Introduction

Shell middens are ubiquitous across the world as places where hunter-gatherer-fisher communities extract resources and dispose of their waste (Gutiérrez-Zugasti et al. 2011, see Fig. 1). As such they present a wealth of data concerning the subsistence economy, health, social organisation, and settlement systems of these hunter-gatherer communities. Along the Baltic coast, shell middens are mainly found in Denmark, and primarily date to the Mesolithic and Early Neolithic.

Fig. 1
figure 1

Map of Mesolithic and Neolithic shell middens along the Atlantic Coast, including the Middle/Late Neolithic shell middens from Germany and the Netherlands (updated from Gutiérrez-Zugasti et al. 2011)

Full size image

Later Neolithic and Bronze Age shell middens do exist, also along the Baltic coast, but the question remains how these later sites should be interpreted (cf. Andersen 2000). Along the North Sea coast, for instance, a concentration of settlement sites consisting of shell middens dates between 3200–2400 BC, and show a mixed economy of farming and fishing, consisting of both small-scale extraction camps and permanent settlements (e.g. Kleijne et al. 2013a). A late 3rd millennium BC shell midden along the Baltic Sea, is the Late Neolithic site of Tegelbarg, in Northern Germany. The shell midden of Tegelbarg is situated in a coastal landscape along the Baltic Sea and was occupied during the second half of the 3rd millennium BC (Höfgen 2018, see Fig. 1), and is, among other things, associated with the production of lanceolate flint daggers (Arnold 1979, 1981a). In order to understand the settlement system at Tegelbarg, it is important to understand the stratigraphy and the specific depositional practices, and postdepositional alterations, or possible hiatuses in occupation, involved in the creation of the midden. Secondly, particular kinds of deposits can inform us on the organisational nature of practices, related to burning and waste disposal.

In this paper, we will focus on site formation processes and internal chronology, using respectively archaeological soil micromorphology and critical stratigraphic analysis, and subsequent Bayesian modelling, of radiocarbon dates.

The site

Archaeological research at the site of Tegelbarg was conducted during the 1970s by the University of Cologne (Arnold 1979, 1981a) and by the University of Kiel and State Museum (Arnold et al. 2011; Höfgen 2018, see Fig. 2). While most shell middens in Northern Europe contain large and extensive accumulations of shells and other materials, up to several thousands of m2 (cf. Andersen 2000; Nobles 2016), the shell midden of Tegelbarg is only c. 60m2 in extent and consists of a finds layer of a maximum of c. 40 cm thick (Höfgen 2018).

Fig. 2
figure 2

The location of the shell midden of Tegelbarg in Schleswig-Holstein (SRTM base-map, with topographic map (OpenStreetMap) of the regional topography and the site of Tegelbarg indicated, and a regional soil map/archaeology map taken from Meurers-Balke et al 1985, fig. 3 and fig. 10)

Full size image

Both Arnold (1981a) and Hartz and Müller (2018) state that this was most likely a short-term extraction camp, primarily aimed at producing flint daggers and gathering shellfish. Their main arguments are based on the archaeobotanical data suggesting that only wild plants were consumed, and the zooarchaeological data finding almost no evidence for fishing (even though the site was intensively wet-sieved). Additionally, they both point to the limited size of the shell midden, in comparison to the Danish shell middens, as an argument for characterising the site as a short-term extraction camp. Hartz and Müller (2018) furthermore sketch that this camp functioned during a time of major agricultural transformations which can be related to the effects of the 4.2 ka BP event. Putting the site in a bigger framework of coastal occupation and resource exploitation, Kleijne et al. (2020) argue that flexibility was an important element of the subsistence strategies of Late Neolithic coastal communities across North-West Europe during the period of climatic instability known as the 4.2 ka event. Short-term extraction camps, therefore, are to be expected, and an important and hitherto not often recognised and understood aspect of Late Neolithic subsistence.

Furthermore, the absolute temporal framework of the site is unclear at the moment. Studies have alternatively dated the site to the Late Neolithic I period (2300–1950 BC; e.g. Hartz and Müller 2018), roughly between 2000 and 1800 BC (Fernandes et al. 2012), and even into the Early Bronze Age (Willroth 1992, 24; Kuhn 1979). The material evidence for Early Bronze Age activity is limited to a single arrowhead found in disturbed context (Arnold 1979, 76, Abb. 112), making clear that habitation definitively ceased in the Late Neolithic. As such, it is unclear how this site can be positioned relative to other Late Neolithic transformation processes, such as socio-economic transformations (e.g. Brozio et al. 2019; Terberger et al. 2018), cultural developments (e.g. Müller and Vandkilde 2020; Kleijne et al. 2021), and the aforementioned climatic changes (e.g. Kleijne et al. 2020). Past attempts have faced problems with understanding the relative stratigraphy of specific elements on the site (Arnold 1979; Höfgen 2018). Next to this, some hints for possible older occupation are present: a fragment of a stone battle-axe and two abraded cord-decorated sherds were found in the shell midden layer (Arnold 1981b, 155), and can be associated with the Corded Ware phenomenon (Schultrich 2018:1380).

Micromorphological analysis was carried out to test the hypotheses proposed by Arnold (1981a) and Hartz and Müller (2018): Are we dealing with short-term occupation? Or do we find activities that fit well with long-term occupation of shell middens, such as platform building, vegetation burning, the repeated use of fire, and the trampling and mixing of sediments and materials (e.g. Villagran 2018; Duarte et al 2019; Aldeias and Bicho 2016; Marchand et al 2018)? At Dutch Late Neolithic shell middens, for instance, evidence for trampling and mixing of occupational layers is found besides the repeated burning of stacked reed deposits (Colenberg 2014), indicative for continued and intense occupation. Another research question includes the possible evidence for pre-midden occupation. Do we find evidence for the disturbance of the Pleistocene soil that can be attributed to such earlier human influence? Finally, regarding the landscape development, thin sections can help understand the environmental characteristics preceding the formation and after abandonment of the shell midden.

Landscape

The landscape of Tegelbarg was primarily formed by Weichselian boulder clay outcrops in a shallow basin during the Pleistocene (see Fig. 2). This dry natural landscape changed through rising postglacial sea-levels, and the appearance of a lagoon along the Baltic coast, filling up the basin with sediments (Meurers-Balke et al. 1985; Exon 1972; Tapfer 1940, 207–209). This lagoon firstly created the opportunities for communities to live in the larger Tegelbarg landscape (also attested by the Funnel Beaker site of Bostholm on an adjacent glacial outcrop; Meurers-Balke et al. 1985). The formation of a beach barrier along the coast (possibly during the Late Bronze Age) closed off the lagoon from the Baltic Sea and lead to the growth of peat (Seer and Schrautzer 2014).

It is unclear what the impact of Late Neolithic communities on the local vegetation was. At the Middle Neolithic site of Bostholm, the local landscape was opened during the intensive habitation, as indicated by palynological indicators, showing an increase in cereal pollen and wild grasses such as Plantago lanceolata, indicating cultivated fields, and a decrease in tree-pollen (described as an Iversen-landnam by Kalis and Meurers-Balke 1998). It remains to be seen whether a similar impact is visible at the later site of Tegelbarg, where archaeological indicators show a different kind of habitation.,.

A detailed micromorphological investigation of the shell midden layers can provide us with information concerning the phasing and character of the occupation and its intensity, and the human practices involved in the formation of the shell midden. Simultaneously, it can further inform us on the landscape development, and post-depositional processes that the assemblage has undergone. To understand these aspects, we will conduct a microfacies analysis on the soil micromorphological thin sections (cf. Courty 2001). A single column of soil micromorphology samples holds only limited interpretive value, but this does provide a starting point and the creation of a model of occupation at Tegelbarg that can be further refined by future micromorphological work. The representativeness of our soil micromorphological study is guided by the variability in the site’s layers that can be analysed. By studying the transition from natural subsoil to midden deposits, the midden deposits themselves, and the transition from midden deposits to natural peat development, we can at least gain insight in these different deposits. In effect, many of the important aspects of the occupation at Tegelbarg are now studied through soil micromorphology.

Materials and methods

Soil micromorphology

The North profile of the 2011 excavation was sampled directly during the excavation, using a plastic container of 2,5 m long, 10 cm wide and 10 cm deep (see Fig. 3 and Höfgen 2018, Fig. 7). In this profile, a 40-cm thick layer (“Muschelschicht”, see Figs. 2 and 3) containing shells and other evidence for human activities is at its highest elevation on the natural boulder clay. The midden layer is covered by a peat layer (“Bruchwaldtorf”). The thickness and the peat layer indicate that a complete sequence of habitation was sampled, without later disturbances affecting the shell midden sequence. Höfgen indicates that within the layer containing shells, a concentration is visible of common mussel shells (Mytilus edulis; “Muschelkonzentration überwigend Miesmuscheln”).

Fig. 3
figure 3

The site of Tegelbarg: excavations carried out, the location of excavation profiles (marked by purple lines), the location of the grave marked in red, the sampling location in the North profile of the 2011 excavation, and the probable extent of the shell midden

Full size image

The 2.5 m long sample was stored cold (-20C) until 2019, when the author sampled it for micromorphology. The top 1,5 m of the sample, containing the peat sequence, was completely dried out and unsuitable for micromorphological analysis. An intact profile was obtained from the lower 1 m, from which 6 small blocks could be sampled, using plaster wraps for stability (see Table 1 and Figs. 4 and 5). In effect, the sampling for micromorphology captured the top of the Pleistocene boulder clay, the complete midden sequence, including one of the shell concentrations, and the lowest part of the peat layer.

Table 1 Archaeological soil micromorphology samples, the depths within the profile column, and the stratigraphic contexts from which they were taken
Full size table
Fig. 4
figure 4

North profile with the location of the profile column used for micromorphology and radiocarbon dating (adapted from Höfgen 2018, Fig. 7)

Full size image
Fig. 5
figure 5

Overview of the thin sections from the column, the discerned microfacies and the location of Figs. 5–8 (note the duplicated slide 6 on the top of the profile, left is slide TEG19_006A, right is slide TEG19_006B)

Full size image

Micromorphological analysis focused on describing groundmass, natural phenomena, and anthropogenic admixtures using standard handbooks (Stoops 2021; Stoops et al. 2010; Goldberg and Macphail 2008; Courty et al. 1989; Macphail and Goldberg 2018), and the characterisation of microfacies (following Courty 2001).

Next to this, the micromorphological analysis was used to critically inform us on the site’s chronology (cf. Toffolo et al. 2012; Goldberg and Arpin 1999). Combining the micromorphological observations concerning post- depositional alterations such as bioturbation, taphonomical considerations, and the possible reworking of sediments, we can gain an understanding of the stratigraphical integrity of the site.

Radiocarbon dating

In total, 33 radiocarbon determinations have been taken throughout the years from material excavated at the site of Tegelbarg (see Table 2).

Table 2 Overview of all the radiocarbon determinations
Full size table

Most of the determinations come from the 1970s excavations and comprise of charcoal of unknown species and own age. These were dated through conventional methods by various institutes (Kiel, Cologne, and Hannover). Despite their history, there is no reason to doubt the measurements. The absence of species information has an effect, known as the Old Wood effect or Inbuilt Age (IA, Dee and Bronk Ramsey 2014). This is taken into consideration in the model (see supplementary information 2). During the 1970s excavations other samples for radiocarbon dating were taken from unspecified fragments of wood, burnt wood, softwood twigs, and bark. These samples have a limited inbuilt age.

Two radiocarbon dates were taken from Cardium edule shells. A single radiocarbon determination was done in the 1980s in the Cologne lab on bulk peat material for an unpublished palynological study (Meurers-Balke pers. comm.). Human bone from the skeleton found in the grave during the 1970s excavations, was directly dated in 2010 in Aarhus [SID-16069], but the determination failed due to a low collagen yield.Footnote 1

During the 2011 excavations, further samples of bone, antler, hazelnut pericarp, and mussels were dated in the Leibniz laboratory (Höfgen 2018). The 2011 measurements however do fall within a problematic period of radiocarbon dating in this lab (Lull et al. 2015; Meadows et al. 2015). A re-dating of these 2011 samples was conducted by the Leibniz Lab in 2021. This led to eight additional dates, taken on the same sample material. Using the R_Combine function to test for the accuracy of the earlier measurements, reveals that 5 out of 8 redated samples fall within the acceptable X2 range (Ward/Wilson 1978; see Table 3). Two of the four dates on shells, and a single date falls only just short of that range. From a statistical standpoint, it is not uncommon that out of the eight re-dated samples, one falls just outside this 5% threshold (KIA-43920/KIA-55748). The two shell dates are more problematic, as their T-values are much higher, and therefore the two 2011 shell dates (KIA-43950 and KIA-43952) are excluded. Concluding we state that most of the 2011-dates are acceptable. Therefore, six of the combined 2011–2021 dates are included in our Bayesian model.

Table 3 The 2011 radiocarbon dated samples that were redated in 2021
Full size table

In addition to the published dates for Tegelbarg, this study added several new dates. Five uncharred seeds of Alnus, taken from a sieved residueFootnote 2 from a part of the 2011 profile column used for micromorphology, were dated by the Poznan radiocarbon laboratory in 2020. And charcoal of QuercusFootnote 3 was taken from the lower levels of the grave pit excavated in 1971. This material was dated by the University of Groningen Centre for Isotope Research (CIO) in 2022.

Both the radiocarbon and the soil micromorphology dataset are used, together with the published (Höfgen 2018; Arnold 1981a) and unpublished (Höfgen 2014; Arnold 1979) excavation accounts, to create a Bayesian model for the occupation.

Results

Soil micromorphology

In Table 4, we present the micromorphological observations per thin section. Here we relate the layering, the groundmass, anthropogenic admixtures, soil structure and soil- and sedimentary phenomena. Subsequently the specific characteristics of groundmass and anthropogenic admixtures, and their distribution across the different thin sections, will be addressed. These observations will result in the grouping of phenomena into distinct microfacies.

Table 4 Micromorphological observations
Full size table

Groundmass and natural phenomena

In the thin sections of Tegelbarg, we can distinguish two separate groundmasses, G1 and G2.

G1 (see Fig. 6) consists of unsorted angular quartz and other rock fragments in sand size, with channels and planar voids, and a b-fabric of clay illuviation in a concave gefuric c/f distribution. The channel voids include dusty grey clay coatings indicative of soil formation processes. These clay coatings are indicative of a barren soil, caused by vegetation clearance, on the Pleistocene boulder clay outcrop.

Fig. 6
figure 6

Groundmass G1 with dusty clay coatings inside the channel voids

Full size image

G2 (see Fig. 7) consists of humified organic matter with vughy and channel voids, indicating that this microstructure was formed under the influence of the welding of aggregates, experienced drying and cracking, and disruption through plant root activity. Following Babel (1975) we name this a mull, spongy fabric. In this groundmass, we see an upward decrease of randomly oriented unsorted angular quartz grains and rock fragments entrapped in the matrix. This decrease of the mineral component creates a diffuse contact between the first and second groundmass, indicating a gradual transition from a sandy, organic poor/dry, environment to an organic rich/wet environment. Randomly sorted sand grains remain present throughout the profile in this groundmass. In several instances this groundmass exhibits dusty grey/brown clay coatings, indicating clay illuviation due to rain splash on barren surfaces.

Fig. 7
figure 7

Groundmass G2 with a spongy mull fabric, quartz grains, clay coatings, and both charred and uncharred organic matter (probably a more recent rootlet)

Full size image

The natural phenomena (see Fig. 8) consist primarily of root activity. This is present throughout the profile, both in G1 and G2 groundmasses, but primarily in G2. This is indicative of peat formation in G2, postdating the occupation of the site. Several fungal sclerotia are present in G2. These are indicative of the breakdown of organic matter such as peat. In the shell midden layers, several shells appear to be stained with Fe or Mn oxides. This is a localised phenomenon, and no clear explanation for its presence and concentration can be given at this point.

Fig. 8
figure 8

Natural phenomena: Fe/Mn-stained shells in G2, fungal sclerotium in G2

Full size image

Anthropogenic admixtures

Shell

The most common admixture at Tegelbarg involves the shells of marine molluscs. In G1, some shells show characteristics of dissolving in an acidic environment. In G2, most shells are either complete, with angular edges, or in a freshly broken state.

Most shells are tightly packed and show a horizontal bedding structure, indicative of dumping and no later reworking or trampling.

Analysis of shell orientation patters through digitisation of their thin section outline and mapping these outlines on top of each other (see for the method also Phillips et al 2011; Menzies and Van Der Meer 2018), indicates that this horizontal bedding is present in slides 3 and 4 (see Fig. 9). In the top of the sequence, more loose packing and random orientation patterns of shells are visible, and simultaneously the shells are more fragmented. This indicates that later reworking of deposited shells has taken place.

Fig. 9
figure 9

Orientation of the shells in the layers at Tegelbarg

Full size image

The human-induced alteration of shells is also visible. Close to, and in, the ash layer, most of the shells show partly blackened surfaces and cracking of the microstructures, indicating secondarily burning. The location of these blackened surfaces on the bottom of the shells, indicate that the shells were only finally deposited after the burning had taken place (see Fig. 10).

Fig. 10
figure 10

Burning and later reworking of a cockle shell

Full size image

Ash

Plant ash is present in the profile in two different forms: firstly, as a 1 cm thick layer, and secondly in the form of small aggregates and cemented around clusters of shells throughout the humified organic matter groundmass.

The 1 cm thick layer of ash can be described as having a horizontal, almost sedimentary, structure. It consists of a massive, homogenous, sometimes even cemented, layer. The sedimentary structure of this thick layer is probably the result of the leaching and sedimentation of primary ash pseudomorphs. This process is indicative of wet post-depositional conditions, probably the result of minor lateral displacement of hearth material, through rain (see Fig. 11). While most formation of secondary calcite is attributed to moist cave conditions and water dripping (Karkanas et al. 2000; Shahack-Gross et al. 2004), it can also be caused by groundwater, where the effects are more massive, and can include sparite (Mallol and Goldberg 2017). Sparite is not observed in the profile here, indicating that we’re not dealing with the unrestricted formation of crystals and groundwater effects, but the flowing cemented structure does indicate wet conditions.

Fig. 11
figure 11

Ash laminated cement (mF4), crystal cluster and minor cementation (both mF2)

Full size image

Small aggregates of ash are found in both the sandy clay groundmass and in the humified organic matter groundmass, indicating that such secondary calcites could also have been reworked. The smaller cemented clusters of ash, sometimes with shells, indicates that the remains of fires and shells were collected and discarded.

Charred organic matter

Charred organic matter is present throughout the profile, both in G1 and in G2. In G1, a concentration of large particles of charred organic matter is present in thin Sect. 004, layer III. Combined with the evidence for burnt shells and aggregates of ash, we can argue that this charred organic matter concentration is part of a reworked combustion feature. Other fragments of charred organic matter are sand sized and angular, indicating relatively minor displacement.

Uncharred organic matter

Uncharred organic matter consists primarily of roots found in the top of the sequence, as discussed above. A particular example of organic matter that must be mentioned here, also from the top of the sequence, is the freshly broken pericarp of Corylus avellana, a hazelnut, characteristic with its stone cells and resin ducts (see Fig. 12). Microscope photos of such pericarp were published by Schweingruber (Schweingruber and Börner 2018, 26, Fig. 4.137), Ismail-Meyer (2017, 130, Fig. 13.9) and most recently by Holguin et al. (2022). At Tegelbarg, several of these hazelnut pericarp fragments were used for 14C dating, producing results inconsistent with the main habitation phase of the shell midden (see Höfgen 2018). The implications of this hazelnut pericarp in the top of the micromorphological sequence here, for the site’s chronology and stratigraphy, will be discussed below.

Fig. 12
figure 12

Pericarp of a hazelnut in the top of G2 (note the stone-cells and the resin ducts)

Full size image

Ceramics

Only two pieces of ceramics were found in the thin sections. In slide 004, a piece of ceramics is part of the shell midden deposit. The second piece of ceramics is found in slide 003 layer I, indicative of anthropogenic input and human activities prior to the deposition of the shell midden in this place. Both pieces show large planar voids and a heterogenous clay composition, indicative of low-burning temperatures. The clay itself is tempered with coarse rock fragments, and burnt in an oxidising environment, considering the yellow and beige colours. Comparatively, Arnold (1981a, 154–155) describes the pottery from the site (most of it, interestingly, coming from below the shell midden) as burnt to low temperature, in an oxygen-rich environment, and tempered with coarse granite grit.

Bone

In total, 21 fragments of bone have been identified throughout the profile column (see Table 5). To understand site taphonomy and changes in environmental conditions before, during, and after occupation, all 21 bone fragments were studied for their bioerosion characteristics (birefringence, fluorescence, shape, structure, colour), and compared to a wide body of literature available (Huisman et al. 2017; Brönnimann et al. 2018, 2020a, 2020b; Turner-Walker 2012, 2019; Turner-Walker and Jans 2008; Ismail-Meyer et al. 2020; Trueman and Martill 2002; Trueman et al. 2004; Booth 2016; Booth and Madgwick 2016).

Table 5 Overview of all the bone fragments from the thin sections and their character
Full size table

Three kinds of degradation are visible (see Fig. 13): Firstly, bones 6, 7, 10, 11, and 20 show signs of a chemical or physical weathering process. Such bone fragments are described as having an angular-blocky structure by Huisman et al. (2017).

Fig. 13
figure 13

A selection of three bone fragments from Tegelbarg, showing blocky degradation (bone 6), cyanobacterial tunneling (bone 12), and modification caused by heat (bone 9)

Full size image

A second type of bone degradation present at Tegelbarg is tunnelling in both bone 1, 12, and 15. The three bones however exhibit two different kinds of tunnelling. In bone 1, this tunnelling consists of small tunnels, interconnected and widespread throughout the bone, consistent with the Wedl-type 1 tunnels (see also Brönnimann et al. 2018, Fig. 1). Wedl-type 1 tunnels are the result of fungal decay, and related to moist (but not waterlogged), oxygenated and neutral to acidic environments. Since the shells provide a basic environment, the presence of oxygen must have caused the fungal decay.

In bones 12 and 15, on the other hand, the tunnels are much larger and more concentrated in the small piece of bone. These tunnels are more consistent with a cyanobacterial impact (see examples in Huisman et al. 2017, Fig. 4; Villagrán et al. 2017; Brönnimann et al. 2018; Ismail-Meyer et al. 2020; Turner-Walker and Jans 2008).

The absence of degradation on the other pieces of bone, indicates an oxygen-poor and basic environment, visible in bone 13 and 14, and that mineralogical alterations have had an effect (e.g. due to heating, as visible in bones 9 and 16, 17, 18, and 19), or that the bones were transported from a different environmental situation. This is possibly the case for the small, rounded bone fragments 4, 8, and 10 (although their size might also limit the visibility of any degradation).

Flint

In total, 5 pieces of flint were observed in the thin sections. All the pieces are small and indicative of the micro-debitage of flint working activities occurring at the site (cf. Arnold 1973, 1974, 1981b, 1990, see Fig. 14). The flint fragments itself are angular in shape, with sharp and prominent curved boundaries, and have a smooth surface (cf. Angelucci 2010, 2017). The flint fragments are found throughout the profile, indicating that flint working activities took place throughout the occupation of the site.

Fig. 14
figure 14

An example of flint debitage (in mF2)

Full size image

Microfacies description

Based on the presence, distribution, abundance and characteristics of anthropogenic admixtures, and their relation to the two groundmasses and the natural phenomena, we can distinguish four distinct microfacies (see Fig. 15).

Fig. 15
figure 15

Photos of the four microfacies, showing the characteristic features of each

Full size image

  • mF1: This microfacies is found in a G1 groundmass, together with dusty clay coatings. If consists of clumped aggregates of anthropogenic material (charred organic matter, undegraded shell, ash, soil material), and loose particles of charred organic matter and shell fragments. This mF is present in M002 (see Figs. 5 and 15). It is clearly the result of reworking of the G1 groundmass, and a barren soil formed after vegetation clearance, in the older sedimentary deposits at the time of the first midden deposition.

  • mF2: This microfacies is found in a G2 groundmass, and a horizontal and sloping bedding layer of shells, showing interconnectedness and freshly broken examples. Dispersed between these shells are fragments of charred organic matter, ash clusters, and highly degraded fragments bone. This mF is present in M003 and M004 (see Figs. 5 and 15). It can be interpreted as the dumping of shells and occupational waste.

  • mF3: This microfacies is found in a G2 groundmass and consists of fragments of shells in a random orientation and not interconnected. Reworking is relatively minor, showing no rounded edges on charred organic matter. Many shells do show signs of degradation. Additionally, this microfacies shows several cases of unburnt organic matter, including plant roots and hazelnut pericarp fragments. This mF is present in M005 and M006A/B (see Figs. 5, 12, and 15). It can be interpreted as the trampling and reworking of dumped shells and occupational waste.

  • mF4: This microfacies is found in a G2 groundmass with dusty clay coatings and consists of locally dumped and secondary precipitated ash (aggregates of ash crystals and massive laminated cementation structures), horizontally embedded fragments of burnt shells, large angular charcoal pieces with dusty clay coatings, and degraded fragments of bone. This mF is visible in M004 (two instances, see Figs. 5 , 11, and 15), and can be interpreted as the reworking of combustion feature deposits, with the local involvement of water. Bone degradation, the result of cyanobacterial attack also indicates the active involvement of lagoon water.

Radiocarbon dating: chronological modelling

Our chronological model is based on the 33 radiocarbon determinations presented above, and the stratigraphy of the site. For the model, we use OxCal 4.2 (Bronk Ramsey 2009) and INTCAL20 (Reimer et al 2020). The stratigraphy of the site is based on both the field observations during excavation and the micromorphological observations presented above. This can be summarised in three aspects worthy of detailed discussion: the position of the grave, the stratigraphy of the shell midden layers, and the development of the peat. All these aspects are dealt with in greater detail in supplementary information 1. Suffice to say here, is that the site can be divided into three phases: a pre-midden phase, a midden phase, and a peat phase. The pre-midden phase consists of the primary and secondary grave fill, and the beach sand layer. The midden phase consists of the tertiary grave fill, the finds layer, and the gyttja layer. The peat phase consists solely of the peat layer. Based on this overall stratigraphy, we constructed a Harris matrix (see Fig. 16), as basis for the Bayesian model.

Fig. 16
figure 16

Harris matrix of Tegelbarg stratigraphy

Full size image

Our Bayesian model is constructed using the above outlined phases, and double boundaries, since hiatuses between the phases are likely. The dated shell samples are considered as terminus post quem within the model, because of likely reservoir effects and other possible problems related to carbonates (see Ascough et al 2004; Douka et al 2010)  In supplementary information 2, we present further details concerning the Bayesian model. Using Span and Date functions in OxCal we calculate the duration of each phase and the time period between the various phases within the model.

The chronological model, graphically represented in Fig. 17, results in our two phases of human occupation: The first phase of human presence is characterised by the digging of the grave and interment of the deceased individual in a coffin, between 2706 and 2477 cal BC (95.4% probability). The secondary grave fill, containing the flint dagger deposited into this grave can be dated between 2622 and 2338 cal BC (95.4% probability).

Fig. 17
figure 17

The Bayesian model for Tegelbarg

Full size image

The interval between the primary grave fill (the digging of the grave) and the secondary grave fill is 0–176 years (95.4% probability). In this same period, we can date the felling of the oak trees and putting them up as posts in the lagoon sediments, between 2774 and 2185 cal BC (95.4% probability). The shell midden occupation itself can be dated between 2186 and 1714 cal BC (95.4% probability), with the lower midden phase dating between 2167 and 1851 cal BC (95.4% probability), and the upper midden phase dating between 1987 and 1690 cal BC (95.4% probability). Finally, the peat formation dates between 1286 and 616 cal BC (95.4% probability). The interval between the end of shell midden occupation and the beginning of the peat formation is between 253 and 913 years (95.4% probability).

Discussion

Landscape and climate

Vegetation clearance

Regarding the formation of the landscape at Tegelbarg, our analysis of the soil micromorphology has produced several results that further our understanding of the landscape developments along the Schleswig–Holstein coast, and on the Pleistocene outcrop on which the site of Tegelbarg is situated. The G1 groundmass, and mF1 microfacies contain dusty clay coatings, indicating soil formation and rain splash on a barren land surface indicative of vegetation clearance (Courty et al 1989, 125). This barren land surface was created during the first midden deposits, containing aggregates of shell, ash, and soil material (see Fig. 5). This is an indication for the clearing of vegetation contemporaneous with the formation of the shell midden.

Furthermore, the combined analysis of soil micromorphology and radiocarbon dating has informed us on the moment of peat formation, a topic that only recently came under closer scrutiny (Quik et al 2023). The presence of an uncharred hazelnut pericarp in a soil micromorphology thin section, the 14C dating of uncharred hazelnut pericarps, and other uncharred plant material, to the Late Bronze Age, indicates that during this period, also simultaneous to the formation of beach barriers along the Baltic coast, peat started growing on the site, prohibiting the desiccation of organic matter. As such, a hiatus of almost a millennium can be assumed between the end of occupation and the start of peat formation. More perspectives on this, from for instance palynology (Meurers-Balke et al. 1985), would be necessary to test this hypothesis.

The chronology of use for the lower-lying lagoon, in which the gyttja was deposited, remains relatively unclear. The chronological model indicates a possibly continued deposition of occupational debris in this environment, extending the occupation well into the 2nd millennium BC, but this can be caused by the inaccuracies of older radiocarbon determinations. The finds from the excavations clearly do not support this late occupation since almost no Bronze Age artefacts (such as flint daggers of late types, or Bronze age pottery) were recorded. Future work in soil micromorphology and radiocarbon dating, on any coastal settlement, should therefore focus on obtaining samples from all depositional environments, in order to understand their formation processes.

Palaeoclimatic evidence

Several indicators from our micromorphological analysis inform us on the palaeoclimate during the Late Neolithic occupation at Tegelbarg. The degradation of the bone fragments from the shell midden sequence informs our understanding of landscape development and climatic changes. The blocky degradation observed in some of the bones is indicative of alternating wet and dry conditions and changing atmospheric redox conditions after deposition, both happening in an iron-rich marine environment. These conditions lead to the dissolution of primary bone minerals and the formation of phosphatic (Ca Fe (III)) minerals such as mitridatite, as observed at the sites of Hazendonk in the Netherlands and Stavanger in Norway (Huisman et al. 2017). A visually similar degradation of bone was also encountered at other sites in Northern Europe, such as at Skanör in Norway (pers. comm. R.I. Macphail).

The tunnelling by cyanobacteria, observed in several bones, holds important implications for understanding the lagoonal waters at Tegelbarg. This type of bioerosion is related to the submersion of bones in shallow clear water where these types of bacteria are present. Cyanobacterial blooms are common in the Baltic Sea, both in past and present, primarily during periods of higher sea surface temperatures. Especially summer Sea Surface Temperatures exceeding 16 °C are needed for N2-fixing cyanobacteria to develop (Warden et al. 2017). A local signal of water temperatures in the shallow lagoon near Tegelbarg, would be needed to understand the bloom of cyanobacteria and its impact on occupation of the shell midden.

Human action

From the thin sections at the siteBoth the archaeological soil micromorphology and Bayesian modelling of radiocarbon dates have given us further insight into the pre-midden activities that have taken place on the Pleistocene outcrop on which later the shell midden would arise. The oldest traces of human presence are represented by the digging of a grave pit, and interment of an individual on a charred coffin or deathbed, between 2706 and 2477 cal BC.

This funerary practice using a wooden coffin (known in German as a Bohlensarg) is not uncommon for the Corded Ware phenomenon in the Cimbrian Peninsula and primarily found in the Younger Neolithic. This possible attribution agrees with our radiocarbon date for the grave (see Hübner 2005, 500–512). Further archaeological traces of the Corded Ware phenomenon at the site of Tegelbarg are scarce. Two cord-decorated sherds and a single fragment of a battle axe were found in the shell midden layer, probably the result of reworking activities of the subsoil similar to mF1, but not in the vicinity of the burial (Arnold 1981b, 155).

The chronological model points to 0–176 years between the grave construction and the dagger inclusion. Based on the significant differences in soil characteristics between the primary fill and the secondary fill (including the amount of shells) and the lack of collagen in the bone from the grave that was sampled for 14C dating, it is more likely towards the longer end of this 0–176-year interval. This indicates that personal, or active memories, were present between the people who buried the deceased and those who provided the flint dagger (cf. Müller 2018). It now becomes clear that depositing a flint dagger into this grave, should be seen as a very significant act by which later communities acknowledge their ties to earlier groups on this outcrop.

Funerary practices are often found on shell midden sites. Well known are the shell midden burials of Téviec and Hoëdic in Brittany, France (e.g. Boulestin 2016; Dupont and Marchand 2020), the shell mounds in Santa Catarina, Brazil (e.g. Villagran 2018, 348–349), the shell midden of Riņņukalns, Latvia (Brinker et al 2020), the Sado and Muge valleys of Portugal (e.g. Peyroteo-Stjerna 2016), and the shell midden of Mienakker in the Netherlands (Kleijne et al. 2013b). This has led some to interpret shell middens in relation to monumental funerary architecture (e.g. Cummings 2003; Nobles 2020). Since the burial from Tegelbarg clearly predates the shell midden formation, and the burial is found outside of the densest accumulation of shell remains (see Fig. 3), it is difficult to follow this interpretation here.

Midden activities

From the thin sections at the site of Tegelbarg, we have uncovered four distinct microfacies which reveal human actions that led to the formation of the midden. Radiocarbon dating allow us to constrain these activities between 2186 and 1714 cal BC. As argued above, it is likely that habitation falls within the first two centuries of this range, so between c. 2200 and 2000 BC. Two kinds of different midden deposition activities could be discerned through micromorphology (mF2 and mF3), that are simultaneously phases in the occupation of Tegelbarg.

During the first phase of midden deposition, associated with mF2, digging activities in the Pleistocene soil were undertaken. At the same time, the first dumping of occupational debris (shells, bones, flint debitage, and pottery) was taking place in distinct midden heaps. These heaps of midden material were moved around, but not trampled upon. Also, combustion structures were placed on the surface. The remains of these midden heaps and combustion structures (shells, aggregates of ash and charred organic matter), ended up mixed with Pleistocene soil material.

The second phase of midden deposition, associated with mF3, is found in the local reworking of midden deposits, the dumping of hearth refuse on top of midden material, and the subsequent evidence for trampling from the top of the midden sequence as evidenced by more fragmented and more randomly oriented shells. It appears that during this second phase less effort was taken to create a living surface and to spatially separate dumping places from places that were lived upon. Chronologically, we can associate mF2 with the lower midden phase, dating between 2167 and 1851 cal BC (95.4% probability), and mF3 with the upper midden phase dating between 1987 and 1690 cal BC (95.4% probability). 

Shell and hearth dumps

Comparing the Tegelbarg shell midden deposits to other shell middens that have been studied micromorphologically in Northern Europe, we can see several similarities and differences that can be explained in spatial and cultural terms, related to dealing with waste material. Most notable is the uniformity of the shell microfacies at Tegelbarg (without omnivore coprolites, larger bone accumulations, or opal phytoliths). This is different from the Dutch shell middens (Colenberg 2014). At the Dutch shell middens, permanent settlements dating to the Late Neolithic, the shell microfacies primarily consists of organic matter, with only a limited accumulation of shells, all trampled and broken, and even here we find some evidence for coprolites. Most notable however, at the Dutch middens, are the massive layers of horizontally bedded phytoliths and charred organic matter. As such, the absence of large bone accumulations, coprolites or phytolith layers, could be indicative of a more short-term occupation at Tegelbarg, where material was clearly tossed instead of reworked (cf. Duarte et al. 2019).

When we follow through this argument, we can see the horizontal bedding of the shell midden layers at Tegelbarg as evidence for a distinct practice in preparing the land surface for walking upon, and a specific way in which shells are collected and discarded after consumption. At first, the living area and the waste disposal area of Tegelbarg were probably spatially separated in some way. Only in later phases do we see evidence for reworking and trampling, like the Dutch shell middens. This is indicative of walking on the shell midden, and possibly scarcity of land surface with a slowly drowning outcrop. Finally, the cleanness of shell deposits at Tegelbarg might be the result of acidity of the shell midden, but it might also be the result of specific waste disposal practices, which did not involve the collection of various categories of domestic waste prior to dumping.

Fish consumption?

One of the notable peculiarities of Tegelbarg, is the near absence of fish remains. Only two bones of gadid were identified by Reichstein (1975), and no fish remains were encountered in the (completely wet-sieved) material from the 2010 excavations (Höfgen 2018). Additionally, no fishing implements were found. All in all, this further reinforced the interpretation that Tegelbarg was a small and specialized site, primarily used as an extraction camp for shellfish and a production place for flint daggers (Hartz and Müller 2018).

However, our analysis presents a different picture. Alternatively, we can relate this near absence of fish remains to the high amount and the varied nature of the degradation and bioerosion patterns in bone fragments from our soil micromorphological thin sections. As several studies have illustrated (e.g. Nicholson 1996; Szpak 2011), fish remains degrade easier than mammal remains, because of differences in collagen chemistry. The absence of fish remains at Tegelbarg, therefore can better be interpreted as a taphonomic effect, rather than a result of specific prehistoric human action. This means that the absence of fish bones cannot be used to label Tegelbarg a special activity site. Simultaneously, the site could very well still be a special activity site, considering the absence of many of the soil micromorphological characteristics for permanent shell midden settlements.

Tegelbarg and the wider 3rd millennium BC

The significance of the grave within the Northern German Corded Ware phenomenon was discussed above. The midden occupation at Tegelbarg falls in the Spätneolithikum 1 or SN1 period, concurrent with the later phase of the European Bell Beaker phenomenon (see Kleijne et al. 2020; Schultrich 2018; Müller and Vandkilde 2020). The Bell Beaker phenomenon in Schleswig-Holstein is poorly known, and primarily consists of several funerary contexts and some settlements, tentatively dating between 2300 and 2000 BC, on which typical characteristic Bell Beaker artefacts (such as spatula-decorated pottery or wristguards) are found. Most graves in Schleswig-Holstein however do not contain any grave goods or are associated with daggers or arrowheads produced from local high-quality flint (Kühn 1979; Sarauw 2007; Kleijne et al. 2020). On the settlement of Tegelbarg, similarly, no indications for Bell Beaker material culture have been found. The production of flint daggers at Tegelbarg, and the addition of such a flint dagger to the older inhumation grave, do point to the positioning of these coastal communities within the wider Late Neolithic world. Regarding the subsistence economy of these Late Neolithic communities, most settlement and palynological evidence points to farming as the main source. While the Bayesian model didn’t provide us with a better date range for the midden formation, between 2200 and 1700 cal BC, it is argued that habitation most likely took place in the beginning centuries of this period, closer in time to the 4.2 ka BP climatic event. However rather than a cause/effect relationship, a flexible coastal economy can be envisaged (cf. Kleijne et al. 2020).

Conclusion

The site of Tegelbarg is exceptional, as no other shell middens are known dating to the Neolithic in Northern Germany. In this study we present a combined archaeological soil micromorphology and Bayesian modelling study of this shell midden, to understand the processes involved in its formation and the general occupation of the site. This combined approach has shown us several things, despite the limitations of a single narrow micromorphology column relative to understanding the occupation of a complete archaeological site.

The first phase of human activity on the site is characterized by a funerary monument of the late Corded Ware phenomenon. This grave was later incorporated into the Late Neolithic occupation, with the deposition of a prestigious flint dagger. Dumping activities on the site commenced in the Late Neolithic with evidence for the reworking of sediment and the formation of distinct midden heaps. Specifically, midden formation took the form of organic waste disposal, together with shells and ash from combustion features, and flint working debris. The continued dumping of waste material on top of these earliest heaps, shell waste and hearth deposits, led to the covering of the outcrop and trampling evidence. Interestingly, we have found widespread evidence for the degradation of bone fragments. This observation holds taphonomical implications, as it can better explain the absence of fish remains, over the choices of prehistoric peoples. Both the bone degradation (with evidence for cyanobacterial blooms) and ash layers further illustrate the increasing role of water on the sandy outcrop, culminating in its drowning, and, several hundreds of years later, the formation of a peat deposit.

Because of the absence of particular elements seen frequently in other, permanently inhabited, shell middens, we still agree with considering Tegelbarg a special activity site, dating to the Late Neolithic. Its significance to these Late Neolithic communities was probably profound, both in terms of the active memories involved, and in terms of the food supply in their flexible subsistence economy.