Aeolian sedimentation in the Rhine and Main area from the Late Glacial until the Mid-Holocene New evidence from the Magdalenien site of Götzenhain (Hesse, Germany)

Aeolian sediments (sandy loess, aeolian sand) were studied by pedological and geochronological (OSL) methods to reconstruct their stratigraphy and age, and to relate these results to archaeological evidence. The results prove loess accumulation on an older ventifact horizon during the Late Glacial period followed by layers of aeolian sand also dating to the Late Glacial period. Holocene aeolian sedimentation (6.9 ka) was recorded within the excavation site, which probably indicates Neolithic human impact that locally disturbed parts of the Magdalenien artefact scatter. The results are discussed by means of similar dated aeolian and archaeological stratigraphies in the Rhine-Main area. [Äolische Sedimentation im Rhein-Main-Gebiet vom Spätglazial bis in das Mittlere Holozän: Neue Nachweise von der Magdalenien Fundstelle in Götzenhain (Hessen, Deutschland)] Kurzfassung: Äolische Sedimente (sandiger Löss, Flugsand) wurden durch pedologische und geochronologische Methoden (OSL) mit dem Ziel untersucht, hieraus Aussagen zur stratigraphischen Abfolge und zum Ablagerungsalter zu gewinnen und diese Ergebnisse auf die Ergebnisse der archäologischen Ausgrabung zu beziehen. Die Ergebnisse zeigen, dass die Ablagerung des Lösses im Spätglazial auf einer älteren Lage von Windkantern erfolgte und dieser am Ende des Pleistozäns von äolischem Sand überdeckt wurde. Untersuchungen der fundführenden Schichten ergaben eine mittelholozäne Datierung (6.9 ka), die mit einer lokalen Störung durch äolische Sedimentumlagerungen in Folge anthropogener Landschaftsveränderungen während des Neolithikums erklärt werden. Diese Untersuchungsergebnisse werden vor dem Hintergrund vergleichbarer äolischer Stratigraphien und Fundplätz im Rhein-Main-Gebiet diskutiert.


Introduction
Few sites provide information about human settlement in the Rhine-Main area after the LGM by the cultural complex of the Magdelenian (Street, Baales & Weninger 1994;Bosinski 2008). New contributions on this topic are presented by the Götzenhain site where artefacts dating to the Magdalenian have been excavated from sandy aeolian sediments (Serangeli & Terberger 2006;Terberger, Serangeli & Woertz 2008). To support the assumed dating and to reconstruct the processes of site formation closely connected to questions concerning the coherence and plausibility of the archaeological results, pedological and geochronological (OSL) investigations were performed. Furthermore we compared our local results with other luminescence dated aeolian stratigraphies in the Rhine-Main area.

Study site and archaeological record
The Götzenhain site (8°44'37''E / 50°00'07''N) is located in northern zone of the exposed foothills of the Odenwald mountains at the transition of the Main River basin (Fig. 1).
It is situated in an agricultural area on the exposed southern slope above a small stream between the villages of Götzenhain and Dietzenbach. The bedrock mainly consists of Rotliegendes of Permian age and covered by layers of loess and aeolian sands.
Based on several surface finds, since 1991 the site in general has been part of the discussion about the sparsely known Magdalenian in this area (using former classification Dreieich-Götzenhain I: Jöris, Schmitz & Thissen 1993: Fig.1;Rosenstein 1998;Bosinski 2008: 365f). Using the opportunity to investigate one of the rare sites of this period, excavations started in 2006 (Serangeli & Terberger 2006) and continued until 2009 by a joint project between the Universities of Greifswald and Tübingen. During these field campaigns a layer of flint artefacts embedded in sandy sediment was recorded. The small assemblage (c. 650 artefacts without chips) reflects a short term occupation by a small group. The artefact distribution has a diameter of about 5 m. Some charcoal remains indicate a fire place in the centre of the find concentration. The stone artefact inventory demonstrates core preparation and blank production at the site. The limited number of about 70 tools underscores the short term character of the site. This is corroborated by the dominance of backed bladelets and burins. They probably indicate activities focusing on the reparation and hafting of hunting weapons. Unfortunately no bones are preserved but analogous to other sites (Stevens et al. 2009a(Stevens et al. , 2009b we assume that horse was the prey of choice. Typological elements such as Lacan burins (Fig. 2) indicate a close relationship of the Götzenhain assemblage to the Magdalenian sites of Andernach and Gönnersdorf (Fig. 5) located about 100 km to the northwest. However, the "Jurahornstein" used as raw material (Fig. 2) indicates an origin of the people in southern Germany who subsequently moved from the south along the river Rhine (Terberger, Serangeli & Woertz 2008).

Methods
The pedological and geochronological investigations are based on the two sampling trenches GOE1 and GOE2 (Fig. 6). While GOE1 was recorded on the western edge of the excavation area and is closely linked to the artefact scattering, GOE2 was recorded 150 m downhill the southern slope. One OSL sample was taken from profile GOE1 and three OSL samples were collected in stratigraphical order from profile GOE2. For pedological analysis every horizon recognisable was sampled for laboratory analysis (samples 1A-C; 2A-G). Gravels recorded in GOE2 were sampled to document their petrographic and surface features (sample GS1).

Pedological analysis
Pedological horizons were described and sampled using the German soil science standard ("KA5"; AG Boden 2005; Tab. 1). The soil colour was documented according to the Munsell Soil Colour Charts under moist conditions. After air drying, careful hand-crushing, humus and carbonate destruction (H 2 O 2 30%, HCl 10%) and dispersing with sodium phosphate, a combined pipette and sieving test was used to determine the grain-size distribution. Total carbon and nitrogen were measured by dry combustion (Elementar vario EL) at 1100°C in duplicates.

Optical dating
Luminescence dating was performed at the Marburg Luminescence Laboratory. All samples were sieved in dry conditions to the fraction of 38-63 µm. Subsequently, samples were treated with HCl (10 %), H 2 O 2 (10 %) and Na 2 C 2 O 4 to remove carbonates, organic matter and clay. Polymineral samples were etched for two weeks in H 2 SiF 6 to separate fine grained quartz. Measurements were carried out on an automated Risø TL/OSL DA 15 reader equipped with a 90Sr/90Y beta source delivering 0.1085 Gy/s to the sample. Equivalent dose (De) was estimated using a single aliquot regenerative (SAR) dose protocol (Murray & Wintle 2000) with 50 s of blue stimulation at 125 °C and preheat temperature of 200 °C. The quartz OSL signal was detected through a Hoya U 340 filter (7.5 mm; transmission spectrum 290-370 nm). A reference sample was used to determine the dose rate (D0) applying Neutron Activation Analysis (NAA). While the H 2 O content measured in the samples seemed far too low, an estimated average moisture of 7 % was used for calculation.

Pedological results
Profiles GOE1 and GOE2 ( Fig. 3; Tab. 1) start with weathered bedrock ("Rotliegendes") and the above-lying sections of both profiles are carbonate-free. The pedological sequence of GOE1, which represents the stratigraphy observed during the excavation, shows homogenous, nearly gravel-free sand of aeolian origin ("Lösssand" according to AG Boden; "Sandlöss" according to Koch & Neumeister 2005) covering the bedrock. The upper part of this sequence is clearly mixed by ploughing. The soil can be classified as (partly eroded) Cambisol showing an Ap/Bv/Cm sequence of soil horizons. During excavation several gravels with ventifact appearance have been observed at the bottom section of the aeolian sand (Serangeli & Terberger 2006, 52) and might be comparable to the blow-out layer with ventifacts in GOE2 by their stratigraphic position. The artefact scattering was documented at the lower part of the aeolian sand though some surface finds indicate that the scattering was already affected by ploughing.
In GOE2 the transition from bedrock to the overlying loess-like sediment ("Sandlöss" according to AG Boden and Koch & Neumeister 2005) is marked by a dense layer of sharp facetted ventifacts (sample GS1) up to 9 cm in diameter, which were determined to consist mainly of Quartz gravels with a few small Lydit gravels (Fig. 4). This blow-out zone is overlain by a loess-like sediment of 20 cm thickness. Above, sand of aeolian origin ("Lösssand" according to AG Boden; "Treibsand" according to Koch & Neumeister 2005) occurs which is comparable to the find layer in profile GOE1. The uppermost 40 cm of profile GOE2 is the ploughing zone characterized by a relatively high organic content. While the zone up to 70 cm below the surface shows the typical sequence of a Cambisol with an (M-) Ap/Bv/lCv sequence, the following section shows properties of stagnant soil water (Sw-Bbt horizon).

Geochronological results
OSL dating (Tab. 2) of the aeolian sand in profile GOE1 yielded an age of 6.906 ± 0.71 ka (MR0762) while the sandy loess at the base of GOE2 yielded an age of 13.915 ± 1.41 ka (MR0765). The overlying aeolian sand in GOE2 was dated to 17.856 ± 1.42 ka (MR0764) while the uppermost aeolian sand in GOE2 yielded an age of 13.262 ± 1.18 ka (MR0763).

Discussion
Luminescence dating is generally accepted as providing reliable results about sedimentation ages especially when performed on quartzes of aeolian deposits (Murray & Olley 2002;Preusser et al. 2008). Nonetheless the ages from profile GOE2 are inconsistent when compared to their stratigraphical position and need explanation. Furthermore the dating of the aeolian sand in GOE1 to the Mid-Holocene seems contradictory to the embedded Late Palaeolithic artefacts.
Regarding the data used to calculate the ages in profile GOE2 (Tab. 2) we argue that the results of the D E -determination are correct giving a smaller equivalent dose (D E ) for sample GOE2b compared to the older sample GOE2c. The reason for the older age of sample GOE2b has to be seen in a significantly lower dose rate (D 0 ) used for age calculation. We argue that due to the bedding of sandy and loamy layers the sample taken from that horizon might not have been representative for a correct D 0 -determination. Therefore we reject the age of 17.856 ± 1.42 ka (MR0764) as an age overestimation.
The remaining datings from GOE2 indicate that the basal sandy loess deposition occurred within a time range of approx. 15.3 to 12.5 ka when including the spans of uncertainty. The underlying layer of ventifacts (GS 1; Fig. 5) and the absence of older loess sediments points to a phase of intensive wind erosion preceding the dated accumulation. A chronological connection to the Beuningen gravel bed as a marker horizon for intensive surface deflation in Western Europe dated to 20 ka to15 ka (Frechen & van den Berg 2002) or 18 ka to 14 ka (Kasse et al. 2007) might be possible and would fit very well into the suggested time model. Comparing this dated accumulation of loess to other luminescence dated stratigraphies along the Rhine (Fig. 5D), continuing aeolian sedimentation during this period is present in all profiles, except Mainz-Weisenau and Nussloch, with considerable rates of mass accumulation until 13 ka as calculated by Frechen, Oches & Kohlfeld (2003). A finer resolution into subphases must be rejected due to the high span of dating uncertainty.
An accumulation of younger aeolian sands after 14.5 ka is in good accordance to supraregional models showing sandy aeolian activity during the colder phases of the Late Glacial Period (Dryas I, II, III) especially in the area of the European sand-belt more to the north (Schirmer 1999). The phases of aeolian activity were interrupted by periods of stability and pedogenesis during warmer periods accompanied by more dense vegetation (e.g. Kasse 2002;Koster 2005;Kolstrup 2007;Hilgers 2007;Kaiser et al. 2009). While the nearby luminescence stratigraphy of Mainz-Gonsenheim (Radtke & Janotta 1998; Fig. 5D) revealed several phases of sandy aeolian accumulation separated by palaeosols and the Laacher See-tephra, a precise attribution of the aeolian sands in Götzenhain to one of the phases / chronozones fails due to the absence of any comparable chronomarker and the span of uncertainty.
Mid-Holocene accumulation of aeolian sand as derived from profile GOE1 (MR0762: 6.906 ± 0.71 ka) can hardly be explained by natural causes like a climatic deterioration during this period but is more likely to result from human impact. Coinciding with the Early Neolithic cultures of Late Linear Pottery (LBK) or Rössen, we suppose that local hu-man impact, triggering forest clearance and land use which is evident for this period (Kalis, Merkt & Wunderlich 2003), has caused sediment reactivation. This assumption might be corroborated by the occurrence of charcoal of oak (Quercus) in the excavation area (analysis: A. Kreuz, Wiesbaden).
A correlation of sedimentation ages to the archaeological results at the site at Götzenhain is supported by the improved chronology of the Late Magdalenien from the sites of Gönnersdorf and Andernach ( Fig. 5B; Street, Baales & Weninger 1994;Stevens et al. 2009a). All sites are comparable in typological aspects. We argue that the settlement occurred during the beginning of the accumulation of the sandy loess dated by sample GOE2c. Ongoing accumulation of aeolian sediments might have supported the preservation of the artefact scatter in the excavation area (GOE1). The archaeological analysis of the artefact distribution supports the idea that the Neolithic disturbance was limited to a small area and did not affect the site in total. A summarizing model concerning the stratigraphy and chronology at Götzenhain based on the aspects discussed here is presented in Fig. 6 Stevens et al. 2009a) indicating the time of Magdalenien settlement; C: OSL-datings from Götzenhain site compared to NGRIP δ 18 O time series as temperature indicator (NGRIP-Members 2004;data source: http://www.iceandclimate.nbi.ku.dk); D: Luminescence datings of loess (Frechen, Oches & Kohlfeld 2003) and aeolian sand (Radkte & Janotta 1998) plotted by the relative probability of aeolian sand accumulation at every sequence based on the 1σ span of luminescence dating uncertainty.