Collapse wedges in periglacial eolian sands imply Late Pleistocene paleoseismic activity of the Vienna Basin Transfer Fault (western Slovakia)

1 – Department of Geology and Paleontology, Faculty of Natural Sciences, Comenius University in Bratislava, Ilkovičova 6, 842 15 Bratislava, Slovakia; michal.sujan@uniba.sk; samuelrybar3@gmail.com 2 – Mining and Geological Survey of Hungary, Stefánia út 14, 1143 Budapest, Hungary; bozso.edit@mbfsz.gov.hu 3 – MM Revital, Šustekova 10, 851 04, Bratislava, Slovakia; tkluciar@gmail.com 4 – Department of Mathematics and Descriptive Geometry, Faculty of Civil Engineering, Slovak University of Technology in Bratislava, Radlinského 11, 810 05 Bratislava, Slovakia; matus.tibensky@gmail.com 5 – Department of Quantitative Methods, Faculty of Management, Comenius University in Bratislava, Odbojárov 10, 820 05 Bratislava, Slovakia 6 – Department of Geology and Meteorology, University of Pécs, Ifjúság ú. 6, Pécs 7624, Hungary; krisztina.sebe@gmail.com * – corresponding author


Introduction
Paleoseismological research focusing on the intensity and recurrence intervals of earthquakes through geological time is of primary importance in the prediction of geohazards related to seismic events. The direct determination of fault activity is routinely carried out for example through the exposure dating of fault scarps (Benedetti et al., 2003;Palumbo et al., 2004;Carcaillet et al., 2008;Tikhomirov et al., 2014), by the dating of successive events using disrupted sediment layers in trenches (Enzel et al., 2000;Decker et al., 2005;Štěpančíková et al., 2008;Štěpančíková et al., 2010;Le Dortz et al., 2011;Hintersberger, 2012;Ansberque et al., 2016;Weissl et al., 2017;Hintersberger et al., 2018), or by dating the deformation of sinters in caves (Plan et al., 2010;Gribovszki et al., 2017;Szczygieł et al., 2021). Significant attention has also been paid to the seismically triggered deformation of sediments related to liquefaction (Guarnieri et al., 2009;Moernaut et al., 2009;Alessio et al., 2012;Holzer et al., 2015;Amorosi et al., 2016). However, intense scientific discussion tends to be focused on ambiguities in the determination of causes in the case of liquefaction -including rapid loading, slope deformations or cryoturbations -all of which may be active concurrently with, or appear in the absence of seismic triggers (van Vliet-Lanoë et al., 2004;Montenat et al., 2007;Moretti and Sabato, 2007;Owen et al., 2011). Liquefaction is usually associated with water saturated sandy deposits, and it is most pronounced in the case of heterolithic strata with the presence of impermeable layers amplifying the effect of the water pressure exceeding the shear strength of sediment (Maltman, 1994).
Much less attention has been paid to the research of seismically triggered deformations in successions without cohesive strata, such as eolian sands. The majority of studies focus on the eolian sandy sediments deformed by liquefaction in the presence of water close to a paleosurface (e.g., Cojan and Thiry, 1992;Moretti, 2000;Hurst and Glennie, 2008;Chan and Bruhn, 2014;Bryant et al., 2016;Ford et al., 2016;Chan et al., 2019;Espinoza et al., 2020;Rodríguez-López and Wu, 2020). Only a few studies describe the slope deformation of eolian facies due to earthquakes (Steidtmann, 1982;Cojan and Thiry, 1992;Ford et al., 2016). The present study investigates collapse structures formed due to the opening of fractures in eolian sand accumulated in the periglacial zone of Central Europe during the late Pleistocene (van Vliet-Lanoë et al., 2004;Ruszkiczay-Rüdiger and Kern, 2016). Collapse structures are accompanied by the folding and disturbance of primary structures in specific horizons. The aim of the present research is to provide a detailed documentation of the structures and interpretation of their formative process, since the description of similar forms in any depositional environment is rare in the literature (Alessio et al., 2012;Caputo et al., 2012;Caputo et al., 2016;Lunina and Gladkov, 2016); what is more, these are also absent from the recent reviews of soft-sediment deformation structures (Owen et al., 2011;Shanmugam, 2017).

Geological setting
The study area is located near the southeastern margin of the Vienna Basin in Central Europe, near the village of Plavecký Štvrtok (Fig. 1). The basin was formed in a pull-apart regime during the middle Miocene due to the lateral escape of the Alpine-Carpathian-Pannonian lithospheric microplate (ALCAPA) northeastwards from the collision zone of Adriatic microplate with the European platform ( Fig. 1A) (Fodor, 1995;Kováč et al., 2004;Strauss et al., 2006;Hölzel et al., 2010;Kováč et al., 2017;Lee and Wagreich, 2017). This movement caused the formation of the Vienna Basin Transfer Fault (VBTF) on the southwestern margin of the basin with oblique extensional or transtensional kinematics and a negative flower structure geometry ( Fig.  1B) Hinsch et al., 2005;. The subsequent late Miocene subsidence of the Vienna Basin took place in an extensional post-rift setting (Hölzel et al., 2010;Kováč et al., 2017;Lee and Wagreich, 2017). The late Miocene extensional phase of the basin evolution was interrupted by basin inversion in a compressional stress regime, which affected the whole Pannonian Basin realm after 8.0-4.0 Ma (Fodor et al., 2005;Horváth et al., 2006;Ruszkiczay-Rüdiger et al., 2007;Balázs et al., 2018). This basin inversion caused large-scale lithospheric folding after ~6.0 Ma in the Danube Basin in the southeast (Fig. 1B) (Šujan et al., 2021), and non-deposition or partial denudation of the Miocene infill in the Vienna Basin (Decker, 1996;Decker et al., 2005;Strauss et al., 2006;Salcher et al., 2017). As a consequence, Quaternary deposits overlie the older basin fill unconformably. Fig. 2. Location of the studied sandpit, Bažantnica. A: Geological map simplified according to Fordinál et al. (2012b). B: Quaternary thickness map according to Maglay (2009). C: Distribution of sandpits with and without the presence of softsediment deformation structures in the NE Vienna Basin. See Fig. 1 for location. Decker et al. (2005) suggested a change from compression during the basin inversion to NE-SW oriented extension during the Early Pleistocene, and ultimately, to a transtensional regime resembling middle Miocene kinematics from the Middle Pleistocene up to recent times. The VBTF experienced transtensional reactivation in this latter phase, accompanied by the formation of small-scale, narrow pull-apart basins. In the deepest of them, the Mitterndorf Basin, the middle-upper Quaternary basin fill reaches a thickness of 140 m ( Fig. 1) Salcher et al., 2012). The marginal faults of the basins are the source of significant seismic activity (Fig. 1B) (Hinsch and Decker, 2003;Hinsch and Decker, 2011). The study area is situated along the margin of the Zohor-Plavecký Basin, which has a Quaternary infill of ~60 m, and represents one of the Quaternary basins formed within the VBTF along the northwestern foothills of the Malé Karpaty Mts. (Fig. 2B) (Maglay et al., 1999;Maglay, 2009). In contrast to the Mitterndorf Basin, which has been studied in a number of papers Hinsch et al., 2005;Salcher et al., 2017), knowledge of the infill of the Zohor-Plavecký Basin is limited. The basin is situated above a negative flower structure of the Vienna Basin transfer fault, as is evident from the reflection seismics (Fig. 3). However, the resolution of the existing seismic lines does not allow the resolution of the contact between the Miocene and the Pliocene to Quaternary successions, which are separated by an unconformity related to the basin inversion. Borehole data show that the infill of the Zohor-Plavecký Basin consists mostly of gravels and sands (Fig. 4). The succession was probably deposited by alluvial fans sourced from the Malé Karpaty Mts., similarly to the sediments outcropping in the area ( Fig. 2A) (Fordinál et al., 2012b). The underlying upper Miocene Gbely Fm. is also of alluvial origin (Kováč et al., 1998;Harzhauser et al., 2004) and despite its floodplain mud-dominated lithology, gravels and sands of channel fills are common. Therefore, the position of the boundary between the upper Miocene and the Pliocene to Quaternary successions may only be tentatively determined, due to the absence of geochronological or biostratigraphic data for these successions (Fig. 4). The alluvial gravels and sands are covered by an eolian succession up to 30 m thick (Fordinál et al. 2012) (Fig. 2B, Fig. 4). The fine to coarse sands and gravelly sands form a topography consisting of a fossil field of lunate dunes covered by recent vegetation (Fig. 5A). These deposits have been and continue to be excavated in the Bažantnica sandpit investigated in this study. Existing luminescence dating ages gained from the sandpit range from 40 ± 5 ka to 13 ± 1 ka (Moravcová and Fordinál, 2010;Fordinál et al., 2012a;Kadlec et al., 2015), indicating that the depositional record spans the uppermost MIS 3 and the whole of MIS 2, including the Last Glacial Maximum (Hughes and Gibbard, 2015). A recent re-evaluation of paleoclimate proxies implies that during glacials the area was characterized by discontinuous permafrost or seasonal frost and relatively dry conditions with the mean annual precipitation reaching 300-490 mm (Ruszkiczay-Rüdiger and Kern, 2016). Soft-sediment deformation structures were first briefly reported from the Bažantnica sandpit by Fordinál et al. (2012a) and by Kadlec et al. (2015) and are investigated in detail in the present study.

Field research
The field campaign of documenting the deformations in the sand pit was performed in December 2016, when air temperatures dropped below 0°C for the whole day. This condition turned out to be very favorable, since frozen pore water in the sediment gave rise to a high degree of cohesion and allowed effective cleaning of the outcrop wall with shovel and spatula without occasioning failure in what is normally a deposit lacking all cohesion, and therefore practically impossible to observe in a clean section. Moreover, variations in the grain size on the outcrop were highlighted by the differential freezing of the beds, leading to increased visibility of the structures after the brushing of the wall.
Active excavation leads to a constantly changing view of the outcrop. After the inspection of all existing outcrops, a wall (referred to in the present study as "the main wall", Fig. 5B) was selected on the basis of the abundance of deformation structures and on its orientation perpendicular to the brittle deformations for detailed cleaning and analysis. The main sedimentary features and deformations of the sediments were documented. The orientation of all fractures related to the wedge-like structures was measured using a geological compass on the main wall, as well as on other walls in the sandpit. The primary dip of the strata was also measured at the position of every wedge-like structure. The main wall was also a subject of sampling for luminescence dating.

Luminescence dating
The sampling strategy for optically stimulated luminescence (OSL) dating was to determine the age of the deformed horizons. Sample preparation for OSL dating was carried out under subdued red-light conditions. The sunlight-contaminated outermost few centimeters of the sediments were removed from the two ends of the sampling tubes and were not used for dating. Wet sieving was applied to separate the 100-200 µm grain size fractions of the samples. Organic material was removed with the use of 20% H2O2, while carbonates were dissolved using 10% HCl. Quartz-rich fractions were separated with the aid of sodium polytungstate (SPT: Na6[H2W12O40]H2O) solution (2.67 g/cm 3 ). The outer 10 μm layer of the quartz grains, which have absorbed a dose of alpha radiation, plus the feldspars, were removed with the use of 40% HF for 60 minutes. As the IR test indicated that these separates were feldspar contaminated, a further 30 minutes' etching with 40% HF was employed. The quartz fractions were then cleaned using 10% HCl. After dry sieving, the 100-160 µm grains were mounted in monolayer on stainless steel discs using silicone oil spray. The size of the aliquots was a 5 mm diameter (medium aliquot). OSL measurements were made by a Risø TL/OSL DA-20 C/D reader with a calibrated 90 Sr/ 90 Y beta source. Blue light stimulated luminescence of quartz was detected through a Hoya-340 filter. The Single-Aliquot Regenerative-dose (SAR) OSL protocol (Wintle and Murray, 2006) was applied, with illumination for 40 s at 280ºC in the last step of each cycle. OSL signals were evaluated using the early-background subtraction method (Cunningham and Wallinga, 2010) when the 0.8-1.6 s integral of the OSL decay curve was subtracted from the initial 0.8 s integral signal to avoid a contribution from medium and slow OSL components. The dose-response curves were fitted using a single saturation exponential function. The purity of the quartz separates was checked by IR test (measurement of OSL depletion due to infrared stimulation) on every aliquot in an additional cycle of the SAR-OSL protocol. The preheat plateau, thermal transfer and dose recovery tests were also carried out on each quartz fraction.
On the basis of the results of the preheat plateau test, a preheat temperature of 240 °C and a cut heat of 200 °C were chosen. The thermal transfer test indicated that the thermal transfer was negligible using a 240 °C preheat temperature. The dose recovery ratio, which was measured on 4 aliquots of each sample, varied between 0.87 and 1.08 (0.98 ± 0.05 on average, n=28), and indicated that the SAR protocol correctly measured the given dose. Only one aliquot, that of sample PS4, had a dose recovery ratio falling outside the acceptable 1.0 ± 0.1 range (0.87). A few aliquots were rejected due to an inadequate result of the IR test, bad recycling ratio, or high equivalent dose error. The recuperation was low (0.16 ± 0.34 on average, n=259).
Sample preparation, gamma spectrometry and water content measurements, the creation and evaluation of the luminescence measurements, and the age calculation were carried out at the Mining and Geological Survey of Hungary (MBFSZ, former Geological and Geophysical Institute of Hungary). The OSL measurements were run at the Eötvös Loránd University, Budapest.
The OSL ages obtained served for age-depth modeling using the Bacon package developed by Blaauw and Christensen (2011) for R statistical software (R Core Team, 2020). The purpose of the model was to obtain a statistically robust age-depth model, what is ensured by the use of Markov Chain Monte Carlo (MCMC) iterations to calculate the accommodation rate for each section of the sedimentary interval under investigation. The model included assumption of a continuous deposition, as the depositional record and ages obtained did not imply existence of a hiatus. The prior assumptions concerning coherence in accommodation rate were set by the Bacon parameters of 'memory', namely, mem.mean and mem.strength. After testing a range of values with very low variability of the output, the model finally selected includes the default setting of the parameters (mem.mean=0.7, mem.strength=4).

Description
A major part of the outcropping walls in the Bažantnica sandpit is formed by subhorizontally arranged, parallel layers of alternating fine, medium and coarse sand, with less frequent layers of granules (Fig. 6). Subhorizontal layers of silty fine sand occasionally appear (<10% of the sediment thickness). Individual layers range between a few millimeters to 2 cm thick. Bedding is interrupted by subhorizontal erosional surfaces, which are covered by conformable or onlapping strata (Figs. 6, 7). In the case of conformable contact of underlying and overlying strata, truncation is visible only as the cross-cutting of deformations. The sand is moderately sorted in individual layers, and grains are well rounded and spherical, with a matte surface. The dip of the strata attains 10-14° in the lower, and 5-8° in the upper part of the main wall, with a general dip direction towards S and SSE (Fig. 8). The measurements across other walls of the sandpit yielded dips of 2-10° and orientations represented almost all directions relatively equally (Fig. 8).
Subhorizontal layers alternate with less frequent low angle planar cross-stratified sand layers 20-40 cm thick (Fig. 6, 9B, G, H). Trough cross-stratified sand lenses are present occasionally, one example in Fig. 6 reaches a thickness of 55 cm and a width of ca. 4.5 m. Large-scale crossstratification with heights of over 4 m also occurs in the sand pit (Fig. 10). Cross-stratified units usually have an erosional base, which is subhorizontal and straight or concave.

Interpretation
The rounding and matte surface of the sand grains indicates eolian transport. The alternation of moderately well-sorted subhorizontal or smoothly dipping layers of sand with various grain sizes is a product of translatent wind ripple stratification (Hunter, 1977), and indicates deposition from subcritically climbing wind ripples on a sand sheet (Loope, 1984;Jordan and Mountney, 2010). Moderate sorting is also a result of a relatively short transport distance, in this case, from the Morava River (Moravcová & Fordinál, 2010, Fordinál et al., 2012. The low angle planar cross-strata could be associated with rippleform lamination, while trough crossstrata represent eolian dunes (Hunter, 1977;Pye and Tsoar, 2009). The observed settings are comparable to those described from a subrecent periglacial eolian sand sheet in Iceland (Mountney and Russell, 2004). The periglacial climate causes the widespread presence of vadose water in sand sheets, increasing cohesivity of transported particles, and hence decreasing the height of wind ripples and dunes (Van Dijk et al., 1996;Jackson and Nordstrom, 1998;Mountney and Russell, 2009). This is in accordance with the observed thickness of cross-strata reaching only 10-55 cm. Cross-stratification to a height of several meters represents the dune landforms also visible in the modern topography (Fig. 5A, Fig. 10). Wide, parabolic dunes are characteristic of areas with low precipitation and restricted, but nonetheless existing, vegetation, typical of cold climates, and the European Pleistocene periglacial belt, as well (Seppälä, 2004;Pye and Tsoar, 2009;Yan & Baas, 2015; and references therein). Their orientation -the arms pointing upwind, the curvature downwind -fits the prevailing ENE Pleistocene wind direction of the area (Sebe et al. 2015). Layers of granules exhibit best sorting and represent a residual lag formed by deflation events with increased wind speed (Fryberger et al., 1992;Mountney and Russell, 2004). This kind of granule lag is especially characteristic of cold-climate eolian environments (Ruz and Allard, 1995). The silty fine sand strata most probably represent events of decreased wind activity (Vandenberghe, 2013).

Deformation structures
Soft-sediment deformation structures are frequent on all walls of the Bažantnica sandpit. All other sandpits in eolian sands of the Slovakian part of the Vienna Basin were checked for the presence of similar deformation structures, but with a negative result (Fig. 2C). However, these sandpits are not located directly in the course of the VBTF, as the Bažantnica sandpit is.

Wedge-shaped structures
Wedge-shaped structures are the most common deformations in the studied outcrops, with 21 wedges documented and measured on the main wall, and 17 wedges on the remaining walls in the sandpit (Figs. 5,6,7,8,9). The wedges are sharply delimited from the surrounding sand sheet strata and narrow downwards. They reach width of a few centimeters to 30 cm with an average width of ~10 cm in the topmost part. The height of individual wedges ranges from 10 cm (Fig. 9A) to >50 cm (Fig. 9I). The top of the wedges is usually overlain by undeformed strata, and as a general rule, their fill is truncated. The internal structure of a wedge can vary. Most wedges are thin (Fig. 7, Fig. 9A, F). In their upper parts the host sand layers bend downward on both sides of the structure. Between them, a structureless sand infill of a few centimeters' thickness can occur, and this may then continue downward, below the bent layers. The lowermost part of the wedges is usually just a fracture, without infill. In some wedges, blocks of the neighboring stratified sand have moved down a few cm-s (Fig. 9B). The largest wedges can partly be filled by the material of the surrounding, chaotically disturbed sand beds (Fig. 9G, I). The overall character of the wedges implies a dominant downward movement of the infill during deformation. Most of the wedges exhibit a curved funnel shape in vertical section (Fig. 9A, B, G, I, J). Each wedge is associated with a fracture running along its center, at least in its lower portion. The inclination of fractures ranges between 46-86° (average 72.9°) on the main wall and 45-85° (average 68.8°) on the remaining walls in the sand pit. Fractures are systematically oriented N-S and NE-SW (Fig. 8). As the fractures are inclined, most of the associated wedges are asymmetrical, and the steeper wall is bent down to a greater degree than the other one (Fig. 7,  Fig. 9G, J). The fractures are planar structures without significant variation in the orientation of a single structure (Fig. 11). A single fracture with an offset of a few centimeters was observed (Fig. 9D), but represents an exception. The offset did not vary within the studied exposure. Most of the fractures continue up to the base of the outcrop, while a minority of them appear to terminate within the succession (e.g., Fig. 7 -middle level).
The wedges are grouped in horizons, clearly visible in Figs. 6, 7 and 9A. In these horizons, several wedges occur next to each other laterally, and they are truncated by the same erosional surface. In many cases, the wedges formed repeatedly along the same fracture (Fig. 9), while some of them terminated within the studied succession (Figs. 6, 7, 9I, J). The upper termination always lies along a major erosional surface. A fracture may occasionally split into two branches, each with developed wedges (Fig. 9I).

Other deformations
Beds a few centimeters to 15 cm thick with chaotic structure appear frequently, locally with poorly visible remains of the original structure, and these appear as fragments of bedded sand a few centimeters long (Fig. 9C, D, H). There is an obvious relationship between the distribution of the layers with chaotic structure and the wedge-shaped deformations: (1) most of the chaotic beds appear in the same horizons as the wedges and are disrupted by the wedges, and (2) the wedges as well as the chaotic beds are truncated by the same erosional surface, when observed together (Figs. 7, 9 A, B, G, I, J). The chaotic beds lie conformably on the underlying strata, without truncating them.
Beside the chaotic layers, and with significantly less frequency, ~10-20 cm thick folded sand layers appear as well (Fig. 12). The folds are usually open, sometimes tight. While many folds are symmetrical and upright, asymmetric -inclined or recumbent -ones are also common. A preferential orientation of asymmetric folds, i.e., the tilt of the axial plane towards a certain direction, could not be observed. The folded layer packages are truncated by the overlying sand beds (Fig. 12). The underlying strata are not affected by the deformation. The folded strata appear in the same horizons as deformation wedges and are truncated by the same erosional surfaces, similarly to the chaotic beds. It should be noted that two types of erosional surfaces were observed in relation to the distribution of deformation structures: (1) truncation at the top of the disturbed surface, which might cut off only the uppermost few centimeters of the sediment, leading to conformable contact with overlying strata (Fig. 7B, C, Fig. 12), and (2) a major erosion surface, cutting off a significant part or the disturbed horizon, while only a small part of the wedges remained intact; the resulting stacking pattern of beds then comprises the truncation and angular contacts of post-deformational strata with the underlying sediments (Fig. 7A, I).
A cross-stratified sandy unit with steep sigmoidal foresets and a complicated concave, sharply defined base also shows deformation structures (Fig. 7 -between erosional surfaces 3a and 3b, and Fig. 9A). The foresets gradually attain a convex shape towards the left, a form that is impossible for them to have attained during sedimentation.  (Fig. 7, Fig. 13). The whole deformed horizon is ca. 85 cm thick. It lies above horizontal sand beds crosscut by two wedges, but its internal deformations are not connected to the underlying fractures. The deformed body has numerous subvertical curved fractures, some of them with displacement (small faults), separating the sediment into blocks. A transition from normal slip along the faults to thrusting could be seen from the right towards the left (Fig. 13). Subvertical curved fractures continue to occur towards the left side with decreasing offset. The body does not exhibit significant deformation more to the right of the described complex deformation (Fig. 7). The basal detachment surface is marked by numerous centimeter scale thrusts oriented towards the left part of the outcrop (Fig. 13).

Luminescence dating
The natural OSL signals indicated relatively bright luminescence and fast OSL decay (Fig. 14A). The natural equivalent doses (De) of the samples lay in the fast-growing region of the dose response curves (Fig. 14B). These are beneficial for OSL dating.  The equivalent doses are between 11 and 53 Gy. The distribution of the equivalent doses of each sample is more or less symmetric (except sample PS3), indicating that they belong to one population on the basis of the histograms and kernel density estimate curves of the abanico plots ( Fig. 15) (Dietze et al., 2016). OSL dating is based on equivalent doses of 29-55 aliquots per sample (Table 1). More than fifty aliquots were measured in the case of samples PS2 and PS6 to obtain a more symmetric equivalent dose distribution. The ages were calculated with the mean or median De-s, depending on which fit better to the peak area of the distribution. The overdispersion (OD) of the equivalent doses varies between 13 and 40% ( Table 1). The sediments in the upper part of the section (sample PS4-7) with lower OD values were probably well bleached at the time of deposition, as laid down in Olley et al. (2004), or better bleached than the samples in the lower part of the section. The environmental dose rates were calculated on the basis of laboratory high-resolution gamma spectrometry measurements using a Canberra GC3020 on 1.0-1.1 kg bulk samples collected in the proximity of each OSL sample. Table 2 shows the resulting U, Th and K concentrations. Dose rate conversion factors given by Guérin et al. (2011) were employed. The cosmic dose rates were determined according to the method set out in Prescott and Stephan (1982) and Prescott and Hutton (1994). The estimated average water content throughout the burial period of the dated eolian sediments was 11 ± 2% according to the results of water content measurements on sediments taken from each OSL sampling tube, and after they were saturated by water in a laboratory. The OSL ages range between 16.7 ± 0.9 ka and 25.9 ± 1.7 ka (Table 1). Within a reasonable margin of error, the ages are in stratigraphic order. This implies a relatively continuous deposition of sand in the area during the Last Glacial Maximum and during MIS2 (Clark et al., 2009).
The standard error σ1 of the OSL ages in the range of 0.7-1.7 ka is relatively high compared to the age range of the studied stratigraphic succession, reaching 9.2 kyr between the mean ages. Hence, the Bayesian age-depth model was considered the most robust approach to the determination of the timing of depositional and deformation events. The resulting model in Fig.  16 displays linear trend, caused by the high degree of uncertainty concerning the ages in comparison to short period dated. The model fits well with samples PS2 to PS7, while the lowermost sample, PS1, lies at the very margin of acceptability, and might be considered as an outlier. Nevertheless, this lowermost sample might also represent a separate succession, comprising predominantly silty sands. The modeled age ranges within σ1 error bars, median and mean ages in 1 cm steps are present in the Supplementary Table 1 for the depth range between dating samples. Fig. 16. Bacon age-depth model (Blaauw and Christen, 2011) applied to the OSL ages of the main wall outcrop in the Bažantnica sandpit. Grey dots indicate 95% probability intervals. The inset graphs show the number of MCMC iterations (upper), distributions of prior (green) and posterior (grey) accommodation rates (center) and memory of the modelling (lower). See Supplementary Table 1 for depicted age values of the dated samples.

Mechanism of deformation
The wedge-shaped deformations under consideration were first interpreted as water escape structures by Fordinál et al. (2012a) and Kadlec et al. (2015). However, the formation of water escape structures is a process characterized by an overall upward transport of the fluidized sandwater mixture (e.g., van Vliet-Lanoë et al., 2004;Owen et al., 2011). In contrast, the wedgeshaped structures here, as a general rule, are indicative of a downward transport of material. Moreover, water escape structures require sediment saturated with pore water. Bedding dips imply that the strata accumulated on a gently dipping slope of a sand sheet, and wedges formed at the surface, which was most probably within the vadose zone and not saturated by water.
The larger, vertical or subvertical wedges, where blocks of the original well-bedded sand were moving down, often along several planes at once, can be readily explained by the opening of extensional (dilation) cracks. However, many of the narrow fractures do not show any obvious downward movement of sand blocks, and they are quite strongly inclined (Fig. 8). As most of them display no displacement, they cannot be interpreted as faults running up to the surface.
The downward bending of the marginal few centimeters of the sand layers could be explained by the lateral opening of the fracture, but in loose sands the overhanging side of the fracture should collapse immediately.
The rheology of loose sand can be significantly altered by frozen vadose water, though, and the strata can exhibit increased cohesion, both as has been shown experimentally and observed in clastic sediments in glacial and periglacial conditions (e.g., Steidtmann, 1973;Youssef and Hanna, 1988;Fitzsimons et al., 2001;Anderson et al., 2007;Waller et al., 2012;Kim et al., 2016;Fitzsimons and Howarth, 2020). According to the OSL dating results the depositional record in this case was formed during the cold MIS2, and within that, partly during the Last Glacial Maximum. The mean annual temperatures went as low as 0 ± 2°C and mean annual precipitation was estimated to attain only 30-50% of the present-day conditions for the lowland areas of the Pannonian Basin region during the time of deposition (Ruszkiczay-Rüdiger and Kern, 2016). Hence, the freezing of vadose water inside the sediment outcropping in the Bažantnica sandpit and increased cohesion of the strata may be considered a likely option.
The high degree of cohesion may also explain the formation of the layers with chaotic structure in otherwise loose sand. The original structure must have been disturbed by an abrupt trigger mechanism related to delivery of kinetic energy. Sharp contact of the disturbed horizons with the underlying undeformed beds implies that some rheological contrast might have existed in between. The fragmented structure with the fragmentary remains of the original bedding fits well to the expected increase of cohesion due to freezing of vadose pore water (Youssef and Hanna, 1988). The folded layers were probably formed by an analogous process, as they appear in comparable settings to the layers with chaotic structure, but with a different level of cohesion and higher degree of plasticity, likely due to their higher content of liquid water at the expense of ice content (Arenson et al., 2007). Folding is not uncommon in seismically deformed deposits (as reported in e.g., Taşgın et al., 2011;Liang et al., 2021, Sağlam Selçuk andKul, 2021). A variety of cohesion settings might be expected in between the end members of loose sand and sand which is frozen solid.
The near-to-surface location of the deformations is indicated by the truncation of their uppermost parts in every documented case. On the other hand, the erosion was mostly gentle, since the overlying strata are generally conformable (Fig. 7). The erosion affected only few centimeters of the deformed surface, which was irregular due to the disturbance and folding of the uppermost layers as well as due to disruption by fractures and wedges. Eolian activity flattened the irregular sandy surface before the formation of new wind ripples. A significantly higher intensity of erosion produced major erosional surfaces, which truncated a significant part of the deformed horizons and led to angular bedding contacts.
The deformation in Fig. 13 does not penetrate below its 85 cm thickness, and pinches out towards the right side, while the body as a whole is delimited by a deformed basal surface; taken together, these imply the movement of a block of sediment as a mechanism of deformation. The observed deformation probably originated in the marginal part of a slide due to the pressure exerted towards the host sediment of the sand sheet, which is not exposed in the section under consideration (Fig. 13 -inserted block diagram). The sliding of eolian loose sand usually results in large (on the meter scale) massive, structureless bodies (Steidtmann, 1982;Loope et al., 1999;Ford et al., 2016). Nevertheless, the niveo-eolian depositional systems of cold regions often preserve slumped bedding in the depositional record (Koster and Dijkmans, 1988;Dijkmans, 1990). Hence, the freezing of sediment might allow the formation of a complex pattern of both brittle and ductile deformation meso-structures, which is much more common in heterolithic deposits than in granulometrically homogeneous sands (e.g., Alsop et al., 2016;2017). The deformation seen in Fig. 13 displays a complex blocky pattern with a change from extensional to compressional regime within tens of centimeters horizontally, a feature frequently observed in slides (Alsop et al., 2020).

Trigger of deformation
The systematic orientation of fractures (Fig. 8) narrows the possible genetic explanations. In the case of eolian sands, bedding indicates the position of the paleosurface. The strike of the fractures is perpendicular to that of the gently dipping sand beds in the main wall and has no regularity in relation to the bedding observed in the remaining walls of the sandpit, therefore, the formation of the fractures by slope mass movement can be excluded. The inclination of the fractures (relatively large in relation to the vertical) and the uniform orientation of the fractures rules out a periglacial (frost or ice wedge) origin, as well.
Surface rupture by planar fractures is a common feature in seismically active areas with unconsolidated sediments. Fractures without vertical displacement, similar to those observed in the Bažantnica sandpit, are usually either dilation cracks or fractures acting as conduits for liquefied sediment to the surface. Both types form above blind faults (Alessio et al., 2012;Caputo et al., 2012;Moretti et al., 2014;Konrád et al., 2021) or as off-fault features in the vicinity of surface-rupturing faults (e.g., Philip and Meghraoui, 1983;Pollak et al., 2021). These ruptures are usually arranged systematically with respect to the orientation of the seismic wave, parallel with the σ1 and perpendicular to the σ3 stress axes (Burbank and Anderson, 2011). The movement of liquified sand upwards along fractures requires appropriate sediments -it is typical, for example, of alluvial successions -and an earthquake with a minimum magnitude of 5.5 (Guarnieri et al., 2009;Alessio et al., 2012;Amorosi et al., 2016). Structures related to water escape were not found in the sand pit examined here, implying that earthquake magnitudes within the area and time interval under investigation were below 5.5. It has been documented that liquefaction appears in eolian sands at the point of contact with water table (e.g., Chan et al., 2019;Espinoza et al., 2020;Rodríguez-López and Wu, 2020;Konrád et al. 2021). However, no study describing how frozen eolian sand behaves when experiencing a seismic shock could be found in the course of the research presented here.
A process-based model of the origin of the deformations described here is illustrated in Figs. 17 and 18. A significant number of the wedges are filled by the material of beds with chaotic fabric, implying that the soft-sediment deformation of these beds appears just before the formation of the wedges. Firstly, the seismic shock causes disturbance or folding of the uppermost layers of the eolian sediment (Fig. 17B). A fracture disrupts the surface of cohesive sand shortly after, and a collapse wedge forms, filled rapidly by the material of already disturbed topmost beds (Fig. 17C). The increased cohesion also allows formation of an overhanging side of an open rupture. The continuous aggradation of the sand sheet allowed the recording of a number of successive deformation events (Fig. 18). The prevailing conformable contact of the strata overlying the deformed horizons was caused by the smooth beveling of the deformed surface (Fig. 18C). The repetition of earthquakes led to the successive formation of collapse wedges above the same fractures (Fig. 18D). The fractures were then reactivated, ceased to be active, or might have undergone splitting in more active fractures (Fig. 18G). The repeated formation of the wedges is related to the propagation of a fracture across the older wedges, which is expressed as planar inhomogeneity in the generally isotropic fill of a wedge (Fig. 9A,F). Major erosion events appeared less frequently and cut off the upper few tens of centimeters, including a major part of the deformed horizons (Fig. 18E). The slide deformation observed in the upper part of the main wall (Figs. 7, 13) was most probably triggered by a seismic shock, in a way similar to other deformations. Fig. 18. A tentative process-based model of the successive formation of the collapse sand wedges in response to seismic shock and rupture of the surface of frozen partly cohesive sand in an aggrading sand sheet.

Spatial relationship of observed deformations to the Vienna Basin Transfer Fault and the recent stress field
Although the neotectonic kinematics of the VBTF in the Slovak part of the Vienna Basin was not investigated in detail, thorough analyses of reflection seismics, focal mechanisms, morphotectonic features and trenching resulted in an accepted kinematic model of the VBTF in the Austrian part of the basin, located ca. 20-40 km towards SW from the Bažantnica sandpit (e.g., Decker et al., 2005;Hinsch and Decker, 2011;Tary et al., 2015;Weissl et al., 2017;Hintersbergher et al., 2018). The VBTF displayed sinistral transtensional strike-slip kinematics with SW-NE trending of faults and prevailing W-E orientation of extension during the period of deposition of the succession in the Bažantnica sandpit (e.g., Decker et al., 2005). The dominant N-S orientation of documented fractures (Fig.  8) is consistent with the suggested stress field and agrees with their interpretation of it having been formed by seismic events originating in the activity of the VBTF.
The deformations described herein are considered to have been formed by a seismic shock. Hence, their repeated occurrence in a vertical succession allows for the determination of recurrence periods of earthquake events. The wedges deforming a single horizon are treated as the result of a single event. The number of distinct deformation events on the main wall is 21 (Suppl. Table 2).
The Bacon age-depth model-based age ranges and mean ages were lined up against the depths of deformed horizons and are included in Supplementary Table 2. Since the age error bars of the age-depth model are related to the error bars of the OSL ages, they reach relatively high values with a mean σ1=837.44 ka (Suppl . Table 1). Therefore, the ages of deformation events do exhibit error bars significantly exceeding the time span between most of the events (Fig.  19A).
The obtained distribution of mean ages of deformation events allows the determination of the recurrence periods of these events (Fig. 19B). It is important to mention that this calculation is biased by the much lower number of dating results in comparison to the number of interpreted events. However, a higher number of OSL ages would most probably not decrease the standard deviations of the age-depth model, as this is mainly related to the ratio between dating error bars and the depositional time span of the succession.  Table 2 for depicted values.
The mean age distribution of collapse wedges shows a relatively uniform pattern in the lower and upper portions of the section, while the central part contains only three events separated by periods of 435 and 492 years, respectively (Fig. 19A). The upper time gap without collapse wedges is fully exposed in the outcrop, therefore it represents a real quiescence of fault activity. The lower gap is partly covered, and it is possible that a few deformation events might have gone unnoticed in this interval. The recurrence period of deformation events is generally between 40 and 190 years, with a mean value of 142.6 ± 114.6 years (Fig. 19B).

Conclusions
This study yielded a detailed analysis of soft-sediment deformation structures (SSDS) present in outcrops of an Upper Pleistocene eolian sand sheet in the eastern Vienna Basin, Central Europe. Sedimentological study was combined with dating using optically stimulated luminescence. The observed deformations include collapse wedges, beds with chaotically disturbed horizons, folded beds and sliding. The mechanism of deformation implies some cohesion of the deformed strata, which is attributed to seasonal frost and the presence of vadose water within the sediment.
The systematic orientation of fractures, their joint presence in specific horizons together with chaotic and folded beds, and the location of the study area on an active Quaternary fault system suggest that the deformations were triggered by repeated seismic shocks. The age distribution of the deformation events based on the Bayesian age-depth model yielded a recurrence period of 142.6 ± 114.6 years. The recurrence period calculation is biased with a relatively high standard deviation of the age-depth model, associated with the high ratio between OSL uncertainties and the duration of dated period, but it does nonetheless give an order of magnitude for the recurrence times.
The formation of SSDS in eolian sands under cold climates is rare and needs to be examined in more detailed studies of periglacial successions elsewhere. Modeling under laboratory conditions is needed to confirm the process-based interpretation of the origin of deformation structures presented here. The findings of present study imply that collapse wedges, chaotically disturbed and folded beds are important proxies for paleoseismic activity, and their recognition will help to identify past earthquakes in the depositional record. An unexpected potential of periglacial eolian deposits to record seismic events is another noteworthy piece of information provided by the study.