Introduction

Today, the Hümmling, an old morainic ridge in the Emsland, is characterised by cultivated land and parcels of semi-natural mixed oak–birch forests. This is largely the result of prolonged farming activity. Farming was introduced by Neolithic settlers in northern Germany ca. 4100 cal. b.c. (Kalis and Meurers-Balke 1998; Hartz et al. 2000; Müller et al. 2010) while intensification of farming took place between ca. 3600 and 3400 cal. b.c. during the phase of megalithic tomb construction (Hartz et al. 2007; Kirleis et al. 2012). The megalithic tombs of the Hümmling are assignable to the west group of the Funnel Beaker Culture (Trichterbecherkultur, TRB), the earliest Neolithic culture known from north-western Germany. These peoples occupied the area during the middle Neolithic (EN II–MN V according to the Northern European Plain chronology) between 3400 and 2800 cal. b.c. and were subsequently displaced by representatives of the Single Grave Culture (SGC; ca. 2800–2200 cal. b.c.) (Brindley 1986; Müller et al. 2010). Little is known about the Neolithic peoples who settled in Hümmling and the way they used and influenced their environment. Limited information about settlement pattern, houses and related features is provided by investigations on a few domestic sites (Nösler et al. 2011). Macrobotanical data from domestic sites that might be expected to contribute to our knowledge of subsistence strategies are hence scarce. Most of the available information from the wider region relates to the Elbe-Weser area where charred cereal remains, mainly Hordeum vulgare, Triticum dicoccum and T. monococcum, have been recorded in the TRB settlement Flögeln (ca. 3050 cal. b.c., details in Behre and Kučan 1994).

The present perception of the Neolithic in north-western Germany differs from that in neighbouring regions such as the Netherlands and Schleswig–Holstein in that these regions have evidence for Neolithic cultures that predate the middle Neolithic (Raemaekers 1999; Out 2009, 2012; Out and Verhoeven 2013; Hartz et al. 2007; Müller et al. 2010). This is not easily explained given that present-day political boundaries can have had no bearing on these prehistoric cultural developments (Raemaekers 2013). Preservation may be important insofar as the sandy soils of north-western Germany do not favour preservation or indeed detection of Neolithic sites (Nösler et al. 2011). The study of vegetation change as detected by pollen analysis, on the other hand, can be a powerful tool in that pollen data give clear insights into human impact and especially farming activity (Behre and Kučan 1994; Kalis and Meurers-Balke 1998; Bakker 2003). Palynological investigations to date in the Emsland, however, have rather poor sample resolution and chronological control is often inadequate or indeed unsatisfactory (Koch 1934; Jonas 1935, 1941, 1943; Kramm 1981) and so the available data are insufficient to enable a critical and reliable reconstruction of early prehistoric farming impact. Investigations at the Dümmer Lake, ca. 60 km to the south-east, show early human impact in connection with archaeological finds at the settlement site Hüde I and a wooden trackway dated to ca. 4800 cal. b.c. (Schütrumpf 1988; Bauerochse 2003; see Kampffmeyer 1991 for further consideration of the function and chronology of the site). Detailed palynological investigations in the Elbe-Weser area (Dörfler 1989; Behre and Kučan 1994; Heider 1995) have provided evidence for Neolithic settlement periods that parallel trends observed in other parts of northern Europe (Troels-Smith 1954; Iversen 1941; Rasmussen 2005; Wieckowska et al. 2012) though here also some aspects of the chronology might be further refined. These investigations show the crucial importance of short distances between coring location and Neolithic settlement sites for the reflection of human activity in pollen spectra (Behre and Kučan 1986).

In this study, two well-dated peat pollen profiles from Hümmling are presented with a focus on reconstructing Neolithic impact. Holschkenfehn, that is expected to reflect mainly local vegetation development and farming activity, was selected because of its close proximity to several megalithic tombs and two possible settlement sites (Fig. 1c). The pollen profile from Bockholter Dose, on the other hand, is expected to provide information on vegetation development and Neolithic impact at a regional level.

Fig. 1
figure 1

Maps relating to the study area and the wider region: a overview of north-western Germany showing location of the study area and Flögeln (indicated by rectangles with non-dashed and dashed boundaries, respectively). NL, Netherlands; LS, Lower Saxony; SH, Schleswig–Holstein: b map of Hümmling showing main features including relevant archaeological sites. Hatching indicates present-day bog distribution. 1 Bockholter Dose, 2 Holschkenfehn; c Detailed map of sampling site Holschkenfehn. Grey shading indicates afforestation. Mapping of megalithic tombs and settlement sites based on current knowledge based on ongoing research

Full size image

Study sites

Hümmling, the study area, is in Emsland, Lower Saxony, between the river Ems and its tributaries Hase and Ohe (Fig. 1). The landscape is largely defined by an east–west orientated morainic (Geest) ridge formed during the Saale, i.e. the penultimate glaciation (Hauschild and Lüttig 1993). The landscape is also influenced by later geological developments. Melting of ice sheets after the last glacial maximum led to a sea-level rise that resulted in a higher groundwater-table. This, in turn, favoured the development of fens and reed swamps in low-lying areas (Behre 2007). At the same time, the progressively advancing North Sea shoreline resulted in a more oceanic climate that favoured the formation of ombrotrophic bogs (raised bogs) on the sandy soils after ca. 6000 cal. b.c. (Petzelberger et al. 1999). Expansion of bog resulted in part displacement of the dominant oak-rich forests that included Betula, Tilia, Ulmus and Acer. The climate is humid (rainfall 815 mm/year), summers are relatively cool and winters are mild (mean annual temperature 9.0 °C; meteorological data relate to the meteorological station at Löningen, Deutscher Wetterdienst 1996).

The present-day woodlands consist mainly of birch and oak (Betulo-Quercetum roboris). In areas with podsols and acid gley soils, Pinus sylvestris is more common on drier stands but also grows on peatlands, while Alnus glutinosa is common on water-saturated mineral soils. On fertile (but podsolised) luvisols, mixed-oak forests are dominated by Fagus sylvatica (Fago-Quercetum petraeae; Pott 2002). F. sylvatica occupies the niche filled by Tilia cordata in the former Atlantic forests and hinders the spread of weak competitors such as Carpinus betulus which is generally pushed to the limit of its ecological range on water-saturated soils that are unsuitable for beech (Pott 2002).

Today, areas no longer suitable for farming due to soil degradation and leaching carry heathlands with Juniperus communis or heather-rich communities with Calluna vulgaris and Erica tetralix (Menke 1963). This process of degradation probably started during the Neolithic when there was substantial activity. More than 80 megalithic tombs are known from the area which, in itself, points to a substantial Neolithic TRB population (Fig. 1b). Heede is the only Neolithic settlement site investigated in Emsland that can with certainty be ascribed to the Neolithic (Fröhlich 1991). It is situated west to Hümmling and the river Ems and has been dated by archaeological finds to TRB and SG cultures. Based on surface finds, five areas with potential settlement sites have recently been identified in Hümmling (Fig. 1b). The small number of possible domestic sites is possibly due to destruction by deep ploughing (Kramer et al. 2012).

Holschkenfehn is a nature reserve characterised by birch-pine woodlands that include oak and heaths that are dominated by Calluna vulgaris (Fig. 1). Within a woodland area, an ephemeral pond (52°46′55.5″N, 7°30′49.14″E; Fig. 1c; dimensions: ~300 m × 80 m) was sampled. The pond today lacks open water but local people are aware of ice skating taking place there as late as the 1990s. Within 100–250 m of the sampling site, seven megalithic tombs and two chambered graves, attributable to TRB, West Group (Schlicht 1972), have been recorded. In the vicinity of the sampling site, two potential TRB settlement sites had been identified based on find scatters (Fig. 1c). Features to the south of the site were originally regarded as evidence for a TRB house (Kaltofen 1991) but this is now regarded as unlikely (Nösler et al. 2011).

Peat from the nature reserve Bockholter Dose was used as source for a regional pollen diagram (52° 51′ 59″ N, 7° 44′ 55″ E). Today, this bog occupies about 130 ha but was much more extensive prior to drainage in the context of intensive agriculture (LBEG 2012). The area today supports heath with C. vulgaris and M. caerulea and is the subject of restoration by raising the prevailing water-table.

Materials and methods

Sampling and dating

At the Holschkenfehn a 1.5 × 2 m2 pit was trenched after several test drillings to locate an appropriate sampling point. A 1.40 m-long peat profile was recovered in overlapping segments at the deepest point within the former pond from a clean profile wall, using three zinc boxes (50 × 10 × 5 cm3). Beneath the top soil, highly decomposed peat was recovered with some sand in the deeper parts. The base of the profile (below 108 cm) consisted of humic sand.

At the Bockholter Dose, a 3.50 m-long peat core was collected using a Russian corer (chamber dimensions: 50 × 4 cm). The core was recovered close to the northern border of the nature reserve. In the present study the interval 66–220 cm, consisting of low to medium-decomposed peat, was analysed.

Seven AMS 14C dates were obtained for the profile from Bockholter Dose and eleven from Holschkenfehn. The samples from Holschkenfehn were soaked in KOH (5 %). Macro-remains suitable for dating were picked from the material, retained by a 200 μm-mesh sieve. Samples from Bockholter Dose yielded insufficient macro-remains for dating. Bulk peat samples were therefore used, rootlets having been removed with the help of a binocular microscope.

Age-depth models were constructed using OxCal 4.1 (Ramsey 1995) based on IntCal09 (Reimer et al. 2009). The programme uses Bayesian statistics that incorporate the prior model (depth and deposition order) and the 14C dating information for the construction of the age-depth model. The applied P_Sequence approaches the unknown deposition rate by introducing the parameter k which represents the number of accumulation processes per depth unit. The parameter was set to 1 deposition unit per cm to avoid circular reasoning and in cognizance of the randomness in deposition processes (Ramsey 2008). In the Holschkenfehn profile, the increase of Secale pollen to more than 1 % was used as an additional date for the construction of the age-depth model. This implies that the upper two samples relate to the Roman Iron Age or are somewhat younger (cf. Behre and Kučan 1994).

Palynology, macro-remains and charcoal

The Holschkenfehn core was sampled continuously using 1 cm-thick samples in the interval relating to the Neolithic (each sample is estimated to represent ~40 years). Elsewhere, the sampling interval was 2–5 cm. The uppermost 15 cm was highly decomposed and was therefore not analysed. In total, 66 samples were analysed. The Bockholter Dose core was sampled at 2–5 cm intervals giving a total of 39 samples and a time resolution of 40–100 years.

The pollen samples (1 cm3) were processed using the standard protocol that includes treatment with KOH (10 %), HCl (10 %), sieving using a 200 μm mesh, HF (40 %) and acetolysis (2.5 min) (Fægri and Iversen 1989). Lycopodium tablets (Batch No. 177745, Lund University) were added at the start of the preparation procedure to enable subsequent calculation of sporomorph and charcoal (size range: 10–200 μm) concentrations (Stockmarr 1971). The samples were mounted in glycerine and analysed using a light microscope at ×400 magnification. A magnification of ×1,000 and phase contrast was used where it was desirable to see the detailed structure and surface pattern of specific pollen grains. Identification and nomenclature of pollen types mainly follow Beug 2004, and in a few instances (Moore et al. 1991). Non-pollen palynomorphs (NPPs) are after van Geel (1978), algae and fungi follow van Geel et al. (1980/1981) and for testate amoebae the publication by Charman et al. (2000) was used.

At least 800 arboreal pollen grains (AP; Corylus excluded) per sample were counted in the Holschkenfehn profile. In samples with abundant Alnus pollen, 1,000 AP grains were counted. Large amounts of plant debris made counting of the Bockholter Dose samples very difficult so that, in general, three slides (24 × 32 mm2) were required to achieve an AP count of 600.

Pollen and NPP percentages were calculated using a total terrestrial pollen sum (TTP). Calluna was excluded from the pollen sum as it was regarded as a local element of the bog vegetation. This is supported by frequent macro-remains of Calluna recorded from both profiles. The data are presented as percentages which are regarded as preferable to concentration or pollen accumulation rate (influx) data (cf. Waller et al. 2012). Other considerations influencing this choice include the possible effects of climate on pollen productivity (van der Knaap et al. 2010) and uncertainties relating to dating uncertainties.

The pollen diagrams were drawn using Tilia ver. 2.0.2. (Grimm 2004). Pollen zone boundaries were placed after careful visual inspection of the profiles.

Macro-botanical remains (>200 μm) obtained by sieving during pollen preparation were examined using a stereo-microscope. Charcoal concentrations are given as particles/cm3.

Ordinations

Pollen percentage data from each profile were analysed using principal components analysis (PCA), as the data sets showed a linear response to theoretical gradients during detrended correspondence analysis (DCA) (gradient length <2; DCA results not shown).

The analyses were run to objectively investigate pollen taxa responses to theoretical gradients (axes of the ordination plot) and to identify taxa that react similarly within the data set. Other points of interest included checking for sample clustering and validation of the pollen zones.

For the analyses, all terrestrial pollen types that were present in at least three samples throughout the profile or showed values >0.5 % in one sample were included. Percentage data were square-root transformed to improve comparability. For these analyses, CANOCO ver. 4.5 and CanoDraw ver. 4.5 were used (ter Braak and Smilauer 2002).

A detrended canonical correspondence analysis (DCCA) was applied to the AP data from the Bockholter Dose to get a semi-quantitative measure of regional forest changes during the Neolithic. Herb pollen was excluded from the analysis as it is under-represented in forested areas (Hicks 1971) and may derive mainly from local vegetation. By introducing the age-depth relationship as a constrained gradient in the analysis, the DCCA enables the detection of total compositional species turnover through time (Birks 2007). Detrending was done by segments and percentages were square-root transformed and scaled non-linearly. The species scores are given as direct standard deviation units (SD) of the compositional turnover (Birks 2007; Hill and Gauche Hill and Gauch 1980).

Results

Age-depth modelling

The results of the radiocarbon dating are presented in Table 1 for both profiles, and age-depth models are shown in Figs. 2 and 3 (Bockholter Dose and Holschkenfehn, respectively). The radiocarbon dates for Holschkenfehn were partly published in Nösler et al. (2011) and Kramer et al. (2012); the radiocarbon date for depth 45 cm was, however, incorrect and has since been re-dated. For construction of the age-depth model for Holschkenfehn, the results from the lowermost sample, 114–115 cm (KIA-42180), were excluded from the analysis. The leached residue contained insufficient carbon for reliable dating (0.1 mg). The humic fraction, on the other hand, returned an unexpectedly young date. This is probably attributable to humic matter deriving from the overlying peat. The calibrated and modelled ages are in good agreement. As regards Bockholter Dose, all dates were used to construct the age-depth model and the modelled ages are in good agreement with the calibrated dates. Ages in the following text are given as the arithmetic mean of the 95.4 significance interval (2 σ range).

Table 1 Dating results
Full size table
Fig. 2
figure 2

Age-depth model for Bockholter Dose, profile DMB

Full size image
Fig. 3
figure 3

Age-depth model for Holschkenfehn, profile HSV

Full size image

Pollen diagrams

Pollen diagrams from Bockholter Dose and Holschkenfehn are presented in Figs. 4 and 5, respectively. In Fig. 6, curves for the wetland taxa and NPP in the Holschkenfehn profile are presented. The PAZs and the macrofossil data relating to Bockholter Dose and Holschkenfehn are summarised in Tables 2, 3, respectively.

Fig. 4
figure 4

Pollen diagram of main taxa from the Bockholter Dose (profile DMB). Percentages based on total terrestrial pollen counts (TTP). Where values are low, silhouettes show the values magnified ×10. Analysis: Kramer 2012

Full size image
Fig. 5
figure 5

Pollen diagram showing curves for AP, shrub taxa and NAP from Holschkenfehn (profile HSV). Percentages are based on TTP. Where values are low, silhouettes show the values magnified ×10. Analysis: Kramer 2012

Full size image
Fig. 6
figure 6

Diagram from Holschkenfehn (profile HSV), showing wetland taxa and non-pollen palynomorphs. Where values are low, silhouettes show the values magnified ×10. Analysis: Kramer 2012

Full size image
Table 2 PAZ descriptions and macrobotanical records, Bockholter Dose profile
Full size table
Table 3 PAZ description of Holschkenfehn pollen spectra and macrobotanical remains
Full size table

Ordination

The ordination diagrams (Figs. 7, 8) show the first two axes of the PCA. Only taxa that fit 25 % or more to the axes are displayed in the ordination plot. Samples are shown using symbols that indicate the PAZs as distinguished by visual inspection of the pollen diagram. The ordination of the terrestrial pollen data from Holschkenfehn is shown in Fig. 7. The first and second axes account for 51 % of the total variance. Both axes seem to reflect vegetation disturbance. The first axis seems to reflect the transition from woodland to human-altered vegetation while the second axis appears to reflect a primary/secondary woodland gradient. The clustering of samples serves to confirm the validity of the zoning, and the transitional character of the vegetation and a trend towards more disturbances in PAZ 2.

Fig. 7
figure 7

Plot of results of PCA analysis of Holschkenfehn pollen data. Spectra relating to the various PAZs are enclosed within envelopes

Full size image
Fig. 8
figure 8

PCA plot from the Bockholter Dose. Spectra relating to the various PAZs are enclosed within envelopes

Full size image

Results of the PCA analysis of Bockholter Dose pollen spectra are shown in Fig. 8. The first axis (eigenvalue 0.35), i.e. the main gradient, separates pollen taxa that reflect woodland from taxa indicative of open vegetation. This axis is therefore used to assess openness of the vegetation (Fig. 9). Quercus separates from Corylus and Betula on the second axis, which suggests that the latter two taxa are indicative of transitional vegetation stages. The ordination shows poor clustering of spectra which suggests that the pollen zones are not distinctive and that changes are gradual.

Fig. 9
figure 9

Summary of vegetation changes and settlement activity as derived from openness of vegetation, compositional species turnover and human indicator pollen curves; main settlement phases are indicated by shading

Full size image

The DCCA carried out on the AP data from the Bockholter Dose profile (Fig. 9) shows a compositional species turnover of 0.6 (eigenvalue 0.03), measured as standard deviation units. Total variance of the data sets is given as 0.29 total inertia.

Discussion

Regional vegetation change inferred from the Bockholter Dose

The pollen spectra from the Bockholter Dose (Fig. 4) are considered to reflect regional vegetation changes as it is a large bog complex (Jacobson and Bradshaw 1981). The pollen record indicates that between 220 and 66 cm (4500 and 1900 cal. b.c.), the vegetation in the Hümmling area consisted of woodland dominated by Quercus, Betula, Fraxinus, Tilia and Ulmus. High Alnus values probably reflect the widespread presence of alder carr that was favoured by much wet habitat (Bauerochse 2003). Pinus, which has relatively high percentage pollen values, was probably largely confined to drier areas. Today, Betulo-Quercetum roboris communities are regarded as the climax stage of forest development on the sandy soils in the region (Pott 2002). It seems therefore that the woodland communities have changed little during the last 6,000 years. Only Tilia, a substantial component of the Atlantic forests of north-western Germany (Behre and Kučan 1994), has declined in importance. In the pollen record, it is greatly underrepresented due to low pollen productivity and poor pollen dispersal (Andersen 1967). Tilia was probably distributed on the better soils that are today occupied by beech or used for agriculture (Behre and Kučan 1994). Generally, low NAP values underline the regional character of the pollen rain at the site but, mostly wind-pollinated species like Artemisia, Rumex and members of the Chenopodiaceae and Poaceae are consistently represented. Changes in the woodland composition are visible throughout the record and the main trends are indicated by the ordination of samples along PCA axis 1 (Figs. 8, 9). The vegetation changed in PAZ 1 with Betula decreasing above 210 cm (4400 cal. b.c.), while increasing Corylus values point to openings in the forest canopy (Kalis et al. 2003). Further opening-up is suggested by Ulmus which declines at the PAZ 1/2 transition after 205 cm (4250 cal. b.c.) while Fagus pollen is recorded for the first time. It has been often discussed whether opening-up of woodland as a result of human activity facilitated local establishment of Fagus (Behre and Kučan 1994; Küster 1997; Pott 1997; Nielsen et al. 2012). Our investigations support the idea that beech benefited from human-induced woodland disturbances in our study area. The woodland openings seem to have been easily reversible so that first Betula and then Ulmus soon recovered again. However, a small change in the forest composition is visible at the beginning of PAZ 4 (after 155 cm, 3500 cal. b.c.) with declining Tilia and Ulmus in the mixed-oak woodlands (Figs 4, 9). The high Corylus values point to woodland clearings that were probably moderate considering the decline of total tree pollen from about 75 to 65 %. Settlement indicators like Plantago lanceolata, Pteridium aquilinum and Poaceae (non-cultivated and cerealia-type) in PAZ 3 above 170 cm (after 3800 cal. b.c.) suggest that the changing woodland composition is probably related to human activities in the region. Also increased macroscopic charcoal in samples from PAZs 3 and 4 (between 160 and 130 cm; 3620–3090 cal. b.c.) points to human activity in the area. In PAZ 5, 123–97 cm (2960–2490 cal. b.c.) woodland regenerated and Betula increased when human activities ceased, indicated by a decrease in Plantago and Pteridium values. Also, late succession trees like Tilia and Ulmus recovered for a short period between 2800 and 2600 cal. b.c. (110 and 100 cm). In PAZ 6 (97–66 cm, 2490–1920 cal. b.c.), Fagus is at about 1 % which suggests local presence of beech.

Local environmental changes at Holschkenfehn

Given the small size of the bog, the pollen spectra from Holschkenfehn (Fig. 5) are assumed to reflect local and extra-local vegetation changes (Jacobson and Bradshaw 1981; Moore et al. 1991). In PAZ 1 (5190–4180 cal b.c.), pollen from mixed-oak woodlands dominates and any changes are small. NAP taxa such as Succisa, Cichorioideae and Melampyrum are regularly recorded in PAZ 1 which points to locally present woodland clearings in the late Atlantic. At the transition PAZ 1/2, Ulmus declined while Tilia and Quercus increased. Whether higher values for Quercus and Tilia reflect greater contribution by oak and lime to the surrounding woodlands, enhanced flowering within an opened woodland or higher pollen influx from the regional vegetation is difficult to assess. The overall changes in PAZs 1 and 2, however, appear to be relatively small. The results of the PCA ordination also support this view (Fig. 7) and so it is concluded that there were only minor vegetation changes during this time.

In PAZ 3 (71–42 cm, 3520–2260 cal. b.c.), increasing values of human settlement indicators such as Rumex acetosella-type, Jasione montana-type, Scleranthus annuus and P. lanceolata as well as a decrease in AP (mainly Alnus, Betula and Tilia) indicates opening-up of woodland cover and suggests local presence of farmers. The NAP pollen types referred to above may be regarded as indicative of arable land as they are frequently found as segetal flora on fallow land (Behre 1981). Pollen from cerealia-type is also regularly present between 70 and 43 cm (3490 and 2310 cal. b.c.). This supports the idea of crop cultivation, even though the possibility that some of this pollen may arise from non-cultivated grasses cannot be excluded (see below). Stronger human impact is particularly evident in the periods between 3490 and 3230 cal. b.c. (70–65 cm), 3050–2870 cal. b.c. (60–54 cm) and from 2870 to 2310 cal. b.c. (54–43 cm). The local hydrological changes, as reflected by the NPP records (Fig. 6), show a shift to more humid conditions in zone 3 (indicators of dry conditions, including Assulina muscorum, Callidina angusticollis (Van Geel 1978) and Byssothecium circinalis (Van Geel et al. 1980/81) decrease). Pollen taxa, such as Hydrocotyle vulgaris, Utricularia, Scheuchzeria palustris and Sparganium-type, that are indicative of shallow dystrophic water bodies, point to increased wetness and probably the formation of an open water-body during this time. Human-induced woodland disturbance, which resulted in reduced evapo-transpiration and increased surface-water run-off, was probably mainly responsible for this development.

Pteridium aquilinum values increase strongly in PAZ 3b, i.e. at ca. 2870 cal. b.c. This suggests vegetation disturbance. Pteridium is common in mixed woodlands today but the ability to produce spores is largely restricted to situations where there are openings in the tree canopy. Furthermore, rhizomes that are resistant to disturbance and with high growth potential enable this fern to quickly expand, given favourable circumstances, by vegetative reproduction. Tree taxa such as Betula, Ulmus, Fraxinus, Pinus and Alnus decline sharply between 2610 and 2300 cal. b.c. (49 and 43 cm), which further suggests severe woodland disturbances. The stable values for Quercus are interpreted as reflecting a steady input of Quercus pollen from the wider region, as local vegetation was cleared. Synchronous with the decline in AP, Corylus increases in PAZ 3c at 2520–2260 cal. b.c. (47–42 cm) which supports the idea of further woodland clearance. A little later at 2420 cal. b.c. (45 cm), this is followed by maximum Poaceae values. This suggests that grasslands have replaced closed-canopy woodland, probably as a result of the impact of grazing animals (Groenman-van Waateringe 1993; Vera 2000). However, elevated Poaceae values as a response to locally drier conditions cannot be excluded from consideration (see below). Woodland recovered after these disturbances and Betula increased and achieved maximum values in PAZ 4a between 2260 and 1570 cal. b.c. (42 and 34 cm). As a high pollen producer, input of Betula pollen probably contributes to the reduction in Quercus values between 1990 and 1570 cal. b.c. (37 and 34 cm). When Betula decreased after 1570 cal. b.c. (34 cm), Quercus values increased again implying presence of well developed mixed-oak woodland.

Since neither human indicators nor high charcoal content is recorded during the Betula maximum in PAZ 4 we regard this as a regeneration phase. This phase lasted ca. 690 years and is comparable to regeneration trends recorded in pollen diagrams from the Elbe-Weser area after human disturbances (Behre and Kučan 1994). The longevity of the regeneration phase is probably due to the considerable leaching of the already poor sandy soils as a result of human disturbance. Betula, a tree that can cope well with poor, highly leached sandy soils, was favoured. Increased leaching also resulted in expansion of heathland during the Neolithic and later periods in this region (Behre and Kučan 1994; Behre 2000), though this development is not clearly reflected in the pollen profiles presented here.

Synchronous with the woodland disturbances a shift in the hydrological regime is suggested by high Sphagnum spore values after 2520 cal. b.c. (47 cm). According to Freund (1994) this might be related to drier conditions as Sphagnum tends to increase spore production under dry (less favourable) conditions. Bauerochse (2003) argued that higher Sphagnum values might reflect lowering of the groundwater table that would favour formation of Sphagnum peat. Typical open-water taxa (Hydrocotyle vulgaris, Utricularia, Scheuchzeria palustris, and Sparganium-type) decreased and other dryness indicators like Assulina muscorum, Callidina angusticollis and Copepoda spermatophores increased so conditions may well have been drier during this time at Holschkenfehn.

Woodland composition remained stable between 1570 and 170 cal. b.c. (34 and 23 cm) (in PAZ 4b), while occurrences of Secale, high values of Rumex acetosella and a decrease in AP, suggest enhanced farming in PAZ 5 (23–16 cm, 170 cal. b.c.–cal. a.d. 390). Since the major focus of the study was the Neolithic, the temporal resolution of the record after 1500 b.c. is lower (~500 years) compared with ~40 years during the Neolithic. This part of the Holschkenfehn record is therefore less detailed but, nevertheless, provides a useful general indication of developments–increased human impact, including rye cultivation–during this time.

Considerations regarding the Neolithic transition

First indications of the presence of Neolithic people in northern Germany are often related to the elm decline. The elm decline is a widespread phenomenon visible in most records from north-western Europe shortly after 4000 cal. b.c. (Tipping et al. 1993; Andersen and Rasmussen 1993; Peglar 1993; Behre and Kučan 1994; Heider 1995; Kalis and Meurers-Balke 1998; O’Connell and Molloy 2001; Molloy and O’Connell 2004; Ghilardi and O’Connell 2013). It is assumed that it originates from the interaction between the spread of a disease of elm and the impact of the first Neolithic farmers. A shift to a more continental climate may have also played a role in the elm decline (Moe and Rackham 1992; Peglar and Birks 1993; Parker et al. 2002).

The elm decline does not constitute a distinctive feature in the pollen profiles presented here as Ulmus does not exceed 3 %. Elm was probably present in the area (cf. Huntley and Birks 1983) but the rather infertile, old morainic soils were hardly favourable for this edaphically demanding tree. Declining elm values as recorded in the Hümmling profiles (cf. PAZ boundaries 1 and 2) predates the classical elm decline in northern Europe by about 300 years. This might be attributable to an earlier (human) impact on elm prior to the spread of an elm disease (cf. Bakker 2003; Feeser et al. 2012). In the Holschkenfehn profile, Ulmus declined rather gradually, from ca. 4270 cal. b.c. (91 cm) onwards. Whether this early decline resulted from Neolithic activity is difficult to say. The PCA ordination suggests little change (Fig. 7) and evidence for pastoral or arable farming is lacking (P. lanceolata is first recorded at 3520 cal. b.c.–71 cm). The records for cerealia-type pollen before the declining of elm probably originate from wild grasses such as Glyceria fluitans (Beug 2004) that is common in wet environments. A potential indicator of human-induced disturbances is Pteridium aquilinum that expanded as Ulmus declined. This fern is considered to reflect wood pasture and fire-induced forest clearings (Behre 1981). The practice of ‘slash and burn’ was probably important for Neolithic people in Germany (Rösch et al. 2002) and new results from the TRB North Group also indicate early use of fire (Feeser et al. 2012). We assume that the Holschkenfehn charcoal record is more related to activities at local settlement sites as no direct correlation between woodland cover, spread of Pteridium and enhanced charcoal input is visible.

Records of spores that may originate from coprophilous fungi, such as Sordaria, Cercophora (Van Geel 1978) and Sporormiella (Davis and Shafer 2006), point to wood pasture but wild animals cannot be excluded. We favour the latter, given that these spore types were also recorded prior to declining elm values and the first occurrence of P. lanceolata, the classic Neolithic settlement indicator (Iversen 1941; Behre 1981; Behre and Kučan 1994).

In the Bockholter Dose, Ulmus declines around 4250 cal. b.c. (205 cm) with a quick recovery after 250 years and a second decline again around 3620 cal. b.c. (160 cm) while settlement indicators did not register before 3800–3620 cal. b.c. (170–160 cm). As pollen spectra from the Bockholter Dose reflect regional woodland vegetation patterns, settlement-indicator pollen taxa and herb pollen in general is underrepresented (Hicks 1971). However, small decreases in Betula, Tilia and Alnus and slightly increasing Poaceae values after 4180 cal. b.c. (195 cm) point to woodland openings that in turn may indicate human activity. The DCCA data suggest a distinct change in woodland composition between 4250 and 4050 cal. b.c., after which there is a distinct change in the opposite direction which may point to woodland regeneration. The appearance of P. lanceolata at 3800 cal. b.c lagged declining elm values by 450 years and is associated with a decrease in Betula and an increase in Poaceae and Quercus. All together, there is no general pattern of vegetation responses to possible human activities. Evaluation of single woodland taxa, settlement indicators and the comparison between pioneer trees (e.g. Corylus, Betula, Fraxinus) and late succession trees (Ulmus, Tilia, Quercus) (Kalis et al. 2003) does not suggest any persistent trend in vegetation dynamics. However, the evidence points to woodland opening and possibly small-scale clearings in areas with differing woodland composition as dictated by successional and other factors.

The Neolithic farming practices that had led to the vegetation changes at the beginning of the Neolithic are difficult to assess by palynological investigations. Troels-Smith (1954) assumed that cutting leaves and twigs (from elm) for fodder for the stabled livestock and small-scale clearing of the woodlands for cultivation of cereals were the main techniques. He excluded wood pasture as Poaceae values were not particularly elevated. This model was also used by Behre and Kučan (1994) for the interpretation of several Neolithic profiles from the Siedlungskammer Flögeln. They connect the so-called leaf-fodder period to a ‘pre-megalithic culture’. Comparable interpretations have been applied to early Neolithic developments as documented at in the Alps (Rasmussen 1993; Akeret et al. 1999) and the Netherlands (Casparie et al. 1977). A recently published work investigated the stable-isotope composition of tooth enamel of livestock from the Neolithic village site Bercy in Paris, France. The results imply that feeding on leaves took place during winter which probably prolonged the period of cattle breeding from three to almost 6 months, and thus possibly securing dairy-product supply throughout the year (Balasse et al. 2012). Seasonal feeding that included wood pasture during summer and leaf fodder during winter might have been possible also in a north-west German context where such practices persisted well into the historical period (Burrichter and Pott 1983; cf. Bakker 2003). Pointers in the Bockholter Dose profile suggestive of woodland pasture include increases in Poaceae and Pteridium as Ulmus declines and P. lanceolata records commence. Increasing Quercus and Corylus values also support this view. In a review of woodland structure during the Linearbandkeramik (LBK) in Central Europe, Kreuz (2008) points out that high Quercus and Corylus values are probably indicative of open woodland suitable for wood pasture. A combination of grazing and leaf-foddering is also assumed by Bakker (2003) from palynological investigations on the Drenthe Plateau, the Netherlands. He found a comparable expression and timing of the elm decline, and sometimes delayed appearance of P. lanceolata followed by increasing Poaceae values. He related “Neolithic Occupation Period I” to the Swifterbant culture that was established in the Netherlands from ~5000 cal. b.c. (Out and Verhoeven 2013; Raemaekers 2013). It seems that the Neolithic transition took place as a gradual process within the Swifterbant culture from ca. 4500 cal. b.c. when animal husbandry began, while the first signs of cereal cultivation date to ca. 4200 cal. b.c (Cappers and Raemearkers 2008; Louwe Kooijmans 2009; Out 2009; Huisman et al. 2009; Raemaekers 2013). A comparable neolithisation process has also been put forward for areas to the east of Hümmling where comparable palynological data have been interpreted as indicative of different stages in the Neolithic transition within the Ertebølle-Ellerbek culture and TRB North Group (Kalis and Meurers-Balke 1998; Hartz et al. 2000; Kirleis et al. 2012).

The pollen data from the Bockholter Dose profile might also be interpreted as indicative of a series of gradual developments in the early Neolithic that included scattered openings in the woodland, these possibly dating to before the classical Elm Decline recorded in other areas in northern Europe, and varying vegetation responses to early farming activities. There are indications of grazing also in the Holschkenfehn profile before the decline of elm. Mesolithic activity, involving improving grazing conditions for wild animals or part adoption of a Neolithic subsistence strategy in the form of livestock breeding, may be involved. Indeed, grazing of wild animals may have produced these disturbances.

While clear archaeological evidence is lacking for early Neolithic in the region, settlement sites such as Hüde I in the Dümmer (Schütrumpf 1988; Kampffmeyer 1991), Boberg at the river Elbe (Hüser 2009) and Sievern 114, in the Cuxhaven district (Nösler et al. 2011; Kramer et al. 2012) point to the presence of early Neolithic settlers in north-western Germany, although details regarding the nature and extent of the activity and the chronology have yet to be resolved. Raemaekers (2013) argues for analogous neolithisation processes in north-western Germany and the Netherlands, a view that receives support in the interpretation of Mesolithic lithic artefacts from Lower Saxony as indicative of a more or less closed cultural group in Lower Saxony and the Netherlands (Mahlstedt 2012).

Human impact ascribable to the TRB West

The vegetation change as recorded in the regional pollen profile and species turnover (Figs. 4, 9) at Bockholter Dose between ca. 3600 and 3500 cal. b.c. involves more long-lasting modifications in woodland composition. Ulmus, Tilia, Fraxinus and Quercus decrease while Corylus increases strongly and Fagus is established in the region (Fig. 4). This indicates that woodlands opened and Corylus quickly occupied the resulting gaps or experienced enhanced flowering due to increased availability of light. Opening-up is also suggested by the PCA ordination plot (Fig. 8) where the distribution of samples on axis 1, which is regarded as reflecting a landscape openness gradient, shows a gradual trend to more open landscape and negative values after 3500 cal. b.c. (Fig. 9). The pollen spectra from Bockholter Dose reflect a regional picture of vegetation changes while spectra from the small site Holschkenfehn enable direct inference of local Neolithic land use (Behre and Kučan 1994). Here, changes in wood composition coincide with sharp increases of human settlement indicators at the site after 3490 cal. b.c. The first period of woodland disturbance lasted ca. 300 years, i.e. until 3230 cal. b.c. (70–65 cm) when tree pollen taxa (Betula, Tilia) recovered and P. lanceolata decreased. Between 3050 and 2870 cal. b.c. (60–54 cm), P. lanceolata values increase again and also AP taxa which suggests that stable forest vegetation, especially Ulmus and Tilia, declined.

Recently, several pedological and pollen analytical studies have been carried out in the vicinity of Neolithic and bronze age burials to investigate the degree of local landscape openings (Dreibrodt et al. 2009; Demnick et al. 2011; Fyfe 2012; Sadovnik et al. 2012). Holschkenfehn is located close to nine megalithic tombs, seven of which lie within 100–250 m (Fig. 1c). It is to be expected therefore that the impact of the construction and use of the tombs on the local environment might be inferable from the Holschkenfehn pollen data. The phases of landscape openness described above probably largely reflect the influence of local settlements. P. lanceolata, Scleranthus annuus, Rumex acetosella-type and also cerealia-type records are presumed to reflect local farming and, in particular, arable activity (Behre 1981). To date, two possible TRB settlement sites have been identified, based on find scatters to the south and west of the Holschkenfehn site (Fig. 1c).

The vegetation changes appear to mirror the pattern of Neolithic land use as reflected by pollen profiles in the Elbe-Weser area, and referred to as Landnam sensu Iversen (1941) and relating to the TRB culture (Dörfler 1989; Behre and Kučan 1994; Heider 1995). Substantial opening-up of the woodland cover indicates greater land use than before, and is probably related to more extensive wood pasture and cereal cultivation with barley, emmer and einkorn as major crops, the latter supported by macro-botanical records (Behre and Kučan 1994; Kirleis et al. 2012). Cerealia records, i.e. Hordeum-type and unidentified cereal-type pollen but excluding Secale which is identifiable with a high degree of certainty, probably also reflect cereal growing. Further differentiation of the cerealia-type pollen as regards size etc. may help in distinguishing pollen of cereal origin from large cereal-like pollen produced by non-cultivated grasses (cf. Ghilardi and O’Connell 2012).

Why the major impact on vegetation took place at ca. 3500 cal. b.c has been much discussed. In north-western Germany and the Netherlands the timing coincides with the emergence of the TRB West Group (Brindley 1986; Müller et al. 2010) as a clearly defined Neolithic culture. Similar changes are visible in Schleswig–Holstein. There, the first evidence for the Neolithic more or less coincides with changes in woodland composition at 4100 cal. b.c. (Müller et al. 2010; Feeser et al. 2012). The much later strong increase in human impact has been interpreted as a change from subsistence, mainly based on husbandry, to a farming economy with increased emphasis on arable activity (Behre and Kučan 1994). Other theories involve technological developments coinciding with the introduction of the ard (Wiethold 1998). The earliest dated plough marks in Lower Saxony, recorded close to Wittenwater (Tegtmeier 1993), point to the introduction of ploughing between the TRB and the early bronze age. However, the ard might be less efficient than hoes for cultivation (Bogaard 2004) and tillage in areas recently under woodland would not have been without difficulty due to persistent root systems (Rösch et al. 2002). In Ireland, changes in Neolithic subsistence strategies have been ascribed to climate change (precipitation and also temperature; Stolze et al. 2012) but it remains uncertain if climate had such a dominant influence (cf. Ghilardi and O’Connell 2012). Evidence gleaned from plant macroremains from Neolithic sites in northern Germany, and especially the shift in emphasis away from plant material collected in the wild to crops, imply a change from a surplus strategy to a full Neolithic lifestyle (Kirleis et al. 2012). Taken in conjunction with the emergence of the TRB West Group in north-western Europe, this appears to be an acceptable explanation also for the intensification of land use at ca. 3500 cal. b.c. recorded in the pollen profiles presented here.

Neolithic land-use pattern

The close proximity of the Holschkenfehn pollen profile to the archaeological evidence facilitates the reconstruction of different phases of Neolithic land use in the area. Small scale agriculture and wood pastures are assumed for the period between 3520 and 2870 cal. b.c. (71–54 cm) with a decrease in settlement activities between 3230 and 3050 cal. b.c. (65–60 cm) and lower impact between 3050 and 2870 cal. b.c. (60–54 cm). This is different to the situation at Flögeln during the Neolithic where Behre and Kučan (1994) assumed uninterrupted arable activity and continuous settlement. Given the poor sandy soils, highly susceptible to leaching, fertility would have rapidly declined if no manuring took place. This, in turn, would lead to new clearances. Further clarification of various aspects of the Neolithic economy, such as shifting cultivation and/or use of permanent plots, requires further research including archaeobotanical investigations (cf. Bogaard 2002; Kreuz and Schäfer 2011).

Between 2870 and 2260 cal. b.c. (54–42 cm), the type of human impact changed. Considering the strong representation of Pteridium and sharp declines in Betula, Ulmus, Pinus and Fraxinus, intense grazing in the context of cleared woodland is assumed. This is also supported by a sharp increase in Poaceae between 2420 and 2310 cal. b.c. (45–43 cm) which is regarded as indicative of grassland expansion. These changes seem to parallel the emergence of the single grave culture (SGC) in northern Europe (Müller et al. 2010). Unfortunately, there is little archaeological evidence of settlement pattern and subsistence practice during the SGC period in the study area (Strahl 1990; Nösler et al. 2011). The results from Bockholter Dose show only minor changes which might imply intensification of land use on a very local level which is also favoured by Bakker (2003) for the Drenthe Plateau. That the soil became exhausted due to crop cultivation has often been cited as an argument for the abandonment of grazed and arable land at the end of the TRB (Behre and Kučan 1994; Wiethold 1998). This may also explain the developments seen in the Holschkenfehn profile at this time. The longevity of the subsequent regeneration phase, involving first birch and then oak, supports the idea of soil exhaustion and ultimately the abandonment of settlement at Holschkenfehn.

The character of the human activity cannot be so easily be reconstructed on the basis of the evidence provided by the Bockholter Dose profile. However, changes in woodland composition and a trend to more open vegetation are in synchrony with phases of TRB settlement intensification and reduction as reconstructed by pollen indicator taxa from Holschkenfehn (Fig. 9). This suggests that settlement patterns are regional in character. There appears to have been high variability in climate in northern Europe at the beginning of the Neolithic (Charman 2010). Investigation of lake sediments and tree-ring data suggest warm summers with mild winters and increasing wetness in the early Neolithic (after ca. 4000–3800 cal. b.c.) (Leuschner et al. 2002; Dreibrodt et al. 2012). After that, the climate became progressively drier until 2600 cal. b.c. while temperatures declined considerably at 3350 cal. b.c. (Leuschner et al. 2002; Dreibrodt et al. 2012). Based on the data obtained so far, no direct connection between regional settlement activities and changing climatic conditions can be made for the Hümmling region.

Conclusions

High percentages of AP pollen indicate that, prior to and during the early Neolithic, closed canopy woodland dominated in Hümmling. Woodland composition changed during the Neolithic but early opening-up of the woodlands may be under-estimated due to shifts in tree composition from low to high pollen producers (cf. Rasmussen 2005) and also increased overall AP production due to increased light availability as a consequence of human disturbance.

The mid Holocene decline in Ulmus pollen representation in Hümmling predates the classical elm decline in northern Europe by ca. 300 years and is only weakly expressed in both profiles. The decline as expressed in these profiles may be attributable to an early farming impact, independent of any disease factor which may not have been important anyhow given the sparse elm population (cf. Bakker 2003).

The timing of the overall vegetation changes as reflected in the shift in compositional turnover of AP taxa and the PCA data from Bockholter Dose are in good agreement with the chronology of Neolithic cultures in north-western Europe. Initially, farming was probably small scale and the responses of the vegetation to farming were accordingly varied.

On the basis of the pollen data, there appears to be little or no change in type of human impact in the region during the early Neolithic (prior to 3520 cal. b.c.), but human impact increased considerably after 3520 cal. b.c. and continued until 2260 cal. b.c. A change from a subsistence-type farming economy with little wood pasture and arable farming to more intensive wood pasture coincides with the emergence of the SGC. The underlying factors that influenced these settlement dynamics have still to be clarified, especially the possibility that they are connected with changing climatic conditions (cf. Berglund 2003, Kalis et al. 2003; Schulting 2010).

The palynological evidence from prior to the emergence of the TRB does not unambiguously prove the presence of Neolithic people in the study area. On the other hand, the vegetation changes parallel developments seen in pollen profiles from neighbouring regions such as the Netherlands and Schleswig–Holstein where the Neolithic transition is regarded as a gradual process that included the adoption of various Neolithic subsistence strategies while essentially a Mesolithic way of life was maintained. It is assumed that vegetation at a regional level was relatively resilient to low human impact but that the influence of the TRB led to compositional change in the woodlands, and to strong leaching and ultimately exhaustion of soils that is visible, especially at a local scale during the early SGC.

Apart from questions concerning vegetation dynamics and settlement history, insights into regional vegetation developments may contribute to a better understanding of early greenhouse effects and climate change by assessing land-cover changes in northern Europe during the mid Holocene (Ganopolski et al. 1998; Ruddiman 2003; Kaplan et al. 2011) and hence deserve further careful attention.