with cold, dry grassland, supports the hypothesis. S. oblonga shows slightly higher δ13C diet 545 values compared to P. muscorum.

546 Unit 5. The highest values of δ18O of Succinella oblonga recorded in sample C7 suggest the establishment

547 of dry and cold conditions. The relatively high δ13C diet values might be related to C3 plants suffering from 548 water stress due to these arid conditions. 549 550 5.5 Palaeoclimate constraints 551 The stable isotope composition of freshwater gastropods shells reflects climate conditions during the 552 warm season when snails are active and build their shells. This period ranges from spring to fall and reflects 553 the average growing season (AGS) temperature (Banak et al., 2016). For aragonitic freshwater molluscs, 554 Grossman and Ku (1986) obtained this temperature equation: 555

18 18 556 (1) T° (°C) = 21.8 – 4.69 (δ Oaragonite - δ Owater)

557 558 A high correlation between air and surface water temperature has been observed on both short and long 559 timescales (McCombie, 1959; von Grafenstein et al., 1996; Livingstone and Dokulil, 2001; Sharma et al., 560 2008). In order to solve the above equation the isotopic composition of the palaeowater needs to be 561 known. We use the value of Late Pleistocene (broadly LGM-age) groundwater from regional aquifers as an 562 approximation, because palaeo-groundwaters reflect the weighted mean of the isotope composition of 563 palaeo-precipitation (Rozanski, 1985; Darling, 2004; Jasechko et al., 2015). LGM-age groundwater in 564 northern Austria, southern Germany and northern Switzerland yielded a rather uniform δ18O value of about 565 -13‰ VSMOW (summarised by Darling, 2004). Under the assumption that the isotopic composition of the 566 water from which the freshwater species G. truncatula secreted its aragonite shell had an isotopic 567 composition equivalent to the weighted mean of the isotope composition of palaeo-precipitation at that 568 time (-13‰) and that secondary effects such as evaporation did not play a major role, a temperature of 569 6.2°C (2.7 to 9.8°C) can be calculated. This range reflects the 2 sigma standard deviation of the measured 570 samples of G. truncatula (Table 3) and is, strictly speaking, only valid for the horizon from which sample C3 571 was obtained (Fig. 3). For comparison, the mean July temperature at Poysdorf about 8 km N of Bullendorf 572 today is of 20.3°C (period of observation 1981-2010, data from ZAMG). 573 574 6. Summary and conclusions 575 The Bullendorf site comprises a 13.6 m-thick succession of fine-grained sediments deposited within a 576 few millennia before (around ca. 34 ka BP; unit 1) and after the LGM (ca. 18 ± 1.8 ka BP, units 2 to 6). The 577 LGM is represented by an erosional discontinuity in the sequence, separating unit 1 from the rest of the 578 succession. OSL ages of units 2 to 6 overlap within their error margins, suggesting that the sediment 579 accumulation rate for the post-LGM period was high (~6.2 m in ca. 2-4 ka). This high rate is tentatively 580 related to neotectonic movements along faults ~250 m southwest of the Bullendorf section, where a 581 paleosol has been tectonically tilted and sheared. Further investigations are needed to constrain the timing, 582 displacement rates and recurrence intervals of these faults and their influence on the palaeotopography of 583 the region. 584 The basal 10.5 m of the succession were deposited in a low-energy aquatic environment in a local 585 depression in the vicinity of the precursor of today’s Zaya River, followed by sheet-flow deposits in the 586 upper part of the section. The sediment succession preserved faunal remains including two tusks of 587 Mammuthus primigenius. The top 3 m of the succession are composed of loess-like sediments from 588 redeposited Neogene deposits. 589 590 The following palaeoecological conclusions can be drawn from the mollusc assemblages: (i) two phases 591 of slightly more favourable climate can be discerned, but their timing cannot be chronologically resolved,

592 (ii) changes in δ18O of both Pupilla muscorum and Succinella oblonga across the section suggest an 593 alternation of dry-cold and moist-cold climate conditions, (iii) the δ18O compositions of the species reflect a 594 range of microenvironments, (iv) the carbon isotope composition of the gastropod shells indicates a C3 595 vegetation, and (v) estimated mean average growing season temperature range suggests a drastic decrease 596 in air temperature compared to today. 597 These palaeoecological observations and the presence of a characteristic cold-steppe fauna at 598 Bullendorf are in-line with the broad palaeoclimatic picture for Central Europe for this time interval (e.g. 599 Heiri et al., 2014; Nigst et al., 2014). With the exception of subunit 2b, which shows convolute bedding, 600 which may also be interpreted as indication of slight cryoturbatic overprinting, no other evidence of 601 cryoturbation was found in the succession. This is surprising given the high water content of the aquatic 602 sediment units and the cold and arid climatic conditions that prevailed during the LGM. This implies that 603 the local climate immediately before (around ca. 34 ka, unit 1) and after the LGM (around ca. 18 ka BP) was 604 sufficiently mild to prevent continuous permafrost from forming in this part of the Vienna basin. 605 606 Acknowledgment 607 We are grateful to ASFINAG and especially to Klaus Schierhackl for permission to perform fieldwork 608 during the road constructions and for support during the excavation. Many thanks to Gernot Rabeder 609 (University Vienna) for identification of the arvicolinae species and to Jürgen Reitner (Geological Survey of 610 Austria) for discussion on the sediment succession. We are grateful to Tobias Sprafke and an anonymous 611 reviewer for their careful reviews and constructive remarks. 612 613 References 614 Abdullatif, O.M., 1989. Channel-fill and sheet-flood facies sequences in the ephemeral terminal River 615 Gash, Kassala, Sudan. Sedimentary Geology 63, 171-184. 616 Adamiec, G., Aitken, M., 1998. Dose-rate conversion factors: update. Ancient TL 16, 37-50. 617 Andrews, J.N., Balderer, W., Bath, A.H., Clausen, H.B., Evans, G.V., Florkowski, T., Goldbrunner, H.E, 618 Ivanovich, M., Loosli, H., Zojer, H., 1984. Environmental isotope studies in two aquifer systems: a 619 comparison of groundwater dating methods. Proceedings of Isotope Hydrology Conference 1983. 620 International Atomic Energy Agency, Vienna, pp. 535–576. 621 Balakrishnan, M., Yapp, C.J., 2004. Flux balance models for the oxygen and carbon isotope compositions 622 of land snail shells. Geochimica et Cosmochimica Acta 68, 2007-2024. 623 Balakrishnan, M., Yapp, C.J., Meltzer, D.J., Theler, J.L., 2005a. Paleoenvironment of the Folsom 624 archaeological site, New Mexico, USA, approximately 10,500 14C yr B.P. as inferred from the stable isotope 625 composition of fossil land snail shells. Quaternary Research 63, 31-44.

626 Balakrishnan, M., Yapp, C.J., Theler, J.L., Carter, B.J., Wyckoff, D.G., 2005b. Environmental significance of 627 13C/12C and 18O/16O ratios of modern land-snail shells from the southern great plains of North America. 628 Quat. Res. 63, 15-30. 629 Banak, A., Mandic, O., Sprovieri, M., Lirer, F., Pavelić, D., 2016. Stable isotope data from loess 630 malacofauna: Evidence for climate changes in the Pannonian Basin during the Late Pleistocene. Quaternary 631 International 415, 15-24. 632 Bednarik, R., 1989. The Galgenberg figurine from Krems, Austria. Rock Art Research 6, 118-125. 633 Björn, K., 1968. Pleistocene Mammals of Europe. Aldine, Chicago. 634 Bøtter-Jensen, L., Mejdahl, V., 1988. Assessment of beta dose-rate using a GM multicounter system. 635 Nuclear Tracks and Radiation Measurements 14, 187-191. 636 Bøtter-Jensen, L., Bulur, E., Duller, G.A.T., Murray, A.S., 2000. Advances in luminescence instrument 637 systems. Radiation Measurements 32, 523-528. 638 Brennan, R., Quade, J., 1997. Reliable late-Pleistocene stratigraphic ages and shorter groundwater travel 639 times from 14C in fossil snails from the southern Great Basin. Quaternary Research 47, 329-336. 640 Bronk, R.C., Higham, T., Bowles, A., Hedges, R., 2004. Improvements to the pretreatment of bone at 641 Oxford. Radiocarbon 46, 155-63. 642 Brown, T.A., Nelson D.E., Vogel J.S., Southon J.R. 1988. Improved collagen extraction by modified 643 Longin method. Radiocarbon 30, 171-7. 644 Bruhn F., Duhr, A., Grootes, PM., Mintrop, A., Nadeau, M.J. 2001. Chemical removal of conservation 645 substances by 'soxhlet'-type extraction. Radiocarbon 43, 229-237. 646 Carobene, D., Harzhauser, M., Mandic, O., Gatto, R., 2018. Taxonomy and palaeoecology of continental 647 Gastropoda (Mollusca) from the Late Pleistocene mammoth-bearing site of Bullendorf in NE Austria. Rivista 648 Italiana di Paleontologia e Stratigrafia 124, 509-534. 649 Colonese, A.C., Zanchetta, G., Fallick, A.E., Martini, F., Manganelli, G., Lo Vetro, D., 2007. Stable isotope 650 composition of Late Glacial land snail shells from Grotta del Romito (Southern Italy): Palaeoclimatic 651 implications. Palaeogeography, Palaeoclimatology, Palaeoecology 254, 550-560. 652 Colonese, A.C., Zanchetta, G., Fallick, A.E., Manganelli, G., Lo Cascio, P., Hausmann, N., Baneschi, I., 653 Regattieri, E., 2014. Oxygen and carbon isotopic composition of modern terrestrial gastropod shells from 654 Lipari Island, Aeolian Archipelago (Sicily). Palaeogeography, Palaeoclimatology, Palaeoecology 394, 119- 655 127. 656 Darling, W.G., 2004. Hydrological factors in the interpretation of stable isotopic proxy data present and 657 past: A European perspective. Quaternary Science Reviews 23, 743-770. 658 Decker, K., Peresson, H., Hinsch, R., 2005. Active tectonics and Quaternary basin formation along the 659 Vienna Basin Transform fault. Quaternary Science Reviews, 24, 307-322.

660 Duller, G.A.T., 2003. Distinguishing quartz and feldspar in single grain luminescence measurements. 661 Radiation Measurements 37, 161-165. 662 Einwögerer, T., Friesinger, H., Händel, M., Neugebauer-Maresch, C., Simon, U., Teschler-Nicola, M., 663 2006. Upper Palaeolithic infant burials. Nature 444, 285. 664 Farquhar, G.D., Ehleringer, J.R.R., Hubick, K.T., 1989. Carbon isotope discrimination and photosynthesis. 665 Annual Review of Plant Physiology 40, 503-537. 666 Fink, J., 1976. Exkursion durch den österreichischen Teil des nördlichen Alpenvorlandes und den 667 Donauraum zwischen Krems und Wiener Pforte. Mitteilungen der Kommission für Quartärforschung der 668 Österreichischen Akademie der Wissenschaften 1, 1-113. 669 Fink, J., Kukla G.J., 1977. Pleistocene climates in central Europe: At least 17 interglacials after the 670 Olduvai event. Quaternary Research 7, 363-371. 671 Folk, R.L., Ward, W.C., 1957. A study in the significance of grain-size parameters. Journal of Sedimentary 672 Petrology 27, 3-26. 673 Frank, C., Terhorst, B., Damm, B., Diel, C., Frechen, M., Peticzka, R., 2011. Pleistocene loess deposits and 674 mollusc assemblages in the Eastern Pre-Alps. Quaternary Science Journal 60, 126-136. 675 Füchtbauer, H., 1959. Zur Nomenklatur der Sedimentgesteine. Erdöl und Kohle, 12, 605-613. 676 Galbraith, R.F., Roberts, R.G., Laslett, G.M., Yoshida, H., Olley, J.M., 1999. Optical dating of single and 677 multiple grains of quartz from Jinmium rock shelter, northern Australia: Part I, experimental design and 678 statistical models. Archaeometry 41, 339-364. 679 Galbraith, R.F., Roberts, R.G., 2012. Statistical aspects of equivalent dose and error calculation and 680 display in OSL dating: An overview and some recommendations. Quaternary Geochronology 11, 1-27. 681 Gardner, T.W., Ritter, J.B., Shuman, C.A., Bell, J.C., Sasowsky, K.C. and Pinter, N., 1991. A periglacial 682 stratified slope deposit in the valley and ridge province of central Pennsylvania, USA: Sedimentology, 683 stratigraphy, and geomorphic evolution. Permafrost and Periglacial Processes 2, 141-162. 684 Goodfriend, G.A., Magaritz, M., 1987. Carbon and oxygen isotope composition of shell carbonate of 685 desert land snails. Earth and Planetary Science Letters 86, 377-388. 686 Goodfriend, G.A., Magaritz, M., Gat, J. R., 1989. Stable isotope composition of land snail body water and 687 its relation to environmental waters and shell carbonate. Geochimica et Cosmochimica Acta 53, 3215-3221. 688 Grill, R., Bachmayer, F., Friedl, K., Janoschek, R., Kapounek, J., 1961. Geologische Karte des nordöstlichen 689 Weinviertels 1:75.000. Geological Survey, Vienna. 690 Grill, R., 1968. Erläuterungen zur Geologischen Karte des nordöstlichen Weinviertels und zu Blatt 691 Gänserndorf. Geological Survey, Vienna. 692 Grossman, E.L., Ku, T.L., 1986. Oxygen and carbon isotope fractionation in biogenic aragonite: 693 temperature effects. Chemical Geology 59, 59-74.

694 Guerin, G., Mercier, N., 2011. Determining gamma dose rates by field gamma spectroscopy in 695 sedimentary media: Results of Monte Carlo simulations. Radiation Measurements 46, 190-195. 696 Händel, M., Einwögerer, T., Simon, U. and Neugebauer-Maresch, C., 2014. Krems-Wachtberg 697 excavations 2005–12: main profiles, sampling, stratigraphy, and site formation. Quaternary international 698 351, 38-49. 699 Harzhauser, M., Daxner-Höck, G., Piller, W., 2004. An integrated stratigraphy of the Pannonian (Late 700 Miocene) in the Vienna Basin. Austrian Journal of Earth Sciences 95/96, 6-19. 701 Harzhauser, M., Piller, W., 2004. Integrated stratigraphy of the Sarmatian (Upper Middle Miocene) in 702 the western Central Paratethys. Stratigraphy 1, 1-22. 703 Havlíček, P., Holásek, O., Smolíková, L., Roetzel, R., 1998. Zur Entwicklung der Quartärsedimente am 704 Südostrand der Böhmischen Masse in Niederösterreich. Jahrbuch Geologische Bundesanstalt 141, 51-71. 705 Havlíček, P., Vachek, M., 2016. Bericht über geologische Aufnahmen in quartären Aufschlüssen mit 706 Paläoböden entlang der A5 Nordautobahn im Bereich Poysdorf. Unpublished report. Scientific archive A 707 18965-R, 39 p., Geological Survey, Vienna. 708 Heiri, O., Koinig, K.A., Spötl, C., Barrett, S., Brauer, A., Drescher-Schneider, R., Gaar, D., Ivy-Ochs, S., 709 Kerschner, H., Luetscher, M., Moran, A., Nicolussi, K., Preusser, F., Schmidt, R., Schoeneich, P., Schwörer, C., 710 Sprafke, T., Terhorst, B., Tinner, W. (2014): Palaeoclimate records 60-8 ka in the Austrian and Swiss Alps 711 and their forelands. Quaternary Science Reviews 106, 186-205. 712 Jacobs, Z., Duller, G.A.T., Wintle, A.G., 2006. Interpretation of single grain De distributions and 713 calculation of De. Radiation Measurements 41, 264-277. 714 Jasechko, S., Lechler, A., Pausata, F.S.R., Fawcett, P.J., Gleeson, T., Cendon, D.I., Galewsky, J., LeGrande, 715 A.N., Risi, C., Sharp, Z.D., Welker, J.M., Werner, M., Yoshimura, K., 2015. Late-glacial to late-Holocene shifts 716 in global precipitation δ18O. Climate of the Past 11, 1375-1393. 717 Jones, M.D., Leng, M.J., Eastwood, W.J., Keen, D.H., Turney, C.S. M., 2002. Interpreting stable-isotope 718 records from freshwater snail-shell carbonate: a Holocene case study from Lake Gölhisar, Turkey. The 719 Holocene 12, 629-634. 720 Kahlke, R.D., 2015. The maximum geographic extension of Late Pleistocene Mammuthus primigenius 721 (Proboscidea, Mammalia) and its limiting factors. Quaternary International 379, 147-154. 722 Kehrwald, N.M., McCoy, W.D., Thibeault, J., Burns, S. J., Oches, E.A., 2010. Paleoclimatic implications of 723 the spatial patterns of modern and LGM European land-snail shell δ18O. Quaternary Research 74, 166-176. 724 Keller, B., 1996. Lithofazies-Codes für die Klassifikation von Lockergesteinen. Mitteilungen der 725 Schweizerischen Gesellschaft für Boden- und Felsmechanik, 132, 5-12. 726 Koenigswald, W.Von., 2002. Lebendige Eiszeit. Theiss, Stuttgart. 727 Kovanda, J., Smolíková, L., Horáček, I., 1995. New data on four classic loess sequences in Lower Austria. 728 Sborník geol. věd, Antropozoikum 22, 63-85.

729 Kukla, G.J., 1975. Loess stratigraphy of central Europe. In: Butzer, K.W., Isaac, L.I. (Eds.), After 730 Australopithecines. Mouton Publishers, The Hague. 731 Lécolle, P., 1985. The oxygen isotope composition of land snail shells as a climatic indicator: applications 732 to hydrogeology and paleoclimatology. Chem. Geol. 58, 157-181. 733 Leng, M.J., Marshall, J.D., 2004. Palaeoclimate interpretation of stable isotope data from lake sediment 734 archives. Quaternary Science Reviews 23, 811-831. 735 Leng, M.J., Lewis, J.P., 2014. Oxygen isotopes in molluscan shell: Applications in environmental 736 archaeology. Environmental Archaeology 21, 295-306. 737 Livingstone, D.M., Dokuli, M.T., 2001. Eighty years of spatially coherent Austrian lake surface 738 temperatures and their relationship to regional air temperature and the North Atlantic Oscillation. 739 Limnology and Oceanography 46, 1220-1227. 740 Martinez De La Torre, H.A., Reyes, A.V., Zazula, G.D., Froese, D.G., Jensen, B.J.L., Southon, J.R. 2019. 741 Permafrost-preserved wood and bone: Radiocarbon blanks from Yukon and Alaska. Nuclear Instruments 742 and Methods in Physics Research Section B: Beam Interactions with Materials and Atoms 455, 154-157. 743 McConnaughey, T.A., Gillikin, D.P., 2008. Carbon isotopes in mollusk shell carbonates. Geo-Marine 744 Letters 28, 287-299. 745 McCombie, M., 1959. Some relations between air temperatures and the surface water temperatures of 746 lakes. Limnology and Oceanography 4, 252-258. 747 Mercier, N., Falguères, C., 2007. Field gamma dose-rate measurement with a NaI(Tl) detector: re- 748 evaluation of the "threshold" technique. Ancient TL 25, 1-4. 749 Meyer-Heintze, S., Sprafke, T., Schulte, P., Terhorst, B., Lomax, J., Fuchs, M., Lehmkuhl, F., Neugebauer- 750 Maresch, C., Einwögerer, T., Händel, M. and Simon, U., 2018. The MIS 3/2 transition in a new loess profile 751 at Krems-Wachtberg East–A multi-methodological approach. Quaternary international 464, 370-385. 752 Miall, A.D., 1984. Principles of Sedimentary Basin Analysis. 460 pp., Springer, New York. 753 Millard, A.R., Hedges, R.E., 1996. A diffusion-adsorption model of uranium uptake by archaeological 754 bone. Geochimica et Cosmochimica Acta 60, 2139-2152. 755 Moore, D.M., Reynolds, R.C., 1997. X-ray diffraction and identification and analysis of clay minerals. 756 Oxford University Press, New York. 757 Müller, G., 1961. Das Sand-Silt-Ton Verhältnis in rezenten marinen Sedimenten. Neues Jahrbuch für 758 Mineralogie 1961, 148-163. 759 Murray, A.S., Wintle, A.G., 2000. Luminescence dating of quartz using an improved single-aliquot 760 regenerative-dose protocol. Radiation Measurements 32, 57-73. 761 Murray, A.S., Wintle, A.G., 2003. The single aliquot regenerative dose protocol: potential for 762 improvements in reliability. Radiation Measurements 37, 377-381.

763 Nehyba, S. and Roetzel, R., 2004. The Hollabrunn-Mistelbach Formation (Upper Miocene, Pannonian) in 764 the Alpine-Carpathian Foredeep and the Vienna Basin in Lower Austria – An example of a coarse-grained 765 fluvial system. Jahrbuch der Geologischen Bundesanstalt 144, 191-221. 766 Nigst, P.R., Viola, T.B., Haesaerts, P., Trnka, G., 2008. Willendorf II. Wissenschaftliche Mitteilungen aus 767 dem Niederösterreichischen Landesmuseum 19, 31-58. 768 Nigst, P.R., Haesaerts, P., Damblonc, F., Frank-Fellnerd, C., Mallole, C., Viola, B., Götzinger, M., Nivena, 769 L., Trnka, G. and Hublin, J.J., 2014. Early modern human settlement of Europe north of the Alps occurred 770 43,500 years ago in a cold steppe-type environment. PNAS 111, 14394-14399. 771 Olley, J.M., Pietsch, T., Roberts, R.G. 2004. Optical dating of Holocene sediments from variety of 772 geomorphic settings using single grains of quartz. Geomorphology 60, 337-358. 773 Pécsi, M., 1990. Loess is not just the accumulation of dust. Quaternary International 7-8, 1-21. 774 Peticzka, R., Holawe, F., Riegler, D., 2010. Structural analyses on the modified paleosol-sequence of 775 “Stillfried B” with high resolution measurements of selected laboratory parameters. Quaternary 776 International 222, 168-177. 777 Prescott, J.R., Hutton, J.T. 1994. Cosmic ray contributions to dose rates for luminescence and ESR dating: 778 Large depths and long-term time variations. Radiation Measurements 23, 497-500. 779 Pye, K. 1995. The nature, origin and accumulation of loess. Quaternary Science Reviews 7-8, 653-667. 780 Rasser, M.W., Harzhauser, M. Anistratenko, O.Y., Anistratenko, V.V., Bassi, D., Belak, M., Berger, J.P., 781 Bianchini, G., Čičić, S., Ćosović, V., Doláková, N., Drobne, K., Filipescu, S., Gürs, K., Hladilová, S., Hrvatović, 782 H., Jelen, B., Kasiński, J.R., Kováč, M., Kralj, P., Marjanac, T., Márton, E., Mietto, P., Moro, A., Nagymarosy, 783 A., Nebelsick, J.H., Nehyba, S., Ogorelec, B., Oszczypko, N., Pavelić, D., Pavlovec, R., Pavšič, J., Petrová, P., 784 Piwocki, M., Poljak, M., Pugliese, N., Redžepović, R., Rifelj, H., Roetzel, R., Skaberne, D., Ľ Sliva, Standke, G., 785 Tunis, G., Vass, D., Wagreich, M., Wesselinghet, F. 2008. Paleogene and Neogene. In: McCann, T., 2008. The 786 Geology of Central Europe: Mesozoic and Cenozoic. The Geological Society of London, London. 787 Reimer, P.J., Bard, E., Bayliss, A., Beck, J.W., Blackwell, P.G., Bronk, R.C., Buck, C.E., Cheng, H., Edwards, 788 R.L., Friedrich, M., Grootes, P.M., Guilderson, T.P., Haflidason, H., Hajdas, I., Hatté, C., Heaton, T.J., 789 Hoffmann, D.L., Hogg, A.G., Hughen, K.A., Kaiser, K.F., Kromer, B., Manning, S.W., Niu, M., Reimer, R.W., 790 Richards, D.A., Scott, E.M., Southon, J.R., Staff, R.A., Turney, C.S.M., van der Plicht, J. 2013. IntCal13 and 791 Marine13 radiocarbon age calibration curves 0–50,000 years cal BP. Radiocarbon 55, 1869-1887. 792 Rozanski, K., 1985. Deuterium and oxygen-18 in European groundwaters - links to atmospheric 793 circulation in the past. Chemical Geology 52, 349-363. 794 Rozanski, K., Araguás-Araguás, L., Gonfiantini, R., 1993. Isotopic patters in modern global precipitation. 795 In: Swart, P.K. et al. (Eds.) Climate change in continental isotopic records. AGU, Washington, pp. 1-36.

796 Ruegg, G.H., 1983. Periglacial eolian evenly laminated sandy deposits in the Late Pleistocene of NW 797 Europe, a facies unrecorded in modern sedimentological handbooks. in: Brookfield, M.E. & Ahlbrandt, T.S. 798 (Eds.), Eolian Sediments and Processes. Developments in Sedimentology 38, 455-482. Elsevier, Amsterdam. 799 Schultz, L. G., 1964. Quantitative interpretation of mineralogical composition from X-ray and chemical 800 data for the Pierre Shale. U.S. Geol. Survey Prof. Paper 391, 1-31 801 Sharma, S., Walker, S. C., Jackson, D. A., 2008. Empirical modelling of lake water-temperature 802 relationships: A comparison of approaches. Freshwater Biology 53, 897-911. 803 Slota, Jr, P.J., Jull, A.J.T., Linick, T.W., Toolin, L.J., 1987. Preparation of small samples for 14C accelerator 804 targets by catalytic reduction of CO. Radiocarbon 29, 303-6. 805 Smolíková, L., Havlíček, P., Roetzel, R., 2010. Stratigraphy of Quaternary fossil soils along Highway A5 806 between Wolkersdorf and Schrick (Vienna Basin, Lower Austria). Abhandlungen der Geologischen 807 Bundesanstalt 65, 221-226. 808 Staal, M.J., Thar, R., Kühl, M., van Loosdrecht, M.C.M., Wolf, G., De Brouwer, J.F.C., Rijstenbil, J.W., 809 2007. Different carbon isotope fractionation in developing natural phototrophic biofilms. Biogeosciences 4, 810 613-626. 811 Spötl, C., Vennemann, T.W., 2003. Continuous‐flow isotope ratio mass spectrometric analysis of 812 carbonate minerals. Rapid Communications in Mass Spectrometry 17, 1004-1006. 813 Spötl, C., Reimer, P., Rabeder, G., Bronk Ramsey, C., 2018. Radiocarbon constraints on the age of the 814 world’s highest-elevation cave-bear population, Conturines Cave (Dolomites, Northern Italy). Radiocarbon, 815 60, 299-307. 816 Sprafke, T., 2016. Löss in Niederösterreich: Archiv quartärer Klima- und Landschaftsveränderungen. 817 Würzburg University Press, Würzburg. 818 Sprafke, T. and Obreht, I., 2016. Loess: rock, sediment or soil – what is missing for its definition? 819 Quaternary International 399, 198-207. 820 Stott, L.D., 2002. The influence of diet on the δ13C of shell carbon in the pulmonate snail Helix aspersa. 821 Earth and Planetary Science Letters 195, 249-259. 822 Stuart, A.J., Sulerzhitsky, L.D., Orlova, L.A., Kuzmin, Y.V., Lister, A.M., 2002. The latest woolly mammoths 823 (Mammuthus primigenius Blumenbach) in Europe and Asia: a review of the current evidence. Quaternary 824 Science Reviews 21, 1559-1569. 825 Stuiver, M., Reimer, P.J., Reimer, R.W. (2019) CALIB 7.1 (WWW program). http://calib.org 826 Stuiver, M., Polach, H.A. 1977. Discussion: reporting of 14C data. Radiocarbon 19, 355-63. 827 Terhorst, B., Kühn, P., Damm, B., Hambach, U., Meyer-Heintze, S., Sedov, S., 2014. Paleoenvironmental 828 fluctuations as recorded in the loess-paleosol sequence of the Upper Paleolithic site Krems-Wachtberg. 829 Quaternary International 351, 67-82.

830 Terhorst, B., Sedov, S., Sprafke, T., Peticzka, R., Meyer-Heintze, S., Kühn, P., Solleiro Rebolledo, E., 2015. 831 Austrian MIS 3/2 loess–palaeosol records—Key sites along a west–east transect. Palaeogeography, 832 Palaeoclimatology, Palaeoecology, 418, 43-56. 833 Thiel, C., Terhorst, B., Jaburová, I., Buylaert, J.P., Murray, A.S., Fladerer, F.A., Damm, B., Frechen, M., 834 Ottner, F., 2011a. Sedimentation and erosion processes in Middle to Late Pleistocene sequences exposed in 835 the brickyard of Langenlois/Lower Austria. Geomorphology 135, 295-307. 836 Thiel, C., Buylaert, J.-P., Murray, A.S., Terhorst, B., Tsukamoto, S., Frechen, M., Sprafke, T., 2011b. 837 Investigating the chronostratigraphy of prominent palaeosols in Lower Austria using post-IR IRSL dating. 838 Quaternary Science Journal 60, 424-7116. 839 Van Husen, D., Reitner, J.M. 2011. An outline of the Quaternary stratigraphy of Austria. Quaternary 840 Science Journal 60, 366-387. 841 von Grafenstein, U., Erlenkeuser, H., Müller, J., Trimborn, P., Alefs, J., 1996. A 200-year mid-European air

18 842 temperature record preserved in lake sediments: An extension of the δ OP-air temperature relation into 843 the past. Geochimica et Cosmochimica Acta 60, 4025-4036. 844 Weissl, M., Hintersberger, E., Lomax, J., Lüthgens, Ch., Decker, K., 2017. Active tectonics and 845 geomorphology of the Gaenserndorf Terrace in the Central Vienna Basin (Austria). In: Blumetti, A.M., 846 Grützner, Ch., Guerrieri, L., Livio, F. (Eds.), Quaternary earthquakes: Geology and palaeoseismology for 847 seismic hazard assessment. Quaternary International, 451, 209-222. 848 Wessely, G., 2006. Geologie der Österreichischen Bundesländer. Niederösterreich. Geological Survey, 849 Vienna. 850 Wintle, A.G., 1997. Luminescence dating: Laboratory procedures and protocols. Radiation 851 Measurements 27, 769-817. 852 Wintle, A.G., 2006. Interpretation of single grain De distributions and calculation of De. Radiation 853 Measurements 41, 264-277. 854 Wintle, A.G., Murray, A. S. 2006. A review of quartz optically stimulated luminescence characteristics 855 and their relevance in single-aliquot regeneration protocols. Radiation Measurements 41, 369-391. 856 Yanes, Y., Romanek, C.S., Delgado, A., Brant, H.A., Noakes, J. E., Alonso, M.R., Ibáňez, M., 2009. Oxygen 857 and carbon stable isotopes of modern land snail shells as environmental indicators from a low-latitude 858 oceanic island. Geochimica et Cosmochimica Acta 73, 4077-4099. 859 Yanes, Y., Romanek, C.S., Molina, F., Antonio, J., Delgado, A., 2011. Holocene paleoenvironment (~7200 - 860 4000 cal BP) of the Los Castillejos archaeological site (SE Spain) inferred from the stable isotopes of land 861 snail shells. Quaternary International 244, 67-75. 862 Yapp, C. J., 1979. Oxygen and carbon isotope measurements of land snail shell carbonate. Geochimica et 863 Cosmochimica Acta 43, 629-635.

864 Zanchetta, G., Leone, G., Fallick, A.E., Bonadonna, F.P., 2005. Oxygen isotope composition of living land 865 snail shells: Data from Italy. Palaeogeography, Palaeoclimatology, Palaeoecology 223, 20-33. 866 Zentralanstalt für Meteorologie und Geodynamik - ZAMG (https://www.zamg.ac.at/cms/de/aktuell - 867 6.10.2019) 868 Zöller, L., Oches, E. A., McCoy, W.D. 1994. Towards a revised chronostratigraphy of loess in Austria with 869 respect to key sections in the Czech Republic and in Hungary. Quaternary Geochronology (Quaternary 870 Science Reviews 13, 465-472.

871 Tables

(1) Dose rate (Gy/ka) Accepted Overdispersion De value Age Sample aliquots (%) (Gy) (ka) Gamma Beta Cosmic Internal Total (n) 0.78 ± 1.25 ± 0.09 ± 0.03 ± 2.15 ± 53.8 ± 25.1 O1 0.04 0.06 0.01 0.01 0.15 29 17 ± 3 1.9 ± 2.0 0.83 ± 1.53 ± 0.08 ± 0.03 ± 2.47 ± 45.2 ± 18.3 O2 0.04 0.08 0.08 0.01 0.15 23 24 ± 4 2.3 ± 1.6 0.91 ± 1.48 ± 0.12 ± 0.03 ± 2.54 ± 46.1 ± 18.1 O3 0.05 0.08 0.01 0.01 0.18 25 12 ± 7 2.4 ± 1.6 0.72 ± 1.35 ± 0.16 ± 0.03 ± 2.25 ± 38.7 ± 17.2 O4 0.04 0.07 0.02 0.01 0.16 19 24 ± 7 3.0 ± 1.8 0.77 ± 1.09 ± 0.16 ± 0.03 ± 2.05 ± 19.1 O5 0.04 0.06 0.02 0.01 0.14 24 25 ± 6 41 ± 2.8 ± 2.0 0.73 ± 1.13 ± 0.16 ± 0.03 ± 2.06 ± 19.7 O5(2) 0.02 0.05 0.02 0.01 0.13 24 25 ± 6 41 ± 2.8 ± 1.9 872 Table 1. Dose rates, central De values and optical ages for the Bullendorf luminescence samples. 873 (1) dose rate calculated via in-situ gamma spectroscopy and beta counting. 874 (2) dose rate calculated via a combination of thick source alpha counting and beta counting. 875

15 δ N cal BP (2 Lab no. Taxon Material Unit δ13C(‰) %N C/N 14C BP (‰) sigma range)

UBA 35576 Mammalia indet. bone fragment 1 -20.5 3.6 1.59 3.19 32840 ±534 35750-38410

UBA 35574 M. primigenius vertebra 2b 0.06

UBA 35575 M. primigenius vertebra 2b 0.09

UBA 35577 R. tarandus humerus 1 0.23

UBA 34810 Equus sp. radius 1 -20.4 3.5 2.05 3.52 34707 ±929 36820-41200

UBA 34811 Mammalia indet. rib 1 0.68

UBA 34812 M. primigenius tusk 2b 0.06

876 Table 2. Radiocarbon dating and stable isotope results of the large mammal remains. 877

δ13C δ18O Sample Species V- Mean SD V- Mean SD Note PDB ‰ PDB ‰ Pupilla B2 muscorum -8.30 -5.14 juvenile -8.07 -4.69 adult -7.84 -5.16 adult -6.58 -3.42 random aliquots -6.98 -4.04 random aliquots -6.20 -7.32 ± 0.86 -3.43 -4.31 ± 0.79 random aliquots Pupilla -8.18 -3.76 C1 muscorum juvenile -7.87 -3.90 adult -9.23 -8.42 ± 0.71 -5.22 -4.29 ± 0.80 adult Pupilla -7.21 -3.79 loessica juvenile -6.29 -4.04 random aliquots -5.82 -4.28 random aliquots -8.98 -7.07 ± 1.39 -3.59 -3.92 ± 0.29 random aliquots Pupilla sterrii -7.16 -3.10 juvenile -7.80 -7.48 ± 0.45 -2.53 -2.81 ± 0.40 random aliquots Succinella -5.93 -4.62 B3 oblonga juvenile -6.93 -3.14 random aliquots -6.41 -6.42 ± 0.50 -2.72 -3.49 ± 0.99 random aliquots Helicopsis -8.23 -5.83 striata - - - - random aliquots Columella -6.92 -6.41 C3 columella juvenile -6.75 -6.34 adult -7.83 -7.16 ± 0.58 -6.02 -6.25 ± 0.20 entire shell Succinella -7.04 -3.95 oblonga random aliquots -6.21 -4.77 random aliquots -6.63 -6.62 ± 0.41 -5.73 -4.81 ± 0.89 random aliquots Trochulus -6.39 -5.21 hispidus juvenile -5.16 -5.77 ± 0.86 -4.22 -4.71 ± 0.70 random aliquots Vallonia -8.25 -5.61 tenuilabris juvenile -7.95 -8.1 ± 0.21 -4.62 -5.11 ± 0.70 random aliquots Pupilla -7.11 -5.20 muscorum juvenile -6.13 -4.55 random aliquots -6.57 -4.84 random aliquots -6.09 -5.28 random aliquots -6.96 -5.28 random aliquots -6.69 -6.59 ± 0.41 -5.10 -5.04 ± 0.29 random aliquots Pupilla -7.82 -5.25 alpicola juvenile -5.86 -6.84 ± 1.38 -5.75 -5.50 ± 0.35 random aliquots Galba truncatula -11.09 -9.50 random aliquots 28

-10.49 -9.81 random aliquots -11.25 -10.5 random aliquots -10.71 -9.25 random aliquots -12.12 -9.61 random aliquots -9.80 -8.04 random aliquots -10.86 -10.59 random aliquots -11.31 -9.62 random aliquots -11.2 -10.98 ± 0.64 -10.17 -9.68 ± 0.76 random aliquots Succinella -5.57 -4.57 B6 oblonga juvenile -6.20 -4.71 adult -5.29 -5.68 ± 0.46 -5.20 -4.82 ± 0.33 entire shell Pupilla -6.49 -4.86 muscorum juvenile -7.30 -4.91 juvenile -7.81 -5.58 adult -6.94 -5.82 random aliquots -6.72 -4.77 random aliquots -6.96 -7.03 ± 0.46 -4.58 -5.08 ± 0.49 random aliquots Succinella -3.36 -3.72 C7 oblonga juvenile -6.47 -2.63 adult -5.06 -3.00 adult -5.59 -5.12 ± 1.30 -3.63 -3.24 ± 0.52 random aliquots 878 Table 3. Stable isotope composition of continental gastropod shells from the Bullendorf section. 879

18 13 δ O V-PDB ‰ δ C V-PDB ‰ Sample Species juvenile adult juvenile adult B2 Pupilla muscorum -5.14 -4.69 -8.30 -8.07 C1 Pupilla muscorum -3.76 -3.9 -8.18 -7.87 B6 Pupilla muscorum -4.91 -5.58 -7.30 -7.81 Mean -4.60 -4.72 -7.93 -7.92 Stand. dev. ±0.08 ‰ ±0.01 ‰ B6 Succinella oblonga -4.57 -4.71 -5.57 -6.20 C7 Succinella oblonga -3.72 -2.63 -3.36 -6.47 Mean -4.14 -4.72 -4.46 -6.33 Stand. dev. ±0.34 ‰ ±1.32 ‰ C3 Columella columella -6.40 -6.34 -6.92 -6.75 Stand. dev. ±0.05 ‰ ±0.12 ‰ 880 Table 4. Variability between juveniles and adults of Pupilla muscorum, Succinella oblonga and Columella 881 columella.

Gravel Sand Silt Clay method of

moments

Ternary

system

mple

Silt

Unit

Clay Sand- Sand

Gravel

< 2< µm

fine silt

fine sand SD

coarse silt coarse

Silt-Clay fine gravel

coarse sand coarse

mean

medium silt

medium sand

medium gravel

percent by weight Phi units clayey S7 6 0.2 29.1 51 19.6 0.2 1.2 5.2 22.8 23.7 15.6 11.8 19.6 5.96 2.99 Sand-Silt clayey O5 6 1.4 33.1 46.5 19 1.4 1.4 7.1 24.5 22.6 13.2 10.7 19 5.18 3.52 Sand-Silt clayey O4 5 0.1 51.2 33.2 15.5 0.1 0.9 14.6 35.7 17.7 7.4 8.1 15.5 5.18 3.52 Silt-Sand clayey S6 5 0 54 33.5 12.4 0.3 14.8 39 17 8.3 8.2 12.4 4.89 3.13 Silt-Sand clayey S5 4 0.3 31.1 45.4 23.1 0.1 0.3 0.5 10.2 20.5 19.6 14.7 11.1 23.1 6.21 3.48 Sand-Silt sandy O3 4 0.1 24.9 48.9 26.1 0.1 0.2 6.7 18 21.4 15.3 12.1 26.1 6.79 3.67 Clay-Silt clayey S4 3b 0.3 43.1 43.1 13.6 0.3 0.7 15.1 27.3 21.3 12.6 9.2 13.6 5.25 3.23 Silt-Sand clayey O1 2b 0.1 38 45.4 16.6 0.1 0.2 10 27.8 22.2 13.2 10 16.6 5.6 3.11 Sand-Silt clayey S3 2b 0 48.4 40.5 11.1 0.2 12.5 35.7 20.7 11.8 8 11.1 5.02 2.97 Silt-Sand clayey O2 2a 0 31.3 51.8 16.9 0.4 9.9 21 23.1 17.3 11.4 16.9 5.91 3.23 Sand-Silt clayey S2 2a 0 41.5 46.4 12.1 0.2 15 26.3 23.9 14 8.6 12.1 5.26 3.2 Sand-Silt

S1 1 Clay-Silt 0.2 3.9 69.3 26.7 0.2 0.2 0.3 3.4 27.2 26.1 16 26.7 7.66 3.22

S0 1 Clay-Silt 0 1.5 65.6 32.9 0.01 0.1 1.4 21.4 25.1 19.1 32.9 8.41 3.42 882 Table 5. Results of grain-size analyses of the Bullendorf section.

Bulk mineralogy Clay mineralogy < 2µm

Unit

Sample

Illite

Albite

Quartz

Calcite

Chlorite

Smectite

Kaolinite

Dolomite

Vermiculite

Alkalifeldsp.

Phyllosilicates S7 6 8 5 41 4 8 34 0 66 20 6 8 O5 6 7 5 40 7 9 32 1 67 17 7 8 O4 5 6 5 47 7 11 25 2 58 20 7 12 S6 5 4 7 54 6 7 21 2 53 26 8 11 S5 4 0 5 46 6 9 33 3 55 26 5 11 O3 4 0 6 41 7 8 38 6 57 21 6 10 S4 3b 4 6 52 5 9 24 1 55 29 5 10 O1 2b 2 5 41 5 10 37 6 60 20 6 8 S3 2b 4 6 52 5 9 24 1 55 28 7 8 O2 2a 5 6 42 7 11 29 3 58 24 7 8 S2 2a 4 5 50 6 11 24 2 57 24 6 10 S1 1 3 8 28 5 7 47 1 75 18 6 0 S0 1 4 7 28 5 5 50 1 73 18 7 1 883 Table 6. Results of the XRD analyses of bulk mineralogy and clay mineralogy of the Bullendorf section.

884 Illustrations 885 886 887 888 889 890 891 892 893 894 Figure 1. Location of Bullendorf (asterisk) and key LPS sites in the state of Lower Austria. 895 896 897 898 899 900 901 902 903 904 905 906 907 908 909 910 911 912 913 914 915 916 917 918 919 920 921

922 Figure 2. Geological map of the Bullendorf area (from Grill, 1961). 923 924 925 926 927 928 929 930 931 932 933 934 935 936 937 938 939

940 Figure 3. (A) Sedimentological log of the Bullendorf succession reporting lithofacies, sample distribution 941 (S: bulk sediment collected by R. Roetzel and I. Wimmer Frey, B: mollusc samples collected by O. 942 Mandic, Bhp: molluscs hand-picked by O. Mandic, C: mollusc samples collected by O. Mandic; O: OSL 943 samples collected by M. Meyer), and ages (green: OSL-dating; red: radiocarbon dating). Lithofacies after 944 Keller (1996). Fl: laminated fine sediments; Hb: bedded alternation of fine sediments and sand; Hl: 945 wavy- laminated alternation of fine sediments and sand; H-r: Alternation of fine sediments and sand 946 with ripple cross-lamination; H-p: alternation of finesediments and sand with pedogenic influence; Sb: 947 bedded sand; Sm: massive sand; Sm-Sb: massive sand and bedded sand; Sm-Fm: massive sand and fine 948 sediment. (B) bulk mineralogy, (C) clay mineralogy <2µm, (D) grain-size distribution. 949

950 Figure 4. Overview of the outcrop and the sampled units. Laterally continuous layering occurs in sub- 951 units 1, 3b, 3c, and 3d. View towards northwest. 952

953 Figure 5. (A) Alternation of clayey silt and fine sand (lithofacies Hb) in unit 2a with sample point OSL O2. (B) 954 Silty fine to medium grained sand alternating with clayey silt (lithofacies Hb) and root casts (lithofacies H-p) 955 in subunit 2b. (C) Layered silty and sandy sediments (lithofacies Hl) with lenses from fine sand and 956 convolute bedding in subunit 2b.(D) Silty fine to medium grained sand alternating with clayey silt 957 (lithofacies Hb, Hl) in subunit 2b. Note convolute-like bedding. (E) Left mammoth tusk at the base of 958 subunit 2b. Scraper approx. 20 cm (A, C, D), decimetre scale (B), centimetre scale (E). 959 960 961 962 963 964 965 966 967 968 969 970 971 972 973 974 975 976 977 978 979 980 981 982 983 984 985 986 987 988 989 990 991 992 993 994 995 996 997 998 999 1000 1001 1002

1003 Figure 6. Natural and regenerated OSL decay curves (A) and sensitivity corrected dose-response curve 1004 fitted with a single exponential function (B) for a characteristics quartz aliquot from Bullendorf. Note the 1005 fast decay of the luminescence signal during the initial 5 seconds of optical stimulation in A. 1006

1007

1008

1009

1010

1011

1012

1013

1014

1015 Figure 7. Stable isotope composition of land snails shells in samples C1 - B3 (A) and C3 (B). Stable isotope 1016 composition of the freshwater species Galba truncatula in sample C3 (C). Stable isotope composition of 1017 Pupilla muscorum in samples B2, C1, C3, B6 (D) and of Succinella oblonga in samples B3, C3, B6, C7 (E).

1018 Supplement material

1019 Figure 1.A. Single-aliquot De distributions visualized as radial plots for samples O1 (A), O2 (B), O3 (C), O4 (D) 1020 and O5 (E). The central De value has been calculated via the central age model and the grey bar indicates 1021 the 2 sigma uncertainty range. For two samples (O2 and 5) high De outliers (shown as triangles) were 1022 rejected before De calculation.