The extent of the Weichselian ice sheet and the environment in the southern Baltic region and bordering areas in northern Germany has received large attention (e.g. Houmark-Nielson, 2010; Kenzler et al., 2017, 2018; Rinterknecht et al., 2014). In contrast to detailed knowledge on the late Weichselian ice dynamics and its climatic implications (e.g. Hughes et al., 2016; Lüthgens et al., 2020), the early Weichselian in this area has not yet been investigated in its entirety. This period, however, is especially interesting due to the changes between cold and warm phases. While these climate variations are well preserved in ice cores (e.g. Seierstad et al., 2014), there is a lack of local terrestrial archives documenting the highly variable conditions at that time.

Among the few terrestrial sites in northern Germany where palaeosols as indicators for warm phases are preserved is the push moraine ridge in Osterbylund, northern Schleswig-Holstein, close to the Danish border (Fig. 1). However, whether the palaeosols represent the Weichselian remains controversially discussed. It has been argued that the soils got deformed by the youngest Saalian (“Warthian”) glacier advance during marine isotope stage (MIS) 6 and had therefore been interpreted as “intra-Saalian”, with an attribution of the lowermost and most intensively developed soil to the intra-Saalian “Treene interglacial” (e.g. Stremme, 1981, 1986; Stremme et al., 1982; Stremme and Weinhold, 1982; Zöller, 1986). In contrast, it has been claimed that the soils are of post-Saalian age, with the lowermost soil representing the Eemian interglacial, not being glacially deformed; the overlying soils represent Weichselian interstadials (Gripp et al., 1965).

To address the debate on the age of the push moraine ridge and with it on the palaeosol formation, optically stimulated luminescence (OSL) dating is the method of choice, as it can be applied to the clastic sediments underlying and covering the soils. In luminescence dating, the time elapsed since the last exposure to daylight is determined; this is in general referred to as the burial age. From that it is evident that not the soil formation itself is being dated, but that age constraints can be provided.

In our study we present detailed sedimentological investigations of the sediment succession in the push moraine complex near Osterbylund, Schleswig-Holstein, Germany, together with OSL ages from the sediment over- and underlying the palaeosols. In addition, pollen analyses from a peat underlying the clastic sediments are presented together with an attempt to date the peat using 230Th/U dating. The combination of the data is then used to correlate the investigated strata and the palaeosols to MISs, with the aim to add another jigsaw piece to the glacial and intraglacial development in the western Baltic region.

For decades, several sand–gravel pits have been opened in a 9 km long, SE–NW stretching push moraine ridge between Wallsbüll and Böxlund, northern Schleswig-Holstein, close to the Danish border (Fig. 1). Two of them (Böxlund 1 and 2; Fig. 1b) have been intensely investigated over decades due to the occurrence of palaeosols (e.g. Stremme and Weinhold, 1980). Because of the presence of several organic-rich horizons, the sand–gravel pit in Osterbylund became of special interest. It is situated on the same ridge (Fig. 1) a few kilometres SW of the Danish border. The push moraine is surrounded by Weichselian sandur plains originating from meltwater streams of the Weichselian end moraines draining westwards into the North Sea. The surfaces of these plains have an elevation no higher than +25 m above present sea level (a.s.l.). The Saalian push moraine complex is 20 to 30 m higher than these plains (topographical map of Germany 1:25 000; sheets Medelby, 1121, and Wallsbüll, 1221), with the highest point, called Lundtop, reaching +53.8 m a.s.l.

The continuous exploitation of the pit revealed not only the complexity of the sand succession and the intercalated soils and organic-rich horizons but also various peat layers. Further, two folded palaeosols in superposition were found (see Stephan, 1988). Based on detailed sedimentary and structural studies, Stephan (1988) showed that the deformation had developed by a periglacial sheet sliding downslope with thrust folding, probably initiated by an undercut of the moraine slope by Weichselian meltwater streams.

In Fig. 2, a generalised view of the sediments, structure, and deformation is given. The lower and median parts are mainly meltwater sands, in most cases strongly deformed by glacial pressure (folded and cut by thrust faults). Blocks of till and folded till beds are incorporated in the sands mostly getting smaller upwards, rarely reaching the surface. The northeastern part of the deformed sequence is covered by a thick undulating till sheet overlain by periglacial sands, which are several metres thick, not deformed, and concordantly filling push synclines, upwards with slightly undulating or increasingly even stratification. Several palaeosols (named OBL 1–4) have been observed within these sandy deposits; the lower two soils show distinct humus-rich Ah horizons (according to AG Boden, 2005). Towards the centres of the synclines the lowermost humus layer is often graded into a peat.

Different outcrop walls have been investigated over the last 2 decades, showing various sedimentological features which can be used to obtain an integrated picture. Detailed investigations and sampling for dating and pollen analyses were carried out in 2003, 2004, and 2005. The field studies in 2012 (see Stephan et al., 2017) and 2016 focussed on the sedimentological description to obtain a better understanding of the geological and geomorphological processes at the site.

The chronological approach is threefold: next to absolute dating of the clastic sediments using OSL, it was attempted to date the peat using 230Th/U. Additionally, pollen analyses were conducted for a qualitative allocation of the peat to possible warm phases.

The correlation between the individual profiles is based on the humic layers representing palaeosols (OBL 1–4) and the OSL ages. The sedimentological investigations in 2016 lead to the conclusion that the lower diamicton is a subglacial traction till (Evans et al., 2006) deposited during a Saalian advance of the Scandinavian Ice Sheet. The sand lenses present (Figs. 5 and 6) could either be formed by supraglacial, englacial, or subglacial drainage systems and had been subsequently incorporated into the till. The faintly stratified upper part of unit I (Fig. 6) could either be a supraglacial melt-out till (Lukas and Rother, 2016) or the result of periglacial reworking processes after or during an ice decaying phase.

OBL 1 is to be correlated with the up to 2 m thick peat found during the excavations in profiles 2003 and 2005-2, filling a depression on the surface of the till (see Figs. 2 and 3). Due to the open-system behaviour of the peat, no 230Th/U ages can be presented. The OSL age of 86 ± 8 ka (LUM 1138; Fig. 3) from well-bleached sediments well above OBL 1 shows that the peat is older than the Odderade interstadial (Fig. 10). The corresponding sample LUM 470 is poorly bleached and cannot add any information. Further age constraint is exclusively provided by the pollen analysis, as the OSL sample (LUM 1137) underlying the peat is both poorly bleached and saturated; no accurate age can be presented for this sample.

The onset of the Corylus, Tilia, and Taxus curves and the maximum distribution of Corylus characteristic for Eemian pollen zone E IVa (Menke and Tynni, 1984) are not recorded in the peat layer of Osterbylund. Peat growth at the investigated location started only during the second half of pollen zone IV, the Tilia phase of Menke and Tynni (1984) which is characterised by the dominance of Tilia and of Corylus, next to Ulmus and Quercus. Taxus only reaches values of about 10 % during E IVb at Osterbylund, further indicating that peat growth started only after the pronounced Taxus phase (Menke and Tynni, 1986) during the younger part of zone IVb. This zone corresponds to pollen zone IVb of Behre (1962) and Müller (1974), respectively. According to varve counts of Eemian deposits at Munster-Breloh and estimates of the duration of pollen zones of Müller (1974), zone IVb should have had a duration of 1100 years.

The following zone, which seems to be fully preserved, corresponds to the Carpinus zone E V (Menke and Tynni, 1984) and can be correlated with pollen zones Va and Vb of Müller (1974). The occurrence of Viscum, Hedera helix, and Buxus in this and the following zones has been recorded from other western European Eemian sites (Litt, 1994; Zagwijn, 1996; Kühl and Litt, 2007) including adjacent areas from northern Germany (Behre, 1962; Behre and Lade, 1986; Müller, 1974; Menke and Tynni, 1984; Ziemus, 1989; Behre et al., 2005) and characterises the thermal optimum of the interglacial. Müller (1974) has estimated the duration of zone E V, Va and Vb, to about 4000 years.

The youngest recorded pollen zone in Osterbylund, correlated with zone E VI (Menke and Tynni, 1984) and zone VI of Müller (1974), shows a decrease of deciduous thermophilous trees and the expansion of Picea, Pinus, Abies, and Betula, marking the beginning of the termination of the interglacial. The overrepresentation of Picea in the profile of Osterbylund compared to other profiles from northern Germany might be caused by local peat bog conditions favouring spruce-rich forest swamps (Menke and Tynni, 1984). This interpretation and the large occurrence of Calluna in the upper zone from about 0.75 m upwards are in agreement with a change in peat composition. Compared to central or southern German locations (Grüger, 1979; Urban et al., 1991; Litt, 1994; Litt et al., 1996), the relative low representation of Abies during this late interglacial phase at Osterbylund and at other northern German sites, e.g. Lichtenberg (Veil et al., 1994; Hein et al., 2021), is caused by the northern limit of the distribution area of Abies which lay in Jutland during that time (Zagwijn, 1989). The transition to pollen zone VII and the entire pollen zone VII, representing the end of the interglacial, are not recorded as the uppermost part of the peat has been eroded. Müller (1974) estimated the duration of Eemian pollen Zone VI (E VI after Menke and Tynni, 1984) to about 1000 years, whereas Kühl and Litt (2007) published a revised extrapolation of 2000 years.

According to these time estimates, the duration of the Eemian peat layer formation at Osterbylund, which comprises pollen zones E IVb, E V, and most parts of E VI (Menke and Tynni, 1984; Müller, 1974), can roughly be estimated to about 6000 years. Turner (2002) pointed out that the entire duration of the terrestrial Eemian north of the Alps or Pyrenees was less than 13 000 years, which matches the results of Urban (2007) who proposed a duration of 10 ka for pollen zones E1–E6 of the Eemian sensu stricto (Zagwijn, 1961) and 10 000 years for the entire interglacial. The terrestrial Eemian has been correlated with MIS5e of the deep-sea chronology (Shackleton, 1969, Mangerud et al., 1979; Fig. 10) and spans the time between about 126 to 115 ka (summarised in Urban, 2007). The age of the peat layer of Osterbylund therefore can be estimated to be about 120 ka.

The reworking of the OBL 1 palaeosol will thus have occurred after MIS5e. Due to the soil formation processes and the resulting post-depositional changes, the interpretation of the depositional environment cannot be reconstructed unambiguously. Slope movements under periglacial conditions are conceivable (Stephan et al., 2017), which filled former depressions of the landscape topography (Guerra et al., 2017). In addition, denudation processes may have partly contributed to the deposition, at least of the lower and middle part of unit II (Fig. 6). In this context, the occurrence of channel-like structures filled with granules is of importance. The better sorting of the upper sandy part of unit II might indicate that aeolian processes have also contributed to the deposition. In unit II the Ah horizon at the top (= OBL 2) illustrates a halt in sedimentation and consequently a hiatus; at least some erosion of the former Ah horizon is likely. The age of 86 ± 8 ka (LUM 1138, Fig. 3) from just below OBL 2 puts the formation of this palaeosol either into the Brörup or the Odderade interstadial (Fig. 10). Considering that there is no indication of a hiatus, it is more likely that OBL 2 is to be correlated with the Brörup interstadial.

The formation of unit III (Fig. 6), which brackets OBL 2 and OBL 3 (Figs. 3 and 4), is probably related to slope movements (Stephan et al., 2017, and their Fig. 7) under periglacial conditions (Millar, 2013). Especially soft sediment deformation structures indicate gravity-induced failures along the basin slope (Blair and McPherson, 2009). This assumption is supported by the outsized clasts within the upper part of this sand package, which have been transported downslope. Furthermore, the sand-filled cracks and fissures suggest freeze–thaw action and the existence of permafrost and thus subarctic conditions. The seasonal thawing of the active layer induced frost creeping of larger stones and boulders down the slope. Some of these boulders show fragmentation due to volume expansion of freezing water in fissures. This clearly implies cold climate conditions during the deposition of the sandy part of this unit. OBL 3 developed most likely under an interstadial climate after 69 ± 6 ka (LUM 1139), which is the age of the underlying sand. The equivalent sample LUM 469 was dated to 68 ± 6 ka, supporting the age constrain. According to these dating results, one could assign the soil formation to either the Odderade or Keller interstadial (Fig. 10).

Above the OBL 3 palaeosol there are aeolian sands, for which good OSL signal resetting had to be expected. While the age of 60 ± 6 ka (LUM 1140) just below the humic horizon of OBL 4 (Fig. 3) seems reliable according to the OSL testing and statistical parameters, the sand covering the palaeosol was poorly bleached, which results in a possibly overestimated age of 43 ± 4 ka (LUM 1141). If that age was the true depositional age, it would fall into MIS3, thus implying a significant hiatus, for which no field evidence was present. From the sedimentological observations it is not possible to reconstruct a distinct depositional environment to these sands (Fig. 6). The burrow-like structures across this unit display the occurrence of digging animals such as rodents. If true, this could indicate a tundra-steppe ecosystem in a periglacial landscape (Dupal et al., 2013). This assumption is supported by the existence of ice wedge casts. If such an ice wedge was accidentally sampled, even younger material than representative for the sandy layer could have been sampled. It is therefore not possible to finally conclude on the age of the sand overlying OBL 4.

However, there are no phases warm enough for soil formation during this period, and despite this unknown, assignment of this humic horizon to the Keller interstadial (Fig. 10) seems feasible, thereby assigning OBL 3 to the Odderade interstadial. The sandy deposits of the upper section of unit VI would therefore belong to the Ellund cold phase or even younger (Fig. 10).

Because of the unequivocal palynological assignment of the peat to the Eemian, an upper limit to the age of the palaeosols (OBL 2–4) above the peat is clear. Thus, the entire push moraine complex (Fig. 1) has to be of Saalian age. This observation makes a Weichselian glacial margin even further west unlikely, i.e. south of the Danish border, as discussed by Houmark-Nielsen (2007). All palaeosols formed during the early Weichselian (MIS 5a–d), which is in agreement with the observations of Gripp et al. (1965). While the pollen data provide a clear view on the age of OBL 1, the luminescence data cannot be interpreted unambiguously; they have to be seen in their entirety and not on their own.

The aim of this study was not only to unravel the age of the push moraine complex Wallsbüll-Böxlund, Schleswig-Holstein, but also to assign the peat and palaeosols present in depressions in Osterbylund to interglacials and interstadials in order to add another jigsaw piece to the understanding of the glacial history in the western Baltic region.

The attempt to date the peat layers using 230Th/U dating failed due to the open-system behaviour of both the brown and the black peat. However, the pollen analyses allowed for clear assignment of the peat and thus of the palaeosol OBL 1 to the Eemian. This coincides with the saturated and correspondingly old age of the sands underlying the peat; the sands above the peat were dated to the early Weichselian. While for palaeosol OBL 2 the assignment to the Brörup interstadial is clear, the correlation of the palaeosols OBL 3 and OBL 4 to any interstadial is hampered due to the limited luminescence data available. Poor luminescence signal resetting especially of the sands above OBL 4 makes an assessment of an upper limit for the palaeosol formation difficult. However, from the field, as well as luminescence data, it is most likely that OBL 3 formed during the Odderade interstadial and OBL 4 during the Keller interstadial. This observation contrasts previous findings claiming that only two post-Eemian thermomers, i.e. Brörup and Odderade interstadials, are to be expected due to the insolation of the Northern Hemisphere. The weakly developed OBL 4 seems to indicate though that warm oceanic currents in the North Atlantic and correspondingly in northern Germany resulted in another – terrestrial – third weak thermomer and is thus not unique to marine settings and ice cores.

With Eemian up to early Weichselian ages of the peat and palaeosols, filling the depression it is evident that the push moraine complex is of Saalian age; a Weichselian ice margin further in the west as assumed in other studies is therefore unlikely. It is certainly worth investigating the position and ages of the Weichselian ice margins in the future as this would complement the studies from Denmark and NE Germany.