Late Pleistocene glaciation history of the southern Black Forest, Germany: 10 Be cosmic-ray exposure dating and equilibrium line altitude reconstructions in Sankt Wilhelmer Tal

: During the Late Pleistocene, an ice cap temporarily rested on the highest summit of the Black Forest, Feldberg, and on the surrounding region. Moraines inside the last glaciation maximum ice extent document subsequent glacial standstills and/or re ‐ advances, but the chronology of the deglaciation remains largely unknown. In Sankt Wilhelmer Tal, moraines were mapped, and suitable moraine boulders were sampled for 10 Be cosmic ray exposure (CRE) dating. Equilibrium line altitudes (ELAs) during moraine formation were reconstructed to evaluate whether these can be used for local stratigraphical correlations. Geomorphological mapping revealed numerous ice ‐ marginal positions in the main valley and in two tributary valleys. CRE ages and ELAs indicate two discrete phases of glacial standstills and/or re ‐ advances by 17 – 16 ka at the latest and no later than 14 ka, respectively. Differing ELAs across the study area preclude the use of ELAs for local stratigraphical correlations. Recalculated 10 Be CRE ages from other localities in Central Europe indicate similar periods of moraine formation, thus raising the question of a common climatic forcing. Additional sets of CRE ages are needed to answer this question. In addition, future studies should concentrate on determining the age of the last glaciation maximum in the Black Forest.


Introduction
The geomorphological record of ice sheets, ice caps and glaciers is commonly employed for the reconstruction of past climatic changes (e.g.Stokes et al., 2015;Rea et al., 2020).If factors other than climate, such as topography (Barr and Lovell, 2014), do not significantly influence their mass balance, variations in ice extent document changes in precipitation and/or temperature (e.g.Kerschner and Ivy-Ochs, 2008).Information on the past extent of smaller ice bodies is particularly valuable, as they dynamically react to both climatic trends on long timescales and short-term climatic shifts (Mackintosh et al., 2017).
Research on the landform record of small-scale Quaternary glaciations in low mountainous areas of Central Europe (Mittelgebirge) has a long tradition.Traces of the glaciations of the Black Forest and Vosges were first recognized and described in the first half of the 19th century (Haase, 1966;Mercier and Jeser, 2004) when the theory of widespread past glaciations was developed in the Alps.Since then, glacial landforms in low mountainous regions in Central Europe have been systematically investigated and mapped (Steinmann, 1902;Ergenzinger, 1967;Duphorn, 1968;Andreoli et al., 2006;Krause and Margold, 2019).The study of sediments in lakes and mires inside of moraines has provided insights into the chronology of the glaciations (Woillard, 1978;Lang et al., 1984;Vočadlová et al., 2015).The emergence of cosmic-ray exposure (CRE) dating in the 1990s greatly facilitated determination of the age of ice-marginal positions.The first CRE ages from a low mountainous area in Central Europe were acquired in the Vosges (Mercier et al., 1999) and the Giant Mountains (Mercier et al., 2000).Further sets of CRE ages of moraines were later obtained for the Bavarian Forest (Reuther, 2007), the Bohemian Forest (Mentlík et al., 2013) and for the Giant Mountains (Engel et al., 2011;).
Well-preserved moraines are found at various localities in the southern Black Forest (Liehl, 1982;Metz and Saurer, 2012;Hemmerle et al., 2016), but they have hitherto not been directly dated.The outermost of these landforms indicate the presence of a 1000-km 2 ice cap during the Late Pleistocene.It covered the highest summit of the Black Forest, Feldberg [1493 m above sea level (a.s.l.)], and the surrounding region (Liehl, 1982;Metz and Saurer, 2012;Hemmerle et al., 2016).The outlet glaciers of this ice cap were up to 25 km long (Hemmerle et al., 2016) and one of them had a maximum thickness of 440 m (Sawatzki, 1992).Groups of morphostratigraphically younger moraines have been preserved inside the supposed maximum ice extent of the last glaciation.A recent geomorphological study revealed numerous ice-marginal positions in the area NW of Feldberg (Hofmann et al., 2020), but the chronology of their formation remains largely unknown.The only conclusions on the deglaciation chronology were drawn by analogy with data from sediment cores from lakes and mires inside of moraines (summarized in Lang, 2005).As the estimation of the age of the lithostratigraphic units in the cores relies only on the apparent, but not properly documented presence of the Laacher See Tephra and not on radiocarbon ages, the suggested chronological correlations should be considered with greatest caution.
The moraines in the southern Black Forest represent prime candidates for palaeoclimatological reconstructions, as the high number of ice-marginal positions suggests highly dynamic glaciers.Dating the former glacier extents could thus provide information on both long-term climatic trends and climatic oscillations on short timescales.In addition, the use of glaciers as palaeoclimatological archives would allow us to compensate for the scarcity of biological proxies in mountainous regions of Central Europe.Sedimentary records from these areas documenting palaeoenvironmental changes before the rapid warming at around 14.6 ka are scarce (Maier et al., 2021), with only a few exceptions (Becker et al., 2006;Duprat-Oualid et al., 2017).
Establishing a robust chronology of glacier fluctuations in the Black Forest represents a prerequisite for glacier-based palaeoclimatological reconstructions.As previously mentioned (Hofmann et al., 2020), suitable boulders for CRE dating exist on most of the moraines in Sankt Wilhelmer Tal, a valley NW of Feldberg.Since quartz-rich lithologies dominate the valley (Hann et al., 2011), this area was deemed particularly suitable to obtain the first 10 Be CRE ages for moraines of the Black Forest.Equilibrium line altitudes (ELAs) during moraine formation were reconstructed to assess whether palaeo-ELAs can be employed for local stratigraphical reconstructions.To evaluate whether the reconstructed chronology is consistent with existing chronologies from mountain regions in Central Europe, previously published 10 Be CRE ages from these areas were recalculated according to recent 10 Be production parameters.

Study area
The study area consists of Sankt Wilhelmer Tal sensu strictu as well as two tributary valleys, Wittenbach and Katzensteig (Fig. 1).The upper part of the main valley is commonly referred to as Napf.The study area is delimited to the south by the Stübenwasen Massif reaching an elevation up to 1386 m a.s.l.(Figs. 1 and 2C).The Hochfahrn-Toter Mann Massif demarcates the study area to the north (Fig. 1).At an elevation of 650 m a.s.l., the main valley merges with Buselbachtal into Bruggatal.Further north, this valley leads into Dreisamtal, an east-west-trending tributary valley of the Upper Rhine Graben (Fig. 1).
Quartz-rich lithologies dominate the study area.Gneiss and migmatite are the predominant lithologies on the north-eastern valley wall, whereas paragneisses, migmatite and isolated patches of amphibolites occur on the south-western valley wall as well as in the Katzensteig, Wittenbach and Napf cirques (Hann et al., 2011).
The main valley is one of the most impressive trough valleys in the entire Black Forest (Fig. 2A).The upper part of the main valley is a well-developed stairway cirque with three steps at 1050, 1200 and 1300 m a.s.l.(Fig. 2C; Schreiner, 2011).Morphologically distinct steps separate the Wittenbach and Katzensteig from the main valley (Fig. 2B,C).The upper ends of these hanging valleys are well-developed staircase cirques (Fig. 2B,C).The Katzensteig cirque has two floors at 1020 and 1050 m a.s.l.The tongue basin on the lowermost cirque floor is today occupied by a mire (Fig. 1).The Wittenbach cirque features three individual cirque floors at 990, 1110 and 1170 m a.s.l.(Schreiner, 2011).Further small glacial cirques have been observed on the southwestern flank of the main valley NW of the Katzensteig cirque (Fig. 3; Hofmann et al., 2020).

Late Pleistocene glacier fluctuations in the study area
An outlet glacier of both the ice cap on the Feldberg and the ice cap on the summit area of Schauinsland covered almost entirely the study area during the local last glaciation maximum (Fig. 1;Liehl, 1982;Schreiner, 2011;Hemmerle et al., 2016).As previously discussed (Schreiner, 2011;Metz and Saurer, 2012;Hofmann et al., 2020), the glacier front was situated somewhere in Bruggatal (Fig. 1).Its exact location remains unknown, as no terminal moraine has been preserved.The outermost ice-marginal landform in the study area lies on the south-western valley wall at the entrance to the main valley.It is a medial moraine consisting of boulders derived from a glacially transported rockslide ('SW-18′ in Fig. 3; Hüttner, 1967;Hofmann et al., 2020).
Multiple terminal moraines have been mapped further upvalley as well as in the tributary valleys.If not explicitly stated otherwise, we refer to the latest study on moraines (Hofmann et al., 2020).A set of moraines (SW-17 to SW-14; Fig. 3C) lies on the NE-facing flank, subparallel to and high above the valley floor (at 900-930 m a.s.l.).The geometry of their crests suggests that the glacier front was located somewhere upvalley from the medial moraine (Fig. 3A).The southernmost of these landforms (SW-14) has an arcuate shape and surrounds a small peat bog.The shape of this moraine indicates that the glacier from the Katzensteig cirque joined the glacier in the main valley during moraine formation (Fig. 3C).A prominent terminal moraine is emplaced at the north-western end of the village of Sankt Wilhelm (SW-13; Figs.2A and 3C).The double-crested moraine of the icemarginal positions SW-12 and -11 occupies the area SE of the village Figs. 2A and 3C).The geometry of the crests suggest that the Katzensteig glacier was disconnected from the main valley glacier during moraine formation (Fig. 3C; Schreiner, 2011).A relatively small moraine is positioned further SW (not shown in Fig. 3).At an elevation of about 780 m a.s.l., upvalley from the double-crested moraine, there is a group of arcuate terminal moraines (SW-10 to -8; Fig. 3C).Another moraine is situated 200 m further east (SW-7; Figs.2A and 3C).The terminal moraine on the southern valley wall 300 m further SE consists almost entirely of boulders (SW-6; Fig. 3C).Two moraines that are only a few metres high lie on the opposite valley wall further upvalley (SW-5 and -4; Fig. 3A).A double-crested terminal moraine is situated at an elevation of 910 m a.s.l (SW-3 and -2; Fig. 3A).The uppermost terminal moraine in the main valley lies on the eastern part of the Napf cirque headwall at an elevation of 1030 m a.s.l.(Fig. 3A).
Multiple moraines are emplaced in the Wittenbach hanging valley (Fig. 3).The northernmost of these progresses from the eastern valley wall in a north-westerly direction down to the bottom of the main valley (Fig. 3A).As the northern end of the ridge is situated on the bottom of the main valley, it is unlikely that the feature is a medial moraine.The landform is rather a terminal moraine of the Wittenbach cirque glacier.Another moraine has been observed at around 950 m a.s.l. on the valley floor.It has a smooth surface and consists of two parts separated by a stream Figs. 2C and 3B).Two moraines are positioned on the eastern valley flank further SW (WB-2 and -1; Fig. 3B).The western moraine leads into an overprinted moraine that surrounds a marshy area on the flat valley floor (WB-1; Fig. 3B).On the western valley flank at an elevation of 1000-1050 m a.s.l., another moraine has been observed (WB-2; Fig. 3B).Further terminal moraines are situated at 1100, 1170 and 1210 m a.s.l.(not shown in Fig. 3).
Numerous moraines have been mapped downvalley from the Katzensteig cirque and in the cirque sensu strictu.A smooth moraine has been observed in the area east of the SW-14 sediment ridge (KS-6; Fig. 3C).The mapped moraines further south and SE are subparallel to the valley floor and only up to a few metres high (KS-5 and -4; Fig. 3C).The most prominent terminal moraines are found in the cirque.The northernmost of these (KS-3 and -2) emerge from the western part of the cirque's headwall and surround the well-developed tongue basin on the lowermost cirque floor (Figs.2B and 3D).A relatively short moraine is found further west (KS-1; Fig. 3D).The upper cirque floor is delimited to the north by an east-west-orientated terminal moraine (KS-1; Figs.2B and  3D).Multiple moraines are observed on the eastern headwall of the cirque (KS-4 to -7; Figs.2B and 3D).
Previous investigators (Erb, 1948;Liehl, 1982;Schreiner, 2011) assigned the moraines in the main valley and the tributary valleys to the last deglaciation and proposed that the outlet glacier probably first evolved into a valley glacier and later split into individual cirque glaciers.Due to the lack of chronological data, these moraines were correlated with glacial stades named after groups of terminal moraines NE of Feldberg (cf.Hofmann et al., 2020).As, so far, no attempts have been made to directly date these moraines with modern techniques, these stratigraphic correlations should be considered with caution.
According to data from sediment cores from the Katzensteig mire obtained by the Baden-Württemberg State Institute for the Environment, Survey and Nature Conservation (LUBW), the Laacher See Tephra (deposited at 13 006 ± 9 cal a BP; Reinig et al., 2021) is apparently present slightly above the bottom of the mire (cf.Hofmann et al., 2020).This suggests that the terminal moraines at the margin of the mire are at least 13 ka old.A bulk sample from the base of a sediment core from a small mire in the upper part of the Wittenbach cirque (Fig. 1) was dated to 11 110-9904 cal a BP [2σ interval; median probability: 10 462 cal a BP; recalculated from Friedmann (1998/9) with the OxCal software (v. 4.4;Bronk Ramsey, 2009) and IntCal20 (Reimer et al., 2020)].This radiocarbon age indicates that the upper floors of the Wittenbach cirque were ice-free at the latest at 10.5 ka and that peat formation started at that time.

Geomorphological mapping and morphostratigraphy
A shaded relief was derived in Esri ArcMap 10.5.1 from a highresolution digital terrain model (DTM) with an xy-resolution of 1 m.Moraines in the study area were mapped in the field with printout versions of the shaded relief (1: 5000 scale).The 'slope' and 'aspect' tools were applied in ArcMap to derive two supplementary raster files for mapping.They were jointly used with scanned field maps.The '3D Analyst' toolbox was used to assess the morphometry of suspected moraines.
Moraines apparently belonging to the same ice-marginal position were then grouped.The following criteria were considered: the geometry of their ends, the size and the elevation on the valley walls.All frontal positions documented by moraines or groups of moraines were then numbered starting with those nearest to the cirques.This allows for adding ice-marginal positions further down the valley in the future, if appropriate.

Be CRE dating
All sufficiently large and stable moraine boulders were sampled for CRE dating.As it has been shown that CRE ages of tall boulders generally cluster better than those of small boulders (Heyman et al., 2016), only large boulders with a height of more than ~1 m were chosen.To avoid underestimated CRE ages due to boulder toppling and rotating or effects of post-depositional and post-stabilization exhumation, only well-embedded boulders on the crest or distal side of moraines were targeted.As landform stability exerts great influence on CRE age distributions (Tomkins et al., 2021), boulders on matrix-poor and debris-rich moraines were considered particularly suitable.
According to sampling guidelines, flat and horizontal surfaces on moraine boulders should preferably be chosen for CRE dating (Ivy-Ochs and Kober, 2008).As this was often not possible, dipping surfaces with a constant angle were targeted.Rock surfaces near edges were avoided to prevent edge effects (Masarik and Wieler, 2003).A battery-powered saw, a chisel and a hammer were used for sampling.To adjust the 10 Be production rate to the sampling sites (n = 25; Table 1), the coordinates of the sampling surfaces were determined with a global navigation satellite system (Leica CS20 controller and Leica Viba GS14 antenna).The elevation of the sampling sites was retrieved from the high-resolution DTM mentioned above.Pairs of azimuth and elevation angles at the sampling sites were recorded with a handheld SUUNTO inclinometer to determine the topographic shielding factor with an online topographic shielding calculator (Balco, 2018).A dense coniferous forest around the SW-17 boulder prevented us from measuring azimuth and elevation angles.The topographic shielding factor for this sampling site was determined with an ArcGIS toolbox (Li, 2018) and a DTM (xy-resolution: about 30 m at the equator) derived from data of the Shuttle Radar Topography Mission (SRTM; NASA Jet Propulsion Laboratory, 2013).As both approaches yielded similar shielding factors for all remaining boulders (R 2 = 0.93; p < 0.01), those determined with the toolbox of Li (2018) were chosen for 10 Be CRE age calculations.The mass and the thickness of the rock fragments in each sample were recorded to compute the mass-weighted average of the sample's thickness for CRE age calculations.A detailed sample documentation is provided in the Supporting Information (Appendix S1).
10 Be geochemistry Quartz separation from the rock samples and 10 Be extraction from the purified quartz were conducted in the laboratory facilities of the University of Freiburg (Germany) and the Laboratoire National des Nucléides Cosmogéniques (LN 2 C) in Aix-en-Provence (France).The separation and purification of quartz follows the procedure outlined in Hofmann et al. (2019), with one additional magnetic separation step to further purify the quartz before etching with 48% HF.For this, the acid-leached samples were spiked with magnetic powder and passed through a Frantz magnetic separator.
The 10 Be concentrations in the samples were derived from accelerator mass spectrometry (AMS) measurements at the French AMS national facility ASTER (Accélérateur pour les Sciences de la Terre, Environnement, Risques; Arnold et al., 2013) at CEREGE (Centre Européen de Recherche et d'Enseignement des Geosciences de l'Environnement) in Aix-en-Provence (France).During AMS measurements on the KS-2c sample, the 9 Be currents were very low, ranging between 0.1 and 0.2 μA.Hence, the 10 Be concentration presented below should not be considered reliable and, thus, no CRE age was computed.The measured 10 Be/ 9 Be ratios were normalized with respect to the in-house standard STD-11 using an assigned 10 Be/ 9 Be ratio of 1.191 ± 0.013 × 10 −11 (Braucher et al., 2015) and the Be halflife of 1.387 ± 0.01 × 10 6 years (Chmeleff et al., 2010;Korschinek et al., 2010).The 10 Be/ 9 Be ratio uncertainty stated in Table 2 comprises the measurement uncertainty (counting statistics), the uncertainty of average standard measures and the systematic error of ASTER (0.5%; Arnold et al., 2010).The 10 Be concentrations in the samples were corrected with the 10 Be concentration in a batch-specific chemical blank.

Determination of 10 Be CRE ages
Apparent 10 Be CRE ages were calculated with the online cosmic ray exposure program (CREp; Martin et al., 2017) available at https://crep.otelo.univ-lorraine.fr(last accessed: 1 September 2021).The following parameters were selected: the 10 Be production rate deduced from rock samples from the Chironico landslide (southern Switzerland; Claude et al., 2014), which is the closest 10 Be production rate reference site to the study area, time-dependent 'Lm' scaling (Nishiizumi et al., 1989;Lal, 1991;Stone, 2000;Balco et al., 2008), the ERA atmosphere model (Uppala et al., 2005), as suggested by Martin et al. (2017), and the atmospheric 10 Be-based geomagnetic database of Muscheler et al. (2005).The assumed spallogenic 10 Be production rate amounts to 4.10 ± 0.10 atoms g −1 quartz a −1 at sea level and high latitudes.As the sampled moraine boulders consist of quartz-rich lithologies, the density of quartz (2.65 g cm −3 ) was assumed.The CRE ages were not corrected for denudation, vegetation cover or snow shielding.
None of the sampled boulders showed protruding quartz veins.This observation tentatively suggests that postglacial denudation did not significantly affect the sampled boulders.Reuther (2007) inferred a denudation rate (0.24 cm a −1 ) in a similar setting in the Bavarian Forest (recalculated from Reuther, 2007, with the parameters listed above).Assuming this denudation rate for CRE age calculations would have resulted, on average, in ~2.8% older CRE ages.The CRE age difference amounts to 0.8 ka for the SW-18a boulder with the oldest CRE age, whereas it is negligible for the boulder with the youngest CRE age (10 a).Note that the sampling surfaces on the SW-18a, SW-18b and SW-18c boulders were covered by a few centimetre-thick organic soil.Soils on rock surfaces enhance chemical weathering rates, as they increase the availability of moisture for weathering (Ahnert, 2009).The The boulders on the moraines of the ice-marginal positions SW-2, -15, -16, -17 and -18 in the main valley lie in forested areas.Vegetation cover alters the 10 Be production rate by a few per cent (e.g.Kubik et al., 1998).Reuther et al. (2011) incorporated the effect of vegetation cover on CRE ages of boulders with biomass estimates.She had already noted that the effect on the resulting CRE ages is rather small (about 4%; Reuther, 2007).Even in extreme cases, such as trees growing on top of boulders, CRE ages did not change significantly.The sampling surfaces on the SW-18a, SW-18b and SW-18b boulders were covered by an about 5-cm-thick organic soil and lichen which may have additionally slowed down the accumulation of in situ produced 10 Be.
According to field observations in 2021 (with a rather cool spring and summer in the region), the sampled boulders in the Katzensteig and Napf cirques as well as in the main valley were covered by a several decimetre-thick snow cover in winter.The sampled boulders are not located in windswept areas where snow drift exposes their surfaces.According to data of the German Weather Service (DWD) from the weather station at Titisee (850 m a.s.l.), located several kilometres NE of the study area, snow cover lasted, on average, for 4 months per year in the 1961-1990 period.Average snow cover amounted to about 30 cm.Assuming a similar snow cover on top of the sampled boulders, a snow density of 0.3 g cm −3 , an attenuation length for fast neutrons in snow of 109 g cm −2 (Zweck et al., 2013), the commonly used equation 3.76 in Gosse and Phillips (2001) yields a snow shielding factor of ~0.97.Incorporating this factor would yield ~2.6% older CRE ages.The climate in the Alps and their forelands strongly varied during the duration of exposure (see Heiri et al., 2014, and references therein), thus raising the question of whether snow cover in the period 1961-1990 is representative.We conclude that the CRE ages presented below may be underestimated by a few hundred years.

Computation of landform ages
Multiple boulders were sampled per terminal moraine, if possible, to allow for detecting potentially underestimated and overestimated CRE ages.The assessment of the CRE ages follows the procedure in version 3 of the online calculator formerly known as the CRONUS-Earth online calculator (Balco et al., 2008).See Balco (2017) for further details.
Reduced chi-squared (χ 2 R ) statistic was first performed for sets of at least three CRE ages.This test allows for determining whether the variation of the CRE ages from the same landform is predominantly caused by analytical uncertainties (χ 2 R ≈ 1) or by geomorphological factors, such as post-depositional exposure (χ 2 R ≫ 1).Reduced chi-squared is then compared with a critical value from a standard χ 2 -table, whereby the degree of freedom is n − 1.In this study, the confidence interval was set to 95%.The critical value is then divided by the degrees of freedom.If χ 2 R is lower than this value, the hypothesis that the data form a single population cannot be excluded at 95% confidence.A higher χ 2 R implies that measurement uncertainties do not fully account for the scatter in the CRE ages (Balco, 2011).
For sets of CRE ages from terminal moraines yielding a χ 2 R lower than the critical value, the arithmetic mean CRE age was chosen as the landform age.Its uncertainty was calculated by adding the average analytical uncertainty and the 10 Be-production † During AMS measurements on the KS-2c sample, the 9 Be currents only ranged between 0.1 and 0.2 μA.Hence, the 10 Be concentration may be out of the 1σ range, and it should not be considered reliable.Thus, no CRE age was computed.
GLACIATION HISTORY OF THE SOUTHERN BLACK FOREST rate error in quadrature, as done elsewhere (e.g.Le Roy et al., 2017).If χ 2 R turned out to be higher than the critical value, the CRE age with the worst χ 2 R was considered an outlier.This procedure was repeated until the remaining CRE ages yielded an acceptable χ 2 R value.Not more than half of the CRE ages from the same landform were removed.

Recalculation of 10 Be CRE ages from key moraine records in Central Europe
To allow for pertinent comparison, 83 previously published 10 Be CRE ages for key sites in mountainous regions in Central Europe and their forelands were recalculated (Tables S1 and  S2).As the climate south of the main weather divide in the Alps differs substantially from the climate further north (Glaser et al., 2013), the area of interest was restricted to the region north and NE of the main weather divide.
All necessary data for recalculation of the 10 Be CRE ages were either retrieved from the original publications or obtained from the authors.Outliers were excluded as suggested by the authors.Recalculation of the 10 Be CRE ages was performed in the CREp calculator with the same parameters listed above.This calculator requires 10 Be concentrations computed with the 07KNSTD standardization as input (Martin et al., 2017).As most of the previously published 10 Be concentrations were determined with other standardizations, they were first recalculated with appropriate conversion factors.These were retrieved from a MATLAB script of version 2.3 of the online exposure age calculator, formerly known as the CRONUS-Earth online exposure age calculator (make_al_be_-consts_v23.m; Balco, 2007).If topographic shielding factors were not reported in the original publications, they were determined with an SRTM-DTM (xy-resolution: about 30 m at the equator; NASA Jet Propulsion Laboratory, 2013) and the ESRI ArcGIS toolbox of Li (2018).The ages were not corrected for snow cover and postglacial denudation.

Glacier and ELA reconstructions
The ELA is commonly defined as the zones of a glacier where the mass balance is zero on an annual timescale (Bakke and Nesje, 2011).Reconstructing ELAs reveals palaeoclimatic information if factors other than climate do not significantly influence its position (Mackintosh et al., 2017).ELAs are also used as a tool for stratigraphic correlations of terminal moraines at the local scale when topographic and lithological conditions are similar (Reitner et al., 2016).
The aim of the ELA reconstructions for this study was twofold: first, we aimed to evaluate whether ELA reconstructions are a suitable tool for relative dating of moraines in the southern Black Forest.Finding an appropriate relative dating technique is of particular interest for future studies, since the lack of suitable boulders hampered CRE dating of the terminal moraines in the Wittenbach hanging valley and will certainly be an obstacle in future studies.ELA reconstructions could be one potential way to circumvent this issue.Several observations suggest similar ELAs in the Napf, Wittenbach and Katzensteig cirques at the same time.These cirques are orientated towards the NE, the cirques' morphology is very similar, as they feature steep western and gently sloping eastern headwalls, and the same bedrock lithologies occur in all cirques.Potentially existing factors other than climate should thus influence the position of the ELA in a similar way.ELA reconstructions should yield similar ELAs for moraines with the same 10 Be CRE age.Second, determining ELAs during moraine formation provides the opportunity for future reconstructions of palaeo-precipitation in the southern Black Forest.
The two ELA reconstruction approaches selected for this study, the accumulation area ratio (AAR) and area-altitude balance ratio (AABR) methods, require glacier surface reconstructions which were performed with the GlaRe ArcGIS toolbox (Pellitero et al., 2016).For the first step, the flowlines of the glaciers in the study area were created as shapefiles in ArcMap 10.5.1.Points with a spacing of 50 m at the flowlines were subsequently created using the 'interval nodes' tool in GlaRe.The 'flowline ice thickness' tool enabled the ice surface elevation at these points to be calculated.This tool requires the basal shear stress as primary input.Thanks to the presence of terminal moraines, reconstruction of the ice margin near the glacier fronts turned out to be straightforward.As suggested (Pellitero et al., 2016), the shear stress was progressively adjusted to make the reconstructed ice thickness fit to these landforms.However, this was not possible further up the glaciers due to the lack of landforms indicative of former ice margins.Thus, the default shear stress of 100 000 Pa in the GlaRe toolbox was assumed in these areas.This is a reasonable assumption, as the basal shear stress beneath glaciers normally varies between 50 000 and 100 000 Pa (Paterson, 1994).Nevertheless, the variation of previously published shear stress values is large and even higher basal shear stress values have been reported (Paterson, 1970;Boulton et al., 1979;Cohen et al., 2000).The choice of the basal shear stress is thus considered the major source of uncertainty in our glacier reconstructions.Since it is unrealistic that the driving stress of topographically constrained glaciers is entirely supported by the basal shear stress (Benn and Hulton, 2010), the ice thickness at points along glacier flowlines in topographically constrained areas was recalculated by incorporating a dimensionless shape factor (f).It was derived from automatically created cross-sections using the 'automatic ice thickness recalculation with f factor' tool in GlaRe.
For the next step, contour lines of today's surface were derived from a high-resolution DTM with an xy-resolution of 5 m.Glacier contours were drawn from the interval nodes to the contour lines of today's surface according to the ice surface elevation at the interval nodes.Following the approach introduced by Sissons (1974), convex and concave contour lines were drawn in the suspected ablation and accumulation areas of the glaciers, respectively.The contour lines were transformed into digital elevation models (DEMs) of the glaciers by applying the 'from contour to DEM' tool in the ArcGIS ELA calculation toolbox of Pellitero et al. (2015).
ELAs in the Alps have traditionally been determined with the surface reconstruction method, which assumes a fixed proportion between the accumulation and ablation areas of a glacier (Pellitero et al., 2015).Gross et al. (1977) found that an AAR of 0.67 is appropriate for glaciers in the European Alps.Kern and László (2010) later showed that this value is only appropriate for larger glaciers and that smaller AARs are sufficient to keep small glaciers in a steady state.We argue that the glaciers in the Alps in their dataset are probably the best analogues for the former glaciers in the study area.Based on the data presented in Kern and László (2010) where s is the surface of the glacier (km 2 ).We assume that the error of the AARs calculated with Equation (1) does not depend on the size.Thus, the differences between the actual AARs and the AARs derived with Equation (1) was computed for every glacier in the Alps included in the study of Kern and László (2010).The standard deviation of these values (0.056) was considered the AAR uncertainty.AAR-ELAs were determined with the ArcGIS toolbox of Pellitero et al. (2015).
The AAR method has been heavily criticized, as it neglects the hypsometry of a glacier which may significantly influence the position of the ELA (Pearce et al., 2017).In contrast, the AABR method explicitly takes this factor into account (Rea, 2009).Rea (2009) published a large global dataset with AABRs.To ensure comparability with the ELAs derived with the AAR method, we chose the mean AABR for the European Alps (1.59) derived from data of 12 glaciers.However, as mentioned by Boston et al. (2015), most of them were in retreat, which may induce errors in ELA reconstructions.To take the high variation of the balance ratios of the glaciers in the Alps into account (1σ = 0.59), each balance ratio from the Alps was used to compute the ELA with the toolbox of Pellitero et al. (2015).The contour line interval was set to 10 m during ELA calculations and, thus, the ELA uncertainty is 5 m (Pellitero et al., 2015).The standard deviation of the ELAs computed for each glacier was considered the ELA uncertainty.The error of the ELA calculation toolbox was added in quadrature.

Mapping of moraines
A detailed geomorphological map of the study area has already been published (Hofmann et al., 2020).Therefore, we only present changes in light of new field evidence.An updated map of terminal moraines and ice-marginal positions in the study area is given in Fig. 3.
Field mapping of the area around the double-crested moraine of the ice-marginal positions SW-11 and -12 revealed that there is not a distinct sediment ridge uphill from the double-crested moraine.Rather, the moraine's outer crest is overlain by a cone-shaped hump.Due to the positioning on the moraine crest and shape of the feature, we propose that the landform was emplaced during a mass movement after moraine formation.Hence, the hump should not be classified as a terminal moraine, as done in previous work (Hofmann et al., 2020).As outcropping bedrock was observed at previously mentioned moraines upvalley from the lowermost floor of the Wittenbach cirque (Hofmann et al., 2020), these landforms should not be classified as terminal moraines.
Overall, moraines of 18 ice-marginal positions have been mapped in the main valley.Moraines of multiple ice-marginal positions are found in the Wittenbach and Katzensteig hanging valleys (Fig. 3).

Be CRE ages
Apparent CRE ages of the sampled moraine boulders in ka (thousand years before 2010 CE) are given in Fig. 4 as well as in Table 2.For landform ages, i.e.CRE ages of terminal moraines as well as χ 2 R values, see Table 3.Three boulders on the SW-18 moraine yielded ages between 19.9 ± 1.0 and 16.9 ± 0.8 ka.As χ 2 R (4.03) exceeded the critical value, the age of the SW-18a boulder with the worst χ 2 R was considered an outlier.The two remaining ages led to a landform age of 17.2 ± 0.8 ka.Only one boulder on the proximal side of the terminal moraine of the ice-marginal position SW-17 proved to be a suitable candidate for 10 Be CRE dating and yielded an age of 16.3 ± 0.7 ka.CRE dating of the only sufficiently large boulder on the SW-16 moraine resulted in a considerably younger age of 3.5 ± 0.3 ka.Two boulders on the moraine of the ice-marginal position 15 (SW-15a and SW-15b) gave ages of 16.4 ± 0.7 and 17.5 ± 0.7 ka, respectively, resulting in a landform age of 17.0 ± 0.7 ka.The lack of suitable boulders prevented sampling of the SW-14 and -13 moraines.Boulder SW-12a from the ice-marginal position SW-12 was dated to 17.0 ± 0.7 ka.The SW-11a boulder from the ice-marginal position SW-11 has an age of 17.4 ± 0.9 ka, whereas boulders SW-11b, SW-11c and SW-11d on the moraine's distal side gave ages of 9.9 ± 0.5, 15.1 ± 0.8 and 17.5 ± 1.0 ka, respectively.The youngest ages from the SW-11b and SW-11c boulders were classified as outliers, resulting in a landform age of 17.5 ± 1.0 ka.
Only one sufficiently large boulder for CRE dating was situated on the SW-10 moraine, yielding a CRE age of 16.0 ± 0.9 ka.A boulder on the morphostratigraphically next oldest moraine gave an indistinguishable age (16.1 ± 0.7 ka).Sufficiently large and stable boulders did not occur on the moraines of the ice-marginal positions SW-7 to -4.The SW-2 boulder on the inner crest of the double-crested terminal moraine further upvalley (ice-marginal position SW-2) was dated to 14.0 ± 0.8 ka.Due to the lack of suitable boulders, the SW-1 moraine was not sampled.
CRE dating was not applied to terminal moraines in the Wittenbach valley and stairway cirque, as only one sufficiently large and stable moraine boulder was recognized during a field survey.
The only sampled boulder on the distal side of the doublecrested moraine of the ice-marginal position KS-3 in the Katzensteig cirque was dated to 14.0 ± 0.7 ka.The ages of six sampled boulders on the inner crest of the moraine and on its proximal side range from 8.8 ± 0.5 to 14.8 ± 0.7 ka.Since χ 2 R (19.18) considerably exceeded the critical value, the ages of boulders KS-2a, KS-2e and KS-2f were discarded, as they yielded the worst χ 2 R .The landform age of 12.8 ± 0.7 ka corresponds to the average of the remaining ages.Boulders KS-1a and KS-1b yielded two differing ages (14.3 ± 0.6 and 4.3 ± 0.4 ka).

Interpretation of 10 Be CRE ages
Considering CRE age uncertainties, the CRE ages presented here are consistent with the morphostratigraphy apart from a few outliers.This study demonstrates that CRE dating is a suitable technique for reconstructing glacier variations in anthropogenically modified environments where the number of suitable boulders is limited.
Geomorphological field evidence suggests that postdepositional processes probably did not affect the sampled boulders on the SW-18 moraine.As the moraine consists almost entirely of boulders, boulder toppling and rotating as well as post-depositional exhumation seem highly unlikely.The consistent CRE ages of boulders SW-18b and SW-18c are thus deemed robust.Pre-exposure to secondary cosmic radiation is the most likely explanation for the outlying CRE age of the SW-18a boulder, as the moraine's sediments probably originate from a rockslide.Since the boulders are partly rounded, we argue that they were transported to their present location by an advancing glacier.The SW-18a boulder may thus have experienced pre-exposure.Recent studies on rockslides demonstrate that the 10 Be CRE ages of boulders may largely exceed the age of the failure (e.g.Hilger et al., 2019).
Apart from the CRE ages of the SW-16, SW-11b and SW-11c boulders, the ages from the ice-marginal positions SW-17 to -2 are consistent with the morphostratigraphy.The age of the SW-16 boulder is considered an outlier.As this boulder is situated on the steep proximal side of the moraine, it may have undergone displacement or toppling.Weathering and spalling may have also affected the sampling surface, as the boulder only has a thin moss cover when compared to the boulders on the moraine of the ice-marginal positions SW-15 and -17 (Appendix S1).As the SW-11b boulder is very angular (Appendix S1), the outlying CRE age of this boulder may be explained by spalling.
The age of the KS-3a boulder is deemed robust, as the boulder is well embedded in the moraine and sufficiently large.The CRE age of the ice-marginal position KS-2 (12.8 ± 0.7 ka) is not considered meaningful.Indeed, the CRE age of the exceptionally large boulder on a moraine of the ice-marginal position KS-1 indicates that this moraine formed not later than 14.3 ± 0.6 ka.Hence, the moraines of the ice-marginal position KS-2 must be at least 14.3 ± 0.7 ka old.Only the CRE ages of the SW-2d and KS-2e boulders are consistent with the morphostratigraphy and, thus, the arithmetic mean CRE age of these boulders (14.2 ± 0.7 ka) is considered the landform age (Fig. 4B).Although the sampling surfaces on boulders KS-2a, KS-2b, KS-2f and KS-2g were considered undisturbed during fieldwork, the sampling surfaces may have been altered by post-depositional processes, such as weathering or spalling.It should also be kept in mind that the boulders are situated in a landscape which has significantly been modified by human activity.The CRE age of the KS-1a boulder is considered particularly robust, as the boulder is, by far, the largest sampled.The outlying CRE age of the KS-1b boulder may be explained by postglacial weathering.

ELAs during moraine formation
Reconstructed ELAs are presented in Table 4; see also Figs. 5-7.As explained above, ELAs determined with the AABR method are considered more robust.Hence, we only comment on them in the following paragraphs.AAR-ELAs are given in parentheses.
As the position of the glacier front in Bruggatal during the emplacement of the SW-18 moraine remains unknown, the ELA during moraine formation could not be reconstructed.The determination of ELAs during the formation of the terminal moraines of the ice-marginal positions SW-17 to -14 proved impossible for the same reason.
Two scenarios were tested for the well-developed terminal moraine of the ice-marginal position SW-13: according to the first scenario, the landform formed in front of a combined glacier from the Napf, Katzensteig and Wittenbach cirques.This glacier would have had an ELA of 1107 ± 22 m a.s.l.(1062-1112 m a.s.l.).Liehl (1982) hypothesized that the glacier from the Katzensteig cirque did not feed any more ).The equilibrium line was situated at a slightly higher elevation (1136 ± 19 m a.s.l.) when the terminal moraine of the ice-marginal position SW-7 formed (AAR-ELA: 1111-1161 m a.s.l.).An ELA of 1134 ± 21 m a.s.l.(1119-1169 m a.s.l.) was reconstructed for the ice-marginal position SW-6.With 1143 ± 19 m a.s.l.(1118-1178 m a.s.l.), the ELA remained similar during emplacement of the terminal moraines of the ice-marginal positions SW-5 and -4.
The Napf cirque glacier had an equilibrium line at a significantly higher elevation (1195 ± 17 m a.s.l.; AAR-ELA: 1230-1290 m a.s.l.) during formation of the double-crested terminal moraine of the ice-marginal positions SW-3 and -2.The ELA was 1268 ± 13 m a.s.l.(1313-1343 m a.s.l.) when the terminal moraines of the ice-marginal position SW-1 formed.
If the terminal moraine at the entrance to the Wittenbach hanging valley pertains to the Wittenbach cirque glacier, the ELA must have been 1133 ± 19 m a.s.l.(1188-1228 m a.s.l.).Under the assumption that the feature formed at the glacier from the Napf cirque, the ELA was 1145 ± 18 m a.s.l.(1140-1190 m a.s.l.).The equilibrium line was situated at an elevation of 1178 ± 15 m a.s.l.(1213-1253 m a.s.l.) during formation of the overprinted terminal moraine of the icemarginal position WB-3.Similar ELAs (1195 ± 11 and 1196 ± 11 m a.s.l.; AAR-ELAs: 1231-1261 and 1230-1260 m a.s.l.) were reconstructed for the moraines of the ice-marginal positions WB-2 and -1.
When the terminal moraines of the ice-marginal position KS-5 of the Katzensteig glacier formed, the ELA was probably 1076 ± 18 m a.s.l.(1131-1171 m a.s.l.).The glacier had an ELA of 1097 ± 15 m a.s.l.(1152-1192 m a.s.l.) during the emplacement of the terminal moraines of the ice-marginal position KS-4.During formation of the moraines of the icemarginal positions KS-3 and -2, the equilibrium line of the Katzensteig cirque glacier was situated at 1145 ± 10 m a.s.l.(1180-1210 m a.s.l.).A similar ELA (1145 ± 9 m a.s.l.; AAR-ELA: 1180-1210 m a.s.l.) was reconstructed for the icemarginal position KS-1.Table 4. Reconstructed equilibrium line altitudes (ELAs) during ice-marginal positions in Sankt Wilhelmer Tal as well as in the Wittenbach and Katzensteig hanging valleys.ELAs were determined with both the area-altitude balance ratio (AABR) and accumulation area ratio (AAR) method.The AABR-ELAs were computed with a balance ratio of 1.59.ELA uncertainties were determined as follows: the ELA during the ice-marginal positions was determined with 12 balance ratios of glaciers in the Alps (see Rea, 2009).The standard deviation of the ELAs was then derived.The uncertainty of each ELA (5 m) was added in quadrature

Discussion
Late Pleistocene glacial history of the study area The moraine of the outermost securely identified ice-marginal position (SW-18) formed no later than ~17 ka.Overlapping CRE ages suggest that emplacement of the moraines of the icemarginal positions SW-17, −16 and -15 probably occurred shortly thereafter.
As the terminal moraine of the ice-marginal position SW-13 is located at a ~200 m lower elevation than the moraines of the ice-marginal position SW-14, the glacier in the main valley must have undergone a drastic thinning when it retreated from the ice-marginal position SW-14.ELA reconstructions cannot unambiguously answer whether the Katzensteig cirque glacier became disconnected from the main valley glacier at this time.As the equilibrium line of the Katzensteig glacier was situated at 1076 ± 18 m a.s.l.during formation moraines of the icemarginal position KS-5, a lower ELA would have been required for joining the main valley glacier.If the moraine in the main valley formed at the margin of a combined glacier from the Wittenbach and Napf cirques, the ELA would have been 1106 ± 24 m a.s.l.Since we do not know whether the ELA was constant across the study area, it remains unclear whether the Katzensteig glacier still joined the glacier in the main valley.The CRE ages of the moraines of the ice-marginal positions SW-12 to -9 are indistinguishable from those of the SW-18, -17 and -15 moraines (range: 17.5 ± 1.0 to 16.0 ± 0.9 ka).ELA reconstructions reveal similar ELAs during the formation of these landforms.Small variations in the ELA were probably sufficient to trigger a progressive retreat of the glacier in the main valley.
If the sediment ridge at the entrance to the Wittenbach hanging valley is a terminal moraine, it is more likely that it formed at the margin of the Napf cirque glacier.Moraine formation at the margin of the Wittenbach cirque glacier would have required a lower ELA.As this view is not supported by the AAR-ELAs, ELA reconstructions do not allow for unambiguously attributing the moraine to one of the glaciers.
Both the CRE age of the SW-2 boulder and the reconstructed ELA suggest that the moraine of the ice-marginal positions SW-3 and -2 is related to a distinct glacial phase.First, the CRE age of the SW-2 boulder (14.0 ± 0.8 ka) does not overlap with that of the moraine of the ice-marginal position 9 (16.1 ± 0.7 ka).Second, the equilibrium line during moraine formation was situated at a significantly higher elevation (1195 ± 17 vs.1126 ± 19 m a.s.l.).
Overlapping CRE ages suggest that the sampled moraines in the Katzensteig cirque probably formed at the same time (Figs. 4, 6 and 7).The ELA was probably 1150 m a.s.l. in the Katzensteig cirque, whereas it was probably higher (~1200 m a.s.l.) in the main valley (Figs. 6 and 7).Two factors could potentially explain the relatively low ELA in the Katzensteig cirque: the headwalls of the cirque are much steeper than the headwalls of the Wittenbach and Napf cirques.They are located east and NE of a hilly plateau extending from Hirschkopf (Fig. 1) in a westerly direction.Shading and snow drift from the west may have thus resulted in a locally lower ELA.The ELA gradient across the study area precludes the idea that reconstructing ELAs of a few glaciers enables stratigraphical reconstructions.Determining a reliable relative age of the unsampled moraines in the Wittenbach hanging valley is therefore not possible.As demonstrated elsewhere (Tomkins et al., 2018), Schmidt-Hammer exposure dating could be a suitable technique to resolve this issue.The equilibrium line then rose to a considerably higher elevation (1268 ± 13 m a.s.l.) and the moraine of the ice-marginal position SW-1 formed.Whether this moraine is related to a distinct glacial event remains unclear, as the lack of boulders prevents CRE dating of this landform.

Correlation with mountain glaciations in Central Europe
Here we compare the first deglaciation chronology from the Black Forest with existing chronologies from key localities in Central Europe (Fig. 8), which we base on 83 recalculated CRE ages from boulders on moraines (Fig. 9).Individual CRE ages are given in the Supporting Information (Table S2).
The timing of the maximum ice extent in the southern Black Forest during the Late Pleistocene remains unknown.Both CRE ages and ELA reconstructions imply two distinct phases of glacial re-advances and/or standstills during the subsequent deglaciation.During the first phase (not later than 17-16 ka), valley glaciers still existed in this region.They subsequently retreated further and disintegrated into small glaciers.The glaciation of the study area was restricted to the cirques by 14 ka at the latest.
Based on reasoning by analogy, Seret et al. (1990) proposed that the last glaciation maximum in the Vosges occurred at around 60 ka and, thus, considerably earlier than the last glacial maximum, irrespective of its definition (Hughes and Gibbard, 2015).Two moraines in the Wormsa Valley (Fig. 8) attributed to the last deglaciation gave CRE ages of 14.4 ± 1.6 and 13.6 ± 1.6 ka (Fig. 9B; Mercier et al., 1999).As already mentioned by Kaltenbrunn and Preusser (2015), it is highly unlikely that a glacier persisted at an elevation of 525-550 m a.s.l. until ~14 ka and, hence, the ages probably reflect postdepositional processes.An erratic boulder on a moraine in the Altenbach Valley further north (Fig. 8) yielded a CRE age of 22.5 ± 1.9 ka (Fig. 9C; Mercier et al., 1999).Moraines in the Missheimle cirque, one of four cirques in the upper reaches of the valley, stabilized at the latest at 19.3 ± 1.5 and at 16.8 ± 1.5 ka (Fig. 9C; Mercier et al., 1999)  re-advance or standstill in the Lac Noir cirque further north (Fig. 8) occurred no later than 14.7 ± 1.5 ka (Fig. 9D; Mercier et al., 1999).
Last glaciation maximum moraines in the northern foreland of the Alps (in the Aare valley and at Lenzburg; Fig. 8) formed not later than 22.9 ± 1.4 and 21.5 ± 1.0 ka, respectively (Fig. 9E,F; Ivy-Ochs et al., 2004;Reber et al., 2014).The CRE ages are thus slightly younger than radiocarbon and optically stimulated luminescence ages from the same region, suggesting a last glaciation maximum at around 25 ka (Gaar et al., 2019, and references therein).The CRE ages are also younger than radiocarbon ages from the southern foreland of the Alps, pointing to a Late Pleistocene maximum glacier extent at around 26-24 ka (Monegato et al., 2007(Monegato et al., , 2017)).Although CRE ages from the area around Lake Starnberg further east (Fig. 8) seemingly suggest a later Late Pleistocene glaciation maximum (Fig. 9G), these ages (18.1 ± 1.2 and 17.7 ± 1.0 ka) are not consistent with independent chronological evidence.Post-depositional processes, such as boulder tilting, probably explain the underestimated CRE ages (Reuther et al., 2011).
CRE ages from moraines inside the Late Pleistocene maximum ice extent in the northern foreland of the Alps (near Wohlen and in the Aare valley around the city of Bern; Fig. 8) document glacial re-advances and/or standstills not later than 19.5 ± 1.3, 18.2 ± 1.0 and 17.4 ± 0.8 ka (Fig. 9E,F; Reber et al., 2014;Wüthrich et al., 2018).The first unequivocal glacial readvance in the Alps took place during the Gschnitz stadial when roughly 80-90% of the ice volume in the Alps had already been disappeared (Ivy-Ochs et al., 2008).According to CRE ages from the stratotype in the Gschnitz Valley (Fig. 8), this stadial occurred at the latest at 17.9 ± 1.0 ka (Fig. 9H; Ivy-Ochs et al., 2006).
The furthest Late Pleistocene glacier advance in the area around Kleiner Arbersee (Bavarian Forest) culminated at the latest at 21.0 ± 1.0 ka (Fig. 9J; Reuther, 2007).Subsequent glacial re-advances and/or standstills occurred not later than 20.3 ± 1.0, 18.8 ± 0.8 and 16.9 ± 1.1 ka (Fig. 9J; Reuther, 2007).An unsampled moraine further upvalley from the lake could be related to a younger glacial phase.Moraines at similar elevations also occur at other localities in the Bavarian Forest, but they have hitherto not been sampled for 10 Be CRE dating (Hauner et al., 2019).
On the Czech side of the mountains (Bohemian Forest), there is more robust evidence for later glacier oscillations.CRE ages from the Laka Valley (Fig. 8) suggest moraine formation at the latest at 15.8 ± 1.8 and 13.8 ± 1.3 ka (Fig. 9K; Mentlík et al., 2013).Concomitant glacier fluctuations also occurred in the Prášilské Valley several kilometres further SE (Fig. 8) where moraines stabilized not later than 15.4 ± 0.7 and 13.4 ± 1.2 ka (Fig. 9L; Mentlík et al., 2013).The CRE ages of the innermost moraine are confirmed by a basal radiocarbon age in a sediment core from a nearby lake (Stará Jímka) pointing to icefree conditions at around 14 ka (Mentlík et al., 2010).The Late Pleistocene glaciation maximum in this area occurred at the latest at 17.5 ± 1.4 ka (Fig. 9L; Mentlík et al., 2013).
Multiple sets of CRE ages from the Giant Mountains have been published.The Late Pleistocene maximum glacier advance in the area downvalley from the snowy cirques (Śnieżne Kotły; Fig. 8) occurred not later than 18.4 ± 1.0 ka (Fig. 9M; Engel et al., 2014).The next further upvalley moraine formed by 18.1 ± 0.9 ka at the latest.Note that Schmidt Hammer rebound values from these moraines and those further upvalley as well as the moraines' morphology suggest that these CRE ages are probably underestimated (Engel et al., 2014).Further periods of moraine formation occurred not later than 18.4 ± 0.7, 13.6 ± 1.2, 12.0 ± 0.5 and 11.5 ± 0.7 ka (Fig. 9M).The snowy cirques are thus the only localities where CRE ages document moraine formation after the climatic downturn in Central Europe at around 12.9 ka evidenced elsewhere (see Heiri et al., 2014, and references therein).The Late Pleistocene maximum glacier advance in the uppermost Elbe Valley (Labský důl) further SE culminated not later than 17.5 ± 1.5 ka (Fig. 9N; Mercier et al., 2000).This  minimum age broadly agrees with CRE ages of two last glaciation maximum moraines in the Łomnica Valley further east (Fig. 8) that stabilized by 16.9 ± 1.0 and 16.8 ± 0.8 ka at the latest (Fig. 9P; Engel et al., 2011).Moraines further upvalley formed not later than 17.5 ± 0.6, 17.4 ± 0.8, 15.4 ± 0.6 and 13.7 ± 0.9 ka (Fig. 9P; Engel et al., 2011).The oldest CRE ages from moraines in the Úpa Valley further SE suggest periods of moraine formation not later than 17.6 ± 0.7, 17.2 ± 0.7, 16.1 ± 1.8 and 14.5 ± 0.6 ka (Fig. 9Q; Engel et al., 2014).CRE dating of a terminal moraine in the Łomniczka Valley on the Polish side of the mountains (Fig. 8) resulted in a CRE age of 15.1 ± 0.7 ka (Fig. 9R; Engel et al., 2011).
Within the limits of available data, the variation of the 10 Be CRE ages from last glaciation maximum moraines tentatively indicates that the Late Pleistocene glaciation maxima in the mountains of Central Europe and their forelands could have been asynchronous.Figure 9 reveals that the last glaciation maximum in the Giant Mountains no later than 18-17 ka (Mercier et al., 2000;Engel et al., 2011Engel et al., , 2014) ) may have occurred later than the Late Pleistocene glaciation maximum in the foreland of the Alps (22.9 ± 1.4 and 21.5 ± 1.0 ka; Ivy-Ochs et al., 2004;Reber et al., 2014).Unfortunately, this study could not clarify whether the Late Pleistocene glaciation maximum in the Black Forest was synchronous with a last glaciation maximum evidenced elsewhere.Testing the hypothesis of asynchronous last glaciation maxima urgently requires CRE ages from moraines of this phase in the southern Black Forest and in other mountainous regions of Central Europe.
The presence of moraines of a similar age at different sites suggest that the periods of moraine formation might be related to climatic phases that affected at least major parts of Central Europe.Ivy-Ochs et al. (2006) already proposed that the Gschnitz stadial should be regarded as the response to Heinrich Event 1, a phase of massive iceberg discharge into the North Atlantic (Heinrich, 1988;Bond et al., 1992;Clark et al., 2012) between 18.0 and 15.6 ka (Fig. 9T; Sanchez Goñi and Harrison, 2010).Relatively low δ 18 O values in the Greenland Ice Core Project ice core indicate that cool conditions prevailed at this time (Fig. 9S; Johnsen et al., 1997).Reuther et al. (2011) speculated that the formation of the innermost sampled moraine at Kleiner Arbersee is potentially related to this cool period in the North Atlantic region.The second phase of glacial readvances and/or standstills during the deglaciation of the study area could be linked to cool phases before the climatic amelioration in Central Europe starting at around 14.6 ka (Heiri et al., 2014, and references therein).
However, these hypotheses remain tentative for now.Before glacial phases can be robustly correlated with climatic events, we propose that three main issues need to be resolved: first, a more extensive set of CRE ages from the southern Black Forest must be acquired to evaluate whether the glacial phases described here are of regional relevance.This will also allow the establishment of a regional stratigraphy of the last glaciation according to the guidelines of Hughes et al. (2005).Second, existing data from other mountainous regions, particularly chronological data from other records and ELAs, need to be reviewed in more detail for a more holistic view of interdependencies.Third, future work will have to clarify whether moraines formed during climatically controlled glacial advances or during standstills triggered by factors other than climate, such as the retreat to more shaded areas (e.g.Lukas, 2007).

Conclusions
Geomorphological mapping, CRE dating and ELA reconstructions allowed for reconstructing the Late Pleistocene glacial history of the southern Black Forest in unprecedented detail.Geomorphological mapping revealed 18 ice-marginal positions in the main valley and multiple ice-marginal positions in the tributary valleys.For the first time, 10 Be CRE dating was applied to moraines in the Black Forest and revealed that two discrete phases of glacial readvances and/or standstills occurred during the deglaciation of the study area.The moraine of the outermost securely identified icemarginal position formed not later than 17 ka.Overlapping CRE ages indicate further periods of moraine formation no later than 17-16 ka.The ELA was situated at 1110-1140 m a.s.l.during these glacial events.Moraines in the cirques formed during the second phase of glacier activity by 14 ka at the latest when the ELA was situated between 1150 and 1200 m a.s.l.The ELA gradient in the study area suggests strongly that ELAs should not be used for local stratigraphical correlations.Future studies will have to clarify whether the periods of moraine formation proposed here are also documented at other sites in the Black Forest.Furthermore, techniques for determining relative ages of moraines, such as Schmidt Hammer exposure dating, should be tested.
The newly acquired CRE ages significantly enhance the knowledge of glaciation in the mountain regions of Central Europe.Periods of moraine formation in the study area are also documented at other sites in Central Europe.These phases may have been triggered by cooling in the North Atlantic region, but a further assessment requires disentangling climatically driven glacier variations and those caused by other factors such as topographic effects.In addition, there is an urgent need for additional chronological data and for dating the last glaciation maximum in the southern Black Forest, as this phase is apparently not preserved in the study area.and in the Faulbach Bannwald.Régis Braucher is thanked for providing unpublished data for the review of CRE ages from moraines in Central Europe.This paper benefited from insightful discussions with Stefan Hergarten on ELA reconstructions.The French AMS national facility ASTER (CEREGE, Aix-en-Provence) is supported by the INSU/CNRS, the ANR through the 'Projets thématiques d'excellence' programme for the 'Equipements d'excellence' ASTER-CEREGE action and IRD.This paper is dedicated to the late Didier Bourlès, who greatly supported this study.Unfortunately, he was not able to discuss the results with us.The reviewers are thanked for their thoughtful comments that greatly helped to improve the quality of the manuscript.The authors declare that they have no conflicts of interest.Open access funding enabled and organized by Projekt DEAL.

Figure 1 .
Figure 1.Map of the study area showing mires, the locations where photos were taken for Fig. 2 and localities mentioned in the text.The last glaciation maximum extent in the southern Black Forest is highlighted in light blue (Hemmerle et al., 2016).The shaded relief in the background was derived from a digital terrain model obtained during the SRTM (NASA Jet Propulsion Laboratory, 2013).© EuroGeographics for the administrative boundaries.[Color figure can be viewed at wileyonlinelibrary.com]

Figure 2 .
Figure 2. Examples of glacial landforms in the study area.See Fig. 1 for the locations from where the photos were taken.Moraines are highlighted with dotted lines.The numbering refers to the ice-marginal position and starts with the innermost moraine(s) in the cirques.(A) Sankt Wilhelmer Tal and Feldberg seen from the area above the entrance to Sankt Wilhelmer Tal.The valley is one of the most impressive trough valleys in the entire Black Forest.(B) The moraine record in the Katzensteig cirque seen from the cirque's headwall.The upper cirque floor in the foreground is delimited to the north by the terminal moraine of the ice-marginal position KS-1.The smooth moraine of the ice-marginal position KS-5 is situated further east.The tongue basin on the lowermost cirque floor is surrounded by double-crested terminal moraines of the ice-marginal positions KS-2 and -3.The terminal moraine of the ice-marginal position KS-4 is located further uphill.(C) The Wittenbach and Napf stairway cirques seen from the saddle between the Hochfahrn and the Toter Mann.The Wittenbach hanging valley is separated by distinct steps from the main valley.Overprinted terminal moraines (dotted lines) occupy the swampy lowermost floor.The Napf stairway cirque directly leads into the main valley.[Color figure can be viewed at wileyonlinelibrary.com]

Figure 3 .
Figure 3. Map of glacial cirques and moraines in (A) the main valley and its tributary valleys, the lower reaches of (B) the Wittenbach hanging valley, (C) the area around the village of Sankt Wilhelm and in (D) the Katzensteig cirque.Inferred ice-margins are marked with dotted lines.See Fig. 1 for the data source of the elevation data in the background.Updated from Hofmann et al. (2020).[Color figure can be viewed at wileyonlinelibrary.com]

Figure 4 .
Figure 4. Glacial cirques, moraines and sampled moraine boulders in (A) the whole study area and (B) the Katzensteig cirque.The 10 Be CRE ages of the moraine boulders in ka (thousand years before 2010 CE) and the external uncertainties are presented in white boxes.Landform ages are given in grey boxes.Outlying CRE ages are shown in grey.[Color figure can be viewed at wileyonlinelibrary.com]

Figure 5 .
Figure 5. Reconstructed glacier extents and AABR-ELAs (m a.s.l.) during moraine formation.Two scenarios were tested for the moraine of the icemarginal position SW-13: according to the first scenario (sc.1), the moraine formed at the front of a combined glacier from the Wittenbach and Napf cirques.In the second scenario (sc.2) the glacier from the Katzensteig cirque still joined the main valley glacier.See Fig. 1 for the data source of the shaded relief in the background.[Color figure can be viewed at wileyonlinelibrary.com]

Figure 6 .
Figure 6.Moraines, glaciers and equilibrium line altitudes (ELAs) in the study area at around 14 ka. 10 Be CRE ages of moraines are presented in ka (thousand years before 2010 CE).[Color figure can be viewed at wileyonlinelibrary.com]

Table 1 .
Characteristics of the samples from moraine boulders in Sankt Wilhelmer Tal.The numbering of the samples from the moraine boulders refers to the morphostratigraphical position of the respective terminal moraines sampled soil-covered boulders may have experienced even stronger postglacial denudation.

Table 2 .
Results of the 10 Be measurements and cosmic-ray exposure (CRE) ages of sampled boulders in the Katzensteig cirque (KS) and in Sankt Wilhelmer Tal sensu strictu (SW).The 10 Be concentrations in the samples were corrected with the 10 Be concentration in a batch-specific chemical blank *CRE age uncertainties comprise both 1σ analytical uncertainties (internal uncertainty) and the error associated with the 10 Be production rate.Internal uncertainties are presented in parentheses.

Table 3 .
10Be CRE ages of terminal moraines in Sankt Wilhelmer Tal and in the Katzensteig cirque As the CRE age of the KS-1a boulder suggests that the KS-2 moraine has a minimum age of 14.3 ± 0.6 ka, the landform age corresponds to the average CRE ages of the KS-2d and KS-2e boulders.All other CRE ages from the KS-2 moraine were considered outliers.†Since the landform ages of the SW-15 and SW-17 moraines indicate that the SW-16 moraine formed between 16.3 ± 0.7 and 17.0 ± 0.7 ka, the only CRE age from the SW-17 moraine (3.5 ± 0.3 ka) is considered largely underestimated. * . A later glacial © 2022 The Authors Journal of Quaternary Science Published by John Wiley & Sons Ltd.